The Immunomodulatory Properties of L-Rhamnose

Mimmi Lundahl M.Sc. Immunology

Supervisors: Prof. Ed Lavelle and Prof. Eoin Scanlan

Report prepared for the degree of

Doctor of Philosophy, Immunology

School of Biochemistry and Immunology, School of Chemistry

Trinity Biomedical Sciences Institute

Trinity College Dublin

2020

I hereby certify that this report is entirely my own work, and that the contents have not been published elsewhere in paper or electronic form unless indicated through referencing.

i

Abstract

L-Rhamnose is a non-mammalian ubiquitously found on the surface of both commensal and pathogenic bacteria. Previous publications had identified that L-rhamnose-rich Mycobacterium tuberculosis glycolipids, and their structural derivatives, pHBADs, were able to aid this pathogen’s ability to escape immune elimination by repressing protective immune responses. A key immune cell for combatting M. tuberculosis is the macrophage, an innate immune cell present in essentially all tissues. A distinguishing feature of macrophages is their polarisation combined with plasticity; the ability to adopt distinct phenotypes. These are simplified into the pro-inflammatory and bactericidal, “classically activated” M1 macrophages and the “alternatively activated” Th2-promoting and anti-inflammatory M2 macrophages. To combat M. tuberculosis, M1 macrophage activation is critical.

In the research presented herein, it is demonstrated that L-rhamnose skews macrophage polarisation away from a bactericidal phenotype and enhances M2 characteristics. Furthermore, it is revealed that L-rhamnose is capable of inducing macrophage innate memory, causing responses elicited by subsequent stimuli, a week after L-rhamnose incubation, to yield a more anti-inflammatory and anti-bactericidal profile. This is presented both as induction of the regulatory cytokine IL-10, as well as reduced expression of iNOS mRNA.

To relate these immunomodulatory properties back to the bacteria, two synthetic chemistry strategies were employed: the conjugation of L-rhamnose onto bacteria-sized particles and the synthesis of pHBAD analogues. The first strategy was used to mimic the bacterial surface presentation of the as a more biorelevant model to investigate immunomodulation by L- rhamnose. The second strategy was used to investigate whether L-rhamnose was an essential structural element for pHBADs to induce immunomodulation. Whilst a particle conjugation protocol was developed, the optimised conditions were not translatable between manufactured batches of particles and thus this strategy had to be abandoned. On the other hand, by investigating synthetic pHBAD analogues it was confirmed that L-rhamnose indeed does confer immunomodulatory properties to the molecule. Summarily, it appears that L-rhamnose confers M. tuberculosis with immunomodulatory properties that protects it from macrophage bactericidal responses. This information can be used in future research, not only to target L- rhamnose containing compounds to ameliorate suffering caused by M. tuberculosis, but also by using L-rhamnose as a starting point, this research could lead to the discovery of more immunomodulatory compounds.

ii

Abbreviations

# 2-DG; 2-deoxy A-B AA; Antimycin-A Abs; absorbance aCD3; anti-CD3 ADCC; -dependent cell-mediated cytotoxicity AKT; protein kinase B ARDS; acute respiratory distress syndrome Arg1; arginase 1 ASGPR; asialoglycprotein receptor ATP; adenosine triphosphate BALF; broncho-alveolar lavage fluid BCG; Bacillus Calmette-Guerin BMDCs; bone marrow-derived DCs BMDMs; bone marrow-derived macrophages BSA; bovine serum albumin C-D C3; complement component 3 CD, cluster of differentiation cGAS; cyclic GMP-AMP synthase COX-2; cyclooxygenase-2 CpG; CpG oligodeoxynucleotide 1826 CR3; complement receptor 3 CT; cycle threshold CTL; cytotoxic T cell CTLA-4; cytotoxic T-lymphocyte-associated Protein 4 DAMPs; danger-associated molecular patterns DAP12; DNAX activating protein 12kDa DC; dentritic cell DCM; dichloromethane Dectin; DC-associated C-type lectin DesTCR; Désiré T cell receptor

iii

DLS; dynamic light scattering DMEM; Dulbecco’s Modified Eagle Medium dNTPs; deoxy nucleotides triphosphates E-F EAE; experimental autoimmune encephalomyelitis EDC; 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide EDTA; ethylenediaminetetraacetic acid ELISA; enzyme-linked immunosorbent assay ESI; electrospray ionisation ETC; electron transport chain EVLP; ex vivo lung perfusion FACS; fluorescent-activated cell sorting FCCP; carbonyl cyanide-p-trifluoromethoxyphenylhydrazone FCS; foetal calf serum FcγR; Fcγ receptor Fizz1; found in inflammatory zone 1 OR resistin-like alpha G-H G-CSF; granulocyte-colony stimulating factor D-GlcNAc; N-acetyl D-glucosamine GLUT1; glucose transporter 1 H37Rv; Gamma-irradiated whole cells of Mycobacterium tuberculosis, strain H37Rv H3K4me3; histone H3 lysine 4 trimethylation HEK; human embryonic kidney cell line Hib; Haemophilus influenzae Type B HIF-1α; hypoxia-inducible factor 1α HMBC; heteronuclear multiple bond correlation HPRA; Health Products Regulatory Authority HRMS; high-resolution mass spectrometry HRP; horseradish-peroxidase HSQC; Heteronuclear Single Quantum Correlation I-J IBD; inflammatory bowel disease

IC50; 50% of maximal inhibitory concentration IFNβ; interferon beta IFNγ; interferon gamma

iv

Ig; Immunoglobulin IL; interleukin IL-13R; IL-13 receptor IL-4R; IL-4 receptor iNOS; inducible nitric oxide synthase INT; tetrazolium salt IR; infrared IRF; interferon regulatory factor ITAM; immunoreceptor tyrosine-based activation motif ITIM; immunoreceptor tyrosine-based inhibitory motif JAK; janus kinase L-M LDH; lactate dehydrogenase LPS; lipopolysaccharide MALDI; matrix assisted laser desorption ionisation ManLAM; -capped lipoarabinomannan MCP1; monocyte chemotactic protein 1 M-CSF; macrophage colony-stimulating factor M-CSFR; macrophage colony-stimulating factor receptor MFI; mean fluorescence intensity Mgl; macrophage -type C-type lectin MHC I; major histocompatibility complex class I MHC II; major histocompatibility complex class II MIP-2; macrophage inflammatory protein-2 m-p-a; methyl-p-anisate MR; mannose receptor MTA; 5′-methylthioadenosine mTOR; mechanistic target of rapamycin mTORC; mTOR complex MyD88; myeloid differentiation primary response 88 N-O NAD+; nicotinamide adenine dinucleotide NADH; nicotinamide adenine dinucleotide NED; N-1-napthylethylenediamine dihydrochloride NF-κB; nuclear factor kappa-light-chain-enhancer of activated B cells

v

NHDF; normal human dermal fibroblasts NHS; N-hydroxysuccinimide NK cell; natural killer cell NLRP3; NLR family pyrin domain containing 3 NMR; nuclear magnetic resonance NOD; non-obese diabetic NOD-2; nucleotide-binding oligomerization domain-containing protein 2 OD; optical density OPD; O-phenylenediamine dihydrochloride OVA; ovalbumin oxphos; oxidative phosphorylation P-Q PALS; phase analysis light scattering PAM3CSK4; tripalmitoyl-S-glyceryl-cysteine PAMPs; pathogen-associated molecular patterns PBMC; peripheral blood monocyte PBS; phosphate buffered saline, pH 7.4 PCR; polymerase chain reaction PDI; polydispersity index PDIM; phthiocerol dimycocerosate PD-L2; programmed cell death 1 ligand 2 PEG; polyethylene glycol PGE2; prostaglandin E2 PGL; phenolic glycolipid pHBAD; para hydroxybenzoic acid derivative pIA; pHBAD I analogue PIM; phosphatidyl-myo-inositol mannosides PLA; poly(lactic acid) PLGA; poly(lactic-co-glycolic acid) PPARγ; peroxisome proliferator-activated receptor gamma PRR; pattern-recognition receptor qPCR; quantitative real-time PCR Q-Tof; Q-Tof Premier Waters Maldi-quadrupole time-of-flight R-S L-Rha; L-Rhamnose

vi

RnaseOUT; Ribonuclease Inhibitor RNS; reactive nitrogen species ROS; reactive oxygen species Rot; rotenone RPMI; Roswell Park Memorial Institute RT; reverse transcription SHP; SH2 domain-containing protein tyrosine phosphatases Siglec; sialic acid-binding immunoglobulin-like lectin STING; cGAS-stimulator of interferon genes SYK; spleen tyrosine kinase T-Y TGF-β; transforming growth factor beta Th; T helper cell TLC; thin layer chromatography TLR; toll-like receptor TNFα; tumour necrosis factor alpha Treg; regulatory T cell TRIF; TIR-domain-containing adapter-inducing interferon-β VEGFa; vascular endothelial growth factor alpha WGP; whole β- particles WNV; west nile virus WT; wild type Ym1; chitinase-like protein 3

Publications

M.L.E. Lundahl, E.M. Scanlan, E.C. Lavelle, Therapeutic potential of as regulators of macrophage activation, Biochem. Pharmacol. 146 (2017) 23–41. doi:10.1016/j.bcp.2017.09.003.

D.D. Barnes, M.L.E. Lundahl, E.C. Lavelle, E.M. Scanlan, The Emergence of Phenolic as Virulence Factors in Mycobacterium tuberculosis, ACS Chem. Biol. 12 (2017) 1969–1979. doi:10.1021/acschembio.7b00394.

vii

Acknowledgements

First and foremost I want to thank my two supervisors Ed and Eoin; you both have shown me the utmost respect, given credit when it was due and most importantly believed in me throughout. You gave me chances to improve, and provided guidance, help and support when it was needed. Thank you both for giving me the opportunity to grow into the researcher, presenter and writer I am today, someone with the confidence to face even greater challenges, and to keep learning.

To everyone past and present in the Scanlan Lab, thank you for making me feel welcome when I started and for teaching me synthetic chemistry – especially Danielle and Moss – I would have been lost in the columns and NMRs without all of you. Also a big thank you to Lauren and Dylan for making compounds for me, especially to Dylan – you were an absolute star throughout and I am so grateful. Thank you also to everyone for your friendship and encouragement throughout these four years.

Also a special thank you to Joanna Vasconcelos and Prof. Paula Colavita for your help with characterising the particles. Joanna, I’m sorry we never got a chance to go dancing, but you were a delight to work with. Thank you also to Dr. John O’Brien and Dr. Manuel Ruether for all your help with the NMRs, especially for all the HSQCs and HMBCs. Similarly, thank you to Dr Gary Hessman for all your help with the mass spectrometry.

In the Lavelle Lab, a big thank you to both Filipa and Natalia for all the help and advice over the years: you helped me when I needed it the most – both emotionally and academically, and helped me grow and have more confidence in myself, for which I will always be grateful. Thanks also to Aoife for introducing the lab to the concept of innate training, which became a cornerstone for my project as well. To everyone, past and present, thank you all for your support, encouragement, friendship, hugs and brainstorming sessions throughout these four years – you all made it a joy to come in every day – thanks for all the lunch chats and pub pints!

To my wonderful family, I fail to find words that express the constant love and support I feel every day thanks to you all. Thanks for all the lyckosparkar, the hugs, the reminders that I’m loved, for all the great food when I come home, and for making every effort to make each visit special. To Materoo and Pappa, thank you for never putting any pressure on me – on the contrary always telling me of how proud you are no matter what, and reminding me of my achievements and how far I’ve come. To Spex, thanks for being the best big sister. You and John, thank you both for all the great times we have when we’re together, and for all your support throughout.

A special thank you to Laura for being the best housemate and friend I could ever ask for.

A big thank you to James, Julie, Anton and Jenny for being the best of friends.

Last but not least, to Niall, I could not have done this without you. You sympathised with every failed experiment, celebrated every victory, no matter how small. You ordered pizza at the end of long days, provided tea during thesis writing, took care of me when I was sick and provided a never-ending supply of humour and hugs to keep me going throughout. You never doubted I’d get through every challenge or obstacle that came my way, giving me the courage I needed to do just that. Thank you for everything you do for me every day.

viii

Table of Contents

Abstract ...... ii Abbreviations ...... iii Publications ...... vii Acknowledgements ...... viii Chapter 1. Introduction ...... 1 1.1 Innate vs Adaptive Immunity ...... 1 1.2 Macrophages and Carbohydrates ...... 5 1.2.1 Sialic Acids Mediate Immunosuppressive Responses ...... 9 1.2.2 D-Mannose shows context-specific immune activation profiles ...... 12 1.2.3 D-Glucose Based Polymers ...... 19 1.2.4 Immunometabolism: a Requisite for Polarisation ...... 26 1.3 L-Rhamnose ...... 33 Chapter 2. Materials and Methods ...... 38 2.1 List of Materials for Chemical Synthesis ...... 38 2.2 Chemical Synthesis ...... 40 2.2.1 General Methods ...... 40 2.2.2 Rhamnose Ligand Synthesis ...... 41 2.2.3 Particle Functionalisation ...... 47 2.2.3 Endotoxin Removal and Dry-Freezing Compounds...... 47 2.2.4 pHBAD I Analogues (pIAs) Characterisation (Synthesis Carried out by Dylan Lynch, B.Sc.) ...... 48 2. 3 Materials for Cell Work ...... 52 2.3.1. General Cell Culture Materials ...... 52 2.3.2 Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR) Materials ...... 54 2.3.3 Enzyme-Linked Immunosorbent Assay (ELISA) Materials ...... 56 2.3.4 Flow Cytometry Materials ...... 57 2.3.5 Seahorse System Assay ...... 58 2.3.6 Measurement of Nitric Oxide (NO) Production ...... 59 2.3.7 Measurement of Cytotoxicity by Release of Lactate dehydrogenase (LDH) ...... 59 2.4 Cell Culture Methods ...... 60 2.4.1 Mice ...... 60 2.4.2 Protocols for Cell Culture ...... 60 2.4.3 Stimulation of Splenocytes with Anti-CD3 ...... 62 2.4.4 Production of M-CSF Conditioned Medium from L929 Cells ...... 62 ix

2.4.5 Isolation and Culture of Bone Marrow-Derived Dendritic Cells (BMDCs) ...... 63 2.4.6 Generation of Bone Marrow-Derived Macrophages (BMDMs) ...... 63 2.4.7 RT-qPCR ...... 66 2.4.8 Enzyme-linked Immunosorbent Assay (ELISA) ...... 68 2.4.9 Flow Cytometry ...... 69 2.4.10 Real-time Metabolic Analysis via Seahorse System ...... 70 2.4.11 Nitrate (NO) Production Analysis ...... 70 2.4.12 Lactate Dehydrogenase (LDH) Cytotoxicity Analysis ...... 71 2.4.13 Statistical Analysis ...... 71 Chapter 3. The Immunomodulatory Properties of L-Rhamnose ...... 73 3.1 Introduction ...... 73 3.2 Results ...... 75 3.2.1 L-Rhamnose Inhibits IFNγ Secretion by Splenocytes ...... 75 3.2.2 L-Rhamnose has a Minimal Effect on TLR ligand-Stimulated TNFα and IL-10 Secretion from BMDCs and BMDMs ...... 76 3.2.3 Impact of L-Rhamnose on Macrophage Polarisation ...... 78 3.2.4 Does L-Rhamnose Inhibit Glycolysis? ...... 87 3.2.5 Innate Memory Induced by L-Rhamnose ...... 89 3.3 Discussion ...... 104 3.3.1 Macrophage Polarisation ...... 104 3.3. 2 Is it a Metabolic Mechanism? ...... 106 3.3.3 Macrophage Innate Memory Studies ...... 107 3.3.4 Conclusions and Future Experiments ...... 112 Chapter 4. The Synthesis of L-Rhamnose Functionalised Polystyrene Particles...... 116 4.1 Introduction to functionalised particles ...... 116 4.1.1 Particle Properties ...... 116 4.1.2 Carbohydrate Coating Strategies – Ligand Density is a Major Concern ...... 120 4.1.3 Hypothesis and Aims ...... 125 4.2 Results ...... 126 4.2.1 L-Rhamnosyl Derivative Synthesis ...... 126 4.2.2 Linker Synthesis ...... 133 4.2.3 Testing Sterility of Compounds ...... 135 4.2.4 Particle Functionalisation by EDC Coupling ...... 136 4.3 Discussion ...... 153 4.3.1 Successful Synthesis of Ligands for Particle Functionalisation ...... 153

x

4.3.2 Functionalisation of aminated 1 μm polystyrene particles ...... 156 4.3.3 Conclusions and Future Experiments ...... 163 Chapter 5. Does L-Rhamnose Convey Immunomodulatory Properties to Mycobacterial PGLs? 165 5.1 Introduction ...... 165 5.1.1 PGL and pHBAD Structures ...... 165 5.1.2 Immunomodulatory Properties of PGLs and pHBADs ...... 167 5.1.3 Hypothesis and Aims ...... 174 5.2 Results ...... 175 5.2.1 Comparison of the Immunomodulatory Properties of L-Rhamnose and pHBADs ...... 175 5.2.2 Synthesis of pHBAD I Analogues ...... 176 5.2.3 Immunological studies ...... 178 5.3 Discussion ...... 188 5.3.1 Choice of Compounds ...... 188 5.3.2 Does Hydrophobicity Affect the Immunomodulatory Properties of pHBAD I? ...... 190 5.3.3 Summary and Conclusions ...... 195 Chapter 6. General Discussion ...... 198 Conclusion ...... 203 References………………………………………………………………………………………………………………… .. 205 Appendix ………………………………………………………………………………………………………… ...... 226

xi

Chapter 1. Introduction

1.1 Innate vs Adaptive Immunity

The following section will briefly summarise the immune system and how specific responses are tailored to meet with various potential threats. As to the threats in question, these can take shape in the form of exogenous intra- vs. extracellular bacteria, viruses, fungi and various parasites, or be endogenous sources such as physical damage and cancer – each scenario requiring a specific response. It is however also important that potent effector responses are only initiated when there is real danger.

The immune system can be roughly divided into the innate and the adaptive immune system.

The adaptive system is only found in vertebrates and is hallmarked by mature cells only found in certain tissues (including the lymph nodes, liver and spleen), the delayed induction of responses against specific epitopes/antigens and the subsequent generation of long-term memory against these particular antigens (protecting against secondary infection or re-infection). The innate system on the other hand is found in all multicellular organisms, present in practically all tissues, first to detect threats in the form of pathogen- and danger associated molecular patterns

(PAMPs and DAMPs respectively) but does not induce long-term antigen-specific memory.

These rapid responses involve initiation, propagation but also resolution of inflammation; including processes such as immune cell recruitment, phagocytosis of for instance bacteria and debris caused by damage, as well as tissue repair and healing. Examples of innate immune cells include macrophages, dendritic cells (DCs), mast cells, and the granulocytes: neutrophils, eosinophils and basophils (of the myeloid origin), as well as natural killer (NK) cells and innate lymphoid cells (ILCs; of the lymphoid origin).When necessary, the innate system is also responsible for activating the adaptive, notably T helper (Th) cells, by supplying three signals: by secreting cytokines, displaying co-stimulatory molecules (such as cluster of differentiation

(CD)80 and CD86) and by presenting a specific antigen on either major histocompatibility

1 complex (MHC) class I or II. If a danger persists, the adaptive immune response will be activated.

Antigen presentation to Th cells (requiring MHC II) can only be carried out by certain antigen presenting cells: macrophages, DCs (Fig. 1.1 A) and B-cells. B-cells are lymphocytes of the adaptive immune response that become activated via antigen recognition by the B cell receptor, present on the surface of non-activated B cells. Upon activation, B cells divide and differentiate to become antibody producing plasma cells or memory B cells. Without the aid of activated Th cells, only low-affinity immunoglobulin (Ig)M are produced, however in the presence of Th cells, antibody class switching can occur, resulting in the production of high-affinity antibodies, which can be IgG, IgE or IgA, depending on the nature of the antigen and signals received by the B cell.

2

Figure 1.1. Innate immune cells establish a cytokine environment that determines which T helper (Th) cell an activated T cell differentiates into. A) Dendritic cell (DC) presenting an antigen on MHC II, to a naive Th0 cell. The cytokine environment then determines which differentiation pathway the Th cell follows. B) Each square is an example of a type of Th cell: Th1, Th2, Th17 and Treg (regulatory T cell), with their corresponding threats, associated immune cells and cytokines. CTL, cytotoxic T cell; Ig, immunoglobulin; IL, interleukin; NK cell, natural killer cell; TGF-β, transforming growth factor-β.

3

T cells are lymphocytes of the adaptive immune system, of which there are cytotoxic T cells

(CTLs), which recognise antigens presented on MHC class I, and the aforementioned Th cells, recognising antigens presented on MHC class II. Whereas CTLs are known for killing cells recognised as either infected or cancerous, Th cells are the major orchestrators of the immune system, the type of Th cell essentially dictating the overall type of response elicited against a threat (examples in Fig. 1.1 B). For instance, Th1 responses are tailored to intracellular threats, such as intracellular bacterial or viral infection, as well as cancer, as the cytokines secreted enhance activity of cytotoxic cells, such as CTLs but also NK cells and activate macrophages to become more bactericidal. Th2 responses are induced by for instance parasites, are typical of allergic responses, and characteristically also induce tissue remodelling and regeneration. Th17 responses cause enhanced neutrophil recruitment, short-lived cells that have phagocytic and bactericidal capabilities. Regulatory T cells (Tregs) however, suppress other Th-related responses, produce anti-inflammatory cytokines and overall induce tolerance towards specific antigens – preventing immune responses being launched against harmless commensals or importantly against self.

With regard to carbohydrates, sugar epitopes are generally considered poorly immunogenic [1].

A proposed reason for this poor immunogenicity is that they cannot be presented to Th cells, because of their inability to bind to the MHC II [2]. The adaptive immune responses elicited to are therefore generally T cell independent [3], resulting in only B cell activation and the production of low-affinity antibodies [4]. A 2003 study, for instance, demonstrated that a digalactoside-conjugated version of a hen lysozyme, had a 200 fold lower affinity to the relevant

MHC II molecule, compared to the free lysozyme [5]. However, there are exceptions: zwitterionic carbohydrates, i.e. carbohydrates containing both positively and negatively charged functional groups, have been shown to be presented to Th cells [6,7]. that are zwitterionic include the type I pneumococcal used for the PneumoVax®

Streptococcus pneumoniae vaccine [8]. Moreover, protein conjugation to carbohydrates is

4 employed as a successful vaccine strategy to induce T cell activation [3], for instance in the vaccine against Haemophilus influenzae Type B (Hib) [9]. Moreover, whilst carbohydrate epitopes are often poor antigens in isolation, they can still possess potent immunomodulatory properties that affect immune cells of either or both the innate and adaptive immune system.

Examples include the potential adjuvant candidates [10] and β-glucan [11], as well as endogenous regulatory sialic acid epitopes [12].

The research presented herein focuses primarily on one innate immune cell: the macrophage, and elucidating its responses towards the monosaccharide L-rhamnose and L-rhamnose-related compounds. As such, the next few sections will cover this particular immune cell in more detail, as well as give examples of other immunomodulatory carbohydrates, to highlight their properties and the nature of such structure-function interactions.

1.2 Macrophages and Carbohydrates

Carbohydrates are ubiquitous in all organisms. They are prominently present on cell surfaces

[13] and it is estimated that as much as 1% of the human genome is dedicated to glycosylation

[14]. The human is primarily built from nine , which may appear to limit diversity [15]. However, due to the numerous stereogenic centres, the possibility of either

α- or β-linkages to any of four hydroxyl functional groups – as few as three different monosaccharides can construct more than 1,000 distinct [15,16]. Carbohydrates

5 are thus a source of diverse information, the decoding of which is important as a means of distinguishing self from non-self, and for tailoring immune responses [17–19].

Macrophages are mononuclear phagocytic and antigen presenting cells of the innate immune system, of myeloid origin. Due to their presence in essentially all tissues, these cells have a key role in responding to exogenous and endogenous factors, from products of tissue damage and pathogenic infections, to the introduction of biomaterials and tumorigenesis. A significant feature of macrophages is dynamic plasticity, expressed by their ability to polarise towards distinct activation states [20,21]. As shown in Fig. 1.2, the two “extremes” of these states are the “classically activated” M1 macrophages that propagate inflammation and have bactericidal activity [20,22], and the “alternatively activated” M2 macrophages, which enhance allergic responses, resolve inflammation and induce tissue remodelling [23]. Whilst the M1 and M2 simplification is useful, the reality is a broad spectrum of differentiation states, continuously regulated by a myriad of signals from the microenvironment [20]. This spectrum is an argument in favour of defining macrophages based on the stimuli used to differentiate them – a system that is also regularly employed. However, for this section, the M1 and M2 extremes will be used to aid clarity. Another important point is that macrophages that have attained a certain differentiation state can furthermore be re-programmed – an exciting avenue for therapeutic discovery [20,24].

6

Figure 1.2. Polarisation towards either M1 or M2 results in distinct functional roles. Top schematic demonstrates typical stimuli for polarisation towards either the M1 or the M2 phenotype, as indicated by the grey arrows. Labelled M1 and M2 macrophages are shown with examples of surface markers and secreted cytokines/effectors (an increase in which is shown by red and blue arrows respectively). DAMPs, danger-associated molecular patterns; MR, mannose receptor; PAMPs, pathogen-associated molecular patterns; RNS, reactive nitrogen species; ROS, reactive oxygen species. Servier Medical Art was used to construct this diagram.

7

The current dogma states that M1 polarisation occurs in response to signalling downstream of cytokines such as tumour necrosis factor-α (TNFα) and Th1-secreted interferon gamma (IFNγ) in concert with the recognition of either endogenous or exogenous danger molecules [23]. This combination induces phagocytic, antigen presenting M1 macrophages that secrete inflammatory cytokines including interleukin (IL)-1β, IL-12, IL-6 and TNFα, as well as secretion of proteolytic enzymes and the production of reactive oxygen and nitrogen species (ROS and RNS respectively) [20,23]. These cells play key roles in protection against bacterial, fungal and viral pathogens [25]. By contrast, the Th2 derived cytokines IL-4, IL-13, IL-10 and transforming growth factor (TGF)β stimulate M2 differentiation [23,26]. M2 macrophages are phagocytic, yet that dampen inflammation, in part by secreting cytokines including IL-10 and TGFβ, promote angiogenesis and scavenge debris [20,24,27]. M2 macrophages can additionally be identified by their expression of surface markers such as the macrophage mannose receptor (MR) [28], as well as up-regulation of proteins such as chitinase-like 3-1 (Ym1), found in inflammatory zone-1

(Fizz1) and arginase 1 (Arg1) [25,29]. M2 macrophages via the induction of Arg1 impede M1 bactericidal responses. M1 macrophages upregulate the expression of inducible nitric oxide synthase (iNOS), which uses the amino acid arginine to produce bactericidal reactive nitrogen species nitric oxide (NO). Arg1 in M2 macrophages however as part of the urea cycle also use arginine as a substrate, converting it to ornithine, thereby directly opposing the mechanism of iNOS its bactericidal NO production [30]. Whilst these established markers are convenient, macrophages can concurrently display characteristic M1 and M2 markers [31,32]. Apart from cytokines, there are numerous additional signals – including carbohydrates – that can influence macrophage polarisation towards either extreme.

When considering cellular interactions with carbohydrates, there are some aspects to bear in mind, some of which are quite distinct from for instance protein interactions. These aspects include size, polarity, charge, a general lack of functional group variability and subsequent low binding affinity. These properties result in epitopes for instance being recognised by several

8 receptors, and conversely, that receptors and enzymes may interact with a range of carbohydrate motifs. The following examples of immune-modifying carbohydrates, taken from disease states where macrophage responses are a key feature, have partly been chosen to illustrate some of these interaction aspects and the bearing they have on the response elicited.

They have also been chosen to demonstrate the range of ways carbohydrates can induce immune effects, from the receptors involved, to training capabilities and metabolic changes.

The scenarios addressed include cancer [33], fungal [34] and importantly Mycobacterium tuberculosis [31,35] infection, which is where our interest in the monosaccharide L-Rhamnose began.

1.2.1 Sialic Acids Mediate Immunosuppressive Responses

Eukaryotic cells are richly coated in sialic acids (Fig. 1.3): structurally diverse nine-carbon monosaccharides that are negatively charged at physiological pH [36–38]. Derived from , these sugars are composed of a six-membered ring, with a three- carbon side chain and a carboxylic acid group at the anomeric position, which can be either α- or β-oriented [37,38]. Sialic acids are important for distinguishing “self” from “non-self” and are detected by sialic acid-binding proteins; prominently by sialic acid-binding immunoglobulin-like lectins (Siglecs) [36,37]. This transmembrane receptor family, of which there are 14 human and nine murine members [39], is a part of the immunoglobulin superfamily-type (I-type) lectins, which is in itself a subset of the C-type (calcium dependent) superfamily of lectins, and are found on numerous immune cells including macrophages [28,37,40]. Because of their immunomodulatory properties, it is not surprising that sialic acids have been implicated in the pathogenesis of many disease states, an important example of which is cancer [12,20,38].

Elevated expression of sialic acids on the surface of cancerous cells (hypersialylation) has been correlated with tumour metastasis [41] and mediates immunosuppressive effects on immune cells, including macrophages via Siglec signalling [12]. 9

Figure 1.3. Structure of the most common sialic acid: N-acetyl neuraminic acid (Neu5Ac, in black), α-2,3-, α-2,6- and α-2,8- linked to either galactose (grey) or to Neu5Ac, and a schematic of a mammalian cell membrane, emphasising the presence of sialic acids on its surface. Ac: acetyl group [42,43]. Structures were drawn using Chemdraw software.

Siglec signalling can be mediated by distinct pathways, depending on the siglec in question. The presence of a basic amino acid residue in the transmembrane region facilitates the association of some siglecs with DNAX activating protein 12kDa (DAP12), which possesses an immunoreceptor tyrosine-based activation motif (ITAM) – this association in turn allows further signal transduction via spleen tyrosine kinase (SYK) recruitment and activation [28,39].

Inhibitory signals, on the other hand, involve the phosphorylation of tyrosine-based motifs on the cytoplasmic tail of the Siglecs themselves, the most common being the immunoreceptor tyrosine-based inhibitory motifs (ITIMS) – the phosphorylation of which results in high affinity interaction with the SH2 domain-containing protein tyrosine phosphatases, SHP-1 and SHP-2

[28,39,44]. These phosphatases become activated when binding to ITIMs and are thought to provide a negative signal [28]. Whilst the initiation of these signalling pathways have been identified, further downstream mediators remain to be elucidated [28,39].

What is known about Siglec signalling is that it can clearly enhance immunosuppressive responses. In murine models of endotoxin tolerance there was an increase in sialyltransferase activity, resulting in heightened surface expression of α-2,3- and α-2,6-linked sialic acids [45].

10

This was accompanied by upregulation of Siglec-1 and enhanced secretion of the anti- inflammatory cytokine TGFβ [45]. Furthermore, knocking down sialyltransferase activity not only abrogated TGFβ secretion, but also heightened TNFα secretion [45]. Similarly, an α-2,6- sialylated tumour antigen has been demonstrated to stimulate TGFβ production from human

THP1 macrophages expressing Siglec-15 – a lectin detected on the surface of macrophages from human cancers in the liver, rectum and lung [46]. In both cases (Siglec-1 and Siglec-15), TGFβ secretion was dependent on DAP12 recruitment and Syk signalling [45,46].

The potential of exploiting inhibitory signalling by macrophage Siglecs has been investigated as a means of dampening inflammation. Targeting murine Siglec-E or its human orthologues,

Siglecs-7 and -9, displayed promising efficacy in a murine model of acute respiratory distress syndrome (ARDS) and a human ex vivo lung perfusion (EVLP) model – inducing a higher level of

IL-10 secretion in both cases [47]. A dimer of the most common endogenous sialic acid in mammals [37], α-2,8- N-acetylneuraminic acid (Fig. 1.3), functionalised onto the surface of 150 nm poly(lactic-co-glycolic acid) (PLGA) particles, attenuated the secretion of the M1-related cytokines IL-6 and TNFα by both murine peritoneal macrophages and human monocyte derived macrophages [47]. Concurrently, these sialic acid-coated particles amplified IL-10 secretion, which further indicates enhancement of an M2 profile. These effects were not induced by soluble form of the dimer nor the uncoated particles, implying either a synergistic property of the two factors combined, or the effect of multimeric expression of the sialic acid. Fluorescent labelling however revealed that the sugar-coated particles interacted with Siglec-E on murine macrophages, and the anti-inflammatory properties of the functionalised particles was abrogated by blocking Siglec-E with an antagonist antibody [47], suggesting the sialic acid dimer had to be presented on particles to be recognised efficiently. This example illustrates how carbohydrate-protein interactions tend to have low affinity, and thus require multivalent presentation to enhance overall binding avidity to elicit a response [48,49]. This is how

11 carbohydrates are generally found in nature: on the surface of cells in this case, and it is therefore not surprising that mimicking its presentation yielded greater effects.

1.2.2 D-Mannose shows context-specific immune activation profiles

D-Mannose is an endogenously expressed that is additionally found on the surface of viral, fungal, parasitic and bacterial pathogens, including HIV [50], Candida albicans [51],

Fasciola hepatica [52] and M. tuberculosis [53] respectively. D-Mannose epitopes can be recognised by macrophages via a number of pattern recognition receptors (PRRs), including C- type lectin receptors such as the MR (CD206) [54], DC-associated C-type lectin (Dectin)-2 [55] and macrophage-inducible C-type lectin (Mincle) [28,56], as well as Toll-like receptors (TLRs) -2 and -4 [56]. As many of these PRRs are linked with internalization, mannosylation is used as a strategy to induce specific uptake by immune cells which can enhance subsequent immune responses [57–59].

D-Mannose motifs that can be recognised by the MR and Dectin-2 are α-linked D-mannose residues [60,61]. As previously mentioned, the MR is a commonly used marker for M2 polarisation, and its expression is upregulated by IL-4 and IL-13, although its role in influencing differentiation is poorly understood [28]. It is a type-I transmembrane receptor that constantly recycles between the cell membrane and the early endosomal compartment [62]. While the MR lacks any identified signalling motifs [28,62], it has been associated with immunomodulatory effects [28,62,63]. Defining its specific role and associated signalling pathway(s) is challenging, especially as some studies suggest that signalling is mediated via collaboration between it and various other PRRs, such as TLR2 [64], TLR4 and Dectin-1 [65].

Dectin-2 is a type II transmembrane receptor of the C-type lectin family [66,67], that is highly conserved between mice and humans, with 75% amino acid identity between the orthologues

[66]. Although it lacks any known signalling domains, upon recognition of high D-mannose- containing epitopes, Dectin-2 associates with the Fcγ receptor (FcγR) to initiate signalling

12 responses via its ITAM motif [51,66,67]. Dectin-2 is prominently present on mature DCs [67,68] and its surface expression has been shown to be upregulated on macrophages in several inflammatory models, including Candida albicans challenge [69,70]. Furthermore, the receptor has been associated with promoting an M1-associated pro-inflammatory profile upon recognition of high D-mannose-containing epitopes [70,71].

ManLAM Attenuates M1 Polarisation

With regard to M. tuberculosis, the outer cell wall is dominated by mannosylated lipoglycoconjugates, comprising phosphatidyl-myo-inositol mannosides (PIMs), lipomannan, and

D-mannose-capped lipoarabinomannan (ManLAM) [53,55,63] the latter of which has been shown to have immunomodulatory properties [54,63,72]. ManLAM is composed of a D-arabinan and D-mannan core, and at the non-reducing termini it can be capped by up to three D- mannose residues [53,55]. This moiety is present on the surface (Fig. 1.4) of the virulent M. tuberculosis laboratory strain H37Rv, as well as attenuated Mycobacterium bovis (Bacillus

Calmette-Guerin, BCG) and it has been demonstrated to contribute to immune evasion by hindering macrophage polarisation towards an M1 profile [54,72]. The outermost (α-1,6) D- mannose residues of ManLAM [55,63] induce peroxisome proliferator-activated receptor γ

(PPARγ) up-regulation in both murine [54] and human [63] macrophages via an MR-mediated mechanism. PPARγ is a nuclear receptor, the activation of which is known to promote M2 differentiation [73–75]. This in turn has been demonstrated to stimulate IL-8 production and

M2-associated cyclooxygenase-2 (COX-2) expression and prostaglandin E2 (PGE2) secretion in human monocyte-derived macrophages. In addition, the use of antagonistic ManLAM ssDNA aptamers to block ManLAM interaction with the MR enhanced secretion of the M1 effectors IL-

1β, IL-12, NO and iNOS expression and decreased secretion of IL-10 by murine peritoneal macrophages [54].

13

Figure 1.4. The structure of Mycobacterium tuberculosis D-mannose-capped lipoarabinomannan (ManLAM, D-mannose residues in black, residues in grey) and a schematic representation of the mycobacterial cell wall: emphasising the presence of ManLAM on the surface. Structure was drawn with Chemdraw software.

Injection of another ManLAM ssDNA aptamer with the TB vaccine BCG significantly enhanced iNOS expression in macrophages, from both the peritoneal cavity and the broncho-alveolar lavage fluid (BALF), enhanced Th1 responses and also enhanced protection of mice from subsequent virulent M. tuberculosis H37Rv aerosol challenge, when compared with BCG alone – as shown by fewer colony forming units in the spleen and lungs [72]. Further investigations in a 14 rhesus monkey model of M. tuberculosis challenge, demonstrated that the addition of the aptamer similarly heightened BCG vaccine efficacy, as measured by increased IFNγ production by peripheral blood mononuclear cells (PBMCs) and reduced lung bacterial burden 5- and 17- weeks post M. tuberculosis H37Rv infection [72]. In vitro activation of naïve murine CD4+ T cells, by CD3-, CD28- and ManLAM stimulation, moreover demonstrated that blocking the ManLAM-

MR interaction boosted intracellular levels of IFNγ and IL-17. Furthermore, it has been reported that surface expression of the MR on DCs, co-cultured with Désiré T cell receptor (DesTCR) T cells – T cells specific for an endogenous MHC I ligand – attenuates IFNγ secretion and cytotoxic activity when compared with MR deficient DCs [76]. The attenuation was attributed to MR- mediated up-regulation of cytotoxic T-lymphocyte-associated Protein 4 (CTLA-4) [76]. Overall, it appears that the MR has immunomodulatory properties and that ManLAM promotes M2 macrophage polarisation, the blocking of which has been demonstrated to enhance BCG vaccine efficacy [72].

α-Mannan Enhances Pro-Inflammatory Responses

Fungal cell walls are largely composed of carbohydrates and typically have an inner layer comprising β- and , with outer layers composed of and glycoprotein structures [34]. In the case of the most of common opportunistic fungal pathogen, C. albicans,

80-90% of its outer cell wall is carbohydrate [51,77] and its outer glycoprotein layer is largely composed of polymers of D-mannose residues forming O- and N-linked [34,61] (Fig.

1.5). N-linked mannans are highly branched structures, comprising of up to 150-200 α-1,6 D- mannose monomers, with α-1,2 and α-1,3 D-mannose as well as phosphomannan side chains, attached to a D-mannose-N-acetyl glucosamine core [77,78]. O-linked mannans are shorter, linear principally comprised of α-1,2 linked D-mannose residues [77,79].

Intraperitoneal injection of mannan extracted from Saccharomyces cerevisiae has been shown

15 to result in enhanced murine macrophage ROS production and TNFα secretion, as well as IL-17A secretion by γδ T cells [80].

Although a direct interaction has not yet been identified, the MR has been implicated in enhancing Th17 responses upon murine Paracoccidioides brasiliensis challenge [65] and employing antagonistic anti-MR antibodies inhibited mannan enhancement of IL-17 secretion from human PBMCs [64]. Moreover α-mannans are specifically recognised by Dectin-2 [61].

Macrophage secretion of the pro-inflammatory cytokines TNFα, IL-6, IL-1α, and IL-1β, following murine in vivo challenge with C. albicans, was attenuated in Dectin-2 deficient mice [81].

Furthermore, knock-down of Dectin-2 by siRNA in the THP1 human macrophage cell line, significantly reduced Aspergillus fumigatus-induced secretion of IL-10, IL-1β and TNFα [70].

These results indicate that α-mannan via either MR and/or Dectin-2 signalling can promote macrophage inflammatory responses.

16

Figure 1.5. The structure of Candida albicans Serotype A mannan, with α-mannosyl residues in black and β-mannosyl residues in grey, and a schematic representation of the C. albicans yeast cell wall: emphasising the presence of mannan on the surface [34,82]. Structure was drawn with Chemdraw software.

Whilst there is clearly compelling evidence indicating that the outer D-mannose epitopes on

ManLAM skew macrophage responses away from an inflammatory M1 profile, the opposite has been demonstrated for α-mannans. That similar D-mannose epitopes are associated with opposing responses seems to be a result of the two structures eliciting signalling via different

PRRs, notably MR and Dectin-2. A possible explanation for this could be differing affinity of the epitopes for the two receptors. However, the same outermost D-mannose residues of ManLAM

[55,63,71] have been shown to induce Dectin-2-dependent secretion of pro-inflammatory IL-6,

TNFα and macrophage inflammatory protein-2 (MIP-2) from dendritic cells [71]. Moreover, inflammatory responses to mannans have been linked to MR interaction [64]. Apart from possible batch variation, this dichotomy would firstly support the hypothesis that the MR

17 collaborates with other receptors to induce signalling and can thus participate in opposing functional outcomes [28,64,65]. This would argue that the macrophage immunomodulatory effects of ManLAM may be caused primarily via signalling through an unidentified receptor.

Another possible explanation for the opposing responses to these similar mannosyl epitopes is contextual receptor availability. Macrophage Dectin-2 expression is heightened by numerous pro-inflammatory stimuli, such as C. albicans challenge [70]. D-Mannose-recognition in a pro- inflammatory environment may thereby bolster macrophage polarisation and function towards an M1 phenotype – at least partly mediated by the up-regulation of Dectin-2. This highlights the importance of not only considering immunomodulatory epitopes in isolation, but also in the specific context of interest, to fully appreciate their immune properties. Furthermore, these studies only consider responses induced by D-mannose when it is part of a polysaccharide, rather than attempting to investigate the monosaccharide alone. Recently published research however, found that a very high concentration of the monosaccharide D-mannose, given orally, was able to both prevent the development of inflammatory type I diabetes in non-obese diabetic (NOD) mice, and suppressed ovalbumin-induced airway inflammation in another mouse model [83]. These anti-inflammatory effects were due to D-mannose inducing TGF-β production and subsequent regulatory T cell differentiation. The mechanisms by which D-mannose induces

TGF-β production could not be fully elucidated; however it was shown to be at least partially dependent on an increase of T cell ROS production. Not only did D-mannose induce higher levels of ROS in naïve T cells in vitro, but it also increased fatty acid oxidation – a metabolic change that was hypothesised to be the source of the ROS [83]. Overall these examples show how an immunomodulatory carbohydrate such as D-mannose can potentially cause numerous immune effects by several mechanisms. Moreover, context and size are important factors to consider – a monosaccharide can have significantly distinct properties to that of its polysaccharide counterparts, a statement that is not only true for D-mannose, but also D- glucose.

18

1.2.3 D-Glucose Based Polymers

D-Glucose based polymers are highly abundant in nature, with (β-1,4-glucose) being the most abundant polysaccharide, followed by chitin (β-1,4-N-acetyl glucosamine). Chitin, and its de-acetylated derivative chitosan, are immunomodulatory polymers [84] that have been extensively investigated as biomaterials [84–86], including scaffold design for bone tissue engineering [86,87] and scaffolds for repair of nerve degeneration [88,89]. Importantly, due to their immunomodulatory, bactericidal and fungicidal properties, chitin, chitosan and other derivatives have been used to develop wound dressings [90]. The distinct physicochemical properties of chitin and chitosan however allow differing interactions and subsequent immunomodulatory effects. Another immunomodulatory glucose-based polymer is β-glucan

(prominently β-1,3-glucose), a highly conserved component of the fungal cell wall [11,34,91], currently considered as a possible antigen for anti-fungal vaccines [92,93].

M1 Re-polarisation and Training by β-Glucan

β-Glucan (Fig. 1.6 A) is the name given to a diverse family of β-glucose-based polymers, which are not selectively found in fungi but also found amongst algae, plants and bacteria [94].

Varying physicochemical properties, such as solubility [95], molecular weight, and helix conformation [96] found amongst differing isolated extracts (e.g. curdlan, and ) and synthesised forms, affect the type and extent of bioactivity that is elicited [94]. As an example, regarding isolated fungal derivatives of β-glucan, it has been argued that only insoluble particulate forms with high molecular weight can efficiently induce Dectin-1 responses

[34,95,97].

19

Figure 1.6. Candida albicans cell wall and the structures of its β-glucose-based carbohydrate polymers. A) Structure of fungal β-Glucan with β-1,3-glycosyl residues in black, and β-1,6-glucosyl residues in grey. B) Schematic representation of the C. albicans yeast cell wall C) Structure of chitin (β-1,4-N-acetyl glucosamine) and its de-acetylated derivative chitosan (β-1,4- glucosamine). [34,84,94,98,99]. Structures were drawn with Chemdraw software.

Fungal glucan is commonly composed of short β-1,6- or β-1,3- glucosyl side chains, β-1,6-linked to the majority β-1,3-glucose core [98,100,101], through which the polymer is also attached to other carbohydrate structures, such as the inner chitin layer [34,66], (Fig. 1.6 B). This polymer is prominently recognised by DCs and macrophages via Dectin-1, which is structurally very similar to Dectin-2, in that it also is a type II transmembrane receptor of the C-type lectin family [66].

Dectin-1 specifically recognises β-1,3-glucan epitopes and mediates signalling and subsequent responses via an ITAM present on its cytoplasmic tail [34,77].

Macrophage recognition of β-glucan through Dectin-1 is a pro-inflammatory event, that promotes M1-associated effectors, including iNOS, IL-6, IL-12, TNFα, and IL-1β, whilst down regulating M2-linked IL-10 and Arg1 [11]. A similar cytokine profile was identified in the serum of mice vaccinated with Saccharomyces cerevisiae-derived whole β-glucan particles (WGPs),

20 compared with vehicle controls [95]. Furthermore, oral WGP administration has been shown to polarise M2-like tumour associated macrophages (TAMs) towards an M1 profile in a murine model of Lewis lung carcinoma (LLC) [11]. Downstream of these initial effects, β-glucan-Dectin-

1 interaction has moreover been demonstrated to stimulate CD4+ and CD8+ T cell proliferation in a mouse tumour model [11], and work synergistically with TLR4, TLR2 and the MR to promote anti-fungal Th17 responses during Paracoccidioides brasiliensis [65,102] and C. albicans infection [64]. Furthermore, intra-muscular vaccination of mice with synthesised β-glucan polymers protected mice from subsequent lethal C. albicans infection: inducing β-glucan -specific protective IgG antibodies [103]. β-glucan therefore can polarise macrophages towards an M1 profile and has been shown to re-polarise M2-like TAMs towards an M1 profile in a cancer setting via Dectin-1 signalling [11].

Fungal β-1,3-glucan has additionally gained a lot of attention in recent years for its ability to protectively prime innate immune cells, to respond more strongly to subsequent stimuli in a non-specific manner, a phenomenon referred to as innate memory or training [104,105]. Pre- incubation of isolated human monocytes with β-glucan 24 hours before heat-killed C. albicans stimulation was shown to enhance TNFα and IL-6 secretion in a dose-dependent manner [106].

In contrast other stimuli, such as mannan, did not exert this effect [106]. Similar experiments with secondary stimulation one week after priming demonstrated that this enhancement of

TNFα was also seen with a range of other stimuli: TLR4 ligand lipopolysaccharide (LPS), TLR1/2 ligand tripalmitoyl-S-glyceryl-cysteine (PAM3CSK4), as well as with heat-killed gram positive and gram negative bacteria [107]. Furthermore, this protective effect has also been observed in vivo, where priming with C. albicans protected mice from lethal secondary challenge [106], and pre- stimulation with β-glucan increased survival in mice with Staphylococcus aureus induced sepsis from 40- to 90% [107]. The vaccine BCG has similar training effects, mediated via nucleotide- binding oligomerization domain-containing protein 2 (NOD-2) [108]. BCG’s non-specific boosting

21 of innate immune responses is already successfully employed clinically to treat bladder cancer

[109,110]. This is referred to as adjuvant immunotherapy [110].

How this training effect is realised has recently been hypothesised to be related to cell viability

[105]. When TNFα and IL-6 concentrations were normalised to the number of viable murine macrophages or human monocytes, it was shown that the cytokine production per cell was not altered, but the number of cells was, suggesting that β-glucan primarily augments cell viability

[105]. However, there is accumulating evidence showing that Dectin-1 recognition of β-glucan does induce epigenetic changes to cytokine, PRR [106] as well as metabolic [107] gene promoters, thus corroborating the training hypothesis [104,106,107,111]. β-glucan induced epigenetic changes in peritoneal macrophages further suggested a skewing of responses towards an inflammatory profile, as histone H3 lysine 4 trimethylation (H3K4me3) was elevated at the promoters of the cytokines TNFα, IL-6 and IL-18 [106]. This was further corroborated by transcriptome analysis of human monocytes, where mRNA levels of TNFα and IL-6 were again elevated, although the M2 markers CD163 (a scavenger receptor) and monocyte chemotactic protein 1 (MCP1) were also heightened [106]. Moreover, the training effect of β-glucan was inhibited with the use of the methyl transferase inhibitor 5′-methylthioadenosine (MTA), further indicating that epigenetic modification is central for the effect to occur [106]. Therefore, the enhanced inflammatory phenotype induced by β-glucan in vivo [105–107] indicates it is able to stimulate non-specific protective immune responses towards secondary stimuli.

This innate training property demonstrates yet another way that carbohydrates can induce immune-modifying responses, inducing more long-term effects. β-glucan is furthermore not the only sugar capable of augmenting innate immune responses in this way. Priming isolated human monocytes with chitin isolated from S. cerevisiae has been shown to enhance secretion of the pro-inflammatory cytokines TNFα and IL-6 following secondary stimulation with TLR3, -8 and -9 ligands in vitro, and protected mice from subsequent C. albicans challenge in vivo [112].

22

The Physicochemical Properties of Chitin and its Derivatives Determine their

Immunomodulatory Properties

Chitin is a polymer of β-1,4-linked N-acetyl D-glucosamine (D-GlcNAc) and it is the second most abundant biopolymer in nature after cellulose [84,99,113] (Fig. 1.6 C). As well as being a conserved component of the fungal cell wall [34], chitin is also a structural entity found in crustaceans, insects and as part of the microfilarial sheath of nematodes [99,113,114]. Chemical de-acetylation of β-1,4-N-acetyl glucosamine yields chitosan, which has distinct physicochemical properties from chitin. For this reason the degree of acetylation is important for distinguishing between the two polysaccharides, which can vary significantly between preparations of both polymers. For the purpose of this section, chitin is defined as 87-100% acetylated.

There is a current lack of consensus regarding chitin induced immune responses. For instance, chitin is associated with mediating downstream Th2 differentiation during fungal Aspergillus fumigatus infection [114] and M2 macrophage polarisation during helminth Nippostrongylus brasiliensis infection [115]. Intraperitoneal administration of ≤ 100 μm chitin particles triggered recruitment of macrophages with upregulated M2 related genes, including Arg1, Ym1 and Fizz1

[115]. Induction of M2 macrophage effectors correlates with Th2-biased responses elicited by helminth infections [116]. However chitin particles of < 70 μm have been shown to induce macrophage pro-inflammatory TNFα secretion [117] and prevent Th2 differentiation [118].

There are numerous factors that can contribute to these discrepancies, such as the degree of acetylation, average molecular weight of the polymer, variation in sources and extraction procedures, as well as the purity versus contamination of preparations; demonstrating a need for full characterisation of the sugars used.

Size is a particle feature that affects the type and intensity of the immune response elicited by numerous materials [119–121] for reasons including differences in uptake [121,122] and

23 changes to surface area-to-volume ratio affecting bioactivity [123]. Vertebrates express chitinases that can digest chitin into smaller particles [124], and the size of these particles seems to affect which PRRs are triggered and thereby has a major impact on subsequent immune responses [113,118]. Chitin particle (< 70 μm) induction of TNFα secretion was significantly reduced in macrophages from TLR2-deficient mice; which was further completely abolished when a Dectin-1 inhibitor was added [117]. However, MR-enhanced phagocytosis of smaller (< 40 μm) particles stimulated IL-10 secretion from murine macrophages [117] and human PBMCs [125]. This IL-10 secretion was lost when macrophages were either TLR9 or NOD-

2 deficient [125]. TLR9 is known to detect double stranded DNA [126], yet pre-treatment with

DNase I did not alter the secretion of IL-10 [125]. That NOD-2 is implicated is also unexpected, given that NOD-2-induced responses to structurally similar peptidoglycans augment pro- inflammatory responses [127,128]. In summary, the physicochemical properties of chitin have a major impact on what immune responses this polysaccharide can induce; the full extent of which are yet to be fully elucidated. Another common modification of chitin is chemical de- acetylation, which yields chitosan – which also has major implications for what immune responses can be induced.

De-Acetylation Causes Chitosan to Activate the NLRP3 Inflammasome

Whereas there is lack of consensus on the nature of macrophage responses to chitin, there are striking differences between the responses triggered by chitin and chitosan. The conversion of chitin into chitosan involves de-acetylation of the NHAc residues (Fig. 1.6 C) furnishing an overall cationic polymer. Importantly, the change in physicochemical properties imparts chitosan with an ability to bind to DNA [129] and induce NLRP3 inflammasome activation and

Th1 responses [10,130–132]. The NLRP3 inflammasome is a cytosolic complex composed of NLR

Family Pyrin Domain Containing 3 (NLRP3) protein and caspase-1, and its two-step activation ultimately results in the secretion of pro-inflammatory IL-1β and IL-18 [131]. Chitosan has been

24 implicated in triggering several mechanisms of NLRP3 activation, including lysosomal destabilization [131,133], potassium ion efflux and mitochondrial stress [131].

Chitosan-triggered mitochondrial stress additionally causes the release of mitochondrial DNA, which in turn activates the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes

(STING) pathway, finally leading to the secretion of type I interferons [10]. Murine studies additionally reveal that chitosan shows great promise as a vaccine adjuvant, capable of stimulating Th1 responses [10,132] which are a key target for vaccines against pathogens such as HIV and M. tuberculosis [134]. Comparatively with chitin, macrophage IL-1β secretion in response to chitosan has been shown to be correlated to size; when comparing chitosan particles from < 20 μm to > 100 μm, as determined by filtration, the smallest particles were the most potent stimulators [130].

By contrast to these pro-inflammatory effects, primary human monocyte-derived macrophages grown on a film of chitosan have been shown to gain an M2-like profile – displaying a moderate increase in IL-10 and TGFβ, compared with same cells grown on tissue culture polystyrene [135].

This would indicate that the form or size of chitosan significantly impacts on the immune responses elicited. Chitosan and also chitin have been extensively investigated as biomaterials used for tissue engineered scaffolds [84,87,136]. Such scaffolds are attractive for the construction of biomedical implants – man-made devices introduced in a patient either to replace or support a damaged biological structure. Accumulating evidence indicates that M2 macrophages are critical for clinical success; by preventing harmful long-lasting inflammation and promoting tissue remodelling and healing [137–139]. Macrophages exposed to polymeric chitosan have been suggested to adopt a more M2-like profile both in vitro [135] and after implantation in vivo [140] when compared to chitin. However, while this is a promising response initially, chitosan’s ability to propagate inflammation once degraded is a potential long-term concern.

25

Whilst a plethora of receptors have been associated with chitin induced responses, few have been associated with chitosan mediated effects. The most recurring lectin mentioned in the literature is the MR, the blocking of which by competitive antagonist or anti-mannose antibody has been observed to decrease chitosan-induced TNFα [141,142]. How the de-acetylation causes such a striking difference in immune profile is not fully understood, although a theory is that steric hindrance caused by the acetyl group prevents chitin interacting in a way that chitosan can. Another, more likely, theory (as an acetyl group is not that bulky) is that the cationic nature of de-acetylated polymer, as opposed to chitin and most other sugars, allows extended interaction with the negative cell surface, causing either enhanced specific or non- specific receptor interactions [121,123,143]. Chitin and chitosan overall are important examples of how physicochemical properties can drastically alter immune responses in general, and towards carbohydrates specifically.

1.2.4 Immunometabolism: a Requisite for Polarisation

Immunometabolism is a rapidly growing field, the importance of which is becoming more and more apparent. Macrophage activation induces proliferation and cytokine production, amongst other effects, that requires an increase in adenosine triphosphate (ATP) and anabolic precursor production [144]. However, as summarised in Fig. 1.7 macrophage differentiation routes are tightly linked with distinct signalling and metabolic programmes [74,75]. M1 macrophage differentiation involves activation of transcription factors, such as nuclear factor kappa-light- chain-enhancer of activated B cells (NF-κB) and interferon regulatory factors (IRFs) [75], as well as the activation of mechanistic target of rapamycin (mTOR) complex 1 (mTORC1) [74,145] and subsequent augmentation of glycolytic metabolism, including hypoxia-inducible factor 1α (HIF-

1α) mediated up-regulation of glucose transporter 1 (GLUT1) and the enzyme lactate dehydrogenase (LDH) [146]. This polarization route also hinders oxidative phosphorylation

26

(oxphos) [147,148], causing electrons to be utilized to make ROS, which in turn aids HIF-1α stabilization [147,149].

Figure 1.7. M1 and M2 macrophages have distinct metabolic programmes. D-Glucose is taken up by glucose transporter 1 (GLUT1), and its oxidation is initiated by glycolysis. Oxidation can then either be finalised by lactate dehydrogenase (LDH), to form lactate as the final product, or can continue via oxidative phosphorylation (oxphos) in the mitochondria, to yield carbon dioxide. Left side of the panel shows M1 polarisation: lipopolysaccharide (LPS) stimulation via toll- like receptor 4 (TLR4) initiates signalling that activates mechanistic target of rapamycin complex 1 (mTORC1) and subsequent hypoxia-inducible factor 1α (HIF-1α) activation. M1 polarisation also hinders oxphos, which leads to heightened reactive oxygen species (ROS) production, which stabilises HIF-1α. HIF-1α induces a higher level of glycolytic metabolism. Right side of the panel shows M2 polarisation: interleukin 4 (IL-4) stimulation via the IL-4 receptor (IL-4R) causes signalling via the Janus kinase (JAK)1/JAK3- signal transducer and activator of transcripton 6 (STAT6) pathway and also seemingly activates mTORC2 and subsequent protein kinase B (AKT). STAT6 and mTORC2 have been proposed to cause the upregulation of interferon regulatory factor 4 (IRF4), which in turn causes increased glycolytic and oxphos activity. [74,75,147–151]. Grey arrows and blue shaded arrow, sequence of events; dashed arrows, possible sequence of events; red or green shaded upward arrows, increase in activity; red shaded downward arrow, decrease in activity. Servier Medical Art was used to construct this diagram.

M2 polarisation contrastingly promotes oxphos activity [148,150] and involves signalling via several pathways, including the Janus kinase (JAK)1/JAK3-signal transducer and activator of transcription (STAT) 6 pathway [74,150,151]. Expression of certain M2 markers, such as Arg1,

Fizz1 and macrophage galactose-type C-type lectin (Mgl) 2 have also been reduced by blocking mTORC2 and protein kinase B (also known as AKT) activation [74,151]. STAT-6 and mTORC2

27 activation have both been proposed to lead to activation of IRF4, which is associated with increased levels of glycolysis, which is central to M2 macrophage differentiation [150].

These metabolic changes are dependent upon and influenced by various carbohydrate metabolites, such as succinate [152], lactate [33] and the availability of D-glucose [153,154]. As such, these carbohydrates have potent immunomodulatory properties, which include influencing macrophage polarisation.

2-Deoxy Glucose

Initial oxidation of D-glucose through glycolysis is hindered by the use of the pharmacological inhibitor 2-deoxy glucose (2-DG) that inhibits the enzyme hexokinase [153,155]. Administration of 2-DG six hours prior to LPS has been shown to significantly protect mice against lethal septic shock, and simultaneously reduced serum levels of M1-associated TNFα and NO [153]. This inhibition of M1 effectors seems to be specifically caused by 2-DG reducing IL-1β expression, as it has been shown in vitro that this sugar blocks IL-1β mRNA production [152], but not IL-6 [152] or TNFα expression [149,152].

Similarly, regarding the other differentiation extreme: macrophage stimulation in vitro with 2-

DG reduced IL-4 induced expression of the M2 activation markers Fizz1 and programmed cell death 1 ligand 2 (PD-L2), and intraperitoneal injection with the helminth Heligmosomoides polygyrus reduced macrophage recruitment and proliferation [150]. These studies corroborate that both M1 and M2 differentiation require a heightened level of glucose uptake and subsequent glycolysis [75,147] – the blocking of which attenuates macrophage activation and function.

Succinate

The metabolism of succinate is a link between the Krebs cycle (also known as citric acid cycle) and the electron transport chain (ETC) via succinate dehydrogenase (also known as Complex II) 28

[156]. It has recently become apparent that heightened levels of succinate in macrophages causes ROS production through reverse electron transport [148,149], which increases expression of the transcription factor HIF-1α and subsequent IL-1β mRNA, as well as pro-IL-1β protein levels [149,152]. This was for instance shown by the use of anti-oxidants, and by blocking succinate oxidation in the mitochondria with the competitive antagonist malonate; both abrogated the succinate induced expression of IL-1β and HIF-1α [149]. Additionally, succinate stimulation inhibited LPS-induced anti-inflammatory IL-10 and IL-1Ra secretion [149], evidently skewing macrophage responses towards an M1 profile.

Lactate

At the end of glycolysis, the final product, pyruvate, can either be used as fuel for oxidative phosphorylation or for fermentation to lactate. Whilst oxphos is the most efficient means to get the greatest yield of ATP per D-glucose molecule, fermentation to lactate is faster. In cancer cells a characteristic metabolic shift, known as the Warburg effect or aerobic glycolysis, is the switch to near exclusive use of glycolysis and subsequent fermentation as the source of ATP, despite the presence of oxygen [157]. This leads to high levels of lactate secreted in the tumour microenvironment [158], which in turn has immunomodulatory effects on macrophages

[33,146].

Stimulating murine macrophages with exogenous lactate in vitro enhanced expression of several M2 markers including Arg1, Fizz1 and prominently vascular endothelial growth factor a

(VEGFa) [33], which is important for promoting angiogenesis. Diminished production of VEGFa, caused by deletion of lactate dehydrogenase, was additionally shown to be rescued by lactate supplementation [146]. High lactate concentration was also shown to correlate with an enhanced presence of M2 macrophages, as identified by the markers CD163 and macrophage

29 colony-stimulating factor receptor (M-CSFR), in human head and neck squamous cell carcinoma patients [158].

Furthermore, blocking lactate production by global conditional deletion of LDH – the enzyme that catalyses pyruvate’s conversion to lactate – has been shown to reduce tumour growth in a murine lung cancer model, whilst also leading to enhanced levels of iNOS in lung tissue [146].

Lactate’s immunoregulatory effects were very similar to those induced by hypoxic conditions

(0.1% oxygen), and lactate was unable to induce expression of M2 markers in HIF-1α deficient cells [33], which would imply that macrophage responses to lactate are connected to hypoxia- related responses. This route of lactate and hypoxia re-programming macrophages to an M2 phenotype via HIF-1α contradicts its proposed role in promoting M1 metabolism [74,147], demonstrating the need for further investigation in this field. Nevertheless, lactate is clearly a potent inducer of M2 macrophage differentiation, which in this case seems to be tightly linked with hypoxia.

These examples clearly demonstrate how affecting metabolism can have potent immune- modifying effects. Moreover, interfering with metabolism is an immunomodulatory mechanism that can be employed by small monosaccharides and carbohydrate metabolites. The primary carbohydrate of interest for this research is a monosaccharide, and as such altering metabolism is a potential mechanism for immunomodulation, along with other mechanisms presented herein (summary of examples in Table 1.1).

30

Table 1.1 Examples of immunomodulatory carbohydrates and a summary of their immune- mediated effects

Sensor(s) Cells Carbohydrate Specific Structure Response Refs implicated investigated

Chitin TR: ↑IL-6, TNFα Monh 100% Ac - [112] TR: ↑IL-1β, IL-6, IL-10, TNFα In vivo

Dectin-1, TLR2 ↑IL-6, TNFα 97% Ac MΦm, PBMCs NOD-2, TLR9 ↑IL-10 [125] Majority 1-10 μm P h, in vivo MR Phagocytosis

87% Ac - ↑Arg1, Ym1, Fizz1, MR MΦm, in vivo [115] ≤ 100 μm P

70-100 μm P - - MΦm in vivo [117]

m [113,1 ≤ 70 μm P Dectin-1, TLR2 ↑TNFα, IL-17A MΦ in vivo 17]

≤ 40 μm P Dectin-1, TLR2, ↑IL-10 MΦm, in vivo [117] MR

20-50 μm P - ↑Arg1, iNOS, PD-1 MΦm [118]

Inflammasome activation Chitosan 20% or 2% Ac - MΦ h [133] ↑IL-1α, IL-1β

Inflammasome activation, 24% Ac - MΦ & DCsm [131] ↑IL-1β

> 100 μm P - ↑IL-1β (minimal) MΦm [130]

20 - 100 μm P - ↑IL-1β (more) MΦm [130]

≤ 20 μm P - ↑IL-1β (most) MΦm [130]

- ↑TNFα, iNOS (modest) MΦm [159] ≤ 25% Ac Oligochitosan - ↑IL-1β, TNFα MΦm [142] (3 - 9 residues) MR ↑TNFα MΦm [141]

m cGAS/STING ↑IFNβ, CXCL10 DCs [10]

10-25 % Ac cGAS/STING, ↑IFNγ In vivo NLRP3 [10]

- ↓IL-1β MΦm [152] 2-Dexoy

Glucose - ↓IL-1β, TNFα In vivo [152]

- ↓TNFα, NO In vivo [160]

31

β-Glucan ↑ IL-1β, IL-6, TNFα, iNOS, CD40, CD86 MΦ, OT T Dectin-1 ↓IL-10, Arg [11] Whole glucan cellsm. ↓CD4+ & CD8+ T cell particles proliferation

↑IL-1β, IL-6, IL-17, MIP1α, TNFα Serumm - [95]

Dectin-1, MR, ↑IL-6, IL-17, TNFα DCs & [65] TLR4 ↑Th17 splenocytesm

TR: ↑IL-6, TNFα, Dectin-1 Monh [106] epigenetic changes β-1,3-glucan - TR: ↑IL-6, IL-10, TNFα Serumm [105]

- TR: ↑TNFα, epigenetic changes Monh [107]

- TR: Epigenetic changes MΦh [111]

Lactate ↑VEGFa, Arg1, Fizz1, Mgl1, - MΦm [33] Mgl2

BLOCK: ↑iNOS, CD86, CD197 - MΦm ↓VEGFa, CXCL10 [146]

↑iNOS, IFNγ - In vivo ↓VEGFa, TGFβ [146]

α-Mannoses BLOCK: ↑IL-1β, IL-12, iNOS, NO MR MΦm in vivo [54] ↓IL-10

Mannose cap of MR BLOCK: ↑iNOS, NO MΦm [72] ManLAM,

↑IL-8, COX , PGE h [63] α-1,6-mannoses MR 2 2 MΦ

m Dectin-2 ↑IL-2, IL-6, IL-10, TNFα DCs [71]

Dectin-2 ↑IL-1α, IL-1β, IL-6, TNFα MΦm in vivo [81]

- ↑TNFα MΦm [80] α-Mannans, - ↑IL-17 In vivo α-1,6- (majority), α-

1,2- & α-1,3- Dectin-1, TLR2 mannoses ↑IL-17 [64] MR PBMCs h ↑IL-17, IL-10

↑IL-1β, IL-6, IL-10, IL-12p40, Dectin-2 DCsm TNFα [61]

Sialic acid ↓TNFα Siglec-1 MΦm, DCsh&m [45] ↑TGFβ α-2,3- and α-2,6-

sialic acids ↑IL-4, IL-10, TGFβ Siglec-Em MΦm ↓IL-2, iNOS, ROS, NO [44]

Siglec-7 & -9h ↓IL-6, TNFα, IL-8 MΦh α-2,8-di-Neu5Ac [161] ↑IL-10

32

m m Siglec-E ↓IL-6,TNFα MΦ [161]

m Siglec-E ↓IL-6, TNFα in vivo [161] ↑IL-10

α-2,6-Neu5Ac Siglec-15 ↑TGFβ MΦh [46]

Siglec-7 & -9 BLOCK: ↑IL-6, IL-10, CD80 DCsh [162]

Global cell surface h Siglec-7 & -9 BLOCK: ↑IFNγ, proliferation DCs & PBLs [162]

Succinate ↑IL-1β, ROS - MΦm [149] ↓IL-10, IL1Ra

- ↑L-1β MΦm [152]

Ac, acetylation; CXCL10, C-X-C motif chemokine 10; FBGC, foreign body giant cell; Mon, monocyte; mw, molecular weight; MΦ, macrophage; Neu, Neutraphils; Neu5Ac, N-acetyl neuraminic acid; P, particles; PD-

1, programmed cell death protein-1; TR, training; h, human; m, murine.

1.3 L-Rhamnose

L-Rhamnose is a deoxy monosaccharide which is anabolically used by plants, fungi and bacteria

[163]. As it is not found naturally in humans, it and structures containing this sugar can be targeted by the immune response. It is for instance commonly part of both the cell wall and surrounding capsule of numerous bacterial pathogens [163], as well as commensals, such as

Lactobacillus rhamnosus [164]. Examples from pathogenic bacteria include the O-antigen of

Salmonella typhimurium [165] and Shigella dysenteriae [166] LPS, as well as mycobacterial phenolic glycolipids (PGLs) [53,167].

Regarding immunomodulatory effects, very little is known about L-rhamnose. Rhamnosoft® is a

L-rhamnose-rich polysaccharide, isolated from Klebsiella pneumoniae (non-pathogenic strain), which seemingly has anti-inflammatory properties, and is marketed as such [168]. Additionally,

L- rhamnose-based encapsulating nanofibrils exerted a suppressive effect on murine antibody responses to co-injected phycoerythrin, a fluorescent protein antigen [169]. Furthermore, L-

33 rhamose is a highly abundant (31.5%) sugar in the cell wall of the commensal Lactobacillus rhamnosus [164]. This, as a probiotic, has been proposed to enhance immunotolerogenic effects, such as increased numbers of regulatory T cells [170,171] and prevent the development of allergy [172,173]. Although none of these effects have been categorically attributed to L- rhamnose, we believe that its abundance in these formulations point towards a decisive role for the sugar.

To date, no human or murine receptor has been identified that can recognise the L-rhamnose monosaccharide or epitope [174]. Rhamnose-binding lectins have been identified in fish [175] and there is some evidence that human cells can also detect this monosaccharide. For example, a synthetically produced, α-L-rhamnose-BSA (bovine serum albumin) conjugate was shown to bind to membrane-bound lectins on human keratinocytes, and was consequently taken up by these cells [176]. Furthermore, L-rhamnose-rich (50-60%) polysaccharides, obtained from

Klebsiella pneumoniae, stimulated both cell proliferation and collagen biosynthesis of normal human dermal fibroblasts (NHDFs) [177]. Increased doses of these L-rhamnose-rich polysaccharides, and also L-rhamnose monosaccharide alone, triggered increased calcium ion influx (characteristic of lectin signalling) in NHDFs and endothelial cells [178]. Recently, L- rhamnose has also been shown to promote browning of white adipocytes, which was impeded when an inhibitor of the β3-adrenergic receptor was used, thus suggesting that L-rhamnose can potentially bind to this receptor [179]. Stimulation of the β3-adrenergic receptor has been shown to reduce LPS mediated iNOS and IL-1β transcription in human macrophages [180]; indicating once more that L-rhamnose has potential to induce immunomodulation.

Our interest in L-rhamnose began with mycobacterial PGLs, which are present in more virulent strains of Mycobacterium tuberculosis, the bacterial pathogen responsible for human tuberculosis [181,182] a disease responsible for the death of 1.6 million people in 2017 [183].

This ancient intracellular bacterium is internalised by a number of immune cells, but prominently macrophages [31], thus shielding itself from extracellular humoral responses. A

34 strong Th1 response can promote an M1 macrophage phenotype and subsequent bactericidal effectors [134,184]. However, M. tuberculosis also possesses a number of means to prevent Th1 and M1 responses [185–187], including the immune-suppressive properties of a number of cell wall components, such as Man-LAM discussed previously [63,72,167,188,189]. These also include PGLs and the structurally related para hydroxybenzoic acid derivatives (pHBADs) [188], that have been shown to suppress TLR2 induced human macrophage secretion of IL-6 and TNFα

[190] and anti-CD3 (aCD3) stimulated murine splenocyte secretion of IL-17 and IFNγ [167] respectively. The pHBADs in question are presented in Fig. 1.8 and, as can be seen, methylated

L-rhamnose is a major component of these compounds.

Figure 1.8. Structures of para-hydroxybenzoic acid derivatives and the monosaccharide L- rhamnose (highlighted in black in all structures), and a schematic representation of the mycobacterial cell wall. Structures were drawn with Chemdraw software.

Unlike PGLs, pHBADs are expressed in all clinical strains of M. tuberculosis [181,191,192]. This evolutionary conservation does not relate to bacterial structural integrity or viability, as a

35 pHBAD deficient mutant can still replicate [192]. This, along with the previously referenced results, could indicate that the immune-suppressive properties of pHBADs provide M. tuberculosis with an evolutionary advantage by contributing to its pathogenesis. Furthermore, when anti-CD3 stimulated splenocytes were incubated with L-rhamnose, the same trend was observed as with the pHBADs: decreased secretion of cytokines IFNγ and IL-17A, whilst the secretion of IL-10 was unchanged [193]. Succinctly, L-rhamnose was identified as a monosaccharide with little-known yet promising immunomodulatory properties.

Objectives and Hypotheses

Carbohydrates essentially are ubiquitous molecules which play major roles in pathogenesis of various diseases, can help immune cells distinguish between self and non-self, as well as regulating elicited immune responses. Based on previous results obtained in our lab, and from results shown by L-rhamnose related compounds and bacteria, we hypothesised that L- rhamnose had immunomodulatory properties, of a regulatory nature. The primary objective of this research, therefore, was to better elucidate these properties. Furthermore, as carbohydrate interactions are usually multivalent, the use of particles to enhance binding avidity is a useful biomimetic model of these interactions [194,195]. We therefore hypothesised that any regulatory properties L-rhamnose had may be enhanced by multivalent expression on bacteria- sized particles. A third and final hypothesis was based on the L-rhamnose prevalence of mycobacterial pHBADs: that L-rhamnose is a necessary structural component for pHBADs, and in turn PGLs, to impart their immunomodulatory properties.

Based on these hypotheses, the first objective was to investigate immunological responses to L- rhamnose, to elucidate whether L-rhamnose had any immunomodulatory properties. The second and third objectives were then to link these properties back to M. tuberculosis by two synthetic strategies. By constructing functionalised particles to see whether multivalent expression enhanced any L-rhamnose induced immune-repressing properties; and by using

36 synthetic derivatives of pHBAD I to investigate the dependency of the L-rhamnose motif for pHBAD I to induce its immunomodulatory properties.

37

Chapter 2. Materials and Methods 2.1 List of Materials for Chemical Synthesis Table 2.1. List of reagents

Name Formula Supplier

Acetic acid AcOH Sigma-Aldrich

Acetic anhydride Ac2O Aldrich Chemistry

2-2(Aminothethoxy)ethanol C4H11NO2 Aldrich Chemistry

Ammonium chloride NH4Cl Sigma-Aldrich

Ammonium molybdate (NH4)2MoO4 Fluka Analytical

Boron trifluoride diethyl etherate BF3OEt2 Aldrich Chemistry

Copper sulphate CuSO4 Sigma-Aldrich

Ethanethiol C2H5SH Sigma-Aldrich

1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide, EDC C8H17N3•HCl Alfa Aesar

Hydrazine hydrate, 88% N2H4 xH2O Sigma-Aldrich

Magnesium sulphate MgSO4 Fisher Chemical

Methyl 4-Hydroxybenzoate C₈H₈O₃ Sigma-Aldrich

Methyl-p-anisate, m-p-a C8H10O3 Sigma-Aldrich

N-hydroxy succinimide, NHS C4H5NO3 Thermo Scientific

Ninhydrin C9H6O4 Sigma-Aldrich

Phenol C6H5OH Sigma-Aldrich

Phthalic anhydride C6H4(CO)2O Sigma-Aldrich

Potassium Carbonate K2CO3 Sigma-Aldrich

Potassium permanganate KMnO4 Sigma-Aldrich

L-Rhamnose, L-Rha C6H12O5 Carbosynth Ltd

Sodium Na Chemical Analytical

Sodium Deoxycholate C24H39NaO4 Sigma-Aldrich

Sodium chloride NaCl Sigma-Aldrich

Sodium hydrogen carbonate NaHCO3 Fisher Chemical

Sodium hydroxide NaOH Fisher Chemical

Sodium methoxide NaOMe Synthesised

Sodium sulphate Na2SO4 Sigma-Aldrich

Succinic anhydride C4H4O3 Sigma-Aldrich

Sulphuric acid (conc) H2SO4 Sigma-Aldrich

38

Table 2.2. List of Solvents

Name Formula Supplier

Acetone C3H6O Trinity

Ammonia NH3•H2O Fisher Chemical

n-Butanol C4H10O Sigma-Aldrich

Chloroform CHCl3 Sigma-Aldrich

Deuterated Chloroform CDCl3 Sigma-Aldrich

Deuterated Methanol CD3OD Sigma-Aldrich

Deuterium oxide D2O Sigma-Aldrich

Dichloromethane, DCM CH2Cl2 Trinity

Ethanol EtOH Trinity

Ethyl acetate EtOAc Trinity

Hexane C6H14 Trinity

Methanol MeOH Sigma-Aldrich, Trinity

Pyridine C5H5N VWR Chemicals

Sterile water (nonpyrogenic, hypotonic) H2O Baxter

Toluene C6H5CH3 Sigma-Aldrich

Particles used were Aminated Polystyrene Latex particles, 1 (± 0.17) μm, 10% solids in water, with a cationic surfactant present. Lot Number: AM3295-0616, supplied by MagSphere (USA).

39

2.2 Chemical Synthesis

2.2.1 General Methods

Thin Layer Chromatography All thin layer chromatography (TLC) were run on Merck 60 F254 silica gel (pre-coated, 0.2 mm thick, 20×20 cm; Merck Millipore).

All Rf values were calculated by dividing the length travelled a compound by that travelled by solvent, from the compound’s starting point.

Visualisation was carried out with short-wave UV light, molybdenium-, ninhydrin- or permanganate stain.

Molybdenium stain: 5 g ammonium molybdate and 5.3 ml concentrated sulphuric acid (96%) in 100 ml water.

Ninhydrin stain: 1.5 g ninhydrin and 3 ml acetic acid in 100 ml of n-butanol.

Permanganate stain: 1.5 g potassium permanganate, 10 g potassium carbonate and 1.25 ml 1M sodium hydroxide in 200 ml water.

Column Chromatography Column chromatography is used to separate and purify compounds due to their differences in solubility. All were carried out using Silica gel Florisil (200 mesh particle size; Fluka-Analytical). Sand was used to help keep silica gel in a moist state and to prevent disruption of the silica upon addition of solvent.

In vacuo Removal of Solvent Removal of solvent in vacuo refers to the use of reduced pressure to aspire solvent, concentrating the desired product. This was carried out by means of rotary evaporators (Buchi) with additional drying with a high vacuum system, involving a pump (Code No: A652-01-903, Edwards) in combination with liquid nitrogen.

Characterisation A Bruker Ultrashield (600 MHz) spectrometer was employed for 1H (600.13 MHz) nuclear magnetic resonance (NMR) and 13C (150.6 MHz) NMR spectra. Resonances δ, are in ppm units downfield of an internal reference in chloroform (δH = 7.26 ppm, δC = 77.0 ppm) or methanol (δH

= 3.31 ppm, δC = 49.0 ppm). Infrared (IR) spectra were recorded on a Spectrum 100, FT-IR Spectrometer (Perkin Elmer, USA). Mass spectrometry was carried out using a Q-Tof Premier 40

Waters Maldi-quadrupole time-of-flight (Q-Tof) mass spectrometer (UK), equipped with Z-spray electrospray ionisation (ESI) and matrix assisted laser desorption ionisation (MALDI) sources.

ζ-potential and dynamic light scattering (DLS) measurements were carried out using a Malvern Zetasizer Nano-ZS, λ = 633 nm He-Ne laser. The signal was detected at 13° and 173° for ζ- potential and DLS measurements respectively. Hydrodynamic size was determined using refractive index n = 1.59 for latex and ζ-potential was obtained using the same instrument which determines particle velocity and mobility via phase analysis light scattering (PALS). Diluent chosen was 0.1 mM PBS in millipore water, as this gave conductivity between 0.1-0.2 mS/cm for ζ-potential measurements. All DLS and ζ-potential measurements were carried out in triplicate.

Measuring Carbohydrate Concentration

In a 96-well high-binding ELISA plate (Grenier Bio-one), 130 µl concentrated sulphuric acid

(Sigma) was added to 50 µl of carbohydrate samples. After which 30 µl of 5% w/v phenol

(Sigma) was added, followed by rapid pipetting up and down to mix the solution thoroughly.

The plate was heated to 90 °C for 5 minutes, then cooled rapidly to room temperature in a water bath before absorbance was measured at 492 nm. Standard curve was generated with a serial dilution of a known amount of the sugar in question.

2.2.2 Rhamnose Ligand Synthesis

Following synthetic procedure was based on [196].1,2,3,4-tetra-O-acetyl-L-rhamnose

Acetic anhydride (Ac2O, 30 ml, 319.4 mmol) was added dropwise to 1 (5 g, 27.8 mmol), dissolved in 30 ml pyridine (C5H5N). This reaction was initiated at 0 °C on ice, and left to reach room temperature. After 17 hours, TLC with 60% hexane, 40% ethyl acetate (EtOAc) confirmed all the starting material (Rf ≈ 0) was consumed, and a product had formed (Rf = 0.65).

Reaction mixture was washed with approximately 1L deionized water and the product was diluted in EtOAc. Copper sulphate was also added, which turns from blue to purple in the

41 presence of pyridine, to aid in the separation. The EtOAc portion was then dried with magnesium sulphate (MgSO4). The solution was then filtered, and remaining solvent was removed in vacuo. This gave clear oil 2 in 83% (7.87 g, 23.7 mmol) yield.

1 HNMR (400 MHz, CDCl3): 5.97 (1H, s, H-1), 5.26 (1H, dd, J1,2 = 2.0 Hz, J2,3 = 7.8 Hz, H-2), 5.22 (1H, d, J2,3 = 7.8 Hz, H-3), 5.09 (1H, app t, H-4), 3.95-3.87 (1H, m, H-5), 2.13, 2.12, 2.03, 1.96 (3H, s,

+ + COCH3), 1.20 (3H, d, J5,6 = 6.3 Hz, H-6) m/z HRMS (ESI ) calcd. for C14H20O9Na 355.1005 (M + Na) , found 355.0999.

1-O-(2(2-Phthalamidoethoxy)ethyl-2,3,4-tri-O-acetyl)-α-L-rhamnopyranoside

To a solution of 2 (6.51 g, 19.6 mmol, 2 eq) and 9 (2.31 g, 9.8 mmol, 1 eq) in anhydrous DCM, under N2 and with molecular sieves, boron trifluoride diethyl etherate (BF3OEt2, 7.3 ml, 58.8 mmol, 6 eq), was added dropwise. Reaction was initiated at 0 °C and then let to reach room temperature. After stirring for 60 hours, the solution was filtered and the remaining BF3OEt2 was quenched with saturated sodium hydrogen carbonate (NaHCO3) solution. The solution was then washed with first NaHCO3 and then brine, diluted with EtOAc, and subsequently dried with sodium sulphate (Na2SO4) and filtered. The filtrate was concentrated in vacuo. Silica chromatography with 2/3 hexane, 1/3 EtOAc separated remaining 2 (Rf = 0.78) from the final product 3 (Rf = 0.65, 3.66 g, 7.2 mmol, 74% yield). By TLC 2 was identified by molybdenium stain, whilst 3 was visualised by both short-wave UV and molybdenium stain.

21 3 -1 -1 -3 1 [α] D = -34.9 (deg cm g dm ) (c = 0.1 g cm in CHCl3); HNMR (400 MHz, CDCl3): δ 7.88-7.84 (2H, m, Ar), 7.75-7.70 (2H, m, Ar), 5.31-5.24 (2H, m, H-2 and H-3), 5.05 (1H, t, J = 9.9 Hz, H-4),

4.78 (1H, d, J1,2 = 1.5 Hz, H-1), 3.93-3.88 (3H, m, H-5, CH2), 3.78-3.73 (3H, m, CH2, CH), 3.69-3.58

(3H, m, CH2, CH), 2.16 (3H, s, OAc), 2.06 (2H, s, OAc), 2.05 (1H, s, OAc), 1.98 (3H, s, OAc), 1.20

13 (3H, d, J5,6 = 6.3 Hz, H-6) ppm; C NMR (100 MHz, CDCl3): δ 170.1, 170.0, 168.3 (C=O), 134.0

(Ar), 132.1 (C=O), 123.3 (Ar), 97.5 (C-1), 71.1 (C-4), 69.8 (C-3), 69.5 (CH2), 69.1 (C-2), 68.1, 66.9

+ (CH2), 66.3 (C-5), 37.2 (CH2), 21.0, 20.9, 20.8 (AcCH3), 17.4 (C-6) ppm; m/z HRMS (ESI ) calcd. for

+ C24H29NO11Na 530.1638 (M + Na) , found 530.1633; νmax (thin film) 2939 (C-H stretch, aromatic), 1743 (C=O stretch), 1709 (C=O stretch), 1467 (C-H bend, alkane), 1393 (C-C stretch, aromatic),

42

1369 (C-H rock, alkane), 1320 (C-O stretch), 1242, 1217 (C-O or C-N stretch), 1043 (C-O stretch), 720 (C-H, aromatic) cm-1.

1-O-(2(2-Aminoethoxy)ethyl)-α-L-rhamnopyranoside

From 3: hydrazine hydrate (50-60%, 2.7 ml, 48.0 mmol, 16 eq.) was added drop-wise to a solution of 3 (1.5 g, 3.0 mmol, 1 eq) in anhydrous MeOH under N2. After 72 hours solution was filtered and concentrated in vacuo, resulting in a thick mixture of yellow oil and white crystals.

Solvent system used for silica chromatography was composed of 60% EtOH, 20% H2O, 15% EtOAc and 5% acetone, to which a small quantity (less than 1% of total) of ammonia solution (in water) was added, to prevent protonation of the amine in the process. Visualised by molybdenium stain, under these conditions product 4 was isolated (Rf = 0.16, 585 mg, 2.3, 77% yield – probable dissolved silica included) from remaining starting material and unknown side product.

From 6: To a solution of 6 (123 mg, 0.32 mmol, 1 eq), dissolved in anhydrous MeOH (under N2 and with molecular sieves), hydrazine hydrate (50-60%, 0.1 ml, 1.72 mmol, 5.4 eq) was added dropwise. After stirring for 2 hours TLC showed all of 6 had been used up and solution was filtered and concentrated in vacuo. Silica chromatography with the same solvent system as described above gave product 4 (15 mg, 0.06 mmol, 19% yield – probable dissolved silica included).

21 3 -1 -1 -3 1 [α] D = -26.9 (deg cm g dm ) (c = 0.1 g cm in CH3OH); HNMR (400 MHz, CD3OD): δ 4.75 (1H, d, J1,2 = 1.5 Hz, H-1), 3.86-3.80 (2H, m, H-2, CH), 3.73-3.58 (7H, m, H-3, H-5, 2CH2, CH), 3.40 (1H,

13 t, J = 9.6 Hz, H-4), 2.98-2.90 (2H, m, CH2), 1.29 (3H, d, J5,6 = 6.2 Hz, H-6) ppm; C NMR (100 MHz,

CD3OD): δ 100.4 (C-1), 72.5 (C-4), 70.9 (C-3), 70.8 (C-2), 70.5 (CH2), 69.9 (CH2), 68.4 (C-5), 66.3

+ + (CH2), 40.3 (CH2), 16.6 (C-6) ppm; m/z HRMS (ESI ) calcd. for C10H22NO6 252.1447 (M + H) , found

252.1447; νmax (thin film) 3356 (O-H stretch), 2981, 2916 (C-H stretch, alkane), 1655, 1567 (N-H bend, amine), 1454 (C-H bend, alkane), 1382 (C-H rock, alkane), 1308, 1261 (C-O or C-N stretch), 1053, 979 (C-O stretch) cm-1.

43

1-O-(2-(2-N-[4-Carboxy]butanamidoethoxy)ethyl)-α-L-rhamnopyranoside

Succinic anhydride (460 mg, 4.6 mmol, 2 eq.) was dissolved with product 4 (585 mg, 2.3 mmol, 1 eq.) in anhydrous MeOH under N2 and with molecular sieves. After 72 hours solution was filtered and concentrated in vacuo. Silica chromatography (35% MeOH, 65% DCM) isolated product 5 (Rf = 0.14) from remaining starting material, monitored by TLC and permanganate stain. As the solvent system employed dissolves silica; after concentrating in vacuo the resulting yellow oil was dissolved in 5% MeOH, 95% CHCl3 and filtered through glass fibre wool. Concentrating in vacuo once more gave product 5 as a white solid, similar to candyfloss in consistency (674 mg, 1.9 mmol, 80% yield).

21 3 -1 -1 -3 1 [α] D = -27.7 (deg cm g dm ) (c = 0.1 g cm in CH3OH); HNMR (400 MHz, CD3OD): δ 4.75 (1H, d, J1,2 = 1.5 Hz, H-1), 3.86-3.84 (1H, m, H-2), 3.83-3.77 (1H, CH), 3.71-3.68 (1H, m, H-3), 3.68-3.59

(4H, m, CH2, CH, H-5), 3.57 (2H, t, J = 5.5 Hz, CH2), 3.41-3.36 (3H, m, CH2, H-4), 2.63-2.57 (2H, m,

13 CH2), 2.54-2.48 (2H, m, CH2), 1.29 (3H, d, J5,6 = 6.3 Hz, H-6) ppm; C NMR (100 MHz, CD3OD): δ 175.6 (C=O, carboxylic acid), 173.5 (C=O, amide), 100.3 (C-1), 72.6 (C-4), 70.9 (C-3), 70.8 (C-2),

+ 69.8, 69.3 (CH2), 68.4 (C-5), 66.4, 39.1, 30.4, 29.4 (CH2), 16.7 (C-6) ppm; m/z HRMS (ESI ) calcd.

- for C14H24NO9 350.1451 (M – H) , found 350.1456; νmax (thin film) 3325 (O-H stretch), 2982 (C-H stretch), 1715 (C=O stretch), 1639 (N-H bend), 1238 (C-N stretch), 1138 (C-N or C-O stretch), 1051 (C-O stretch), 979 (C-O stretch) cm-1.

44

1-O-(2(2-Phthalamidoethoxy)ethyl)-α-L-rhamnopyranoside

To a solution of 3 (160 mg, 0.32 mmol, 1 eq) in MeOH, drops of sodium methoxide (NaOMe) in MeOH was added, turning the solution clear green, until pH 10 was reached, measured by pH paper. After 1 hour, TLC with ½ hexane, ½ EtOAc, showed no starting material remained, whilst a new spot (both UV active and positive for molybdenium stain at Rf ≈ 0) demonstrated 4 had formed. Dowex® (50WX8) proton donor was added to the solution until pH 7 was reached. Solution was subsequently filtered and the rose-gold filtrate was concentrated in vacuo. This gave product 4 (123 mg, 0.32 mmol, 100% yield) as a red oil.

21 3 -1 -1 -3 1 [α] D = -24.3 (deg cm g dm ) (c = 0.1 g cm in CH3OH); HNMR (400 MHz, CD3OD): δ 7.97 (1H, dd, J = 1.1, J = 7.8 Hz, Ar), 7.66-7.60 (1H, m, Ar), 7.55 (1H, dt, J = 1.4, J = 7.6 Hz, Ar), 7.48 (1H, dd,

J = 1.1, J = 7.6 Hz, Ar), 4.75 (1H, d, J1,2 = 1.5 Hz, H-1), 3.85-3.79 (2H, m, H-2, CH), 3.71 (4H, t, J =

4.7 Hz, 2 CH2), 3.68-3.61 (3H, H-2, H-3, CH), 3.56 (2H, t, J = 5.5 Hz, CH2), 3.42-3.38 (1H, m, H-4),

13 1.28 (3H, d, J5,6 = 6.3 Hz, H-6) ppm; C NMR (100 MHz, CDCl3): δ 138.3 (C=O), 131.6, 129.8,

129.2, 127.5 (Ar), 100.4 (C-1), 72.6 (C-4), 70.9 (C-3), 70.8 (C-2), 69.9, 69.0 (CH2), 68.4 (C-5), 66.5,

+ - 39.5 (CH2), 16.7 (C-6) ppm; m/z HRMS (ESI ) calcd. for C18H22NO8 380.1345 (M - H) , found

380.1345; νmax (thin film) 2820 (C-H stretch, aromatic), 1672 (C=O), 1567 (C=C, Aromatic), 1442 (C-H), 1325 (C-O), 1031 (C-O) cm-1.

4-(2-Ethanoyl-2-ethoxyamide)-butanoic acid

2-(2-aminoethoxy)ethanol (7, 1.5 ml, 15 mmol, 1 eq.) and succinic anhydride (1.5 g, 15 mmol, 1 eq.) were dissolved in anhydrous MeOH, under N2 and with molecular sieves. After leaving it to stir overnight, the solution was filtered and concentrated in vacuo. Silica chromatography, with

50% MeOH and 50% DCM, monitored by permanganate stained TLC, gave product 8 (Rf = 0.54, 1.5 g, 7.3 mmol, 49% yield).

45

1 HNMR (400 MHz, CD3OD): δ 3.72-3.68 (2H, m, CH2), 3.58-3.53 (4H, m, 2CH2), 3.42-3.38 (2H, m,

13 CH2), 2.64-2.57 (2H, m, CH2), 2.54-2.48 (2H, m, CH2) ppm; C NMR (100 MHz, CD3OD): δ 174.9

(C=O, carboxylic acid), 173.3 (C=O, amide), 72.0, 69.2, 60.8, 39.0, 30.1, 29.0 (CH2) ppm; m/z

+ - HRMS (ESI ) calcd. for C8H14NO5 204.0872 (M-H) , found 204.0874; νmax (thin film) 3300 (O-H stretch, alcohol), 2931 (O-H stretch, carboxylic acid), 1715 (C=O stretch), 1632 (N-H bend), 1551 (N-H bend), 1208 (C-N stretch), 1168 (C-O stretch), 1122 (C-O stretch), 1060 (C-O stretch), 885 (O-H bend, carboxylic acid) cm-1.

N-(2-(2-ethanol)ethoxy)phthalimide

Phthalic anhydride (37 g, 250 mmol, 1 eq.) and 7 (26.3 g, 250 mmol, 1 eq.) were dissolved in 450 ml toluene and heated to reflux, with condenser and Dean Stark apparatus set up to eliminate the water produced, as previously reported [197]. After 24 hours, the solution was concentrated in vacuo, with DCM as a co-evaporating solvent, giving an orange, oily yet solid residue. Sticky consistency was credited to the presence of remaining amine starting material. The residue was diluted in DCM and washed with minimal volume of saturated ammonium chloride aqueous solution. This gave product 9 as a cream-coloured powder (38.5 g, 163 mmol, minimum 65% yield).

1 HNMR (400 MHz, CDCl3): δ 7.86-7.81 (2H, m, 2CH), 7.73-7.68 (2H, m, 2CH), 3.89 (2H, t, J = 5.1

Hz, CH2), 3.73 (2H, t, J = 5.1 Hz, CH2), 3.69-3.64 (2H, m, CH2), 3.61-3.56 (2H, m, CH2); m/z HRMS

+ - (ESI ) calcd. for C12H12NO4 234.0766 (M-H) , found 234.0764; νmax (thin film) 3658 (O-H stretch), 2981 (C-H stretch, aromatic), 2886 (C-H stretch, alkane), 1767 (C=O stretch), 1706 (C=O stretch), 1432 (C-C bend, aromatic), 1397 (C-C stretch, aromatic), 1357 (C-N stretch), 1322 (C-N stretch), 1082 (C-O stretch), 1023 (C-O stretch), 718 (C-H “oop”, aromatic) cm-1.

46

2.2.3 Particle Functionalisation

Particle Preparation 6.2 mg of 1 μm aminated polystyrene latex particles (AM-PS, Magsphere, Lot number: AM3295- 0616, standard deviation 0.17 μm) were suspended in 1.6 ml sterile water (nonpyrogenic, hypotonic; Baxter) and centrifuged at 6,000 G for 15 minutes, after which the supernatant was carefully decanted by pipetting, and the particles were re-suspended in 155 μl of 10 mM sodium phosphate buffer (pH 9.3) and sonicated: 70% amplitude, 30 seconds on and 30 seconds off pulse rate and total sonication time of 2 minutes. The particles were then left stirring in the pH 9.3 buffer at room temperature overnight (now at 40 mg/ml).

Particle Functionalisation 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide, (EDC•HCl, 1.15 mg, 6 μmol, 2 eq.) was dissolved in water and equal μmol of NaOH was added to neutralise the hydrochloric acid, giving a final concentration of 11.5 mg/ml. 100 μl of this EDC solution was mixed with 100 μl aqueous 30 mM ligand 8 (0.62 mg, 3 μmol, 1 eq.) or ligand 5 (1.05 mg, 3 μmol, 1 eq.) and 50 μl aqueous 69.1 mg/ml N-hydroxysuccinimide (NHS, 2.76 mg, 60 μmol, 20 eq.). This solution was left stirring at room temperature for 15-20 minutes, at which point 150 μl of 40 mg/ml AM-PS (prepared as specified above) in 10 mM sodium phosphate buffer (pH 9.3) was added to the reaction. This final mixture was left stirring at room temperature for 2 hours. Next, the particles were washed twice, by adding sterile-filtered Phosphate Buffered Saline (PBS, pH 7.4, 10 mM) w/o magnesium, w/o calcium (Biosera) to reach a total volume of 1.6 ml and the particles centrifuged at 6,000 G for 15 minutes and the supernatant consequently removed. After these two washes, the particles were re-suspended in 400 μl PBS (15 mg/ml of particles) and the particles were sonicated: 70% amplitude, 30 seconds on and 30 seconds off pulse rate and total sonication time of 2 minutes.

2.2.3 Endotoxin Removal and Dry-Freezing Compounds

Potential endotoxin contamination was removed from synthesised compounds 5 and 8 as follows: 674 mg of compound 5 and 1.51 g of compound 8 were dissolved in nonpyrogenic and hypotonic sterile water (Baxter; 3 ml and 5 ml respectively), were filter sterilised (0.22 μm, Millipore) and subsequently passed through endotoxin binding columns (Detoxi-Gel column, 1 ml, Thermofisher Scientific), following manufacturer’s set protocol. In short, each column was regenerated with 5 ml 1% sodium deoxycholate in sterile water, washed with 5 ml sterile water, then the sample solutions with dissolved compounds (volumes specified above, one column per 47 compound). More sterile water was added to ensure the column did not dry and that all solution containing each compound was collected.

The solutions were then dry frozen. 2 ml screw-capped tubes were labelled and weighed before dissolved compounds 5 and 8 were divided into 0.5 ml fractions in these tubes. Following freezing with liquid nitrogen, caps were removed and replaced with parafilm. The parafilm was then punctured with a needle before the tubes were placed in the freeze dryer (Christ alpha 1-2 LD plus) and subsequently dried under vacuum, at condenser temperature -55 °C. After freeze- drying overnight, caps were put on again and tubes, now with dried compounds, were weighed once more to establish the amount of compound in each tube; total amount of regained, dried product: 772 mg compound 8 and 411 mg compound 5.

2.2.4 pHBAD I Analogues (pIAs) Characterisation (Synthesis Carried out by

Dylan Lynch, B.Sc.)

The synthesis of each compound is summarised in 5.2.2.

1 H NMR (400 MHz, CDCl3): δ 7.98 (2H, d, J = 8.9 Hz, ortho Ar-CH), 7.07 (2H, d, J = 8.9 Hz, meta

Ar-CH), 5.59 (1H, s, H-1), 4.17 (1H, s, H-2), 4.03-3.97 (1H, m, H-3), 3.89 (3H, s, OCH3), 3.71 (1H, dq, J = 12.1, 6.1 Hz, H-5), 3.58 (1H, t, J = 9.4 Hz, H-4), 1.27 (3H, d, J = 6.1 Hz, H-6) ppm; m/z

+ HRMS: C14H19O7 calcd. for 299.1131 (M+H) , found 299.1123.

48

1 H NMR (400 MHz, CDCl3): δ 6.96 (2H, d, J = 9.2 Hz, ortho Ar-CH)), 6.81 (2H, d, J = 9.2 Hz, meta

Ar-CH), 5.40 (1H, s, H-1), 4.15 (1H, s, H-2), 3.99 (1H, dd, J = 9.3, 3.0 Hz, H-3), 3.85 – 3.77 (1H, m,

13 H-5), 3.76 (3H, s, OCH3), 3.57 (1H, t, J = 9.7 Hz, H-4), 1.29 (1H, d, J5-6 = 6.2 Hz, H-6) ppm; C NMR

(101 MHz, CDCl3): δ 155.1 (Ar-qC), 150.3 (Ar-qC), 117.8 (ortho Ar-C), 114.8 (meta-Ar-C), 98.7 (C-

1), 73.4 (C-5), 71.7 (C-4), 71.1 (C-2), 68.7 (C-3), 55.8 (OCH3), and 17.7 (C-6) ppm; m/z HRMS:

- C13H17O6 calcd. for 269.1025 (M-H) , found 269.1037; νmax 2943 (C-H stretch), 2833 (C-H stretch),

1508 (ArC=C stretch), 1024 (ArC-H stretch) cm-1.

1 H NMR (400 MHz, CDCl3): δ 7.25 – 7.19 (2H, m, meta Ar-CH), 7.02 – 6.94 (3H, m, ortho and para

Ar-CH), 5.50 (1H, s, H-1), 4.17 (1H, s, H-2), 4.03 (1H, dd, J = 9.3, 2.7 Hz, H-3), 3.79-3.72 (1H, m, H-

13 5), 3.64 – 3.57 (1H, m, H-4), and 1.26 (1H, d, J5-6 = 6.1 Hz, H-6) ppm; C NMR (101 MHz, CDCl3): δ

156.2 (Ar-C), 129.7 (meta Ar-C), 122.5 (para Ar-C), 116.5 (ortho Ar-C), 97.9 (C-1), 73.1 (C-5), 71.7

- (C-4), 71.1 (C-2), 69.0 (C-3), and 17.7 (C-6) ppm; m/z HRMS: C12H15O5 calcd. for 239.0919 (M-H) , found 239.0931; νmax 2943 (C-H stretch), 2833 (C-H stretch) 1750 (C=O stretch), 1224 (ArC-H stretch) cm-1.

49

1 H NMR (400 MHz, CDCl3): δ 4.65 (1H, d, J1-2 = 1.0 Hz, H-1), 3.92 (1H, dd, J = 3.2, 1.4 Hz, H-2),

3.73 (1H, dd, J = 9.4, 3.2 Hz, H-3), 3.64 – 3.55 (1H, m, H-5), 3.49 – 3.42 (1H, m, H-4), 3.35 (3H, s,

- OCH3), and 1.31 (3H, d, J5-6 = 6.2 Hz, H-6) ppm; m/z HRMS: C7H13O5 calcd. for 177.0763 (M-H) , found 177.0768.

1 H NMR (400 MHz, CDCl3): δ 5.25 (1H, s, H-1), 4.08-3.99 (2H, m, H-2 and H-5), 3.74 (1H, dd, J =

9.4, 2.9 Hz, H-3), 3.52 (1H, t, J = 9.4 Hz, H-4), 2.70 – 2.52 (2H, m, CH2), 1.32 (3H, d, J5-6 = 6.2 Hz,

- H-6) 1.28 (3H, t, J = 7.4 Hz, CH2CH3); m/z HRMS: C8H15O4S calcd. for 207.0691 (M-H) , found

207.0697.

1 HNMR (400 MHz, CDCl3): δ 7.30 (2H, t, J = 7.9 Hz, meta Ar-CH), 7.07 (2H, d, J = 7.9 Hz, ortho Ar-

CH), 7.05 (1H, t, J = 7.4 Hz, para Ar-CH), 5.52 (1H, dd, J = 10.1, 3.5 Hz, H-3), 5.46 (1H, d, J1-2 = 1.8

Hz, H-1), 5.43 (1H, dd, J = 3.3, 1.8 Hz, H-2), 5.15 (1H, t, J = 10.0 Hz, H-4) 4.00 (1H, dq, J = 12.5, 6.2

Hz, H-5), 2.19 (3H, s, OAc), 2.06 (3H, m, OAc), 2.03 (3H, s, OAc), 1.20 (1H, d, J5-6 = 6.2 Hz, H-6)

13 ppm; C NMR (101 MHz, CDCl3): δ 170.2 (OAc C=O), 156.0 (Ar-qC), 129.8 (Ar-C), 122.9 (Ar-C),

116.5 (Ar-C), 95.8 (C-1), 71.2 (C-3), 69.9 (C-4), 69.1 (C-5), 67.3 (C-2), 21.1 (OAc-CH3), 21.0 (OAc-

- CH3), 20.9 (OAc-CH3), 17.6 (C-6) ppm; m/z HRMS: C18H21O8 calcd. for 365.1236 (M-H) , found

50

365.1242; νmax (thin film) 2946 (C-H stretch), 2834 (C-H stretch), 1736 (C=O stretch), 1019 (ArC-H stretch) cm-1.

1 H NMR (400 MHz, CDCl3): δ 7.00 (2H, d, J = 9.2 Hz, 2 x ortho Ar-CH), 6.83 (2H, d, J = 9.2 Hz, 2 x meta Ar-CH), 5.50 (1H, dd, J = 10.1, 3.5 Hz, H-3), 5.42 (1H, dd, J = 3.5, 1.8 Hz, H-2), 5.33 (1H, d, J1-

2 = 1.8 Hz, H-1), 5.13 (1H, t, J = 10.1 Hz, H-4), 4.02 (1H, dq, J = 9.9, 6.2 Hz, H-5), 3.77 (3H, s, Ar-

OCH3), 2.18 (3H, s, OAc-CH3), 2.06 (3H, s, OAc-CH3), 2.02 (3H, s, OAc-CH3), 1.21 (3H, d, J5-6 = 6.3

13 Hz, H-6) ppm; C NMR (101 MHz, CDCl3): δ 170.2 (OAc C=O), 155.4 (Ar-qC), 150.1 (Ar-qC), 117.8

(Ar-C), 114.8 (Ar-C), 96.6 (C-1), 71.2 (C-3), 69.9 (C-4), 69.1 (C-5), 67.1 (C-2), 55.8 (Ar-OCH3), 21.1

(OAc-CH3), 21.0 (OAc-CH3), 20.9 (OAc-CH3), 17.6 (C-6); m/z HRMS: C19H23O9 calcd. for 395.1342

- (M-H) , found 395.1345; νmax: 2945 (C-H stretch), 2834 (C-H stretch), 1749 (C=O stretch), 1025

(ArC-H stretch) cm-1.

1 H NMR (400 MHz, CDCl3): δ 8.00 (2H, d, J = 9.2 Hz, ortho Ar-CH), 7.11 (2H, d, J = 9.2 Hz, meta Ar-

CH), 5.53 (1H, d, J1-2 = 1.8 Hz, H-1), 5.50 (1H, dd, J = 10.1, 3.5 Hz, H-3), 5.43 (1H, dd, J = 3.5, 1.8

Hz, H-2), 5.16 (1H, t, J = 10.8 Hz, H-4), 3.97-3.90, (1H, m, H-5), 3.89 (3H, s, Ar-COCH3), 2.20 (3H,s,

OAc-CH3), 2.05 (3H, s, OAc-CH3), 2.03 (3H, s, OAc-CH3), and 1.20 (3H, d, J5-6 = 6.3 Hz, H-6) ppm;

13 C NMR (101 MHz, CDCl3): δ 170.2 (C=O), 166.7 (C=O), 159.4 (C=O), 131.8 (Ar-C), 124.7 (Ar-qC),

116.0 (Ar-C), 95.5 (C-1), 90.8 (Ar-qC), 70.9 (C-4), 69.6 (C-2), 68.9 (C-3), 67.6 (C-5), 52.2 (OCH3),

21.0 (OAc-CH3), 20.9 (OAc-CH3), 20.9 (Ac-CH3), and 17.6 (C-6) ppm; m/z HRMS: C20H23O10 calcd.

51

- for 423.1291 (M-H) , found 423.1306; νmax 2947 (C-H stretch), 2835 (C-H stretch), 1735 (C=O stretch), 1071 (ArC-H stretch) cm-1.

1 H NMR: (400 MHz, CDCl3) δ 5.33 (1H, dd, J = 3.4, 1.5 Hz, H-2), 5.23 (1H, dd, J = 10.0, 3.4 Hz, H-

3), 5.19 (1H, d, J1-2 = 1.5 Hz, H-1), 5.09 (1H, t, J = 10.0 Hz, H-4), 4.23 (1H, dq, J = 9.7, 6.2 Hz, H-5),

2.71 – 2.55 (2H, m, CH2), 2.15 (3H, s, OAc CH3), 2.05 (3H, s, OAc CH3), 1.98 (3H, s, OAc CH3), 1.29

(3H, t, J = 7.4 Hz, CH2CH3), 1.23 (3H, d, J5-6 = 6.3 Hz, H-6) ppm; m/z HRMS: C14H21O7S calcd. for

- -1 333.1008 (M-H) , found 333.1016; νmax 2975 – 2902 (C-H stretch), 1738 (C=O stretch) cm .

2. 3 Materials for Cell Work

All materials are from Sigma-Aldrich unless stated otherwise.

2.3.1. General Cell Culture Materials

Complete RPMI 1640 medium (cRPMI) Roswell Park Memorial Institute (RPMI) 1640 medium (Biosera) was supplemented with 2 mM L-Glutamine (Gibco), 50 units/ml penicillin (Gibco), 50 μg/ml streptomycin (Gibco) and 8% (v/w) filter-sterilised, heat-inactivated (56 °C for 30 minutes) foetal calf serum (FCS, Biosera).

Complete DMEM cell culture medium (cDMEM) Complete 4500 mg/L D-glucose Dulbecco’s Modified Eagle Medium (DMEM, Sigma Aldrich) was supplemented with 2 mM L-Glutamine (Gibco), 50 units/ml penicillin (Gibco), 50 μg/ml streptomycin (Gibco) and 8% (v/w) filter-sterilised, heat-inactivated (56 °C for 30 minutes) FCS (Biosera).

Complete DMEM w/o D-glucose (cDMEMG↓, Gibco) was supplemented with 50 units/ml penicillin (Gibco), 50 μg/ml streptomycin (Gibco) and 8% (v/w) filter-sterilised, heat-inactivated (56 °C for 30 minutes) FCS (Biosera).

52

0.88% ammonium chloride (NH4Cl) red blood cell lysis solution 8.8 ammonium chloride was dissolved in 1 litre of sterile water (Baxter) and filter-sterilised with 0.22 μm syringe-driven filter (Millipore).

Trypsin-EDTA 0.25% sterile-filtered 2.5 g porcine trypsin and 0.2 g ethylenediaminetetraacetic acid (EDTA)×4Na/L in Hanks′ Balanced Salt Solution.

Table 2.3. Stimuli used for in vitro cell work

Stimulus Abbreviation Target Supplier CpG oligodeoxynucleotide 1826 CpG TLR9 Oligos Gamma-irradiated whole cells of Mycobacterium tuberculosis, strain H37Rv Various BEI Resources H37Rv Local 2% Lidocaine Hydrochloride Braun anaesthetic Lipopolysaccharide from , LPS TLR4 Enzo Life Sciences serotype R515 Tripalmitoyl-S-glyceryl-cysteine PAM3CSK4 TLR1/2 Invivogen IL-4 Recombinant murine interleukin 4 IL-4 receptor Peprotech type I and II IL-4 Recombinant murine interleukin 13 IL-13 receptor Peprotech type II IL-10 Recombinant murine interleukin 10 IL-10 Peprotech receptor IFNγ Recombinant murine interferon gamma IFNγ Peprotech receptor

53

2.3.2 Reverse Transcription Quantitative Polymerase Chain Reaction (RT- qPCR) Materials

RNA isolation High pure RNA isolation Kit (Roche, IN 46250-0414, USA) was used per manufacturer’s instructions although the volume of DNAse was halved to 50 μl. The elution volume used was 50 μl.

Table 2.4 Reverse Transcription reagents

Reagent Supplier Concentration dNTPs Promega 2.5 mM/NTP Reverse Transcriptase Buffer Promega - (1x) Random primers 5’-NNNNNN-3’/Random Hexamers MWG Biotech 1 μg/μl Ribonuclease Inhibitor (RnaseOUT) Invitrogen 40 U/μl M-MLV Reverse Transcriptase Promega 200 U/μl

qPCR reagents Kapa SYBR® Fast Rox low qPCR mix (KK4622, Kapa Biosystems [Sigma.Aldrich]), using ROX low reference dye in combination with nuclease-free water.

Primers (Invitrogen) against mouse genes were designed to span exon-exon junctions to avoid amplification of genomic DNA. Dissociation curve analysis was performed after each real-time qPCR to exclude non-specific products.

54

Table 2.5. Primers for qPCR

Product Protein mRNA Primer, 5’-3’ * RefSeq(s) mRNA Source Length

FP TACAAGACAGGGCTCCTTTCAG Primer Arginase 1 Arg1 128 NM_007482.3 RP TGAGTTCCGAAGCAAGCCAA BLAST

FP GCTTCTTTGCAGCTCCTTCGT Primer β-Actin Actb 60 NM_007393.5 RP CGTCATCCATGGCGAACTG BLAST

FP CAGCTGATGGTCCCAGTGAAT Primer Fizz 1 Retnla 99 NM_020509.3 RP AGTGGAGGGATAGTTAGCTGG BLAST

FP TCCTGGACATTACGACCCCT NM_010927.4 Primer iNOS Nos2 110 NM_001313921.1 RP CTCTGAGGGCTGACACAAGG BLAST NM_001313922.1

FP ATGGTGGTCCGAGCAGAGAT IFNβ Ifnb1 107 NM_010510.1 [10] RP CCACCACTCATTCTGAGGCA

FP AGGCGCTGTCATCGATTTCTC Primer IL-10 Il10 93 NM_010548.2 RP GACACCTTGGTCTTGGAGCTTAT BLAST

FP GTGTAACCAGAAAGGTGCGTT Primer IL-12p40 Il12p40 142 NM_001303244.1 RP TCGGACCCTGCAGGGAAC BLAST

FP ACGGTGTGCAGACACAACTA Primer MHC II H2-Ab1 74 NM_207105.3 RP ACGACATTGGGCTGTTCAAG BLAST

TATAbox- FP CAGGAGCCAAGAGTGAAGAACA Primer Tbp 120 NM_013684.3 binding protein RP AAGAACTTAGCTGGGAAGCCC BLAST

FP AAGCTCTCCAGAAGCAATCC Primer Ym 1 Chil3 71 NM_009892.3 RP AGAAGAATTGCCAGACCTGTGA BLAST

FP TGAACTTTCTGCTCTCTTGGGT NM_001025250.3 Primer VEGFa Vegfa 124 NM_009505.4 RP AACTTGATCACTTCATGGGACT BLAST NM_001025257.3 * Forward Primer, FP; Reverse Primer, RP. Fizz 1, found in inflammatory zone 1 or resistin-like alpha; IFNβ, interferon beta; IL, interleukin; iNOS, inducible nitric oxide synthase; MHC II, major histocompatibility complex class II; VEGFa, vascular endothelial growth factor; Ym 1, Chitinase-like protein 3.

55

2.3.3 Enzyme-Linked Immunosorbent Assay (ELISA) Materials

For all ELISAs, appropriate solvents (Table 4), concentrations of antibodies and horseradish- peroxidase (HRP)-conjugated streptavidin were dictated by the ELISA kit for each cytokine. The only discrepancies lay in adjustments of top standard concentrations (Table 5) and the volumes employed. All volumes were halved (from 100 μl to 50 μl), except the blocking step, which remained at 200 μl.

ELISA OPD Substrate 20 mg O-phenylenediamine dihydrochloride (OPD) tablet was dissolved in 50 ml phosphate citrate buffer containing 14 μl hydrogen peroxide (30% w/w).

Table 2.6. Solutions for ELISAs

Solution Composition 1:10 diluted 10x PBS in millipore water, with 0.5% 0.5% PBS-Tween/ Wash buffer (v/w) Tween 20®

1% Bovine Serum Albumin (BSA)/Reagent 10 g/L fatty acid free, heat-shocked BSA (Fischer diluent (for all ELISAs except IFNγ) Scientific), dissolved in 1x PBS (Gibco, pH 7.4)

80 g/L NaCl, 11.6 g/L Na2HPO4, 2 g/L KH2PO4 and 10x PBS 2 g/L KCl dissolved in millipore water, pH 7.4

0.1% BSA (v/w), 0.05% Tween 20® (v/w) in Tris IFNγ reagent diluent buffered saline (20 mM Trizma base and 150 mM NaCl), pH 7.2

10.19 g/L citric acid and 14.6 g/L Na2HPO4, Phosphate Citrate buffer dissolved in millipore water, pH 5.0

1.68 g/L NaHCO3 and 0.712 g/L Na2CO3, dissolved Sodium Carbonate buffer in millipore water, pH 9.5

Stop solution 1M H2SO4

56

Table 2.7. Antibodies and top standard concentrations used to measure cytokine concentrations by ELISA

Antibody Top standard concentration, pg/ml Supplier IFNγ 2000 R&D Systems IL-6 2000 Biolegend IL-10 2000 Biolegend IL-17A 1000 Biolegend TNFα 2000 R&D Systems

2.3.4 Flow Cytometry Materials

FACS buffer Sterile PBS (Gibco) with 1% (v/w) filter-sterilised, heat-inactivated (56 °C for 30 minutes) FCS (Biosera).

Live/Dead® Fixable Dead Stain Live/Dead® Fixable Dead Stain Kit (Invitrogen) was used to exclude dead cells from every fluorescent-activated cell sorting (FACS) experiment.

Table 2.8. Fluorescent antibodies and LIVE/DEAD stain used to assess macrophage activation by flow cytometry

Concentration, Antibody/Stain Fluorochrome Clone Supplier μl/ 50 μl mastermix

CD11b APC-eFluor780 0.1 M1/70 Thermo Fischer

F4/80 PE-Cy7 0.1 BM8 BD Biosciences

CD11c* PerCP-Cy5.5 0.3 HL3 BD Biosciences

CD86 PE 0.1 GL1 eBioscience

MHC II eFlour 450 0.3 M5/114.15.2 eBioscience

Zombie aquaTM Brilliant Violet 510 0.25 / 250 μl PBS - Invitrogen 0.5 / 100 μl FACS CD16/CD32 N/A 2.4G2 BD Bioscience buffer

57

Table 2.9. Fluorescent antibodies and LIVE/DEAD stain used to assess macrophage polarisation by flow cytometry

Amount, μl /50 μl Antibody/Stain Fluorochrome Clone Supplier of Mastermix

CD80 FITC 0.15 16-10A1 BD Biosciences

CD206 PE 0.4 C068C2 Biolegend

F4/80 PerCP-Cy5.5 0.25 BM8 eBioscience

CD11b APC-eFluor780 0.1 M1/70 eBioscience

MHCII eFlour 450 0.2 M5/114.15.2 eBioscience

Zombie aquaTM Brilliant Violet 510 0.25 / 250 μl PBS - Invitrogen 0.5 / 100 μl FACS CD16/CD32 N/A 2.4G2 BD Bioscience buffer

2.3.5 Seahorse System Assay

All real-time measurements of oxygen consumption rate (OCR) and extra cellular acidification rate (ECAR) were measured by using Seahorse system: Seahorse XFe96 Analyser (Agilent).

Materials required for the assay are listed in Table 2.10.

Table 2.10 Materials used to carry out seahorse assay

Material Supplier Seahorse 96-well plate Agilent Seahorse cartridge for 96-well plate Agilent Calibration Fluid, pH 7.4 Agilent Seahorse XF DMEM Medium, pH 7.4 Agilent D-Glucose Sigma-Aldrich L-Glutamine Sigma-Aldrich Pyruvate Sigma-Aldrich Sterile water Baxter

The media used for each assay was Seahorse DMEM, supplemented D-glucose, L-glutamine and pyruvate to reach final concentrations of 10, 2 and 1 mM respectively. The reagents and inhibitors used during the assay are listed in Table 2.11.

58

Table 2.11 Reagents and inhibitors used for carrying out Seahorse Assay experiments

Name Abbreviation Target Effect Supplier Inhibit ATP synthesis via Sigma- Oligomycin OM ATP-synthase ox-phos Aldrich

carbonyl cyanide-p- Mitochondria Shuttles protons from Sigma- trifluoromethoxyphen FCCP l inner the intermembrane Aldrich yl hydrazone membrane space back to the matrix

Inhibit the electron Sigma- Rotenone Rot Complex I transport chain Aldrich

Inhibit the electron Sigma- Antimycin-A AA complex III transport chain Aldrich

Sigma- 2-Deoxy Glucose 2-DG hexokinase Inhibit glycolysis Aldrich

2.3.6 Measurement of Nitric Oxide (NO) Production

Griess Reagent System (G2930, Promega) was used to measure NO secretion, with 100 μM as top standard. The kit came with the nitrate standard and the necessary reagents: sulfanilamide solution and N-1-napthylethylenediamine dihydrochloride (NED) solution. Cut-off limit used was 2.5 μM.

2.3.7 Measurement of Cytotoxicity by Release of Lactate dehydrogenase (LDH)

PierceTM LDH Assay Kit (Thermo-Scientific) was used to measure amount of LDH released from dead cells as per the manufacturer’s specifications, except for the sample volume. 25 μl of sample was used, diluted in 25 μl PBS to account for high amount of foetal calf serum in the media used (8%) which leads to high background if not diluted. Sterile water (Baxter) was used to dissolve the lyophilised substrate mix and absorbance was measured at 492 nm.

59

2.4 Cell Culture Methods

2.4.1 Mice

Female and male C57BL/6 mice were obtained from Harlan Olac (Bicester, United Kingdom) and female TLR4-defective C3H/HeJ mice were bred in the Trinity Biomedical Sciences Institute (TBSI) Bioresources Unit. All mice were euthanized at 8-16 weeks old.

Animals were maintained according to the regulations of the Health Products Regulatory Authority (HPRA).

2.4.2 Protocols for Cell Culture

Cells were cultured at 37 °C, at 95% humidity and 5% carbon dioxide. All incubations (unless specified otherwise) were also in these conditions.

Cell Viability and Counting Samples were diluted in 1:1 trypan blue (0.4%, Sigma Aldrich):PBS solution before being loaded onto a hemocytometer: a KOVA Glasstic cell counter slide with grids (Hycor Biomedical Inc.). Viewed in a light microscope, cell viability was assessed by dye exclusion. A hemocytometer is typically divided into a grid of 9 larger squares (each composed of 9 smaller squares), and the volume that fills one of these larger squares is fixed and thus the concentration of cells could be determined (number of viable cells in a large square × 104 cells/ml). Samples were diluted to give counts of around 100 per large square.

Pelleting Cells by Centrifugation Each time cells had to be spun down/pelleted by centrifugation, the conditions used were 400 G, at room temperature, for 5 minutes.

Lysing Red Blood Cells When cells were isolated from the bone marrow or from the spleen, the red blood cells were removed by lysis. Cells pelleted by centrifugation were re-suspended in 1 ml of 0.88% NH4Cl red blood cell lysis solution. After 2 minutes, media was added. After further centrifugation the supernatant was discarded to remove remaining lysis buffer solution, and the pelleted cells were re-suspended with media.

Detaching Adherent Cells by Trypsinisation Trypsin is a protease that breaks down cell surface proteins non-specifically, causing proteins used to attach cells to the surface to be digested, and the detached cells can then be collected. Both magnesium and calcium ions can inhibit trypsin. Therefore, the medium was first removed

60 and the cells were washed with PBS (w/o Mg2+ and Ca2+), after which trypsin with EDTA (a magnesium and calcium ion chelator) was added and the cells were incubated for 5-8 minutes, before the trypsin was diluted in PBS. Cells were detached by pipetting up and down, then pelleted by centrifugation. Remaining trypsin was removed with the supernatant and the cells were re-suspension in fresh media.

Preparation of Gamma-irradiated Mycobacteium tuberculosis, Strain H37Rv Roughly 1 g of γ-irradiated M. tuberculosis, strain H37Rv (BEI Resources) was dissolved in 2 ml PBS by repeated pipetting up-and-down with a P1000 micropipette (cut off tip). The resulting 500 mg/ml suspension was divided, some kept for long-term storage at -80 °C, 1 ml was diluted further to prepare a 100 mg/ml suspension. This 100 mg/ml suspension was divided into 200 μl aliquots and stored at -80 °C – one aliquot was used per experiment.

For each 200 μl, ~100 mg/ml aliquot, the sample was sonicated for 12 minutes, at room temperature, at 45% amplitude and with 40 sec/15 sec on/off pulser, with a 750 Watt Ultrasonic Processor (VCX750, Sonics Vibra-CellTM) to break up cell aggregates. This made the beige/brown cell suspension turn white. The PBS was then replaced with DMEM by centrifuging the sample (4,000 G, 8.5 minutes, 4 °C), discarding the supernatant and re-suspension in 200 μl DMEM. The suspension was then passed through a 23G needle 8-10 times with a 1 ml syringe, to form a single-cell suspension. This solution was spun at 22.4 G for 3 minutes to pellet any remaining cell aggregates. The supernatant was carefully removed and used as a 100 mg/ml single cell suspension.

A standard curve was generated by serially diluting a sample and measuring the optical density (OD) at 600 nm (2 μl, NanoDropTM 2000c spectrophotometer, ThermoFisher). The standard curve can be seen in the graph below. OD 600 nm of 0.1 roughly translated as 1*107 bacteria per ml. For every subsequent aliquot, the final single cell solution was serially diluted, OD 600 nm was measured and compared to this curve, to standardise the concentration over experimental repeats.

61

2.4.3 Stimulation of Splenocytes with Anti-CD3

Round-bottomed 96-well plates were coated with 50 µl of 0.4 µg/ml anti-CD3e antibody (anti- mouse CD3ε, BD Pharmingen), diluted in sterile PBS and left at 4 °C overnight. A mouse spleen was homogenised by mechanical crushing and a single cell suspension made by filtering through a sterile cell strainer (40 µm, FisherScientific) and the red blood cells were lysed.

Remaining PBS was removed from the anti-CD3-coated 96-well plates and cell suspension was added (200 μl of 1  10 6 cells/well). After incubation for one hour, further stimuli were added (specified in figure legends) and after 72-hours incubation, supernatants were collected.

2.4.4 Production of M-CSF Conditioned Medium from L929 Cells

The murine gene encoding macrophage colony-stimulating factor (M-CSF) was cloned into a mammalian expression vector and transfected into the murine aneuploidy fibrosarcoma cell line L929.

L929 cells were seeded at 0.5 × 106 cells/ml, 40 ml per T175 flask (Grenier BioOne), grown in cDMEM (devoid of penicillin and streptomycin). For each passage, after incubation for 7 days, supernatant (containing M-CSF) was collected. Cells were detached by trypsinisation and counted, then seeded again to repeat the procedure. The collected supernatant over was then filter sterilised and stored (at -20 °C). Passage number was recorded each time cells were seeded and the passage number never exceeded 30.

Cells were also frozen for long-term storage, at which point 1-2 million cells were suspended in 90% FCS, 10% DMSO (Sigma-Aldrich) and frozen with a “Mr Frosty” at -80 °C. When cells were taken from one of these frozen vials to initiate the M-CSF production, cells were rapidly thawed in a 37 °C water bath and diluted in 9 ml pre-warmed DMEM. The cells were pelleted by centrifugation and re-suspended in 10 ml cDMEM to wash off the DMSO. Cells were then seeded directly in a T75 flask and growth was monitored daily. When 90% confluency was reached, the supernatant was discarded, cells were detached by trypsinisation, counted and seeded as explained above.

BMDMs are typically differentiated using roughly 20% L929-conditioned media. The amount necessary was determined for each batch by differentiating BMDMs at different percentages of the conditioned medium, and markers of macrophage differentiation were measured by FACS. Each new batch was compared with previous ones. The flow cytometry panel used in presented in Table 2.8 and the staining procedure is outlined in section 2.4.9.1.

62

2.4.5 Isolation and Culture of Bone Marrow-Derived Dendritic Cells (BMDCs)

Murine BMDCs were prepared based on previously reported protocols [198]. Briefly, female or male mice from either a C57BL/6 or C3H/HeJ background; bone marrow cells in their femurs and tibiae was flushed out with cRPMI medium using a 27G needle. Cell aggregates were broken up by repeated passage through a 19 G needle and the cell suspension was then filtered (40 μm filter) to remove residual bone fragments, before red blood cells were lysed.

To differentiate the bone marrow cells into DCs, cells were counted and seeded in T175 flasks (Greiner BioOne) at 30-35 ml of 4 × 105 cells/ml in cRPMI medium, supplemented with 20 ng/ml granulocyte-macrophage colony stimulating factor (GM-CSF, Peprotech). After 3 days incubation, 30 ml of fresh medium supplemented with 20 ng/ml GM-CSF was added to the flask. On day 6, the cell culture supernatant, containing non-adherent cells, was carefully removed and was replaced with 30 ml of cRPMI, supplemented with 20 ng/ml GM-CSF. After 2 days (day 8), a further 30 ml of cRPMI, with 20 ng/ml GM-CSF was added to the flask. On day 10, the loosely adherent cells were collected, by gentle repeated pipetting and counted. Cells were plated in cRPMI supplemented with 10 ng/ml GM-CSF in 96-well round-bottomed plates at a concentration of 0.625 × 106 cells/ml (200 μl/well) and stimulated the following day for 24 hours. The specific treatments and conditions for each experiment are specified in each relevant figure legend.

2.4.6 Generation of Bone Marrow-Derived Macrophages (BMDMs)

Bone marrow cells from C57BL/6 mice were flushed from murine bone marrow as described in section 2.4.5, only with cDMEM instead of cRPMI.

2.4.6.1 Short-Term Stimulations for Cytokine Analysis via ELISA 35 ml at 1 × 106 cells/ml, supplemented with 10% L929 conditioned DMEM (hereafter referred to as L929) containing M-CSF was added to a T175 flask. After 3 days incubation, 30 ml of cDMEM, supplemented with 10% L929. 2 or 3 days later (at 2 days if cells had reached > 90% confluency), the media was removed (to displace all non-adherent cells) and cells were removed by trypsinisation. Cells were plated at 1 × 106 cells/ml in cDMEM supplemented with 10% L929, in flat-bottomed 96-well plates (200 μl/well). Cells were stimulated the following day, for 24 hours before supernatants were collected for ELISA analysis. The specific treatments and conditions for each experiment are specified in each relevant figure legend.

63

2.4.6.2 Short-Term Stimulations for Flow Cytometry Analysis Cells were seeded at 1 × 106 cells/ml, in cDMEM supplemented with 10% L929. 125 mm non- treated tissue culture dishes (Corning) were rinsed with enough cDMEM to cover the full surface of each dish, before 20 ml of cell suspension was added. After 3 days incubation, 20 ml of cDMEM, supplemented with 10% L929 was added to the dish. 3 days after (day 6), the media was removed (to displace all non-adherent cells) and cells were removed by trypsinisation. BMDMs were plated at 0.8 × 106 cells/ml in cDMEM supplemented with 5% L929 into 35 mm non-treated tissue culture dishes (VWR) that were rinsed with enough media to cover the surface before use. The cells were stimulated for 24 hours before the media was taken off and cells were washed in 1 ml PBS, before they were incubated with 5 mM EDTA in PBS and placed on ice. After 30 minutes, the cells were removed by gentle pipetting up-and-down with a P1000 micropipette and transferred to FACS tubes. The flow cytometry panel used in presented in Table 2.9 and the staining procedure is outlined in section 2.4.9.2.

2.4.6.3 Short-Term Stimulations for mRNA Analysis via RT-qPCR and Nitric Oxide Production Analysis using Griess Reagent Cells were seeded at 1 × 106 cells/ml, in cDMEM supplemented with 10% L929. 125 mm non- treated tissue culture dishes (Corning) were rinsed with enough cDMEM to cover the full surface of each dish, before 20 ml of cell suspension was added. After 3 days incubation, 20 ml of cDMEM, supplemented with 10% L929 was added to the dish. 3 days after (day 6), the BMDMs were removed by trypsinisation. Counted BMDMs were plated at 1 × 106 cells/ml in cDMEM supplemented with 5% L929, in flat-bottomed 24-well plates (900 μl/well) for RNA isolation, and in 12-well plates (450 μl/well) for nitric oxide (NO) production analysis. Cells were stimulated the following day, and in the case of mRNA analysis, the cells were lysed for RNA extraction at certain time points (specified in relevant figure legends) for RT-qPCR analysis. For NO secretion analysis, cells were stimulated for 24 hours before supernatant was collected and samples were immediately analysed for their NO content by Griess reagent system (Promega).

2.4.6.4 Innate Memory Experiments On Day -8, bone marrow cells were seeded at 1 × 106 cells/ml, in cDMEM supplemented with 10% L929. 125 mm non-treated tissue culture dishes (Corning) were rinsed with enough cDMEM to cover the full surface of each dish, before 20 ml of cell suspension was added. After 3 days incubation, 20 ml of cDMEM, supplemented with 10% L929 conditioned DMEM was added to

64 the dish. 3 days after (Day -2), the media was removed (to displace all non-adherent cells) and cells were detached by trypsinisation.

Protocol 1: Plating Twice On Day -2, 92 mm non-treated tissue dishes (Sarstedt) were rinsed with enough media to cover the entire surface of each dish, before 6.5 ml of 0.8 × 106 cells/ml was added. The BMDMs were seeded in cDMEM or cDMEMG↓ (see 2.3.1 for specification) supplemented with 5% L929. In every case, the same media (cDMEM or cDMEMG↓) was used from Day -2 until the end of the protocol. Cells were stimulated the following day (Day -1) for 24 hours (stimuli specified in relevant figure legends). After 24-hour stimulation (Day 0), the media was taken off and replaced with 7 ml of media supplemented with 5% L929 conditioned DMEM. On Day 3, 7 ml of media, supplemented with 5% L929, was added to each dish. On Day 5, the BMDMs were detached by trypsinisation and plated at 1 × 106 cells/ml in flat-bottomed 96-well plates (200 μl/well) and stimulated the following day (Day 6) for 24-hours with secondary stimuli. The specific treatments and conditions for each experiment are specified in each relevant figure legend.

Protocol 2: Plating Once On Day -2, the 0.8 ml of BMDMs were seeded in 12-well plates, at 0.6 × 106 cells/ml in cDMEM or at 0.8 × 106 cells/ml in cDMEMG↓, both supplemented with 5% L929. Media was added on top to reach 1.3 ml. The following day (Day -1), the cells were incubated with primary stimuli for 24 hours (specified in figures). The media was removed and replaced 1.5 ml of fresh media supplemented with 5% L929. The cells were fed three days later (Day 3): media was taken off and replaced with 1.3 ml fresh DMEM supplemented with 5% L929.

On Day 6, the BMDMs were incubated with secondary stimuli (specified in relevant figures) and the supernatant was collected and the cells were lysed after 24 hours (Day 7).

Protocol 3: Different Batch of L929

The following protocol is distinct from the previous in that a new batch of L929 conditioned media (hereafter referred to as L929) was produced, which was less potent than the previous one, and amounts were adjusted accordingly. Media used was cDMEM.

On Day -8, bone marrow cells were seeded at 1 × 106 cells/ml, in cDMEM supplemented with 25% L929. 125 mm non-treated tissue culture dishes (Corning) were rinsed with enough cDMEM to cover the full surface of each dish, before 20 ml of cell suspension was added. After 3 days incubation, 20 ml of cDMEM, supplemented with 25% L929 was added. On Day -2, the media was removed (to displace all non-adherent cells) and cells were detached by trypsinisation.

65

For these studies all tested conditions were in duplicate in 24-well plates. Cells were seeded at 0.5 × 106 cells/ml with 0.4 ml per well, supplemented with 10% L929. The cells were stimulated the following day (Day -1) for 24 hours. The BMDMs were then washed (Day 0) and the media was replaced with 600 µl cDMEM per well, supplemented with 7% L929. On Day 3, the cells were fed, by replacing the media with 450 µl fresh media, supplemented with 5% L929. On Day 6, the cells were given their secondary stimulation (as specified in figure legends) and samples were collected for analysis: supernatant for ELISA and cells were lysed, duplicates pooled together, for RT-qPCR.

2.4.7 RT-qPCR

2.4.7.1 RNA Isolation Total RNA from cells was isolated using a High Pure RNA Isolation Kit (Roche) by following the manufacturer’s instructions. Following stimulations and incubations at appropriate time points, cells were lysed, added to a filter column, spun to remove solvent and incubated with 50 μl DNase I to digest any DNA present. After three washing steps, RNA was eluted in 50 μl nuclease- free water into nuclease-free microcentrifuge tubes. RNA concentrations were measured by Nanodrop spectrophotometer; absorbance measured at 260 nm gave the approximate concentration and the ratio of 260 nm/280 nm to provide an indication of protein contamination. Samples with a ratio of 2 or higher were used. The RNA could then be stored long-term at -80 °C.

2.4.7.2 Reverse Transcription The isolated RNA (20 ng/µl) was used to produce complementary DNA (cDNA). The RNA concentration was adjusted by the addition of more nuclease-free water to match the lowest concentration of the set of samples, with a minimum amount of 100 ng. 5 μl total of RNA sample was mixed with 5 μl mastermix containing random hexamers to act as non-specific primers, deoxy nucleotides triphosphates (dNTPs) as DNA building blocks, the enzyme Reverse transcriptase (RT) to generate the cDNA and RNase OUT to inhibit ribonuclease activity and thus prevent RNA degradation during the process (specified in section 2.3.2, Table 2.4 ). To confirm there was no DNA contamination, a control sample was chosen at random where no RT was included (replaced with nuclease-free water). Additionally, to make sure none of the components included in the mastermix were contaminated/could affect the qPCR analysis, there was an additional control where no RNA sample was added (replaced with nuclease-free

66 water). The PCR cycle used for the reverse transcription is detailed in Table 2.12. Upon the conclusion of the cycle, 20 μl of nuclease-free water was added to each sample and control; all of which could then be stored long-term at -20 °C.

Table 2.12. cDNA Synthesis Thermocycler Program

Temperature, °C Time, minutes 25 10 42 60 95 3 4 10

2.4.7.3 qPCR Real-time quantitative polymerase chain reaction (qPCR) was performed with previously made cDNA and forward and reverse primers and using KAPA SYBR® Fast qPCR Kit (section 2.3.2). Samples were run on an Applied Biosystems 7500 Fast system or the QuantStudio 3 System (programs specified in Table 2.13). For either system, 3 μl of cDNA sample was mixed with 5 μl KAPA SYBR fast ROX low (Sigma-Aldrich), 0.5 μl of each primer preparation (final concentration of 5 pmol/μl) and 1 μl of nuclease-free water (total volume of 10 μl).

Table 2.13. qPCR amplification Thermocycler Programs for each system employed

System Time Temperature, °C Cycles AB7500 Fast System 2 min Up to 95 - 3 sec 95 40 30 sec 60 40 QuantStudio 3 System 2 min Up to 50 -

2 min Up to 95 -

3 sec 95 40

30 sec 60 40

Dissociation analysis was performed after each complete qPCR to exclude non-specific products. The change in gene expression was normalised to Actb and/or Tbp expression for each individual sample, and additionally compared to the unstimulated control by use of the 2ΔΔCT method [199]. The cut off cycle threshold (CT) value was ≥33. 67

Efficacy of the primers was measured as reported previously [199]. Briefly, standards for the mRNA target of interest were diluted in a 1:2 dilution series (four dilutions). qPCR was carried out for each concentration with the respective primers. The CT value measured for each concentration was then plotted against the logarithm of the dilution factor. The slope of the curve (k) can then be used to calculate the efficiency (E) by the following equation: E = 10-(1/k) – 1. Primers yielding slopes of ≥ -3.6 were accepted, as they demonstrated at least 90% efficacy or higher.

2.4.8 Enzyme-linked Immunosorbent Assay (ELISA)

Every ELISA was carried out in triplicate – biological replicates when 96-well plates were used, technical replicates when 12-well or 24-well plates were used.

For each ELISA performed, solvents, concentrations, reagents, incubation times and protocol were specified by each kit (section 2.3.3). 96-well high-binding ELISA plates (Grenier Bio-one) were coated with capture antibody (50 μl/well) and incubated overnight. The ELISA plates were washed with wash buffer (Table 2.6) three times and incubated with 200 μl/well blocking buffer for 1 hour. The plates were washed three more times in wash buffer and 50 μl sample or serially diluted standard (top standard, Table 2.7) was added to each well and incubated for 2 hours at room temperature whilst shaking gently. Any necessary dilutions at this step were done in reagent diluent. Plates were subsequently washed three times with wash buffer before adding biotinylated detection antibody (50 μl/well) and incubation for the appropriate time in the dark. After washing three times in wash buffer, HRP-conjugated streptavidin/avidin diluted in appropriate reagent diluent was added to the plates (50 μl/well) and incubated in the dark for the appropriate time. Plates were then washed three times in wash buffer and left submerged in wash buffer for 10 minutes. For IL-10 ELISAs, the plates were washed an additional 6 times before carrying on to the next step. Finally O-phenylenediamine dihydrochloride (OPD) substrate, dissolved in phosphocitrate buffer with hydrogen peroxide (section 2.3.3) was added to the plates (50 μl/well) and allowed to develop for the required time in the dark. The absorbance at 492 nm was measured spectrophotometrically using a Versa Max microtitre plate reader. The standards of known concentrations generated standard curves which were used to determine the cytokine concentration of each sample. In cases where the absorbance was higher than that for the top standard, the ELISA was repeated and the sample was appropriately diluted to be within the range of the standard. The detection limit applied was < 30 pg/ml.

68

2.4.9 Flow Cytometry

All data was analysed using FlowjoTM software (Treestar, Oregon).

2.4.9.1 Testing L929 Batches Macrophages were differentiated as outlined in section 2.4.6.2 with cDMEM supplemented with 10-30% L929 conditioned media from Day 0 until Day 6, at which point the media was taken off, the dishes were washed with 20 ml PBS, then 10 ml PBS was added and the dishes were placed on ice for 5 minutes. 15 ml of ice-cold PBS was added and the cells were detached by repeated pipetting. The cells were pelleted by centrifugation, counted and divided into FACS tubes (Falcon) to yield 2 × 106 cells/sample.

Cells were pelleted and then were re-suspended in 250 μl of 1:100 diluted Live/dead zombie aquaTM (BV510, Table 2.8) in PBS. After 10 minutes of incubation at room temperature in the dark, the tubes were washed with 1 ml PBS and pelleted. Following decanting of the supernatant, cells were re-suspended in 1:200 diluted anti-mouse CD16/CD32 monoclonal antibody diluted in FACS buffer. After incubation for 10 minutes at 4 °C, 50 μl of mastermix containing fluorchrome-labelled antibodies specific for CD11b, F4/80 (macrophage markers), CD11c (DC marker), MHC II and CD86 was added. After 30 minutes incubation in the dark at 4 °C, the samples were washed with 1 ml FACS buffer once, pelleted and re-suspended in 250 μl FACS buffer. Cells were kept on ice before being analysed using the BD LSRFortessa flow cytometer (when CD11c was included) or BD FACS Canto II flow cytometer (all other times).

2.4.9.2 Macrophage Polarisation Macrophages were detached from 35 mm non-treated tissue culture dishes (VWR) as described in section 2.4.6.2 and transferred into FACS tubes. 1 ml PBS was added to each sample and the cells were pelleted. Cells were re-suspended in 250 μl of 1:100 diluted Live/dead zombie aquaTM (BV510, Table 2.9) in PBS. After 15 minutes of incubation at room temperature in the dark, the tubes were washed with 1 ml PBS and spun as previously. Following decanting of the supernatant, cells were re-suspended in 1:200 diluted anti-mouse CD16/CD32 monoclonal antibody diluted in FACS buffer. After incubation for 10 minutes at on ice, 50 μl of mastermix containing fluorochrome-labelled antibodies specific for CD11b, F4/80 (macrophage markers), MHC II, CD80 and CD206 was added. After 30 minutes incubation in the dark on ice, the samples were washed with 1 ml FACS buffer once, spun down as before and re-suspended in 200 μl FACS buffer. Cells were kept on ice before being analysed using the BD FACS Canto II flow cytometer.

69

2.4.10 Real-time Metabolic Analysis via Seahorse System

BMDMs (6 days after bone marrow isolation) were plated in 96-well seahorse plates at 100,000 cells/well in regular cDMEM the day prior to the assay. Also, the day before, the Seahorse

Cartridge (Agilent) was left overnight in 200 µl/well sterile water (Baxter) and incubated at 37 C֯ without CO2. The following day, the sterile water was removed and replaced with 200 µl/well

calibration fluid, pH 7.4, and incubated at 37 C֯ without CO2 for 90 minutes before the assay began. An hour before the assay, the regular cDMEM was replaced with 180 µl/well Seahorse XF

DMEM, supplemented with 10 mM D-glucose, 1 mM pyruvate and 2 mM L-glutamine. The

BMDMs with this new media were incubated at 37 C֯ without CO2 an hour before the assay. 15 minutes before the assay, the cartridge was removed and reagents/inhibitors were added to their respective ports (see Table 2.14).

Table 2.14 Layout and Concentration of Reagents Used for Seahorse Assays

Port Reagent/Inhibitor Final Concentration Volume Added A Oligomycin 10 µM 20 µl B FCCP 10 µM 22 µl C Rotenone and Antimycin-A 5 µM each 25 µl D 2-DG or L-rhamnose 25 mM or 5 mM 28 µl

2.4.11 Nitrate (NO) Production Analysis

NO was measured using the Griess Reagent System (G2930, Promega) as per the manufacturer’s protocol. Briefly, BMDMs were stimulated for 24-hours before supernatant was collected and immediately analysed. All reagents involved were left to reach room temperature for 30 minutes before analysis. A serial dilution in cDMEM was generated with the NO provided in the kit, with 100 μM as top standard and added to a 96-well ELISA plate. Samples were added to the 96-well plates in triplicate (50 μl/well) to which 50 µl of the sulfanilamide solution was added to samples and standards. After 10 minutes of incubation at room temperature in the dark, 50 μl NED solution was added to each well. After another 10-minute incubation at room temperature and in the dark, absorbance was measured at 560 nm. Absorbance values for the known concentrations of the diluted NO provided was used to generate a standard curve, from which NO concentration in the samples could be calculated.

70

2.4.12 Lactate Dehydrogenase (LDH) Cytotoxicity Analysis

LDH was measured from samples in 96-well plates using PierceTM LDH Assay Kit (Thermo-

Scientific) as per the manufacturer’s specifications. Briefly, a lyophilised substrate mix provided in the kit, containing nicotinamide adenine dinucleotide (NAD+), diaphorase and a tetrazolium salt (INT) was dissolved in 11.4 m sterile water (Baxter) and combined with 0.6 ml provided assay buffer, making the reagent mix. To make a positive control of maximum LDH activity, to a triplicate of wells of cells 10 μl of 10x lysis buffer was added and incubated for 45 minutes.

Supernatants were then collected and 25 μl of each sample was diluted in 25 μl PBS in a 96-well

ELISA plate. 50 μl of reagent mix was then added on top, and the resulting solution was mixed by pipetting up and down gently. Lactate dehydrogenase (LDH) catalyses the conversion of pyruvate to lactic acid, which reduces NAD+ to nicotinamide adenine dinucleotide (NADH), which diaphorase then uses to reduce INT to a red formazan product which is measured spectrophotometrically. The plate was incubated at room temperature for 30 minutes, protected from light, before absorbance (Abs) of the formazan product was measured at 492 nm. To measure % cytotoxicity, firstly the background levels of LDH activity (absorbance of media treated cells) was subtracted and the following formula was used: %Cytotoxicity =

Sample Abs/Maximum Activity Abs

2.4.13 Statistical Analysis

All statistical analysis was performed by using GraphPad Prism 7 software. Several statistical tests were used depending on the type of data (specified in figure legends). In all cases when a test was used, a minimum of three values in each data set was a requirement (either technical or biological replicates).

For analysing differences between groups where multiple related groups within an experiment were used (such as increasing concentrations) one-way ANOVA was employed, comparing each mean with all the others, with Tukey as a post test to correct for multiple comparisons. When such data sets were collected from several individual experiments, two-way ANOVA was employed to account for differences between individual mice and procedural differences

71 between experiments; as previously, when all means were compared with each other, Tukey was used to account for multiple comparisons.

In cases where only two data sets were compared, t-tests were employed. Every test was two- tailed and Gaussian distribution was assumed, given that the mice used had the same genetic background, were bred in the same facility and were roughly the same age. If the data sets were collected from several individual experiments, paired t-test was used to account for differences as outlined above. When comparing mRNA expression when standardised to media-stimulated cells, ratio paired t-test was employed to account for the exponential aspect of the 2ΔΔCT method, otherwise a student t-test was used.

72

Chapter 3. The Immunomodulatory Properties of L-Rhamnose

3.1 Introduction

To reiterate from Chap. 1, L-rhamnose (L-Rha) is a monosaccharide ubiquitously present in plants, fungi and bacteria, but is not endogenous to humans or mice [163]. Although very little is known regarding this carbohydrate’s relationship to the immune response, a link has been proposed between bacterial expression of L-Rha-rich compounds, including mycobacterial phenolic glycolipids (PGLs) and induction of regulatory immune responses. These are components of the cell wall of mycobacteria such as Mycobacterium tuberculosis [188,200,201].

It has been observed that when murine BMDMs were infected with four strains of M. tuberculosis, expressing various amounts of PGLs, there was a negative correlation between increased levels of PGL expression and secretion of the pro-inflammatory cytokines IL-6, TNFα,

IL-12 and MCP-1 [201]. Furthermore, the addition of purified PGLs further reduced TNFα and

MCP-1 secretion in a dose-dependent manner [201]. PGLs have been likewise demonstrated to suppress TLR2 ligand-induced human macrophage secretion of IL-6 and TNFα [190]. There is moreover accumulating evidence indicating that it is the glycan portion of PGLs – of which L-Rha is a key component – that is the motif responsible for conveying its immunosuppressive properties [200,202].

PGLs from Mycobacterium leprae and M. tuberculosis, as well as synthetic forms of their respective phenolic components inhibited TLR2 ligand-induced NFκB activation in human macrophages, leading to reduced TNFα secretion [200]. Crucially, when the cells were stimulated with the common lipid PGL core, phthiocerol dimycocerosates (PDIMs), this inhibitory activity was not observed. Furthermore, infection of murine BMDMs with recombinant strains of Mycobacterium bovis, that expressed PGLs (either from M. leprae or M. tuberculosis) induced lower iNOS expression and NO production, compared to a strain with no

PGLs [202]. This inhibition was dependent upon the phenolic trisaccharide component, as

73 isolated PGLs from both of these pathogenic mycobacteria were able to reduce iNOS and NO production in IFNγ and LPS-stimulated BMDMs, yet the lipid component PDIM was not [202].

These studies thereby demonstrate the importance of the trisaccharide moiety for conveying immunosuppressive effects upon human and murine macrophages, and in the case of PGLs from M. tuberculosis, this moiety is composed of two L-rhamnosyl pyranosides, and a terminal [200].

Further study of the glycan component from M. tuberculosis have been carried out using PGL structural derivatives: the L-Rha-rich para hydroxybenzoic acid derivatives (pHBADs) have demonstrated inhibition of IL-17A and IFNγ secretion from murine anti-CD3-stimulated splenocytes [167]. Furthermore, a transposon mutant form of M. tuberculosis that lacked pHBAD synthesis was shown to induce higher levels of IL-6, TNFα and IL-12p40 BMDM secretion compared with the wild type bacterium [189]. This is further evidence demonstrating the importance of the sugar moiety of the PGLs in conferring immunomodulatory properties to their host bacterium – linking yet again the presence of L-Rha presence to tolerogenic immune responses.

Given that regulatory immune profiles are enhanced by L-Rha-rich PGLs and their glycan units, it was hypothesised that L-Rha itself possessed immunoregulatory properties and the aim was to elucidate the nature of those properties.

74

3.2 Results

3.2.1 L-Rhamnose Inhibits IFNγ Secretion by Splenocytes

The immunomodulatory properties of L-rhamnose (L-Rha) were first addressed using an anti-

CD3 stimulated splenocyte assay. The spleen is a secondary lymphoid organ containing a mix of both innate and adaptive immune cells. With low-level anti-CD3 stimulation, this causes some non-specific T cell activation and cytokine secretion, which in turn triggers a cascading effect of activation in neighbouring immune cells. The batch of L-Rha used had not been endotoxin- purified at this stage and to eliminate possible LPS contamination, C3H/HeJ mice were used. The secretion of IFNγ, IL-17A and IL-10 was measured and the data show that L-Rha significantly inhibited secretion of IFNγ. There was also a trend towards inhibition seen for IL-17A (Fig. 3.1

A). Secretion of IL-10 was more variable than the other two cytokines and no clear pattern was established.

Figure 3.1 L-Rha selectively reduces IFNγ and IL-17A secretion from anti-CD3 stimulated murine splenocytes. Cytokine secretion from splenocytes (C3H/HeJ mice) stimulated with 0.4 μg/ml anti-CD3 antibody and L-Rha for 72 hours. Results (n = 3) are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with 0 mM L-Rha control.

75

3.2.2 L-Rhamnose has a Minimal Effect on TLR ligand-Stimulated TNFα and IL-

10 Secretion from BMDCs and BMDMs

Since splenocytes are a mix of numerous types of immune cells, the next step was to investigate dendritic cell (DC)-specific responses. As can be seen in Fig. 3.2 A, L-Rha (not endotoxin purified) induced secretion of the cytokine IL-6 by bone marrow-derived DCs (BMDCs) from C57BL/6 mice, compared with those given media alone. This effect was not observed when BMDCs from

C3H/HeJ mice (LPS non-responsive) were stimulated. Following the splenocyte assays, therefore, L-Rha preparations were sterilised by filtration (0.22 µm), after which L-Rha neither induced IL-6 secretion (Fig. 3.2 B) or affected PAM3CSK4 induced IL-6 production (Fig. 3.2 C).

Figure 3.2 L-Rha does not affect BMDC IL-6 secretion once sterilised. IL-6 secretion A) BMDCs from C57BL/6 and C3H/HeJ (LPS non-responsive) mice stimulated with non-sterilised L- Rha, media or positive controls: LPS (1 ng/ml) and CpG (10 µg/ml). B) BMDCs from C57BL/6 mice stimulated with filter sterilised L-Rha, media or CpG (10 µg/ml). C) BMDCs from C57BL/6 mice stimulated with PAM3CSK4 (200 ng/ml), filter sterilised L-Rha or media. Results (n = 1) are shown as mean ± SD and analysed by one-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, ***p < 0.001, ****p < 0.0001, compared with media control group.

76

With this sterilised batch of L-Rha, all subsequent in vitro studies were carried out with BMDCs and BMDMs generated from C57BL/6 mice. BMDCs were stimulated with various TLR-agonists in combination with L-Rha, and the cytokines TNFα and IL-10 were measured after 24 hours by

ELISA. L-Rha did not induce TNFα or IL-10 secretion but did reduce LPS-, PAM3CSK4- and CpG- induced TNFα production in BMDCs (Fig. 3.3), as well as LPS- and CpG-induced TNFα in BMDMs

(Fig. 3.4). By contrast, IL-10 secretion was unchanged by the addition of L-Rha to TLR- stimulated BMDCs. In BMDMs, L-Rha reduced LPS-mediated IL-10 secretion at the highest concentration.

Figure 3.3 At high concentrations, L-Rha reduces TNFα but not IL-10 secretion by TLR- stimulated BMDCs. Cytokine secretion from BMDCs stimulated with L-Rha in combination with media, LPS (100 ng/ml), PAM3CSK4 (1 μg/ml) or CpG (200 ng/ml) for 24 hours. Representative results (n = 1) are shown as mean ± SD and analysed by one-way ANOVA, with Tukey’s multiple comparisons post test. **p < 0.01, ***p < 0.001, compared with 0 mM L-Rha control.

77

Figure 3.4 L-Rha reduces TNFα secretion from LPS and CpG-stimulated BMDMs. Cytokine secretion from BMDMs stimulated with L-Rha in combination with media, LPS (10 ng/ml) or CpG (200 ng/ml) for 24 hours. Representative results (n = 1) are shown as mean ± SD and analysed by one-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, ***p < 0.001, compared with 0 mM L-Rha control.

3.2.3 Impact of L-Rhamnose on Macrophage Polarisation

3.2.3.1 Investigating Surface Markers Indicative of Macrophage Activation States

To investigate the effect of L-Rha on macrophages further, the impact on macrophage polarisation was next addressed. For this experiment BMDMs from three male C57BL/6 mice were stimulated with L-Rha at 1 mM and 5 mM, recombinant murine cytokines IL-4 and IL-13

(M(IL-4, IL-13)) to induce M2 macrophage polarisation, or IFNγ and LPS (M(IFNγ, LPS)) to induce

M1 polarisation. The gating strategy used to identify CD11b+ F4/80+ macrophages can be seen in Fig. 3.5 A. Cells were identified based on forward scatter and side scatter. Live/dead staining was used to identify viable cells and of these, the final gate identified CD11b+ F4/80+ macrophages. The effect of 24-hour incubation with each stimulus on fluorescence intensity of the surface markers CD80, MHC II and CD206 can be seen in Fig. 3.5 B. These results show that

M(IFNγ, LPS) have higher CD80 and MHC II expression, and lower CD206 expression than media

78 stimulated BMDMs. M(IL-4, IL-13) show reduced levels of CD80 and increased levels of CD206 and MHC II (Fig. 3.5 B). For L-Rha, incubation of cells with both concentrations induced a heightened expression of MHC II, whilst the higher concentration also induced enhanced levels of CD80. Exposure of cells to the lower concentration of rhamnose also led to elevated CD206 expression.

79

Figure 3.5 Stimulation of murine BMDMs with recombinant cytokines alters surface expression of macrophage polarisation markers CD80, MHC II and CD206. Flow cytometry analysis of BMDMs stimulated with media, L-Rha as specified, recombinant murine cytokines IL-4 (40 ng/ml) and IL-13 (20 ng/ml), or recombinant murine IFNγ and LPS (50 ng/ml each) for 24 hours. A) Gating strategy used to identify macrophages: cells, live cells and F4/80+ CD11b+ cells. B) Expression of CD80, MHC II and CD206 by F4/80+ CD11b+ cells. On the left side: representative histograms of cell counts against fluorescence intensity (n = 1). Right side: the average mean fluorescence intensity (MFI) shown as mean ± SD (n = 3) and analysed with paired t-test. *p < 0.05, **p < 0.01, ***p < 0.001, compared with media control.

80

3.2.3.2 Investigating Macrophage Activation by mRNA Expression

As these flow cytometry results showed that L-Rha enhanced markers of M(IFNγ, LPS) and M(IL-

4, IL-13) associated phenotypes, further investigation into macrophage activation was performed by reverse transcription quantitative PCR (RT-qPCR). Inducible nitric oxide synthase

(iNOS; Nos2), MHC II (H2-Ab1) and IL-12p40 (il12p40) were used as markers for M1 polarisation, whereas arginase 1 (Arg 1; Arg1), chitinase-like 3 (Ym 1; Chil3), found in inflammatory zone 1 or resistin-like alpha (Fizz 1; Retnla), vascular endothelial growth factor alpha (VEGFa; Vegfa) and

IL-10 (Il10) were used as markers for M2 polarisation (primers found in section 2.3.2). The first step was to establish the optimal time point for expression of each marker of interest. For this purpose, murine BMDMs were stimulated with recombinant murine cytokines: IL-4 and IL-13

(M(IL-4, IL-13)) or IL-10 (M(IL-10)) to promote M2-like profiles, recombinant murine IFNγ in combination with LPS (M(IFNγ, LPS)) to promote a pro-inflammatory M1-like profile, medium only as a negative control or L-Rha. RNA was isolated at various time points and reverse transcription quantitative PCR (RT-qPCR) was used to analyse the difference in expression for each mRNA. Fold change of each mRNA of interest was determined by standardising to β-actin mRNA (Actb) expression (Fig 3.6).

81

Figure 3.6 Optimum time points for measuring mRNA of interest after stimulation of BMDMs. RT-qPCR on BMDMs stimulated with media, L-Rha (0.5 mM) or recombinant murine cytokines IL-4 and IL-13 (40 ng/ml and 20 ng/ml respectively), IL-10 (40 ng/ml) or IFNγ (50 ng/ml) in combination with LPS (50 ng/ml), with Actb as internal reference, standardised to media control. Results (n = 3) are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with media control.

As can be seen in Fig. 3.6, only M(IL-4, IL-13) showed an increase in mRNA expression for Arg1,

Chil3 and Retnla. Whereas the expression of both Chil3 and Retnla mRNA only increased significantly at the 24-hour time point, the expression of Arg1 mRNA in M(IL-4, IL-13) reached significantly higher levels compared with media-stimulated BMDMs at the 6-hour time point, peaked after 12 hours, yet with a high degree of variation, then reduced by 24-hours but remained significantly elevated. Regarding Il10 expression, the peak was found to be at 6 hours, and only M(IFNγ, LPS) and M(IL-10) exhibited an increase at this time point, M(IFNγ, LPS)

82 inducing higher expression than M(IL-10). Furthermore, M(IFNγ, LPS) also exhibited significantly higher mRNA expression of Vegfa at both the 12-hour time point and the 24-hour time point, with more variability at the 12-hour time point.

Regarding M1-associated markers, only M(IFNγ, LPS) displayed an enhanced expression of Nos2,

H2-Ab1 and Il12p40. Whereas Il12p40 expression peaked at 3 hours and then reduced back to unstimulated cell-levels, H2-Ab1 showed the opposite trend, only reaching significant levels at the 24-hour time point. Nos2 mRNA expression on the other hand was significantly higher for

M(IFNγ, LPS) than any other stimulus at all time points, peaking at the 24-hour time point.

L-Rha (0.5 mM) did not directly induce expression of any of these M1 or M2 macrophage markers (Fig. 3.6), but having identified optimum time-points and stimuli, the next step was to see whether the addition of L-Rha to these stimuli could alter macrophage polarisation.

Additionally, BMDMs stimulated separately with IFNγ M(IFNγ) and LPS M(LPS) were investigated in parallel to investigate the effects on each individual stimulant.

As can be seen in Fig. 3.7 A, M(IFNγ, LPS) expressed higher levels of Nos2, Il12p40, Vegfa and

Il10 mRNA compared with media-stimulated controls, whilst M(IFNγ) showed enhanced expression of Nos2 and H2-Ab1. Furthermore, M(IFNγ, LPS) and M(LPS) expressed higher levels of Nos2 compared with M(IFNγ). Regarding the effect of L-Rha, it can be seen in Fig. 3.7 B that in isolation L-Rha at this concentration did not induce expression of any of the mRNAs tested, at these time points. However, L-Rha significantly reduced M(IFNγ) Nos2 expression, whilst the expression of H2-Ab1 was unchanged (Fig. 3.7 A).

83

Figure 3.7 L-Rha reduces expression of Nos2 in IFNγ-stimulated BMDMs. RT-qPCR on BMDMs, with ActB as internal reference, standardised to media incubation. RNA was isolated at intervals: Nos2, H2-Ab1 and Vegfa after 24 hours, Il10 after 6 hours and Il12p40 after 3 hours. A) BMDMs were stimulated with ether recombinant murine IFNγ and LPS (50 ng/ml each), IFNγ (100 ng/ml) or LPS (50 ng/ml) in combination with L-Rha as specified. Results (n = 3) are shown as mean and analysed by ratio-paired t-test. *p < 0.05, compared with 0 mM L-Rha control. B) BMDMs stimulated with 0.5 mM L-Rha or media alone (dotted line =1). Results (n = 3) are shown as mean.

To probe the effect of L-Rha upon Nos2 expression further, a range of concentrations was used to investigate the effect of the sugar on IFNγ and LPS induced Nos2. As can be seen in Fig. 3.8 A, when the concentrations were reduced to 0.5 ng/ml for each activating stimulus (M(IFNγ,

LPS0.5)) then L-Rha caused a similar reduction in Nos2 expression to that seen for M(IFNγ), 84 although with greater variability, as can be seen by the deviation when considering the proportional difference (bottom graphs in Fig. 3.8 A). In both cases 0.5 mM L-Rha reduced Nos2 expression by almost 50%. To see whether this L-Rha-induced reduction in Nos2 expression had a functional effect, nitric oxide (NO) production was measured by the Griess reagent system. As can be seen in Fig. 3.8 B, L-Rha reduced NO secretion from M(IFNγ, LPS0.5) and M(IFNγ), to a significant extent in the case of the latter.

85

Figure 3.8 L-Rha reduces Nos2 expression and nitrate (NO) secretion in IFNγ stimulated BMDMs. BMDMs stimulated for 24 hours with L-Rha as specified, recombinant murine IFNγ alone (100 ng/ml), with LPS (0.5 ng/ml each), or LPS alone (50 ng/ml). A) RT-qPCR with ActB as internal reference. Results (n = 3) are shown in top row by mean, standardised to media-stimulated BMDMs, analysed by ratio-paired t-test; in bottom row by mean ± SD standardised to 0 mM L-Rha with IFNγ or IFNγ and LPS, analysed by student t-test. B) NO secretion (Griess reagent). Results (n = 3) are shown as mean ± SD, analysed by student t-test. *p < 0.05, **p < 0.01, compared with 0 mM L-Rha control.

L-Rha alone did not alter expression of Arg1, Chil3 and Retnla, as can be seen in Fig. 3.9 A. The effect of L-Rha upon M(IL-4, IL-13), expression of Arg1 was measured at both 12 hours (not shown) and 24 hours (Fig. 3.9 B). Whilst there was no change in Arg1 at the 12-hour time point

86

(not shown), after 24-hour stimulation, L-Rha induced significantly higher levels of Arg1 in M(IL-

4, IL-13), whereas Retnla expression was unchanged and Chil3 expression was reduced – although not to a significant extent. What can also be seen in Fig. 3.9 B is that L-Rha induced more than 50% greater Arg1 expression and contrastingly Chil3 expression was roughly halved.

Figure 3.9 L-Rha selectively enhances BMDM Arg1 expression RT-qPCR on BMDMs after 24 hour stimulation, with ActB as internal reference. A) BMDMs stimulated with 0.5 mM L-Rha, standardised to media stimulated BMDMs (dotted line =1). Results (n = 3) are shown by mean. B) BMDMs stimulated with L-Rha as specified and recombinant murine cytokines IL-4 and IL-13 (40 ng/ml and 20 ng/ml respectively). Results (n = 3) are shown in top row by mean, standardised to media-stimulated BMDMs, analysed by ratio-paired t-test; the bottom row by mean ± SD, standardised to 0 mM L-Rha with IL-4 and IL-13, analysed by student t-test. **p < 0.01, ***p < 0.001, compared with 0 mM L-Rha control.

3.2.4 Does L-Rhamnose Inhibit Glycolysis?

Inhibition of glycolysis with 2-deoxyglucose (2-DG) has previously been shown to reduce serum levels of TNFα and NO in a murine model of sepsis [153]. As the monosaccharide L-Rha was

87 shown to reduce both TNFα (Fig. 3.4) and NO secretion (Fig. 3.8) a possible mechanism causing these effects could have been interference with glycolysis. A mitochondrial stress test was used, measured in real-time by the Seahorse System. To carry out this test, BMDMs were incubated with inhibitors of oxidative phosphorylation (oligomycin, rotenone and antimycin A) at different time points, as controls to ensure all processes regarding oxidative phosphorylation and glycolysis were functioning (Fig. 3.10). Oligomycin inhibits ATP synthase, whereas rotenone and antimycin A block complex I and complex III respectively of the electron transport chain, all of which thereby inhibit the electron transport chain. This inhibition prevents the reduction of oxygen to water, by the electron transport chain, as shown by a reduction of the oxygen consumption rate (OCR).

Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) is a mitochondrial proton transporter, which bypasses the inhibition of the electron transport chain caused by oligomycin, thus allowing the OCR to increase again, before the electron transport chain was further inhibited by rotenone and antimycin A. Following oligomycin’s blockade of ATP synthase, and a rapid consumption of energy induced by FCCP, these inhibitors concurrently forced an increased rate of glycolysis to make more ATP, as measured by extracellular acidification rate (ECAR) - an indirect measurement of lactic acid (the final product of anaerobic glycolysis) secretion. The final step of the test was the addition of either the glycolysis inhibitor 2-DG or L-Rha; as one can see in Fig. 3.10 L-Rha did not reduce the ECAR, whereas addition of 2-DG did to a significant extent. Up until either sugar was added, the BMDMs that were given L-Rha as the final stimulus, showed higher averages of both OCR and ECAR. As all BMDMs had been treated the same way up until the addition of the sugars, this was considered an artefact caused by these BMDMs being placed in the outer wells of the 96-well plate used – the wells with the highest evaporation rate – thereby increasing the concentrations slightly.

88

Figure 3.10 L-Rha does not inhibit glycolysis. Seahorse System real time measurement of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) from BMDMs. Inhibitors were injected at 19-minute intervals: oligomycin (OM), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), rotenone (Rot) with antimycin- A (AA) and either 2-deoxy glucose (2-DG) or L-Rha (annotated as Sugar at time point added). Results (n = 3) are shown as mean ± SD. ECAR (19 minutes before and after Sugar was added) was analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. ****p < 0.0001.

3.2.5 Innate Memory Induced by L-Rhamnose

Previous results have demonstrated that L-Rha enhanced M(IL-4, IL-13) Arg1 expression and reduced Nos2 expression and NO secretion in M(IFNγ) and M(IFNγ, LPS0.5). The next step was to investigate whether L-Rha could imprint a similar skewing of macrophage responses by inducing innate memory. Innate memory is a phenomenon were prior exposure to a compound or an organism alters subsequent elicited immune responses in innate immune cells, such as macrophages and monocytes. This phenomenon is distinct from adaptive (T- and B-cell) immune memory; whereas adaptive memory targets specific antigens and altered responses are therefore restricted to a related subsequent challenge, innate memory is non-specific – the result being that responses elicited by a range of stimuli can be affected.

3.2.5.1 Optimising the Protocol

The basic protocol can be seen in Fig. 3.11: BMDMs were stimulated for 24 hours (Day -1 to Day

0), then given a secondary stimulation on Day 6 for 24 hours, after which samples were collected on Day 7. In the initial protocol used (Protocol 1, specified in 2.4.6.4) BMDMs were plated twice after maturation, once for the primary stimulation, and then again before the

89 secondary stimulation to account for possible differences in cell number. For the primary stimulation phase, macrophages were incubated with L-Rha at 1 mM and 5 mM or IFNγ (25 ng/ml) with LPS (10 ng/ml; M(IFNγ25, LPS10)). The secondary stimuli used, were irradiated whole cells of Mycobacterium tuberculosis strain H37Rv (henceforth referred to as H37Rv), as well as the TLR4 agonist LPS. Preliminary results generated using this protocol can be seen in Fig. 3.12.

Figure 3.11 Timeline for investigating innate memory in BMDMs, using Protocol 1. Isolated bone-marrow cells were differentiated in 10% L929 conditioned medium, in cell culture dishes. On Day -2, BMDMs were plated in smaller cell culture dishes, and incubated with primary stimuli the following morning (Day -1) for 24 hours. On Day 0, the supernatant was replaced to remove primary stimuli. On Day 5, BMDMs were re-plated in 96-well plates, and on Day 6 they were stimulated with secondary stimuli for 24 hours.

90

Figure 3.12 L-Rha has no marked effect on BMDM TNFα and IL-10 secretion following secondary stimulation with 50 µg/ml H37Rv or 10 ng/ml LPS, using Protocol 1. Cytokine secretion from BMDMs incubated for with media, L-Rha as specified or recombinant murine IFNγ with LPS (25 ng/ml and 10 ng/ml respectively) for 24 hours on Day -1, re-plated Day 5, then stimulated with H37Rv or LPS for 24 hours on Day 6. Results (n = 2) are shown as mean.

As can be seen in Fig. 3.12, the preliminary results showed that L-Rha had no effect upon the response to a secondary stimulation a week later, whereas M(IFNγ25, LPS10) displayed lower IL-

10 secretion after secondary stimulation with LPS. However, these concentrations of H37Rv and

LPS used for the secondary stimulation induced lower concentrations of TNFα and IL-10 than observed following TLR stimulation previously (see Fig. 3.4). To reach similar levels of cytokine secretion to those previously observed, a range of concentrations of H37Rv, LPS as well as

PAM3CSK4 was tested (Fig. 3.13). Based on this, the concentration range of all secondary stimuli was narrowed to 100 or 150 µg/ml of H37Rv, 25 ng/ml LPS and 25 or 35 ng/ml

PAM3CSK4.

91

Figure 3.13 Finding the optimal concentrations for secondary stimulations on Day 6. Cytokine secretion from BMDMs incubated with media on Day -1 for 24 hours, re-plated on Day 5, and stimulated on Day 6 for 24 hours with H37Rv, LPS or PAM3CSK4 at the concentrations indicated. Results (n = 1) are shown as mean ± SD.

During this optimisation stage, another factor taken into account with the protocol was the D- glucose content of the media: as can be seen in Fig. 3.14 A, once the cells were plated the first time (Day -2), the BMDMs were either given regular cDMEM with high glucose content (4500 mg/L D-glucose DMEM supplemented with foetal calf serum and 5% L929 conditioned DMEM) or D-glucose depleted cDMEM (0 mg/L D-glucose DMEM, supplemented with foetal calf serum

[D-glucose content unknown] and 5% L929 conditioned cDMEM) – referred to as cDMEMG↓.

The lower D-glucose content routinely gave a higher number of cells by Day 5 (second plating) following incubation with all primary stimuli, except for M(IFNγ25, LPS10), where the macrophages did not survive in these low D-glucose conditions; an example of which can be seen in Fig. 3.14 B.

92

Figure 3.14 Timeline for investigating innate memory in BMDMs, using Protocol 1, and the effect of D-glucose content of the media on cell number. Isolated bone-marrow cells were differentiated in 10% L929 conditioned media in cell culture dishes. On Day -2, BMDMs were plated in smaller cell culture dishes. Here, the media could be changed to D-glucose deprived DMEM (cDMEMG↓), as opposed to regular cDMEM. The BMDMs were incubated with primary stimuli on Day -1 for 24 hours. On Day 5, the BMDMs were re-plated in 96- well plates, and on Day 6 they were stimulated with secondary stimuli for 24 hours, after which supernatant was collected A) Schematic of Protocol 1 and timeline B) Cell numbers counted on Day 5 by hemocytometer. Results (n = 3) are shown as mean ± SD.

As can be seen in Fig. 3.15, in D-glucose depleted conditions with Protocol 1, L-Rha reduced

TLR1/2 induced responses: both pro-inflammatory TNFα and regulatory IL-10 secretion was reduced by prior exposure to the sugar.

93

Figure 3.15 In D-glucose depleted conditions, L-Rha reduces PAM3CSK4 induced TNFα. Cytokine secretion from BMDMs incubated with media or L-Rha as specified on Day -1 for 24 hours, re-plated on Day 5, and stimulated on Day 6 for 24 hours with H37Rv (100 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (25 ng/ml). Media used from Day -2 was D-glucose depleted (cDMEMG↓). Results (n = 2) are shown as mean and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, compared with media control group.

Following these preliminary experiments, a new protocol was tested – where the BMDMs were only plated once, the day before primary stimulation (see Fig. 3.16). This choice was made for two reasons. Firstly the previous section indicated that L-Rha may skew macrophages towards an M2-like profile – a characteristic of which is enhanced proliferation – and the second plating was done to make sure differences in cell number did not influence the results; however, as can be seen in Fig. 3.14 B, L-Rha stimulation did not significantly affect cell numbers, making this additional plating step redundant. Lastly, it was hypothesised that the stress induced by detaching the cells with trypsin and the subsequent re-plating was altering any memory effect induced by the primary stimulus. As such, Protocol 2 was used henceforth (Fig. 3.16).

94

Figure 3.16 Timeline for investigating innate memory in BMDMs using Protocol 2. Isolated bone-marrow cells were differentiated in 10% L929 conditioned media in cell culture dishes. On Day -2, BMDMs were plated in 12-well plates. Here, the media could be changed to D- glucose deprived DMEM (cDMEMG↓), as opposed to regular cDMEM. The BMDMs were incubated with primary stimuli on Day -1 for 24 hours. On Day 6 they were stimulated with secondary stimuli for 24 hours, after which supernatant was collected and cells were lysed.

3.2.5.2 L-Rhamnose-Induced Memory Inhibits Nos2 Expression and Enhances IL-10

Secretion

Using Protocol 2 with D-glucose depleted media, prior L-Rha exposure enhanced IL-10 secretion in a concentration-dependent manner, following stimulation with H37Rv, LPS or PAM3CSK4 (Fig

3.17 A). Moreover, L-Rha further reduced PAM3CSK4-induced TNFα secretion. Moreover, incubation of cells with L-Rha significantly reduced the minor Nos2 induction when cells were re-stimulated with PAM3CSK4 (Fig. 3.17 B). M(IFNγ25, LPS10) are not shown in this graph as too little RNA was isolated to use for RT-qPCR, and neither TNFα nor IL-10 were detected by ELISA.

95

Figure 3.17 L-Rha induced memory enhances IL-10 secretion in D-glucose depleted conditions. BMDMs incubated with media or L-Rha as specified on Day -1 for 24 hours and stimulated on Day 6 for 24 hours with media, H37Rv (100 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (25 ng/ml). Media used from Day -2 was D-glucose depleted DMEM (cDMEMG↓). A) Cytokine secretion. Results (n = 3) are shown as mean and are analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. B) RT-qPCR on BMDMs, with Actb as internal reference, standardised to BMDMs incubated with media throughout. Results (n=4) are shown as mean and analysed by ratio-paired t-test. C) RT-qPCR on same BMDMs as in (B), with Actb as internal reference, standardised to BMDMs incubated with media control (Day -1) for each secondary stimulant (Day 6). Results (n=4) are shown as mean ± SD and analysed by student t-test. **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with media control group.

96

Prior experimentation using Protocol 1 had shown that BMDMs grown in high D-glucose media gave a low cell yield on Day 5 (Fig 3.4 B), compared to BMDMs incubated in cDMEMG↓. Based on microscopy (not shown), this appeared to be due the BMDM in regular cDMEM being over confluent by Day 4, and starting to die and detach before the secondary stimulation. This issue was solved by lowering the seeding density when plating on Day -2. Following this adjustment, it can be seen in Fig. 3.18 that L-Rha induced memory, enhanced IL-10 secretion following secondary stimulation with H37Rv, LPS or PAM3CSK4. By contrast, BMDMs incubated with IFNγ and LPS as primary stimulus secrete TNFα, but IL-10 secretion is abrogated. At the point of secondary stimulation (Day 6) BMDMs activated with IFNγ and LPS generally had fewer numbers of cells per well (seen by microscopy, not shown) than the media and L-Rha primed BMDMs did; therefore this reduction in IL-10 could be partially caused by reduced cell number, as well as change in phenotype.

Figure 3.18 L-Rha-induced innate memory enhances BMDM capacity for IL-10 secretion. Cytokine secretion from BMDMs incubated with media, L-Rha as specified or recombinant murine IFNγ and LPS (25 ng/ml and 10 ng/ml respectively) on Day -1 for 24 hours, and stimulated on Day 6 for 24 hours with media, H37Rv (150 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (35 ng/ml). Results are shown as mean (n = 3) and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with media control group.

97

L-Rha, using this optimised protocol, also reduced H37Rv, LPS and PAM3CSK4 induced Nos2 expression (Fig. 3.19 A & C) by about 50% at the highest concentration compared to BMDMs given media as primary stimulus (Fig. 3.19 C). Moreover, both concentrations of L-Rha induced memory that significantly attenuated H37Rv induced interferon beta (IFNβ, Ifnb1) mRNA expression (Fig. 3.19 B & D). M(IFNγ25, LPS10) on the other hand enhanced Nos2 for all secondary stimuli and Ifnb1 following PAM3CSK4 stimulation. Next, the basal expression level of

Arg1, Il10, Nos2, Tnfa and Ifnb1 one week after stimulation with 5 mM L-Rha or IFNγ & LPS was addressed (Fig. 3.19 E). M(IFNγ25, LPS10) had heightened levels of Nos2 and Tnfa levels compared to BMDMs given media alone, whilst prior stimulation with L-Rha did not alter the expression level of any of the mRNA measured. Given that L-Rha as primary stimulus reduced

Nos2 up-regulation by secondary stimuli, NO secretion was also measured by Griess Reagent, however only M(IFNγ25, LPS10) produced NO levels that were detectable (not shown).

98

Figure 3.19 L-Rha-induced memory reduces H37Rv induced expression of Nos2 and Ifnb1. RT-qPCR with Actb as internal reference, on BMDMs incubated with media, L-Rha as specified or recombinant murine IFNγ and LPS (25 ng/ml and 10 ng/ml respectively) on Day -1 for 24 hours, and stimulated on Day 6 for 12 or 24 hours with media, H37Rv (150 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (35 ng/ml). A) Nos2 fold change after 24 hours, standardised to BMDMs given media throughout. Results (n = 4) are shown as mean and analysed by ratio-paired t-test. B) Ifnb1 fold change after 12 hours, standardised to BMDMs given media throughout. Results (n = 3) are shown as mean and analysed by ratio-paired t-test. C) Nos2 fold change after 24 hours, standardised media control group (Day -1) for each secondary stimulus. Results (n = 4) are shown as mean ± SD and analysed by student t-test. D) Ifnb1 fold change after 12 hours, standardised by the media control group (Day -1) for H37Rv stimulation. Results (n = 3) are shown as mean ± SD and analysed by student-test. E) Fold change of several mRNA after 24 hours, following media as secondary stimulant (Day 6), standardised to BMDMs given media throughout (=1). Results (n = 3) are shown as mean and analysed by student t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with media control group.

99

3.2.5.3 Does L-Rhamnose Affect Histone Methylation?

Innate training is a term used for innate memory, when the phenomenon is induced by epigenetic modification [106,108]. Of the epigenetic changes possible, histone methylation especially has been highlighted in the literature as one of the most common changes that, if blocked, prevents innate training from developing. This is the case for β-glucan, where use of the pan methyl transferase inhibitor 5′-methylthioadenosine (MTA) blocks its innate training effects but an inhibitor of de-methylation, pargyline, had no such effect [106]. As such, these two epigenetic inhibitors were investigated herein: pargyline (Fig. 3. 20) and MTA (Fig. 3.21). As can be seen in Fig. 3.19, pargyline did not alter the secretion of TNFα or IL-10 or the expression of Nos2 following stimulation with media, H37Rv, LPS or PAM3CSK4, regardless of primary stimulant. Incubation of cells with MTA reduced IL-10 secretion following H37Rv stimulation, regardless of primary stimulus (Fig. 3.20 A). Moreover, for BMDMs that were given L-Rha, the prior incubation with MTA also reduced IL-10 secretion induced by the other two secondary stimuli: LPS and PAM3CSK4. Secretion of TNFα (Fig. 3.20 A) and expression of Nos2 (Fig. 3.20 B) were not affected by MTA incubation, following all combinations of primary and secondary stimuli.

100

Figure 3.20 Pargyline does not affect responses induced by secondary stimuli. BMDMs incubated with either media or pargyline one hour before more media, L-Rha (5 mM) or recombinant murine IFNγ with LPS (25 ng/ml and 10 ng/ml respectively) were added (Day -1 for 24 hours). BMDMs were stimulated again on Day 6 for 24 hours with media, H37Rv (150 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (35 ng/ml). A) Cytokine secretion. B) RT-qPCR measured Nos2 fold change, with Actb as internal reference, standardised to BMDMs given media throughout. Results (n = 2) are shown as mean ± SD.

101

Figure 3.21 MTA reduces IL-10 induction by secondary stimuli. BMDMs incubated with either media or MTA one hour before more media, L-Rha (5 mM) or recombinant murine IFNγ with LPS (25 ng/ml and 10 ng/ml respectively) were added (Day -1 for 24 hours). BMDMs were stimulated again on Day 6 for 24 hours with media, H37Rv (150 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (35 ng/ml). A) Cytokine secretion. B) RT-qPCR measured Nos2 fold change, with Actb as internal reference, standardised to BMDMs given media throughout. Results (n = 2) are shown as mean ± SD.

3.2.5.4 Can L-Rhamnose Re-polarise M(IFNγ25, LPS10)?

Having observed that L-Rha-induced memory enhances IL-10 secretion and reduces expression of Nos2, whilst activation with IFNγ and LPS reduces IL-10 secretion and enhances Nos2 expression, the next question was whether L-Rha could interfere with IFNγ and LPS activation, i.e. if L-Rha could re-polarise M1 macrophages (Fig. 3.22). Therefore, BMDMs given the primary stimulation with IFNγ & LPS on Day -1, were given a secondary stimulation with either media or

L-Rha on Day 3 for 24 hours. Cells were washed 24 hours later to remove the stimulus and fresh medium added. On Day 6, as previously, the BMDMs were stimulated with H37Rv, LPS or PAM.

As can be seen in Fig. 3.23 incubation of cells with L-Rha reduced all H37Rv-induced responses measured: TNFα and IL-10 secretion, as well as Nos2 expression. Stimulating M(IFNγ25, LPS10) 102 with L-Rha on Day 3 also reduced LPS-induced TNFα and IL-10 secretion (Fig. 3.23 A), as well as

PAM3CSK4-induced Nos2 expression (Fig. 3.23 B).

Figure 3.22 Schematic for investigating how L-Rha could alter BMDM activation induced by recombinant murine IFNγ with LPS. Isolated bone-marrow cells were differentiated in 10% L929 conditioned media in cell culture dishes. On Day -2, BMDMs were plated in 12-well plates. The BMDMs were incubated with primary stimuli on Day -1 for 24 hours. On Day 3 the BMDMs were incubated with media alone or with L-Rha for 24 hours. On Day 6 they were stimulated with H37Rv, LPS or PAM3CSK4 for 24 hours, after which supernatant was collected and cells were lysed.

Figure 3.23 Incubation of BMDMs with L-Rha reduces M(IFNγ, LPS) responses. BMDMs were incubated with recombinant murine IFNγ with LPS (25 ng/ml and 10 ng/ml respectively) on Day -1 for 24 hours, media or L-Rha (5 mM) on Day 3 for 24 hours. BMDMs were stimulated again on Day 6 for 24 hours with media, H37Rv (100 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (25 ng/ml). A) Cytokine secretion. Results (n = 3) are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. B) RT-qPCR measured Nos2 fold change, with Actb as internal reference, standardised to BMDMs given media throughout. Results (n = 3) are shown as mean and analysed by ratio-paired t-test. *p < 0.05, **p < 0.01, ****p < 0.0001), compared with media control group for each secondary stimulant.

103

3.3 Discussion

The aim of this chapter was to investigate whether the monosaccharide L-Rha has immunomodulatory properties. Prior to this work, there were indications that this motif may have such properties, given that structurally related compounds have been demonstrated to have anti-bactericidal effects. M. tuberculosis is internalised by a number of immune cells, but prominently macrophages [31], thus shielding itself from extracellular humoral responses. A strong Th1 response, crucially driven by the cytokine IFNγ, promotes M1 macrophage activation and subsequent bactericidal effectors [134,184]. Given the importance of the macrophage in immunity to mycobacteria, elucidating the effects of L-Rha on the effector functions of these cells was the primary focus of this chapter.

3.3.1 Macrophage Polarisation

As discussed in Chap. 1, macrophages polarise towards M1 and M2 phenotypes, although the reality is a broad spectrum of differentiation states where characteristics of either phenotype can be concurrently expressed [21,26,27]. Herein, M1 macrophages were induced, as were two types of M2 macrophages, sometimes referred to as M2a and M2c [26,203]. M1 macrophages were differentiated by incubation with IFNγ, with or without LPS, and were identified by amplified surface expression of CD80 and MHC II with downregulation of the M2 marker CD206, enhanced expression of H2-Ab1, Il12p40 (component of both IL-12 and IL-23) and Nos2, and increased production of nitric oxide (NO). M2a macrophages are differentiated with IL-4 and IL-

13 [203], these cells down-regulate pro-inflammatory cytokine secretion, induce healing and

Th2 responses [26,27]. M2c macrophages on the other hand are induced by IL-10 [203] and typically secrete regulatory cytokines [27]. Along with surface expression of CD206 [28], the

104 expression of Arg1, Il10, angiogenesis-promoting Vegfa, Chil3 and Retnla were chosen as identifying M2 markers [25,29,33].

The M2c macrophages however did not display any clear differences from the unpolarised

BMDMs, and this control was subsequently omitted. Moreover, surprisingly only BMDMs stimulated with LPS enhanced expression of Vegfa at the 12- and 24-hour time points, probably due to enhanced Il10 after 6 hours (Fig. 3.6). As the nomenclature regarding these different polarisation phenotypes is not globally established, the various macrophage phenotypes induced herein will be specified by their differentiation stimuli. In the case of LPS and IFNγ stimulation, the concentration (ng/ml) will sometimes also be included where relevant.

L-Rha in isolation did not affect BMDM differentiation. However, when incubated with M(IFNγ), whilst H2-Ab1 expression was unchanged, L-Rha significantly reduced the expression of Nos2 by about 50% (Fig. 3.8). Furthermore, a similar reduction of Nos2 was observed at low concentrations of IFNγ and LPS (Fig. 3.8). It was moreover found that L-Rha reduced NO production in both M(IFNγ, LPS0.5) and M(IFNγ) – to a significant extent for the latter.

The indication that L-Rha skews macrophage polarisation away from an M1 phenotype is further supported by the finding that it enhances Arg1 expression in M(IL-4, IL-13), especially as Arg1 directly inhibits iNOS function. The same was not seen for Retnla or Chil3, however, where L-

Rha had no effect on the expression of the former and reduced the expression of the latter (Fig.

3.9). Although the reduction of Chil3 was not significant, due to variability, a marked reduction was nevertheless observed in BMDM from all mice tested. The selective enhancement of Arg1 therefore suggests that L-Rha does not enhance a typical Th2-promoting response, but instead alters macrophage differentiation more selectively away from bactericidal responses (Fig. 3.24).

105

Figure 3.24 L-Rha promotes an anti-inflammatory macrophage phenotype. L-Rha stimulation reduces IFNγ-, as well as IFNγ and LPS-induced Nos2 expression and NO secretion. Similarly, L-Rha enhances anti-inflammatory Arg1 expression, whilst having no effect upon Retnla and reducing Chil3 expression. Schematic was constructed with the use of Servier Medical Art.

3.3. 2 Is it a Metabolic Mechanism?

Macrophage differentiation routes are tightly linked with distinct signalling and metabolic programmes [74,75], and interfering with these programmes is a mechanism of immunomodulation available to small carbohydrate metabolites and monosaccharides. An increase in D-glucose uptake and glycolysis, combined with decreased oxidative phosphorylation, is characteristic of pro-inflammatory macrophage differentiation [75,148].

Administration of the glycolysis inhibitor 2-DG six hours prior to LPS has been shown to reduce serum levels of M1- associated TNFα and NO [153] – results that are similar to those observed here for L-Rha. Despite there being very little structural similarity between these two monosaccharides, as D-glucose has one more hydroxyl group and is in the D-conformation, it is nevertheless a typical characteristic of carbohydrate-protein interactions that the protein can recognise a range of carbohydrate structures, with varying degrees of affinity. For instance, the mannose receptor’s C-type lectin-like domain 2 recognises terminal D-mannoses, L-fucoses and

D-N-acetyl glucosamines [62]. As such, given the similarities in responses induced by L-Rha and

2-DG, it was important to consider that L-Rha may be interfering with glycolysis. Therefore, real- 106 time analysis of glycolytic activity was measured by the Seahorse System (Fig. 3.10). In case L-

Rha efficacy was poor, to avoid any minor inhibition of glycolysis (ECAR) going undetected by low basal levels of glycolytic activity, the activity was first enhanced using inhibitors of oxidative phosphorylation (corresponding to reduced OCR and increased ECAR prior to L-Rha addition).

However, L-Rha stimulation – at the highest concentration used in this project (5 mM) – gave no indication of affecting glycolysis (Fig. 3.10).

One could argue that the concentration of L-Rha used (5 mM) may have been too low to detect any inhibition, given that the concentration of 2-DG, the positive control, was 25 mM. However, the amount of 2-DG used is in excess compared to the 10 mM D-glucose present in the media, to ensure near complete inhibition of glycolytic activity. The added L-Rha concentration was still comparatively very high – one L-Rha molecule for every two D-glucose – and if L-Rha did affect glycolytic activity one would expect it to have been detectable under these experimental conditions (outlined above). As such, based on this experiment, it appeared to be unlikely that

L-Rha inhibited glycolysis.

However, there was still the possibility that the presence of 10 mM glucose in the cell culture media could be interfering with the immunomodulatory mechanism of L-Rha. As such, this factor was considered moving forward, i.e. lowering the glucose concentration in the cell culture media to investigate whether this could enhance L-Rha-induced responses. Thus, when the innate memory experiments began, BMDMs were incubated in either high [D-glucose] cDMEM or low [D-glucose] cDMEMG↓.

3.3.3 Macrophage Innate Memory Studies

Having observed that L-Rha altered acute M1 macrophage activation, the question of whether

L-Rha could induce an innate memory response was next addressed. Of interest here was to investigate how L-Rha could alter responses towards M. tuberculosis, as the hypothesis was that

107 this monosaccharide suppresses anti-mycobacterial immune responses. As such, relevant TLR- agonists and irradiated whole cells of M. tuberculosis H37Rv (hereafter referred to as H37Rv), were investigated.

A key factor that was considered when designing this protocol was whether to plate the

BMDMs after the primary stimulation, prior to the secondary stimulation. It has been proposed that changes in immune responses following innate memory studies could be due to differences cell number [105] and to ensure, therefore, that any changes were caused by a change in macrophage phenotype, this strategy was initially employed (Protocol 1, Fig. 3.14 A). However, as a result of counting the cells for re-plating (an example of which can be seen in Fig. 3.14 B) it became apparent that there were no differences in cell number between BMDMs given media or L-Rha, at either concentration. At the same time there were consistently fewer M(IFNγ25,

LPS10), which could be due their activation state. As can be seen in Fig. 3.18 E, M(IFNγ25, LPS10) retain a heightened basal level of expression of both Nos2 and Tnfa, suggesting a heightened level of activity which may be affecting their viability over time. However, for the purpose of these studies, the M(IFNγ25, LPS10) are used as a positive control for the M1 typical phenotype which was still apparent, regardless of this difference. Moreover, by re-plating the BMDMs before the secondary stimulation, there was the possibility that the stress caused by cell detachment and re-plating may alter any innate memory induced, thus masking any effect induced by L-Rha. As such, the protocol was changed (Protocol 2, Fig. 3.16) so that the BMDMs were only plated once, before incubation with media, L-Rha or IFNγ with LPS.

To investigate how L-Rha affects immune responses to M. tuberculosis, BMDMs were stimulated with H37Rv, along with agonists of TLR4 and TLR2, as M. tuberculosis is known to induce immune responses via both of these TLRs [204,205]. Regarding the choice of strain, M. tuberculosis H37Rv do not express phenolic glycolipids (PGLs) [201]. This was a deliberate choice, as strains without PGLs are more pro-inflammatory, to observe how the response induced by L-Rha can affect the overall response to the pathogen. Indeed, these studies

108 revealed that L-Rha did skew responses away from a pro-inflammatory profile, supporting the role of PGLs in protecting the pathogen, and also supporting the hypothesis for this work, that L-

Rha is an important motif for this effect to occur.

As discussed in section 3.3.2, the reason for incubating BMDMs in low [D-glucose] cDMEMG↓ was to elucidate whether doing so would enhance the efficacy of L-Rha. The results presented in this chapter indicate that the presence of D-glucose in the culture media does not affect L-

Rha induced responses. In both conditions L-Rha skews the macrophages towards a more tolerogenic response profile: the anti-inflammatory and regulatory cytokine IL-10 is enhanced, whilst the secretion of pro-inflammatory cytokine TNFα is either unaffected (Fig. 3.18) or even reduced following PAM3CSK4 stimulation (Fig. 3.17 A).

Another significant result was that the expression of Nos2 induced by all secondary stimuli was reduced when BMDMs were incubated with 5 mM L-Rha one week beforehand. This effect was most striking when the BMDMs were incubated in regular high [D-glucose] media, where the induction of Nos2 was essentially halved (Fig. 3.19 C). This reduction was also observed for

BMDMs cultured in low D-glucose cDMEMG↓ (Fig. 3. 17 C) but only to a marked extent when stimulated with LPS and PAM3CSK4, but not with H37Rv. This latter observation, was a key determinant for abandoning future experimentation with cDMEMG↓. Overall, these studies investigating whether the D-glucose content influenced L-Rha efficacy, along with the glycolysis activity experiment outlined in the previous section (3.3.2) reveal that these sugars do not affect each other’s functions.

Having discovered that L-Rha skews macrophages towards a more tolerogenic profile, a key question was whether this effect was due to epigenetic changes. As introduced previously, epigenetic changes can drive innate memory, which is then referred to as innate training. For instance, vaccination with BCG has been shown to induce a memory response that was protective against not only M. tuberculosis but also Candida albicans and Staphylococcus

109 aureus: PBMCs isolated from vaccinated volunteers showed enhanced expression of TNFα, IL-1β and IFNγ to all three pathogens, three months after BCG vaccination [108]. This non-specific memory response could not be attributed to BCG induced antigen-specific adaptive immune memory, instead, it was found that innate immune cells, particularly monocytes, were driving this response [108]. Further research from this group demonstrated that Candida albicans as well as its cell wall sugar, β-glucan, could induce a similar memory response: enhancing pro- inflammatory monocyte secretion of IL-6 and TNFα whilst IL-10 secretion was unaffected [106].

The innate memory responses induced by the BCG vaccine were accompanied by histone methylation at the promoter regions for both TNFα and IL-6 [108], leading to the hypothesis that epigenetic changes may be responsible for the enhanced responses observed. Indeed, use of the global methylation inhibitor MTA blocked the memory response induced by not only BCG

[108], but also C. albicans and β-glucan [106], whereas use of the de-methylation inhibitor pargyline did not affect the memory responses [106,108].

As epigenetic modifications, crucially histone methylation, had been identified as a critical mechanism for these innate memory responses, the inhibitors pargyline and MTA were tested to see whether they affected the responses induced by L-Rha. Neither pargyline nor MTA altered L-Rha induced reduction Nos2. Furthermore, whilst MTA attenuated L-Rha-induced enhancement of IL-10 secretion, MTA had a global negative effect upon IL-10 secretion – regardless of primary stimulus or secondary stimulus. Whereas these experiments by no means eliminate the possibility of an epigenetic component driving the memory induced by L-Rha, as there are numerous epigenetic modifications apart from methylation that can alter gene transcription, they do suggest that histone methylation is not a key driver of the response observed.

Given that L-Rha induced memory responses were the opposite of those induced by IFNγ and

LPS tied in with the earlier macrophage polarisation studies, the next step was to consider whether L-Rha could re-polarise M(IFNγ25, LPS10) towards a more M2-type phenotype. An

110 experiment was carried out, outlined in Fig. 3. 22, where M(IFNγ25, LPS10) were incubated with

L-Rha three days prior to the final stimulation. The results revealed that L-Rha incubation dampened M(IFNγ25, LPS10) responses to all subsequent stimuli. In particular, the results following PAM3CSK4 stimulation were especially compelling, where the induction of Nos2 was reduced by a half and TNFα secretion was somewhat reduced, whilst IL-10 secretion was unaffected.

Not only has this chapter outlined how L-Rha interferes with IFNγ induced macrophage responses, but the very first experiment shows that L-Rha also potently inhibits anti-CD3 induced secretion of the cytokine IFNγ (Fig. 3.1 A). Given this antagonism towards the type II interferon, another question was how L-Rha may affect the induction of the type I interferon,

IFNβ. Macrophages are known to produce IFNβ in response to M. tuberculosis infection [206] which in turn up-regulates pro-inflammatory responses [207]. For instance, in a mouse model of

M. tuberculosis infection, inhibition of IFNβ signalling has been shown to significantly reduce lung homogenate levels of pro-inflammatory cytokines such as IL-6, TNFα and IL-1β [207]. Most importantly for this work, IFNβ has also been shown to enhance macrophage production of NO

[208] as well as prevent M2 activation and subsequent up-regulation of Arg1 during M. tuberculosis mouse model infection [209]. Given the capacity of IFNβ to activate pro- inflammatory and antibacterial macrophage responses, the mRNA expression of IFNβ was addressed following L-Rha memory induction and secondary stimulus stimulation. As can be seen in Fig. 3.19 B, only the irradiated Mycobacterium tuberculosis (H37Rv) induced a noticeable enhancement of Ifnb1 in BMDMs. Prior incubation with L-Rha eliminated this enhancement of Ifnb1 completely (Fig. 3.19 B & D). Again, this response is contradictory to the response induced by prior IFNγ and LPS activation, where Ifnb1 expression was either maintained or even enhanced following stimulation with PAM3CSK4. As summarised in Fig 3. 25 below, this inhibition of Ifnb1 is another example of how L-Rha can target the induction and the inducers of bactericidal responses.

111

Figure 3.25 L-Rha skews macrophage responses towards a tolerogenic profile. L-Rha is presented on the surface of M. tuberculosis and has been shown to reduce IFNγ secretion and IFNγ-induced Nos2 expression and NO secretion – responses that kill mycobacteria. L-Rha also enhances Arg1 expression, following IL-4 and IL-3 activation, a response that blocks bactericidal NO secretion. Furthermore, L-Rha stimulation alters responses elicited by killed M. tuberculosis, enhancing anti-inflammatory IL-10 and repressing Nos2 and Ifnb1 expression. Structure was drawn with Chemdraw software. Schematic was constructed with the use of Servier Medical Art.

3.3.4 Conclusions and Future Experiments

Does L-Rhamnose Protect M. tuberculosis from Immune-Mediated Elimination?

This work aimed to elucidate whether L-Rha can exert immunomodulatory activity that may facilitate immune evasion by mycobacteria. As highlighted in the previous section, L-Rha inhibits

IFNγ induced macrophage responses. IFNγ is a vital cytokine for combatting Mycobacterium

112 tuberculosis infection [134,186,210], prominently for its role in polarising macrophages towards a pro-inflammatory M1 phenotype [20,211,212]. Therefore, the specific inhibition of IFNγ and its associated responses by L-Rha is a promising sign that this monosaccharide conveys immunosuppressive properties that would be beneficial for M. tuberculosis pathogenesis. These results are furthermore corroborated by previously published studies investigating the L-Rha- rich phenolic glycolipids (PGLs) and the related para hydroxybenzoic acid derivatives (pHBADs).

For instance, as mentioned previously, strains of M. tuberculosis that do not express L-Rha-rich

PGLs, such as H37Rv, induced a greater BMDM pro-inflammatory cytokine production (TNFα and IL-6) than those that did [201]. It has also been demonstrated that the M. tuberculosis PGL trisaccharide component alone (composed of two L-rhamnose pyranosides and a terminal fucose) reduced iNOS and NO production in IFNγ and LPS-stimulated BMDMs [202]. Given that

L-Rha induces similar immunomodulation, it may responsible for the prior results obtained for the M. tuberculosis PGL trisaccharide.

Other studies have found that PGLs suppress TLR2 induced immune responses [190,200].

Synthetic PGL analogues have been demonstrated to reduce PAM3CSK4-induced TNFα [190] and phenolic trisaccharide components from M. tuberculosis PGLs attenuated PAM3CSK4 stimulated NFκB activation [200]. M. tuberculosis possesses a large repertoire of TLR2 agonists, and its signalling has been implicated in several important aspects of M. tuberculosis pathogenesis, including phagolysosome escape [204], inhibited MHC II antigen processing and presentation [213], as well as Treg recruitment [214]. Therefore, the way L-Rha induced memory affects not only H37Rv, but specifically PAM3CSK4 induced responses via TLR1/2, skewing the macrophage phenotype away from pro-inflammatory responses, is not only highly relevant in this context, but aligns well with what has been shown for PGLs. To conclude, the results outlined in this chapter suggest that L-Rha indeed may be an important factor contributing to the immune evasive effects of M. tuberculosis. To investigate this possibility further, the continued aim of this project was to link these L-Rha responses back to the bacterial

113 pathogen through synthetic means: via particle conjugation to mimic the bacterial surface, and by investigating pHBAD analogues to determine whether L-Rha is a critical epitope for its overall immunosuppressive properties.

Effect of Particle Conjugation

A characteristic of carbohydrate-protein interactions is, as noted previously, low affinity.

Carbohydrates possess vast structural diversity, but not much variation with regard to functional groups, which consist almost exclusively of the polar hydroxyl groups. There are some notable exceptions to this, such as negatively charged sialic acids, associated with numerous immunomodulatory responses induced by specific sialic-acid binding lectins [37] or the positively charged chitosan that can specifically activate the NLRP3 inflammasome and hence exerts more potent responses than its neutral counterpart chitin [130,131]. For those sugars that do not possess such distinguishing functional groups, another approach to increase overall avidity is branching. As can be seen with structures presented in Chap. 1, ManLAM, mannan and β-glucan are all branching sugars, able to present multiple epitopes at once, and hence activate receptors such as CD206 and Dectin-2 [54,64,71,215] or Dectin-1 respectively

[106]. Even if the isolated carbohydrate does not have this structural feature, its multimeric presentation on the surface of cells can produce the same effect. When such carbohydrates are investigated for in vitro or in vivo work, supraphysiological concentrations need to be used to see an effect. For instance, D-mannose was shown to induce Treg differentiation in vitro, at the very high concentration of 25 mM [83]. For this research the maximum concentration used is 5 mM, which is also relatively high. A hypothesis for this work is therefore that conjugation of L-

Rha onto bacteria-sized nanoparticles will mimic such multimeric presentation, and thus enhance its efficacy (Fig. 3.26).

114

Figure 3.26 Multimeric presentation of L-Rha on the surface of bacteria can be imitated by particle conjugation. Low affinity of carbohydrate-protein interactions is solved in nature by presenting multiple epitopes at once (left panel). When researching L-Rha, very high concentrations (middle panel) need to be used to see immunomodulatory effects; mimicking bacterial presentation by particle conjugation (right panel) may enhance L-Rha efficacy. Schematic was drawn with use of Servier Medical Art.

115

Chapter 4. The Synthesis of L-Rhamnose Functionalised

Polystyrene Particles

4.1 Introduction to Carbohydrate functionalised particles

Carbohydrate-based particulates and nanocomposites being utilized in numerous fields: from biomedicine, drug delivery and tissue engineering, to more industrial applications [195,216].

Polysaccharide covered polymeric materials have, for instance, been popularly used for the purpose of wound dressing and drug delivery, as they are highly stable, non-toxic, hydrophilic and biodegradable [217,218]. As well as being considered as drug delivery vehicles, particulates are also very promising for vaccine design, as they offer the possibility to mimic pathogen- relevant size and structure [219], as well as co-delivery of additional adjuvants combined with the antigen of interest [220]. Carbohydrate functionalised particles are also utilised as a means to develop improved adjuvants by targeted DC activation [57,221,222] as well as investigated as a scaffold for immunosuppressive therapeutics [161]. However, to design carbohydrate- functionalised particles, several parameters need to be considered, including particle physicochemical properties, the methodology for functionalisation and the proportion/density of ligands of interest – all of which need to be optimised for the application in question. In this section the implications of these factors will be discussed.

4.1.1 Particle Properties

Elemental Composition

Having identified a ligand of interest, the first aspect of particle conjugation to consider is the particle itself. Elemental properties affects for instance optical and magnetic properties [223], surface charge and immunogenicity of the particle [224,225]. For the purposes of drug design,

116 and other clinical applications, it is attractive to use particles of biodegradable polymers, such as poly(lactic-co-glycolic acid) (PLGA), which hydrolyse into lactic and glycolic acid – both of which are natural human metabolites [120]. Other examples of such polymers are poly(lactic acid)

(PLA) and poly(alkylcyanoacrylates) [226], as well as polyanhydride [225] and polysaccharide particles, such as chitosan [123].

Size

Apart from elemental composition, other particle factors to consider include shape, size and their toxicological effects. The degree to which particulate structures exert either a beneficial or adverse effect, is dependent upon these physicochemical properties – where size has been argued to be the most important [123]. The larger surface area proportionally of nanoparticles compared to microparticles, is proposed to support enhanced bioactivity. Furthermore, the size of particulates affects uptake [122] and distribution of same [227]. Considering the extent of internalisation, one investigation found that uptake of polystyrene particles by murine DMBM-2 macrophages was approximately 50,000-fold greater for 20 nm sized-particles, compared with 1

µm, and that this uptake was mostly via passive transport – whilst larger particles (500 nm and

1µm) were taken up mostly via active transport [119]. This tendency of nanoparticles to disperse in an uncontrollable manner can be a major shortcoming, depending on the desired function. Another example, considering polyethylene glycol (PEG) coated PLGA particles, found that 3 µm sized particles translocated efficiently through damaged, inflamed mucosal tissue, yet not through healthy tissue; whilst 300 nm sized particles could disperse regardless of inflammatory status [228]. For the purpose of developing a possible drug delivery vehicle for inflammatory bowel disease (IBD) patients, 3 µm sized particles thus emerged as a superior option, as they had a more disease-specific and targeted delivery profile.

With regard to directly elicited immune responses, these can be influenced by particle size as well [120,121,229,230]. For instance, mouse studies on polystyrene found that particles of 50 nm and 500 nm both induced pro-inflammatory IL-1α, IL-1β, TNFα – measured three days after

117 intratracheal administration; yet only 50 nm particles additionally stimulated elevated IL-6, granulocyte-colony stimulating factor (G-CSF) and macrophage granulocyte-CSF, as well as a two-fold increase number of lung leukocytes at all time points tested (ranging from one to 31 days) [121]. Furthermore, it was also found that the largest particles (500 nm) were the only size to induce a modest amount of the regulatory cytokine transforming growth factor (TGF)-β [121].

As will be discussed in more detail in section 4.1.3, results from our lab showing the effect of particle size on the immune response was one of the key factors considered for designing the L- rhamnose (L-Rha) coated particles.

Shape

All examples discussed in this section, unless stated otherwise, concern spherical particles, yet there exist a broad variety of particle shapes. Shape is a factor that affects biodistribution and internalisation – including the propensity for phagocytic or endocytic uptake [231]. For instance, comparing IgG adsorbed spherical with worm-like particles of same volume (1 µm or 3 µm spheres), it was found that the worm-like particles were internalized to a lesser extent by rat alveolar macrophages [232]. By contrast, a different study found than rod-shaped gold nanoparticles were taken up to a greater extent in vitro by RAW264.7 macrophages than corresponding spherical particles of similar volume (40 nm spheres), both coated with an immunogenic West Nile virus protein [233]. The difference between these two studies in the shape preference for uptake is probably due in part to difference in size. The latter investigation also observed that the spherical nanoparticles, despite internalised to a lesser degree, induced higher antigen-specific serum IgG levels, after immunization in C3H/HeJ mice. This research additionally revealed an inverse correlation between surface area to volume ratio, against antibody production – sphere being the shape with the lowest surface area to volume proportion [233]. This is surprising, as more surface area would logically translate into more antigen being presented – perhaps the rod shape in this case negatively affected how the epitopes were presented on the surface. The use of viral capsids, with e.g. icosahedral shape,

118 has also been shown to enhance immunogenicity towards conjugated carbohydrate antigens

[219]. Essentially, shape of particulate structures is evidently a key feature that certainly can affect immunological responses; however, as the work in our lab was primarily focused on spherical particles, and commercially these are the most abundantly available, the research herein, and the rest of this section, focuses on spherical particles.

Surface Charge

Surface charge is another key characteristic that must be taken into consideration when carrying out biological studies with particles in general, and when designing functionalised particles. The charge affects specificity of interactions [143,234], association with cells [228], internalization [235] and the tendency to stimulate an immune response [121,123,236]. With regard to immunomodulatory properties, it has been proposed that positively charged particles stimulate pro-inflammatory responses, whilst neutral and negatively charged particles do not

[123]. By contrast, a recent investigation has demonstrated that negatively charged carboxylated polystyrene particles (500 nm) have anti-inflammatory effects in mouse models of west nile virus (WNV) encephalitis, inflammatory bowel disease and experimental autoimmune encephalomyelitis (EAE) [237]. A proposed reason for why cationic particles induce pro- inflammatory responses, as opposed to neutral or anionic particles, is that they prolong cellular interactions (both specific and non-specific).

In a comparison between cationic liposomes with near neutral liposomes – both galactose ligated – uptake by human hepatoma HepG2 cells, bearing the galactose-specific asialoglycoprotein receptor (ASGPR), was measured [143,234]. When an ASGPR inhibitor was used, uptake of the neutral liposomes was 20%, whereas the corresponding intake of the cationic ones was 40% - indicating that a positive charge can increase the extent of cellular interaction and uptake of particulate structures [143]. That a positive charge enhances uptake is further supported by other studies, for instance functionalisation of polyanhydride particles with either a glycosidic acid linker or di-D-mannose augmented particle uptake by murine

119

C57BL/6 alveolar macrophages, compared to bare particles – cases where the surface charge changed from negative to positive [57]. Similarly, another study utilised gelatin functionalisation with the aim of adjusting the surface charge of PLGA particles from negative to positive, as a hypothesised approach to enhance particle internalisation by human monocyte THP-1 cells, which indeed it did [235]. A positive surface charge on particles seemingly enhances association with cells, likely due to electrostatic attraction to the negatively charged cell membrane

[228,238].

4.1.2 Carbohydrate Coating Strategies – Ligand Density is a Major Concern

Regarding the design of carbohydrate functionalized particles, how the ligand(s) of choice is attached is a key feature. Surfaces of particulates tend to, with some selectivity, adsorb biomolecules – typically forming a so-called protein corona in biological fluids [239,240]. This is a potential issue with particles once present in serum, as serum proteins can become adsorbed and affect particle charge, (partially) block presentation and thus downstream interactions of an attached ligand, as well as impact particle uptake by cells [119,218,241]. This kind of non- specific association can however also sometimes be used to coat particles [221], but this method can cause structural changes to the ligand in question, particularly when the ligand is large and complicated, and the particle is nano-sized [242]. Physical adsorption additionally makes the presented ligand less defined. A common approach to ensure functionalisation and to better control surface presentation, is to synthesise covalently bound derivatives using a linker, terminated in a way to form a stable bond to the particle used [222,243]. A number of factors need to be addressed when designing such a linker, including its length, hydrophilicity,

“bulkiness” and how it is terminated [226,243,244].

A number of different types of particles, such as uncapped PLGA nanoparticles [221], ethylcellulose microparticles [245], and polystyrene nano- and microparticles [246] have

120 carboxylic acid groups on their surface. This functional group is useful for covalent attachment of ligands, for instance via the formation of an ester [221], or an amide [57,245]. Seemingly the most common approach is amide formation, for which an amine terminating linker is used. The use of carbodiimide reagents, such as 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)

[241,246] is often employed. This reagent reacts with the carboxyl group to form a reactive O- acylisourea intermediate – this intermediate readily reacts with amines to irreversibly form an amide and an urea side product [245]. However, this intermediate is susceptible to side reactions, such as hydrolysis, thus this amide coupling is often carried out in combination with

N-hydroxysuccimide (NHS). This addition makes the reaction more efficacious, as NHS reacts with the intermediate to form an amine reactive NHS-ester, which is less susceptible to side reactions [245]. The end result is a stable amide covalent bond.

With regard to the linker itself, there are a number of aspects that can affect the particle’s properties, for instance the linker’s length, charge and hydrophobicity. For example, PEG is used both as a linker and for functionalisation alone to improve particle water-solubility, increase the hydrodynamic radius and reduce non-specific binding [120,228,241]. As another example, one study using sulphated carbohydrate mimetics as potential P- and L-selectin ligands, found that the removal of an amide bond within the linker (see Fig. 4.1), lowered the 50% of maximal inhibitory concentration (IC50) values for binding to P-selectin, despite the amide in question not being directly solvent exposed [244]. By contrast, the presence or absence of this amide group had no effect on IC50 values for L-selectin. In this case, it was hypothesised that the more hydrophobic linker conveyed greater affinity to the relevant binding pocket on the P-selectin

[244]. This study also compared two different sizes of gold nanoparticles, namely 6 and 14 nm, and found that proportionally, less ligand could be attached to the 14 nm particles. While the disproportionate increase in ligand amount with particle size is partly due to the decrease in surface area to volume ratio that occurs with increased size, in this case it is also thought to be due to reduced surface curvature. This reduces the distance between the terminal binding ends,

121 where the linkers are attached (Fig. 4.2 AII). This is proposed to cause a loss of binding sites on the surface, rather than a denser shell [244]. However, aiming to have as high density of ligand as possible on the particle can also backfire; if the layer is too dense it can for negatively affect binding affinity [247]. Furthermore, the linker length also affects packing density of the presented epitope at the surface in general (see Fig. 4.2 AI), and for this reason it has been argued that a longer linker is better to ensure “free” binding [244]. This was demonstrated by

Roskamp et al. where an 11-atom long alkyl chain, as opposed to a 6-atom chain, reduced the density of the ligand, and resulted in better binding to the target P-selectin.

Figure 4.1 Schematic representation of the linker variations used by Roskamp et al. This study tested different sulphated carbohydrate mimetics (A), whether the linker had an internal amide bond (B) and the length of its alkyl chain (n = 4, 6 or 9) and how these changes affected P- and E-selectin binding. This was performed on particles of sizes 6 nm and 14 nm. Figure taken from [244].

122

Figure 4.2 Various strategies to reduce dense ligand particle presentation – upper left quadrant, marked in red. A) The proportion between the particle and the linker can be changed favourably, either by increasing the length of the linker (I), or (II) by changing to a smaller particle size [244]. B) By adding a protuberant entity to the linker, the overall density can be decreased [222,248]. C) Adding a spacer aids epitope dispersal at the surface. Two spacer examples are firstly glucose (green) on a short linker (I) [50,243], and secondly, a modified form of the same linker used to present the ligand (II) [249].

Whilst longer linkers may be recommended, polyanhydride particles as large as 8 µm have however successfully been di-D-mannose functionalised by means of a short two-carbon glycolic acid linker [222,248]. Glycolic acid is considerably shorter than the minimum six carbon alkyl spacer utilized by Roskamp et al. While at first glance this can be taken as a counter argument against the need to design longer rather than shorter linkers for the ligation of epitopes to the surface of particles, there is another aspect to consider, namely the “bulkiness” of the linker.

Whilst an alkyl spacer is a simple hydrocarbon chain, which allows very close packaging with neighbouring linkers, glycolic acid contains a protuberant carboxylic acid group, which thus does not allow such dense association (Fig. 4.2 B). In the case of di-D-mannose presentation, it could also be argued that possibly the first D-mannose monosaccharide is essentially part of the linker

– which adds both length, but also “bulk”, as D-mannose is not a linear structure. This “bulk” principal can be compared to membrane lipid fatty acids, where the introduction of a double bond (unsaturated fatty acids) causes a bend in the hydrocarbon chain, causing less compact

123 structure, which increases fluidity and lowers melting temperature compared with fully saturated fatty acid [250]. Adding “bulk” to linkers can thus aid dispersal to an extent that allows the attached ligand to not be too densely presented at the surface of the particle.

Another strategy to avoid a densely packed outer shell, is to design a particle with a spacer: a ligand or linker alone is added to disperse the epitope of interest on the particle surface

[50,243,247]. For two studies, sets of three ligands were attached to gold nanoparticles with varying linker lengths. In both cases, one ligand was used as a spacer: D-glucose, at the end a short thiol-terminating pentenyl linker. This strategy was employed to construct a proposed S. pneumonia type 14 vaccine, where the antigen was presented on a 31-atom linker, and resulted in particles that successfully stimulated IgG production in BALB/c mice [243]. Similarly, a short

D-glucose spacer was used as part of an anti-cancer vaccine particle construct, where the

Thomsen Friedenreich antigen was presented on a 29-atom linker [195]. Brinas et. al., whilst also targeting this antigen, had a different construct strategy. The spacer in this case was a modified version of the 33-atom linker used to present both the Thomsen Freidenreich bearing peptide and a complement peptide (Fig. 4.2 CII). These particles still significantly enhanced antigen-specific IgG and IgM production in BALB/c mice [249]. As the construct in this latter example successfully enhanced the immune response against the antigen of choice, this would suggest that, whilst the use of “bulky” spacer epitopes with smaller liner length is a useful particle design feature, using the linker alone is also a practical strategy to avoid dense ligand presentation.

124

4.1.3 Hypothesis and Aims

Results in Chap. 3 revealed that the sugar L-Rha had immunosuppressive properties that could potentially shield mycobacteria from harmful immune responses. To investigate this further, it was hypothesised that conjugating L-Rha onto a particle could enhance the efficacy of these immunosuppressive properties. This hypothesis was based on two premises: firstly that mimicking mycobacterial presentation of L-Rha on the surface would enhance overall L-Rha binding avidity, if L-Rha was binding to an undiscovered lectin, and secondly because prior work in our lab had identified that particles of a certain size range could also enhance tolerogenic immune responses.

Results from Hearnden et.al (unpublished), carried out in in vivo mouse models, revealed that size of polystyrene particles modified both innate and adaptive immune responses: smaller particles (50 nm) tended to induce neutrophil infiltration and pro-inflammatory cytokine production to a greater extent than larger (up to 100 µm). Contrastingly, larger particles, for instance 1 μm polystyrene particles, were shown to induce heightened levels of IL-10 from TLR- stimulated murine bone marrow-derived DCs and macrophages (McCluskey, unpublished). This work with polystyrene particles demonstrated that larger particle sizes tend to induce a regulatory immune profile, whilst smaller particles induce a more pro-inflammatory response

(unpublished, McCluskey & Muñoz et al.). As demonstrated in Chap. 3, L-Rha was shown to induce a memory response that enhanced IL-10 and repressed iNOS expression and nitric oxide

(NO) production. As such, 1 μm-sized polystyrene particles emerged as a clear choice for attempting L-Rha conjugation; not only because this size is relatively close to the size of mycobacteria (roughly 0.5 μm width, 2-4 μm length [251]), but also because of the immunoregulatory property of the particles themselves, which could potentially have enhanced this effect further. The aim of this chapter was therefore to conjugate L-Rha onto 1 μm polystyrene particles.

125

4.2 Results

4.2.1 L-Rhamnosyl Derivative Synthesis

Synthetic Strategy

Scheme 4.1 Synthetic strategy for the preparation of an L-rhamnosyl derivative displaying a carboxylic acid-terminating linker 5, to enable particle functionalisation. Yields presented are the highest obtained following optimisation. Scheme was generated via Chemdraw software.

The synthetic strategy used to prepare the L-rhamnosyl derivative 5, with a carboxylic acid- terminating linker, can be seen in Scheme 4.1. This strategy was based on a synthesis published previously [196] but was further optimised.

Initially L-Rha 1 was treated under acetylating conditions, in order to globally protect the hydroxyl groups and thereby prevent unwanted side reactions. Pyridine (C5H5N) was used as a

Lewis base to form a reactive ester intermediate upon the addition of acetic anhydride (Ac2O).

This enables the nucleophilic attack by the hydroxyl groups, finally yielding the acetylated compound 2 at near quantitative yield: 97%.

Before making 3, N-(2-(2-ethanol)ethoxy)phthalimide 9 had to be synthesised from 2-(2- aminoethoxy)ethanol 7 and phthalic anhydride [197]. To form 3, boron trifluoride diethyl etherate (BF3OEt2) was employed as a Lewis acid activator, to promote formation of the electrophilic oxonium ion intermediate at the anomeric position of 2. The subsequent 126 nucleophilic attack of 9 furnished the glycosylated product 3, in the α-conformation (as depicted in Scheme 4.1). This restriction in conformation is due to the formation of a second cationic intermediate, an acetonium ion, rendering it impossible for the phthalimide to attack from the

β-position. This application of the neighbouring group effect is a common strategy for controlling stereoselectivity in glycosylation reactions. Trials of reaction conditions can be seen in Table 4.1. Increasing the number of equivalents of both 2 and BF3OEt2 increased the yield from 27% to 39% and then further to the maximum yield of 74%. Repeating these conditions on a larger scale gave a yield of 67%. It was found that using fresh BF3OEt2 was critical in achieving good yields of the desired product.

Table 4.1 Chronological list of results from the trials of the glycosylation reaction between 2 and 9, using boron trifluoride diethyl etherate as lewis acid activator

Amount of Equivalents of Equivalents of Equivalents Time, Trial Yield 2, g 2 9 of BF3OEt2 hours 1 3 1 1.2 2.5a 24 27% 2 6 2 1 5a 60 39% 3 6.5 2 1 6a 60 74% 4 18.2 2 1 6a 60 67% 5 5.5 2 1 6a 72 48% 6 18.4 2 1 6a 36 11% 7 1.3 1.5 1 6b 60 31% 8 2.15 1.1 1 6b 60 54% a b and : batches of BF3OEt2

The next step in the synthetic pathway was the de-protection of both the acetyl groups and the phthalimide. This was carried out according to two different strategies, either all groups were concomitantly de-protected upon treatment with hydrazine hydrate, or sequentially, by first de- acetylating the hydroxyl groups with sodium methoxide (NaOMe), followed by de-protection of the phthalimide by hydrazinolysis. For the two-step de-protection, the methoxide ion functioned as a nucleophile, displacing the acetyl group as the ester methyl methanoate, yielding the de-protected hydroxyl groups in quantitative yield. Hydrazine hydrate removed the

127 phthalimide protecting group through both amines of the hydrazine sequentially attacking both carbonyls of the phthalimide by nucleophilic attack, releasing a terminal primary amine on the linker and forming a phthalhydrazide by-product. However, this two-step de-protection route gave a maximum yield of 15%, and was thus avoided. For the concomitant de-protection, the acetyl groups were removed by trans-acetylation upon treatment with hydrazine – nucleophilic attack by the amines of the hydrazine to the carbonyl carbons yields the free hydroxyl groups and the by-products hydrazine acetate and hydrazine diacetate. The results of screening reaction conditions for the complete de-protection with hydrazine hydrate can be seen in Table

4.2.

Table 4.2 Chronological list of results from the trials of complete de-protection of 3 by hydrazine hydrate

Trial Amount of 3, g Equivalents of 3 Equivalents of Hydrazine Time, hours Yield 1 1.00 1 12.5a 60 40% 2 1.76 1 29a 20 34% 3 3.56 1 29a 18 26% 4 9.50 1 20a 20 1% 5 2.07 1 12.5a 48 - 6 0.15 1 12b 20 20% 7 0.60 1 12.5b 18 29% 8 1.49 1 16b 60 77% a and b: batches of Hydrazine hydrate

Another complication was that an unidentified by-product was being formed (isolated at trial 5,

Table 4.2), as can be seen in Fig. 4.3. Silica chromatography was initially carried out in methanol, with a small percentage of ammonia added to keep the amine protonated, but the by-product had the same Rf-value as 4 in these conditions, therefore new column conditions were investigated. After testing several combinations, column conditions were found that could separate the final product successfully from the contamination: 60% EtOH, 20% H2O, 15%

EtOAc, 5% acetone and <1% ammonia. These column conditions were employed after trial 5, as

128 was a new bottle of hydrazine hydrate, resulting in a higher yield of 4. The number of equivalents of hydrazine was furthermore increased to 16, and the time given for the reaction was extended, giving a final yield of the completely de-protected compound 4 at 77%.

Figure 4.3 The structure of synthesised compound 4 and its proton NMR (top NMR) and the proton NMR of the unidentified by-product of de-protection reaction with hydrazine (bottom NMR). Boxes highlight the presence of protons at carbon 10 in the 1H-NMR for the desired product 5, which is absent in the by-product. Images were copied from Bruker TopSpin software. Structure was drawn with Chemdraw software

The final synthetic step was the addition of succinic anhydride in methanol, resulting in amide bond formation, via nucleophilic attack of the primary amine to either carbonyl, opening the chain and furnishing a terminal carboxylic acid, compound 5. Reaction conditions screened are shown in Table 4.3. Increasing the equivalents of succinic anhydride increased the efficacy of the reaction, giving the final product 5 at 80%: L-Rha with a linker, at the anomeric position, terminating in a carboxylic acid. 129

Table 4.3 Chronological list of trials of adding succinic anhydride to 4, to yield compound 5

Amount of 4, Equivalents of Succinic Trial Equivalents of 3 Time, hours Yield mg Anhydride 1 300 1 1 18 45% 2 458 1 1 18 35% 3 30 1 1.5 18 60% 4 585 1 2 18 83%

Characterisation

Confirmation of the correctly assigned structures for all compounds discussed above was carried out by 1H-NMR, 13C-NMR, mass spectrometry, optical rotation and IR-analysis (see section 2.2.2). Both NMRs for compound 5 can be seen in Fig. 4.4, including examples of which peaks correspond to which proton or carbon. 1H-NMR peaks at around 5.5, 5 and 3.3 were residual solvent peaks: dichloromethane, water and methanol respectively; similarly, the unlabelled peaks in the 13C-NMR are solvent peaks. Of the remaining peaks, a two L-Rha proton peaks were easy to identify as part of the desired structure. The H-6 protons were assigned as the doublet at δ 1.29 ppm (1H-NMR), as the doublet signified there was only one neighbouring proton, the integration indicated the presence of three protons and moreover the chemical shift was typical of the protons in this chemical environment in L-Rha. It was thus straightforward to assign this proton peak throughout all synthetic steps. Another characteristic proton, was the doublet at δ 4.75 ppm, again a doublet signifies only one neighbouring proton, and an integration corresponding to a single proton. Additionally, this chemical shift was typical of a proton at the anomeric position, and was again, easily identified throughout the synthesis and confirmed by additional 2D-NMR experiments. Having assigned these protons, integration indicated that the structure had a total of 20 protons (bonded to carbon), as did the desired final compound 5. Apart from these characteristic protons, other peaks were identified by their appearance after a synthetic step (protons at carbon 12 and carbon 13 in particular).

130

As to useful functional groups, in the 13C-NMR for this structure there were two peaks at δ

175.6 ppm and δ 173.5 ppm which were both characteristic of carbonyl carbons (C=O), clearly indicating the presence of two distinct carbonyl groups – the amide and the carboxylic acid.

Furthermore, also in the 13C-NMR, the chemical shifts at δ 100.3 ppm and δ 16.7 ppm are also typical values for carbons 1 and 6 respectively, and aided structural assignment identification.

Apart from these characteristic peaks, the 13C-NMR also showed 14 carbons present – the same number as compound 5.

Figure 4.4 Annotated proton (1H) and carbon (13C) NMRs, as well as the structure, for the synthesised compound 5: L-Rha with carboxylic acid-terminating linker. For the 1H NMR, red numbers beneath the axis is the area under each peak, standardised to the number of protons present; solvent peaks are not included. For the 13C, the (hyphenated) numbers specify the chemical shift of each peak; solvent peaks are not included. For both NMRs: drawn boxes, with coloured numbers (corresponding to the structure), display which proton or carbon gave that peak. Structure was drawn with Chemdraw software. Images were copied from Bruker TopSpin Software.

131

Further confirmation of the structural assignment was obtained by heteronuclear multiple bond correlation (HMBC) and heteronuclear single quantum correlation (HSQC) experiments. Both correlation experiments show 13C-NMR on the y-axis and 1H-NMR on the x-axis. A HMBC will reveal protons that are usually either two or three carbons away from the carbon in question, and a HSQC will directly show which protons correspond to which carbon. By analysing both experiments, with the additional information provided by each NMR, each proton and carbon was assigned to the structure. This, along with the other characterisation techniques previously mentioned, confirmed the synthesis of compound 5. Part of the HSQC for the synthesised L-Rha with carboxylic acid-terminating linker, 5, can be seen in Fig. 4.5.

Figure 4.5 Heteronuclear Single Quantum Correlation (HSQC) experiment on the synthesised L-Rhamnosyl derivative with carboxylic acid-terminating linker (5); 13C-NMR is on the y-axis, 1H-NMR is on the x-axis.

Darker spots show single protons (CH), lighter spots show two protons (CH2). Image was copied from Bruker TopSpin Software.

132

4.2.2 Linker Synthesis

4.2.2.1 Synthetic Strategy

The synthetic strategy for making the chosen linker can be seen in Scheme 2. Succinic anhydride was added to a solution of 2-(2-aminoethoxy)ethanol (7) in methanol, where the primary amine acted as a nucleophile to form an amide bond and the final product 8, in 49% yield.

Scheme 4.2 Synthetic strategy to make the 11-atom linker 8. Scheme was generated with Chemdraw software.

4.2.2.2 Characterisation

Full characterisation of the linker was carried out with 1H-NMR, 13C-NMR, mass spectrometry and IR analysis (see section 2.2.2). The proton NMR and the HSQC for 8 can be seen in Fig. 4.6.

Solvent peaks around δ 5.00 ppm and δ 3.30 ppm are water and methanol respectively.

Integration shows that 12 protons are present in the structure, which matches with 8.

Moreover, when comparing with the 2-(2-aminoethoxy)ethanol starting material, after the reaction two new peaks were now present (labelled 6 and 7 in Fig. 4.6). HMBC and HSQC experiments enabled full characterisation of the synthesised compound. The HSQC showed the presence of six CH2-groups, which correlated exactly with 8. Mass spectrometry and IR further confirmed that the linker had been synthesised.

133

Figure 4.6 Proton (1H) NMR, Heteronuclear Single Quantum Correlation (HSQC) experiment and drawn structure of synthesised 11-atom linker with hydroxyl group at one end and a terminal carboxylic acid. For the proton (1H) NMR: red numbers beneath the axis is area under each peak, standardised to the number of protons (solvent peaks not included); drawn in box with coloured numbers, corresponding to the structure, displays which protons gave those peaks. For the HSQC, x-axis is 1H- NMR and the y-axis is the carbon (13C) NMR. Dark spots signify single protons (CH, in this case solvent), lighter spots signify two protons (CH2). Structure was drawn with Chemdraw software. Images were copied from Bruker TopSpin software.

134

4.2.3 Testing Sterility of Compounds

The 1 μm aminated polystyrene particles (AM-PS, Magsphere) were incubated with bone marrow-derived dendritic cells (BMDCs) and IL-6 secretion measured (Fig. 4.1 A). BMDCs secrete high levels of IL-6 upon TLR4 stimulation with bacterial lipopolysaccharides (LPS); as LPS is a common contaminant of non-sterile materials/stimuli, this is a useful test to see whether the ordered particles were sterile. As can be seen in Fig. 4.7 A the particles did not cause an enhancement of IL-6 secretion compared to media stimulated BMDCs. As the synthesised ligands 5 and 8 had been synthesised in a non-sterile environment, these compounds were sterilised by filtration (0.22 µm) and endotoxin removal. After subsequent dry-freezing, compounds were tested for IL-6 secretion. As can be seen in Fig. 4.7 B, the sterilised synthesised ligands did not induce BMDC IL-6 secretion, nor did they affect CpG-induced IL-6 secretion (Fig.

4.7 C).

Figure 4.7 BMDC IL-6 secretion induced by synthesised ligands 5 and 8, and the 1 μm AM-PS used for EDC coupling. IL-6 secretion from BMDCs (C57BL/6 mice) stimulated for 24 hours. A) Stimulation with media, CpG (10 μg/ml) or 1 μm AM-PS (100 µg/ml). Results (n = 4) are shown as mean. B) Stimulation with synthesised ligands 5 and 8 (filter sterilised and endotoxin removed) as specified, media or CpG (10μg/ml). Results (n = 1) are shown as mean ± SD. C) Stimulation with CpG (10 μg/ml) combined with media or sterilised synthesised ligands 5 and 8 (filter sterilised and endotoxin removed) as specified. Results (n = 1) are shown as mean ± SD. Results were analysed by one-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, *** p < 0.001, **** p < 0.0001, compared to media control group.

135

4.2.4 Particle Functionalisation by EDC Coupling

As mentioned in the introduction, EDC coupling was identified as an efficient method for coating amine-functionalised 1 μm particles with carboxylic acid-terminating synthetic compounds 5 or 8, by covalent amide bond formation.

4.2.4.1 Theoretical Maximum Number of Binding Sites

Aminated polystyrene latex 1 μm particles (AM-PS) were purchased from Magsphere. Because little information was given regarding either the coverage of the amino groups on the particles, or how the particles had been amine-functionalised, the first step was to determine how many possible amino groups (and thus a theoretical maximum number of conjugation sites) could be present on the surface of the particles.

The first aspect to consider was the structure of the particle itself. As can be seen in Fig. 4.8 A, polystyrene can form polymers in two distinct ways: either as syndiotactic polystyrene or as atactic polystyrene – which form determines the surface density of phenyl groups of the polymer. As there is no available information regarding which polymer structure these particles are composed of, nor which structure would be more structurally stable, equal proportions of the two will be assumed for these calculations.

Regarding the density of the amine-functionalisation, the manufacturer only revealed the amination is of an aliphatic nature (not as depicted in Fig. 4.8 A); however, as the actual amine- functionalisation was not specified, and the purpose was to determine a maximum number of possible amino groups, these calculations were carried out assuming the amino group was in the para position of the phenyl portion, as this is the densest type of coverage possible.

The distance between binding sites (assuming para position specified) is roughly either 1 or 1.5 benzene widths for the syndiotactic and atactic polystyrene forms respectively (as estimated by chemdraw software). The C-C bonds within a benzene ring are 1.40 Å in length. As the sum of all angles within a hexagon totals 720°, the triangle connecting three neighbouring carbon atoms

136 has a maximum angle of 120°, as seen in Fig. 4.8 B. Using the sine formula, that gives a width of benzene of approximately 2.4 Å. This would mean that the distance between possible conjugation sites is about 2.4 Å for the syndiotactic polystyrene and 3.6 Å for the atactic polystyrene, but as the typical C-C bond length in a carbon chain is slightly longer than those found in a benzene, for the purpose of the following calculations, these values were rounded up to 3 Å and 4 Å respectively.

Figure 4.8 Structure of polystyrene (A) and the width of benzene (B). A) Structures of syndiotactic and atactic polystyrene, aminated at the para position of the phenyl epitope. B) Schematic of relevant bond length and angles to calculate the width of benzene.

Calculations: Maximum Number of Conjugation Sites per mg of AM-PS

Density of the polystyrene latex polymer: 1.05 g/ml Radius (r): 500 nm (or 5,000 Å or 0.000005 cm) Avogadro’s number: 6.023 × 1023 atoms or molecules in one mole

Volume of each particle (4/3πr3): 5.24 × 10-13 cm3 Weight per particle: 5.24×10-13 cm3 × 1.05 g/ml = 5.50 × 10-13 g How many particles per 1 mg: 0.001 / (5.50 × 10-13) = 1.82 × 109 particles Moles of particles per mg: 1.82 × 109 / (6.023 × 1023) = 3.02 × 10-15 mol/mg

Surface area of each particle (4πr2): 3.14 × 108 Å2. Surface covered by each reactive site, assuming full rotation of the phenyl groups, and where “a” is syndiotactic and “b” is Atactic: ra: 1.5 Å, rb: 2 Å 137

Surface area (πr2): a) 7.1 Å2, b) 12.6 Å2

Number of available conjugation sites: a) 3.14 × 108 Å2 / 7.1 Å2 ≈ 4.42 × 107 sites b) 3.14 × 108 Å2 / 12.6 Å2 ≈ 2.49 × 107 sites Average: 3.46 × 107 conjugation sites per particle

Maximum moles of binding sites per mg of particles: 3.46 × 107 × 3.02 × 10-15 ≈ 0.1 μmoles/mg

For reactions of 6 mg of particles (estimated 0.6 μmol) to ensure there would be an excess of ligand and reagents present, ligand concentrations of twice this estimate was chosen initially (1.2 μmol) and increased from there, up to 3 μmol.

138

4.2.4.2 Synthetic Strategy

Mechanism 4.1 EDC coupling mechanism, specifically between synthesised carboxylic-acid terminating linkers (5 and 8) and AM-PS. EDC, 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide; NHS, N-hydroxy succinimide. Mechanism was generated via Chemdraw software.

The proposed mechanism for EDC coupling can be seen in Mechanism 4.1 and involves two main steps: firstly an activation step, with the formation of reactive esters using the carbodiimide coupling reagents EDC and NHS. This activation enables the second step: nucleophilic attack by the primary amines located on the surface of the AM-PS, resulting in the formation of a stable amide bond between the ligand and the particle. A critical component of this reaction is pH, as the functional groups ultimately furnishing the amide bond can be either protonated or de-protonated. For the formation of reactive esters, the carboxylic acid is initially

139 de-protonated by the EDC, enabling a nucleophilic attack onto the protonated carbodiimide by the linker, and subsequent re-arrangement of electrons, causes the formation of an unstable O- acylisourea ester intermediate. NHS is often utilised when carrying out EDC coupling; the hydroxyl group of the NHS attacks the O-acylicourea ester at the carbonyl, to form a more semi- stable NHS-ester through displacement of a stable isourea by-product. For the formation of these reactive esters, a buffer is often used at the same pH as the pKa of the carboxylic acid, to ensure both protonated and de-protonated molecules are present. The pKa of a carboxylic acid is typically 4. Therefore, for the initial trials of this reaction, a sodium phosphate buffer at pH 4.9 was used for this activation step.

For the second step of the amide coupling, the amino groups of the particles have to carry out a nucleophilic attack at the carbonyl of either reactive ester present in the solution – displacing either an isourea by-product (EDC alone) or re-generating NHS (via NHS) – to form a stable amide bond between the particle surface and the carboxylic acid-terminating linker. This nucleophilic attack is dependent on the amino groups of the particles having a free electron pair, and thus cannot be protonated. As the pKa of a protonated amine is roughly between 9 and 11, sodium phosphate buffers around pH 9 are employed for this step of the coupling, to encourage de-protonation of the particle amino groups. More extreme pH environments were avoided as too acidic or basic conditions can cause hydrolysis of amide bonds.

4.2.4.3 Coupling Trials

To monitor the various coupling trials, zeta potential was primarily used; as the AM-PS were positively charged due to the presence of protonated amino groups, and functionalising them by forming the amide bond would cause the loss of these protonation sites. Successful functionalisation should therefore have resulted in an overall drop in surface charge.

Additionally, measuring hydrodynamic size by dynamic light scattering (DLS) would reveal if the particle size was changing, and thus eliminated reaction conditions causing aggregation (which could also lead to a decrease in zeta potential). To find optimal zeta potential and DLS

140 conditions, concentration and solvent were both considered. As can be seen in Fig. 4.9, all conditions tested gave similar zeta potentials, however only the two higher concentrations gave a size measurement close to the manufacturer’s specifications (1000 nm) and with less variation. Another parameter considered was conductivity (not shown); whilst using millipore water as solvent resulted in a conductivity range of 0.02-0.1 mScm-1, the 0.2 mM PBS gave a consistent conductivity of 0.4 mScm-1. As conductivity needs to remain consistent to accurately compare samples, the 0.2 mM PBS was chosen as the solvent for future measurements. Finally

100 × 10-6 v/v particle concentration in 0.2 mM PBS was chosen, as these parameters gave satisfactory size, zeta potential and conductivity measurements, as well as the lowest polydispersity index (PDI). However, once the functionalisation trials began, due to the particles being dissolved in buffer systems, the concentration of PBS was lowered to 0.1 mM to compensate for a measured increase in conductivity (not shown).

Figure 4.9 Zeta potential, hydrodynamic size and polydispersity index (PDI) of 1 μm AM-PS. Concentrations used were 50-200 × 10-6 v/v, in either millipore water or PBS diluted in millipore water.

Before measuring the zeta potential and DLS after functionalisation trials, there had to be separation of by-products and remaining reagents from the particles by washing steps. As can be seen in Fig. 4.10 washing involved diluting the particles in a solvent (either water or PBS), pelleting the particles by centrifugation, then removal of supernatant and subsequent re- suspension of the particles. This was carried out twice after every reaction. Sonication was then also carried out to break up aggregates and encourage full re-suspension.

141

Figure 4.10 Washing 1 μm AM-PS. Following Dilution of the reagent mixture, the particles were pelleted by centrifugation at 6,000 G for 15 minutes and the supernatant was removed. Particles were then re-suspended. Servier Medical art Powerpoint image bank was used to construct this diagram.

As synthesised linker 8 required only one synthetic step, and was thus more readily available,

EDC coupling optimisation was carried out using this linker as the ligand. As can be seen in Table

4.4 a number of parameters were considered for the EDC coupling reaction, including length of time, quantity of reagents and equivalents, various buffer systems – or the lack thereof – as well as particle priming steps prior to the reaction. What is not shown is that initially (trials 1-5), the post-reaction washes were carried out in nonpyrogenic and hypotonic sterile water. However, as can be seen in Fig. 4.11 A, when the particles were washed twice with water they all, regardless of reaction conditions, had gained a negative surface charge, as measured by zeta potential. Upon investigation of the washing procedure (Fig. 4.11 B) it was revealed washing the

AM-PS in water once caused the surface charge to reduce by about half, and further reducing with two washes, reaching a zeta potential of below zero. However, when the particles were instead washed twice with 10 mM PBS (Fig. 4.11 C), though there is still a drop in zeta potential, the surface charge remains positive. Therefore, all post-reaction washing steps were henceforth carried out in 10 mM PBS, from trial 6 and onward.

142

Table 4.4 Trials of EDC coupling between synthesised linker 8 and 1 μm AM-PS; including amounts of all reagents involved, particle preparation, time, resulting zeta potential and whether there was a significant change in zeta potential compared to control particles. The rows highlighted in black shows the trials with the optimised conditions.

Equivalents of AM- Solvents* Pre- Pre- ζ- Different Linker Time, PS, (W, water; Wash Eq potential, from Trial , μmol hours Linker EDC NHS mg B, buffers) ** *** mV control**** 0 ------64.7 - 1 1 1.2 20 6 1.2 W, W - - 48 -10.7 No 2 1 1.2 20 6 1.2 BI, BII 9.19 - - 48 -5.1 No 3 1 1.2 20 6 1.2 BI, BII 9.19 - - 48 -5.6 No 4 1 1.2 20 6 1.2 W, W - - 48 1.37 No 5 1 1.2 20 6 1.2 BI, BII 9.19 - - 48 -12.2 Yes 6 1 2 20 6 1.2 BI, BII 9.19 - - 48 17.4 No 7 1 2 20 6 3 BI, BII 9.19 - - 48 27.1 No 8 1 2 20 6 1.2 BI, BII 9.19 - + 48 38.5 No 9 1 2 20 6 3 BI, BII 9.19 - + 48 33.2 No 10 1 2 20 6 1.2 BI, BII 9.19 + - 48 23.8 Yes 11 1 2 20 6 3 BI, BII 9.19 + - 48 25.8 Yes 12 1 2 20 6 3 BI, BII 9.15 + + 48 -0.5 No 13 1 2 20 6 3 BI, BII 9.60 + + 48 2.3 No 14 1 2 20 6 3 BI, BII 9.15 + + 72 -2.11 Yes 15 1 2 20 6 3 BI, BII 9.60 + + 72 12.2 No 16 1 2 20 6 3 W, BII 8.60 + + 48 -11.7 No 17 1 2 20 6 3 W, BII 8.60 + + 2.5 28.1 No 18 1 2 20 6 3 BI, BII 8.60 + + 48 -23.5 Yes 19 1 2 20 6 3 BI, BII 8.60 + + 2.5 31.3 No 20 1 2 20 6 3 W, BII 9.30 + + 48 -6.5 No 21 1 2 20 6 3 W, BII 9.30 + + 2.5 0.1 Yes 22 1 2 20 6 3 BI, BII 9.30 + + 48 19.9 No 23 1 2 20 6 3 BI, BII 9.30 + + 2.5 20.6 No 24 1 2 20 6 3 W, BII 9.30 + + 2.5 -8.57 Yes 25 1 2 20 6 3 W, BII 9.30 + + 2.5 -21.4 Yes * W, sterile water; BI, sodium phosphate buffer used for linker and reagents EDC and NHS, at pH 4.9 and 0.2 mM; BII, sodium phosphate buffer used for particles (pH specified at each trial) and at either 50 mM (trials 1-15) or 10 mM (trials 16-25). **Particles washed once in water, pelleted by centrifugation and re-suspended before added to reaction mixture ***Particles left overnight in BII (specified in Solvents column) overnight before added to reaction ****Specifying whether the zeta potential was significantly different compared to particles in same reaction conditions, except lacking the presence of the ligand, linker 8. Statistical test used was student t- test. Significance p ≤ 0.05.

143

Figure 4.11 Washing particles in water reduces zeta potential of 1 μm AM-PS. Zeta potential measurements of 100 10-6 v/v 1 μm AM-PS measured in 0.1 mM PBS. Particle reaction conditions: A and C) Buffer, solvents only; EDC & NHS, EDC and NHS only; Linker 8, EDC, NHS and synthesised linker 8; B) 1 wash, particles were washed (dilution, pelleting, removal of supernatant) once; 2 washes, particles were washed twice. A, B and C) Untreated, untreated particles. A) Treated particles were washed twice in water. B) Particles were washed in water as specified. C) Treated particles were washed twice in 10 mM PBS before zeta potential measurements. Results are shown as mean ± SD.

As seen in Table 4.4 although trial 5 initially gave a significant drop in zeta potential, this was not reproducible (trials 6 and 7). After trial 7, particle preparation steps were tested, such as washing the particles once in water and leaving the particles in the alkaline buffer overnight before the reaction. Results from trials 10 and 11 can be seen in Fig. 4.12 A; for these trials there was a ligand-dependent reduction in zeta potential, when the particles were washed in water once before the reaction. The combined effect of washing particles in water and leaving them in alkaline buffer overnight can be seen in Fig. 4.12 B where either 1 wash with water or leaving the particles in alkaline buffer overnight gave a small decrease in surface charge, whereas combining the two reduced the surface charge to near -30 mV. With the aim of possibly de-protonating the particles further, several alkaline buffers were tested after the water wash. Furthermore, sonication was also tested, after initial immersion in the buffer, to prevent possible aggregation, and ensure thorough mix and dispersion of the particles. As can be seen in Fig. 4.12 C, immersion in sodium phosphate buffer at pH 9.3 and sonication before being left overnight gave the greatest reduction in zeta potential, intimating that these

144 preparation conditions were the best means to de-protonate the 1 μm AM-PS prior to conjugation.

145

Figure 4.12 Optimising particle preparation (B) and EDC coupling between synthesised ligand 8 and 1 μm AM-PS (A, C-E). Measured zeta potential and hydrodynamic size of 1 μm AM-PS at 100 10-6 v/v in 0.1 mM PBS. Trial numbers correspond to Table 4.4. Particle Reaction conditions: buffer, solvents only (all buffers at 10 mM); EDC & NHS, EDC and NHS only; Linker 8, EDC, NHS and linker 8; pH Equilibrated, particles left in sodium phosphate (pH 9.19) buffer overnight; Prepared, particles washed once in water and left in sodium phosphate buffer (pH 9.30) overnight; Untreated, untreated particles; water wash, particles diluted in water, pelleted and re-suspended. A) Linker 8 as specified was mixed with EDC in sodium phosphate buffer (pH 4.9) for 2 hours before NHS was added in the same buffer. This was left stirring overnight. 6 mg of water-washed particles were added in sodium phosphate buffer (pH 9.19) and was left stirring overnight. B) Particles were either washed in water (number of times specified) and/or left in sodium phosphate buffer (pH 9.19) overnight. C) Particles were washed once in water then specified samples were sonicated. The particles were then left in sodium phosphate buffer (pH specified, 10 mM) overnight before zeta potential was measured. D & E) Linker 8, EDC (HCl neutralised with NaOH) and NHS dissolved in sterile water, stirred for 15- 20 minutes, 6 mg of prepared particles were added. 2 washes with PBS preceded measurement of zeta potential (D), hydrodynamic size and PDI (E). Results are shown as mean ± SD, and analysed by student t-test. *p < 0.05, *** p < 0.001, **** p < 0.0001, compared to EDC & NHS only group. 146

Of the remaining trials that gave a ligand-dependent significant difference in zeta potential, reaction conditions for trials 14 and 18 were eliminated due to severe aggregation. The next trial (21) that showed a significant decrease in surface charge occurred during a screening of several conditions and upon repetition of these conditions (trials 24 and 25) this result was reproducible. As can be seen in Fig. 4.12 D, all three trials give a significant decrease in zeta potential when the ligand is present (linker 8 bars) compared when all reaction conditions are the same, except the ligand is absent (EDC & NHS bars). As trials 24 and 25 were carried out in parallel, these have the same controls. Upon measurement of DLS of these two latter trials (Fig.

4.12 E) it was shown that all particles had most particles in the desired size range (1000 nm) and that all prepared particles (regardless of reaction conditions) had an increased PDI.

Having found reproducible conjugation conditions, these were repeated once more with the synthesised L-Rha derivative 5. As can be seen in Fig. 4.13, AM-PS incubated with 5 or 8 and coupling reagents EDC and NHS had significantly reduced zeta potential measurements compared with particles given EDC and NHS alone. Moreover, particles incubated with either 5 or 8 without EDC or NHS remained positive, indicating minimal coating without covalent bond formation. All particles also had a hydrodynamic size that was roughly 1000 nm and PDI below

0.7. As seen in Fig. 4.13 C, only particles of this size were detected.

147

Figure 4.13 Successful conjugation of L-Rha derivative onto 1 µm polystyrene particles. Measured zeta potential (A), hydrodynamic size and PDI (B & C) of 1 μm AM-PS at 100 10-6 v/v in 0.1 mM PBS. Reaction conditions: 3 µmoles of linker 8 (Linker-PS) or L-Rha derivative 5 (L-Rha-PS), 2 equivalents EDC (HCl neutralised with NaOH) and 20 equivalents of NHS dissolved in sterile water, stirred for 15-20 minutes, 6 mg of prepared AM-PS were added. 2 washes with PBS and sonication preceded measurements. Untreated, untreated AM-PS; EDC & NHS, full reaction conditions with EDC and NHS alone. A) AM-PS were left untreated (untreated), were subjected to reaction conditions with 8 or 5 as outlined with (+) or without (-) reagents EDC and NHS, or were subjected to full reaction conditions with EDC and NHS alone (no ligand). Results shown are mean ± SD and analysed by student t-test. **p < 0.01, **** p < 0.0001, compared with “no ligand” group. B). Results (n = 3) are shown as mean ± SD. C) Size distribution as presented by Malvern Zetasizer Software.

148

To confirm the conjugation of L-Rha onto the surface of the 1 μm AM-PS, the amount of leftover

5 in the reaction mix was measured by spectrophotometry, by using the phenol-sulphuric acid method [252]. As can be seen in Table 4.5 when 5 was added alone, the amount was measured to be 3.15 µmol, whereas when EDC and NHS was included, the amount of 5 in the reaction solution was reduced to 2.11 µmol, indicating roughly 1 µmol of L-Rha derivative 5 had been conjugated onto the 6 mg of particles used.

Table 4.5 Amount of L-Rha conjugated to 1 µm AM-PS

L-Rha derivative EDC & Solution Amount of 5, µmol 5 NHS

PBS wash N Y 0 PBS wash Y N 3.15 PBS wash Y Y 2.11 Total conjugated to 6 mg of particles - - 1.03 Amount per mg of particle - - 0.17 Calculated maximum amount capable - - 0.1 of binding to 1 mg particles (4.2.4.1) N, no; Y, yes

4.2.4.4 Optimised EDC Coupling

Particle Preparation

As can be seen in Scheme 4.3, 1 μm AM-PS were first washed in hypotonic sterile water and then re-suspended in 10 mM sodium phosphate buffer, at pH 9.30, and left stirring overnight.

These particle preparation step caused a significant drop in zeta potential (Fig. 4.12 B), assumed to be due to de-protonation of the amino groups (Scheme 4.3).

149

Scheme 4.3 Particle Preparation Step: de-protonation of 1 μm AM-PS in preparation for EDC coupling reaction – particle preparation step. 1. Particles are diluted in water, pelleted and re-suspended in 10 mM sodium phosphate buffer at pH 9.3, at concentration 40 mg/ml, sonicated and then left stirring at room temperature overnight (2.). Scheme was generated by Chemdraw software.

EDC Coupling

Linker 8 or L-Rha-derivative 5 was diluted in sterile water with first EDC, followed by NHS. As

EDC contains equal equivalents of hydrochloric acid, this acid was neutralised with sodium hydroxide before the EDC was added to the ligand. After 15 minutes activation, where the reactive esters were formed, prepared particles (previous section) were added. The de- protonated amino groups then formed the amide bond by nucleophilic attack, as described previously. A summary of these coupling conditions can be seen in Scheme 4.4. After being left stirring for 2.5 hours, the by-products and leftover reagents were removed by washing the particles twice in 10 mM PBS, and finally re-suspended and sonicated in this same solvent.

150

Scheme 4.4 EDC coupling between synthesised L-Rha derivative 5 or linker 8, and 1 μm AM-PS. Scheme was generated by Chemdraw software.

4.2.4.5 Variation between Particle Batches

The batch of particles used for optimising the conjugation conditions described above (batch 1), expired shortly after the optimisation studies were concluded. As such a new batch was ordered from the manufacturer (batch 2). As can be seen in Fig. 4.14 A the second batch did not have a positive surface charge and it was thereby questionable whether or not these particles were aminated. Therefore, a third batch was ordered (batch 3), which did have a positive charge, that was significantly higher than the previous batch 1. The conjugation conditions specified in the previous section were attempted with this new batch (batch 3) and as can be seen in Fig. 4.14 B no decrease in zeta potential was measured compared to particles given EDC and NHS alone. As 151 the optimised conjugation conditions were not translatable to other batches of particles – as a results of major inconsistencies between commercial batches from the same supplier – this part of the project had to be abandoned.

Figure 4.14 Different batches of particles have differing zeta potentials and optimised EDC conditions were not translatable to other batches tested. Measured Zeta potential of 1 μm AM-PS at 100 10-6 v/v in 0.1 mM PBS. A) 3 different untreated particle batches. Results are shown as mean ± SD and analysed by student t- test, compared to batch 1. B) 3 µmoles of linker 8 (Linker-PS) or L-Rha derivative 5 (L-Rha-PS), 2 equivalents EDC (HCl neutralised with NaOH) and 20 equivalents of NHS dissolved in sterile water, stirred for 15-20 minutes, 6 mg of batch 3 prepared AM-PS were added. 2 washes with PBS preceded measurement of zeta potential. Reaction conditions: Prepared, prepared particles as specified in previous sections; EDC & NHS, EDC and NHS only; Linker-PS, EDC, NHS and linker 8; L-Rha-PS, EDC, NHS and L-Rha derivative 5. Results are shown as mean ± SD.

152

4.3 Discussion

4.3.1 Successful Synthesis of Ligands for Particle Functionalisation

As presented, two ligands were synthesised for particle functionalisation: an L-rhamnosyl derivative with an 11-atom linker attached, terminating in a carboxylic acid. The linker itself was also synthesised with a hydroxyl group in place of the L-Rha. This synthetic strategy was based on linkers previously used in a mice models, where L-Rha was conjugated onto proteins to induce and measure L-Rha-specific antibodies in mouse serum [196]. As the linker seemingly had no immunogenic properties, was not reported to cause toxicity in a mouse model and could be used to conjugate L-Rha onto aminated particles, this linker was deemed a suitable target compound for functionalising particles with L-Rha. Furthermore, the synthesis for the linker alone was also included, and required only one synthetic step. All these factors were ideal for the purpose of making functionalised AM-PS.

Whilst the literature reported conditions were used as a starting point, these reactions had to be optimised further to increase yields and aid better separation. For instance, the glycosylation reaction, to form 3, was reported to give a 73% yield when peracetylated rhamnose 2, synthesised phthalimide acceptor 9 and BF3OEt2 were at 1, 1.2 and 2.5 equivalents each; however when these same proportions were tested herein (Table 4.1), the yield was only 27%.

Whereas this reduction can partially be explained by the synthesis here being carried out at a larger scale (3 g of 2 rather than 1 g as published), a comparable yield of 74% was gained from using 6 g starting material and 67% from 18.2 g starting material using different conditions, namely 2:1:6 equivalents of 2:9: BF3OEt2 respectively. In addition, the time was increased from an overnight to a two-day reaction. These conditions are arguably more efficient timewise for preparing these large amounts of product 3, as it would take longer overall to make an equivalent amount by using the previously published conditions, especially as scaling up the amount of material lowered the yield considerably.

153

It could also be considered wasteful, using an excess 2. However, whilst any excess BF3OEt2 is lost, excess glycosyl donor 2 was easily recovered when carrying out the column chromatography separation, as this compound was more polar, the fractions with 2 were collected before the desired product 3. Furthermore, the two compounds that were the most difficult to separate when carrying out the column chromatography was product 3 from synthesised reagent 9. By using a large excess of BF3OEt2 and also having an excess of 2, this resulted in essentially all 9 being consumed in glycosylation reaction, and thus the main separation necessary was to separate 2 from 3 which, as stated, was easy with the column conditions employed. Another consideration for this glycosylation reaction was the need for fresh BF3OEt2, as the obtained results additionally showed how yields dropped steadily over time, until old BF3OEt2 was replaced with a fresh batch of reagent (Table 4.1). This reagent readily reacts with water, and whilst it was stored and used in as anhydrous conditions as possible, it would appear that over time the reagent degraded and a fresh batch had to be used.

Overall, this optimisation yielded reaction conditions that worked better on a larger scale and made isolation of the desired product easier.

The other reaction that was problematic was the de-protection step: de-acetylation and de- protection of the phthalimide by hydrazine to yield hydroxyl groups and a primary amine

(product 4). Similarly to BF3OEt2, hydrazine is also a reactive reagent that in this case seemingly lost potency over time, as can be seen in Table 4.2. Hydrazine is known to become unstable when de-hydrated, which could account for this loss of potency over time, and subsequent reduction in reaction efficacy. However, for this reaction there was an additional problem of an unidentified by-product being produced and isolated instead of the wanted product 4. This was problematic for several reasons; with the initial column chromatography conditions (100%

MeOH with less than 1% ammonia) this by-product had the same Rf-value as product 4.

Moreover, it was detectable with molybdenium stain (indicating the presence of a sugar) but not detectable with UV (indicative of the phenyl portion of the phthalimide) – which was all very

154 similar to product 4, however the proton-NMR showed clearly that this was not the desired product. As can be seen in Fig. 4.3, there was a peak missing. This peak was indicative of protons at the carbon neighbouring the newly formed primary amine and was an important peak for characterising compound 4. Its absence meant the amine somehow was no longer attached to the linker, and essentially that the wrong compound had formed.

To overcome this hurdle, alternative column chromatography conditions were investigated. A problem with the previous column conditions (100% MeOH with less than 1% ammonia) was that the Rf-value for product 4 was near 0 (just above the baseline), making it indistinguishable from any other similarly polar compound. As compound 4 is very hydrophilic, polar solvents had to be used to identify column conditions where it would have a higher Rf-value. However, a balance had to be found, as silica is also soluble in water and to a certain extent in other polar solvents like methanol and ethanol; and as silica was the solid phase used for the chromatography, some inorganic solvent had to be included to gain some stability. Eventually, a solvent system was found where product 4 had an Rf-value of 0.16 and where the silica was still stable: 60% EtOH, 20% H2O, 15% EtOAc and 5% acetone and less than 1% of ammonia solution.

With this solvent system, the by-product appeared at a lower Rf-value and could thus be separated from product 4. Overall, the same reaction conditions as previously reported [196] gave the highest yield of 77%. Although this is significantly less than the reported 96%, the amount of starting material was also more than three times larger (1.4 g compared with 0.4 g) which could explain this difference.

For the last reaction, the addition of succinic anhydride to form compound 5, some slight changes to the methodology improved both yields and the separation of this compound.

Whereas 1:1 equivalents of 4 to succinic anhydride gave 96% yield when 50 mg of starting material (4) was used [196], scaling up the amount of starting material to 300- and 458 mg, using the same conditions, gave yields of only 45% and 35% respectively. Instead, by increasing the amount of succinic anhydride to two equivalents resulted in an 83% yield, when 585 mg of

155 starting material was used. Succinctly, the desired L-rhamnosyl derivative with carboxylic acid- terminating linker was successfully synthesised, and fully characterised.

Regarding the synthesised linker 8, although the highest yield for this reaction was only 49%, this still provided sufficient material to carry on with the next step: particle functionalisation. As such, there was no need to optimise this reaction further. To summarise, two ligands for particle functionalisation were successfully synthesised and fully characterised.

4.3.2 Functionalisation of aminated 1 μm polystyrene particles

By choosing a linker with a terminal carboxylic acid, to form an amide bond by EDC coupling, the choice of particle was limited to amine-functionalised particles. Whereas polystyrene particles are not ideal for human use, as biodegradable polymers are far more favourable, they do however have a couple of advantages for research purposes, namely that they are readily available in a broad range of sizes and with little deviation. As could be seen by the DLS measurements (Fig. 4.12 E) the PDI was very low (0.3), and only one size of particle was detected (Fig. 4.13 C). Furthermore, when the particles were tested for TLR-contamination, none was observed (Fig. 4.7 A). Thereby, to investigate the effect of multivalent expression of L-

Rha with a specific size of particle, 1 μm, these aminated particles were ideal.

Another factor to consider however was the charge. As mentioned in the introduction to this chapter, surface charge of particles is a factor that is postulated to affect their immunogenicity.

Positively charged particles have been proposed to be more pro-inflammatory [123] whereas neutral and negatively charged particles are not [123,237]. This could at least partly be explained by the positive charge extending the interaction between the particle (and its ligands) and the negatively charged surface of cells, thereby enhancing both specific and non-specific interactions [143,228,234,238], leading also to an enhanced degree of induced response and particle uptake [57,235].

156

As the primary hypothesis for this research is that combining immunotolerogenic properties of

L-Rha and polystyrene particles of 1 μm will in combination synergistically enhance these immunomodulatory properties, the use of potentially pro-inflammatory AM-PS may seem counterintuitive. However, if one considers that these responses result from the extended non- specific interactions with the cell surface, functionalisation should alter this effect in at least two ways. Firstly, the coupling reduces the amount of positive charge, as it is the amino groups that are responsible for the positive charge and these are “consumed” through the formation of amide bonds. Secondly, even if the coupling was sparse enough that some positive charge remained, enhancing interaction with the cell membrane could help extend the interaction between L-Rha and a receptor, if that is indeed how L-Rha is able to induce its immunoregulatory responses. Essentially, the particles being positively charged is not necessarily an issue and could even help enhance the responses to L-Rha.

Unlike the previous synthesis described, there are very few previously published methods for conjugating ligands onto AM-PS by EDC coupling, and none for polystyrene particles of this size.

The closest reported literature protocol was where 11 nm iron oxide particles were amine- functionalised and subsequently conjugated with a polysaccharide, where EDC coupling was utilized as a ligation strategy [253]. Other related examples are when chitosan was coupled to a ligand and particles were subsequently formed with the modified polymer [254–256]. As such, very little was known regarding appropriate amounts of ligand, what solvent/buffer system to use or what timeframe would be the best. On the other hand, EDC coupling chemistry is well documented and consulting such literature provided insight regarding which factors needed to be considered (buffers, concentrations and time) as well as the kind of systems that should prove successful. The task therefore was to investigate these conditions, as well as to confront the greater challenge of taking into account the unknown information regarding the amination of the particles.

157

The first challenge with the EDC coupling was therefore to determine what amounts of ligand and reagents to use. As mentioned in section 4.2.4.1, the manufacturer of the particles was unable to provide any information regarding coverage of the amino groups or even how the particles had been modified to display the amine at the surface. However, knowing the particle size and having an idea of its structure, allowed a theoretical maximum number of conjugation sites to be calculated: 0.1 μmol/mg. These calculations gave a starting point to ensure an excess of ligand and reagents were applied, starting at twice the maximum, and increasing to five times this maximum for the amount of particles used. Following the successful trial of L-Rha conjugation (Fig. 4.13), the amount of remaining L-Rha was measured, which revealed roughly

0.1 μmol L-Rha was conjugated per mg of particle – the same as the calculated estimate. Given that there was twice this amount of L-Rha remaining in the reaction mixture, this may be the maximum amount of L-Rha to conjugate to these particles. This result affirms that calculating a theoretical maximum was a useful strategy to estimate amount of ligand to use.

Having calculated an amount of ligand and reagents to use the coupling trials began. The initial coupling conditions employed were partly based on the work of Dr Muñoz-Wolf (unpublished), where she conjugated carboxylated particles with a lipoprotein ligand. The conditions allowed for the formation of reactive esters over a period of 24 hours, by first incubating EDC with the carboxylic acid ligand for two hours before adding the NHS reagent overnight. The AM-PS were then added for an additional 24 hours. Regarding the initial buffers used, the pH of buffer I and buffer II (used for the reactive ester formation and the amide bond formation respectively) were chosen based on the approximate pKa of a carboxylic acid and a primary amine respectively, i.e. pH 5 and 9 respectively, as aiming for these conditions was recommended

[257]. The last step to be considered was the removal of leftover reagents and ligand in the reaction mixture. For this, the simplest solution was washing the particles in sterile water (Fig.

4.10).

158

To monitor the reactions, measuring surface charge by zeta potential was used. As the AM-PS have a positive charge (when the amino groups are protonated), when functionalised this positive charge is lost and should correspond with a drop in zeta potential. EDC coupling trials commenced with synthesised ligand 8. It was assumed that exchanging the hydroxyl group of 8 with the L-Rha of 5 would not change the reactivity of the terminal carboxylic acid and thus the optimal conditions found for 8 should essentially be the same as for 5.

The initial conditions chosen for the EDC coupling at first appeared to have worked: the zeta potential measurements showed a drop in surface charge. However, this decrease occurred regardless of reaction conditions and turned out to be a result of the post-reaction washing in sterile water. As can be seen in Fig. 4.11 B two washes in sterile water caused the zeta potential to decrease to below 0 mV. The hypothesised cause for this observation was that the dilution of

AM-PS in a solution with very low ionic strength drove the de-protonation of the amino groups, and an equilibrium being reached where significantly fewer amino groups were protonated. The subsequent removal of the solvent would then result in a loss of available protons, and the following re-emersion in more sterile water would induce further de-protonation.

It is important to note here that all zeta potential measurements were carried out in at least 0.1 mM PBS as solvent, and that this buffer was not able to protonate the particles fully. The reason for this is most likely the pKa of the amines in question. The only specification given by the manufacturer regarding the amine functionalisation was that the amine was aliphatic, i.e. not directly connected to the phenyl component of the polystyrene. Aliphatic amines have a pKa between 9 and 11, so at pH 7.4, if the pKa of the amine was at the lower end of the scale (pKa of

9) the ratio of de-protonated:protonated amino groups would approximately be 1:4

(Henderson-Hasselbalch equation), increasing up to 1:4000 if the pKa was around 11. As PBS was evidently not a good buffer for protonating the particles, this suggested that the pKa of the amino groups was around 9. As a phenomenon this was quite useful; as it meant that the

159 solvent used for the surface charge measurements was having minimal effect on this characteristic, and was therefore ideal to measure any differences.

Going back to the issue that the water wash was potentially causing particle de-protonation; whilst PBS was unable to effectively protonate the particles after a water wash, changing the wash solvent from water to a buffered system would in theory help prevent the particles becoming de-protonated in the first place. To test this hypothesis, particles were washed in an identical manner, but with 10 mM PBS (pH 7.4) as the diluent. As can be seen in Fig. 4.11 C, when the particles were washed with PBS the surface charge did remain positive. Consequently, henceforth all post-reaction washes were carried out with 10 mM PBS. However, another observation was that the PBS-washed particles were markedly less positive than the unwashed particles. This difference was most likely due to the presence of cationic surfactant in the untreated particles, which had been washed away by either water or PBS wash.

Once the PBS wash was employed it was revealed that the initial reaction conditions were not causing a noticeable difference in zeta potential between particles exposed to coupling conditions with ligand and reagents present, compared with particles exposed to the same conditions without any ligand. The first component considered was the particle nucleophilic attack phase. As stated, for the amino groups on the particles to partake in the reaction they need to have a free electron pair available, therefore they cannot be protonated. The solvent employed for this part of the reaction was initially therefore an alkaline buffer, to induce de- protonation. However, immersing particles in this alkaline buffer overnight only caused a minor drop in surface charge. From this it was concluded that the reaction probably was not occurring because there were hardly any de-protonated amino groups available for the coupling to take place. As such, the next task became to try to increase the number of available amino groups.

A method for de-protonating the particles had serendipitously already been found, namely washing the particles in sterile water. When this was tested for the first time (Fig. 4.12 A) it was

160 found that washing the particles once before the reaction indeed did result in a ligand- dependent drop in surface charge, indicating some degree of coupling had occurred. This was tested with 1.2 μmol and 3 μmol of ligand (same equivalents of EDC and NHS), and there was essentially no difference in zeta potential between the two, indicating that it was indeed the amino group availability that was the limiting factor for the coupling to occur. Therefore, means of de-protonating the particles were investigated further. As has been stated, washing this amount (6 mg) of particles twice in water could reduce the surface charge to around 0 mV, which would indeed indicate more amino groups available for de-protonation. However, a zeta potential of near 0 mV is cause for concern from a stability point of view.

For particles in a solution to not aggregate with each other, and thus maintain their shape and size, they need to repel each other. The current rule-of-thumb is that when the zeta potential is approximately above 30 mV or below -30 mV, the surface charge is high or low enough to yield electrostatic repulsion at an extent where the particles in the formulation are stable [258].

Contrastingly, when the surface charge of particles nears 0 mV, there is little to no repulsion keeping the particles separate, resulting in a high risk of aggregation. Therefore, more than two water washes was not used to avoid this complication. Instead, the combination that gave the highest level of de-protonation was an initial water wash, then re-suspension in alkaline sodium phosphate buffer (pH 9.30, 10 mM), followed by sonication, and then stirring overnight before addition to the reaction. This gave particles that had a much more stable surface charge, below

-30 mV; which additionally intimated that the majority of amino groups on the particles were de-protonated after being prepared in this fashion, and thus now available for the amide coupling (Fig. 4.15). Fortunately, washing the particles in 10 mM PBS after the reaction protonated the majority of unconjugated amino groups again, therefore a drop in zeta potential could still be used to monitor successful conjugation. This was no doubt due to the higher concentration (10 mM) of PBS, as the 0.1 mM PBS used for the zeta measurements did not have this effect.

161

Figure 4.15 Degree of protonation affects both particle stability, and availability for amide coupling. A) For particles in solution to retain their shape and avoid aggregation, they need sufficient electrostatic repulsion, i.e. surface charge of ≤ -30 mV or ≥ 30 mV. B) The amino group on AM-PS needs to be de-protonated to furnish them with a free electron pair available for the amide coupling.

Having developed a post-reaction washing procedure and promising particle preparation technique, there followed further optimisation by considering other factors, such as solvent systems and the time for the reaction. Eventually, the best conditions used were rather different from the starting ones. For instance, instead of using a slightly acidic buffer for the reactive ester formation, using water proved to be better. Changing the solvent to water also meant the hydrochloric acid associated with the EDC had to be neutralised before added to the reaction. Acidic conditions can potentially cause lysis of amide bonds and it could therefore be the case that the pH 5 buffer in combination with the EDC associated HCl was interfering with the reaction, making water a better solvent. Additionally, it was found that shortening the time, 162 to a total reaction time of 2.5 hours also gave more promising results. The change in time from

24 hour reactive ester formation and 24 hour coupling to 15 minutes and 2 hours respectively was based on previously published guidelines [257], which stated that this is a rapid reaction that should not require any more time than this. Apart from this protocol being more efficient, having a shorter timespan for the reaction also seemed to reduce particle aggregation (seen visually and measured by DLS, not shown).

The coupling conditions found caused a significant ligand-dependent drop in zeta potential, intimating successful particle functionalisation with both ligands 5 and 8. In the case of the L-

Rha linker, the conjugation was further confirmed by spectrophotometry. Regarding the hydrodynamic size, measured by DLS, it was revealed that there was still a majority of particles of the desired size, and no other size was picked up by the zetasizer (Fig. 4.13 C). However, what was more of a concern was that the PDI increased, compared with untreated particles, from roughly 0.3 to 0.6. This increase was universal for all treated particles, regardless of reaction conditions, and was therefore unlikely to be a by-product of the functionalisation. The fact that no other particle size was detected, but that the PDI increased indicated that the centrifugation and re-suspension caused a certain degree of aggregation, yielding particles larger than the detection limit (10 μm) as well as the desired particles. This being the case, further optimisation of the protocol would have been necessary to try to remove these aggregates from the solution, or alter washing conditions, such as the centrifugation speed, to avoid the aggregates in the final product.

4.3.3 Conclusions and Future Experiments

Unfortunately, at this stage the batch of particles (batch 1) used for the optimisation had passed their expiry date and could not be used. The new batches of manufactured particles however did not have the same zeta potential when untreated, as batch 1 had. Even though batch 3 was at least aminated (judging from the positive charge), the particle preparation conditions did not

163 appear to de-protonate the particles (-20 mV) to the same extent as batch 1 (below -30 mV) and the optimised reaction conditions outlined in 4.2.4.4 did not cause any noticeable conjugation of either 5 or 8, as the zeta potential was not lower than those given reagents EDC and NHS (Fig.

4.14 B). These results indicated that each batch of particles required its own optimisation for conjugation to occur. Even though, through this project it was found that de-protonating the particles before the reaction was a key step to achieve conjugation, and ways of doing this were found, there were still too many conditions that had to be tested to find a protocol. Therefore, because of the lack of reproducibility by the manufacturer, the issue of having to optimise a protocol for each particle batch meant this part of the project had to be abandoned.

Given that not enough L-Rha coated particles were made for immunological studies, to relate L-

Rha immunosuppressive properties to Mycobacterium tuberculosis pathogenesis, the second synthetic strategy was next pursued, namely to link the L-Rha induced responses to the mycobacteria-specific pHBADs.

164

Chapter 5. Does L-Rhamnose Convey Immunomodulatory

Properties to Mycobacterial PGLs?

5.1 Introduction

The aim of this research was to explore the hypothesis that L-rhamnose (L-Rha) exerts immunomodulatory effects that could shield Mycobacterium tuberculosis from harmful protective host immune responses. The previous two chapters explored the effect of L-Rha on macrophage activation and innate memory, and the use of synthetic chemistry to connect the soluble form of L-Rha back to the surface expression of the mycobacterial phenolic glycolipids

(PGLs). This final result chapter addresses whether L-Rha is critical for pHBAD I to induce its immunomodulatory effects.

5.1.1 PGL and pHBAD Structures

Mycobacterial PGLs are found on the outer cell wall of several mycobacterial species:

Mycobacterium tuberculosis, M. bovis, M. leprae, M. ulcerans, and M. kansasii [188] and are composed of a lipid core, phthiocerol dimycocerosate (PDIM), an aromatic ring and an outermost sugar moiety [53,188]. The outermost sugar moieties for the listed mycobacteria all share an α-linked 2-O-methyl L-rhamnopyranoside [188,201,202,259]. The major M. tuberculosis PGL (PGL-tb) and the M. leprae PGL (PGL-1) both have a trisaccharide sugar motif, with an additional L-Rha residue, followed by a terminal 2,3,4-tri-O-methyl fucose or 3,6-di-O- methyl glucose respectively (Fig. 5.1) [188,259,260]. The minor M. tuberculosis PGL is also found as the major PGL on M. bovis (mycoside B), and has only the first L-rhamnosepyranoside sugar moiety [260,261] (Fig. 5.1).

165

Figure 5.1 L-Rha (in black) is a prominent structural component of mycobacterial PGLs and pHBADs. Structures of the mycobacterial species-specific sugar moieties and the common lipid core (PDIM) that composes the major PGLs produced by M. leprae (PGL-1), M. tuberculosis (PGL-tb) and M. bovis (mycoside B).

Only the most virulent strains of M. tuberculosis express PGLs on their surface, due to most strains containing a pks 15/1 frameshift mutation, which prevents the formation of the phenol- terminating lipid core [181]. However, all clinical strains of M. tuberculosis produce and secrete para hydroxybenzoic acid derivatives (pHBADs) which, instead of the PDIM lipid core, possess a para methyl ester group [181,192] (Fig. 5.1). The major M. tuberculosis pHBAD is known as pHBAD II (PGL-tb glycan), and its counterpart in M. bovis is known as pHBAD I (mycoside B glycan) [167,188,192]. The biosynthesis of the saccharide residue of PGL-tb or mycoside B, as well as the pHBADs, starts with either the para hydroxyphenyl PDIM, or methyl para hydroxybenzoate precursors [188], which are glycosylated with a L-Rha residue, forming the α bond [262]. The L-Rha residue is then O-methylated at the second carbon [263], giving the structure of either mycoside B or pHBAD I, depending on which phenolic precursor was glycosylated [188]. Both the pHBAD I and the prior un-methylated pHBAD I (UM-pHBAD I) have been shown to be secreted from strains of M. tuberculosis [189]. To form the trisaccharide of

PGL-tb and pHBAD II, the two remaining sugar residues are added (exact sequence of the

166 addition is presently unknown), followed by O-methylation of the terminal L-fucose residue

[188,264]. The addition of the second L-Rha residue is carried out by an enzyme encoded by

Rv2958c [265]. The presence of a frameshift mutation in this gene is the reason why M. bovis only expresses monosaccharide PGL: mycoside B [266].

5.1.2 Immunomodulatory Properties of PGLs and pHBADs

As most of the research regarding the immunomodulatory properties of PGLs have looked at M. tuberculosis, M. bovis and M. leprae, the research presented in this section will focus primarily on these mycobacterial species. As can be seen in Table 5.1, mycobacterial PGLs from M. leprae,

M. tuberculosis and M. bovis, their phenolic glycan derivatives, as well as synthesised structural analogues of the former have been shown to possess immunomodulatory properties

[167,190,200–202,267]. Of these responses, there is a clear pattern of inhibition of both human and murine macrophage pro-inflammatory responses, such as secretion of TNFα

[167,190,200,201,267], IL-6 [167,190,267], IL-1β and MCP-1 [190,201,267] and reduce iNOS

[202] and NO production [190,202,267]. Moreover, a key observation regarding these immunomodulatory compounds can be made: that the phenolic glycan portion of the molecule appears to be critical for the immunosuppressive properties.

167

Table 5.1 The Immunomodulatory Properties of PGLs, their Derivatives and Synthesised Analogues

Immune Mycobacterium Structure Stimulus Response Refs Cell(s) M. tuberculosis IFNγ & LPS ↓ iNOS, NO BMDMsm [202] X = PDIM H37Rv ↓ TNFα, MCP-1 BMDMsm [201]

h BCG ↓ TNFα MDM [200] PGL-tb H37Rv ↓ IL-1β, TNFα PBMCsh [268] aCD3 ↓ IFNγ, IL-17, IL-10 Splenocytesm [167]

↓ TNFα, IL-1β, IL-6, PAM3CSK4 THP-1h [267] pHBAD II MCP-1, NO

PAM3CSK4 ↓ NFκB activation THP-1h [200] ND TNFα, IL-1β, IL- LPS THP-1h [267] 6, MCP-1, NO PDIM IFNγ & LPS ND iNOS, NO BMDMsm [202] M.tuberculosis and X = PDIM IFNγ & LPS ↓iNOS, NO BMDMsm [202] M. bovis Mycoside B ND TNFα, IL-1β, IL- PAM3CSK4 THP-1h [190] 6, MCP-1, NO aCD3 ↓ IFNγ, IL-17, IL-10 Splenocytesm [167] ↓ TNFα, IL-6, IL- H37Rv BMDMsm [167] pHBAD I 12p40

PAM3CSK4 ND NFκB activation THP-1h [200]

aCD3 ↓ IFNγ, IL-17, IL-10 Splenocytesm [167]

m H37Rv ↓ TNFα BMDMs [167] UM-pHBAD I M. leprae ↓ iNOS, NO, IFNβ, IFNγ & LPS BMDMsm [202] CXCL10 ↓ TNFα, IL-1β, IL-6, PAM3CSK4 THP-1h [190] MCP-1, NO ND TNFα, IL-1β, IL- X = PDIM LPS THP-1h [190] 6, MCP-1, NO

[200,2 PGL-1 BCG ↓ TNFα MDMh 69]

h BCG ↓ NFκB activation THP-1 [269] PAM3CSK4 ↓ NFκB activation HEK [200]

↓ TNFα, IL-1β, IL-6, PAM3CSK4 THP-1h [190] MCP-1, NO ND TNFα, IL-1β, IL- LPS THP-1h [190] 6, MCP-1, NO

168

↓ TNFα, IL-1β, IL-6, PAM3CSK4 THP-1h [190] MCP-1, NO

ND TNFα, IL-1β, IL- [190,2 PAM3CSK4 THP-1h 6, MCP-1, NO 70]

↓ TNFα, IL-1β, IL-6, PAM3CSK4 THP-1h [190] MCP-1, NO

ND TNFα, IL-1β, IL- LPS THP-1h [190] 6, MCP-1, NO

ND TNFα, IL-1β, IL- PAM3CSK4 THP-1h [190] PDIM 6, MCP-1, NO PAM3CSK4 ND NFκB activation HEK [200] ↓, reduced; ND, no detectable effect; h, human; m, murine; HEK, human embryonic kidney reporter cell line for TLR-2 signalling;

As can be seen in the stimulus column of Table 5.1, PGL and related structures reduced macrophage pro-inflammatory cytokine secretion, as well as bactericidal NO production, when stimulated with mycobacteria [167,200,201] or the TLR2 ligand PAM3CSK4 [190,267]. M. tuberculosis expresses a large repertoire of TLR2 agonists, signalling from which has been implicated in several important aspects of its pathogenesis, including phagolysosomal escape

[204], inhibited MHC II antigen processing and presentation [213], as well as Treg recruitment

[214]. Interestingly, these prior studies also identified that PGLs and related structures inhibit

TLR2 mediated responses quite specifically, as they also found that responses induced by the

TLR4 agonist, ultrapure LPS, were not affected [190,267].

With regard to this latter result, there appears to be some contradictory findings; as can moreover be seen in Table 5.1, PGLs have also been shown to inhibit LPS induced macrophage

NO and IFNβ when given concomitantly with IFNγ [202]. Furthermore, this inhibition was proposed to be TIR-domain-containing adapter-inducing interferon-β (TRIF)-dependent. TRIF mediates signalling upon TLR3 and TLR4 agonist recognition [271]. Isolated PGL-1 was shown to reduce TRIF protein levels and affect specifically TRIF-dependent responses, such as NO and

IFNβ [202], whilst TRIF independent responses, such as IL-6 and Arg1 expression, were

169 unaffected [202]. Apart from this dependency on TRIF, the inhibition of IFNγ and LPS mediated

NO by isolated PGL-1, PGL-tb and mycoside B, was demonstrated to be impeded in myeloid differentiation primary response 88 (MyD88) and CD11b knock-out mice. MyD88 mediates signalling from all TLRs apart from TLR3 [271].

Whilst this study by Oldenburg et al. was the first publication to identify a potential role for

MyD88, the possible role of CD11b to mediate immune response inhibition by PGLs has been proposed previously [269]. CD11b with CD18 forms complement receptor 3 (CR3), and blocking

CR3 with an inhibitory antibody enhanced human monocyte derived macrophage secretion of

TNFα upon challenge with recombinant BCG expressing PGL-1, compared to when no anti-CR3 antibody was present [269]. Interest in CR3 dates back to 1991, where it was demonstrated that its ligand, complement component 3 (C3), bound to PGL-1 [272] and that CR3, along with other complement receptors, were mediators of human macrophage phagocytosis of M. leprae [273].

These latter studies have identified a number of possible components for the mechanism that drives immunomodulation by PGLs, including identifying roles for MyD88 and CR3 to mediate reduced levels of macrophage NO and TNFα secretion [202,269], as well as demonstrating that

PGL-1 reduces TRIF protein levels and that responses targeted are TRIF dependent [202].

However, it is important to note two issues: firstly as mentioned that these studies contradict others carried out by Lowary and co-workers that showed immunomodulation caused by PGLs and related synthesised structures were tied to TLR2 signalling (which are TRIF independent)

[190,267] and secondly that all studies that identify a role for CR3 and TRIF involve experimentation with isolated PGLs, where the lipid portion is included [202,269]. On the one hand, it was shown that isolated M. tuberculosis PDIM (the lipid portion) did not inhibit IFNγ and LPS induced NO, when PGL-tb did, indicating that the outermost aromatic sugar portion of the molecule is important for immunomodulation [202]. On the other hand, the aromatic sugar component was not investigated on its own. This is important to consider, as the initial study that identified binding of C3 to PGL-1, found that both the lipid and the outermost sugar were

170 important for C3 binding [272]. Sequential removal of the each of the sugar monomers building the outermost trisaccharide sequentially also reduced the binding of C3 – however removal of the PDIM lipid portion had the same effect on C3 binding; indicating both components were necessary. As this work considers the potential role of L-rhamnose, therefore, the remainder of this section will focus on the existing literature that outlines the potential structure-function relationship of the L-rhamnose-rich structures in isolation, such as the pHBADs and related structural analogues.

Phenolic Glycan is Essential for Immunomodulation

As shown previously, PGLs have three structural components, a lipid core bound to the membrane, the phenol and the outermost sugar moiety. Of these structural elements, previous studies have demonstrated that the phenolic glycan is critical for conveying immunomodulatory properties [190,200,202]. Work carried out by Lowary and co-workers demonstrated that pHBAD II [267], the native PGL-1 as well as a synthetic analogue of the phenolic glycan – which had a para methyl ester group (akin to pHBAD structure) instead of the M. leprae PDIM lipid core [190] – reduced PAM3CSK4 induced THP-1 secretion of TNFα, IL-6, IL-1β and NO, whereas stimulation with isolated PDIM alone had no such effect [190,267]. These observations tie in with results obtained in our lab, where pHBAD I and pHBAD II reduced irradiated M. tuberculosis

H37Rv mediated TNFα, as well as aCD3 induced splenocyte secretion of IFNγ, IL-17A and IL-10

[167]. Given that the phenolic glycans have immunomodulatory properties [167,190,202,267], whereas M. tuberculosis PDIM [202] and M. leprae PDIM [190] alone do not, these results overall display that the outermost phenolic glycans are a key motif for furnishing mycobacterial

PGLs with their immunomodulatory properties.

Another observation is that increasing the number of saccharides in the sugar motif appears to enhance the immunosuppressive effects [190,200,202]. When comparing the capacity of the aromatic glycan components from M. tuberculosis, M. leprae and M. bovis to inhibit PAM3CSK4 induced NFκB activation, only the compounds with trisaccharides displayed any inhibitory

171 effect, whereas the monosaccharide compound based on M. bovis pHBAD I did not [200].

Furthermore, when M. bovis was genetically engineered to express other mycobacterial PGLs, the change from mycoside B to either PGL-1 or PGL-tb reduced the elicited secretion of TNFα from human monocyte derived macrophages (hMDMs) [200]. Similarly, when comparing analogues of the phenolic glycans, the trisaccharide motif of PGL-1 reduced PAM3CK4 mediated pro-inflammatory responses by THP-1 cells to a greater extent than a analogue of

L-Rha, which still mediated an immunosuppressive effect that was not detected for the monosaccharide aromatic glycan pHBAD I at the concentration used (0.16 mM) [190]. Whilst these studies demonstrate that an increased number of saccharides can enhance the immunomodulatory effects of these aromatic glycan structures, it appears to not always be the case. For instance, this was not seen in our lab, where pHBAD I, but not pHBAD II, reduced irradiated M. tuberculosis H37Rv stimulated BMDMs secretion of TNFα and IL-12p40 at 0.1 mM and 0.5 mM [167]. These studies addressed various experimental models – different cell types and stimuli – and these discrepancies could therefore be attributed to such differences. Overall, prior work would suggest that both pHBAD I and pHBAD II have immunomodulatory properties, but that their potency can vary depending on which experimental model is used.

Increased Hydrophobicity Amplifies Immunosuppression

Whilst the phenolic glycan is pivotal for PGLs to have immune modifying properties, research has also revealed that the presence of the lipid core amplified their immunosuppressive potential [190,200]. When investigating PGL-1 induced repression of PAM3CSK4 stimulated

NFκB activation and subsequent TNFα secretion, it was found that both the native PGL-1 and the aromatic glycan component had these attenuating properties, however PGL-1 was 10 times more potent [200]. Similarly, when the phenolic glycan of PGL-1 (with a para methyl ester group instead of the PDIM) was compared to the same phenolic glycan, but with a para 19-atom long hydrocarbon tail instead of a methyl ester group, it was found that the addition of the hydrophobic hydrocarbon tail enhanced the suppression of PAM3CSK4 induced pro-

172 inflammatory responses [190]. As with the lipids PDIM (M. leprae) [190,270] and PDIM (M. tuberculosis) [202], the hydrocarbon tail – in this case with a terminal phenol – did not suppress pro-inflammatory responses [190].

That less hydrophilic variants of these structures have enhanced immunosuppressive properties is further corroborated by prior work by Lowary and co-workers. They investigated how synthetically altering the methylation pattern of sugar moieties of M.tuberculosis pHBAD II [267]

– and the corresponding phenolic glycans of M. leprae [190] and M. kansaii [270] – affected their immunomodulatory properties. When inhibition of PAM3CSK4-induced TNFα, IL-6, IL-1β,

MCP-1 and NO secretion by THP-1 cells was compared between variants of these phenolic glycans, variant structures with methoxy groups were more inhibitory than their hydroxyl group counterparts [190,267,270]. Summarily these findings indicate that whilst the aromatic glycan is essential for immunomodulatory properties to be conveyed to mycobacterial PGLs, increasing hydrophobic properties of the molecule enhances their potency.

173

5.1.3 Hypothesis and Aims

In Chap. 3 it was found that L-Rha has immunomodulatory properties, such as inhibition of IFNγ secretion by splenocytes and IFNγ-induced macrophage activation (Nos2 expression and subsequent NO). This aligned well with what has been shown for the L-Rha-rich mycobacterial

PGLs and pHBADs [202,267]. Given the prior literature highlighting the importance of the phenolic glycan to induce immunomodulatory effects, the hypothesis for this chapter was that the presence of L-Rha was essential for pHBAD I to display immunomodulatory properties. The aim for the work presented herein was therefore to repeat the key immunological experiments carried out in Chap. 3 with not only L-Rha and pHBAD I, but with several synthetic structural derivatives of pHBAD I, to examine the necessity of L-Rha for immunomodulatory properties to be detected. Moreover, given that L-Rha was shown to induce innate memory responses which enhanced irradiated M. tuberculosis induction of anti-inflammatory IL-10 and reduced its induction of Nos2 and Ifnb1, another aim was to investigate whether the mycobacterial pHBAD I also held similar innate memory stimulating capabilities.

174

5.2 Results

5.2.1 Comparison of the Immunomodulatory Properties of L-Rhamnose and pHBADs

As previous work in the Scanlan lab had shown that pHBAD II and pHBAD I could reduce splenocyte secretion of IFNγ [167], this was the first experiment carried out with L-Rha in

Chap.3 – a comparison of the three compounds can be seen in Fig. 5.2 A. This comparison shows that all three compounds reduce IFNγ secretion by splenocytes at concentrations of 0.1 mM or higher. The synthesis of pHBAD I and pHBAD II, as well as their data in Fig. 5.2 A, were carried out by Danielle Barnes, PhD. With the remaining pHBAD I synthesised by Danielle, a direct comparison with L-Rha was carried out next, investigating the effect upon IFNγ induced macrophage activation. Both pHBAD I and L-Rha significantly reduced IFNγ induced Nos2 expression (Fig. 5.2 B) and IFNγ and LPS induced NO production (Fig. 5.3 C). Stimulation with the sugar D-glucose had no significant effect upon Nos2 or NO production at the same concentration as L-Rha and pHBAD I (Fig. 5.3 B and C). Having confirmed that pHBAD I possessed similar immunomodulatory properties to L-Rha, the next step was to explore the structure-function relationship in greater detail by synthesis of pHBAD I analogues.

175

Figure 5.2 L-Rha and pHBAD I attenuate IFNγ secretion and IFNγ-induced Nos2 and NO. A) % IFNγ secretion from splenocytes (C3H/HeJ mice) stimulated with 0.4 μg/ml anti-CD3 antibody and pHBAD II, pHBAD I or L-Rha as specified for 72 hours. Results (n = 3) are shown as mean ± SD and analysed with one-way ANOVA, with Tukey’s multiple comparisons post test. B) RT-qPCR on BMDMs (C57BL/6 mice) stimulated for 24 hours with recombinant murine IFNγ (100 ng/ml) or IFNγ with LPS (0.5 ng/ml each), with either media or 1 mM of pHBAD I, L-Rha or D- glucose. Fold change in comparison to BMDMs incubated with media alone, with ActB (shown) and Tbp as internal references. Results (n = 3) are shown as mean ± SD and analysed by ratio-paired t- test. C) NO secretion (Griess reagent) from BMDMs (C57BL/6 mice) stimulated for 24 hours with IFNγ and LPS (0.5 ng/ml each), in the presence of either media or 1 mM of pHBAD I, L-Rha or D-glucose. Results (n = 3) are shown as mean ± SD and analysed by student t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with 0 mM or media control groups.

5.2.2 Synthesis of pHBAD I Analogues

The structures of the synthetic pHBAD I derivatives investigated in this chapter can be seen in

Fig. 5.3. These compounds included not only the mycobacteria-specific O-methylated pHBAD I and its un-methylated precursor (UM-pHBAD I), but also analogues of these, to investigate the structure-function relationship between the methylation of the L-Rha, the para methyl benzoate group, the aromatic ring, the α-configuration at the anomeric carbon (C-1) of L-Rha, as well as L-Rha itself to the immunomodulatory properties held by pHBAD I. Prior work from the

176

Scanlan lab demonstrated the synthesis of both pHBAD I and UM-pHBAD I [167]. These compounds, used for the remainder of this chapter, were synthesised by Lauren McSweeney,

PhD, and Dylan Lynch, B.Sc., respectively, following a further optimised synthetic procedure developed by Danielle Barnes, PhD. The aromatic compound methyl-p-anisate (m-p-a, Fig. 5.3) was commercially available, and the pHBAD I analogues (pIAs) 1-4 were prepared synthetically.

Figure 5.3 Structures of L-Rha, pHBAD I, UM-pHBAD I and the analogues of pHBAD I (pIAs) used in this chapter. m-p-a, methyl-p-anisate. Structures drawn with Chemdraw software.

Synthetic Strategy

Scheme 5.1 Synthetic strategy to prepare pIAs 1-4. Scheme was drawn with Chemdraw software

177

The synthetic strategies for preparation of the pHBAD I analogues pIAs 1-4 can be seen in

Scheme 5.1. This synthesis was carried out by Dylan Lynch, B.Sc., and involved two strategies.

To prepare the acetylated precursors of pIAs 1, 2 and 4, peracetylated L-Rha (2) was used as a

Lewis-activated glycosyl donor, to glycosylate methyl 4-hydroxybenzoate, phenol or ethanethiol respectively. As described in Chap. 4, boron trifluoride diethyletherate (BF3OEt2) was employed as the Lewis acid activator, to promote formation of the oxonium ion intermediate at the anomeric position of 2. The formation of a second cationic intermediate, an acetonium ion, restricted the subsequent nucleophilic attack to the axial face, furnishing exclusively the desired

α-products. This glycosylation reaction was followed by de-acetylation upon treatment with sodium methoxide: the methoxide ion functioned as a nucleophile, displacing the acetyl group as the ester methyl methanoate, furnishing the de-protected hydroxyl groups in quantitative yield.

The second synthetic strategy, to make pIA.3, involved Fischer glycosylation over two steps.

First, 2 equivalents of acetyl chloride were reacted with an excess of methanol, forming a solution of methanolic HCl. The acid then activated the anomeric (C-1) hydroxyl group of L-Rha, which was eliminated as water. Subsequently, the methanol hydroxyl group attacked the resulting oxocarbenium-type intermediate of L-Rha, furnishing it with the methoxy group at the anomeric position. The major product formed was in the α-conformation, and the β side product was removed by column chromatography.

5.2.3 Immunological studies

Cytotoxicity and Sterility of Compounds

Prior to cell work, all compounds were dissolved in sterile water, except for the more hydrophobic m-p-a, pIA.1 and pIA.2 that were dissolved in either ethanol (m-p-a) or 1:1 water:ethanol (pIAs 1 and 2). As the compounds pHBAD I, UM-pHBAD I and the analogues had

178 been synthesised in a non-sterile environment, they were all sterile filtered (0.22 μm) prior to cell work. Similarly, because purchased m-p-a was not specified as sterile, this compound was also sterile filtered before it was used for cell work. To check that there was no sign of endotoxin contamination for all compounds, BMDM IL-6 secretion was measured by ELISA after

24-hour incubation with each compound (Fig. 5.4). LPS is a potent inducer of IL-6 and thus measuring IL-6 secretion is useful method to check that compounds are endotoxin free. None of the compounds tested induced significant amounts of IL-6, compared to media-stimulated

BMDMs.

Figure 5.4 Sterile-filtered synthesised compounds and L-Rha do not induce IL-6 secretion by BMDMs. IL-6 secretion from BMDMs stimulated with media, 1 mM of each tested compound or recombinant murine IFNγ with LPS (at 0.5 ng/ml each) for 24 hours. Results (n = 1) are shown as mean ± SD.

Given that the immunological studies with L-Rha had employed concentrations between 0.5 mM and 5 mM, the cytotoxicity of the new synthetic compounds was investigated at concentrations from 0.5 mM up to 10 mM (except for m-p-a, where 5 mM was the maximum concentration at which it was soluble in water-based media). As can be seen in Fig. 5.5, m-p-a,

UM-pHBAD I, pIA.1 and pIA.2 were all significantly more cytotoxic than media stimulated

BMDMs at concentrations of 5 mM or above. pHBAD I was significantly cytotoxic at 10 mM, and pIA.2 was more cytotoxic than media from concentrations of 2 mM and higher. Meanwhile, L- 179

Rha, pIA.3 and pIA.4 were not significantly more cytotoxic than media at any concentration tested (Fig. 5.5). Vehicle controls (ethanol and 1:1 water:ethanol) were also tested at the highest volume added, and the induced levels of LDH secretion were comparable to the media control (not shown).

Figure 5.5 Cytotoxicity of pHBAD I, UM-pHBAD I, m-p-a, L-Rha and synthesised pHBAD I analogues (pIAs) 1-4. Formazan absorbance (492 nm) measurement of lactate dehydrogenase (LDH) release from BMDMs stimulated with media or tested compounds (0.5 – 10 mM). Lysed cells were used to measure 100% cytotoxicity. Representative results (n = 1) shown are mean ± SD and analysed by one-way ANOVA for each compound, with Tukey’s multiple comparisons post test. *p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, compared with media control group.

Effect on Macrophage Polarisation

As a finite amount of pHBAD I had been made, and its synthesis took several months to complete, due to time constraints only the most critical experiments were carried out, starting with BMDM activation.

Insufficient RNA was recovered from BMDMs following 24-hour stimulation with pIA.2 to carry out cDNA synthesis so the effect of this compound could not be analysed by RT-qPCR at this time point. As can be seen in Fig. 5.6, whereas none of the measured compounds induced expression of either the M1-associated Nos2 or M2-associated Arg1, (Fig. 5.6 A), m-p-a and pIA.4 along with L-Rha enhanced IL-4 and IL-13 induced expression of Arg1 to a statistically

180 significant extent (Fig. 5.6 B). Moreover, pHBAD I, pIA.1, pIA.4 and L-Rha reduced IFNγ-induced

Nos2 expression and pHBAD I and pIA.1 inhibited Nos2 expression induced by IFNγ with LPS

(Fig. 5.6 D). All compounds except pIA.3 and pIA.4 also significantly reduced IFNγ and LPS induced nitric oxide (NO) production by BMDMs (Fig. 5.6 F) and all compounds except pIA.3 reduced IFNγ and LPS induced IL-6 secretion (Fig. 5.6 G). Compound pIA.2 also induced a significant amount of NO production when given to BMDMs alone (Fig. 5.6 F).

181

Figure 5.6 pHBAD I, L-Rha and most pHBAD I analogues (pIAs) reduce IFNγ and LPS-induced macrophage responses. BMDMs (n =3) stimulated with media, recombinant murine cytokines IL-4 with IL-13 (40 ng/ml and 20 ng/ml respectively), IFNγ (100 ng/ml), or IFNγ with LPS (0.5 ng/ml each), either alone or combined with 1 mM of each compound for 24 hours. RT-qPCR internal references were Actb and Tbp. A) RT-qPCR standardised to media incubated BMDMs. Results are shown as mean ± SD. B & C) RT-qPCR following IL-4 and IL-13 stimulation, standardised to BMDMs given media (B) or IL- 4 and IL-13 (C) alone. Results are shown as mean (B) or mean ± SD (C) and analysed by ratio-paired t-test (B) or student t-test (C), compared to the BMDMs given IL-4 and IL-13 alone. D & E) RT-qPCR following stimulation with either IFNγ or IFNγ with LPS, standardised to BMDMs given media (D) or IFNγ, with or without LPS, alone (E). Results are shown as mean (D) or mean ± SD (E) and analysed by ratio-paired t-test (D) or student t-test (E), compared to the BMDMs given IFNγ, with or without LPS, alone. F) NO secretion (Griess reagent) following media or IFNγ and LPS stimulation. Results are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. G) IL-6 secretion. Results are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, compared with media control per group.

182

Innate Memory Studies

As the results for macrophage polarisation by pHBAD I and synthesised analogues showed similar inhibition of IFNγ-induced responses to L-Rha, the next step was to investigate their ability to trigger macrophage innate memory. The protocol used for these studies can be seen in

Fig. 5.7, which is the same basic protocol as the optimised protocol used in Chap. 3 except for two changes: the batch of L929 conditioned media used and the concentration of the compounds. In Chap. 3 a more potent batch of L929 conditioned media was used, where 10% supplementation was enough to differentiate bone marrow-derived precursor cells into macrophages. However, with this new batch, 25% had to be used to differentiate the bone marrow cells to a similar degree. Furthermore, given that four of the compounds of interest had shown significant cytotoxicity at 5 mM – the concentration previously used for L-Rha that induced the most consistent innate training profile – the concentration chosen was 2 mM, to avoid the confounding effects of cytotoxicity as much as possible.

Figure 5.7 Timeline for investigating innate memory in BMDMs, using Protocol 3. Isolated bone-marrow cells were differentiated in 25% L929 conditioned media in cell culture dishes. On Day -2, cells were plated in 24-well pates. The BMDMs were stimulated with primary stimuli the following morning (Day -1) for 24 hours. On Day 6, BMDMs were incubated with secondary stimuli for 24 hours.

183

BMDMs were stimulated with either media, 2 mM of each compound, or IFNγ with LPS for 24 hours, before these primary stimuli were washed away. Six days after the first stimulation the

BMDMs were stimulated a second time, either with media, irradiated M. tuberculosis strain

H37Rv (henceforth referred to as H37Rv) or the TLR1/2 ligand PAM3CSK4. Secretion of the cytokines TNFα, IL-6 and IL-10 after 24 hours of the secondary stimulation can be seen in Fig

5.8. Whereas there was no induction of any of the measured cytokines following secondary incubation in media alone (Fig. 5.8 A), stimulation with m-p-a, pIA.2 and IFNγ with LPS (primary stimuli) enhanced H37Rv and PAM3CSK4 induced secretion of TNFα, and reduced H37Rv mediated IL-10 secretion (Fig 5.8 B & C). Moreover, m-p-a also enhanced PAM3CSK4 induced IL-

6 secretion whereas pIA.2 and IFNγ with LPS reduced its stimulation of IL-10. For the other compounds tested, there were no statistically significant changes in cytokine secretion. This was a change from results in Chap.3, where L-Rha induced memory enhanced IL-10 secretion following both H37Rv and PAM3CSK4 stimulation. There was another observable change, namely a detectable level of background TNFα secretion (Fig. 5.8 A). Given that the protocol had not changed apart from the batch of L929 cells used to make the L929 conditioned media – which had to be used at a higher percentage than the previous batch – these changes in cytokine secretion were attributed to this new batch priming the macrophages differently than previously, perhaps altering the innate memory induced by L-Rha.

184

Figure 5.8 m-p-a and pIA.2 enhance TNFα and reduce IL-10 secretion following stimulation with H37Rv. Cytokine secretion from BMDMs incubated with media, each compound (2 mM) or recombinant murine IFNγ with LPS (25 ng/ml and 10 ng/ml respectively) on Day -1 for 24 hours, and stimulated Day 6 for 24 hours with media, H37Rv (150 μg/ml) or PAM3CSK4 (35 ng/ml). A) Media incubation on Day 6, with PAM3CSK4 as positive control. No statistical analysis. B) H37Rv incubation on Day 6. C) PAM3CSK4incubation on Day 6. Results (n = 3) are shown as mean ± SD and analysed by two-way ANOVA, with Tukey’s multiple comparisons post test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, compared with media control group.

Other results in Chap.3 had shown that L-Rha induced memory reduced H37Rv induced expression of Ifnb1 and both H37Rv and PAM3CSK4 induced expression of Nos2. Whilst none of the compounds tested induced memory that significantly altered the expression of Nos2, Arg1 or Ifnb1 alone (Fig. 5.9 A), pHBAD I, L-Rha, pIA.3 and pIA.4 all significantly reduced H37Rv and

185

PAM3CSK4 induced Nos2 expression (Fig. 5.9 B and C) and all compound except m-p-a, pIA.1 and pIA.2 reduced H37Rv mediated Ifnb1 upregulation. Arg1 expression was also measured but as none of the secondary stimuli or the compounds alone caused any enhancement, these results are not included. Overall these results for L-Rha induced memory align with what was observed in Chap.3.

186

Figure 5.9 Altering pHBAD I structure affects its ability to induce innate memory that reduces Nos2 and Ifnb1 expression in response to H37Rv stimulation. RT-qPCR with Actb and Tbp as internal references, on BMDMs (n = 3) incubated with media, each compound (2 mM) or recombinant murine IFNγ and LPS (25 ng/ml and 10 ng/ml respectively) on Day -1 for 24 hours, and stimulated on Day 6 for 24 hours with media, H37Rv (150 μg/ml), LPS (25 ng/ml) or PAM3CSK4 (35 ng/ml). A) BMDMs given media as secondary stimulus, standardised to BMDMs given media throughout (dotted line =1). Results are shown as mean ± SD. B) Nos2 or Ifnb1 fold change following stimulation with H37Rv or PAM3CSK4 on Day 6, standardised to BMDMs given media throughout. Results are shown as mean ± SD and analysed by ratio-paired t- test. C) Nos2 or Ifnb1 fold change following stimulation with H37Rv or PAM3CSK4 on Day 6, standardised to the media (Day -1) control group for each stimulus. Results are shown as mean ± SD and analysed by student t-test. *p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, compared with media control group (dotted line).

187

5.3 Discussion

5.3.1 Choice of Compounds

The premise for the work presented in this chapter was to investigate whether L-Rha was a critical structural element for pHBAD I to exert immunomodulatory properties. Moreover, as L-

Rha had been demonstrated in previous chapters to induce innate memory, another aim was to address whether pHBAD I could do the same. Therefore such properties were assessed for pHBAD I and L-Rha, as well as modified analogues of pHBAD I structure – excluding each component to investigate their role in pHBAD I immunomodulation: the 2-O-methylation of L-

Rha (UM-pHBAD I), the para methyl benzoate group (pIA.1), any para benzyl group (pIA.2), the aromatic ring – whilst maintaining the α-configuration at the anomeric carbon (C-1) of L-Rha –

(pIA.3 and pIA.4), as well as exclusion of L-Rha itself (m-p-a, Fig 5.10).

188

Figure 5.10 Schematic of the pHBAD I structural elements investigated All outlined structural components on pHBAD I were investigated by using synthetic analogues where each component was either modified or excluded. The role of L-Rha and the α-configuration of the anomeric carbon (left side panel), the role of the aromatic component and specifically the para methyl ester (right side panel), the role of the 2-O-methylated L-Rha residue (bottom panel). Schematic was drawn using Chemdraw software.

m-p-a was chosen in place of the pHBAD I precursor methyl p-hydroxybenzoate as a compound of interest. This choice to replace L-Rha with an O-methyl group as opposed to a hydroxyl group

– which was the strategy employed in the previous chapter – was made for two reasons: to increase steric hindrance and to avoid the presence of a potentially reactive hydroxyl group.

Although the mechanism(s) of action are unclear, aromatic compounds are known to be mutagenic/carcinogenic and to drive the formation of reactive oxygen species [274]. In the case of phenolic compounds especially, it has furthermore been proposed that the hydroxyl group allows the formation of a very reactive hydroxyl radical, which has the potential to cause

189 oxidative damage to cell constituents indiscriminately [275]. Furthermore, planar aromatic compounds, including phenols, are common DNA intercalating agents [276,277]. It was therefore considered that the presence of an O-methyl group instead of a hydroxyl group would not only prevent potential reactivity, but also add some steric hindrance, and would thus overall be a more relevant representative for the pHBAD I without L-Rha.

Another choice was to use two variants of L-Rha with the anomeric carbon fixed in the α- configuration: pIA.3 and pIA.4. This choice was made to explore the necessity of this atom arrangement – an O-linked group, as opposed to an S-linked one – for L-Rha and subsequently the pHBAD I to induce its responses. Moreover, prior work in our lab had shown that L-Rha derivatives with a terminal thiol had heightened cytotoxicity [193], and there was an interest for future work to address whether incorporation of a thiol itself would alter cytotoxicity or immunomodulatory properties of a compound of interest. One could argue that for the purposes of this work, an anomeric α-S-methyl group would have been a more relevant compound to use, however methanethiol – which would have been the compound to be glycosylated – is extremely toxic, and therefore ethanethiol was used a safe alternative.

5.3.2 Does Hydrophobicity Affect the Immunomodulatory Properties of pHBAD I?

Prior work investigating the structure-function relationships of mycobacterial PGLs and their elicited immune-modifying properties had shown that the phenolic glycan was an essential component for the immune-modifying properties to be induced, but that the presence of the

PDIM lipid core, or another hydrophobic substitute, enhanced their potency [190,202,270].

Furthermore, investigation of the methylation of for instance pHBAD II revealed that replacing

O-methyl groups with hydroxyl groups – and correspondingly reducing the hydrophobicity – also attenuated its immunomodulatory properties [267]. As prior work had identified that increased

190 hydrophobicity could enhance the immune-modifying properties of pHBADs, this was considered when analysing the structure-function relationships of pHBAD I and its analogues.

As mentioned in the previous section, aromatic compounds are known to be cytotoxic, which can also be seen by the results presented in this chapter. A striking difference between all compounds which contain an aromatic ring (m-p-a, pHBAD I, UM-pHBAD I, pIA.1 and pIA.2) as opposed to those variants which did not (L-Rha, pIA.3 and pIA.4) was that the former compounds were all cytotoxic at 5-10 mM concentration, whereas the latter compounds were not (Fig. 5.5). Of the compounds tested, pIA.2 was by far the most cytotoxic – showing a mean of 37- and 52% cytotoxicity (BMDM) at concentrations 1- and 2 mM respectively, as opposed to the corresponding 5- and 25% measured for pIA.1 and 6- and 23% measured for UM-pHBAD I.

Moreover, this cytotoxicity was also apparent following BMDM activation studies: pIA.2 was the only compound to induce NO production on its own, it abrogated IL-6 secretion and not enough

RNA was isolated from stimulated BMDMs to carry out cDNA synthesis – all indicative of cell death. Interestingly, pIA.1 also markedly reduced IL-6 secretion compared to the other compounds tested, which could be an indication that this compound was inducing a higher degree of cytotoxicity than pHBAD I that was not detected by LDH release (Fig. 5.6 G).

The question of why the presence of a mono-substituted benzene ring (pIA.2) gave such a high degree of cytotoxicity, which in turn was shown to be severely curtailed by the presence of a para structural element, could be explained by either the damage caused by the intercalating properties of a free planar aromatic ring, the accompanying increase of hydrophobicity, or a combination of both. Increased hydrophobicity aids the mobility of aromatic compounds within a cell, allows interaction with non-polar constituents such as poly-unsaturated fatty acids and overall has been shown to be a key factor determining the toxicity of aromatic compounds

[275]. However, pIA.2 was not the most hydrophobic aromatic compound used in this study, m- p-a was (Fig. 5.11), and this compound did not induce the same degree of cytotoxicity as pIA.2.

This would suggest that the planar structural aspect of pIA.2 was the major physicochemical

191 contributor that induced cell death. Whilst the degree of cytotoxicity differed, m-p-a and pIA.2 on the other hand did induce similar innate memory.

Figure 5.11 Schematic of all compounds of interest in order of increasing hydrophobicity. m-p-a, methyl-p-anisate. Schematic done with Chemdraw software

Of all the compounds tested (Fig. 5.11), m-p-a and pIA.2 were the only two to induce an innate memory response that enhanced pro-inflammatory cytokine secretion, and repressed anti- inflammatory IL-10 following both H37Rv and PAM3CSK4 stimulation – akin to IFNγ and LPS activated BMDMs (Fig. 5.8 B and C). Moreover, these were the only two compounds (apart from pIA.1) whose memory response did not reduce Nos2 and Ifnb1 expression (Fig. 5.9 C). In the case of pIA.2 this pro-inflammatory phenotype could be attributed to its cytotoxicity, as the release of danger-associated molecular patterns (DAMPs), are reported to induce pro- inflammatory responses and subsequent macrophage M1 activation [23,278]. However this does not explain why m-p-a was inducing the same memory response. This observation lead to the question whether the increase in hydrophobicity may also be altering the specific/non- specific interaction profile of these compounds, which could also be contributing to the overall shift to an M1-associated memory response.

That increased hydrophobicity caused a more M1-like profile could be further supported by a related observation: there was a minor, yet consistent trend throughout the results for innate memory induced by UM-pHBAD I, pIA.1 and pIA.2, that showed a correlation between increasing hydrophobicity and increasing secretion of TNFα, IL-6, Nos2 and Ifnb1 combined with 192 reduced IL-10 in response to both H37Rv and PAM3CSK4 (Fig. 5.8 B and C). However, this trend could also be attributed to cytotoxicity; although there was essentially no difference in LDH release from BMDMs stimulated with UM-pHBAD I or pIA.1, pIA.1 – as mentioned previously – did markedly reduce IL-6 secretion compared to UM-pHBAD I, which could be a sign of toxicity.

However, m-p-a again did not reduce IL-6 secretion to the same extent, leaving increased hydrophobicity as a common denominator. Therefore, based on these results, one can argue that the most hydrophobic aromatic compounds were able to induce M1 type innate memory responses via an unidentified mechanism(s). As this memory response was however not shown for pHBAD I this route of enquiry was not pursued further.

Going back to the role of L-Rha, when compared to pHBAD I and pHBAD II, all three compounds repressed aCD3 induced IFNγ to a similar extent (Fig. 5.2 A). Moreover, when UM-pHBAD I was compared to the less polar pHABD I, there were no marked differences in immune responses elicited. However, pHBAD I did appear to more potently reduce IFNγ and LPS induced Nos2 expression, NO production and IL-6 secretion during 24-hour BMDM activation (Fig. 5.2 C and

Fig. 5.6 D and E), demonstrating that pHBAD I more potently repressed M1 macrophage activation. This increase in efficacy could be attributed to the overall increase in hydrophobicity, furnished by the aromatic component. However, these experiments also revealed that m-p-a alone induced these same IFNγ interfering responses. Furthermore, both m-p-a and L-Rha enhanced IL-4 and IL-13 induced expression of Arg1 (Fig. 5.6 B) – an effect also seen for pHBAD I stimulated BMDMs in two out of three mice. These results would therefore suggest that both the aromatic component and L-Rha contributed to the immunomodulatory effect that pHBAD I had on M1 macrophage activation.

Although no prior publication has specifically investigated the role of the para methyl benzylester, a para phenyl compound with a hydrocarbon chain was found not to affect

PAM3CSK4 induced IL-6 or NO secretion, whilst addition of the M. lepre trisaccharide (forming an analogue of PGL-1) did [190,270], supporting the reasoning that it is unlikely that the

193 aromatic component of PGLs contributes to their immunosuppression. Furthermore, when L-

Rha and pHBAD I were compared to D-glucose, this latter sugar did not affect M1 macrophage activation, indicating that there must be some degree of structural specificity for L-Rha and pHBAD I to induce their similar effects (Fig. 5.2 B). Therefore, the discovery that m-p-a also attenuated M1 macrophage activation was very surprising. Because L-Rha and m-p-a were structurally distinct molecules with hugely dissimilar polarities from one another, it was highly unlikely that they induced their effects through a similar mechanism. The question arose therefore whether the aromatic component was conveying immunomodulation to pHBAD I, or could m-p-a be coincidentally attenuating M1 macrophage activation via an unrelated mechanism? Unfortunately, due to time constraints, we were unable to fully explore this question within the scope of this thesis.

However, the same cannot be said for the innate memory induced by these compounds. Having found that pHBAD I, as hypothesised, did induce innate immune memory of similar type and intensity as L-Rha (Fig. 5.9 C), these responses were drastically different from those induced by m-p-a. Whilst it is not clear why m-p-a induced responses that impede M1 activation initially yet promote M1-type memory, at least it can be said that the addition of 2-O-methylated L-Rha to m-p-a (forming pHBAD I) blocks its ability to induce a pro-inflammatory memory response and instead drives concurrent reduction of Nos2 and Ifnb1 expression (Fig. 5.9 C). That L-Rha was the driver of this response was further supported by its close structural analogues pIA.3 and pIA.4. Neither compound was more potent than L-Rha when looking at 24 hour BMDM activation – in fact pIA.3 was the only compound tested that did not reduce NO production or

IL-6 secretion. However, pIA.4 induced the most potent memory response with regard to suppressed stimulation of Nos2 expression (Fig. 5.9 B) and even induced a marked increase in

IL-10 secretion (Fig. 5.8 B and C). This suggested that the addition of a small hydrophobic ethyl group may have enhanced the potency of L-Rha. Overall, L-Rha and pHBAD I triggered similar innate memory – both qualitatively and quantitatively – that was different from that induced by

194 m-p-a. This would indicate L-Rha is a critical structural element underlying the capacity of pHBAD I to promote innate memory responses.

5.3.3 Summary and Conclusions

This work addressed the concept that systematically removing structural components one at a time to explore the structure-function relationship of pHBAD I would provide key insights into the immunomodulatory properties of this molecule. However, its translation in practice turned out to be more complicated than anticipated. Essentially, for some of the compounds tested, the removal of structural features altered their physicochemical properties to a degree where their response profile was drastically altered. As highlighted here, the removal of the para methyl ester of the aromatic ring (pIA.2) led to both increased cytotoxicity and an altered innate memory response that was very different from that induced by pHBAD I; changes that could be attributed to increased hydrophobicity and/or possibly the change to a freer, potentially more intercalating aromatic structure. Similarly, pIA.1 cannot be used to comment on the role of the methyl ester, because its responses are akin to both pHBAD I and pIA.2, showing a skewing towards M1-type innate memory and reduced IL-6 secretion compared to pHBAD I.

Furthermore, pIA.4 was the compound that induced the most potent innate memory response, comparable to pHBAD I and L-Rha. By contrast, pIA.3, which was structurally more similar to L-

Rha than pIA.4, did not show enhanced efficacy. Whilst the innate memory responses were similar, pIA.3 was the only compound that did not reduce either NO and IL-6 secretion following

IFNγ and LPS activation, and it did not reduce IFNγ induced Nos2 – which both L-Rha and pIA.4 did (Fig. 5.6 D). As these three compounds are structurally related, with similarly reactive functional groups, one would expect them to have similar physicochemical properties and thus induce similar responses. However, for reasons unknown, in this case, the addition of a 1-O-

195 methyl group appears to have reduced the potency of L-Rha, demonstrating the unpredictable nature of an investigation of this kind.

Another unexpected complication lay with the discrepancy between responses seen for IFNγ- induced macrophage activation and innate memory induction. Throughout this section there have been examples of compounds inducing similar memory responses but not affecting acute macrophage activation the same way, or vice versa, with m-p-a being the clearest example. m- p-a and pIA.2 induced similar memory responses (Fig. 5.12), which were different to those induced by pHBAD I; yet m-p-a and L-Rha inhibited IFNγ-induced macrophage activation in the same way as pHBAD I did. This dichotomy in response profile for m-p-a complicates matters, as at this point it is not known whether the responses are linked or caused by two distinct mechanisms. If the sequence of events by which m-p-a or L-Rha prevented IFNγ-induced macrophage activation, and enhanced Arg1, were the same events that induced their respective memory responses, it would intimate that L-Rha is the critical component for pHBAD I to induce its responses, and the hydrophobicity of the aromatic component was enhancing its potency.

However, if the responses were not linked, then both components could be synergistically causing the acute macrophage activation profile of pHBAD I. Overall, however, it can be concluded that for the measured innate memory induced by pHBAD I, L-Rha appears to be a critical structural element, confirming the hypothesis that L-Rha does furnish mycobacterial pHBADs – and possibly PGLs – with immunomodulatory properties.

196

Figure 5.12 Venn diagram summarising immunomodulatory properties demonstrated by pHBAD I, L-Rha and related synthetic analogues. Structures of m-p-a, L-Rha and pHBAD I are coloured in black to highlight their structure-function relationship. All other structures coloured in grey. Left circle: inhibition of IFNγ-induced macrophage activation, by reduced upregulation of Nos2, NO and IL-6. Right circle: stimulation with compound 6 days prior to incubation with irradiated M. tuberculosis reduces induction of Nos2 and Ifnb1. Bottom circle: stimulation with compound 6 days prior to incubation with irradiated M. tuberculosis enhances induction of TNFα and IL-6. Compounds pIA.2 and potentially pIA.1 are believed to induce cytotoxicity.

197

Chapter 6. General Discussion

Mycobacterium tuberculosis is the causative agent of tuberculosis (TB), and is arguably the deadliest pathogen in human history, which remains a major issue to this day. In 2017, 10 million people were newly infected with the disease and 1.6 million died [183]. There are a number of factors that contribute to why this disease continues to be a global health threat, including the absence of an efficient vaccine, evolving antibiotic resistance and the immunomodulatory properties of the bacteria. In this final chapter, the newfound role of L- rhamnose (L-Rha) will be considered in the context of each of these factors.

The majority of M. tuberculosis infected individuals develop latent TB, where pulmonary infection culminates in the formation of granulomas (amalgamations of both adaptive and innate immune cells) where the bacteria are contained [279]. Latent TB is asymptomatic and instead is characterised by a lifelong standoff between immune-mediated killing and bacterial replication [279]. It is estimated that roughly one third of the World’s population is infected with latent TB, and in 5-10% of cases, an active and contagious form of disease evolves [279], where the patient experiences fever and weight loss, and can spread the bacterium through coughing [280]. Moreover, in immunocompromised individuals, as well as young children and the elderly, the infection can spread uncontrollably from the lungs (disseminated TB), causing more severe forms of infection, including meningitis [281].

A schematic of the initial stages of infection, leading to granuloma formation can be seen in Fig.

6.1. Upon inhalation, alveolar macrophages are believed to be the initial hosts of these intracellular bacteria [282]. Bacteria taken up by phagocytosis are usually enclosed in an early phagosome, which then matures into a bactericidal phagolysosome. Phagosome maturation involves acidification, accumulation of degrading enzymes, antimicrobial peptides, as well as production of reactive nitrogen species through iNOS [283]. However, M. tuberculosis interferes with phagosome maturation, ultimately escaping the bactericidal mechanisms and end up

198 residing in the cytosol, where they can replicate, protected by their host cell [283,284]. The infected alveolar macrophages then migrate out of the alveolar space into the surrounding interstitium, where they have been observed to form small aggregates, which could be start of granuloma formation [282] (Fig. 6.1). Upon the death of the infected alveolar macrophages, M. tuberculosis are released and taken up by new host cells, prominently circulating monocytes/macrophages and neutrophils that are recruited to the site by chemokine secretion

[284]. Following cycles of immune cell recruitment and release of bacteria from dead host cells, the infection is eventually contained within granulomas to prevent the bacteria spreading further through the body. Within these granulomas the cycle continues in a stalemate, where the immune system is seldom able to eradicate the bacteria completely.

Figure 6.1 Schematic of latent tuberculosis pathogenesis, focusing on the primary infection stage. Inhaled M. tuberculosis are phagocytosed by alveolar macrophages (1). The bacteria escape the phagolysosome, enter the cytosol and replicate inside the host macrophages (2). These infected alveolar macrophages migrate into the interstitium (3). Upon the release of bacteria from dead host cells, recruited monocytes/macrophages and neutrophils (4) become new hosts for the bacteria. Eventually a granuloma is formed (5), where both adaptive (T- and B-cells) and innate (macrophages, neutrophils, foam cells and dendritic cells) immune cells work together to contain the infection.

199

To control TB infection, induction of a strong IFNγ response is vital [279,285], because of its ability to induce macrophage bactericidal responses, including iNOS induced NO production

[23,134,184,286]. However, M. tuberculosis possesses means of modifying macrophage activation to its advantage, including the presentation of immunomodulatory PGLs [188,202].

The research presented herein has shown that the monosaccharide L-Rha appears to contribute to these immunomodulatory properties, by both preventing IFNγ secretion, as well as IFNγ- mediated macrophage activation. Furthermore, L-Rha endows pHBAD I with the ability to induce macrophage innate memory, which alters responses elicited by subsequent recognition of M. tuberculosis. The expression of L-Rha and L-Rha-rich PGLs and pHBADs therefore appears to aid M. tuberculosis pathogenesis by evasion of host immune responses.

Having identified that L-Rha induces innate memory that impedes macrophage induction of

Nos2 and Ifnb1 in response to killed M. tuberculosis, strain H37Rv, one of the questions remaining is what mechanism is involved. As discussed in the previous chapter, previous work has found that immunomodulation induced by isolated PGL-tb, PGL-1 and mycoside B is dependent upon complement receptor 3 (CR3) and that the targeted responses are specifically

TRIF-mediated [202]. Furthermore, PGL-1 was shown to reduce protein levels. Given that the innate memory responses demonstrated for L-Rha and pHBAD I affect TRIF-dependent macrophage Nos2 and Ifnb1, future work should address whether L-Rha in particular can affect

TRIF expression in a similar way, and to what extent this may be maintained over a prolonged period of time, such as the case in our innate memory protocol.

Having discovered that L-Rha has these properties another important question is: how can we use this information beneficially? A growing problem with trying to combat TB infection is that multidrug resistant (MDR) and extensively drug resistant (XDR) strains of M. tuberculosis are on the rise [287]. In 2017, it was estimated that 558 000 people developed TB that was resistant to rifampicin (RR-TB) – the most effective first-line drug – and of these, 82% had multidrug-

200 resistant TB (MDR-TB) [183]. Because of this growing crisis, targets for new antibiotic research is critical. Together with prior research regarding the immunomodulatory properties of PGLs, this newly identified role of L-Rha suggests that targeting PGL biosynthesis could be a good target to enhance macrophage killing capacity. A recent study showed that recombinant BCG strains that expressed no PGLs induced higher levels of macrophage NO secretion than strains that expressed PGLs from BCG (mycoside B) or M. tuberculosis (PGL-tb) [202], indicating that this approach could prove useful in the short term.

For long-term eradication of TB, however, the best strategy is vaccination. The current vaccine against TB is BCG, which is a live attenuated strain of M. bovis and was developed in the 1920s

[284]. This vaccine is able to protect children from severe disseminated forms of TB, but does not protect adults from becoming infected, nor prevent activation of latent TB [284]. Therefore new vaccine candidates are desperately needed and a number are at advanced stages of clinical testing [288]. One suggested approach is genetic modification of the current BCG vaccine to more efficiently induce Th1 adaptive immune memory [289,290] – the primary immune cells responsible for IFNγ secretion. To differentiate activated CD4+ T helper cells towards a Th1 phenotype, secretion of the cytokine IL-12p70 by antigen presenting cells has been shown to be critical [291]. Furthermore, it has been demonstrated that autocrine IFNβ stimulation enhances the amount of IL-12p70 produced by macrophages [292,293]. For instance, in IFN alpha/beta receptor (IFNAR) knock out mice, the IL-12p40 secretion and Il12b mRNA expression was reduced compared to WT mice [292]. Although a direct effect upon IL-12p70 has not been identified herein, the results presented have shown that L-Rha and pHBAD I reduce the macrophage induction of Ifnb1 following killed M. tuberculosis stimulation. As such, L-Rha may be negatively affecting Th1 differentiation; a potential immunomodulatory effect that remains to be investigated. This hypothesis is further supported by the way L-Rha and pHBADs I and II significantly reduced splenocyte secretion of IFNγ following anti-CD3 stimulation – which is primarily targeted towards T cell activation. This latter result could indicate further that L-Rha

201 may attenuate Th1 activation. If this is the case, for the development of a more efficient BCG vaccine, a strategy could be to modify the bacterium to eliminate mycoside B and pHBAD I, with the aim to induce a greater Th1 response.

Another factor regarding BCG vaccination is that it is not only used to induce long-term adaptive immune memory, but through innate memory induction has been demonstrated to enhance pro-inflammatory responses to both M. tuberculosis as well as non-related pathogens, such as the extracellular bacterium Staphylococcus aureus and the fungus Candida albicans [108]. This non-specific innate memory induction has been recently proposed to be critical for the efficacy of the BCG vaccine, by enhancing the innate immune cell bactericidal effector functions against

TB [294,295]. Therefore, the discovery that L-Rha and pHBAD I induce an innate memory phenotype that potentially undermines bactericidal capabilities of macrophages, by inhibiting

Nos2 induction, provides another argument in favour of modifying the current BCG vaccine. If these sugars were absent, the innate memory induced by BCG could perhaps induce an even greater bactericidal innate memory response.

Apart from using the information obtained regarding L-Rha induced immunomodulation to improve strategies to combat TB infection, another route is to use the immunomodulation beneficially in other settings, such as the making of biomaterials. To reiterate from Chap. 1, biomaterials are used for tissue engineered scaffolds; man-made devices introduced in a patient to replace or support a damaged biological structure. Accumulating evidence proposes that M2 macrophage activation is critical for clinical success, preventing harmful long-lasting inflammation and promoting tissue remodelling and healing [137–139]. In Chap. 3 it was found that L-Rha skewed macrophage responses towards an M2-like profile, and enhanced macrophage secretion of IL-10 following secondary challenge. Although this latter property could not be demonstrated in Chap. 5, which is thought to be due to the change brought on by a new batch of L929 conditioned media, nevertheless these results suggest that L-Rha could induce an anti-inflammatory macrophage innate memory. As such, incorporation of this sugar

202 to make novel biomaterials is a feasible strategy to use the immunomodulatory properties of L-

Rha to enhance clinical efficacy of bioengineered scaffolds.

Conclusion

This work explored the potential of the monosaccharide, L-rhamnose to exert immunomodulatory properties and what that may mean for M. tuberculosis pathogenesis. This exploration has led to the novel discovery that L-rhamnose indeed does have immunomodulatory properties, and through the use of synthesised novel analogues has been confirmed to convey at least its innate memory inducing properties to pHBAD I. With regard to

TB pathogenesis, more research remains to be carried out regarding how this impact on macrophages may for instance be inhibiting Th1 cell activation and macrophage bactericidal properties. However, this work has provided a solid foundation upon which further research in the field can be built. Lastly, this role for L-rhamnose, a monosaccharide ubiquitously expressed amongst numerous pathogenic fungal and bacterial species, is an important starting point for the discovery of more immune modifying compounds, and their potential roles in pathogeneses.

203

204

References

[1] R.D. Astronomo, D.R. Burton, Carbohydrate vaccines: developing sweet solutions to sticky situations?, Nat Rev Drug Discov. 9 (2010) 308–324. doi:10.1038/nrd3012.

[2] G.Y. Ishioka, A.G. Lamont, D. Thomson, N. Bulbow, F.C. Gaeta, A. Sette, H.M. Grey, MHC interaction and T cell recognition of carbohydrates and glycopeptides, J Immunol. 148 (1992) 2446–2451.

[3] E.T. Clarke, N.A. Williams, J. Findlow, R. Borrow, R.S. Heyderman, A. Finn, Polysaccharide-specific memory B cells generated by conjugate vaccines in humans conform to the CD27+IgG+ isotype- switched memory B Cell phenotype and require contact-dependent signals from bystander T cells activated by bacterial proteins to differentiat, J Immunol. 191 (2013) 6071–6083. doi:10.4049/jimmunol.1203254.

[4] K.P. Murphy, Janeway’s Immunobiology, 8th ed., Garland Science, Taylor & Francis Group, LLC, United States of America, 2011.

[5] M. Mogemark, T.P. Cirrito, P. Sjolin, E.R. Unanue, J. Kihlberg, Influence of saccharide size on the cellular immune response to glycopeptides, Org Biomol Chem. 1 (2003) 2063–2069.

[6] B.A. Cobb, Q. Wang, A.O. Tzianabos, D.L. Kasper, Polysaccharide processing and presentation by the MHCII pathway, Cell. 117 (2004) 677–687. doi:10.016/j.cell.2004.05.001.

[7] J. Duan, F.Y. Avci, D.L. Kasper, Microbial carbohydrate depolymerization by antigen-presenting cells: deamination prior to presentation by the MHCII pathway, Proc Natl Acad Sci U S A. 105 (2008) 5183–5188. doi:10.1073/pnas.0800974105.

[8] C.D. Velez, C.J. Lewis, D.L. Kasper, B.A. Cobb, Type I Streptococcus pneumoniae carbohydrate utilizes a nitric oxide and MHC II-dependent pathway for antigen presentation, Immunology. 127 (2009) 73–82. doi:10.1111/j.1365-2567.2008.02924.x.

[9] V. Verez-Bencomo, V. Fernandez-Santana, E. Hardy, M.E. Toledo, M.C. Rodriguez, L. Heynngnezz, A. Rodriguez, A. Baly, L. Herrera, M. Izquierdo, A. Villar, Y. Valdes, K. Cosme, M.L. Deler, M. Montane, E. Garcia, A. Ramos, A. Aguilar, E. Medina, G. Torano, I. Sosa, I. Hernandez, R. Martinez, A. Muzachio, A. Carmenates, L. Costa, F. Cardoso, C. Campa, M. Diaz, R. Roy, A synthetic conjugate polysaccharide vaccine against Haemophilus influenzae type b, Science (80-. ). 305 (2004) 522–525. doi:10.1126/science.1095209.

[10] E.C. Carroll, L. Jin, A. Mori, N. Muñoz-Wolf, E. Oleszycka, H.B.T. Moran, S. Mansouri, C.P. McEntee, E. Lambe, E.M. Agger, P. Andersen, C. Cunningham, P. Hertzog, K.A. Fitzgerald, A.G. Bowie, E.C. Lavelle, The Vaccine Adjuvant Chitosan Promotes Cellular Immunity via DNA Sensor cGAS-STING- Dependent Induction of Type I Interferons., Immunity. 44 (2016) 597–608. doi:10.1016/j.immuni.2016.02.004.

[11] M. Liu, F. Luo, C. Ding, S. Albeituni, X. Hu, Y. Ma, Y. Cai, L. McNally, M.A. Sanders, D. Jain, G. Kloecker, M. Bousamra, H. Zhang, R.M. Higashi, A.N. Lane, T.W.-M. Fan, J. Yan, Dectin-1 Activation by a Natural Product β-Glucan Converts Immunosuppressive Macrophages into an M1-like Phenotype, J. Immunol. 195 (2015) 5055–5065. doi:10.4049/jimmunol.1501158.

[12] C. Büll, M.H. den Brok, G.J. Adema, Sweet escape: Sialic acids in tumor immune evasion, Biochim. Biophys. Acta - Rev. Cancer. 1846 (2014) 238–246. doi:10.1016/j.bbcan.2014.07.005.

[13] A.K. Adak, B.Y. Li, L.D. Huang, T.W. Lin, T.C. Chang, K.C. Hwang, C.C. Lin, Fabrication of antibody microarrays by light-induced covalent and oriented immobilization, ACS Appl Mater Interfaces. 6 (2014) 10452–10460. doi:10.1021/am502011r.

[14] Y. van Kooyk, G.A. Rabinovich, Protein-glycan interactions in the control of innate and adaptive immune responses, Nat Immunol. 9 (2008) 593–601. doi:10.1038/ni.f.203.

205

[15] R.L. Schnaar, Glycans and glycan-binding proteins in immune regulation: A concise introduction to glycobiology for the allergist, J Allergy Clin Immunol. 135 (2015) 609–615. doi:10.1016/j.jaci.2014.10.057.

[16] K. Marino, J. Bones, J.J. Kattla, P.M. Rudd, A systematic approach to protein glycosylation analysis: a path through the maze, Nat Chem Biol. 6 (2010) 713–723. doi:10.1038/nchembio.437.

[17] J.D. Marth, P.K. Grewal, Mammalian glycosylation in immunity, Nat Rev Immunol. 8 (2008) 874– 887. doi:10.1038/nri2417.

[18] I.D. Mascanfroni, J.P. Cerliani, S. Dergan-Dylon, D.O. Croci, J.M. Ilarregui, G.A. Rabinovich, Endogenous lectins shape the function of dendritic cells and tailor adaptive immunity: mechanisms and biomedical applications, Int Immunopharmacol. 11 (2011) 833–841. doi:10.1016/j.intimp.2011.01.021.

[19] J.M. Ilarregui, D.O. Croci, G.A. Bianco, M.A. Toscano, M. Salatino, M.E. Vermeulen, J.R. Geffner, G.A. Rabinovich, Tolerogenic signals delivered by dendritic cells to T cells through a galectin-1- driven immunoregulatory circuit involving interleukin 27 and interleukin 10, Nat Immunol. 10 (2009) 981–991. doi:10.1038/ni.1772.

[20] A. Sica, M. Erreni, P. Allavena, C. Porta, Macrophage polarization in pathology, Cell. Mol. Life Sci. 72 (2015) 4111–4126. doi:10.1007/s00018-015-1995-y.

[21] M.L.E. Lundahl, E.M. Scanlan, E.C. Lavelle, Therapeutic potential of carbohydrates as regulators of macrophage activation, Biochem. Pharmacol. 146 (2017) 23–41. doi:10.1016/j.bcp.2017.09.003.

[22] C.J. Ferrante, S.J. Leibovich, Regulation of Macrophage Polarization and Wound Healing, in: Adv Wound Care (New Rochelle), Department of Cell Biology and Molecular Medicine and The Cardiovascular Research Institute, New Jersey Medical School, University of Medicine and Dentistry of New Jersey , Newark, New Jersey., 2012: pp. 10–16. doi:10.1089/wound.2011.0307.

[23] J. Bystrom, I. Evans, J. Newson, M. Stables, I. Toor, N. van Rooijen, M. Crawford, P. Colville-Nash, S. Farrow, D.W. Gilroy, Resolution-phase macrophages possess a unique inflammatory phenotype that is controlled by cAMP, Blood. 112 (2008) 4117–4127. doi:10.1182/blood-2007-12-129767.

[24] S. Gordon, F.O. Martinez, Alternative activation of macrophages: mechanism and functions, Immunity. 32 (2010) 593–604. doi:10.1016/j.immuni.2010.05.007.

[25] P.J. Murray, T.A. Wynn, Protective and pathogenic functions of macrophage subsets, Nat. Rev. Immunol. 11 (2011) 723–737. doi:10.1038/nri3073.

[26] A. Mantovani, A. Sica, S. Sozzani, P. Allavena, A. Vecchi, M. Locati, The chemokine system in diverse forms of macrophage activation and polarization, Trends Immunol. 25 (2004) 677–686. doi:10.1016/j.it.2004.09.015.

[27] S.L.M. van Loon, A.I.P. M. Smits, A. Driessen-Mol, F.P. T. Baaijens, C. V. C. Boute, The Immune Response in In Situ Tissue Engineering of Aortic Heart Valves, in: Calcif. Aortic Valve Dis., InTech, 2013. doi:10.5772/54354.

[28] G.D. Brown, P.R. Crocker, Lectin Receptors Expressed on Myeloid Cells, Microbiol. Spectr. 4 (2016). doi:10.1128/microbiolspec.MCHD-0036-2016.

[29] D.I. Gabrilovich, S. Ostrand-Rosenberg, V. Bronte, Coordinated regulation of myeloid cells by tumours, Nat. Rev. Immunol. 12 (2012) 253–268. doi:10.1038/nri3175.

[30] M. Orecchioni, Y. Ghosheh, A.B. Pramod, K. Ley, Macrophage Polarization: Different Gene Signatures in M1(LPS+) vs. Classically and M2(LPS–) vs. Alternatively Activated Macrophages, Front. Immunol. 10 (2019) 1084. doi:10.3389/fimmu.2019.01084.

[31] M.V.S. Rajaram, B. Ni, C.E. Dodd, L.S. Schlesinger, Macrophage immunoregulatory pathways in tuberculosis, Semin. Immunol. 26 (2014) 471–485. doi:10.1016/j.smim.2014.09.010.

206

[32] K.A. Jablonski, S.A. Amici, L.M. Webb, J. de D. Ruiz-Rosado, P.G. Popovich, S. Partida-Sanchez, M. Guerau-de-Arellano, Novel Markers to Delineate Murine M1 and M2 Macrophages., PLoS One. 10 (2015) e0145342. doi:10.1371/journal.pone.0145342.

[33] O.R. Colegio, N.-Q. Chu, A.L. Szabo, T. Chu, A.M. Rhebergen, V. Jairam, N. Cyrus, C.E. Brokowski, S.C. Eisenbarth, G.M. Phillips, G.W. Cline, A.J. Phillips, R. Medzhitov, Functional polarization of tumour-associated macrophages by tumour-derived lactic acid., Nature. 513 (2014) 559–563. doi:10.1038/nature13490.

[34] L.P. Erwig, N.A.R. Gow, Interactions of fungal pathogens with phagocytes, Nat. Rev. Microbiol. 14 (2016) 163–176. doi:10.1038/nrmicro.2015.21.

[35] B. Dey, W.R. Bishai, Crosstalk between Mycobacterium tuberculosis and the host cell., Semin. Immunol. 26 (2014) 486–496. doi:10.1016/j.smim.2014.09.002.

[36] G.P. Bhide, K.J. Colley, Sialylation of N-glycans: mechanism, cellular compartmentalization and function., Histochem. Cell Biol. 147 (2017) 149–174. doi:10.1007/s00418-016-1520-x.

[37] B.S. Bochner, N. Zimmermann, Role of siglecs and related glycan-binding proteins in immune responses and immunoregulation., J. Allergy Clin. Immunol. 135 (2015) 598–608. doi:10.1016/j.jaci.2014.11.031.

[38] I. Fraschilla, S. Pillai, Viewing Siglecs through the lens of tumor immunology, Immunol. Rev. 276 (2017) 178–191. doi:10.1111/imr.12526.

[39] M.S. Macauley, P.R. Crocker, J.C. Paulson, Siglec-mediated regulation of immune cell function in disease, Nat. Rev. Immunol. 14 (2014) 653–666. doi:10.1038/nri3737.

[40] G.A. Rabinovich, M.A. Toscano, Turning “sweet” on immunity: galectin-glycan interactions in immune tolerance and inflammation, Nat Rev Immunol. 9 (2009) 338–352. doi:10.1038/nri2536.

[41] C. Büll, M.A. Stoel, M.H. den Brok, G.J. Adema, Sialic acids sweeten a tumor’s life., Cancer Res. 74 (2014) 3199–204. doi:10.1158/0008-5472.CAN-14-0728.

[42] A.N. Samraj, H. Läubli, N. Varki, A. Varki, Involvement of a Non-Human Sialic Acid in Human Cancer, Front. Oncol. 4 (2014) 33. doi:10.3389/fonc.2014.00033.

[43] J.E. Stencel-Baerenwald, K. Reiss, D.M. Reiter, T. Stehle, T.S. Dermody, The sweet spot: defining virus-sialic acid interactions., Nat. Rev. Microbiol. 12 (2014) 739–49. doi:10.1038/nrmicro3346.

[44] S. Roy, C. Mandal, S. Sun, L. Mao, X. Chu, H. Deng, M. Tremblay, Leishmania donovani Utilize Sialic Acids for Binding and Phagocytosis in the Macrophages through Selective Utilization of Siglecs and Impair the Innate Immune Arm, PLoS Negl. Trop. Dis. 10 (2016) e0004904. doi:10.1371/journal.pntd.0004904.

[45] Y. Wu, C. Lan, D. Ren, G.-Y. Chen, Induction of Siglec-1 by Endotoxin Tolerance Suppresses the Innate Immune Response by Promoting TGF-β1 Production, J. Biol. Chem. 291 (2016) 12370– 12382. doi:10.1074/jbc.M116.721258.

[46] R. Takamiya, K. Ohtsubo, S. Takamatsu, N. Taniguchi, T. Angata, The interaction between Siglec-15 and tumor-associated sialyl-Tn antigen enhances TGF-β secretion from monocytes/macrophages through the DAP12-Syk pathway., Glycobiology. 23 (2013) 178–87. doi:10.1093/glycob/cws139.

[47] S. Spence, M.K. Greene, F. Fay, E. Hams, S.P. Saunders, U. Hamid, M. Fitzgerald, J. Beck, B.K. Bains, P. Smyth, E. Themistou, D.M. Small, D. Schmid, C.M. O’Kane, D.C. Fitzgerald, S.M. Abdelghany, J.A. Johnston, P.G. Fallon, J.F. Burrows, D.F. McAuley, A. Kissenpfennig, C.J. Scott, Targeting Siglecs with a sialic acid-decorated nanoparticle abrogates inflammation, Sci Transl Med. 7 (2015) 303ra140. doi:10.1126/scitranslmed.aab3459.

[48] A.G. Barrientos, J.M. de la Fuente, T.C. Rojas, A. Fernández, S. Penadés, Gold glyconanoparticles: synthetic polyvalent ligands mimicking glycocalyx-like surfaces as tools for glycobiological studies., Chemistry. 9 (2003) 1909–21. doi:10.1002/chem.200204544.

207

[49] R. Cao-Milan, L.M. Liz-Marzan, Gold nanoparticle conjugates: recent advances toward clinical applications, Expert Opin Drug Deliv. 11 (2014) 741–752. doi:10.1517/17425247.2014.891582.

[50] F. Chiodo, P.M. Enriquez-Navas, J. Angulo, M. Marradi, S. Penades, Assembling different antennas of the gp120 high mannose-type glycans on gold nanoparticles provides superior binding to the anti-HIV antibody 2G12 than the individual antennas, Carbohydr Res. 405 (2015) 102–109. doi:10.1016/j.carres.2014.07.012.

[51] N.A.R. Gow, F.L. van de Veerdonk, A.J.P. Brown, M.G. Netea, Candida albicans morphogenesis and host defence: discriminating invasion from colonization, Nat. Rev. Microbiol. 10 (2011) 112. doi:10.1038/nrmicro2711.

[52] A. Ravidà, A.M. Aldridge, N.N. Driessen, F.A.H. Heus, C.H. Hokke, S.M. O’Neill, P. Robinson, P. Lobel, Z. Misulovin, M. Nussenzweig, T. Feizi, Fasciola hepatica Surface Coat Glycoproteins Contain Mannosylated and Phosphorylated N-glycans and Exhibit Immune Modulatory Properties Independent of the Mannose Receptor, PLoS Negl. Trop. Dis. 10 (2016) e0004601. doi:10.1371/journal.pntd.0004601.

[53] T.L. Lowary, Twenty Years of Mycobacterial Glycans: Furanosides and Beyond, Acc. Chem. Res. 49 (2016) 1379–1388. doi:10.1021/acs.accounts.6b00164.

[54] Q. Pan, J. Yan, Q. Liu, C. Yuan, X.-L. Zhang, An ssDNA aptamer against mannose-capped lipoarabinomannan enhances anti-tuberculosis activity of macrophages through downregulation of lipid-sensing nuclear receptor peroxisome proliferator-activated receptor γ expression, Microbiol. Immunol. (2017). doi:10.1111/1348-0421.12470.

[55] J.B. Torrelles, A.K. Azad, L.S. Schlesinger, Fine discrimination in the recognition of individual species of phosphatidyl-myo-inositol mannosides from Mycobacterium tuberculosis by C-type lectin pattern recognition receptors., J. Immunol. 177 (2006) 1805–16. doi:10.4049/jimmunol.177.3.1805.

[56] R.A. Hall, N.A.R. Gow, Mannosylation in Candida albicans: role in cell wall function and immune recognition., Mol. Microbiol. 90 (2013) 1147–61. doi:10.1111/mmi.12426.

[57] A. V Chavez-Santoscoy, R. Roychoudhury, N.L. Pohl, M.J. Wannemuehler, B. Narasimhan, A.E. Ramer-Tait, Tailoring the immune response by targeting C-type lectin receptors on alveolar macrophages using “pathogen-like” amphiphilic polyanhydride nanoparticles, Biomaterials. 33 (2012) 4762–4772. doi:10.1016/j.biomaterials.2012.03.027.

[58] F. Al-Barwani, S.L. Young, M.A. Baird, D.S. Larsen, V.K. Ward, Mannosylation of Virus-Like Particles Enhances Internalization by Antigen Presenting Cells, PLoS One. 9 (2014) e104523. doi:10.1371/journal.pone.0104523.

[59] W. Yao, Y. Peng, M. Du, J. Luo, L. Zong, Preventative Vaccine-Loaded Mannosylated Chitosan Nanoparticles Intended for Nasal Mucosal Delivery Enhance Immune Responses and Potent Tumor Immunity, Mol. Pharm. 10 (2013) 2904–2914. doi:10.1021/mp4000053.

[60] V. Kéry, J.J.F. Křepinský, C.D. Warren, P. Capek, P.D. Stahl, Ligand recognition by purified human mannose receptor, Arch. Biochem. Biophys. 298 (1992) 49–55. doi:10.1016/0003-9861(92)90092- B.

[61] S. Saijo, S. Ikeda, K. Yamabe, S. Kakuta, H. Ishigame, A. Akitsu, N. Fujikado, T. Kusaka, S. Kubo, S. Chung, R. Komatsu, N. Miura, Y. Adachi, N. Ohno, K. Shibuya, N. Yamamoto, K. Kawakami, S. Yamasaki, T. Saito, S. Akira, Y. Iwakura, Dectin-2 Recognition of α-Mannans and Induction of Th17 Cell Differentiation Is Essential for Host Defense against Candida albicans, Immunity. 32 (2010) 681–691. doi:10.1016/j.immuni.2010.05.001.

[62] L. Martinez-Pomares, The mannose receptor, J. Leukoc. Biol. 92 (2012) 1177–1186. doi:10.1189/jlb.0512231.

[63] M.V.S. Rajaram, M.N. Brooks, J.D. Morris, J.B. Torrelles, A.K. Azad, L.S. Schlesinger,

208

Mycobacterium tuberculosis Activates Human Macrophage Peroxisome Proliferator-Activated Receptor Linking Mannose Receptor Recognition to Regulation of Immune Responses, J. Immunol. 185 (2010) 929–942. doi:10.4049/jimmunol.1000866.

[64] F.L. van de Veerdonk, R.J. Marijnissen, B.J. Kullberg, H.J.P.M. Koenen, S.-C. Cheng, I. Joosten, W.B. van den Berg, D.L. Williams, J.W.M. van der Meer, L.A.B. Joosten, M.G. Netea, The Macrophage Mannose Receptor Induces IL-17 in Response to Candida albicans, Cell Host Microbe. 5 (2009) 329–340. doi:10.1016/j.chom.2009.02.006.

[65] F. V Loures, E.F. Araújo, C. Feriotti, S.B. Bazan, V.L.G. Calich, TLR-4 cooperates with Dectin-1 and mannose receptor to expand Th17 and Tc17 cells induced by Paracoccidioides brasiliensis stimulated dendritic cells., Front. Microbiol. 6 (2015) 261. doi:10.3389/fmicb.2015.00261.

[66] S. Saijo, Y. Iwakura, Dectin-1 and Dectin-2 in innate immunity against fungi, Int. Immunol. 23 (2011) 467–472. doi:10.1093/intimm/dxr046.

[67] K. Sato, X. Yang, T. Yudate, J.-S. Chung, J. Wu, K. Luby-Phelps, R.P. Kimberly, D. Underhill, P.D. Cruz, K. Ariizumi, Dectin-2 is a pattern recognition receptor for fungi that couples with the Fc receptor gamma chain to induce innate immune responses., J. Biol. Chem. 281 (2006) 38854–66. doi:10.1074/jbc.M606542200.

[68] K. Ariizumi, G.L. Shen, S. Shikano, R. Ritter, P. Zukas, D. Edelbaum, A. Morita, A. Takashima, Cloning of a second dendritic cell-associated C-type lectin (dectin-2) and its alternatively spliced isoforms., J. Biol. Chem. 275 (2000) 11957–63. doi:10.1074/jbc.275.16.11957.

[69] P.R. Taylor, D.M. Reid, S.E.M. Heinsbroek, G.D. Brown, S. Gordon, S.Y.C. Wong, Dectin-2 is predominantly myeloid restricted and exhibits unique activation-dependent expression on maturing inflammatory monocytes elicited in vivo., Eur. J. Immunol. 35 (2005) 2163–74. doi:10.1002/eji.200425785.

[70] H. Sun, X. Xu, X. Tian, H. Shao, X. Wu, Q. Wang, X. Su, Y. Shi, Activation of NF-κB and respiratory burst following Aspergillus fumigatus stimulation of macrophages, Immunobiology. 219 (2014) 25–36. doi:10.1016/j.imbio.2013.06.013.

[71] A. Yonekawa, S. Saijo, Y. Hoshino, Y. Miyake, E. Ishikawa, M. Suzukawa, H. Inoue, M. Tanaka, M. Yoneyama, M. Oh-Hora, K. Akashi, S. Yamasaki, Dectin-2 is a direct receptor for mannose-capped lipoarabinomannan of mycobacteria., Immunity. 41 (2014) 402–13. doi:10.1016/j.immuni.2014.08.005.

[72] X. Sun, Q. Pan, C. Yuan, Q. Wang, X.-L. Tang, K. Ding, X. Zhou, X.-L. Zhang, A Single ssDNA Aptamer Binding to Mannose-Capped Lipoarabinomannan of Bacillus Calmette–Guérin Enhances Immunoprotective Effect against Tuberculosis, J. Am. Chem. Soc. 138 (2016) 11680–11689. doi:10.1021/jacs.6b05357.

[73] T. Weichhart, M. Hengstschläger, M. Linke, Regulation of innate immune cell function by mTOR, Nat. Rev. Immunol. 15 (2015) 599–614. doi:10.1038/nri3901.

[74] V. Byles, A.J. Covarrubias, I. Ben-Sahra, D.W. Lamming, D.M. Sabatini, B.D. Manning, T. Horng, The TSC-mTOR pathway regulates macrophage polarization, Nat. Commun. 4 (2013) 1237–1247. doi:10.1038/ncomms3834.

[75] P.K. Langston, M. Shibata, T. Horng, Metabolism Supports Macrophage Activation, Front. Immunol. 8 (2017) 61. doi:10.3389/fimmu.2017.00061.

[76] V. Schuette, M. Embgenbroich, T. Ulas, M. Welz, J. Schulte-Schrepping, A.M. Draffehn, T. Quast, K. Koch, M. Nehring, J. König, A. Zweynert, F.L. Harms, N. Steiner, A. Limmer, I. Förster, F. Berberich- Siebelt, P.A. Knolle, D. Wohlleber, W. Kolanus, M. Beyer, J.L. Schultze, S. Burgdorf, Mannose receptor induces T-cell tolerance via inhibition of CD45 and up-regulation of CTLA-4., Proc. Natl. Acad. Sci. U. S. A. 113 (2016) 10649–54. doi:10.1073/pnas.1605885113.

[77] M.G. Netea, G.D. Brown, B.J. Kullberg, N.A.R. Gow, An integrated model of the recognition of

209

Candida albicans by the innate immune system., Nat. Rev. Microbiol. 6 (2008) 67–78. doi:10.1038/nrmicro1815.

[78] J.E. Cutler, N-glycosylation of yeast, with emphasis on Candida albicans., Med. Mycol. 39 Suppl 1 (2001) 75–86. doi:10.1080/mmy.39.1.75.86.

[79] C. Timpel, S. Strahl-Bolsinger, K. Ziegelbauer, J.F. Ernst, Multiple functions of Pmt1p-mediated protein O-mannosylation in the fungal pathogen Candida albicans., J. Biol. Chem. 273 (1998) 20837–46. doi:10.1074/jbc.273.33.20837.

[80] I. Khmaladze, T. Kelkka, S. Guerard, K. Wing, A. Pizzolla, A. Saxena, K. Lundqvist, M. Holmdahl, K.S. Nandakumar, R. Holmdahl, Mannan induces ROS-regulated, IL-17A-dependent psoriasis arthritis- like disease in mice, Proc. Natl. Acad. Sci. 111 (2014) E3669–E3678. doi:10.1073/pnas.1405798111.

[81] D.C. Ifrim, J. Quintin, F. Courjol, I. Verschueren, J.H. van Krieken, F. Koentgen, C. Fradin, N.A.R. Gow, L.A.B. Joosten, J.W.M. van der Meer, F. van de Veerdonk, M.G. Netea, The Role of Dectin-2 for Host Defense Against Disseminated Candidiasis, J. Interf. Cytokine Res. 36 (2016) 267–276. doi:10.1089/jir.2015.0040.

[82] J.P. Martínez, M.L. Gil, J.L. López-Ribot, W.L. Chaffin, Serologic response to cell wall mannoproteins and proteins of Candida albicans., Clin. Microbiol. Rev. 11 (1998) 121–41.

[83] D. Zhang, C. Chia, X. Jiao, W. Jin, S. Kasagi, R. Wu, J.E. Konkel, H. Nakatsukasa, P. Zanvit, N. Goldberg, Q. Chen, L. Sun, Z.-J. Chen, W. Chen, D-mannose induces regulatory T cells and suppresses immunopathology, Nat Med. advance on (2017). http://dx.doi.org/10.1038/nm.4375.

[84] D. Elieh-Ali-Komi, M.R. Hamblin, Chitin and Chitosan: Production and Application of Versatile Biomedical Nanomaterials., Int. J. Adv. Res. 4 (2016) 411–427.

[85] V. V. Mutsenko, O. Gryshkov, L. Lauterboeck, O. Rogulska, D.N. Tarusin, V. V. Bazhenov, K. Schütz, S. Brüggemeier, E. Gossla, A.R. Akkineni, H. Meißner, A. Lode, S. Meschke, J. Fromont, A.L. Stelling, K.R. Tabachnik, M. Gelinsky, S. Nikulin, S. Rodin, A.G. Tonevitsky, A.Y. Petrenko, B. Glasmacher, P.J. Schupp, H. Ehrlich, Novel chitin scaffolds derived from marine sponge Ianthella basta for tissue engineering approaches based on human mesenchymal stromal cells: Biocompatibility and cryopreservation, Int. J. Biol. Macromol. (2017). doi:10.1016/j.ijbiomac.2017.03.161.

[86] Z. Jamalpoor, H. Mirzadeh, M.T. Joghataei, D. Zeini, S. Bagheri-Khoulenjani, M.R. Nourani, Fabrication of cancellous biomimetic chitosan-based nanocomposite scaffolds applying a combinational method for bone tissue engineering, J. Biomed. Mater. Res. Part A. 103 (2015) 1882–1892. doi:10.1002/jbm.a.35320.

[87] S. Saravanan, R.S. Leena, N. Selvamurugan, Chitosan based biocomposite scaffolds for bone tissue engineering, Int. J. Biol. Macromol. 93 (2016) 1354–1365. doi:10.1016/j.ijbiomac.2016.01.112.

[88] W. Wang, S. Itoh, A. Matsuda, S. Ichinose, K. Shinomiya, Y. Hata, J. Tanaka, Influences of mechanical properties and permeability on chitosan nano/microfiber mesh tubes as a scaffold for nerve regeneration, J. Biomed. Mater. Res. Part A. 84A (2008) 557–566. doi:10.1002/jbm.a.31536.

[89] T. Freier, R. Montenegro, H. Shan Koh, M.S. Shoichet, Chitin-based tubes for tissue engineering in the nervous system, Biomaterials. 26 (2005) 4624–4632. doi:10.1016/j.biomaterials.2004.11.040.

[90] R. Jayakumar, M. Prabaharan, P.T. Sudheesh Kumar, S.V. Nair, H. Tamura, Biomaterials based on chitin and chitosan in wound dressing applications, Biotechnol. Adv. 29 (2011) 322–337. doi:10.1016/j.biotechadv.2011.01.005.

[91] R.A. Hall, N.A.R. Gow, Mannosylation in Candida albicans : role in cell wall function and immune recognition, Mol. Microbiol. 90 (2013) 1147–1161. doi:10.1111/mmi.12426.

[92] A. Torosantucci, C. Bromuro, P. Chiani, F. De Bernardis, F. Berti, C. Galli, F. Norelli, C. Bellucci, L. Polonelli, P. Costantino, R. Rappuoli, A. Cassone, A novel glyco-conjugate vaccine against fungal 210

pathogens., J. Exp. Med. 202 (2005) 597–606. doi:10.1084/jem.20050749.

[93] M.E. Danielson, M. Martinez, M.A. Antonysamy, K.S. Michel, D.A. Stevens, K. V. Clemons, V. Chen, Whole glucan particles as a vaccine against systemic coccidioidomycosis, J. Med. Microbiol. 64 (2015) 1237–1243. doi:10.1099/jmm.0.000138.

[94] Q. Wang, X. Sheng, A. Shi, H. Hu, Y. Yang, L. Liu, L. Fei, H. Liu, β-Glucans: Relationships between Modification, Conformation and Functional Activities, Molecules. 22 (2017) 257. doi:10.3390/molecules22020257.

[95] K. V. Clemons, M.E. Danielson, K.S. Michel, M. Liu, N.C. Ottoson, S.M. Leonardo, M. Martinez, V. Chen, M.A. Antonysamy, D.A. Stevens, Whole glucan particles as a vaccine against murine aspergillosis, J. Med. Microbiol. 63 (2014) 1750–1759. doi:10.1099/jmm.0.079681-0.

[96] L. Zhang, X. Li, X. Xu, F. Zeng, Correlation between antitumor activity, molecular weight, and conformation of lentinan., Carbohydr. Res. 340 (2005) 1515–21. doi:10.1016/j.carres.2005.02.032.

[97] H.S. Goodridge, C.N. Reyes, C.A. Becker, T.R. Katsumoto, J. Ma, A.J. Wolf, N. Bose, A.S.H. Chan, A.S. Magee, M.E. Danielson, A. Weiss, J.P. Vasilakos, D.M. Underhill, Activation of the innate immune receptor Dectin-1 upon formation of a “phagocytic synapse”., Nature. 472 (2011) 471–5. doi:10.1038/nature10071.

[98] G. Liao, S. Burgula, Z. Zhou, Z. Guo, A Convergent Synthesis of 6- O -Branched β-Glucan Oligosaccharides, European J. Org. Chem. 2015 (2015) 2942–2951. doi:10.1002/ejoc.201500229.

[99] D. Elieh Ali Komi, L. Sharma, C.S. Dela Cruz, Chitin and Its Effects on Inflammatory and Immune Responses, Clin. Rev. Allergy Immunol. (2017). doi:10.1007/s12016-017-8600-0.

[100] D.J. Manners, A.J. Masson, J.C. Patterson, The structure of a beta-(1 leads to 3)-D-glucan from yeast cell walls., Biochem. J. 135 (1973) 19–30. doi:10.1042/bj1350019.

[101] D.J. Manners, A.J. Masson, J.C. Patterson, H. Björndal, B. Lindberg, The structure of a beta-(1--6)- D-glucan from yeast cell walls., Biochem. J. 135 (1973) 31–6. doi:10.1042/bj1350031.

[102] S. Walachowski, G. Tabouret, G. Foucras, J. Bain, H. Mora-Montes, R. Barker, Triggering Dectin-1- Pathway Alone Is Not Sufficient to Induce Cytokine Production by Murine Macrophages, PLoS One. 11 (2016) e0148464. doi:10.1371/journal.pone.0148464.

[103] G. Liao, Z. Zhou, J. Liao, L. Zu, Q. Wu, Z. Guo, 6-O-Branched Oligo-β-glucan-Based Antifungal Glycoconjugate Vaccines, ACS Infect. Dis. 2 (2016) 123–131. doi:10.1021/acsinfecdis.5b00104.

[104] J.W.M. van der Meer, L.A.B. Joosten, N. Riksen, M.G. Netea, Trained immunity: A smart way to enhance innate immune defence., Mol. Immunol. 68 (2015) 40–4. doi:10.1016/j.molimm.2015.06.019.

[105] P. Garcia-Valtanen, R.M. Guzman-Genuino, D.L. Williams, J.D. Hayball, K.R. Diener, Evaluation of trained immunity by β-1, 3 (d)-glucan on murine monocytes in vitro and duration of response in vivo., Immunol. Cell Biol. (2017). doi:10.1038/icb.2017.13.

[106] J. Quintin, S. Saeed, J.H.A. Martens, E.J. Giamarellos-Bourboulis, D.C. Ifrim, C. Logie, L. Jacobs, T. Jansen, B.-J. Kullberg, C. Wijmenga, L.A.B. Joosten, R.J. Xavier, J.W.M. van der Meer, H.G. Stunnenberg, M.G. Netea, Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes., Cell Host Microbe. 12 (2012) 223–32. doi:10.1016/j.chom.2012.06.006.

[107] S.-C. Cheng, J. Quintin, R.A. Cramer, K.M. Shepardson, S. Saeed, V. Kumar, E.J. Giamarellos- Bourboulis, J.H.A. Martens, N.A. Rao, A. Aghajanirefah, G.R. Manjeri, Y. Li, D.C. Ifrim, R.J.W. Arts, B.M.J.W. van der Veer, B.M.J.W. van der Meer, P.M.T. Deen, C. Logie, L.A. O’Neill, P. Willems, F.L. van de Veerdonk, J.W.M. van der Meer, A. Ng, L.A.B. Joosten, C. Wijmenga, H.G. Stunnenberg, R.J. Xavier, M.G. Netea, mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity., Science. 345 (2014) 1250684. doi:10.1126/science.1250684. 211

[108] J. Kleinnijenhuis, J. Quintin, F. Preijers, L.A.B. Joosten, D.C. Ifrim, S. Saeed, C. Jacobs, J. van Loenhout, D. de Jong, H.G. Stunnenberg, R.J. Xavier, J.W.M. van der Meer, R. van Crevel, M.G. Netea, Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes., Proc. Natl. Acad. Sci. U. S. A. 109 (2012) 17537–42. doi:10.1073/pnas.1202870109.

[109] R. Pichler, G. Gruenbacher, Z. Culig, A. Brunner, D. Fuchs, J. Fritz, H. Gander, A. Rahm, M. Thurnher, Intratumoral Th2 predisposition combines with an increased Th1 functional phenotype in clinical response to intravesical BCG in bladder cancer, Cancer Immunol. Immunother. 66 (2017) 427–440. doi:10.1007/s00262-016-1945-z.

[110] P.F. Severino, M. Silva, M. Carrascal, N. Malagolini, M. Chiricolo, G. Venturi, A. Astolfi, M. Catera, P.A. Videira, F. Dall’Olio, Expression of sialyl-Tn sugar antigen in bladder cancer cells affects response to Bacillus Calmette Guérin (BCG) and to oxidative damage, Oncotarget. (2017). doi:10.18632/oncotarget.17138.

[111] S. Saeed, J. Quintin, H.H.D. Kerstens, N.A. Rao, A. Aghajanirefah, F. Matarese, S.-C. Cheng, J. Ratter, K. Berentsen, M.A. van der Ent, N. Sharifi, E.M. Janssen-Megens, M. Ter Huurne, A. Mandoli, T. van Schaik, A. Ng, F. Burden, K. Downes, M. Frontini, V. Kumar, E.J. Giamarellos- Bourboulis, W.H. Ouwehand, J.W.M. van der Meer, L.A.B. Joosten, C. Wijmenga, J.H.A. Martens, R.J. Xavier, C. Logie, M.G. Netea, H.G. Stunnenberg, Epigenetic programming of monocyte-to- macrophage differentiation and trained innate immunity., Science. 345 (2014) 1251086. doi:10.1126/science.1251086.

[112] L. Rizzetto, D.C. Ifrim, S. Moretti, N. Tocci, S.-C. Cheng, J. Quintin, G. Renga, V. Oikonomou, C. De Filippo, T. Weil, B.A. Blok, M.S. Lenucci, M.A.S. Santos, L. Romani, M.G. Netea, D. Cavalieri, Fungal Chitin Induces Trained Immunity in Human Monocytes during Cross-talk of the Host with Saccharomyces cerevisiae., J. Biol. Chem. 291 (2016) 7961–72. doi:10.1074/jbc.M115.699645.

[113] C.A. Da Silva, D. Hartl, W. Liu, C.G. Lee, J.A. Elias, TLR-2 and IL-17A in chitin-induced macrophage activation and acute inflammation., J. Immunol. 181 (2008) 4279–86. doi:10.4049/jimmunol.181.6.4279.

[114] N. Amarsaikhan, S.P. Templeton, Co-recognition of β-glucan and chitin and programming of adaptive immunity to Aspergillus fumigatus., Front. Microbiol. 6 (2015) 344. doi:10.3389/fmicb.2015.00344.

[115] T. Satoh, O. Takeuchi, A. Vandenbon, K. Yasuda, Y. Tanaka, Y. Kumagai, T. Miyake, K. Matsushita, T. Okazaki, T. Saitoh, K. Honma, T. Matsuyama, K. Yui, T. Tsujimura, D.M. Standley, K. Nakanishi, K. Nakai, S. Akira, The Jmjd3-Irf4 axis regulates M2 macrophage polarization and host responses against helminth infection., Nat. Immunol. 11 (2010) 936–44. doi:10.1038/ni.1920.

[116] S. Tundup, L. Srivastava, D.A. Harn, Polarization of host immune responses by helminth-expressed glycans., Ann. N. Y. Acad. Sci. 1253 (2012) E1–E13. doi:10.1111/j.1749-6632.2012.06618.x.

[117] C.A. Da Silva, C. Chalouni, A. Williams, D. Hartl, C.G. Lee, J.A. Elias, Chitin Is a Size-Dependent Regulator of Macrophage TNF and IL-10 Production, J. Immunol. 182 (2009). doi:10.4049/jimmunol.0802113.

[118] C.J. Wagner, S. Huber, S. Wirth, D. Voehringer, Chitin induces upregulation of B7-H1 on macrophages and inhibits T-cell proliferation., Eur. J. Immunol. 40 (2010) 2882–90. doi:10.1002/eji.201040422.

[119] B. Prietl, C. Meindl, E. Roblegg, T.R. Pieber, G. Lanzer, E. Frohlich, Nano-sized and micro-sized polystyrene particles affect phagocyte function, Cell Biol Toxicol. 30 (2014) 1–16. doi:10.1007/s10565-013-9265-y.

[120] L.J. Cruz, P.J. Tacken, F. Rueda, J.C. Domingo, F. Albericio, C.G. Figdor, Targeting nanoparticles to dendritic cells for immunotherapy, Methods Enzym. 509 (2012) 143–163. doi:10.1016/b978-0-12- 391858-1.00008-3.

212

[121] C.L. Hardy, J.S. Lemasurier, R. Mohamud, J. Yao, S.D. Xiang, J.M. Rolland, R.E. O’Hehir, M. Plebanski, Differential uptake of nanoparticles and microparticles by pulmonary APC subsets induces discrete immunological imprints, J Immunol. 191 (2013) 5278–5290. doi:10.4049/jimmunol.1203131.

[122] X. Liu, N. Huang, H. Li, Q. Jin, J. Ji, Surface and size effects on cell interaction of gold nanoparticles with both phagocytic and nonphagocytic cells, Langmuir. 29 (2013) 9138–9148. doi:10.1021/la401556k.

[123] R. Mohamud, S.D. Xiang, C. Selomulya, J.M. Rolland, R.E. O’Hehir, C.L. Hardy, M. Plebanski, The effects of engineered nanoparticles on pulmonary immune homeostasis, Drug Metab Rev. 46 (2014) 176–190. doi:10.3109/03602532.2013.859688.

[124] L.K. Dubey, J.B. Moeller, A. Schlosser, G.L. Sorensen, U. Holmskov, Induction of innate immunity by Aspergillus fumigatus cell wall polysaccharides is enhanced by the composite presentation of chitin and beta-glucan, Immunobiology. 219 (2014) 179–188. doi:10.1016/j.imbio.2013.10.003.

[125] J. Wagener, R.K.S. Malireddi, M.D. Lenardon, M. Köberle, S. Vautier, D.M. MacCallum, T. Biedermann, M. Schaller, M.G. Netea, T.-D. Kanneganti, G.D. Brown, A.J.P. Brown, N.A.R. Gow, Fungal Chitin Dampens Inflammation through IL-10 Induction Mediated by NOD2 and TLR9 Activation, PLoS Pathog. 10 (2014) e1004050. doi:10.1371/journal.ppat.1004050.

[126] R.M. Clancy, A.J. Markham, J.P. Buyon, Endosomal Toll-like receptors in clinically overt and silent autoimmunity, Immunol. Rev. 269 (2016) 76–84. doi:10.1111/imr.12383.

[127] R. Caruso, N. Warner, N. Inohara, G. Núñez, NOD1 and NOD2: Signaling, Host Defense, and Inflammatory Disease, Immunity. 41 (2014) 898–908. doi:10.1016/j.immuni.2014.12.010.

[128] J. Kleinnijenhuis, M. Oosting, L.A.B. Joosten, M.G. Netea, R. Van Crevel, Innate immune recognition of Mycobacterium tuberculosis., Clin. Dev. Immunol. 2011 (2011) 405310. doi:10.1155/2011/405310.

[129] Y. Yang, M. Guo, R. Qian, C. Liu, X. Zong, Y.-Q. Li, W. Li, Binding efficacy and kinetics of chitosan with DNA duplex: The effects of deacetylation degree and nucleotide sequences, Carbohydr. Polym. 169 (2017) 451–457. doi:10.1016/j.carbpol.2017.04.040.

[130] C.L. Bueter, C.K. Lee, V.A.K. Rathinam, G.J. Healy, C.H. Taron, C.A. Specht, S.M. Levitz, Chitosan but Not Chitin Activates the Inflammasome by a Mechanism Dependent upon Phagocytosis, J. Biol. Chem. 286 (2011) 35447–35455. doi:10.1074/jbc.M111.274936.

[131] C.L. Bueter, C.K. Lee, J.P. Wang, G.R. Ostroff, C.A. Specht, S.M. Levitz, Spectrum and mechanisms of inflammasome activation by chitosan., J. Immunol. 192 (2014) 5943–51. doi:10.4049/jimmunol.1301695.

[132] A. Mori, E. Oleszycka, F.A. Sharp, M. Coleman, Y. Ozasa, M. Singh, D.T. O’Hagan, L. Tajber, O.I. Corrigan, E.A. McNeela, E.C. Lavelle, The vaccine adjuvant alum inhibits IL-12 by promoting PI3 kinase signaling while chitosan does not inhibit IL-12 and enhances Th1 and Th17 responses, Eur. J. Immunol. 42 (2012) 2709–2719. doi:10.1002/eji.201242372.

[133] D. Fong, P. Grégoire-Gélinas, A.P. Cheng, T. Mezheritsky, M. Lavertu, S. Sato, C.D. Hoemann, Lysosomal rupture induced by structurally distinct either promotes a type 1 IFN response or activates the inflammasome in macrophages, Biomaterials. 129 (2017) 127–138. doi:10.1016/j.biomaterials.2017.03.022.

[134] M.P. Etna, E. Giacomini, M. Severa, E.M. Coccia, Pro- and anti-inflammatory cytokines in tuberculosis: a two-edged sword in TB pathogenesis, Semin Immunol. 26 (2014) 543–551. doi:10.1016/j.smim.2014.09.011.

[135] M.I. Oliveira, S.G. Santos, M.J. Oliveira, A.L. Torres, M.A. Barbosa, Chitosan drives anti- inflammatory macrophage polarisation and pro-inflammatory dendritic cell stimulation., Eur. Cell. Mater. 24 (2012) 136–52; discussion 152-3. doi:10.22203/eCM.v024a10.

213

[136] T. Freier, H.S. Koh, K. Kazazian, M.S. Shoichet, Controlling cell adhesion and degradation of chitosan films by N-acetylation, Biomaterials. 26 (2005) 5872–5878. doi:10.1016/j.biomaterials.2005.02.033.

[137] S. Franz, S. Rammelt, D. Scharnweber, J.C. Simon, Immune responses to implants - a review of the implications for the design of immunomodulatory biomaterials, Biomaterials. 32 (2011) 6692– 6709. doi:10.1016/j.biomaterials.2011.05.078.

[138] B.N. Brown, S.F. Badylak, Expanded applications, shifting paradigms and an improved understanding of host-biomaterial interactions, Acta Biomater. 9 (2013) 4948–4955. doi:10.1016/j.actbio.2012.10.025.

[139] B.N. Brown, R. Londono, S. Tottey, L. Zhang, K.A. Kukla, M.T. Wolf, K.A. Daly, J.E. Reing, S.F. Badylak, Macrophage phenotype as a predictor of constructive remodeling following the implantation of biologically derived surgical mesh materials, Acta Biomater. 8 (2012) 978–987. doi:10.1016/j.actbio.2011.11.031.

[140] D.P. Vasconcelos, A.C. Fonseca, M. Costa, I.F. Amaral, M.A. Barbosa, A.P. Águas, J.N. Barbosa, Macrophage polarization following chitosan implantation, Biomaterials. 34 (2013) 9952–9959. doi:10.1016/j.biomaterials.2013.09.012.

[141] Y. Han, L. Zhao, Z. Yu, J. Feng, Q. Yu, Role of mannose receptor in oligochitosan-mediated stimulation of macrophage function, Int. Immunopharmacol. 5 (2005) 1533–1542. doi:10.1016/j.intimp.2005.04.015.

[142] J. Feng, L. Zhao, Q. Yu, Receptor-mediated stimulatory effect of oligochitosan in macrophages, Biochem. Biophys. Res. Commun. 317 (2004) 414–420. doi:10.1016/j.bbrc.2004.03.048.

[143] A. Ueki, K. Un, Y. Mino, M. Yoshida, S. Kawakami, H. Ando, H. Ishida, F. Yamashita, M. Hashida, M. Kiso, Synthesis and evaluation of glyco-coated liposomes as drug carriers for active targeting in drug delivery systems, Carbohydr Res. 405 (2015) 78–86. doi:10.1016/j.carres.2014.06.028.

[144] E. Mills, L.A.J. O’Neill, Succinate: a metabolic signal in inflammation, Trends Cell Biol. 24 (2014) 313–320. doi:10.1016/j.tcb.2013.11.008.

[145] W.K.E. Ip, N. Hoshi, D.S. Shouval, S. Snapper, R. Medzhitov, Anti-inflammatory effect of IL-10 mediated by metabolic reprogramming of macrophages, Science (80-. ). 356 (2017) 513–519. doi:10.1126/science.aal3535.

[146] P. Seth, E. Csizmadia, A. Hedblom, M. Vuerich, H. XIE, M. Li, M.S. Longhi, B. Wegiel, Deletion of lactate dehydrogenase-A in myeloid cells triggers antitumor immunity., Cancer Res. (2017) canres.2938.2016. doi:10.1158/0008-5472.CAN-16-2938.

[147] J. Van den Bossche, L.A. O’Neill, D. Menon, Macrophage Immunometabolism: Where Are We (Going)?, Trends Immunol. (2017). doi:10.1016/j.it.2017.03.001.

[148] E.L. Mills, B. Kelly, L.A.J. O’Neill, Mitochondria are the powerhouses of immunity, Nat. Immunol. 18 (2017) 488–498. doi:10.1038/ni.3704.

[149] E.L. Mills, B. Kelly, A. Logan, A.S.H. Costa, M. Varma, C.E. Bryant, P. Tourlomousis, J.H.M. Däbritz, E. Gottlieb, I. Latorre, S.C. Corr, G. McManus, D. Ryan, H.T. Jacobs, M. Szibor, R.J. Xavier, T. Braun, C. Frezza, M.P. Murphy, L.A. O’Neill, Succinate Dehydrogenase Supports Metabolic Repurposing of Mitochondria to Drive Inflammatory Macrophages, Cell. 167 (2016) 457-470.e13. doi:10.1016/j.cell.2016.08.064.

[150] S.C.-C. Huang, A.M. Smith, B. Everts, M. Colonna, E.L. Pearce, J.D. Schilling, E.J. Pearce, Metabolic Reprogramming Mediated by the mTORC2-IRF4 Signaling Axis Is Essential for Macrophage Alternative Activation., Immunity. 45 (2016) 817–830. doi:10.1016/j.immuni.2016.09.016.

[151] A.J. Covarrubias, H.I. Aksoylar, J. Yu, N.W. Snyder, A.J. Worth, S.S. Iyer, J. Wang, I. Ben-Sahra, V. Byles, T. Polynne-Stapornkul, E.C. Espinosa, D. Lamming, B.D. Manning, Y. Zhang, I.A. Blair, T. Horng, Akt-mTORC1 signaling regulates Acly to integrate metabolic input to control of 214

macrophage activation., Elife. 5 (2016). doi:10.7554/eLife.11612.

[152] G.M. Tannahill, A.M. Curtis, J. Adamik, E.M. Palsson-McDermott, A.F. McGettrick, G. Goel, C. Frezza, N.J. Bernard, B. Kelly, N.H. Foley, L. Zheng, A. Gardet, Z. Tong, S.S. Jany, S.C. Corr, M. Haneklaus, B.E. Caffrey, K. Pierce, S. Walmsley, F.C. Beasley, E. Cummins, V. Nizet, M. Whyte, C.T. Taylor, H. Lin, S.L. Masters, E. Gottlieb, V.P. Kelly, C. Clish, P.E. Auron, R.J. Xavier, L.A.J. O’Neill, Succinate is an inflammatory signal that induces IL-1β through HIF-1α., Nature. 496 (2013) 238– 42. doi:10.1038/nature11986.

[153] L. Liu, Y. Lu, J. Martinez, Y. Bi, G. Lian, T. Wang, S. Milasta, J. Wang, M. Yang, G. Liu, D.R. Green, R. Wang, Proinflammatory signal suppresses proliferation and shifts macrophage metabolism from Myc-dependent to HIF1α-dependent., Proc. Natl. Acad. Sci. U. S. A. 113 (2016) 1564–9. doi:10.1073/pnas.1518000113.

[154] S.J. Lawless, N. Kedia-Mehta, J.F. Walls, R. McGarrigle, O. Convery, L. V Sinclair, M.N. Navarro, J. Murray, D.K. Finlay, Glucose represses dendritic cell-induced T cell responses., Nat. Commun. 8 (2017) 15620. doi:10.1038/ncomms15620.

[155] S. Pavlou, L. Wang, H. Xu, M. Chen, Higher phagocytic activity of thioglycollate-elicited peritoneal macrophages is related to metabolic status of the cells, J. Inflamm. 14 (2017) 4. doi:10.1186/s12950-017-0151-x.

[156] L. Tretter, A. Patocs, C. Chinopoulos, Succinate, an intermediate in metabolism, signal transduction, ROS, hypoxia, and tumorigenesis., Biochim. Biophys. Acta. 1857 (2016) 1086–101. doi:10.1016/j.bbabio.2016.03.012.

[157] B. Kalyanaraman, Teaching the basics of cancer metabolism: Developing antitumor strategies by exploiting the differences between normal and cancer cell metabolism, Redox Biol. 12 (2017) 833–842. doi:10.1016/j.redox.2017.04.018.

[158] T. Ohashi, M. Aoki, H. Tomita, T. Akazawa, K. Sato, B. Kuze, K. Mizuta, A. Hara, H. Nagaoka, N. Inoue, Y. Ito, M2-like macrophage polarization in high lactic acid-producing head and neck cancer, Cancer Sci. (2017). doi:10.1111/cas.13244.

[159] Z. Yu, L. Zhao, H. Ke, Potential role of nuclear factor-kappaB in the induction of nitric oxide nd tumor necrosis factor-alpha by oligochitosan in macrophages, Int. Immunopharmacol. 4 (2004) 193–200. doi:10.1016/j.intimp.2003.12.001.

[160] L. Liu, Y. Lu, J. Martinez, Y. Bi, G. Lian, T. Wang, S. Milasta, J. Wang, M. Yang, G. Liu, D.R. Green, R. Wang, Proinflammatory signal suppresses proliferation and shifts macrophage metabolism from Myc-dependent to HIF1α-dependent., Proc. Natl. Acad. Sci. U. S. A. 113 (2016) 1564–9. doi:10.1073/pnas.1518000113.

[161] S. Spence, M.K. Greene, F. Fay, E. Hams, S.P. Saunders, U. Hamid, M. Fitzgerald, J. Beck, B.K. Bains, P. Smyth, E. Themistou, D.M. Small, D. Schmid, C.M. O’Kane, D.C. Fitzgerald, S.M. Abdelghany, J.A. Johnston, P.G. Fallon, J.F. Burrows, D.F. McAuley, A. Kissenpfennig, C.J. Scott, Targeting Siglecs with a sialic acid-decorated nanoparticle abrogates inflammation, Sci Transl Med. 7 (2015) 303ra140. doi:10.1126/scitranslmed.aab3459.

[162] C. Büll, E. Collado-Camps, E.D. Kers-Rebel, T. Heise, J.N. Søndergaard, M.H. den Brok, B.M. Schulte, T.J. Boltje, G.J. Adema, Metabolic sialic acid blockade lowers the activation threshold of moDCs for TLR stimulation., Immunol. Cell Biol. 95 (2017) 408–415. doi:10.1038/icb.2016.105.

[163] M.F. Giraud, J.H. Naismith, The rhamnose pathway, in: Curr Opin Struct Biol, Institut de Biochimie et de Genetique Cellulaires du CNRS, Universite Victor Segalen, Bordeaux 2, 1 rue Camille Saint- Saens, F-33 077 cedex, Bordeaux, France., 2000: pp. 687–696.

[164] S. Lebeer, T.L.A. Verhoeven, G. Francius, G. Schoofs, I. Lambrichts, Y. Dufrêne, J. Vanderleyden, S.C.J. De Keersmaecker, Identification of a Gene Cluster for the Biosynthesis of a Long, Galactose- Rich Exopolysaccharide in Lactobacillus rhamnosus GG and Functional Analysis of the Priming Glycosyltransferase., Appl. Environ. Microbiol. 75 (2009) 3554–63. doi:10.1128/AEM.02919-08.

215

[165] X.M. Jiang, B. Neal, F. Santiago, S.J. Lee, L.K. Romana, P.R. Reeves, Structure and sequence of the rfb (O antigen) gene cluster of Salmonella serovar typhimurium (strain LT2), Mol Microbiol. 5 (1991) 695–713. http://www.ncbi.nlm.nih.gov/pubmed/1710759.

[166] D.F. Macpherson, P.A. Manning, R. Morona, Characterization of the dTDP-rhamnose biosynthetic genes encoded in the rfb locus of Shigella flexneri, Mol Microbiol. 11 (1994) 281–292. http://www.ncbi.nlm.nih.gov/pubmed/8170390.

[167] J. Bourke, C.F. Brereton, S. V Gordon, E.C. Lavelle, E.M. Scanlan, The synthesis and biological evaluation of mycobacterial p-hydroxybenzoic acid derivatives (p-HBADs), Org Biomol Chem. 12 (2014) 1114–1123. doi:10.1039/c3ob42277a.

[168] Solabia, Rhamnosoft® polysaccharide Solabia Report NO AC046, Solabia, France, 2001.

[169] F. Zhao, B.A. Heesters, I. Chiu, Y. Gao, J. Shi, N. Zhou, M.C. Carroll, B. Xu, L-Rhamnose-containing supramolecular nanofibrils as potential immunosuppressive materials, Org Biomol Chem. 12 (2014) 6816–6819. doi:10.1039/c4ob01362j.

[170] P. V. Licciardi, I.H. Ismail, A. Balloch, M. Mui, E. Hoe, K. Lamb, M.L.K. Tang, Maternal Supplementation with LGG Reduces Vaccine-Specific Immune Responses in Infants at High-Risk of Developing Allergic Disease, Front. Immunol. 4 (2013) 381. doi:10.3389/fimmu.2013.00381.

[171] I.H. Ismail, R.J. Boyle, L.-J. Mah, P. V Licciardi, M.L.K. Tang, Reduced neonatal regulatory T cell response to microbial stimuli associates with subsequent eczema in high-risk infants., Pediatr. Allergy Immunol. 25 (2014) 674–84. doi:10.1111/pai.12303.

[172] R.J. Boyle, I.H. Ismail, S. Kivivuori, P. V. Licciardi, R.M. Robins-Browne, L.-J. Mah, C. Axelrad, S. Moore, S. Donath, J.B. Carlin, S.J. Lahtinen, M.L.K. Tang, Lactobacillus GG treatment during pregnancy for the prevention of eczema: a randomized controlled trial, Allergy. 66 (2011) 509– 516. doi:10.1111/j.1398-9995.2010.02507.x.

[173] K.-C. Hsiao, A.-L. Ponsonby, C. Axelrad, S. Pitkin, M.L.K. Tang, W. Burks, S. Donath, F. Orsini, D. Tey, M. Robinson, E.L. Su, Long-term clinical and immunological effects of probiotic and peanut oral immunotherapy after treatment cessation: 4-year follow-up of a randomised, double-blind, placebo-controlled trial, Lancet Child Adolesc. Heal. 1 (2017) 97–105. doi:10.1016/S2352- 4642(17)30041-X.

[174] R.B. Zheng, S.A.F. Jégouzo, M. Joe, Y. Bai, H.-A. Tran, K. Shen, J. Saupe, L. Xia, M.F. Ahmed, Y.-H. Liu, P.S. Patil, A. Tripathi, S.-C. Hung, M.E. Taylor, T.L. Lowary, K. Drickamer, Insights into Interactions of Mycobacteria with the Host Innate Immune System from a Novel Array of Synthetic Mycobacterial Glycans, ACS Chem. Biol. 12 (2017) 2990–3002. doi:10.1021/acschembio.7b00797.

[175] T. Shirai, Y. Watanabe, M.S. Lee, T. Ogawa, K. Muramoto, Structure of rhamnose-binding lectin CSL3: unique pseudo-tetrameric architecture of a pattern recognition protein, in: J Mol Biol, Department of Bioscience, Nagahama Institute of Bio-science and Technology, and JST-BIRD, Tamura, Nagahama, Japan. [email protected], 2009: pp. 390–403. doi:10.1016/j.jmb.2009.06.027.

[176] D. Cerdan, C. Grillon, M. Monsigny, G. Redziniak, C. Kieda, Human keratinocyte membrane lectins: characterization and modulation of their expression by cytokines, in: Biol Cell, Departement de biochimie des glycoconjugues, Centre de biophysique moleculaire, CNRS, Orleans, France., 1991: pp. 35–42.

[177] E. Andres, J. Molinari, G. Peterszegi, B. Mariko, E. Ruszova, V. Velebny, G. Faury, L. Robert, Pharmacological properties of rhamnose-rich polysaccharides, potential interest in age- dependent alterations of connectives tissues, Pathol Biol. 54 (2006) 420–425. doi:10.1016/j.patbio.2006.07.004.

[178] G. Faury, E. Ruszova, J. Molinari, B. Mariko, S. Raveaud, V. Velebny, L. Robert, The alpha-L- Rhamnose recognizing lectin site of human dermal fibroblasts functions as a signal transducer:

216

modulation of Ca2+ fluxes and gene expression, in: Biochim Biophys Acta, Laboratoire Physiopathologies vasculaires: interactions cellulaires, signalisation et vieillissement, iRTSV-APV, CEA-Grenoble, 17 rue des Martyrs, 38054 Grenoble Cedex 9, France. [email protected], 2008: pp. 1388–1394. doi:10.1016/j.bbagen.2008.07.008.

[179] M. Choi, S. Mukherjee, N.H. Kang, J.L. Barkat, H.A. Parray, J.W. Yun, L-rhamnose induces browning in 3T3-L1 white adipocytes and activates HIB1B brown adipocytes, IUBMB Life. 70 (2018) 563– 573. doi:10.1002/iub.1750.

[180] T. Hadi, R. Douhard, A.M.M. Dias, M. Wendremaire, M. Pezzè, M. Bardou, P. Sagot, C. Garrido, F. Lirussi, Beta3 adrenergic receptor stimulation in human macrophages inhibits NADPHoxidase activity and induces catalase expression via PPARγ activation, Biochim. Biophys. Acta - Mol. Cell Res. 1864 (2017) 1769–1784. doi:10.1016/j.bbamcr.2017.07.003.

[181] P. Constant, E. Perez, W. Malaga, M.A. Laneelle, O. Saurel, M. Daffe, C. Guilhot, Role of the pks15/1 gene in the biosynthesis of phenolglycolipids in the Mycobacterium tuberculosis complex. Evidence that all strains synthesize glycosylated p-hydroxybenzoic methyl esters and that strains devoid of phenolglycolipids harbor a frameshift , J Biol Chem. 277 (2002) 38148– 38158. doi:10.1074/jbc.M206538200.

[182] C.J. Cambier, S. Falkow, L. Ramakrishnan, Host Evasion and Exploitation Schemes of Mycobacterium tuberculosis, Cell. 159 (2014) 1497–1509. doi:10.1016/j.cell.2014.11.024.

[183] Global Tuberculosis Report Executive Summary, 2018. https://www.who.int/tb/publications/global_report/tb18_ExecSum_web_4Oct18.pdf?ua=1.

[184] J.L. Flynn, J. Chan, K.J. Triebold, D.K. Dalton, T.A. Stewart, B.R. Bloom, An essential role for interferon gamma in resistance to Mycobacterium tuberculosis infection, J Exp Med. 178 (1993) 2249–2254.

[185] D. Huang, L. Bao, Mycobacterium tuberculosis EspB protein suppresses interferon-gamma- induced autophagy in murine macrophages, J Microbiol Immunol Infect. (2014). doi:10.1016/j.jmii.2014.11.008.

[186] J.N. Sondergaard, J.M. Laursen, L.B. Rosholm, S. Brix, Mycobacterium tuberculosis promotes Th17 expansion via regulation of human dendritic cells toward a high CD14 and low IL-12p70 phenotype that reprograms upon exogenous IFN-gamma, Int Immunol. 26 (2014) 705–716. doi:10.1093/intimm/dxu085.

[187] A.A. Awomoyi, A. Marchant, J.M. Howson, K.P. McAdam, J.M. Blackwell, M.J. Newport, Interleukin-10, polymorphism in SLC11A1 (formerly NRAMP1), and susceptibility to tuberculosis, J Infect Dis. 186 (2002) 1808–1814. doi:10.1086/345920.

[188] D.D. Barnes, M.L.E. Lundahl, E.C. Lavelle, E.M. Scanlan, The Emergence of Phenolic Glycans as Virulence Factors in Mycobacterium tuberculosis, ACS Chem. Biol. 12 (2017) 1969–1979. doi:10.1021/acschembio.7b00394.

[189] G. Stadthagen, M. Jackson, P. Charles, F. Boudou, N. Barilone, M. Huerre, P. Constant, A. Liav, I. Bottova, J. Nigou, T. Brando, G. Puzo, M. Daffe, P. Benjamin, S. Coade, R.S. Buxton, R.E. Tascon, A. Rae, B.D. Robertson, D.B. Lowrie, D.B. Young, B. Gicquel, R. Griffin, Comparative investigation of the pathogenicity of three Mycobacterium tuberculosis mutants defective in the synthesis of p- hydroxybenzoic acid derivatives, Microbes Infect. 8 (2006) 2245–2253. doi:10.1016/j.micinf.2006.04.008.

[190] H.R.H. Elsaidi, D.R. Barreda, C.W. Cairo, T.L. Lowary, Mycobacterial Phenolic Glycolipids with a Simplified Lipid Aglycone Modulate Cytokine Levels through Toll-Like Receptor 2, ChemBioChem. 14 (2013) 2153–2159. doi:10.1002/cbic.201300505.

[191] D. Kaur, M.E. Guerin, H. Skovierova, P.J. Brennan, M. Jackson, Chapter 2: Biogenesis of the cell wall and other glycoconjugates of Mycobacterium tuberculosis, Adv Appl Microbiol. 69 (2009) 23– 78. doi:10.1016/s0065-2164(09)69002-x.

217

[192] G. Stadthagen, J. Kordulakova, R. Griffin, P. Constant, I. Bottova, N. Barilone, B. Gicquel, M. Daffe, M. Jackson, p-Hydroxybenzoic acid synthesis in Mycobacterium tuberculosis, J Biol Chem. 280 (2005) 40699–40706. doi:10.1074/jbc.M508332200.

[193] M.L.E. Lundahl, E.C. Lavelle, E.M. Scanlan, An Investigation of the Murine Immune Response to L- Rhamnose and Synthesised Rhamnosyl Derivatives, Trinity College Dublin, 2015.

[194] I. Garcia, M. Marradi, S. Penades, Glyconanoparticles: multifunctional nanomaterials for biomedical applications, Nanomedicine (Lond). 5 (2010) 777–792. doi:10.2217/nnm.10.48.

[195] S. Biswas, S.H. Medina, J.J. Barchi Jr., Synthesis and cell-selective antitumor properties of amino acid conjugated tumor-associated carbohydrate antigen-coated gold nanoparticles, Carbohydr Res. 405 (2015) 93–101. doi:10.1016/j.carres.2014.11.002.

[196] W. Chen, L. Gu, W. Zhang, E. Motari, L. Cai, T.J. Styslinger, P.G. Wang, L-rhamnose antigen: a promising alternative to alpha-gal for cancer immunotherapies, ACS Chem Biol. 6 (2011) 185–191. doi:10.1021/cb100318z.

[197] P. Rouge, M. Becuwe, A. Dassonville-Klimpt, S. Da Nascimento, J.-F. Raimbert, D. Cailleu, E. Baudrin, P. Sonnet, Efficient synthesis of amino-protected calix[4]arenes selectively functionalized with iron chelator ICL670 designed as platform for iron recognition, Tetrahedron. 67 (2011) 2916– 2924. doi:10.1016/j.tet.2011.02.059.

[198] M.B. Lutz, N. Kukutsch, A.L. Ogilvie, S. Rössner, F. Koch, N. Romani, G. Schuler, An advanced culture method for generating large quantities of highly pure dendritic cells from mouse bone marrow., J. Immunol. Methods. 223 (1999) 77–92. http://www.ncbi.nlm.nih.gov/pubmed/10037236.

[199] M. Kubista, J.M. Andrade, M. Bengtsson, A. Forootan, J. Jonák, K. Lind, R. Sindelka, R. Sjöback, B. Sjögreen, L. Strömbom, A. Ståhlberg, N. Zoric, The real-time polymerase chain reaction, Mol. Aspects Med. 27 (2006) 95–125. doi:10.1016/j.mam.2005.12.007.

[200] A. Arbués, W. Malaga, P. Constant, C. Guilhot, J. Prandi, C. Astarie-Dequeker, Trisaccharides of Phenolic Glycolipids Confer Advantages to Pathogenic Mycobacteria through Manipulation of Host-Cell Pattern-Recognition Receptors, ACS Chem. Biol. 11 (2016) 2865–2875. doi:10.1021/acschembio.6b00568.

[201] M.B. Reed, P. Domenech, C. Manca, H. Su, A.K. Barczak, B.N. Kreiswirth, G. Kaplan, C.E. Barry, A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response, Nature. 431 (2004) 84–87. doi:10.1038/nature02837.

[202] R. Oldenburg, V. Mayau, J. Prandi, A. Arbues, C. Astarie-Dequeker, C. Guilhot, C. Werts, N. Winter, C. Demangel, Mycobacterial Phenolic Glycolipids Selectively Disable TRIF-Dependent TLR4 Signaling in Macrophages, Front. Immunol. 9 (2018) 2. doi:10.3389/fimmu.2018.00002.

[203] K.L. Spiller, E.A. Wrona, S. Romero-Torres, I. Pallotta, P.L. Graney, C.E. Witherel, L.M. Panicker, R.A. Feldman, A.M. Urbanska, L. Santambrogio, G. Vunjak-Novakovic, D.O. Freytes, Differential gene expression in human, murine, and cell line-derived macrophages upon polarization, Exp. Cell Res. 347 (2016) 1–13. doi:10.1016/j.yexcr.2015.10.017.

[204] M.A. Rahman, P. Sobia, N. Gupta, L. Van Kaer, G. Das, Mycobacterium tuberculosis Subverts the TLR-2 - MyD88 Pathway to Facilitate Its Translocation into the Cytosol, PLoS One. 9 (2014) e86886. doi:10.1371/journal.pone.0086886.

[205] E. Mortaz, I.M. Adcock, P. Tabarsi, M.R. Masjedi, D. Mansouri, A.A. Velayati, J.L. Casanova, P.J. Barnes, Interaction of Pattern Recognition Receptors with Mycobacterium Tuberculosis, J Clin Immunol. (2014). doi:10.1007/s10875-014-0103-7.

[206] A.-R. Jang, J.-H. Choi, S.J. Shin, J.-H. Park, Mycobacterium tuberculosis ESAT6 induces IFN-β gene expression in Macrophages via TLRs-mediated signaling, Cytokine. 104 (2018) 104–109. doi:10.1016/J.CYTO.2017.10.006.

218

[207] A. Dorhoi, V. Yeremeev, G. Nouailles, J. Weiner, S. Jörg, E. Heinemann, D. Oberbeck-Müller, J.K. Knaul, A. Vogelzang, S.T. Reece, K. Hahnke, H.-J. Mollenkopf, V. Brinkmann, S.H.E. Kaufmann, Type I IFN signaling triggers immunopathology in tuberculosis-susceptible mice by modulating lung phagocyte dynamics., Eur. J. Immunol. 44 (2014) 2380–93. doi:10.1002/eji.201344219.

[208] E. Müller, M. Speth, P.F. Christopoulos, A. Lunde, A. Avdagic, I. Øynebråten, A. Corthay, Both Type I and Type II Interferons Can Activate Antitumor M1 Macrophages When Combined With TLR Stimulation, Front. Immunol. 9 (2018) 2520. doi:10.3389/fimmu.2018.02520.

[209] L. Moreira-Teixeira, J. Sousa, F.W. McNab, E. Torrado, F. Cardoso, H. Machado, F. Castro, V. Cardoso, J. Gaifem, X. Wu, R. Appelberg, A.G. Castro, A. O’Garra, M. Saraiva, Type I IFN Inhibits Alternative Macrophage Activation during Mycobacterium tuberculosis Infection and Leads to Enhanced Protection in the Absence of IFN-γ Signaling., J. Immunol. 197 (2016) 4714–4726. doi:10.4049/jimmunol.1600584.

[210] D. Ordway, M. Henao-Tamayo, M. Harton, G. Palanisamy, J. Troudt, C. Shanley, R.J. Basaraba, I.M. Orme, The hypervirulent Mycobacterium tuberculosis strain HN878 induces a potent TH1 response followed by rapid down-regulation, J Immunol. 179 (2007) 522–531. http://www.ncbi.nlm.nih.gov/pubmed/17579073.

[211] J. Chan, Y. Xing, R.S. Magliozzo, B.R. Bloom, Killing of virulent Mycobacterium tuberculosis by reactive nitrogen intermediates produced by activated murine macrophages, J Exp Med. 175 (1992) 1111–1122. http://www.ncbi.nlm.nih.gov/pubmed/1552282.

[212] Z. Huang, Q. Luo, Y. Guo, J. Chen, G. Xiong, Y. Peng, J. Ye, J. Li, Mycobacterium tuberculosis- Induced Polarization of Human Macrophage Orchestrates the Formation and Development of Tuberculous Granulomas In Vitro, PLoS One. 10 (2015) e0129744. doi:10.1371/journal.pone.0129744.

[213] H. Su, S. Zhu, L. Zhu, W. Huang, H. Wang, Z. Zhang, Y. Xu, Recombinant Lipoprotein Rv1016c Derived from Mycobacterium tuberculosis Is a TLR-2 Ligand that Induces Macrophages Apoptosis and Inhibits MHC II Antigen Processing, Front. Cell. Infect. Microbiol. 6 (2016) 147. doi:10.3389/fcimb.2016.00147.

[214] A. McBride, J. Konowich, P. Salgame, Host defense and recruitment of Foxp3+ T regulatory cells to the lungs in chronic Mycobacterium tuberculosis infection requires toll-like receptor 2., PLoS Pathog. 9 (2013) e1003397. doi:10.1371/journal.ppat.1003397.

[215] D.C. Ifrim, J.M. Bain, D.M. Reid, M. Oosting, I. Verschueren, N.A.R. Gow, J.H. van Krieken, G.D. Brown, B.-J. Kullberg, L.A.B. Joosten, J.W.M. van der Meer, F. Koentgen, L.P. Erwig, J. Quintin, M.G. Netea, Role of Dectin-2 for Host Defense against Systemic Infection with Candida glabrata, Infect. Immun. 82 (2014) 1064–1073. doi:10.1128/IAI.01189-13.

[216] Y. Zheng, J. Monty, R.J. Linhardt, Polysaccharide-based nanocomposites and their applications, Carbohydr Res. 405 (2015) 23–32. doi:10.1016/j.carres.2014.07.016.

[217] K. Varaprasad, K. Vimala, S. Ravindra, N. Narayana Reddy, G. Venkata Subba Reddy, K. Mohana Raju, Fabrication of silver nanocomposite films impregnated with curcumin for superior antibacterial applications, J Mater Sci Mater Med. 22 (2011) 1863–1872. doi:10.1007/s10856-011- 4369-5.

[218] B. Kang, T. Opatz, K. Landfester, F.R. Wurm, Carbohydrate nanocarriers in biomedical applications: functionalization and construction, Chem Soc Rev. 44 (2015) 8301–8325. doi:10.1039/c5cs00092k.

[219] Z. Yin, M. Comellas-Aragones, S. Chowdhury, P. Bentley, K. Kaczanowska, L. Benmohamed, J.C. Gildersleeve, M.G. Finn, X. Huang, Boosting immunity to small tumor-associated carbohydrates with bacteriophage qbeta capsids, ACS Chem Biol. 8 (2013) 1253–1262. doi:10.1021/cb400060x.

[220] A.S. Cordeiro, M.J. Alonso, M. de la Fuente, Nanoengineering of vaccines using natural polysaccharides, Biotechnol Adv. 33 (2015) 1279–1293. doi:10.1016/j.biotechadv.2015.05.010.

219

[221] Z. Ghotbi, A. Haddadi, S. Hamdy, R.W. Hung, J. Samuel, A. Lavasanifar, Active targeting of dendritic cells with mannan-decorated PLGA nanoparticles, J Drug Target. 19 (2011) 281–292. doi:10.3109/1061186x.2010.499463.

[222] Y. Phanse, B.R. Carrillo-Conde, A.E. Ramer-Tait, R. Roychoudhury, N.L. Pohl, B. Narasimhan, M.J. Wannemuehler, B.H. Bellaire, Functionalization of polyanhydride microparticles with di-mannose influences uptake by and intracellular fate within dendritic cells, Acta Biomater. 9 (2013) 8902– 8909. doi:10.1016/j.actbio.2013.06.024.

[223] M. Marradi, M. Martin-Lomas, S. Penades, Glyconanoparticles polyvalent tools to study carbohydrate-based interactions, Adv Carbohydr Chem Biochem. 64 (2010) 211–290. doi:10.1016/s0065-2318(10)64005-x.

[224] Y. Men, C. Thomasin, H.P. Merkle, B. Gander, G. Corradin, A single administration of tetanus toxoid in biodegradable microspheres elicits T cell and antibody responses similar or superior to those obtained with aluminum hydroxide, Vaccine. 13 (1995) 683–689.

[225] L.K. Petersen, A.E. Ramer-Tait, S.R. Broderick, C.S. Kong, B.D. Ulery, K. Rajan, M.J. Wannemuehler, B. Narasimhan, Activation of innate immune responses in a pathogen-mimicking manner by amphiphilic polyanhydride nanoparticle adjuvants, Biomaterials. 32 (2011) 6815–6822. doi:10.1016/j.biomaterials.2011.05.063.

[226] H. Hillaireau, P. Couvreur, Nanocarriers’ entry into the cell: relevance to drug delivery, Cell Mol Life Sci. 66 (2009) 2873–2896. doi:10.1007/s00018-009-0053-z.

[227] S.T. Reddy, A.J. van der Vlies, E. Simeoni, V. Angeli, G.J. Randolph, C.P. O’Neil, L.K. Lee, M.A. Swartz, J.A. Hubbell, Exploiting lymphatic transport and complement activation in nanoparticle vaccines, Nat Biotechnol. 25 (2007) 1159–1164. doi:10.1038/nbt1332.

[228] C. Lautenschlager, C. Schmidt, C.M. Lehr, D. Fischer, A. Stallmach, PEG-functionalized microparticles selectively target inflamed mucosa in inflammatory bowel disease, Eur J Pharm Biopharm. 85 (2013) 578–586. doi:10.1016/j.ejpb.2013.09.016.

[229] V.E. Schijns, E.C. Lavelle, Trends in vaccine adjuvants, Expert Rev Vaccines TA - Expert Rev, Vaccines. 10 (2011) 539–550.

[230] K.S. Devi, B. Sahoo, B. Behera, T.K. Maiti, Nanoparticle and polysaccharide conjugate: A potential candidate vaccine to improve immunological stimuli, Int J Biol Macromol. 72 (2015) 1254–1264. doi:10.1016/j.ijbiomac.2014.10.026.

[231] S. Venkataraman, J.L. Hedrick, Z.Y. Ong, C. Yang, P.L. Ee, P.T. Hammond, Y.Y. Yang, The effects of polymeric nanostructure shape on drug delivery, Adv Drug Deliv Rev. 63 (2011) 1228–1246. doi:10.1016/j.addr.2011.06.016.

[232] J.A. Champion, S. Mitragotri, Shape induced inhibition of phagocytosis of polymer particles, Pharm Res. 26 (2009) 244–249. doi:10.1007/s11095-008-9626-z.

[233] K. Niikura, T. Matsunaga, T. Suzuki, S. Kobayashi, H. Yamaguchi, Y. Orba, A. Kawaguchi, H. Hasegawa, K. Kajino, T. Ninomiya, K. Ijiro, H. Sawa, Gold nanoparticles as a vaccine platform: influence of size and shape on immunological responses in vitro and in vivo, ACS Nano. 7 (2013) 3926–3938. doi:10.1021/nn3057005.

[234] S. Kawakami, J. Wong, A. Sato, Y. Hattori, F. Yamashita, M. Hashida, Biodistribution characteristics of mannosylated, fucosylated, and galactosylated liposomes in mice, Biochim Biophys Acta. 1524 (2000) 258–265.

[235] C. Lawlor, M.P. O’Sullivan, N. Sivadas, S. O’Leary, P.J. Gallagher, J. Keane, S.A. Cryan, The application of high-content analysis in the study of targeted particulate delivery systems for intracellular drug delivery to alveolar macrophages, Mol Pharm. 8 (2011) 1100–1112. doi:10.1021/mp1004178.

[236] M. Roser, D. Fischer, T. Kissel, Surface-modified biodegradable albumin nano- and microspheres. 220

II: effect of surface charges on in vitro phagocytosis and biodistribution in rats, Eur J Pharm Biopharm. 46 (1998) 255–263.

[237] D.R. Getts, R.L. Terry, M.T. Getts, C. Deffrasnes, M. Muller, C. van Vreden, T.M. Ashhurst, B. Chami, D. McCarthy, H. Wu, J. Ma, A. Martin, L.D. Shae, P. Witting, G.S. Kansas, J. Kuhn, W. Hafezi, I.L. Campbell, D. Reilly, J. Say, L. Brown, M.Y. White, S.J. Cordwell, S.J. Chadban, E.B. Thorp, S. Bao, S.D. Miller, N.J. King, Therapeutic inflammatory monocyte modulation using immune-modifying microparticles, Sci Transl Med. 6 (2014) 219ra7. doi:10.1126/scitranslmed.3007563.

[238] S.A. Abouelmagd, Y.J. Ku, Y. Yeo, Low molecular weight chitosan-coated polymeric nanoparticles for sustained and pH-sensitive delivery of paclitaxel, J Drug Target. 23 (2015) 725–735. doi:10.3109/1061186x.2015.1054829.

[239] M.P. Monopoli, C. Aberg, A. Salvati, K.A. Dawson, Biomolecular coronas provide the biological identity of nanosized materials, Nat Nanotechnol. 7 (2012) 779–786. doi:10.1038/nnano.2012.207.

[240] M. Azhdarzadeh, A.A. Saei, S. Sharifi, M.J. Hajipour, A.M. Alkilany, M. Sharifzadeh, F. Ramazani, S. Laurent, A. Mashaghi, M. Mahmoudi, Nanotoxicology: advances and pitfalls in research methodology, Nanomedicine (Lond). 10 (2015) 2931–2952. doi:10.2217/nnm.15.130.

[241] M. Moros, B. Hernaez, E. Garet, J.T. Dias, B. Saez, V. Grazu, A. Gonzalez-Fernandez, C. Alonso, J.M. de la Fuente, Monosaccharides versus PEG-functionalized NPs: influence in the cellular uptake, ACS Nano. 6 (2012) 1565–1577. doi:10.1021/nn204543c.

[242] G.A. Barhate, S.M. Gaikwad, S.S. Jadhav, V.B. Pokharkar, Structure function attributes of gold nanoparticle vaccine association: effect of particle size and association temperature, Int J Pharm. 471 (2014) 439–448. doi:10.1016/j.ijpharm.2014.06.002.

[243] D. Safari, M. Marradi, F. Chiodo, H.A. Th Dekker, Y. Shan, R. Adamo, S. Oscarson, G.T. Rijkers, M. Lahmann, J.P. Kamerling, S. Penades, H. Snippe, Gold nanoparticles as carriers for a synthetic Streptococcus pneumoniae type 14 conjugate vaccine, Nanomedicine (Lond). 7 (2012) 651–662. doi:10.2217/nnm.11.151.

[244] M. Roskamp, S. Enders, F. Pfrengle, S. Yekta, V. Dekaris, J. Dernedde, H.U. Reissig, S. Schlecht, Multivalent interaction and selectivities in selectin binding of functionalized gold colloids decorated with carbohydrate mimetics, Org Biomol Chem. 9 (2011) 7448–7456. doi:10.1039/c1ob05583f.

[245] A.O. Adebisi, B.R. Conway, Lectin-conjugated microspheres for eradication of Helicobacter pylori infection and interaction with mucus, Int J Pharm. 470 (2014) 28–40. doi:10.1016/j.ijpharm.2014.04.070.

[246] A. Palma, L.A. Alvarez, D. Scholz, D.O. Frimannsson, M. Grossi, S.J. Quinn, D.F. O’Shea, Cellular uptake mediated off/on responsive near-infrared fluorescent nanoparticles, J Am Chem Soc. 133 (2011) 19618–19621. doi:10.1021/ja208086e.

[247] M.J. Marin, A. Rashid, M. Rejzek, S.A. Fairhurst, S.A. Wharton, S.R. Martin, J.W. McCauley, T. Wileman, R.A. Field, D.A. Russell, Glyconanoparticles for the plasmonic detection and discrimination between human and avian influenza virus, Org Biomol Chem. 11 (2013) 7101– 7107. doi:10.1039/c3ob41703d.

[248] B. Carrillo-Conde, E.H. Song, A. Chavez-Santoscoy, Y. Phanse, A.E. Ramer-Tait, N.L. Pohl, M.J. Wannemuehler, B.H. Bellaire, B. Narasimhan, Mannose-functionalized “pathogen-like” polyanhydride nanoparticles target C-type lectin receptors on dendritic cells, Mol Pharm. 8 (2011) 1877–1886. doi:10.1021/mp200213r.

[249] R.P. Brinas, A. Sundgren, P. Sahoo, S. Morey, K. Rittenhouse-Olson, G.E. Wilding, W. Deng, J.J. Barchi Jr., Design and synthesis of multifunctional gold nanoparticles bearing tumor-associated glycopeptide antigens as potential cancer vaccines, Bioconjug Chem. 23 (2012) 1513–1523. doi:10.1021/bc200606s.

221

[250] M. V Bell, R.J. Henderson, J.R. Sargent, The role of polyunsaturated fatty acids in fish, Comp Biochem Physiol B. 83 (1986) 711–719.

[251] K. Todar, Mycobacterium tuberculosis and Tuberculosis, Todar’s Online Textb. Bacteriol. (2012) 1. http://textbookofbacteriology.net/tuberculosis.html (accessed August 23, 2019).

[252] T. Masuko, A. Minami, N. Iwasaki, T. Majima, S.-I. Nishimura, Y.C. Lee, Carbohydrate analysis by a phenol-sulfuric acid method in microplate format., Anal. Biochem. 339 (2005) 69–72. doi:10.1016/j.ab.2004.12.001.

[253] A.C.A. Roque, A. Bicho, I.L. Batalha, A.S. Cardoso, A. Hussain, Biocompatible and bioactive gum Arabic coated iron oxide magnetic nanoparticles, J. Biotechnol. 144 (2009) 313–320. doi:10.1016/j.jbiotec.2009.08.020.

[254] I. Khan, C.N. Tango, S. Miskeen, D.-H. Oh, Evaluation of nisin-loaded chitosan-monomethyl fumaric acid nanoparticles as a direct food additive, Carbohydr. Polym. 184 (2018) 100–107. doi:10.1016/j.carbpol.2017.11.034.

[255] C.-M. Lee, H.-J. Jeong, S.-L. Kim, E.-M. Kim, D.W. Kim, S.T. Lim, K.Y. Jang, Y.Y. Jeong, J.-W. Nah, M.- H. Sohn, SPION-loaded chitosan–linoleic acid nanoparticles to target hepatocytes, Int. J. Pharm. 371 (2009) 163–169. doi:10.1016/j.ijpharm.2008.12.021.

[256] H. Onishi, H. Takahashi, M. Yoshiyasu, Y. Machida, Preparation and in vitro properties of N- succinylchitosan- or carboxymethylchitin-mitomycin C conjugate microparticles with specified size., Drug Dev. Ind. Pharm. 27 (2001) 659–67. http://www.ncbi.nlm.nih.gov/pubmed/11694013.

[257] M.J.E. Fischer, Amine Coupling Through EDC/NHS: A Practical Approach, in: Methods Mol. Biol., 2010: pp. 55–73. doi:10.1007/978-1-60761-670-2_3.

[258] I. Nallamuthu, A. Parthasarathi, F. Khanum, Thymoquinone-loaded PLGA nanoparticles: antioxidant and anti-microbial properties, Int. Curr. Pharm. J. 2 (2013) 202–207. http://www.icpjonline.com/documents/Vol2Issue12/02.pdf.

[259] A. Arbues, G. Lugo-Villarino, O. Neyrolles, C. Guilhot, C. Astarie-Dequeker, Playing hide-and-seek with host macrophages through the use of mycobacterial cell envelope phthiocerol dimycocerosates and phenolic glycolipids., Front. Cell. Infect. Microbiol. 4 (2014) 173. doi:10.3389/fcimb.2014.00173.

[260] M. DAFFE, C. LACAVE, M.-A. LANEELLE, G. LANEELLE, Structure of the major triglycosyl phenol- phthiocerol of Mycobacterium tuberculosis (strain Canetti), Eur. J. Biochem. 167 (1987) 155–160. doi:10.1111/j.1432-1033.1987.tb13317.x.

[261] M. Daffé, M.A. Lanéelle, C. Lacave, G. Lanéelle, Monoglycosyldiacylphenol-phthiocerol of Mycobacterium tuberculosis and Mycobacterium bovis., Biochim. Biophys. Acta. 958 (1988) 443– 9. doi:10.1016/0005-2760(88)90230-5.

[262] E. Perez, P. Constant, A. Lemassu, F. Laval, M. Daffe, C. Guilhot, Characterization of three glycosyltransferases involved in the biosynthesis of the phenolic glycolipid antigens from the Mycobacterium tuberculosis complex, J Biol Chem. 279 (2004) 42574–42583. doi:10.1074/jbc.M406246200.

[263] E. Pérez, P. Constant, F. Laval, A. Lemassu, M.-A. Lanéelle, M. Daffé, C. Guilhot, Molecular dissection of the role of two methyltransferases in the biosynthesis of phenolglycolipids and phthiocerol dimycoserosate in the Mycobacterium tuberculosis complex., J. Biol. Chem. 279 (2004) 42584–92. doi:10.1074/jbc.M406134200.

[264] R. Simeone, G. Huet, P. Constant, W. Malaga, A. Lemassu, F. Laval, M. Daffé, C. Guilhot, C. Chalut, Functional characterisation of three o-methyltransferases involved in the biosynthesis of phenolglycolipids in Mycobacterium tuberculosis., PLoS One. 8 (2013) e58954. doi:10.1371/journal.pone.0058954.

[265] S.T. Cole, R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris, S. V. Gordon, K. Eiglmeier, S. Gas, 222

C.E. Barry, F. Tekaia, K. Badcock, D. Basham, D. Brown, T. Chillingworth, R. Connor, R. Davies, K. Devlin, T. Feltwell, S. Gentles, N. Hamlin, S. Holroyd, T. Hornsby, K. Jagels, A. Krogh, J. McLean, S. Moule, L. Murphy, K. Oliver, J. Osborne, M.A. Quail, M.-A. Rajandream, J. Rogers, S. Rutter, K. Seeger, J. Skelton, R. Squares, S. Squares, J.E. Sulston, K. Taylor, S. Whitehead, B.G. Barrell, Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence, Nature. 393 (1998) 537–544. doi:10.1038/31159.

[266] W. Malaga, P. Constant, D. Euphrasie, A. Cataldi, M. Daffé, J.-M. Reyrat, C. Guilhot, Deciphering the genetic bases of the structural diversity of phenolic glycolipids in strains of the Mycobacterium tuberculosis complex., J. Biol. Chem. 283 (2008) 15177–84. doi:10.1074/jbc.M710275200.

[267] H.R.H. Elsaidi, T.L. Lowary, Inhibition of Cytokine Release by Mycobacterium tuberculosis Phenolic Glycolipid Analogues, ChemBioChem. 15 (2014) 1176–1182. doi:10.1002/cbic.201402001.

[268] D. Sinsimer, G. Huet, C. Manca, L. Tsenova, M.S. Koo, N. Kurepina, B. Kana, B. Mathema, S.A.E. Marras, B.N. Kreiswirth, C. Guilhot, G. Kaplan, The phenolic glycolipid of Mycobacterium tuberculosis differentially modulates the early host cytokine response but does not in itself confer hypervirulence, Infect. Immun. 76 (2008) 3027–3036. doi:10.1128/IAI.01663-07.

[269] G. Tabouret, C. Astarie-Dequeker, C. Demangel, W. Malaga, P. Constant, A. Ray, N. Honoré, N.F. Bello, E. Perez, M. Daffé, C. Guilhot, Mycobacterium leprae Phenolglycolipid-1 Expressed by Engineered M. bovis BCG Modulates Early Interaction with Human Phagocytes, PLoS Pathog. 6 (2010) e1001159. doi:10.1371/journal.ppat.1001159.

[270] H.R.H. Elsaidi, T.L. Lowary, Effect of phenolic glycolipids from Mycobacterium kansasii on proinflammatory cytokine release. A structure-activity relationship study., Chem. Sci. 6 (2015) 3161–3172. doi:10.1039/c4sc04004j.

[271] C. Brikos, L.A.J. O’Neill, Signalling of Toll-Like Receptors, in: 2008: pp. 21–50. doi:10.1007/978-3- 540-72167-3_2.

[272] L.S. Schlesinger, M.A. Horwitz, Phenolic glycolipid-1 of mycobacterium leprae binds complement component C3 in serum and mediates phagocytosis by human monocytes, J. Exp. Med. 174 (1991) 1031–1038. doi:10.1084/jem.174.5.1031.

[273] L.S. Schlesinger, M.A. Horwitz, Phagocytosis of Mycobacterium leprae by human monocyte- derived macrophages is mediated by complement receptors CR1 (CD35), CR3 (CD11b/CD18), and CR4 (CD11c/CD18) and IFN-gamma activation inhibits complement receptor function and phagocytosis of this bacterium., J. Immunol. 147 (1991) 1983–94. http://www.ncbi.nlm.nih.gov/pubmed/1679838 (accessed December 18, 2019).

[274] J. Michałowicz, W. Duda, Polish journal of environmental studies., HARD, n.d. http://www.pjoes.com/Phenols-Sources-and-Toxicity,87995,0,2.html.

[275] C.D. Selassie, T. V DeSoyza, M. Rosario, H. Gao, C. Hansch, Phenol toxicity in leukemia cells: a radical process?, Chem. Biol. Interact. 113 (1998) 175–90. doi:10.1016/s0009-2797(98)00027-1.

[276] K.K.K. Leung, B.H. Shilton, Binding of DNA-Intercalating Agents to Oxidized and Reduced Quinone Reductase 2, Biochemistry. 54 (2015) 7438–7448. doi:10.1021/acs.biochem.5b00884.

[277] M. Mohamadi, D. Afzali, S. Esmaeili-Mahani, A. Mostafavi, M. Torkzadeh-Mahani, Spectroscopic and electrochemical studies of the interaction between oleuropein, the major bio-phenol in olives, and salmon sperm DNA, Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 148 (2015) 260– 265. doi:10.1016/j.saa.2015.03.123.

[278] A.M. Grabiec, T. Hussell, The role of airway macrophages in apoptotic cell clearance following acute and chronic lung inflammation, Semin. Immunopathol. 38 (2016) 409–423. doi:10.1007/s00281-016-0555-3.

[279] J.A. Philips, J.D. Ernst, Tuberculosis pathogenesis and immunity, Annu Rev Pathol. 7 (2012) 353–

223

384. doi:10.1146/annurev-pathol-011811-132458.

[280] M. Pai, M.A. Behr, D. Dowdy, K. Dheda, M. Divangahi, C.C. Boehme, A. Ginsberg, S. Swaminathan, M. Spigelman, H. Getahun, D. Menzies, M. Raviglione, Tuberculosis, Nat. Rev. Dis. Prim. 2 (2016) 16076. doi:10.1038/nrdp.2016.76.

[281] R.L. Hunter, The Pathogenesis of Tuberculosis: The Early Infiltrate of Post-primary (Adult Pulmonary) Tuberculosis: A Distinct Disease Entity, Front. Immunol. 9 (2018) 2108. doi:10.3389/fimmu.2018.02108.

[282] S.B. Cohen, B.H. Gern, J.L. Delahaye, K.N. Adams, C.R. Plumlee, J.K. Winkler, D.R. Sherman, M.Y. Gerner, K.B. Urdahl, Alveolar Macrophages Provide an Early Mycobacterium tuberculosis Niche and Initiate Dissemination., Cell Host Microbe. 24 (2018) 439-446.e4. doi:10.1016/j.chom.2018.08.001.

[283] J.A. Awuh, T.H. Flo, Molecular basis of mycobacterial survival in macrophages, Cell. Mol. Life Sci. 74 (2017) 1625–1648. doi:10.1007/s00018-016-2422-8.

[284] M.B. Ryndak, S. Laal, Mycobacterium tuberculosis Primary Infection and Dissemination: A Critical Role for Alveolar Epithelial Cells., Front. Cell. Infect. Microbiol. 9 (2019) 299. doi:10.3389/fcimb.2019.00299.

[285] E. Jouanguy, S. Lamhamedi-Cherradi, D. Lammas, S.E. Dorman, M.C. Fondanèche, S. Dupuis, R. Döffinger, F. Altare, J. Girdlestone, J.F. Emile, H. Ducoulombier, D. Edgar, J. Clarke, V.A. Oxelius, M. Brai, V. Novelli, K. Heyne, A. Fischer, S.M. Holland, D.S. Kumararatne, R.D. Schreiber, J.L. Casanova, A human IFNGR1 small deletion hotspot associated with dominant susceptibility to mycobacterial infection., Nat. Genet. 21 (1999) 370–8. doi:10.1038/7701.

[286] R. Pasula, W.J. Martin, B.R. Kesavalu, M.Y. Abdalla, B.E. Britigan, Passive transfer of interferon-γ over-expressing macrophages enhances resistance of SCID mice to Mycobacterium tuberculosis infection, Cytokine. 95 (2017) 70–79. doi:10.1016/j.cyto.2017.02.009.

[287] N.R. Gandhi, P. Nunn, K. Dheda, H.S. Schaaf, M. Zignol, D. van Soolingen, P. Jensen, J. Bayona, Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of tuberculosis, Lancet. 375 (2010) 1830–1843. doi:10.1016/S0140-6736(10)60410-2.

[288] P. Andersen, T.J. Scriba, Moving tuberculosis vaccines from theory to practice, Nat. Rev. Immunol. 19 (2019) 550–562. doi:10.1038/s41577-019-0174-z.

[289] A. Khan, P. Bakhru, S. Saikolappan, K. Das, E. Soudani, C.R. Singh, J.L. Estrella, D. Zhang, C. Pasare, Y. Ma, J. Sun, J. Wang, R.L. Hunter, N. Tony Eissa, S. Dhandayuthapani, C. Jagannath, An autophagy-inducing and TLR-2 activating BCG vaccine induces a robust protection against tuberculosis in mice, Npj Vaccines. 4 (2019) 34. doi:10.1038/s41541-019-0122-8.

[290] H. Ning, L. Wang, J. Zhou, Y. Lu, J. Kang, T. Ding, L. Shen, Z. Xu, Y. Bai, Recombinant BCG With Bacterial Signaling Molecule Cyclic di-AMP as Endogenous Adjuvant Induces Elevated Immune Responses After Mycobacterium tuberculosis Infection, Front. Immunol. 10 (2019) 1519. doi:10.3389/fimmu.2019.01519.

[291] L. Moreira-Teixeira, K. Mayer-Barber, A. Sher, A. O’Garra, Type I interferons in tuberculosis: Foe and occasionally friend., J. Exp. Med. 215 (2018) 1273–1285. doi:10.1084/jem.20180325.

[292] F.W. McNab, J. Ewbank, A. Howes, L. Moreira-Teixeira, A. Martirosyan, N. Ghilardi, M. Saraiva, A. O’Garra, Type I IFN induces IL-10 production in an IL-27-independent manner and blocks responsiveness to IFN-γ for production of IL-12 and bacterial killing in Mycobacterium tuberculosis-infected macrophages., J. Immunol. 193 (2014) 3600–12. doi:10.4049/jimmunol.1401088.

[293] A. Howes, C. Taubert, S. Blankley, N. Spink, X. Wu, C.M. Graham, J. Zhao, M. Saraiva, P. Ricciardi- Castagnoli, G.J. Bancroft, A. O’Garra, Differential Production of Type I IFN Determines the Reciprocal Levels of IL-10 and Proinflammatory Cytokines Produced by C57BL/6 and BALB/c

224

Macrophages., J. Immunol. 197 (2016) 2838–53. doi:10.4049/jimmunol.1501923.

[294] S.A. Khader, M. Divangahi, W. Hanekom, P.C. Hill, M. Maeurer, K.W. Makar, K.D. Mayer-Barber, M.M. Mhlanga, E. Nemes, L.S. Schlesinger, R. van Crevel, R. Vankalayapati, R.J. Xavier, M.G. Netea, Bill and Melinda Gates Foundation Collaboration for TB Vaccine Discovery Innate Immunity Working Group18, Targeting innate immunity for tuberculosis vaccination, J. Clin. Invest. 129 (2019) 3482–3491. doi:10.1172/JCI128877.

[295] P. Steigler, A.J. Verrall, J.R. Kirman, Beyond memory T cells: mechanisms of protective immunity to tuberculosis infection, Immunol. Cell Biol. 97 (2019) 647–655. doi:10.1111/imcb.12278.

225

Appendix

226

Biochemical Pharmacology 146 (2017) 23–41

Contents lists available at ScienceDirect

Biochemical Pharmacology

journal homepage: www.elsevier.com/locate/biochempharm

Research update Therapeutic potential of carbohydrates as regulators of macrophage activation ⇑ Mimmi L.E. Lundahl a,b, Eoin M. Scanlan b, Ed C. Lavelle a, a Adjuvant Research Group, School of Biochemistry and Immunology, Trinity Biomedical Sciences Institute, Trinity College Dublin, D02 R590 Dublin 2, Ireland b School of Chemistry and Trinity Biomedical Sciences Institute, Trinity College, Pearse St, Dublin 2, Ireland article info abstract

Article history: It is well established for a broad range of disease states, including cancer and Mycobacterium tuberculosis Received 4 July 2017 infection, that pathogenesis is bolstered by polarisation of macrophages towards an anti-inflammatory Accepted 6 September 2017 phenotype, known as M2. As these innate immune cells are relatively long-lived, their re-polarisation Available online 8 September 2017 to pro-inflammatory, phagocytic and bactericidal ‘‘classically activated” M1 macrophages is an attractive therapeutic approach. On the other hand, there are scenarios where the resolving inflammation, wound Chemical compounds cited in this article: healing and tissue remodelling properties of M2 macrophages are beneficial – for example the successful Chitin (PubChem CIDs: 6857375, 5288898, introduction of biomedical implants. Although there are numerous endogenous and exogenous factors 70824173 & 71314665) that have an impact on the macrophage polarisation spectrum, this review will focus specifically on Chitosan (PubChem CIDs: 71853 & 21896651) prominent macrophage-modulating carbohydrate motifs with a view towards highlighting structure– 2-Deoxy Glucose (PubChem CID: 40) function relationships and therapeutic potential. b-Glucan (PubChem CIDs: 439262, Ó 2017 Elsevier Inc. All rights reserved. 46173706 & 92024379) Lactic acid (PubChem CID: 612) Mannan (PubChem CID: 870) N-acetyl neuraminic acid (PubChem CIDs: 445063 & 439197) Sialic Acid (PubChem CID: 906) Succinate (PubChem CID: 160419)

Keywords: Macrophages Carbohydrates Polarisation Inflammation Immunomodulation

Abbreviations: 2-DG, 2-deoxy glucose; Akt, protein kinase B; ARDS, acute respiratory distress syndrome; Arg, arginase; BALF, broncho-alveolar lavage fluid; BCG, Bacillus Calmette–Guérin; CD, cluster of differentiation; cGAS, cyclic GMP-AMP synthase; COX-2, cyclooxygenase-2; DAMP, danger-associated molecular pattern; DAP12, DNAX activating protein 12 kDa; DC, dendritic cell; Dectin, DC-associated C-type lectin; DAT, di-O-acylated ; DesTCR, Désiré T cell receptor; ETC, electron transport chain; EVLP, ex vivo lung perfusion; FccR, Fc receptor for immunoglobulin G; Fizz1, found in inflammatory zone 1; GAG, ; GLUT1, glucose transporter 1; He3K4me3, histone H3 lysine 4 trimethylation; HIF-1a, hypoxia-inducible factor 1a; IFNc, interferon c; IL, interleukin; IRF, interferon regulatory factor; JAK, Janus kinase; LDH, lactate dehydrogenase; LLC, Lewis lung carcinoma; ManLAM, mannose-capped lipoarabinomannan; MCP1, monocyte chemotactic protein 1 ; M-CSFR, macrophage colony-stimulating factor receptor; Mgl, macrophage galactose-type C-type lectin; MHC, major histocompatibility complex; MHC II, MHC class II; Mincle, macrophage- inducible C-type lectin; MIP, macrophage inflammatory protein; MR, mannose receptor; Mtb, Mycobacterium tuberculosis; mTOR, mechanistic target of rapamycin; mTORC, mTOR complex; Neu5Ac, N-acetyl neuraminic acid; NF-jB, nuclear factor kappa-light-chain-enhancer of activated B cells; NO, nitric oxide; NOD2, nucleotide-binding oligomerization domain-containing protein 2; oxphos, oxidative phosphorylation; Pam3Cys, tripalmitoyl-S-glyceryl-cysteine; PAMP, pathogen-associated molecular pattern; PD-L2, programmed cell death 1 ligand 2; PGE-2, prostaglandin E2; PIM, phosphatidyl-myo-inositol mannosides; PLGA, poly lactic-co-glycolic acid; RNS, reactive nitrogen species; ROS, reactive oxygen species; PPARc, peroxisome proliferator-activated receptor c; PRR, pattern recognition receptor; SHP, SH2 domain-containing protein tyrosine phosphatase; Siglec, sialic acid-binding immunoglobulin-like lectin; STAT, signal transducer and activator of transcription; STING, stimulator of interferon genes; TAM, tumour associated macrophages; TAT, tri-O-acylated trehalose; TDB, trehalose-6,6-dibehenate; TDM, trehalose-6,6-dimycolate; TGFb, transforming growth factor b; TLR, toll- like receptor; TNFa, tumour necrosis factor a; VEGFa, vascular endothelial growth factor a; Ym1, chitinase-like 3. ⇑ Corresponding author. E-mail address: [email protected] (E.C. Lavelle). http://dx.doi.org/10.1016/j.bcp.2017.09.003 0006-2952/Ó 2017 Elsevier Inc. All rights reserved. 24 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

Contents

1. Introduction ...... 24 2. Sialic acids mediate immunosuppressive responses ...... 24 3. Mannose engagement with cells can either promote or inhibit immune responses depending on the specific context ...... 27 3.1. The dichotomy of a-linked mannosyl motifs ...... 27 3.1.1. ManLAM attenuates M1 polarisation ...... 28 3.1.2. a-Mannan enhances pro-inflammatory responses ...... 29 4. Glucose based polymers ...... 29 4.1. The M1 re-polarisation and training opportunities with b-glucan...... 29 4.2. The physicochemical properties of chitin and its derivatives determine their immunomodulatory properties ...... 30 4.3. De-acetylation causes chitosan to activate the NLRP3 inflammasome ...... 30 5. Immunometabolism: a requisite for polarisation...... 31 5.1. 2-Deoxy glucose ...... 31 5.2. Succinate ...... 31 5.3. Lactate ...... 31 6. Discussion: the therapeutic potential of macrophage-activating carbohydrates ...... 32 7. Conclusion ...... 37 Conflict of interest disclosure...... 37 Acknowledgements ...... 37 References ...... 37

1. Introduction markers such as the macrophage mannose receptor (MR) [8],as well as up-regulation of proteins such as chitinase-like 3-1 Macrophages are mononuclear phagocytic and antigen present- (Ym1), found in inflammatory zone-1 (Fizz1) and arginase (Arg) ing cells of the innate immune system. Due to their presence in [5,9]. Whilst these established markers are convenient, macro- essentially all tissues, these cells have a key role in responding to phages can concurrently display characteristic M1 and M2 markers exogenous and endogenous factors, from products of tissue dam- [10,11]. Apart from cytokines, there are numerous additional sig- age and pathogenic infections, to the introduction of biomaterials nals, including carbohydrates, that can influence macrophage and tumorigenesis. A significant feature of macrophages is polarisation towards either extreme. As macrophage responses dynamic plasticity, expressed by their ability to polarise towards are a key feature in cancer [12], fungal [13] and Mycobacterium distinct activation states [1]. As shown in Fig. 1, the two ‘‘ex- tuberculosis [10,14] infection, as well as the successful introduction tremes” of these states are the ‘‘classically activated” M1 macro- of implants [15,16], this review will primarily focus on examples of phages that propagate inflammation, have bactericidal activity immunomodulatory carbohydrate epitopes in these clinically rele- and are highly phagocytic [1,2], and the ‘‘alternatively activated” vant settings. M2 macrophages, which enhance allergic responses, resolve Carbohydrates are ubiquitous in all organisms. They are promi- inflammation and induce tissue remodelling [3]. Whilst the M1 nently present on cell surfaces [17] and it is estimated that as and M2 simplification is useful, it is important to bear in mind that much as 1% of the human genome is dedicated to glycosylation the reality is a broad spectrum of differentiation states, continu- [18]. The human glycome is primarily built from nine monosaccha- ously regulated by a myriad of signals from the microenvironment rides, which may appear to limit diversity [19]. However, due to [1]. This spectrum is an argument in favour of defining macro- the numerous stereogenic centres, the possibility of either a-or phages based on the stimuli used to differentiate them – a system b-linkages to any of four hydroxyl functional groups, as well as that is also regularly employed. However, in this review, the M1 forming either linear or branched structures – as few as three dif- and M2 extremes will be used to aid clarity. Due to their relatively ferent monosaccharides can construct more than 1,000 distinct long lifespan, macrophages that have attained a certain differenti- trisaccharides [19,20]. Carbohydrates are thus a source of diverse ation state can furthermore be re-polarised – a fundamental neces- information, the decoding of which is important as a means of dis- sity for instance in tissue healing, and an exciting avenue for tinguishing self from non-self, and for tailoring immune responses therapeutic discovery [1,4]. [21–23]. Characterising these immunomodulatory properties The current dogma suggests that M1 polarisation occurs in offers great potential for realising therapeutic applications. response to signalling downstream of cytokines such as tumour necrosis factor-a (TNFa) and interferon-c (IFNc) in concert with 2. Sialic acids mediate immunosuppressive responses the recognition of either endogenous or exogenous danger mole- cules [3]. This combination induces phagocytic, antigen presenting Eukaryotic cells are richly coated in sialic acids (Fig. 2): struc- M1 macrophages that secrete inflammatory cytokines including turally diverse nine-carbon monosaccharides that are negatively interleukin (IL)-1b, IL-12, IL-6, IL-23 and TNFa, as well as secretion charged at physiological pH [24–26]. Derived from neuraminic of proteolytic enzymes and the production of reactive oxygen and acid, these sugars are composed of a six-membered pyranose ring, nitrogen species (ROS and RNS respectively) [1,3]. These cells play with a three-carbon side chain and a carboxylic acid group at the key roles in protection against bacterial, fungal and viral pathogens anomeric position, which can be either a-orb-oriented [25,26]. [5]. By contrast, the cytokines IL-4, IL-13, IL-10 and transforming Sialic acids are important for distinguishing ‘‘self” from ‘‘non- growth factor (TGF)b stimulate M2 differentiation [3,6]. M2 macro- self” and are detected by sialic acid-binding proteins; prominently phages are poor antigen presenting cells that dampen inflamma- by sialic acid-binding immunoglobulin-like lectins (Siglecs) tion, in part by secreting cytokines including IL-10 and TGFb, [24,25]. This transmembrane receptor family, of which there are promote angiogenesis and scavenge debris [1,4,7]. M2 macro- 14 human and nine murine members [27], is a part of the phages can additionally be identified by their expression of surface immunoglobulin superfamily-type (I-type) lectins, which is in M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 25

Fig. 1. Polarisation towards either M1 or M2 results in distinct functional roles. Top schematic demonstrates typical stimuli for polarisation towards either the M1 or the M2 phenotype, as indicated by the blue arrows. Labelled M1 and M2 macrophages are shown with examples of surface markers and secreted cytokines/effectors (an increase of which is shown by red and green arrows respectively). Abbreviations: DAMPs, danger-associated molecular patterns; MR, mannose receptor; PAMPs, pathogen-associated molecular patterns; RNS, reactive nitrogen species; ROS, reactive oxygen species. Servier Medical art Powerpoint image bank was used to construct this diagram. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) itself a subset of the C-type (calcium dependent) superfamily of lated with tumour metastasis [30] and mediates immunosuppres- lectins, and are found on numerous immune cells including macro- sive effects on immune cells, including macrophages via Siglec phages [8,25,28]. Because of their immunomodulatory properties, signalling [29]. it is not surprising that sialic acids have been implicated in the Siglec signalling can be mediated by distinct pathways, depend- pathogenesis of many disease states, an important example of ing on the siglec in question. The presence of a basic amino acid which is cancer [1,26,29]. Elevated expression of sialic acids on residue in the transmembrane region facilitates the association of the surface of cancerous cells (hypersialylation) has been corre- some siglecs with DNAX activating protein 12 kDa (DAP12), which 26 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

Fig. 2. Structure of the most common sialic acid: N-acetyl neuraminic acid (Neu5Ac, in black), a-2,3-, a-2,6- and a-2,8- linked to either galactose (grey) or to Neu5Ac, and a schematic of a mammalian cell membrane, emphasising the presence of sialic acids on its surface. Ac: acetyl group [179,180]. possesses an immunoreceptor tyrosine-based activation motif syndrome (ARDS) and a human ex vivo lung perfusion (EVLP) model (ITAM) – this association in turn allows further signal transduction – inducing a higher level of IL-10 secretion in both cases [37]. via spleen tyrosine kinase (SYK) recruitment and activation [8,27]. Similar results were obtained when sialic acid expression on Inhibitory signals, on the other hand, involve the phosphorylation Leishmania donovani promastigotes was investigated [31]. Enzy- of tyrosine-based motifs on the cytoplasmic tail of the Siglecs matic removal of sialic acids from the promastigote surface led themselves, the most common being the immunoreceptor to enhanced macrophage production of ROS and RNS, as well as tyrosine-based inhibitory motifs (ITIMS) – the phosphorylation of increased expression of inducible nitric oxide synthase (iNOS), which results in high affinity interaction with the SH2 domain- characteristics of M1 macrophages [31]. By contrast, the presence containing protein tyrosine phosphatases, SHP-1 and SHP-2 of a-2,6 sialic acids on the promastigote surface induced more [8,27,31]. These phosphatases become activated when binding to M2-typical cytokines: heightened expression of IL-4, IL-10 and ITIMs and are thought to provide a negative signal [8]. Inhibitors TGFb [31]. of either SYK (Entospletinib) [32,33] and SHP-1 (Sodium Stiboglu- Siglec E-deficient mice furthermore have been shown to exhibit conate) [34–36] are pre-clinical compounds that have so far increased neutrophil-mediated in vivo killing of cancer cells reached phase II clinical trials for treating various cancers and, in compared to wild type mice [40]. In this model however, the case of SYK, also been tested for treating chronic graft versus Siglec-E-deficiency enhanced an M2-like macrophage profile – an host disease [32]. It should however be noted that in all cases enhancement that was reversed by transgenic expression of the effect upon Siglec signalling was not primarily considered. human Siglec-9 [40]. This is surprising, not simply because it Whilst the initiation of these signalling pathways have been iden- contradicts the findings reported above, but also since heightened tified, further downstream mediators remain to be elucidated and sialylation is linked with numerous cancers and subsequent pro- could reveal further therapeutic targets [8,27]. tumour responses, including macrophage polarisation to an M2 The potential of exploiting inhibitory signalling by macrophage phenotype [24,26,29]. On a similarly contradictory note, agonistic Siglecs has been investigated as a means of dampening inflamma- cross-linking by Siglec-7-specific monoclonal antibody has been tion. A dimer of the most common endogenous sialic acid in mam- shown to stimulate pro-inflammatory IL-6, TNFa and IL-1a secre- mals [25], a-2,8-N-acetylneuraminic acid (Fig. 2), functionalised tion by cluster of differentiation (CD)14+ human monocytes [41]. onto the surface of 150 nm poly(lactic-co-glycolic acid) (PLGA) par- A key difference between these studies is that these latter ticles, attenuated the secretion of the M1-related cytokines IL-6 and results do not investigate the effects of the sialic acids themselves. TNFa by both murine peritoneal macrophages and human mono- Siglec-E deficiency may have secondary effects, such as allowing cyte derived macrophages [37]. In parallel, these sialic acid- heightened surface expression of other signalling receptors – an coated particles enhanced IL-10 secretion, which further indicates effect then lost when Siglec-9 is introduced. Moreover, the mono- enhancement of an M2 profile. These effects were not induced by clonal antibody-mediated activation of Siglec-7 may differ from soluble form of the dimer nor the uncoated particles, implying that caused by the engagement of sialic acid epitopes. When de either a synergistic property of the two factors combined, or the novo a-2,3- and a-2,6-linked sialic acid synthesis was blocked in effect of multimeric expression of the sialic acid. Fluorescent label- human DCs, the subsequent reduction in sialic acid, Siglecs-7 and ling revealed that the sugar-coated particles interacted with Siglec- -9 expression was associated with enhanced DC maturation E on murine macrophages, and the anti-inflammatory properties of (measured by CD80 and CD86 expression), as well as increased the functionalised particles were abrogated by blocking Siglec-E IL-6 secretion [42], highlighting the importance of addressing the with an antagonist antibody [37]. That the sialic acid dimer had specific carbohydrate ligand in question to fully understand the to be presented on particles may reflect the low affinity character- carbohydrate structure-immune function relationship in each case. istic of carbohydrate-protein interactions requiring multivalent Siglecs-7 and -9 are not the only Siglecs capable of enhancing binding to enhance overall binding avidity [38,39]. Targeting mur- immunosuppressive responses. In murine models of endotoxin tol- ine Siglec-E or its human orthologues, Siglecs-7 and -9, displayed erance there was an increase in sialyltransferase activity, resulting promising efficacy in a murine model of acute respiratory distress in heightened surface expression of a-2,3- and a-2,6-linked sialic M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 27 acids [43]. This was accompanied by upregulation of Siglec-1 and pathogens, including HIV [45], Candida albicans [46], Fasciola hepat- enhanced secretion of the anti-inflammatory cytokine TGFb [43]. ica [47] and M. tuberculosis [48] respectively. Mannose epitopes can Furthermore, knocking down sialyltransferase activity not only be recognised by macrophages via a number of pattern recognition abrogated TGFb secretion, but also heightened TNFa secretion receptors (PRRs), including C-type lectin receptors such as the MR [43]. Similarly, an a-2,6-sialylated tumour antigen has been (CD206) [49], DC-associated C-type lectin (Dectin)-2 [50] and demonstrated to stimulate TGFb production from human THP1 macrophage-inducible C-type lectin (Mincle) [8,51], as well as macrophages expressing Siglec-15 – a lectin detected on the sur- Toll-like receptors (TLRs) -2 and -4 [51]. As many of these PRRs face of macrophages from human cancers in the liver, rectum are linked with internalization, mannosylation is used as a strategy and lung [44]. In both cases (Siglec-1 and Siglec-15), TGFb secre- to induce specific uptake by immune cells which can enhance sub- tion was dependent on DAP12 recruitment and Syk signalling sequent immune responses [52–54]. [43,44]. To summarise, certain sialic acid moieties are clearly implicated in polarising macrophages towards an M2 phenotype, 3.1. The dichotomy of a-linked mannosyl motifs a property which could potentially be harnessed for therapeutic applications. Mannose motifs that can be recognised by the MR and Dectin-2 are a-linked mannose residues [55,56]. As previously mentioned, 3. Mannose engagement with cells can either promote or the MR is a commonly used marker for M2 polarisation, and its inhibit immune responses depending on the specific context expression is upregulated by IL-4 and IL-13, although its role in influencing differentiation is poorly understood [8]. It is a type-I Mannose is an endogenously expressed hexose that is addition- transmembrane receptor that constantly recycles between the cell ally found on the surface of viral, fungal, parasitic and bacterial membrane and the early endosomal compartment [57]. Whilst the

Fig. 3. The structure of Mycobacterium tuberculosis mannose-capped lipoarabinomannan (ManLAM, mannose residues in black, arabinose residues in grey) and a schematic representation of the mycobacterial cell wall: emphasising the presence of ManLAM on the surface. 28 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

MR lacks any identified signalling motifs [8,57], it has been associ- evasion by hindering macrophage polarisation towards an M1 pro- ated with immunomodulatory effects [8,57,58]. Nevertheless, file [49,67]. The outermost (a-1,6) mannose residues of ManLAM defining its specific role and associated signalling pathways is chal- [50,58] induce peroxisome proliferator-activated receptor c lenging, especially as some studies suggest that signalling is medi- (PPARc) up-regulation in both murine [49] and human [58] macro- ated via collaboration between it and various other PRRs, such as phages via an MR-mediated mechanism. PPARc is a nuclear recep- TLR-2 [59], TLR-4 and Dectin-1 [60]. tor, the activation of which is known to promote M2 differentiation Dectin-2 is a type II transmembrane receptor of the C-type lec- [68–70]. This in turn has been demonstrated to stimulate IL-8 pro- tin family [61,62], that is highly conserved between mice and duction and M2-associated cyclooxygenase-2 (COX-2) expression humans, with 75% amino acid identity between the orthologues and prostaglandin E2 (PGE2) secretion in human monocyte- [61]. Although it lacks any known signalling domains, upon recog- derived macrophages. In addition, the use of antagonistic ManLAM nition of high mannose-containing epitopes, Dectin-2 associates ssDNA aptamers to block ManLAM interaction with the MR with the Fcc receptor (FccR) to initiate signalling responses via enhanced secretion of the M1 effectors IL-1b, IL-12, NO and iNOS the ITAM motif [46,61,62]. Dectin-2 is prominently present on expression and decreased secretion of IL-10 by murine peritoneal mature DCs [62,63] and its surface expression has been shown to macrophages [49]. be upregulated on macrophages in several inflammatory models, Injection of another ManLAM ssDNA aptamer with the TB vac- including Candida albicans challenge [64,65]. Furthermore, the cine BCG significantly enhanced iNOS expression in macrophages, receptor has been associated with promoting an M1-associated from both the peritoneal cavity and the broncho-alveolar lavage pro-inflammatory profile upon recognition of high mannose- fluid (BALF), enhanced Th1 responses and also enhanced protection containing epitopes [65,66]. of mice from subsequent virulent Mtb H37Rv aerosol challenge, when compared with BCG alone – as shown by fewer colony form- 3.1.1. ManLAM attenuates M1 polarisation ing units in the spleen and lungs [67]. Further investigations in a With regard to M. tuberculosis (Mtb), the outer cell wall is dom- rhesus monkey model of Mtb challenge, demonstrated that the inated by mannosylated lipoglycoconjugates, comprising addition of the aptamer similarly heightened BCG vaccine efficacy, phosphatidyl-myo-inositol mannosides (PIMs), lipomannan, and as measured by increased IFNc production by peripheral blood mannose-capped lipoarabinomannan (ManLAM) [48,50,58] the mononuclear cells (PBMCs) and reduced lung bacterial burden 5- latter of which has been shown to have immunomodulatory prop- and 17-weeks post Mtb H37Rv infection [67]. In vitro activation erties [49,58,67]. ManLAM is composed of a D-arabinan and D- of naïve murine CD4+ T cells, by CD3-, CD28- and ManLAM stimu- mannan core, and at the non-reducing termini it can be capped lation, moreover demonstrated that blocking the ManLAM-MR by up to three mannose residues [48,50]. This moiety is present interaction boosted intracellular levels of IFNc and IL-17. Further- on the surface (Fig. 3) of the virulent Mtb H37Rv strain, as well more, it has been reported that surface expression of the MR on as attenuated Mycobacterium bovis (Bacillus Calmette-Guerin, DCs, co-cultured with Désiré T cell receptor (DesTCR) T cells – T BCG) and it has been demonstrated to contribute to immune cells specific for an endogenous major histocompatibility complex

Fig. 4. The structure of Candida albicans Serotype A mannan, with a-mannosyl residues in black and b-mannosyl residues in grey, and a schematic representation of the C. albicans yeast cell wall: emphasising the presence of mannan on the surface [13,181]. M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 29

(MHC) class I ligand – attenuates IFNc secretion and cytotoxic and TNFa [65]. These results indicate that a-mannan via either activity when compared with MR deficient DCs [71]. The attenua- MR and/or Dectin-2 signalling can promote macrophage inflamma- tion was attributed to MR-mediated up-regulation of cytotoxic tory responses. T-lymphocyte-associated Protein 4 (CTLA-4) [71]. Overall, it appears that the MR has immunomodulatory properties and that 4. Glucose based polymers ManLAM promotes M2 macrophage polarisation, the blocking of which has been demonstrated to enhance BCG vaccine Glucose based polymers are highly abundant in nature, with efficacy [67]. cellulose (b-1,4-glucose) being the most abundant polysaccharide, followed by chitin (b-1,4-N-acetyl glucosamine). Chitin, and its 3.1.2. a-Mannan enhances pro-inflammatory responses de-acetylated derivative chitosan, are immunomodulatory poly- Fungal cell walls are largely composed of carbohydrates and mers [77] that have been extensively investigated as biomaterials typically have an inner layer comprising b-glucans and chitin, with [77–79], including scaffold design for bone tissue engineering outer layers composed of glycan and glycoprotein structures [13]. [79,80] and scaffolds for reparation of nerve degeneration In the case of the most of common opportunistic fungal pathogen, [81,82]. Importantly, due to their immunomodulatory, bactericidal C. albicans, 80–90% of its outer cell wall is carbohydrate [46,72] and and fungicidal properties, chitin, chitosan and other derivatives its outer glycoprotein layer is largely composed of polymers of have been used to develop wound dressings [83]. The distinct mannose residues forming O- and N-linked mannans [13,56] physicochemical properties of chitin and chitosan however allow (Fig. 4). N-linked mannans are highly branched structures, com- differing interactions and subsequent immunomodulatory effects. prising of up to 150–200 a-1,6 mannose monomers, with a-1,2 Another immunomodulatory glucose-based polymer is b-glucan and a-1,3 mannose as well as phosphomannan side chains, (prominently b-1,3-glucose), a highly conserved component of attached to a mannose-N-acetyl glucosamine core [72,73]. O- the fungal cell wall [13,84,85], currently considered as a possible linked mannans are shorter, linear oligosaccharides principally antigen for anti-fungal vaccines [86,87]. comprised of a-1,2 linked mannose residues [72,74]. Intraperi- toneal injection of mannan extracted from Saccharomyces cerevisiae 4.1. The M1 re-polarisation and training opportunities with b-glucan has been shown to result in enhanced murine macrophage ROS production and TNFa secretion, as well as IL-17A secretion by cd b-Glucan (Fig. 5A) is the name given to a diverse family of T cells [75]. b-glucose-based polymers, which are not selectively found in fungi Although a direct interaction has not yet been identified, the MR but also found amongst algae, plants and bacteria [88]. Varying has been implicated in enhancing T helper (Th)17 responses upon physicochemical properties, such as solubility [89], molecular murine Paracoccidioides brasiliensis challenge [60] and employing weight, and helix conformation [90] found amongst differing iso- antagonistic anti-MR antibodies inhibited mannan enhancement lated extracts (e.g. curdlan, lentinan and zymosan) and synthesised of IL-17 secretion from human PBMCs [59]. Moreover a-mannans forms, affect the type and extent of bioactivity that is elicited [88]. are specifically recognised by Dectin-2 [56]. Macrophage secretion As an example, regarding isolated fungal derivatives of b-glucan, it of the pro-inflammatory cytokines TNFa, IL-6, IL-1a, and IL-1b, fol- has been argued that only insoluble particulate forms with high lowing murine in vivo challenge with C. albicans, was attenuated in molecular weight can efficiently induce Dectin-1 responses Dectin-2 deficient mice [76]. Furthermore, knock-down of Dectin-2 [13,89,91]. by siRNA in the THP1 human macrophage cell line, significantly Fungal glucan is commonly composed of short b-1,6- or b-1,3- reduced Aspergillus fumigatus-induced secretion of IL-10, IL-1b glucosyl side chains, b-1,6-linked to the majority b-1,3-glucose

Fig. 5. Candida albicans cell wall and the structures of its b-glucose-based carbohydrate polymers. A) Structure of fungal b-Glucan with b-1,3-glycosyl residues in black, and b- 1,6-glucosyl residues in grey. B) Schematic representation of the C. albicans yeast cell wall. C) Structure of chitin (b-1,4-N-acetyl glucosamine) and its de-acetylated derivative chitosan (b-1,4-glucosamine) [13,77,88,92,102]. 30 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 core [92–94], through which the polymer is also attached to other able to stimulate non-specific protective immune responses carbohydrate structures, such as the inner chitin layer [13,61], towards secondary stimuli. (Fig. 5B). This polymer is prominently recognised by DCs and macrophages via Dectin-1, which is structurally very similar to 4.2. The physicochemical properties of chitin and its derivatives Dectin-2, in that it also is a type II transmembrane receptor of determine their immunomodulatory properties the C-type lectin family [61]. Dectin-1 specifically recognises b- 1,3-glucan epitopes and mediates signalling and subsequent Chitin is a polymer of b-1,4-linked N-acetyl glucosamine responses via an ITAM present on its cytoplasmic tail [13,72]. (GlcNAc) and it is the second most abundant biopolymer in nature Macrophage recognition of b-glucan through Dectin-1 is a pro- after cellulose [77,102,103] (Fig. 5C). As well as being a conserved inflammatory event, that promotes M1-associated effectors, component of the fungal cell wall [13], chitin is also a structural including iNOS, IL-6, IL-12, TNFa, and IL-1b, whilst down regulat- entity found in crustaceans, insects and as part of the microfilarial ing M2-linked IL-10 and arginase I [85]. A similar cytokine profile sheath of nematodes [102–104]. Chemical de-acetylation of b-1,4- was identified in the serum of mice vaccinated with Saccharomyces N-acetyl glucosamine yields chitosan, which has distinct physico- cerevisiae-derived whole b-glucan particles (WGPs), compared chemical properties from chitin. For this reason the degree of with vehicle controls [89]. Furthermore, oral WGP administration acetylation is important for distinguishing between the two has been shown to polarise M2-like tumour associated macro- polysaccharides, which can vary significantly between prepara- phages (TAMs) towards an M1 profile in a murine model of Lewis tions of both polymers. For the purpose of this review, chitin is lung carcinoma (LLC) [85]. Downstream of these initial effects, b- defined as 87–100% acetylated. glucan-Dectin-1 interaction has moreover been demonstrated to There is a current lack of consensus regarding chitin induced stimulate CD4+ and CD8+ T cell proliferation in a mouse tumour immune responses. For instance, chitin is associated with mediat- model [85], and work synergistically with TLR-4, TLR-2 and the ing downstream Th2 differentiation during fungal Aspergillus fumi- MR to promote anti-fungal Th17 responses during Paracoccidioides gatus infection [104] and M2 macrophage polarisation during brasiliensis [60,95] and C. albicans infection [59]. Furthermore, helminth Nippostrongylus brasiliensis infection [105]. Intraperi- intra-muscular vaccination of mice with synthesised b-glucan toneal administration of 100 lm chitin particles triggered polymers protected mice from subsequent lethal C. albicans infec- recruitment of macrophages with upregulated M2 related genes, tion: inducing b-glucan oligosaccharide-specific protective IgG including Arg1, Ym1 and Fizz1 [105]. Induction of M2 macrophage antibodies [96]. b-Glucan therefore shows promise as an antifungal effectors correlates with Th2-biased responses elicited by helminth vaccine adjuvant, and as a therapeutic to polarise macrophages infections [106]. However chitin particles of <70 lm have been towards an M1 profile and potentially re-polarise M2-like TAMs shown to induce macrophage pro-inflammatory TNFa secretion towards an M1 profile in a cancer setting [85]. [107] and prevent Th2 differentiation [108]. There are numerous Fungal b-1,3-glucan has additionally gained a lot of attention in factors that can contribute to these discrepancies, such as degree recent years for its ability to protectively prime innate immune of acetylation, average molecular weight of the polymer, variation cells, to respond more strongly to subsequent stimuli in a non- in sources and extraction procedures as well as the purity versus specific manner, a phenomenon referred to as innate training contamination of preparations; demonstrating a need for greater [97,98]. Pre-incubation of isolated human monocytes with b- standardisation. glucan 24 h before heat-killed C. albicans stimulation was shown Size is a particle feature that affects the type and intensity of the to enhance TNFa and IL-6 secretion in a dose-dependent manner immune response elicited by numerous materials [109–111] for [99]. In contrast other stimuli, such as mannan, did not exert this reasons including differences in uptake [111,112] and changes to effect [99]. Similar experiments with secondary stimulation one surface area-to-volume ratio affecting bioactivity [113]. Verte- week after priming demonstrated that this enhancement of TNFa brates express chitinases that can digest chitin into smaller parti- was also seen with a range of other stimuli: LPS, TLR2 ligand cles [114], and the size of these particles seems to affect which tripalmitoyl-S-glyceryl-cysteine (Pam3Cys), as well as with heat- PRRs are triggered and thereby has a major impact on subsequent killed gram positive and gram negative bacteria [100]. Furthermore, immune responses [103,108], as summarised in Table 1. <70 lm this protective effect has also been observed in vivo, where priming chitin particle induction of TNFa secretion was significantly with C. albicans protected mice from lethal secondary challenge reduced in macrophages from TLR-2-deficient mice; which was [99], and pre-stimulation with b-glucan increased survival in mice further completely abolished when a Dectin-1 inhibitor was added with Staphylococcus aureus induced sepsis from 40 to 90% [100]. [107]. However, MR-enhanced phagocytosis of smaller (<40 lm) How this training effect is realised has recently been hypothe- particles stimulated IL-10 secretion from murine macrophages sised to be related to cell viability [98]. When TNFa and IL-6 con- [107] and human PBMCs [115]. This IL-10 secretion was lost when centrations were normalised to the number of viable murine macrophages were either TLR-9 or Nucleotide-binding macrophages or human monocytes, it was shown that the cytokine oligomerization domain-containing protein 2 (NOD2) deficient production per cell was not altered, but the number of cells was, [115]. TLR-9 is known to detect double stranded DNA [116], yet suggesting that b-glucan primarily augments cell viability [98]. pre-treatment with DNase I did not alter the secretion of IL-10 However, there is accumulating evidence showing that Dectin-1 [115]. That NOD-2 is implicated is also unexpected, given that recognition of b-glucan does induce epigenetic changes to cyto- NOD-2-induced responses to structurally similar peptidoglycans kine, PRR [99] as well as metabolic [100] gene promoters, thus cor- augment pro-inflammatory responses [117,118]. In summary, roborating the training hypothesis [97,99–101]. b-Glucan induced further investigation into the physicochemical characteristics of epigenetic changes in peritoneal macrophages further suggest a chitin, and how these affect PRR-recognition and downstream skewing of responses towards an inflammatory profile, as histone effects, is needed in order to fully comprehend the immunomodu- H3 lysine 4 trimethylation (H3K4me3) was elevated at the promot- latory properties of this naturally abundant polysaccharide. ers of the cytokines TNFa, IL-6 and IL-18 [99]. This was further cor- roborated by transcriptome analysis of human monocytes, where 4.3. De-acetylation causes chitosan to activate the NLRP3 mRNA levels of TNFa and IL-6 were again elevated, although the inflammasome M2 markers CD163 and monocyte chemotactic protein 1 (MCP1) were also heightened [99]. Therefore, the enhanced inflammatory Whereas there is lack of consensus on the nature of macrophage phenotype observed in vivo [98–100] indicates that b-glucan is responses to chitin, there are striking differences between the M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 31 responses triggered by chitin and chitosan. The process to convert fact electrons to be utilized to make ROS, which in turn aids HIF- chitin into chitosan involves de-acetylation of the NHAc residues 1a stabilization [133,135]. (Fig. 5C) furnishing an overall cationic polymer. Importantly, the M2 polarisation contrastingly induces oxphos activity [134,136] change in physicochemical properties imparts chitosan with an and involves signalling via several pathways, including the Janus ability to bind to DNA [119] and induce NLRP3 inflammasome acti- kinase (JAK)1/JAK3-signal transducer and activator of transcription vation and Th1 responses [120–123]. The NLRP3 inflammasome is (STAT) 6 pathway [69,136,137]. Expression of certain M2 markers, a cytosolic complex composed of NLR Family Pyrin Domain Con- such as Arg1, Fizz1 and macrophage galactose-type C-type lectin taining 3 (NLRP3) protein and caspase-1, and its two-step activa- (Mgl) 2, have also been reduced by blocking mTORC2 and protein tion ultimately results in the secretion of pro-inflammatory IL-1b kinase B (also known as AKT) activation [69,137]. STAT-6 and and IL-18 [121]. Chitosan has been implicated in triggering several mTORC2 activation have both been proposed to lead to activation mechanisms of NLRP3 activation, including lysosomal destabiliza- of IRF4, which is associated with increased levels of glycolysis, tion [121,124], potassium ion (K+) efflux, and mitochondrial stress which is central to M2 macrophage differentiation [136]. [121]. Much of the detail of these metabolic and signalling pathways, Chitosan-triggered mitochondrial stress additionally causes the and how they are intertwined, remains to be determined, and for a release of mitochondrial DNA, which in turn activates the cyclic more in-depth assessment there are numerous recent reviews cov- GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) ering this subject matter [70,130,133]. However, it remains clear pathway, finally leading to the secretion of type I interferons [122]. that these metabolic changes are dependent upon and influenced Murine studies additionally reveal that chitosan shows great pro- by various carbohydrate metabolites, such as succinate [138], lac- mise as a vaccine adjuvant, capable of stimulating Th1 responses tate [12] and the availability of glucose [139,140]. As such, these [122,123] which are a key target for vaccines against pathogens carbohydrates have potent immunomodulatory properties, which such as HIV and M. tuberculosis [125]. Furthermore, oligochitosan include influencing macrophage polarisation. has been demonstrated to induce modest nitric oxide (NO) produc- tion and TNFa secretion by RAW264.7 macrophages [126–128]. All 5.1. 2-Deoxy glucose of these results are indicative of chitosan driving M1 responses. By contrast, primary human monocyte-derived macrophages grown Initial oxidation of glucose through glycolysis is hindered by the on a film of chitosan have been shown to gain an M2-like profile use of the pharmacological inhibitor 2-deoxy glucose (2-DG) – displaying a moderate increase in IL-10 and TGFb, compared with [139,141]. Administration of 2-DG six hours prior to LPS has been same cells grown on tissue culture polystyrene [129]. This would shown to significantly protect mice against lethal septic shock, and indicate that the form or size of chitosan significantly impacts on simultaneously reduced serum levels of M1-associated TNFa and the immune responses elicited. The extent of macrophage IL-1b NO [139]. This inhibition of M1 effectors seems to be specifically secretion in response to chitosan was shown to be correlated to caused by 2-DG prevention of IL-1b expression, as it has been size; when comparing chitosan particles from <20 lmto shown in vitro that this sugar blocks IL-1b mRNA production >100 lm, as determined by filtration, the smallest particles were [138], but not IL-6 [138] or TNFa expression [135,138]. the most potent stimulators [120]. Similarly, regarding the other differentiation extreme: 2D-G Whilst a plethora of receptors have been associated with chitin macrophage stimulation in vitro reduced IL-4 induced expression induced responses, few have been associated with chitosan of M2 activation markers Fizz1 and programmed cell death 1 mediated effects. The most recurring lectin mentioned in the liter- ligand 2 (PD-L2), and intraperitoneal injection with the helminth ature is the mannose receptor, the blocking of which by competi- Heligmosomoides polygyrus reduced macrophage recruitment and tive antagonist or anti-mannose antibody has been observed to proliferation [136]. These studies corroborate that both M1 and decrease chitosan-induced TNFa [126,127]. However, further stud- M2 differentiation require a heightened level of glucose uptake ies investigating chitosan induced inflammatory responses and and subsequent glycolysis [70,133] – the blocking of which atten- adjuvanticity in vivo are required to substantiate the roles of speci- uates macrophage activation and function. fic receptors. Overall, results thus far indicate that chitosan induces a M1 phenotype in macrophages and pro-inflammatory effectors 5.2. Succinate that lead to Th1 responses. The metabolism of succinate is a link between the Krebs cycle (also known as citric acid cycle) and the electron transport chain 5. Immunometabolism: a requisite for polarisation (ETC) via succinate dehydrogenase (also known as Complex II) [142]. It has recently become apparent that heightened levels of Immunometabolism is a rapidly growing field, the importance succinate in macrophages causes ROS production through reverse of which is becoming more and more apparent for numerous electron transport [134,135], which increases expression of the immune cells, including macrophages. Macrophage activation transcription factor hypoxia-inducible factor-1a (HIF-1a) and sub- induces proliferation and cytokine production, amongst other sequent IL-1b mRNA, as well as pro-IL-1b protein levels [135,138]. effects, that requires an increase in ATP and anabolic precursor This was for instance shown by the use of anti-oxidants, and by production [130]. However, as summarised in Fig. 6 macrophage blocking succinate oxidation in the mitochondria with the compet- differentiation routes are tightly linked with distinct signalling itive antagonist malonate; both abrogated the succinate induced and metabolic programmes [69,70]. M1 macrophage differentia- expression of IL-1b and HIF-1a [135]. Additionally, succinate stim- tion involves activation of transcription factors, such as nuclear ulation inhibited LPS-induced anti-inflammatory IL-10 and IL-1Ra factor kappa-light-chain-enhancer of activated B cells (NF-jB) secretion [135], evidently skewing macrophage responses towards and interferon regulatory factors (IRFs) [70], as well as the activa- an M1 profile. tion of mechanistic target of rapamycin (mTOR) complex 1 (mTORC1) [69,131] and subsequent augmentation of glycolytic 5.3. Lactate metabolism, including hypoxia-inducible factor 1a (HIF-1a) medi- ated up-regulation of glucose transporter 1 (GLUT1) and enzyme At the end of glycolysis, the final product, pyruvate, can either lactate dehydrogenase (LDH) [132]. This polarization route also be used as fuel for oxidative phosphorylation or for fermentation hinders oxidative phosphorylation (oxphos) [133,134], causing in to lactate. Whilst oxphos is the most efficient means to get the 32 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

Fig. 6. M1 and M2 macrophages have distinct metabolic programmes. Glucose is taken up by glucose transporter 1 (GLUT1), and its oxidation is initiated by glycolysis. Oxidation can then either be finalised by lactate dehydrogenase (LDH), to form lactate as the final product, or can continue via oxidative phosphorylation (oxphos) in the mitochondria, to yield carbon dioxide. Left side of the panel shows M1 polarisation: lipopolysaccharide (LPS) stimulation via toll-like receptor 4 (TLR4) initiates signalling that activates mechanistic target of rapamycin complex 1 (mTORC1) and subsequent hypoxia-inducible factor 1a (HIF-1a) activation. M1 polarisation also hinders oxphos, which leads to heightened reactive oxygen species (ROS) production, which stabilises HIF-1a. HIF-1a induces higher level of glycolytic metabolism. Right side of the panel shows M2 polarisation: interleukin 4 (IL-4) stimulation via the IL-4 receptor (IL-4R) causes signalling via the Janus kinase (JAK)1/JAK3 signal transducer and activator of transcripton 6 (STAT6) pathway and also seemingly activates mTORC2 and subsequent protein kinase B (AKT). STAT6 and mTORC2 have been proposed to cause the upregulation of interferon regulatory factor 4 (IRF4), which in turn causes increased glycolytic and oxphos activity. [69,70,133–137]. Grey arrows and blue shaded arrow, sequence of events; dashed arrows, possible sequence of events; red or green shaded upward arrows, increase in activity; red shaded downward arrow, decrease in activity. Servier Medical art Powerpoint image bank was used to construct this diagram. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) greatest yield of ATP per glucose molecule, fermentation to lactate 6. Discussion: the therapeutic potential of macrophage- is faster. In cancer cells a characteristic metabolic shift, known as activating carbohydrates the Warburg effect and aerobic glycolysis, is the switch to near exclusive use of glycolysis and subsequent fermentation as the A striking observation when reviewing the glycoscience litera- source of ATP, despite the presence of oxygen [143]. This leads to ture, such as the carbohydrates and polysaccharides discussed high levels of lactate secreted in the tumour microenvironment herein, is that there is often contradictory evidence regarding the [144], which in turn has immunomodulatory effects on macro- responses elicited; and seemingly a degree of cooperation or even phages [12,132]. redundancy amongst receptors that recognise them, as sum- Stimulating murine macrophages with exogenous lactate marised in Table 1. Chitin, as an example is shown to promote both in vitro enhanced expression of several M2 markers including M1 and M2 characteristic effectors, via a spectrum of PRRs, includ- Arg1, Fizz1 and prominently vascular endothelial growth factor a ing Dectin-1, MR, NOD-2, TLR-2 and TLR-9 [103,105,107,108]. (VEGFa) [12], which is important for promoting angiogenesis. Moreover, the specificity of epitope-receptor interaction is some- Diminished production of VEGFa, caused by deletion of lactate times called into question, given that structurally distinct carbohy- dehydrogenase, was additionally shown to be rescued by lactate drates can trigger the same PRR. For instance a-mannans and b- supplementation [132]. High lactate concentration was also shown glucans can both be associated with responses at least partly to correlate with an enhanced presence of M2 macrophages, as dependent on Dectin-1 and the MR [59,60]. These discrepancies identified by the markers CD163 (a scavenger receptor) and macro- in the literature can be caused by numerous factors; contamination phage colony-stimulating factor receptor (M-CSFR), in human head being a constant concern – certain b-glucan preparations such as and neck squamous cell carcinoma patients [144]. zymosan can be contaminated with numerous TLR agonists [95]. Furthermore, blocking lactate production by global condi- Furthermore, differences between preparations also occur because tional deletion of LDH – the enzyme that catalyses pyruvate’s these carbohydrates are isolated from various sources using varied conversion to lactate – has been shown to reduce tumour extraction and purification protocols. However, a major contribut- growth in a murine lung cancer model, whilst also leading to ing factor is a lack of full characterisation of the carbohydrates enhanced levels of iNOS in lung tissue [132]. Lactate’s used. immunoregulatory effects were very similar to those induced Differences in physical properties such as solubility [89] and by hypoxic conditions (0.1% oxygen), and lactate was unable to molecular weight [120,145] have been shown to impact both the induce the M2 markers in HIF-1a deficient cells [12], which type and intensity of immune responses elicited. Furthermore, dif- would imply that macrophage responses to lactate are connected ferences in composition resulting from the use of varying sources to hypoxia-related responses. This route of lactate and hypoxia for carbohydrate extraction can affect the results. When comparing re-programming macrophages to an M2 phenotype via HIF-1a responses to mannans from the yeast, S. cerevisiae and C. albicans, contradicts its proposed role in promoting M1 metabolism only the Candida isolated mannan enhanced IL-17 secretion from [69,133], demonstrating the need for further investigation in this human PBMCs [59]. As fungal mannan is composed of both a- field. Nevertheless, lactate is clearly a potent inducer of M2 and b-mannans, and a-mannan is more pro-inflammatory [76], macrophage differentiation, which in this case seems to be the proportion of the two in a preparation of ‘‘mannan” can affect tightly linked with hypoxia. immune responses elicited. Table 1 Examples of immunomodulatory carbohydrates and a summary of their immune-mediated effects.

Carbohydrate Specific Structure Sensor(s) implicated Response Cells investigated Refs Chitin 100% Ac – TR: "IL-6, TNF a Monh [152] TR: "IL-1b, IL-6, IL-10, TNF a In vivo 97% Ac Dectin-1, TLR-2 "IL-6, TNF a M U m, PBMCs h, in vivo [115] Majority 1–10lmP NOD2, TLR-9 "IL-10 MR Phagocytosis 87% Ac – "Arg1, Ym1, Fizz1, MR M U m, in vivo [105] 100lmP 70–100lmP – – M U m in vivo [107] l70m P Dectin-1, TLR-2 "TNF a, IL-17A M U m in vivo [103,107] l40m P Dectin-1, TLR-2, MR "IL-10 M U m, in vivo [107] 20–50lmP – "Arg1, iNOS, PD-1 M U m [108] Chitosan 20% or 2% Ac – Inflammasome activation M U h [124] "IL-1 a, IL-1b 24% Ac – Inflammasome activation, "IL-1b MU & DCsm [121] m >100lmP – "IL-1b (minimal) M U [120] 23–41 (2017) 146 Pharmacology Biochemical / al. et Lundahl M.L.E. 20–100lmP – "IL-1b (more) M U m [120] lmP20 – "IL-1b (most) M U m [120] 25% Ac Oligochitosan – "TNFa, iNOS (modest) M U m [128] (3–9 residues) – "IL-1b, TNFa MU m [127] MR "TNFa MU m [126] 10–25% Ac cGAS/STING "IFNb, CXCL10 DCsm [122] cGAS/STING, NLRP3 "IFNc In vivo [122] 2-Dexoy Glucose – ;IL-1b MU m [138] – ;IL-1b, TNF a In vivo [138] – ;TNFa,NO In vivo [182]

b-Glucan Whole glucan particles Dectin-1 " IL-1b, IL-6, TNF a, iNOS, CD40, CD86 M U, OT T cells m [85] ;IL-10, Arg ;CD4+ & CD8+ T cell proliferation – "IL-1b, IL-6, IL-17, MIP1a, TNFa Serumm [89] b-1,3-Glucan Dectin-1, MR, TLR-4 "IL-6, IL-17, TNFa DCs & splenocytes m [60] "Th17 Dectin-1 TR: "IL-6, TNFa, epigenetic changes Mon h [99] – TR: "IL-6, IL-10, TNFa Serumm [98] – TR: "TNFa, epigenetic changes Mon h [100] – TR: Epigenetic changes MU h [101] Glycosamino-glycans (GAGs) Heparin – "Angiogenesis, neo-vascularisation In vivo [175] – ;IL-1b, FBGC formation MU h [177] Hyaluronan – ;IL-1b, FBGC formation MU h [177] High mw – ;IL-6, MCP-1 MU h [145] Low mw – – MU h [145] Sulfated, low mw – ;IL-6, IL-12p40, MCP-1 MU h [145] Chondroitin sulfate – ;IL-1b, FBGC formation MU h [177] Lactate – "VEGFa, Arg1, Fizz1, Mgl1, Mgl2 M U m [12] – BLOCK: "iNOS, CD86, CD197 MU m [132] – ;VEGFa, CXCL10 In vivo [132] "iNOS, IFNc

(continued on next page) 33 34 Table 1 (continued)

Carbohydrate Specific Structure Sensor(s) implicated Response Cells investigated Refs ;VEGFa, TGFb a-Mannoses Mannose cap of ManLAM, a-1,6-mannoses MR BLOCK: "IL-1b, IL-12, iNOS, NO M U m in vivo [49] ;IL-10 MR BLOCK: "iNOS, NO MU m [67] h MR "IL-8, COX2, PGE2 MU [58] Dectin-2 "IL-2, IL-6, IL-10, TNFa DCsm [66] a-Mannans, a-1,6- (majority), a-1,2- & a-1,3-mannoses Dectin-2 "IL-1 a, IL-1b, IL-6, TNF a MU m in vivo [76] – "TNFa M U m [75] – "IL-17 In vivo Dectin-1, TLR-2 "IL-17 PBMCs h [59] MR "IL-17, IL-10 Dectin-2 "IL-1b, IL-6, IL-10, IL-12p40, TNFa DCsm [56] Sialic acid a-2,3- and a-2,6- sialic acids Siglec-1 ;TNF a M U m, DCsh&m [43] "TGFb

Siglec-Em "IL-4, IL-10, TGFb M U m [31] 23–41 (2017) 146 Pharmacology Biochemical / al. et Lundahl M.L.E. ;IL-2, iNOS, ROS, NO a-2,8-di-Neu5Ac Siglec-7 & -9 h ;IL-6, TNFa, IL-8 M U h [183] Siglec-E m "IL-10 MU m [183] Siglec-E m ;IL-6,TNFa in vivo [183] ;IL-6, TNFa "IL-10 a-2,6-Neu5Ac Siglec-15 "TGFb M U h [44] Global cell surface Siglec-7 & -9 BLOCK: "IL-6, IL-10, CD80 DCs h [42] Siglec-7 & -9 BLOCK: "IFNc, proliferation DCs & PBLs h [42] Succinate – "IL-1b, ROS M U m [135] O ;IL-10, IL1Ra HO m OH – "L-1b MU [138] O

Trehalose Trehalose-6,6-dimycolate (TDM) Mincle, Fc cR "IL-1b, IL-6 NO M U m [166] Mincle ;Phagosome maturation MU m [168] Mincle,c FcR "IL-10 M U m [169] ;IL-12p40 Trehalose-6,6-dibehenate (TDB) Mincle,c FcR "IL-1b, IL-6 NO M U m [166] Mincle, FccR "IFN c, IL-17 In vivo [166] Di-O-acylated trehalose (DAT) – ;NOS, NO MU m [184] Tri-O-acylated trehalose (TAT) – ;NOS, NO MU m [184]

Ac, acetylation; CXCL10, C-X-C motif chemokine 10; FBGC, foreign body giant cell; Mon, monocyte; mw, molecular weight;U M, macrophage; Neu, Neutrophils; Neu5Ac, N-acetyl neuraminic acid; P, particles; PD-1, programmed cell death protein-1; TR, training. h human. m murine. M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 35

Moreover, when investigating b-glucan-PRR interactions by accessing more homogenous biopolymers from natural sources. means of HEK-BlueTM-hTLR2, -hTLR4 and hDectin-1 cells: linear b- All these concerns provide a strong argument in favour of moving 1,3-glucan extracted from Alcaligenes faecalis was found to interact towards the use of synthetic forms of carbohydrates, to be able to with TLR-2 and TLR-4 but not Dectin-1, whereas a b-(1,6)(1,3)- control and fully characterise all these physicochemical attributes glucan preparation from S. cerevisiae could interact with Dectin- whilst avoiding contamination. Yet, despite all these issues, 1, but not TLRs-2 and -4 [95]. However, C. albicans b-1,3-glucan numerous immunomodulatory carbohydrates have been identified [98,99] has been proposed to boost innate training [98–100] via that can influence macrophage activation, and are thus attractive a Dectin-1 dependent mechanism, which was inhibited when b- candidates for therapeutic applications. (1,6)(1,3)-glucan (lentinan) from Laminaria digitata was added as Macrophages are ubiquitously present innate immune cells that an antagonist [99]. Differences in these results could be due to con- play an important role in numerous settings (Fig. 7). For instance tamination or the structural differences of the b-glucan polymers tumour-associated macrophages are actively polarised towards from these varying sources. As also discussed, the degree of acety- an M2-like phenotype by the cancer cell altered glycosylation pat- lation of chitin and chitosan is another example, and can vary sig- tern, such as intensified sialylation on the cell surface [26,29], and nificantly between different sources and preparations [115]. the tumour mediated microenvironment orchestrating macro- For these reasons there is a need to fully and accurately charac- phage metabolic re-programming [12,146]. These macrophages terise the carbohydrates used, with regard to physical properties, greatly aid tumorigenesis by promoting tumour growth, angiogen- such as molecular weight, but also structural and chemical differ- esis, matrix degradation and suppression of anti-cancerous ences, such as the degree of acetylation and branching, the propor- immune responses by secretion of anti-inflammatory cytokines, tional differences of various linkages found in polymers and so such as IL-10 and TGFb [1]. For this reason, sialic acid epitopes, forth. This type of characterisation remains a very significant chal- amongst other tumour-associated carbohydrate antigens, have lenge due to the natural heterogeneity of biological oligo- and been suggested as targets for the design of anti-cancer vaccines polysaccharides. However, recent advances in extraction technolo- [29]. gies, such as ultrasound, microwave and enzyme-assisted extrac- Another possible therapeutic strategy is the re-polarisation of tion, as well as new purification techniques (e.g. membrane macrophages towards an M1 profile [12,85]. b-Glucan, for instance, separation), in concert with advances in analytics may assist in has been proposed as a polysaccharide capable of inducing pre-

Fig. 7. Examples of carbohydrate-recognising receptors present on macrophages, and a summary of M1 and M2 polarising stimuli, markers and secreted products, as well as the responses elicited. Abbreviations: Arg, Arginase; DAMPs, danger-associated molecular patterns; Dectin, DC-associated C-type lectin; FccR, Fc receptor for immunoglobulin G; Fizz1, found in inflammatory zone 1; IFNc, interferon c; IL-, interleukin; MHC II, major histocompatibility complex class II; MR, mannose receptor; NO, nitrogen oxide; NOD2, Nucleotide-binding oligomerization domain-containing protein 2; PAMPs, pathogen-associated molecular patterns; ROS, reactive oxygen species; Siglec: Sialic acid-binding immunoglobulin-like lectin; TGFb, transforming growth factor b; TLR, Toll-like receptor; TNFa, tumour necrosis factor a; Ym1, chitinase-like 3. h: human, m: mouse. Servier Medical art Powerpoint image bank was used to construct this diagram. 36 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 cisely this effect in a cancer model, via Dectin-1 signalling [85] and affinity of the epitopes for the two receptors. However, the same is currently included in phase I clinical trials for metastatic neurob- outermost mannose residues of ManLAM [50,58,66] have been lastoma [147] and in phase I/II clinical trials for chronic lympho- shown to induce Dectin-2-dependent secretion of pro- cytic leukaemia [148]. The interest in utilizing b-glucan is also inflammatory IL-6, TNFa and macrophage inflammatory protein- explained by its innate training properties. Epigenetic modifica- 2 (MIP-2) from dendritic cells [66]. Moreover, inflammatory tions induced by Dectin-1 signalling enhanced macrophage inflam- responses to mannans have been linked to MR interaction [59]. matory responses to a range of secondary stimuli [99,100]. The Apart from possible batch variation, this dichotomy would firstly vaccine BCG has similar training effects, mediated via NOD-2 support the hypothesis that the MR collaborates with other recep- [149]. BCG’s non-specific boosting of innate immune responses is tors to induce signalling and can thus participate in opposing func- already successfully employed clinically to treat bladder cancer tional outcomes [8,59,60]. This would argue that the macrophage [150,151]. This is referred to as adjuvant immunotherapy [151]. immunomodulatory effects of ManLAM may be caused primarily Moreover, b-glucan is not the only sugar capable of augmenting via signalling through an unidentified receptor. innate immune responses. Priming isolated human monocytes An additional explanation for the opposing responses to these with chitin isolated from S. cerevisiae has been shown to enhance similar mannosyl epitopes is contextual receptor availability. secretion of the pro-inflammatory cytokines TNFa and IL-6 follow- Macrophage Dectin-2 expression is heightened by numerous pro- ing secondary stimulation with TLR-3, -8 and -9 ligands in vitro, inflammatory stimuli, such as C. albicans challenge [65]. and protected mice from subsequent C. albicans challenge in vivo Mannose-recognition in a pro-inflammatory environment may [152]. Clearly there is accumulating evidence to support further thereby bolster macrophage polarisation and function towards an investigation of these carbohydrates and how their training prop- M1 phenotype – at least partly mediated by the up-regulation of erties can be used therapeutically. Dectin-2. This illustrates the importance of not only studying Macrophage polarisation is tightly linked with distinct meta- immunomodulatory epitopes in isolation, but also in disease- bolic programmes that are also influenced by carbohydrate specific contexts, and vice versa, to fully appreciate their therapeu- metabolites such as succinate [138], lactate [12] and the availabil- tic potential. ity of glucose [139]. As such understanding these pathways can To date, mannosylation has been extensively explored as a reveal therapeutic targets to treat for instance inflammatory dis- means of inducing uptake by antigen presenting cells and enhanc- eases and cancer [153]. Succinate induced pro-inflammatory IL- ing subsequent immune responses [52–54,162,163]. For instance, 1b can be inhibited by its competitive antagonist malonate [135] human DC and macrophage uptake of virus-like particles was aug- and blocking glycolysis with 2-DG protected mice from lethal mented by a-di-mannosylation [53] and mannosylation of a pro- LPS-mediated septic shock [139]. On the other hand, a heightened posed chitosan-plasmid anti-cancer vaccine resulted in increased level of lactate is associated with tumour malignancy [12,144], serum IFNc and reduced tumour growth, compared with the enhancing pro-tumorigenic M2 effectors, such as angiogenesis- non-mannosylated version, in a murine subcutaneous prostate promoting VEGFa [12,132]. In a murine lung cancer model, condi- cancer model [54]. Similarly, specifically targeting the MR by mon- tional deletion of the enzyme LDH caused tumour regression on oclonal antibody has been investigated as a delivery system for an the same scale regardless of whether the deletion was global or epithelial cell cancer antigen [164]. This vaccine candidate reached limited to macrophages specifically [132]. This dually demon- phase II clinical trials for bladder cancer [164]. strated the key role of M2-like macrophages in tumour progression For the development of new vaccines against diseases such as and the potential of targeting lactate in macrophages. So far, the cancer and M. tuberculosis infection, much research continues to fact that these pathways are fundamental to all cells is a pharma- try to identify adjuvants with the ability to stimulate Th1 codynamic challenge for metabolic-targeted drugs. responses [125]. An example is mycobacterial cell wall component Another disease setting that promotes M2 polarisation is M. trehalose-6,6-dimycolate (TDM), a glycolipid capable of inducing tuberculosis infection [10,14]. This pathogen is an ancient intracel- both Th1 and Th17 responses [165,166]. For this reason TDM, its lular bacterium that is internalised by a number of immune cells, synthetic analogue trehalose-6,6-dibehenate (TDB) [166] and other but prominently macrophages [10], thus shielding itself from derivatives [167] are investigated as vaccine adjuvants. Their adju- extracellular humoral responses. A strong Th1 response can pro- vanticity seems to be in part due to combined Mincle-FccR recog- mote an M1 macrophage phenotype and subsequent bactericidal nition which induces macrophage NO production and T cell IFNc effectors [125,154]. However, M. tuberculosis also possesses a num- secretion [166]. Mincle-FccR recognition has however also been ber of means to prevent Th1 and M1 responses [155–157], includ- linked with prevention of macrophage phagosome maturation ing the immune-suppressive properties of a number of cell wall [168] and enhanced IL-10 secretion [169]. As such, these mycobac- components [58,67,158–160]. These include phenolic glycolipids terial glycolipids show promise as adjuvants, yet their effect upon and the structurally related para hydroxybenzoic acid derivatives macrophages remains to be fully elucidated. Another promising [158], that have been shown to suppress TLR-2 induced human adjuvant, capable of enhancing Th1 responses is chitosan macrophage secretion of IL-6 and TNFa [161] and anti-CD3 stimu- [122,123]. lated murine splenocyte secretion of IL-17 and IFNc [159] respec- Chitosan and also chitin have been extensively investigated as tively. An important example is the ManLAM epitope [67], which biomaterials used for tissue engineered scaffolds [77,80,170]. Such has in isolation been shown to induce the expression of M2- scaffolds are attractive for the construction of biomedical implants promoting PPARc in human monocyte derived macrophages [58]. – man-made devices introduced in a patient either to replace or Therapeutic opportunities are thus found in hindering these support a damaged biological structure. Accumulating evidence immune-suppressing effects [29,49,67], such as blocking the Man- indicates that M2 macrophages are critical for clinical success; by LAM epitope to improve BCG vaccine efficacy [67]. preventing harmful long-lasting inflammation and promoting tis- Whilst there is compelling evidence indicating that the outer sue remodelling and healing [15,16,171]. Macrophages exposed mannose epitopes on ManLAM skew macrophage responses away to polymeric chitosan have been suggested to adopt a more M2- from an inflammatory M1 profile, the opposite has been demon- like profile both in vitro [129] and after implantation in vivo strated for a-mannans. That similar mannose epitopes are associ- [172] when compared to chitin. However, whilst this is a promis- ated with opposing responses seems to be a result of the two ing response initially, chitosan’s ability to propagate inflammation structures eliciting signalling via different PRRs, notably MR and once degraded is a potential long-term concern that requires fur- Dectin-2. A possible explanation for this could be the differing ther consideration for therapeutic applications. M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 37

Furthermore, the immunoregulatory properties of certain car- is funded by Science Foundation Ireland (SFI) under Grant number bohydrate epitopes could potentially be used to bolster M2 effects 15/CDA/3310. Ed Lavelle is funded by Science Foundation Ireland beneficially. For example, using sialic acid epitopes to target (SFI) under Grant number 12/IA/1421 and the SFI Research Centre, Siglecs-1 [43], -15 [44],-7or-9[37] to induce anti-inflammatory Advanced Materials and BioEngineering Research (AMBER) under TGFb [43,44] and IL-10 [37]. For the latter, this approach showed Grant number SFI/12/RC/2278. promise in inducing protective IL-10 secretion in a murine ARDS model and human EVLP model [37]. Another example of endoge- References nous immunomodulatory carbohydrates are (GAGs): these are universally present on cell surfaces, but also [1] A. Sica, M. Erreni, P. Allavena, C. Porta, Macrophage polarization in pathology, prominently make up a major portion of the extracellular matrix Cell. Mol. Life Sci. 72 (2015) 4111–4126, http://dx.doi.org/10.1007/s00018- 015-1995-y. glycome [173–175]. GAGs are a highly diverse family of polysac- [2] C.J. Ferrante, S.J. Leibovich, Regulation of macrophage polarization and wound charides, composed of repeating disaccharide units; each disaccha- healing, in: Adv Wound Care (New Rochelle), Department of Cell Biology and ride comprising of a hexuronic acid (e.g. ) and a Molecular Medicine and The Cardiovascular Research Institute, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, Newark, hexosamine, such as glucosamine or galactosamine [173]. Due to New Jersey, 2012: pp. 10–16. doi:10.1089/wound.2011.0307. sulfation, most GAGs are highly negatively charged, all except [3] J. Bystrom, I. Evans, J. Newson, M. Stables, I. Toor, N. van Rooijen, M. Crawford, hyaluronan [173]. This family, as well as structural derivatives P. Colville-Nash, S. Farrow, D.W. Gilroy, Resolution-phase macrophages possess a unique inflammatory phenotype that is controlled by cAMP, and analogues are investigated for their anti-inflammatory proper- Blood 112 (2008) 4117–4127, http://dx.doi.org/10.1182/blood-2007-12- ties [145,174]. 129767. Much of GAG’s immune-regulatory activity comes from seques- [4] S. Gordon, F.O. Martinez, Alternative activation of macrophages: mechanism and functions, Immunity 32 (2010) 593–604, http://dx.doi.org/10.1016/j. tering messenger molecules, such as chemokines, cytokines and immuni.2010.05.007. growth factors [173,176]. Due to their recognised role in regulating [5] P.J. Murray, T.A. Wynn, Protective and pathogenic functions of macrophage inflammation, tissue repair and wound healing, they are attractive subsets, Nat. Rev. Immunol. 11 (2011) 723–737, http://dx.doi.org/10.1038/ nri3073. epitopes for biomaterial construction [173,175–177]. THP-1 [6] A. Mantovani, A. Sica, S. Sozzani, P. Allavena, A. Vecchi, M. Locati, The macrophages grown on various GAG-chitosan materials secreted chemokine system in diverse forms of macrophage activation and lower levels of pro-inflammatory IL-1b in response to LPS stimula- polarization, Trends Immunol. 25 (2004) 677–686, http://dx.doi.org/ tion than the poly (ethylene imine) control surface [177]. Of the 10.1016/j.it.2004.09.015. [7] S.L.M. van Loon, A.I.P.M. Smits, A. Driessen-Mol, F.P.T. Baaijens, C.V.C. Boute, GAGs tested, heparin was the most anti-inflammatory [177] and The Immune Response in In Situ Tissue Engineering of Aortic Heart Valves, in: has also been shown to induce macrophage secretion of and then Calcif. Aortic Valve Dis., InTech, 2013, http://dx.doi.org/10.5772/54354. sequester VEGFa along with other growth factors, which resulted [8] G.D. Brown, P.R. Crocker, Lectin receptors expressed on myeloid cells, Microbiol. Spectr. 4 (2016), http://dx.doi.org/10.1128/microbiolspec.MCHD- in enhanced angiogenesis in vivo in a proposed hydrogel implant 0036-2016. [175]. The clinical relevance of employing endogenous ligands [9] D.I. Gabrilovich, S. Ostrand-Rosenberg, V. Bronte, Coordinated regulation of such as these sugars is attractive also for their excellent biocom- myeloid cells by tumours, Nat. Rev. Immunol. 12 (2012) 253–268, http://dx. doi.org/10.1038/nri3175. patibility, hydrophilicity and safety profile [163,178]. [10] M.V.S. Rajaram, B. Ni, C.E. Dodd, L.S. Schlesinger, Macrophage immunoregulatory pathways in tuberculosis, Semin. Immunol. 26 (2014) 471–485, http://dx.doi.org/10.1016/j.smim.2014.09.010. 7. Conclusion [11] K.A. Jablonski, S.A. Amici, L.M. Webb, J. de, D. Ruiz-Rosado, P.G. Popovich, S. Partida-Sanchez, M. Guerau-de-Arellano, Novel markers to delineate murine To summarise, this review highlights a number of recurring car- M1 and M2 macrophages, PLoS One 10 (2015) e0145342, http://dx.doi.org/ 10.1371/journal.pone.0145342. bohydrate epitopes in nature, both endogenous and exogenous to [12] O.R. Colegio, N.-Q. Chu, A.L. Szabo, T. Chu, A.M. Rhebergen, V. Jairam, N. Cyrus, humans, which have immunomodulatory properties that could C.E. Brokowski, S.C. Eisenbarth, G.M. Phillips, G.W. Cline, A.J. Phillips, R. be harnessed therapeutically. Regarding these immune-altering Medzhitov, Functional polarization of tumour-associated macrophages by tumour-derived lactic acid, Nature 513 (2014) 559–563, http://dx.doi.org/ effects, the primary focus has been on macrophages – a key innate 10.1038/nature13490. immune cell in numerous disease states, such as M. tuberculosis [13] L.P. Erwig, N.A.R. Gow, Interactions of fungal pathogens with phagocytes, Nat. infection and cancer. Furthermore, macrophages appear to be an Rev. Microbiol. 14 (2016) 163–176, http://dx.doi.org/10.1038/ nrmicro.2015.21. essential determinant for successful introduction of clinically rele- [14] B. Dey, W.R. Bishai, Crosstalk between Mycobacterium tuberculosis and the vant biomaterials. A great deal remains to be discovered and eluci- host cell, Semin. Immunol. 26 (2014) 486–496, http://dx.doi.org/10.1016/j. dated regarding carbohydrate-recognising receptors, their smim.2014.09.002. [15] B.N. Brown, S.F. Badylak, Expanded applications, shifting paradigms and an signalling pathways and their collaborative nature. However improved understanding of host-biomaterial interactions, Acta Biomater. 9 greater attention to the complete physicochemical characterisation (2013) 4948–4955, http://dx.doi.org/10.1016/j.actbio.2012.10.025. and elimination of contaminants is essential to fully elucidate [16] B.N. Brown, R. Londono, S. Tottey, L. Zhang, K.A. Kukla, M.T. Wolf, K.A. Daly, J. these carbohydrates’ immunomodulatory properties. For this latter E. Reing, S.F. Badylak, Macrophage phenotype as a predictor of constructive remodeling following the implantation of biologically derived surgical mesh reason we strongly encourage the use and continued development materials, Acta Biomater. 8 (2012) 978–987, http://dx.doi.org/10.1016/j. of synthetic glycans to accommodate this need for clarity and actbio.2011.11.031. specificity in future research. Carbohydrates are a rich and valuable [17] A.K. Adak, B.Y. Li, L.D. Huang, T.W. Lin, T.C. Chang, K.C. Hwang, C.C. Lin, Fabrication of antibody microarrays by light-induced covalent and oriented immunomodulatory resource and research should continue to elu- immobilization, ACS Appl. Mater. Interfaces 6 (2014) 10452–10460, http://dx. cidate their roles in macrophage activation and the potential for doi.org/10.1021/am502011r. subsequent therapeutic applications. [18] Y. van Kooyk, G.A. Rabinovich, Protein-glycan interactions in the control of innate and adaptive immune responses, Nat. Immunol. 9 (2008) 593–601, http://dx.doi.org/10.1038/ni.f.203. Conflict of interest disclosure [19] R.L. Schnaar, Glycans and glycan-binding proteins in immune regulation: A concise introduction to glycobiology for the allergist, J. Allergy Clin. Immunol. 135 (2015) 609–615, http://dx.doi.org/10.1016/j.jaci.2014.10.057. No potential conflicts of interest were disclosed. [20] K. Marino, J. Bones, J.J. Kattla, P.M. Rudd, A systematic approach to protein glycosylation analysis: a path through the maze, Nat. Chem. Biol. 6 (2010) 713–723, http://dx.doi.org/10.1038/nchembio.437. Acknowledgements [21] J.D. Marth, P.K. Grewal, Mammalian glycosylation in immunity, Nat. Rev. Immunol. 8 (2008) 874–887, http://dx.doi.org/10.1038/nri2417. [22] I.D. Mascanfroni, J.P. Cerliani, S. Dergan-Dylon, D.O. Croci, J.M. Ilarregui, G.A. The authors would like to acknowledge Trinity College Dublin Rabinovich, Endogenous lectins shape the function of dendritic cells and for a postgraduate studentship to Mimmi Lundahl. Eoin Scanlan tailor adaptive immunity: mechanisms and biomedical applications, Int. 38 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

Immunopharmacol. 11 (2011) 833–841, http://dx.doi.org/10.1016/j. [44] R. Takamiya, K. Ohtsubo, S. Takamatsu, N. Taniguchi, T. Angata, The intimp.2011.01.021. interaction between Siglec-15 and tumor-associated sialyl-Tn antigen [23] J.M. Ilarregui, D.O. Croci, G.A. Bianco, M.A. Toscano, M. Salatino, M.E. enhances TGF-b secretion from monocytes/macrophages through the Vermeulen, J.R. Geffner, G.A. Rabinovich, Tolerogenic signals delivered by DAP12-Syk pathway, Glycobiology 23 (2013) 178–187, http://dx.doi.org/ dendritic cells to T cells through a galectin-1-driven immunoregulatory 10.1093/glycob/cws139. circuit involving interleukin 27 and interleukin 10, Nat. Immunol. 10 (2009) [45] F. Chiodo, P.M. Enriquez-Navas, J. Angulo, M. Marradi, S. Penades, Assembling 981–991, http://dx.doi.org/10.1038/ni.1772. different antennas of the gp120 high mannose-type glycans on gold [24] G.P. Bhide, K.J. Colley, Sialylation of N-glycans: mechanism, cellular nanoparticles provides superior binding to the anti-HIV antibody 2G12 compartmentalization and function, Histochem. Cell Biol. 147 (2017) 149– than the individual antennas, Carbohydr. Res. 405 (2015) 102–109, http://dx. 174, http://dx.doi.org/10.1007/s00418-016-1520-x. doi.org/10.1016/j.carres.2014.07.012. [25] B.S. Bochner, N. Zimmermann, Role of siglecs and related glycan-binding [46] N.A.R. Gow, F.L. van de Veerdonk, A.J.P. Brown, M.G. Netea, Candida albicans proteins in immune responses and immunoregulation, J. Allergy Clin. morphogenesis and host defence: discriminating invasion from colonization, Immunol. 135 (2015) 598–608, http://dx.doi.org/10.1016/j.jaci.2014.11.031. Nat. Rev. Microbiol. 10 (2011) 112, http://dx.doi.org/10.1038/nrmicro2711. [26] I. Fraschilla, S. Pillai, Viewing Siglecs through the lens of tumor immunology, [47] A. Ravidà, A.M. Aldridge, N.N. Driessen, F.A.H. Heus, C.H. Hokke, S.M. O’Neill, Immunol. Rev. 276 (2017) 178–191, http://dx.doi.org/10.1111/imr.12526. P. Robinson, P. Lobel, Z. Misulovin, M. Nussenzweig, T. Feizi, Fasciola hepatica [27] M.S. Macauley, P.R. Crocker, J.C. Paulson, Siglec-mediated regulation of surface coat glycoproteins contain mannosylated and phosphorylated N- immune cell function in disease, Nat. Rev. Immunol. 14 (2014) 653–666, glycans and exhibit immune modulatory properties independent of the http://dx.doi.org/10.1038/nri3737. mannose receptor, PLoS Negl. Trop. Dis. 10 (2016) e0004601, http://dx.doi. [28] G.A. Rabinovich, M.A. Toscano, Turning ‘‘sweet” on immunity: galectin-glycan org/10.1371/journal.pntd.0004601. interactions in immune tolerance and inflammation, Nat. Rev. Immunol. 9 [48] T.L. Lowary, Twenty years of mycobacterial glycans: furanosides and beyond, (2009) 338–352, http://dx.doi.org/10.1038/nri2536. Acc. Chem. Res. 49 (2016) 1379–1388, http://dx.doi.org/10.1021/acs. [29] C. Büll, M.H. den Brok, G.J. Adema, Sweet escape: Sialic acids in tumor accounts.6b00164. immune evasion, Biochim. Biophys. Acta Rev. Cancer 2014 (1846) 238–246, [49] Q. Pan, J. Yan, Q. Liu, C. Yuan, X.-L. Zhang, An ssDNA aptamer against http://dx.doi.org/10.1016/j.bbcan.2014.07.005. mannose-capped lipoarabinomannan enhances anti-tuberculosis activity of [30] C. Büll, M.A. Stoel, M.H. den Brok, G.J. Adema, Sialic acids sweeten a tumor’s macrophages through downregulation of lipid-sensing nuclear receptor life, Cancer Res. 74 (2014) 3199–3204, http://dx.doi.org/10.1158/0008-5472. peroxisome proliferator-activated receptor c expression, Microbiol. CAN-14-0728. Immunol. (2017), http://dx.doi.org/10.1111/1348-0421.12470. [31] S. Roy, C. Mandal, S. Sun, L. Mao, X. Chu, H. Deng, M. Tremblay, Leishmania [50] J.B. Torrelles, A.K. Azad, L.S. Schlesinger, Fine discrimination in the donovani utilize sialic acids for binding and phagocytosis in the macrophages recognition of individual species of phosphatidyl-myo-inositol mannosides through selective utilization of siglecs and impair the innate immune arm, from Mycobacterium tuberculosis by C-type lectin pattern recognition PLoS Negl. Trop. Dis. 10 (2016) e0004904, http://dx.doi.org/10.1371/journal. receptors, J. Immunol. 177 (2006) 1805–1816, http://dx.doi.org/10.4049/ pntd.0004904. jimmunol.177.3.1805. [32] G. Sciences, A Phase 2, Randomized, Double-Blind, Placebo-Controlled Study [51] R.A. Hall, N.A.R. Gow, Mannosylation in Candida albicans: role in cell wall to Assess the Efficacy and Tolerability of Entospletinib, a Selective SYK function and immune recognition, Mol. Microbiol. 90 (2013) 1147–1161, Inhibitor, in Combination With Systemic Corticosteroids as First-Line Therapy http://dx.doi.org/10.1111/mmi.12426. in Subjects With Chronic Graft Versus Host Disease (cGVHD), [52] A.V. Chavez-Santoscoy, R. Roychoudhury, N.L. Pohl, M.J. Wannemuehler, B. ClinicalTrials.gov, 2017. doi: NCT02701634. Narasimhan, A.E. Ramer-Tait, Tailoring the immune response by targeting C- [33] J. Sharman, M. Hawkins, K. Kolibaba, M. Boxer, L. Klein, M. Wu, J. Hu, S. Abella, type lectin receptors on alveolar macrophages using ‘‘pathogen-like” C. Yasenchak, An open-label phase 2 trial of entospletinib (GS-9973), a amphiphilic polyanhydride nanoparticles, Biomaterials 33 (2012) 4762– selective spleen tyrosine kinase inhibitor, in chronic lymphocytic leukemia, 4772, http://dx.doi.org/10.1016/j.biomaterials.2012.03.027. Blood 125 (2015) 2336–2343, http://dx.doi.org/10.1182/blood-2014-08- [53] F. Al-Barwani, S.L. Young, M.A. Baird, D.S. Larsen, V.K. Ward, Mannosylation of 595934. virus-like particles enhances internalization by antigen presenting cells, PLoS [34] M.D. Anderson Cancer Center, A Phase I, Open-Label, Dose Escalation Study to One. 9 (2014) e104523, http://dx.doi.org/10.1371/journal.pone.0104523. Evaluate the Safety, Tolerability, and Pharmacokinetics of Sodium [54] W. Yao, Y. Peng, M. Du, J. Luo, L. Zong, Preventative vaccine-loaded Stibogluconate in Combination With Interferon Alpha-2b for Patients With mannosylated chitosan nanoparticles intended for nasal mucosal delivery Advanced Malignancies, ClinicalTrials.gov, 2011. doi: NCT00629200. enhance immune responses and potent tumor immunity, Mol. Pharm. 10 [35] Case Comprehensive Cancer Center, Phase I Evaluation of Sodium (2013) 2904–2914, http://dx.doi.org/10.1021/mp4000053. Stibogluconate in Combination With Interferon a-2b Followed by [55] V. Kéry, J.J.F. Krˇepinsky´ , C.D. Warren, P. Capek, P.D. Stahl, Ligand recognition Cisplatin, Vinblastine and Dacarbazine for Patients With Melanoma or by purified human mannose receptor, Arch. Biochem. Biophys. 298 (1992) Malignancies Potentially Responsive to SSG and/or Interferons, 49–55, http://dx.doi.org/10.1016/0003-9861(92)90092-B. ClinicalTrials.gov. (2015). [56] S. Saijo, S. Ikeda, K. Yamabe, S. Kakuta, H. Ishigame, A. Akitsu, N. Fujikado, T. [36] M.K. Pathak, T. Yi, Sodium stibogluconate is a potent inhibitor of protein Kusaka, S. Kubo, S. Chung, R. Komatsu, N. Miura, Y. Adachi, N. Ohno, K. tyrosine phosphatases and augments cytokine responses in hemopoietic cell Shibuya, N. Yamamoto, K. Kawakami, S. Yamasaki, T. Saito, S. Akira, Y. lines, J. Immunol. 167 (2001) 3391–3397, http://dx.doi.org/10.4049/ Iwakura, Dectin-2 recognition of a-mannans and induction of Th17 cell jimmunol.167.6.3391. differentiation is essential for host defense against Candida albicans, [37] S. Spence, M.K. Greene, F. Fay, E. Hams, S.P. Saunders, U. Hamid, M. Fitzgerald, Immunity 32 (2010) 681–691, http://dx.doi.org/10.1016/j. J. Beck, B.K. Bains, P. Smyth, E. Themistou, D.M. Small, D. Schmid, C.M. O’Kane, immuni.2010.05.001. D.C. Fitzgerald, S.M. Abdelghany, J.A. Johnston, P.G. Fallon, J.F. Burrows, D.F. [57] L. Martinez-Pomares, The mannose receptor, J. Leukoc. Biol. 92 (2012) 1177– McAuley, A. Kissenpfennig, C.J. Scott, Targeting Siglecs with a sialic acid- 1186, http://dx.doi.org/10.1189/jlb.0512231. decorated nanoparticle abrogates inflammation, Sci. Transl. Med. 7 (2015) [58] M.V.S. Rajaram, M.N. Brooks, J.D. Morris, J.B. Torrelles, A.K. Azad, L.S. 303ra140, http://dx.doi.org/10.1126/scitranslmed.aab3459. Schlesinger, Mycobacterium tuberculosis activates human macrophage [38] A.G. Barrientos, J.M. de la Fuente, T.C. Rojas, A. Fernández, S. Penadés, Gold peroxisome proliferator-activated receptor linking mannose receptor glyconanoparticles: synthetic polyvalent ligands mimicking glycocalyx-like recognition to regulation of immune responses, J. Immunol. 185 (2010) surfaces as tools for glycobiological studies, Chemistry 9 (2003) 1909–1921, 929–942, http://dx.doi.org/10.4049/jimmunol.1000866. http://dx.doi.org/10.1002/chem.200204544. [59] F.L. van de Veerdonk, R.J. Marijnissen, B.J. Kullberg, H.J.P.M. Koenen, S.-C. [39] R. Cao-Milan, L.M. Liz-Marzan, Gold nanoparticle conjugates: recent advances Cheng, I. Joosten, W.B. van den Berg, D.L. Williams, J.W.M. van der Meer, L.A.B. toward clinical applications, Expert Opin. Drug Deliv. 11 (2014) 741–752, Joosten, M.G. Netea, The macrophage mannose receptor induces IL-17 in http://dx.doi.org/10.1517/17425247.2014.891582. response to Candida albicans, Cell Host Microbe 5 (2009) 329–340, http://dx. [40] H. Läubli, O.M.T. Pearce, F. Schwarz, S.S. Siddiqui, L. Deng, M.A. Stanczak, doi.org/10.1016/j.chom.2009.02.006. L. Deng, A. Verhagen, P. Secrest, C. Lusk, A.G. Schwartz, N.M. Varki, J.D. [60] F.V. Loures, E.F. Araújo, C. Feriotti, S.B. Bazan, V.L.G. Calich, TLR-4 cooperates Bui, A. Varki, Engagement of myelomonocytic Siglecs by tumor-associated with Dectin-1 and mannose receptor to expand Th17 and Tc17 cells induced ligands modulates the innate immune response to cancer, Proc. Natl. by Paracoccidioides brasiliensis stimulated dendritic cells, Front. Microbiol. 6 Acad. Sci. USA 111 (2014) 14211–14216, http://dx.doi.org/10.1073/ (2015) 261, http://dx.doi.org/10.3389/fmicb.2015.00261. pnas.1409580111. [61] S. Saijo, Y. Iwakura, Dectin-1 and Dectin-2 in innate immunity against fungi, [41] S. Varchetta, E. Brunetta, A. Roberto, J. Mikulak, K.L. Hudspeth, M.U. Mondelli, Int. Immunol. 23 (2011) 467–472, http://dx.doi.org/10.1093/intimm/dxr046. D. Mavilio, Engagement of Siglec-7 receptor induces a pro-inflammatory [62] K. Sato, X. Yang, T. Yudate, J.-S. Chung, J. Wu, K. Luby-Phelps, R.P. Kimberly, D. response selectively in monocytes, PLoS One. 7 (2012) e45821, http://dx.doi. Underhill, P.D. Cruz, K. Ariizumi, Dectin-2 is a pattern recognition receptor for org/10.1371/journal.pone.0045821. fungi that couples with the Fc receptor gamma chain to induce innate [42] C. Büll, E. Collado-Camps, E.D. Kers-Rebel, T. Heise, J.N. Søndergaard, M.H. den immune responses, J. Biol. Chem. 281 (2006) 38854–38866, http://dx.doi.org/ Brok, B.M. Schulte, T.J. Boltje, G.J. Adema, Metabolic sialic acid blockade 10.1074/jbc.M606542200. lowers the activation threshold of moDCs for TLR stimulation, Immunol. Cell [63] K. Ariizumi, G.L. Shen, S. Shikano, R. Ritter, P. Zukas, D. Edelbaum, A. Morita, A. Biol. 95 (2017) 408–415, http://dx.doi.org/10.1038/icb.2016.105. Takashima, Cloning of a second dendritic cell-associated C-type lectin [43] Y. Wu, C. Lan, D. Ren, G.-Y. Chen, Induction of Siglec-1 by endotoxin tolerance (dectin-2) and its alternatively spliced isoforms, J. Biol. Chem. 275 (2000) suppresses the innate immune response by promoting TGF-b1 production, J. 11957–11963, http://dx.doi.org/10.1074/jbc.275.16.11957. Biol. Chem. 291 (2016) 12370–12382, http://dx.doi.org/10.1074/jbc. [64] P.R. Taylor, D.M. Reid, S.E.M. Heinsbroek, G.D. Brown, S. Gordon, S.Y.C. Wong, M116.721258. Dectin-2 is predominantly myeloid restricted and exhibits unique activation- M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 39

dependent expression on maturing inflammatory monocytes elicited in vivo, [84] R.A. Hall, N.A.R. Gow, Mannosylation in Candida albicans : role in cell wall Eur. J. Immunol. 35 (2005) 2163–2174, http://dx.doi.org/10.1002/ function and immune recognition, Mol. Microbiol. 90 (2013) 1147–1161, eji.200425785. http://dx.doi.org/10.1111/mmi.12426. [65] H. Sun, X. Xu, X. Tian, H. Shao, X. Wu, Q. Wang, X. Su, Y. Shi, Activation of NF- [85] M. Liu, F. Luo, C. Ding, S. Albeituni, X. Hu, Y. Ma, Y. Cai, L. McNally, M.A. jB and respiratory burst following Aspergillus fumigatus stimulation of Sanders, D. Jain, G. Kloecker, M. Bousamra, H. Zhang, R.M. Higashi, A.N. Lane, macrophages, Immunobiology 219 (2014) 25–36, http://dx.doi.org/10.1016/j. T.W.-M. Fan, J. Yan, Dectin-1 activation by a natural product b-glucan imbio.2013.06.013. converts immunosuppressive macrophages into an M1-like phenotype, J. [66] A. Yonekawa, S. Saijo, Y. Hoshino, Y. Miyake, E. Ishikawa, M. Suzukawa, H. Immunol. 195 (2015) 5055–5065, http://dx.doi.org/10.4049/ Inoue, M. Tanaka, M. Yoneyama, M. Oh-Hora, K. Akashi, S. Yamasaki, Dectin-2 jimmunol.1501158. is a direct receptor for mannose-capped lipoarabinomannan of mycobacteria, [86] A. Torosantucci, C. Bromuro, P. Chiani, F. De Bernardis, F. Berti, C. Galli, F. Immunity 41 (2014) 402–413, http://dx.doi.org/10.1016/j. Norelli, C. Bellucci, L. Polonelli, P. Costantino, R. Rappuoli, A. Cassone, A novel immuni.2014.08.005. glyco-conjugate vaccine against fungal pathogens, J. Exp. Med. 202 (2005) [67] X. Sun, Q. Pan, C. Yuan, Q. Wang, X.-L. Tang, K. Ding, X. Zhou, X.-L. Zhang, A 597–606, http://dx.doi.org/10.1084/jem.20050749. Single ssDNA aptamer binding to mannose-capped lipoarabinomannan of [87] M.E. Danielson, M. Martinez, M.A. Antonysamy, K.S. Michel, D.A. Stevens, K.V. bacillus calmette-guérin enhances immunoprotective effect against Clemons, V. Chen, Whole glucan particles as a vaccine against systemic tuberculosis, J. Am. Chem. Soc. 138 (2016) 11680–11689, http://dx.doi.org/ coccidioidomycosis, J. Med. Microbiol. 64 (2015) 1237–1243, http://dx.doi. 10.1021/jacs.6b05357. org/10.1099/jmm.0.000138. [68] T. Weichhart, M. Hengstschläger, M. Linke, Regulation of innate immune cell [88] Q. Wang, X. Sheng, A. Shi, H. Hu, Y. Yang, L. Liu, L. Fei, H. Liu, b-Glucans: function by mTOR, Nat. Rev. Immunol. 15 (2015) 599–614, http://dx.doi.org/ relationships between modification, conformation and functional activities, 10.1038/nri3901. Molecules 22 (2017) 257, http://dx.doi.org/10.3390/molecules22020257. [69] V. Byles, A.J. Covarrubias, I. Ben-Sahra, D.W. Lamming, D.M. Sabatini, B.D. [89] K.V. Clemons, M.E. Danielson, K.S. Michel, M. Liu, N.C. Ottoson, S.M. Leonardo, Manning, T. Horng, The TSC-mTOR pathway regulates macrophage M. Martinez, V. Chen, M.A. Antonysamy, D.A. Stevens, Whole glucan particles polarization, Nat. Commun. 4 (2013) 1237–1247, http://dx.doi.org/10.1038/ as a vaccine against murine aspergillosis, J. Med. Microbiol. 63 (2014) 1750– ncomms3834. 1759, http://dx.doi.org/10.1099/jmm.0.079681-0. [70] P.K. Langston, M. Shibata, T. Horng, Metabolism supports macrophage [90] L. Zhang, X. Li, X. Xu, F. Zeng, Correlation between antitumor activity, activation, Front. Immunol. 8 (2017) 61, http://dx.doi.org/ molecular weight, and conformation of lentinan, Carbohydr. Res. 340 (2005) 10.3389/fimmu.2017.00061. 1515–1521, http://dx.doi.org/10.1016/j.carres.2005.02.032. [71] V. Schuette, M. Embgenbroich, T. Ulas, M. Welz, J. Schulte-Schrepping, A.M. [91] H.S. Goodridge, C.N. Reyes, C.A. Becker, T.R. Katsumoto, J. Ma, A.J. Wolf, N. Draffehn, T. Quast, K. Koch, M. Nehring, J. König, A. Zweynert, F.L. Harms, N. Bose, A.S.H. Chan, A.S. Magee, M.E. Danielson, A. Weiss, J.P. Vasilakos, D.M. Steiner, A. Limmer, I. Förster, F. Berberich-Siebelt, P.A. Knolle, D. Wohlleber, Underhill, Activation of the innate immune receptor Dectin-1 upon formation W. Kolanus, M. Beyer, J.L. Schultze, S. Burgdorf, Mannose receptor induces T- of a ‘‘phagocytic synapse”, Nature 472 (2011) 471–475, http://dx.doi.org/ cell tolerance via inhibition of CD45 and up-regulation of CTLA-4, Proc. Natl. 10.1038/nature10071. Acad. Sci. USA 113 (2016) 10649–10654, http://dx.doi.org/10.1073/ [92] G. Liao, S. Burgula, Z. Zhou, Z. Guo, A convergent synthesis of 6-O-branched b- pnas.1605885113. glucan oligosaccharides, Eur. J. Org. Chem. 2015 (2015) 2942–2951, http://dx. [72] M.G. Netea, G.D. Brown, B.J. Kullberg, N.A.R. Gow, An integrated model of the doi.org/10.1002/ejoc.201500229. recognition of Candida albicans by the innate immune system, Nat. Rev. [93] D.J. Manners, A.J. Masson, J.C. Patterson, The structure of a beta-(1 leads to 3)- Microbiol. 6 (2008) 67–78, http://dx.doi.org/10.1038/nrmicro1815. D-glucan from yeast cell walls, Biochem. J. 135 (1973) 19–30, http://dx.doi. [73] J.E. Cutler, N-glycosylation of yeast, with emphasis on Candida albicans, Med. org/10.1042/bj1350019. Mycol. 39 (Suppl. 1) (2001) 75–86, http://dx.doi.org/10.1080/ [94] D.J. Manners, A.J. Masson, J.C. Patterson, H. Björndal, B. Lindberg, The mmy.39.1.75.86. structure of a beta-(1–6)-D-glucan from yeast cell walls, Biochem. J. 135 [74] C. Timpel, S. Strahl-Bolsinger, K. Ziegelbauer, J.F. Ernst, Multiple functions of (1973) 31–36, http://dx.doi.org/10.1042/bj1350031. Pmt1p-mediated protein O-mannosylation in the fungal pathogen Candida [95] S. Walachowski, G. Tabouret, G. Foucras, J. Bain, H. Mora-Montes, R. Barker, albicans, J. Biol. Chem. 273 (1998) 20837–20846, http://dx.doi.org/10.1074/ Triggering dectin-1-pathway alone is not sufficient to induce cytokine jbc.273.33.20837. production by murine macrophages, PLoS One 11 (2016) e0148464, http:// [75] I. Khmaladze, T. Kelkka, S. Guerard, K. Wing, A. Pizzolla, A. Saxena, K. dx.doi.org/10.1371/journal.pone.0148464. Lundqvist, M. Holmdahl, K.S. Nandakumar, R. Holmdahl, Mannan induces [96] G. Liao, Z. Zhou, J. Liao, L. Zu, Q. Wu, Z. Guo, 6-O-branched oligo-b-glucan- ROS-regulated, IL-17A-dependent psoriasis arthritis-like disease in mice, based antifungal glycoconjugate vaccines, ACS Infect. Dis. 2 (2016) 123–131, Proc. Natl. Acad. Sci. 111 (2014) E3669–E3678, http://dx.doi.org/10.1073/ http://dx.doi.org/10.1021/acsinfecdis.5b00104. pnas.1405798111. [97] J.W.M. van der Meer, L.A.B. Joosten, N. Riksen, M.G. Netea, Trained immunity: [76] D.C. Ifrim, J. Quintin, F. Courjol, I. Verschueren, J.H. van Krieken, F. Koentgen, a smart way to enhance innate immune defence, Mol. Immunol. 68 (2015) C. Fradin, N.A.R. Gow, L.A.B. Joosten, J.W.M. van der Meer, F. van de Veerdonk, 40–44, http://dx.doi.org/10.1016/j.molimm.2015.06.019. M.G. Netea, The role of Dectin-2 for host defense against disseminated [98] P. Garcia-Valtanen, R.M. Guzman-Genuino, D.L. Williams, J.D. Hayball, K.R. candidiasis, J. Interf. Cytokine Res. 36 (2016) 267–276, http://dx.doi.org/ Diener, Evaluation of trained immunity by b-1, 3 (d)-glucan on murine 10.1089/jir.2015.0040. monocytes in vitro and duration of response in vivo, Immunol. Cell Biol. [77] D. Elieh-Ali-Komi, M.R. Hamblin, Chitin and chitosan: production and (2017), http://dx.doi.org/10.1038/icb.2017.13. application of versatile biomedical nanomaterials, Int. J. Adv. Res. 4 (2016) [99] J. Quintin, S. Saeed, J.H.A. Martens, E.J. Giamarellos-Bourboulis, D.C. Ifrim, C. 411–427. Logie, L. Jacobs, T. Jansen, B.-J. Kullberg, C. Wijmenga, L.A.B. Joosten, R.J. [78] V.V. Mutsenko, O. Gryshkov, L. Lauterboeck, O. Rogulska, D.N. Tarusin, V.V. Xavier, J.W.M. van der Meer, H.G. Stunnenberg, M.G. Netea, Candida albicans Bazhenov, K. Schütz, S. Brüggemeier, E. Gossla, A.R. Akkineni, H. Meißner, A. infection affords protection against reinfection via functional reprogramming Lode, S. Meschke, J. Fromont, A.L. Stelling, K.R. Tabachnik, M. Gelinsky, S. of monocytes, Cell Host Microbe 12 (2012) 223–232, http://dx.doi.org/ Nikulin, S. Rodin, A.G. Tonevitsky, A.Y. Petrenko, B. Glasmacher, P.J. Schupp, H. 10.1016/j.chom.2012.06.006. Ehrlich, Novel chitin scaffolds derived from marine sponge Ianthella basta for [100] S.-C. Cheng, J. Quintin, R.A. Cramer, K.M. Shepardson, S. Saeed, V. Kumar, E.J. tissue engineering approaches based on human mesenchymal stromal cells: Giamarellos-Bourboulis, J.H.A. Martens, N.A. Rao, A. Aghajanirefah, G.R. Biocompatibility and cryopreservation, Int. J. Biol. Macromol. (2017), http:// Manjeri, Y. Li, D.C. Ifrim, R.J.W. Arts, B.M.J.W. van der Veer, B.M.J.W. van dx.doi.org/10.1016/j.ijbiomac.2017.03.161. der Meer, P.M.T. Deen, C. Logie, L.A. O’Neill, P. Willems, F.L. van de Veerdonk, [79] Z. Jamalpoor, H. Mirzadeh, M.T. Joghataei, D. Zeini, S. Bagheri-Khoulenjani, M. J.W.M. van der Meer, A. Ng, L.A.B. Joosten, C. Wijmenga, H.G. Stunnenberg, R.J. R. Nourani, Fabrication of cancellous biomimetic chitosan-based Xavier, M.G. Netea, mTOR- and HIF-1a-mediated aerobic glycolysis as nanocomposite scaffolds applying a combinational method for bone tissue metabolic basis for trained immunity, Science 345 (2014) 1250684, http:// engineering, J. Biomed. Mater. Res. Part A 103 (2015) 1882–1892, http://dx. dx.doi.org/10.1126/science.1250684. doi.org/10.1002/jbm.a.35320. [101] S. Saeed, J. Quintin, H.H.D. Kerstens, N.A. Rao, A. Aghajanirefah, F. Matarese, [80] S. Saravanan, R.S. Leena, N. Selvamurugan, Chitosan based biocomposite S.-C. Cheng, J. Ratter, K. Berentsen, M.A. van der Ent, N. Sharifi, E.M. Janssen- scaffolds for bone tissue engineering, Int. J. Biol. Macromol. 93 (2016) 1354– Megens, M. Ter Huurne, A. Mandoli, T. van Schaik, A. Ng, F. Burden, K. 1365, http://dx.doi.org/10.1016/j.ijbiomac.2016.01.112. Downes, M. Frontini, V. Kumar, E.J. Giamarellos-Bourboulis, W.H. [81] W. Wang, S. Itoh, A. Matsuda, S. Ichinose, K. Shinomiya, Y. Hata, J. Tanaka, Ouwehand, J.W.M. van der Meer, L.A.B. Joosten, C. Wijmenga, J.H.A. Influences of mechanical properties and permeability on chitosan nano/ Martens, R.J. Xavier, C. Logie, M.G. Netea, H.G. Stunnenberg, Epigenetic microfiber mesh tubes as a scaffold for nerve regeneration, J. Biomed. Mater. programming of monocyte-to-macrophage differentiation and trained Res. Part A. 84A (2008) 557–566, http://dx.doi.org/10.1002/jbm.a.31536. innate immunity, Science 345 (2014) 1251086, http://dx.doi.org/ [82] T. Freier, R. Montenegro, H. Shan Koh, M.S. Shoichet, Chitin-based tubes for 10.1126/science.1251086. tissue engineering in the nervous system, Biomaterials 26 (2005) 4624–4632, [102] D. Elieh Ali Komi, L. Sharma, C.S. Dela Cruz, Chitin and Its Effects on http://dx.doi.org/10.1016/j.biomaterials.2004.11.040. Inflammatory and Immune Responses, Clin. Rev. Allergy Immunol. (2017), [83] R. Jayakumar, M. Prabaharan, P.T. Sudheesh Kumar, S.V. Nair, H. Tamura, http://dx.doi.org/10.1007/s12016-017-8600-0. Biomaterials based on chitin and chitosan in wound dressing applications, [103] C.A. Da Silva, D. Hartl, W. Liu, C.G. Lee, J.A. Elias, TLR-2 and IL-17A in chitin- Biotechnol. Adv. 29 (2011) 322–337, http://dx.doi.org/10.1016/j. induced macrophage activation and acute inflammation, J. Immunol. 181 biotechadv.2011.01.005. (2008) 4279–4286, http://dx.doi.org/10.4049/jimmunol.181.6.4279. 40 M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41

[104] N. Amarsaikhan, S.P. Templeton, Co-recognition of b-glucan and chitin and [126] Y. Han, L. Zhao, Z. Yu, J. Feng, Q. Yu, Role of mannose receptor in programming of adaptive immunity to Aspergillus fumigatus, Front. oligochitosan-mediated stimulation of macrophage function, Int. Microbiol. 6 (2015) 344, http://dx.doi.org/10.3389/fmicb.2015.00344. Immunopharmacol. 5 (2005) 1533–1542, http://dx.doi.org/10.1016/j. [105] T. Satoh, O. Takeuchi, A. Vandenbon, K. Yasuda, Y. Tanaka, Y. Kumagai, T. intimp.2005.04.015. Miyake, K. Matsushita, T. Okazaki, T. Saitoh, K. Honma, T. Matsuyama, K. Yui, [127] J. Feng, L. Zhao, Q. Yu, Receptor-mediated stimulatory effect of oligochitosan T. Tsujimura, D.M. Standley, K. Nakanishi, K. Nakai, S. Akira, The Jmjd3-Irf4 in macrophages, Biochem. Biophys. Res. Commun. 317 (2004) 414–420, axis regulates M2 macrophage polarization and host responses against http://dx.doi.org/10.1016/j.bbrc.2004.03.048. helminth infection, Nat. Immunol. 11 (2010) 936–944, http://dx.doi.org/ [128] Z. Yu, L. Zhao, H. Ke, Potential role of nuclear factor-kappaB in the induction 10.1038/ni.1920. of nitric oxide nd tumor necrosis factor-alpha by oligochitosan in [106] S. Tundup, L. Srivastava, D.A. Harn, Polarization of host immune responses by macrophages, Int. Immunopharmacol. 4 (2004) 193–200, http://dx.doi.org/ helminth-expressed glycans, Ann. N. Y. Acad. Sci. 1253 (2012) E1–E13, http:// 10.1016/j.intimp.2003.12.001. dx.doi.org/10.1111/j.1749-6632.2012.06618.x. [129] M.I. Oliveira, S.G. Santos, M.J. Oliveira, A.L. Torres, M.A. Barbosa, Chitosan [107] C.A. Da Silva, C. Chalouni, A. Williams, D. Hartl, C.G. Lee, J.A. Elias, Chitin is a drives anti-inflammatory macrophage polarisation and pro-inflammatory size-dependent regulator of macrophage TNF and IL-10 production, J. dendritic cell stimulation, Eur. Cell. Mater. 24 (2012) 136–152, http://dx.doi. Immunol. 182 (2009), http://dx.doi.org/10.4049/jimmunol.0802113. org/10.22203/eCM.v024a10. [108] C.J. Wagner, S. Huber, S. Wirth, D. Voehringer, Chitin induces upregulation of [130] E. Mills, L.A.J. O’Neill, Succinate: a metabolic signal in inflammation, B7–H1 on macrophages and inhibits T-cell proliferation, Eur. J. Immunol. 40 Trends Cell Biol. 24 (2014) 313–320, http://dx.doi.org/10.1016/j. (2010) 2882–2890, http://dx.doi.org/10.1002/eji.201040422. tcb.2013.11.008. [109] B. Prietl, C. Meindl, E. Roblegg, T.R. Pieber, G. Lanzer, E. Frohlich, Nano-sized [131] W.K.E. Ip, N. Hoshi, D.S. Shouval, S. Snapper, R. Medzhitov, Anti-inflammatory and micro-sized polystyrene particles affect phagocyte function, Cell Biol. effect of IL-10 mediated by metabolic reprogramming of macrophages, Toxicol. 30 (2014) 1–16, http://dx.doi.org/10.1007/s10565-013-9265-y. Science (80-.) 356 (2017) 513–519, http://dx.doi.org/10.1126/science. [110] L.J. Cruz, P.J. Tacken, F. Rueda, J.C. Domingo, F. Albericio, C.G. Figdor, Targeting aal3535. nanoparticles to dendritic cells for immunotherapy, Methods Enzym. 509 [132] P. Seth, E. Csizmadia, A. Hedblom, M. Vuerich, H. XIE, M. Li, M.S. Longhi, B. (2012) 143–163, http://dx.doi.org/10.1016/b978-0-12-391858-1.00008-3. Wegiel, Deletion of lactate dehydrogenase-A in myeloid cells triggers [111] C.L. Hardy, J.S. Lemasurier, R. Mohamud, J. Yao, S.D. Xiang, J.M. Rolland, R.E. antitumor immunity, Cancer Res (2017), http://dx.doi.org/10.1158/0008- O’Hehir, M. Plebanski, Differential uptake of nanoparticles and microparticles by 5472.CAN-16-2938. pulmonary APC subsets induces discrete immunological imprints, J. Immunol. [133] J. Van den Bossche, L.A. O’Neill, D. Menon, Macrophage immunometabolism: 191 (2013) 5278–5290, http://dx.doi.org/10.4049/jimmunol.1203131. where are we (Going)? Trends Immunol. (2017), http://dx.doi.org/10.1016/j. [112] X. Liu, N. Huang, H. Li, Q. Jin, J. Ji, Surface and size effects on cell interaction of it.2017.03.001. gold nanoparticles with both phagocytic and nonphagocytic cells, Langmuir [134] E.L. Mills, B. Kelly, L.A.J. O’Neill, Mitochondria are the powerhouses of 29 (2013) 9138–9148, http://dx.doi.org/10.1021/la401556k. immunity, Nat. Immunol. 18 (2017) 488–498, http://dx.doi.org/10.1038/ [113] R. Mohamud, S.D. Xiang, C. Selomulya, J.M. Rolland, R.E. O’Hehir, C.L. Hardy, ni.3704. M. Plebanski, The effects of engineered nanoparticles on pulmonary immune [135] E.L. Mills, B. Kelly, A. Logan, A.S.H. Costa, M. Varma, C.E. Bryant, P. homeostasis, Drug Metab. Rev. 46 (2014) 176–190, http://dx.doi.org/ Tourlomousis, J.H.M. Däbritz, E. Gottlieb, I. Latorre, S.C. Corr, G. McManus, 10.3109/03602532.2013.859688. D. Ryan, H.T. Jacobs, M. Szibor, R.J. Xavier, T. Braun, C. Frezza, M.P. Murphy, L. [114] L.K. Dubey, J.B. Moeller, A. Schlosser, G.L. Sorensen, U. Holmskov, Induction of A. O’Neill, Succinate dehydrogenase supports metabolic repurposing of innate immunity by Aspergillus fumigatus cell wall polysaccharides is mitochondria to drive inflammatory macrophages, Cell 167 (2016), http:// enhanced by the composite presentation of chitin and beta-glucan, dx.doi.org/10.1016/j.cell.2016.08.064. Immunobiology 219 (2014) 179–188, http://dx.doi.org/10.1016/j. [136] S.C.-C. Huang, A.M. Smith, B. Everts, M. Colonna, E.L. Pearce, J.D. Schilling, E.J. imbio.2013.10.003. Pearce, Metabolic reprogramming mediated by the mTORC2-IRF4 signaling [115] J. Wagener, R.K.S. Malireddi, M.D. Lenardon, M. Köberle, S. Vautier, D.M. axis is essential for macrophage alternative activation, Immunity 45 (2016) MacCallum, T. Biedermann, M. Schaller, M.G. Netea, T.-D. Kanneganti, G.D. 817–830, http://dx.doi.org/10.1016/j.immuni.2016.09.016. Brown, A.J.P. Brown, N.A.R. Gow, Fungal chitin dampens inflammation [137] A.J. Covarrubias, H.I. Aksoylar, J. Yu, N.W. Snyder, A.J. Worth, S.S. Iyer, J. Wang, through IL-10 induction mediated by NOD2 and TLR9 activation, PLoS I. Ben-Sahra, V. Byles, T. Polynne-Stapornkul, E.C. Espinosa, D. Lamming, B.D. Pathog. 10 (2014) e1004050, http://dx.doi.org/10.1371/journal. Manning, Y. Zhang, I.A. Blair, T. Horng, Akt-mTORC1 signaling regulates Acly ppat.1004050. to integrate metabolic input to control of macrophage activation, Elife 5 [116] R.M. Clancy, A.J. Markham, J.P. Buyon, Endosomal Toll-like receptors in (2016), http://dx.doi.org/10.7554/eLife.11612. clinically overt and silent autoimmunity, Immunol. Rev. 269 (2016) 76–84, [138] G.M. Tannahill, A.M. Curtis, J. Adamik, E.M. Palsson-McDermott, A.F. http://dx.doi.org/10.1111/imr.12383. McGettrick, G. Goel, C. Frezza, N.J. Bernard, B. Kelly, N.H. Foley, L. Zheng, [117] R. Caruso, N. Warner, N. Inohara, G. Núñez, NOD1 and NOD2: signaling, Host A. Gardet, Z. Tong, S.S. Jany, S.C. Corr, M. Haneklaus, B.E. Caffrey, K. Pierce, Defense Inflam. Dis. Immun. 41 (2014) 898–908, http://dx.doi.org/10.1016/j. S. Walmsley, F.C. Beasley, E. Cummins, V. Nizet, M. Whyte, C.T. Taylor, H. immuni.2014.12.010. Lin, S.L. Masters, E. Gottlieb, V.P. Kelly, C. Clish, P.E. Auron, R.J. Xavier, L.A.J. [118] J. Kleinnijenhuis, M. Oosting, L.A.B. Joosten, M.G. Netea, R. Van Crevel, Innate O’Neill, Succinate is an inflammatory signal that induces IL-1b through immune recognition of Mycobacterium tuberculosis, Clin. Dev. Immunol. HIF-1a, Nature 496 (2013) 238–242, http://dx.doi.org/ 2011 (2011) 405310, http://dx.doi.org/10.1155/2011/405310. 10.1038/nature11986. [119] Y. Yang, M. Guo, R. Qian, C. Liu, X. Zong, Y.-Q. Li, W. Li, Binding efficacy and [139] L. Liu, Y. Lu, J. Martinez, Y. Bi, G. Lian, T. Wang, S. Milasta, J. Wang, M. Yang, G. kinetics of chitosan with DNA duplex: the effects of deacetylation degree and Liu, D.R. Green, R. Wang, Proinflammatory signal suppresses proliferation and nucleotide sequences, Carbohydr. Polym. 169 (2017) 451–457, http://dx.doi. shifts macrophage metabolism from Myc-dependent to HIF1a-dependent, org/10.1016/j.carbpol.2017.04.040. Proc. Natl. Acad. Sci. USA 113 (2016) 1564–1569, http://dx.doi.org/10.1073/ [120] C.L. Bueter, C.K. Lee, V.A.K. Rathinam, G.J. Healy, C.H. Taron, C.A. Specht, S.M. pnas.1518000113. Levitz, Chitosan but not chitin activates the inflammasome by a mechanism [140] S.J. Lawless, N. Kedia-Mehta, J.F. Walls, R. McGarrigle, O. Convery, L.V. dependent upon phagocytosis, J. Biol. Chem. 286 (2011) 35447–35455, Sinclair, M.N. Navarro, J. Murray, D.K. Finlay, Glucose represses dendritic cell- http://dx.doi.org/10.1074/jbc.M111.274936. induced T cell responses, Nat. Commun. 8 (2017) 15620, http://dx.doi.org/ [121] C.L. Bueter, C.K. Lee, J.P. Wang, G.R. Ostroff, C.A. Specht, S.M. Levitz, Spectrum 10.1038/ncomms15620. and mechanisms of inflammasome activation by chitosan, J. Immunol. 192 [141] S. Pavlou, L. Wang, H. Xu, M. Chen, Higher phagocytic activity of (2014) 5943–5951, http://dx.doi.org/10.4049/jimmunol.1301695. thioglycollate-elicited peritoneal macrophages is related to metabolic [122] E.C. Carroll, L. Jin, A. Mori, N. Muñoz-Wolf, E. Oleszycka, H.B.T. Moran, S. status of the cells, J. Inflamm. 14 (2017) 4, http://dx.doi.org/10.1186/ Mansouri, C.P. McEntee, E. Lambe, E.M. Agger, P. Andersen, C. Cunningham, P. s12950-017-0151-x. Hertzog, K.A. Fitzgerald, A.G. Bowie, E.C. Lavelle, The vaccine adjuvant [142] L. Tretter, A. Patocs, C. Chinopoulos, Succinate, an intermediate in chitosan promotes cellular immunity via dna sensor cGAS-STING-dependent metabolism, signal transduction, ROS, hypoxia, and tumorigenesis, Biochim. induction of type I interferons, Immunity 44 (2016) 597–608, http://dx.doi. Biophys. Acta 1857 (2016) 1086–1101, http://dx.doi.org/10.1016/j. org/10.1016/j.immuni.2016.02.004. bbabio.2016.03.012. [123] A. Mori, E. Oleszycka, F.A. Sharp, M. Coleman, Y. Ozasa, M. Singh, D.T. [143] B. Kalyanaraman, Teaching the basics of cancer metabolism: developing O’Hagan, L. Tajber, O.I. Corrigan, E.A. McNeela, E.C. Lavelle, The vaccine antitumor strategies by exploiting the differences between normal and adjuvant alum inhibits IL-12 by promoting PI3 kinase signaling while cancer cell metabolism, Redox. Biol. 12 (2017) 833–842, http://dx.doi.org/ chitosan does not inhibit IL-12 and enhances Th1 and Th17 responses, Eur. 10.1016/j.redox.2017.04.018. J. Immunol. 42 (2012) 2709–2719, http://dx.doi.org/10.1002/eji.201242372. [144] T. Ohashi, M. Aoki, H. Tomita, T. Akazawa, K. Sato, B. Kuze, K. Mizuta, A. Hara, [124] D. Fong, P. Grégoire-Gélinas, A.P. Cheng, T. Mezheritsky, M. Lavertu, S. Sato, C. H. Nagaoka, N. Inoue, Y. Ito, M2-like macrophage polarization in high lactic D. Hoemann, Lysosomal rupture induced by structurally distinct chitosans acid-producing head and neck cancer, Cancer Sci. (2017), http://dx.doi.org/ either promotes a type 1 IFN response or activates the inflammasome in 10.1111/cas.13244. macrophages, Biomaterials 129 (2017) 127–138, http://dx.doi.org/10.1016/j. [145] F. Jouy, N. Lohmann, E. Wandel, G. Ruiz-Gómez, M.T. Pisabarro, A.G. Beck- biomaterials.2017.03.022. Sickinger, M. Schnabelrauch, S. Möller, J.C. Simon, S. Kalkhof, M. von Bergen, [125] M.P. Etna, E. Giacomini, M. Severa, E.M. Coccia, Pro- and anti-inflammatory S. Franz, Sulfated hyaluronan attenuates inflammatory signaling pathways in cytokines in tuberculosis: a two-edged sword in TB pathogenesis, Semin. macrophages involving induction of antioxidants, Proteomics 17 (2017) Immunol. 26 (2014) 543–551, http://dx.doi.org/10.1016/j.smim.2014.09.011. 1700082, http://dx.doi.org/10.1002/pmic.201700082. M.L.E. Lundahl et al. / Biochemical Pharmacology 146 (2017) 23–41 41

[146] D. Liu, C. Chang, N. Lu, X. Wang, Q. Lu, X. Ren, P. Ren, D. Zhao, L. Wang, Y. Zhu, invasive bladder cancer, Expert. Rev. Vaccin. 10 (2011) 733–742, http://dx. F. He, L. Tang, Comprehensive proteomics analysis reveals metabolic doi.org/10.1586/erv.11.20. reprogramming of tumor-associated macrophages stimulated by the tumor [165] K. Werninghaus, A. Babiak, O. Gross, C. Hölscher, H. Dietrich, E.M. Agger, J. microenvironment, J. Proteome Res. 16 (2017) 288–297, http://dx.doi.org/ Mages, A. Mocsai, H. Schoenen, K. Finger, F. Nimmerjahn, G.D. Brown, C. 10.1021/acs.jproteome.6b00604. Kirschning, A. Heit, P. Andersen, H. Wagner, J. Ruland, R. Lang, Adjuvanticity [147] Memorial Sloan Kettering Cancer Center, Beta-Glucan and Monoclonal of a synthetic cord factor analogue for subunit Mycobacterium tuberculosis Antibody 3F8 in Treating Patients With Metastatic Neuroblastoma, vaccination requires FcRgamma-Syk-Card9-dependent innate immune ClinicalTrials.gov, 2017. doi:NCT00492167. activation, J. Exp. Med. 206 (2009) 89–97, http://dx.doi.org/ [148] Mayo Clinic, Imprime PGG, Alemtuzumab, and Rituximab in Treating Patients 10.1084/jem.20081445. With High Risk Chronic Lymphocytic Leukemia, ClinicalTrials.gov, 2016. doi: [166] H. Schoenen, B. Bodendorfer, K. Hitchens, S. Manzanero, K. Werninghaus, F. NCT01269385. Nimmerjahn, E.M. Agger, S. Stenger, P. Andersen, J. Ruland, G.D. Brown, C. [149] J. Kleinnijenhuis, J. Quintin, F. Preijers, L.A.B. Joosten, D.C. Ifrim, S. Saeed, C. Wells, R. Lang, Cutting edge: mincle is essential for recognition and Jacobs, J. van Loenhout, D. de Jong, H.G. Stunnenberg, R.J. Xavier, J.W.M. van adjuvanticity of the mycobacterial cord factor and its synthetic analog der Meer, R. van Crevel, M.G. Netea, Bacille Calmette-Guerin induces NOD2- trehalose-dibehenate, J. Immunol. 184 (2010) 2756–2760, http://dx.doi.org/ dependent nonspecific protection from reinfection via epigenetic 10.4049/jimmunol.0904013. reprogramming of monocytes, Proc. Natl. Acad. Sci. USA 109 (2012) 17537– [167] A. Huber, R.S. Kallerup, K.S. Korsholm, H. Franzyk, B. Lepenies, D. Christensen, 17542, http://dx.doi.org/10.1073/pnas.1202870109. C. Foged, R. Lang, Trehalose diester glycolipids are superior to the monoesters [150] R. Pichler, G. Gruenbacher, Z. Culig, A. Brunner, D. Fuchs, J. Fritz, H. Gander, A. in binding to Mincle, activation of macrophages in vitro and adjuvant activity Rahm, M. Thurnher, Intratumoral Th2 predisposition combines with an in vivo, Innate Immun. 22 (2016) 405–418, http://dx.doi.org/10.1177/ increased Th1 functional phenotype in clinical response to intravesical BCG 1753425916651132. in bladder cancer, Cancer Immunol. Immunother. 66 (2017) 427–440, http:// [168] E.C. Patin, A.C. Geffken, S. Willcocks, C. Leschczyk, A. Haas, F. Nimmerjahn, R. dx.doi.org/10.1007/s00262-016-1945-z. Lang, T.H. Ward, U.E. Schaible, Trehalose dimycolate interferes with FccR- [151] P.F. Severino, M. Silva, M. Carrascal, N. Malagolini, M. Chiricolo, G. Venturi, A. mediated phagosome maturation through Mincle, SHP-1 and FccRIIB Astolfi, M. Catera, P.A. Videira, F. Dall’Olio, Expression of sialyl-Tn sugar signalling, PLoS One 12 (2017) e0174973, http://dx.doi.org/10.1371/journal. antigen in bladder cancer cells affects response to Bacillus Calmette pone.0174973. Guérin (BCG) and to oxidative damage, Oncotarget (2017), http:// [169] E.C. Patin, S. Willcocks, S. Orr, T.H. Ward, R. Lang, U.E. Schaible, Mincle- dx.doi.org/10.18632/oncotarget.17138. mediated anti-inflammatory IL-10 response counter-regulates IL-12 in vitro, [152] L. Rizzetto, D.C. Ifrim, S. Moretti, N. Tocci, S.-C. Cheng, J. Quintin, G. Renga, V. Innate Immun. 22 (2016) 181–185, http://dx.doi.org/10.1177/ Oikonomou, C. De Filippo, T. Weil, B.A. Blok, M.S. Lenucci, M.A.S. Santos, L. 1753425916636671. Romani, M.G. Netea, D. Cavalieri, Fungal chitin induces trained immunity in [170] T. Freier, H.S. Koh, K. Kazazian, M.S. Shoichet, Controlling cell adhesion and human monocytes during cross-talk of the host with Saccharomyces degradation of chitosan films by N-acetylation, Biomaterials 26 (2005) 5872– cerevisiae, J. Biol. Chem. 291 (2016) 7961–7972, http://dx.doi.org/10.1074/ 5878, http://dx.doi.org/10.1016/j.biomaterials.2005.02.033. jbc.M115.699645. [171] S. Franz, S. Rammelt, D. Scharnweber, J.C. Simon, Immune responses to [153] E.M. Palsson-McDermott, A.M. Curtis, G. Goel, M.A.R. Lauterbach, F.J. Sheedy, implants - a review of the implications for the design of immunomodulatory L.E. Gleeson, M.W.M. van den Bosch, S.R. Quinn, R. Domingo-Fernandez, D.G. biomaterials, Biomaterials 32 (2011) 6692–6709, http://dx.doi.org/10.1016/j. W. Johnston, J. Jiang, W.J. Israelsen, J. Keane, C. Thomas, C. Clish, M. Vander biomaterials.2011.05.078. Heiden, R.J. Xavier, L.A.J. O’Neill, R.J. Xavier, L.A.J. O’Neill, Pyruvate kinase M2 [172] D.P. Vasconcelos, A.C. Fonseca, M. Costa, I.F. Amaral, M.A. Barbosa, A.P. Águas, regulates Hif-1a activity and IL-1b induction and is a critical determinant of J.N. Barbosa, Macrophage polarization following chitosan implantation, the warburg effect in LPS-activated macrophages, Cell Metab. 21 (2015) 65– Biomaterials 34 (2013) 9952–9959, http://dx.doi.org/10.1016/j. 80, http://dx.doi.org/10.1016/j.cmet.2014.12.005. biomaterials.2013.09.012. [154] J.L. Flynn, J. Chan, K.J. Triebold, D.K. Dalton, T.A. Stewart, B.R. Bloom, An [173] K. Gulati, K.M. Poluri, Mechanistic and therapeutic overview of essential role for interferon gamma in resistance to Mycobacterium glycosaminoglycans: the unsung heroes of biomolecular signaling, tuberculosis infection, J. Exp. Med. 178 (1993) 2249–2254. Glycoconj. J. 33 (2016) 1–17, http://dx.doi.org/10.1007/s10719-015-9642-2. [155] D. Huang, L. Bao, Mycobacterium tuberculosis EspB protein suppresses [174] I.C. Severin, A. Soares, J. Hantson, M. Teixeira, D. Sachs, D. Valognes, A. Scheer, interferon-gamma-induced autophagy in murine macrophages, J. Microbiol. M.K. Schwarz, T.N.C. Wells, A.E.I. Proudfoot, J. Shaw, Glycosaminoglycan Immunol. Infect. (2014), http://dx.doi.org/10.1016/j.jmii.2014.11.008. analogs as a novel anti-inflammatory strategy, Front. Immunol. 3 (2012) 293, [156] J.N. Sondergaard, J.M. Laursen, L.B. Rosholm, S. Brix, Mycobacterium http://dx.doi.org/10.3389/fimmu.2012.00293. tuberculosis promotes Th17 expansion via regulation of human dendritic [175] Y. Feng, Q. Li, D. Wu, Y. Niu, C. Yang, L. Dong, C. Wang, A macrophage- cells toward a high CD14 and low IL-12p70 phenotype that reprograms upon activating, injectable hydrogel to sequester endogenous growth factors for exogenous IFN-gamma, Int. Immunol. 26 (2014) 705–716, http://dx.doi.org/ in situ angiogenesis, Biomaterials 134 (2017) 128–142, http://dx.doi.org/ 10.1093/intimm/dxu085. 10.1016/j.biomaterials.2017.04.042. [157] A.A. Awomoyi, A. Marchant, J.M. Howson, K.P. McAdam, J.M. Blackwell, M.J. [176] J. Iijima, K. Konno, N. Itano, Inflammatory alterations of the extracellular Newport, Interleukin-10, polymorphism in SLC11A1 (formerly NRAMP1), and matrix in the tumor microenvironment, Cancers (Basel). 3 (2011) 3189–3205, susceptibility to tuberculosis, J. Infect. Dis. 186 (2002) 1808–1814, http://dx. http://dx.doi.org/10.3390/cancers3033189. doi.org/10.1086/345920. [177] G. Zhou, M.S. Niepel, S. Saretia, T. Groth, Reducing the inflammatory [158] D.D. Barnes, M.L.E. Lundahl, E.C. Lavelle, E.M. Scanlan, The emergence of responses of biomaterials by surface modification with glycosaminoglycan phenolic glycans as virulence factors in Mycobacterium tuberculosis, ACS multilayers, J. Biomed. Mater. Res. Part A 104 (2016) 493–502, http://dx.doi. Chem. Biol. 12 (2017) 1969–1979, http://dx.doi.org/10.1021/ org/10.1002/jbm.a.35587. acschembio.7b00394. [178] B. Kang, T. Opatz, K. Landfester, F.R. Wurm, Carbohydrate nanocarriers in [159] J. Bourke, C.F. Brereton, S.V. Gordon, E.C. Lavelle, E.M. Scanlan, The synthesis biomedical applications: functionalization and construction, Chem. Soc. Rev. and biological evaluation of mycobacterial p-hydroxybenzoic acid derivatives 44 (2015) 8301–8325, http://dx.doi.org/10.1039/c5cs00092k. (p-HBADs), Org. Biomol. Chem. 12 (2014) 1114–1123, http://dx.doi.org/ [179] A.N. Samraj, H. Läubli, N. Varki, A. Varki, Involvement of a non-human sialic 10.1039/c3ob42277a. acid in human cancer, Front. Oncol. 4 (2014) 33, http://dx.doi.org/ [160] G. Stadthagen, M. Jackson, P. Charles, F. Boudou, N. Barilone, M. Huerre, P. 10.3389/fonc.2014.00033. Constant, A. Liav, I. Bottova, J. Nigou, T. Brando, G. Puzo, M. Daffe, P. Benjamin, [180] J.E. Stencel-Baerenwald, K. Reiss, D.M. Reiter, T. Stehle, T.S. Dermody, The S. Coade, R.S. Buxton, R.E. Tascon, A. Rae, B.D. Robertson, D.B. Lowrie, D.B. sweet spot: defining virus-sialic acid interactions, Nat. Rev. Microbiol. 12 Young, B. Gicquel, R. Griffin, Comparative investigation of the pathogenicity (2014) 739–749, http://dx.doi.org/10.1038/nrmicro3346. of three Mycobacterium tuberculosis mutants defective in the synthesis of p- [181] J.P. Martínez, M.L. Gil, J.L. López-Ribot, W.L. Chaffin, Serologic response to cell hydroxybenzoic acid derivatives, Microbes Infect. 8 (2006) 2245–2253, wall mannoproteins and proteins of Candida albicans, Clin. Microbiol. Rev. 11 http://dx.doi.org/10.1016/j.micinf.2006.04.008. (1998) 121–141. [161] H.R.H. Elsaidi, D.R. Barreda, C.W. Cairo, T.L. Lowary, Mycobacterial phenolic [182] L. Liu, Y. Lu, J. Martinez, Y. Bi, G. Lian, T. Wang, S. Milasta, J. Wang, M. Yang, G. glycolipids with a simplified lipid aglycone modulate cytokine levels through Liu, D.R. Green, R. Wang, Proinflammatory signal suppresses proliferation and toll-like receptor 2, ChemBioChem 14 (2013) 2153–2159, http://dx.doi.org/ shifts macrophage metabolism from Myc-dependent to HIF1a-dependent, 10.1002/cbic.201300505. Proc. Natl. Acad. Sci. 113 (2016) 1564–1569, http://dx.doi.org/10.1073/ [162] B. Carrillo-Conde, E.H. Song, A. Chavez-Santoscoy, Y. Phanse, A.E. Ramer-Tait, pnas.1518000113. N.L. Pohl, M.J. Wannemuehler, B.H. Bellaire, B. Narasimhan, Mannose- [183] S. Spence, M.K. Greene, F. Fay, E. Hams, S.P. Saunders, U. Hamid, M. Fitzgerald, functionalized ‘‘pathogen-like” polyanhydride nanoparticles target C-type J. Beck, B.K. Bains, P. Smyth, E. Themistou, D.M. Small, D. Schmid, C.M. O’Kane, lectin receptors on dendritic cells, Mol. Pharm. 8 (2011) 1877–1886, http:// D.C. Fitzgerald, S.M. Abdelghany, J.A. Johnston, P.G. Fallon, J.F. Burrows, D.F. dx.doi.org/10.1021/mp200213r. McAuley, A. Kissenpfennig, C.J. Scott, Targeting Siglecs with a sialic acid- [163] Y. Phanse, B.R. Carrillo-Conde, A.E. Ramer-Tait, R. Roychoudhury, N.L. Pohl, B. decorated nanoparticle abrogates inflammation, Sci. Transl.Med. 7 (2015), Narasimhan, M.J. Wannemuehler, B.H. Bellaire, Functionalization of http://dx.doi.org/10.1126/scitranslmed.aab3459. polyanhydride microparticles with di-mannose influences uptake by and [184] P. Espinosa-Cueto, M. Escalera-Zamudio, A. Magallanes-Puebla, L.M. López- intracellular fate within dendritic cells, Acta Biomater. 9 (2013) 8902–8909, Marín, E. Segura-Salinas, R. Mancilla, Mycobacterial glycolipids di-O-acylated http://dx.doi.org/10.1016/j.actbio.2013.06.024. trehalose and tri-O-acylated trehalose downregulate inducible nitric oxide [164] M.A. Morse, D.A. Bradley, T. Keler, R.J. Laliberte, J.A. Green, T.A. Davis, B.A. synthase and nitric oxide production in macrophages, BMC Immunol. 16 Inman, CDX-1307: a novel vaccine under study as treatment for muscle- (2015) 38, http://dx.doi.org/10.1186/s12865-015-0102-3. Reviews

pubs.acs.org/acschemicalbiology

The Emergence of Phenolic Glycans as Virulence Factors in Mycobacterium tuberculosis Danielle D. Barnes,† Mimmi L. E. Lundahl,†,‡ Ed C. Lavelle,‡ and Eoin M. Scanlan*,†

† School of Chemistry and Trinity Biomedical Sciences Institute, Trinity College, Pearse St., Dublin 2, Ireland ‡ Adjuvant Research Group, School of Biochemistry and Immunology, Trinity Biomedical Sciences Institute, Trinity College Dublin, D02 R590, Dublin 2, Ireland

ABSTRACT: Tuberculosis is the leading infectious cause of mortality worldwide. The global epidemic, caused by Mycobacterium tuberculosis, has prompted renewed interest in the development of novel vaccines for disease prevention and control. The cell envelope of M. tuberculosis is decorated with an assortment of glycan structures, including glycolipids, that are involved in disease pathogenesis. Phenolic glycolipids and the structurally related para- hydroxybenzoic acid derivatives display potent immunomodula- tory activities and have particular relevance for both understanding the interaction of the bacterium with the host immune system and also in the design of new vaccine and therapeutic candidates. Interest in glycobiology has grown exponentially over the past decade, with advancements paving the way for effective carbohydrate based vaccines. This review highlights recent advances in our understanding of phenolic glycans, including their biosynthesis and role as virulence factors in M. tuberculosis. Recent chemical synthesis approaches and biochemical analysis of synthetic glycans and their conjugates have led to fundamental insights into their roles in host−pathogen interactions. The applications of these synthetic glycans as potential vaccine candidates are discussed.

Mycobacterium tuberculosis. The global epidemic that is contributions will pave the way for innovative therapeutics for tuberculosis (TB) is the most prevalent of all potentially fatal TB in the future. bacterial infections worldwide and is caused by Mycobacterium The only human vaccine currently available for M. tb is the tuberculosis (M. tb). In 2015, M. tb infected 10.5 million bacille Calmette−Gueriń (BCG) vaccine, which is composed of people and accounted for 1.8 million deaths worldwide, making a live attenuated strain of M. bovis.11 The BCG vaccine was 1 it one of the leading causes of death from an infectious agent. introduced in 1921 and has been employed globally ever since, Multidrug resistant (MDR) and extensively drug resistant with 120 million doses administered annually.12 While the

Downloaded via TRINITY COLG DUBLIN on June 26, 2020 at 08:24:57 (UTC). − (XDR) strains of M. tb are on the rise,2 4 with an estimated 1 vaccine provides a high rate of protection against childhood 480 000 new cases of MDR infections being reported in 2015. and disseminated TB, many studies have shown that it displays Furthermore, an increase in the frequency of cases involving ffi See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. suboptimal e cacy against pulmonary TB in adult populations, fi TB and human immunode ciency virus (HIV) coinfection has with protection rates being highly variable, ranging from − made the treatment of TB more problematic, contributing to 0 to 80%.12 15 the global TB pandemic and high mortality rate associated with ffi 1−3,5 A more e cacious vaccine for TB is urgently needed and the active disease. would represent the most cost-effective way to control the Accordingly, there is an urgent need for the development disease. In particular, an effective vaccine would hinder the of new and more effective antitubercular therapeutics.6 In 2012, spread of MDR-tb and TB/HIV coinfection.16 A number of Bedaquiline became the first novel anti-TB drug approved in comprehensive reviews covering advances in TB vaccine develop- over 40 years and is the only drug available for the treatment 17−19 fi of MDR-tb.7,8 A promising anti-TB drug pipeline has recently ment have been published. In order for signi cant progress emerged, with several candidates currently in the phase II and to be made, an improved understanding of the interaction phase III clinical trials.8,9 Notably, Baulard et al. have reported between the mycobacterium and the host immune system is the discovery of novel drug-like spiroisoxazoline molecules needed. In particular, detailed structural analysis of cell wall known as Small Molecules Aborting Resistance (SMARt). components and an improved understanding of their biosynthesis Remarkably, resistance to the anti-TB drug, ethionamide, was fully reversed upon treatment with SMARt-420.10 The Received: May 11, 2017 development of new treatments for TB remains a significant Accepted: July 10, 2017 challenge for the field; however, it is anticipated that the current Published: July 10, 2017

© 2017 American Chemical Society 1969 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

Figure 1. Schematic representation of M. tb cell envelope. The plasma membrane, the cell wall core (peptidoglycan, mycolyl-arabinogalactan complex), and the outer capsule are depicted along with their comprising biomolecules (LAM, lipoarabinomannan, extractable glycolipids (PGLs, phenolic glycolipids shown) and glycoproteins). The structures and component molecules are not drawn to scale, and the numbers of carbohydrate residues shown are not representative of the actual molecules. is required. Biological evaluation of fully synthetic derivatives capsule (Figure 1). This unique multilayered cell envelope is enables an improved understanding of how these compounds highly hydrophobic with limited permeability and therefore acts modulate the immune response at a molecular level and pro- as an effective barrier against antibiotic drugs and chemo- vides fascinating prospects for therapeutic and vaccine develop- theraptic agents. This intricate cell envelope is critical for ment. In this review, we focus on the biosynthesis, chemical bacterial survival and infection and has been implicated in the synthesis, and immunomodulatory properties of mycobacterial ability of the mycobacterium to develop resistance to several cell wall phenolic glycoconjugates (PGLs and pHBADs). We antitubercular drugs.29 discuss the significant potential of these compounds in the The plasma membrane is composed of a symmetrical development of novel vaccine candidates. phospholipid bilayer. The cell wall surrounding this consists M. tb and the Host Immune System. M. tb possesses an of a lower segment made up of peptidoglycan, which is extraordinary ability to resist and evade the bactericidal covalently linked to the mycolyl-arabinogalactan (AG) layer mechanisms of the human immune system. Mycobacteria via phosphoryl-N-acetylglucosaminosylrhamnosyl.30 The AG infect host innate immune cells, prominently macrophages, and layer is composed of the oligosaccharides galactofuranose and subsequently inhibit several host antibacterial processes, arabinofuranose and is esterified by α-alkyl and β-hydroxyl allowing them to survive harsh intracellular conditions.20,21 long chain fatty acids, known as mycolic acids. Last, the upper Various ligands on the mycobacterium cell surface are recognized segment of the cell wall is made up of noncovalently attached by a series of receptors on antigen presenting cells of the glycolipids intercalated into the mycolic acid layer, known as immune system, such as Toll-like receptors (TLRs), thereby the mycomembrane.30 The outer capsule is composed of directly triggering the innate immune response.22,23 noncovalently attached polysaccharides, glycolipids, and Protective immunity against M. tb is mediated by T-cells.24 proteins, which are secreted across the cell membrane.30 By far the most important antimicrobial effectors are The surface of the cell wall is decorated with a variety orchestrated by T helper 1 (Th1) cells that secrete the of species-specific complex extractable glycolipids, such as pivotal cytokine interferon-γ (IFN-γ). This messenger, among phosphatidylinositol mannosides (PIMs), lipomannan (LM), other effects, promotes macrophage activation and subsequent lipoarabinomannan (LAM), mannose capped lipoarabinomannan bactericidal responses. Inhibition of both T cell prolifera- (Man-LAM), dimycolyl trehalose (TDM), sulfolipids (SL), tion and IFN-γ mediated effects are key components of M. tb diacyltrehaloses (DAT), polyacyltrehaloses (PAT), and phenolic pathogenesis.24,25 A number of reviews extensively detail and glycolipids (PGLs). In other mycobacteria species, extractable discuss the effects of M. tb on the host immune system, to glycolipids can include lipooligosaccharides (LOS)31,32 and − which the reader is directed.21,26 28 glycopeptidolipids (GPLs).33 The Cell Wall of M. tb. The cell envelope of M. tb is a The outermost components of the cell envelope are the first highly complex structure, composed of three major compo- to interact with a target cell and as such can exert potent nents: the plasma membrane, the cell wall core, and the outer immunomodulatory effects. Many of these glycolipids, such as

1970 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

− TDM and LAM,34 36 have been implicated as important M. tb indicates their critical importance in the pathogenesis of virulence factors and are involved in the pathogenesis of TB, the mycobacterium.46 triggering a downregulation of T-cell proliferation, inhibiting Biological Activity of PGLs and p-HBADs. PGLs are secretion of the pivotal cytokine IFN-γ, and its subsequent involved in many biological activities; for example, PGL-1, activation of macrophages.37 produced by M. leprae, is implicated in the invasion of Schwann PGLs and p-HBADs. Structure of PGLs and p-HBADs. cells and in suppressing the secretion of pro-inflammatory Certain mycobacteria, such as M. leprae, M. kansasii, M. marinum, cytokines by host immune cells.48 Additionally, the PGLs found and M. bovis, are known to produce PGLs, with several structural in M. marinum are known to play a role in cell permeability and variations depending on the species.38,39 PGL-tb, synthesized in evading the host immune system.49,50 However, the exact only by a subset of M. tb strains, was first isolated in 1987 by role of PGL-tb is less clear.23 Dafféet al. from four Canetti strains of M. tb (M. tb “canetti”), A subset of clinical isolates of M. tb, known as the W. Beijing and since then interest in the glycolipid has surged.39 family, are prevalent in southeast Asia and have spread world- Furthermore, their presentation on the cell surface of M. tb wide, causing a number of epidemics.51,52 These strains were makes these compounds ideally located to interact with host found to have a hypervirulent phenotype in mice; they failed to immune cells. induce protective Th1 responses in early infection, resulting in PGLs are made up of a lipid core common to that of PDIMs, 53,54 − a more rapid death. Interestingly, most of the W. Beijing which are known effectors of virulence in M. tb infection.40 42 family were found to produce PGLs, therefore supporting The lipid core is composed of a long chain β-diol, naturally the hypothesis that these glycolipids may be involved in the occurring as a diester of polymethyl-branched fatty acids/ hypervirulence associated with these strains of M. tb.43,55,56 phenolphthiocerol esterified by two chains of multiple methyl- A study by Reed et al. found that disruption of PGL synthesis in branched fatty acids, mycocerosic acids (Figure 2).23,43 W. Beijing strains led to a loss in their hypervirulence, without affecting bacterial load during the disease.53 In this study, an increase in the release of pro-inflammatory cytokines, interleukin 6 (IL-6), interleukin 12 (IL-12), and tumor necrosis factor α (TNF-α), in vivo was also observed upon the loss of PGL production. IL-12 plays a central role in the induction of Th1 responses. Overproduction of PGL also led to the inhibition of the pro-inflammatory mediators in a dose dependent manner, strongly suggesting that the PGLs are involved in suppressing the immediate cytokine response of the host.53 The activity of PGL is postulated to be derived from the saccharide moiety as there is no hypervirulent response associated with the structurally similar monosaccharide derivative, mycoside B.53 Work by Stadthagen et al. found that the glycans, p-HBADs, inhibit pro-inflammatory responses in infected macrophages, thereby having a significant impact on the host response to infection.23 Moreover, failure to produce p-HBADs reduced the ability of the mycobacterium to suppress the innate immune system of the host, which is typically Figure 2. Structures of the major phenolic glycolipid of M. tb 23 (PGL-tb) and structurally related glycans, para-hydroxybenzoic induced by M. tb during infection. PGL-tb is synthesized in acid derivatives I and II, p-HBAD I and p-HBAD II; red = few M. tb strains, compared to the p-HBADs, which are phenolphthiocerol; blue = mycocerosic acids. produced by all strains. Therefore, it is not surprising that the carbohydrate residue is imperative for biological activity.46 Sera The lipid core is ω-terminated by an aromatic nucleus, which from TB patients contain antibodies that specifically recognize is glycosylated with one to four sugar residues, depending on PGL-tb, despite the low incidence of PGL producing strains, the species producing it, typically O-methylated deoxysugars. therefore indicating that the seroactivity is dependent on the In the case of M. tb, the glycosylated domain of the major carbohydrate moiety (i.e., p-HBADs).46 form of PGL-tb is a trisaccharide substituent, consisting of Significant progress has been made toward elucidating the 2,3,4-tri-O-methyl-α-L-fucopyranosyl-(1→3)-α-L-rhamnopyra- immunomodulatory role of PGLs, with a large body of work 44 nosyl-(1→3)-2-O-methyl-α-L-rhamnopyranosyl. The minor providing supporting evidence that PGL is a vital virulence PGL found in M. tb is mycoside B, also found as the major factor.22,23,53,55 One such study found evidence that PGL may PGL in M. bovis, whose glycosylated domain consists of enhance the infectivity of M. tb at the earliest stage of infection, 2-O-methyl-α-L-rhamnopyranosyl, with the lipid core identical through CC chemokine receptor 2 (CCR2)-mediated recruit- to that of PGL-tb.45 ment of permissive macrophages.22 It has also been suggested The small carbohydrate molecules, para-hydroxybenzoic acid that the lack of PGL production in BCG M. bovis strains derivatives (p-HBADs), are structurally related to PGLs and are may account for the suboptimal protection provided by depicted in Figure 2 as p-HBAD I and p-HBAD II. They are the vaccine.57 However, the exact correlation between the secreted around the cell wall of M. tb and are found in culture glycolipids and related virulence remains unclear.58 isolates of all strains of M. tb and M. bovis BCG.46 The aromatic Biogenesis of PGLs and p-HBADs. Understanding the core is derived from the methyl ester of p-hydroxybenzoic acid biosynthesis of PGLs and related phenolic glycans is critical to (p-HBA), and the glycosyl moieties in both p-HBAD I and the development of an understanding of how these molecules p-HBAD II are identical to that of mycoside B and PGL-tb, modulate immune response. In addition, potential protein respectively.47 The production of these glycans by all strains of targets for small-molecule inhibitors may be identified through

1971 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews careful elucidation of the genes involved in cell wall biosynthesis. Biosynthesis of the Saccharide Residue. The synthesis The biosynthesis of PGLs in M. tb is a complex 25 step of the saccharide residue of PGL and the p-HBADs involves the enzymatic process, with 1.5% of the M. tb genome dedicated to same set of enzymes, starting from either the p-hydroxyphenyl the synthesis and transport of these molecules, providing PDIM or methyl p-hydroxybenzoate precursors, for PGL-tb additional evidence for their essential role in the mycobacte- and p-HBADs, respectively (Figure 3B). Due to the structural rium.30,59 Consistent with their conserved structures, the similarity between the two, it was initially thought that PGL-tb and p-HBADs share a similar biosynthetic pathway glycosylation was an early step in the biosynthesis of PGLs and common enzymatic steps.60 Sequencing of the M. tb and that p-HBADs were simply intermediates in the biosynthetic genome has greatly aided the elucidation of the biosynthetic pathway.46 However, as the p-HBADs are found outside the cell steps involved in the production of these molecules.61 The wall and not in the cytosol, where the enzymatic biomachinery genes involved in the biosynthesis are clustered on a 73-kb for the PGL synthesis is located, it is unlikely that these glycans fragment of the M. tb chromosome with the organization of the are intermediates of PGL biosynthesis. This is further confirmed locus being highly conserved in all PGL producing strains.30 by the production of p-HBADs in all species of M. tb, including Biosynthesis of p-Hydroxyphenyl PDIM. The common those unable to produce PGLs.46 Moreover, intracellular lipid core of PGLs and PDIMs, phenolphthiocerol and attachment of the sugar residue to the lipid occurs before trans- phthiocerol, respectively, involve similar biosynthetic steps, location of the PGLs to the cell wall. Inactivation of the related the genes for which are located on the PDIM/PGL locus of transporter proteins, which are involved in transporting the the M. tb chromosome.41 molecules across the cell wall, results in blocked secretion of the The aromatic core in phenylphthiocerol is derived from PDIMs and PGLs, without affecting their biosynthesis.42 chorismate, which is converted to p-HBA by a pyruvate-lygase The first rhamnose residue is transferred onto p-hydroxyphenyl enzyme encoded by Rv2949c (Figure 3A).47 p-HBA is activated PDIM by the glycosyltransferase encoded by Rv2962c,60 followed by FadD22, a fatty acid ligase involved in the activation and by the methylation of the 2-OH, catalyzed by the product transfer of long chain fatty acids, to yield p-HBA-S-FadD22, an of Rv2959c.59 A second rhamnose residue is subsequently acyl-S-enzyme covalent intermediate.62 This is subsequently transferred by the rhamnosyltransferase enzyme encoded by elongated by a type 1 polyketide synthase to give malonyl-CoA Rv2958c.60 A frameshift mutation in the Rv2958c gene is units attached to p-hydroxyphenylalkenoate.46 Type 1 polyke- responsible for the formation of the truncated minor PGL, tide synthase is encoded for by the pks 15/1 gene, and it was mycoside B, as well as the accumulation of the monosaccharide found that disruption of pks 15/1 in PGL producing strains p-HBAD I in M. tb.44 Rv2957 encodes for the glycosytransferase, eradicated their ability to produce PGLs.46 Most strains of which is responsible for the transfer of the terminal fucose M. tb naturally contain a pks 15/1 frameshift mutation, which residue to form the trisaccharide of PGL-tb or p-HBAD II. The explains why these strains only produce p-HBADs and not exact order in which the trisaccharide is formed is yet to be fully PGLs. Interestingly, when pks 15/1 was introduced to a non- elucidated. Possibly, the disaccharide is formed first, followed PGL producing strain, the ability to produce PGLs was fully by its transfer to the p-HBAD I and 2-O-methyl-rhamnosyl- restored, highlighting the fact that this gene is vital for the phenyl PDIM to form p-HBAD II and PGL-tb, respectively. formation of the lipid precursor for PGL, through the elongation Alternatively, the two enzymes may sequentially catalyze the of the putative p-HBA precursor.46 Moreover, Tsenova et al. transfer of the rhamnosyl and fucosyl appendages, controlled by found that disruption of pks 15/1 in PGL producing strains of a regulation loop, to prevent the accumulation of biosynthetic M. tb resulted in a reduced virulence in the rabbit model of intermediates containing the disaccharide PGL or p-HBAD, TB.55 which have never been isolated.60 The final steps in this bio- The enzyme, FadD29, activates p-hydroxyphenylalkenoate, synthetic pathway involve the methylation of the terminal which is transferred onto the polyketide synthase, ppsA, and fucose. This occurs after glycosylation, orchestrated by three elongated with both malonyl-CoA and methyl malonyl-CoA methyltransferases, encoded by Rv2956, Rv2954c,andRv2955c units by ppsA-E to form phenolphthiocerol.46,49 Finally, the for the methylation of the 2-OH, 3-OH, and 4-OH of fucose, phenolphthiodiolone residue is released from ppsE by type II respectively. Evidence has suggested that the methylation occurs thioesterase, TesA.63 The enzymes responsible for the unusual sequentially, beginning with the 2-OH position, followed by the methylation of the third hydroxy group of the PDIM are 4-OH and finally the 3-OH residue.73 encoded for by two genes, Rv2951c and Rv2952. The reduction Translocation of PGL to the Cell Wall. Both genetic and of phenolphthiodiolone to phenolphthiotriol is catalyzed by the biochemical strategies have shown that the assembly of PGLs product of Rv2951c. The subsequent methylation of phenolph- and p-HBADs occurs in the cytosol. Furthermore, some of the thiotriol to form phenolphthiocerol is catalyzed by the product enzymes involved in the biosynthesis are known to be located of Rv2952.43,59 Phenolphthiocerol is subsequently esterified by intracellularly (e.g., FadD enzymes); however, their presence at mycocerosic acids to yield p-hydroxyphenyl PDIM. the outermost layers of the cell wall and capsule indicate the 74 Mycocerosic acids are derived from C16−20 chain fatty acids, existence of dedicated PDIM/PGL translocation machinery. which are elongated by successive reactions to incorporate Most work on the translocation of these molecules has to propionate units by the multifunctional enzyme, mycocerosic date focused on PDIM; however, it is possible that the same − acid synthase, encoded by mas.64 68 The use of methyl transporters are involved in the transport of PGLs across the malonyl-CoA as a substrate by this enzyme introduces methyl cell membrane. branches during the fatty acid elongation to form tetra-methyl The mmpL7 gene encodes for a transporter of the resistance- branched mycocerosic acids.64,69 The fatty-acyl AMP ligase, nodulation-division (RND) permease family and is located at FadD28, is involved in activating the fatty acid chain precursors the PGL/PDIM locus on the M. tb chromosome; it is implicated to form mycocerosates,70 which are transferred directly onto in the formation of the outer membrane of M. tb.74,75 Three the diol of the phenolphthiocerol by the acyl transferase, PapA5, genes, drrA, drrB,anddrrC, encode for an ABC transporter in the final esterification step.71,72 found in the cell membrane, with the subunit encoded for by

1972 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

Figure 3. Biosynthetic pathway of M. tb PGLs and p-HBADs. (A) Biosynthetic pathway and intermediates in the synthesis of PDIM. (B) Biosynthetic pathway and intermediates in the synthesis of PGLs and p-HBADs; m =12−16; n =18−20; R = H, Me; orange box = enzymes; blue box = genes; purple box = acyl carrier proteins; PDIM, phthiocerol dimycoserosate; PGL-tb = phenolic glycolipid of M. tb; p-HBAD I, para-hydroxybenzoic acid derivative I; p-HBAD II, para-hydroxybenzoic acid derivative II. drrC being required for the proper localization of PDIMs.74 that the synthesis and transport of these molecules may be Inactivation of these transporters blocks the transport of related processes.76 Furthermore, the mutants of these genes, both PDIM and PGL across the cell membrane, while the which were unable to properly transport the PDIMs, showed biosynthesis of the glycolipids remains unaffected.42,74 Interest- increased cell wall permeability, indicating that these molecules ingly, the A domain of the mmpL7 transporter was found may play an integral role in the architecture and permeability to interact biochemically with ppsE, the polyketide synthase of the cell wall, in addition to being important virulence involved in the synthesis of PDIM and PGL, suggesting factors.63,74

1973 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

Finally, the lipoprotein encoded by LppX is required for the Lowary et al. and Astarie-Dequeker and co-workers reported transport of PDIMs from the periplasm to the outer layers of the synthesis and biological evaluation of the trisaccharide moiety the cell wall. The crystal structure of this protein revealed of PGL. a large hydrophobic cavity suitable for accommodating a PDIM Lowary and co-workers prepared a methoxy derivative of the molecule. Lipid analysis revealed that mutants of LppX fails to p-HBAD II trisaccharide using a convergent synthesis, starting 77 release PDIMs, without affecting their biosynthesis. from three monosaccharide building blocks (1−3, Figure 4). Chemical Synthesis and Biological Studies of PGLs and p-HBADs. As outlined above, PGLs and p-HBADs are important virulence factors in M. tb infection as they modulate the host immune response. Due to the complex pattern and variable ratios of these molecules in vivo, isolation of appreciable amounts of the pure compounds has proven to be extremely challenging. Therefore, chemical synthesis is necessary to provide sufficient quantities of pure material in order to fully elucidate the immune response in isolation from the parent bacterium and to develop potential vaccine candidates. However, these complex molecules represent challenging synthetic targets, encompassing lengthy synthetic routes and intricate protecting group strategies. Moreover, a major obstacle to be considered in the chemical synthesis of these compounds is the selective formation of the α- linkages, which are found in the natural products.78 Since the elucidation of the molecular structures of the PGL family, several elegant approaches have been reported regarding their total chemical synthesis. As the biological specificity of the PGLs is thought to be related to the carbohydrate residue, much of this work has focused on obtaining synthetic analogues of the saccharide domain. The synthesis of the trisaccharide epitope of PGL-1 found in M. leprae was first described by Brennan et al. in 1984.79 Since then, other derivatives of the PGL-1 carbohydrate have since been published by Brennan 80,81 82 et al. and Pinto et al. Synthesis of the glycan segment Figure 4. Synthesis of methoxy derivative of the p-HBAD II of the PGL found in M. kanasii was initially published by trisaccharide reported by Lowary and co-workers.92 (a) NIS, AgOTf, 83 − ° · Reddy et al. in 1992. More recently, Lowary and co-workers CH2Cl2, 20 C, 30 min, 87%; (b) NH2NH2 HOAc, CH2Cl2,4h, − ° described the synthesis of an epitope of this PGL saccharide 77%; (c) 3, NIS, AgOTf, CH2Cl2, 40 C, 30 min; (d) NaOCH3, 84,85 accompanied by extensive immunological studies. CH3OH/CH2Cl2 (1:1), 5 h, 70% over 2 steps; (e) CH3I, NaH, DMF, The synthesis of the trisaccharide unit of the PGL found in 1 h; (f) Pd/C, H2,CH3OH/CH2Cl2 (1:1), 72 h, 71% over 2 steps. M. tb was first reported in 1990 by Van Boom, who introduced the α-glycosyl linkages using an iodonium ion-promoted glyco- These building blocks were sequentially added, starting from sylation approach in order to achieve the desired stereo- the reducing end to the nonreducing end of the molecule to selectivity.86 This was subsequently followed by the synthesis of assemble the trisaccharide. The rhamnose building blocks, 1 a PGL-tb glycan epitope conjugated to Bovine Serum Albumin and 2, were both prepared from per-acetylated rhamnose in (BSA) by Fujiwara in 1991, who demonstrated that the glycan eight synthetic steps. The trimethylated thiofucoside starting 87 could be used as a potential tool for serodiagnosis of M. tb. material, 3, was synthesized from L-fucose via a four-step In the same year, Viswanadam et al. also published the synthesis synthetic approach. of trisaccharide segment of PGL-tb.88 The initial glycosylation of 1 and thioglycoside 2 was Impressively, the total chemical synthesis of PGL-tb was achieved using NIS-AgOTf-promoted conditions to furnish the reported by Minnaard et al. in 2012 and represents a landmark α-linked disaccharide. Typically, in the synthesis of p-HBAD in glycoconjugate synthesis, marking the first fully synthetic molecules, the methyl group of the 2-OH of 1 is introduced PGL compound.89 The synthetic approach utilized a Sonogashira prior to glycosylation with the 2 residue.91,93 However, in this coupling to conjugate the phenolic trisaccharide onto the case, it was possible to introduce the methyl functionality after phthiocerol. Asymmetric Cu-catalyzed 1,4-additions to unsatu- the formation of the trisaccharide. The use of a benzoyl (Bz) rated thioesters and cyclic enones were employed to introduce protecting group provided the α-linked disaccharide exclusively, the methyl groups. Minnaard et al. subsequently published the via neighboring group participation. total chemical synthesis of mycoside B the following year.90 In The levulinoyl (Lev) protecting group was selectively 2013, the same group notably reported the synthesis of p-HBAD removed using hydrazine acetate to furnish acceptor 5. This 90 I, starting from L-rhamnose, in only five synthetic steps. In was subsequently coupled to thiofucoside 3 by an NIS-AgOTf- 2014, Scanlan et al. published the chemical synthesis of both promoted glycosylation, using an “inverse glycosylation” p-HBAD I and p-HBAD II, accompanied by biological studies procedure, to provide the α-linked trisaccharide exclusively. on the immunological effects of the molecules. This was the Schmidt and co-workers introduced the concept of “inverse first time that the p-HBADs were studied in isolation from the glycosylation” whereby the acceptor and activator are maintained parent bacterium; remarkably, they were found to down-regulate at a higher concentration than that of the donor, thus providing cytokine production by host immune cells.91 More recently, both higher yielding reactions with greater selectivity.94 The benzoyl

1974 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews group was subsequently removed using Zempleń deacetylation The stereoselectivity of the glycosylation was achieved by the conditions, and the resulting 2-OH was methylated on treatment neighboring group participation of the adjacent acetyl (Ac) with sodium hydride and methyl iodide. The final compound, 7, protecting group, to furnish the α-linked disaccharide, 11, was obtained following global deprotection of 6 via hydro- exclusively. The 3′-O-Bn protecting group of the nonreducing genolytic cleavage of the benzyl (Bn) ethers. rhamnose unit was removed using hydrogenation conditions to This compound was subsequently tested in THP-1 (human give alcohol 12, and the terminal fucose residue was introduced acute leukemic monocyte/macrophage) cells. They were found upon NIS-TMSOTf-promoted glycosylation with 10 to yield to inhibit the release of nitric oxide (NO) and the cytokines, the trisaccharide 13. Finally, deprotection of the Ac protecting TNF-α, interleukin 1β (IL-1β), and IL-6, in a concentration group using Zempleń conditions, followed by the removal of the dependent manner, supporting previous biological studies on these tert-butyldimethylsilyl (TBS) protecting groups using NBu4F, compounds.91 In this study, further testing with other analogues of furnished the final compound 14. p-HBAD II showed that the correct methylation pattern on these This synthetic epitope of PGL-tb was subsequently utilized molecules is essential. Derivatives lacking the methoxy group at to investigate the molecular mechanisms by which the PGLs therhamnose2-OHandthefucose3-OH″ position displayed act on the host immune system. From this study, it was shown a substantial loss in the inhibitory activity, compared to that of that the trisaccharide moiety of the PGL-tb acts as a pathogen- the naturally methylated analogue, 7.Furthermore,Lowaryet al. associated molecular pattern (PAMP) recognized by the host demonstrated that the inhibitory activity of these synthetic PGL immune system. This enables the mycobacterium to exploit analogues is mediated through Toll-like receptor 2 (TLR2), and host pattern recognition receptors (PRRs) and consequently not Toll-like receptor 4 (TLR4).92 modulate the immune response. The trisaccharide epitope is Astarie-Dequeker and co-workers have recently published the directly responsible for the specificity of binding to TLR2, and synthesis of a methyl analogue of the p-HBAD II trisaccharide the common lipid core of the PGL family may be involved in using a convergent synthetic route. This was achieved using enhancing the affinity of the receptor for the glycan antigen. three monosaccharide starting materials, 8−10, depicted in Binding of the glycan to TLR2 causes the inhibition of TLR2- Figure 5. The rhamnose starting materials 8 and 9 were both dependent signaling cascades and the secretion of inflammatory cytokines (TNF-γ), resulting in an overall dampening of the host immune response.93 This study also demonstrated that the expression of PGLs on the surface of the mycobacterium enhances infectivity and furthermore provided important insight into the mechanism of action of PGLs, which may aid the future development of improved therapeutics. The Development of TB Vaccines Based on Cell Wall Carbohydrates. There is significant precedent for the application of carbohydrates in vaccine development, especially those found on the cell surface of pathogens.13,95,96 Carbohydrates can play key roles in the initial stages of many pathogenic infections and are therefore often located on the surface of the cell, making them ideal targets for vaccine design. However, saccharides alone are generally T-cell independent antigens, rendering them poorly immunogenic, facilitating bacterial immune evasion by a mechanism known as “glycan shielding.”95 The application of chemical synthesis toward the preparation of PGLs and related compounds has provided significant benefit to the field of mycobacterial glycobiology. The chemical synthesis approach enables the preparation of homogeneous compounds suitable for immunological evalua- tion, protein conjugation, labeling, and further modification. The biological evaluation of these compounds in isolation from the host bacterium has provided fundamental new insights into Figure 5. Synthetic route used by Astarie-Dequeker et al. for the the roles of these compounds in modulating immune response synthesis of the methyl derivative of the p-HBAD II.93 (a) TMSOTf, and offers a template for the development of novel vaccine − ° CH2Cl2, 20 to 0 C, 2 h, 89%; (b) Pd/C, H2,CH3OH/CH2Cl2 candidates. One well established technique used to overcome − ° (1:1), rt, 72 h, 95%; (c) 10, NIS, TMSOTf, CH2Cl2, 20 to 0 C, 2 h; the low immunogenicity of carbohydrate antigens is to con- (d) NaOCH3,CH3OH, rt, 5 h; (e) NBu4F, THF, rt, 5 h, 69% over jugate the glycan to a carrier protein; this can elicit a T-cell 2 steps. immune response, aiding B cells, allowing for class switching and conferring long lasting protection against the pathogen. synthesized from the orthoester of L-rhamnose via nine and five One of the earliest conjugate vaccines developed was for step synthetic routes, respectively. In a similar approach to the Haemophilus influenza (Hib).97 Prevnar is a carbohydrate− synthetic route used by Lowary, the fucose building block, 10, protein conjugate vaccine against Streptococcus pneumoniae.98 was a trimethylated thiofucoside. However, in contrast, Astarie- Other carbohydrate based vaccines include the meningococcal Dequeker introduced the pivotal methyl group at the 2-OH of C conjugate vaccine.99 Another method that can increase 8 prior to the initial glycosylation step.93 the immunogenicity of carbohydrates is the “glycoside cluster The rhamnosyl acceptor, 8, was coupled to the trichloroace- effect;” this involves the multivalent presentation of the timidate donor, 9, by NIS-TMSOTf -promoted glycosylation. carbohydrates and has been employed in the virosomal vaccine

1975 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews for Hepatitis A, Epaxal.96,100 Various methods have been ■ ABBREVIATIONS utilized to achieve this multivalent glycan display, such as gold Ac, acetyl; AG, arabinogalactan; BCG, Bacille−Calmette− nanoparticles and liposomes, allowing for substantial progress to Guerin;́ Bn, benzyl; Bz, benzoyl; CCR2, CC chemokine be made toward the development of a more effective carbo- 101,102 receptor 2; DAT, diacyltrehaloses; GPL, glycopeptidolipids; hydrate based vaccine for Streptococcus pneumoniae. HIV, human immunodeficiency virus; IL-1β, interleukin 1β; IL- The success of these carbohydrate based vaccines has 6, interleukin 6; IL-12, interleukin 12; IFN-γ, interferon-γ; generated considerable interest in utilizing the immunomodu- LAM, lipoarabinomannan; Lev, levulinoyl; LM, lipomannan; latory glycolipids at the cell surface of M. tb as potential vaccine 13 LOS, lipooligosaccharide; Man-LAM, mannose capped lip- candidates. Recently, a study by Sun et al. highlighted the oarabinomannan; MDR, multidrug resistant; M. tb, Mycobacte- potential of blocking Man-LAM by an antagonistic/neutralizing rium tuberculosis; NO, nitric oxide; PAMP, pathogen-associated aptamer to increase the protective effect of the BCG vaccine 103 molecular pattern; PAT, polyacyltrehaloses; PDIM, phthiocerol against M. tb. Furthermore, both PIMs and LAM have dimycoserosate; p-HBA, para-hydroxybenzoic acid; p-HBADs, recently been shown to be promising candidates for a potential 104,105 para-hydroxybenzoic acid derivatives; p-HBAD I, para-hydrox- TB vaccine. Recently, Zhu and co-workers have ybenzoic acid derivative I; p-HBAD II, para-hydroxybenzoic synthesized a PGL-tb epitope conjugated to the protein carrier, 106 acid derivative II; PGL, phenolic glycolipid; PIMs, phosphati- CRM197. Interestingly, this monovalent glycoconjugate has fi dylinositol mannosides; PRR, host pattern recognition been found to induce high antigen-speci c IgG levels in the receptor; SL, sulfolipids; SMARt, Small Molecules Aborting serum of mice, rabbits, and guinea pigs and has highlighted the Resistance; TB, tuberculosis; TBS, tert-butyldimethylsilyl; utility of the PGL-tb glycan as a valuable tool for the genera- TDM, dimycolyl trehalose; TLR, Toll like receptor; TLR2, tion of PGL specific monoclonal antibodies. Furthermore, the α ff Toll like receptor 2; TLR4, Toll like receptor 4; TNF- , tumor immunoprotective e ect associated with this PGL-tb epitope necrosis factor α; RND, resistance-nodulation-division; XDR, has reinforced the potential of a synthetic PGL based vaccine extensively drug resistant for the treatment of TB.106 ■ KEYWORDS ■ CONCLUSION Tuberculosis: an infectious bacterial disease caused by PGLs and the related glycans, p-HBADs, play a key role in Mycobacterium tuberculosis (M. tb) M. tb pathogenesis. They are located on the outermost layers of Glycolipid: a lipid connected to one or more sugar residues the cell wall, where they act as important virulence factors, linked by a interacting with the host immune system. A combination of Vaccine: an antigenic substance prepared from the causative genetic and biochemical studies has enabled the elucidation agent of a disease or a synthetic substitute, used to provide of the key steps in the biogenesis of these molecules. These protective immunity against one or several diseases complex molecules represent challenging synthetic targets for Glycosylation: the addition of a glycan onto a protein, lipid, chemists, with a number of impressive total chemical syntheses or other organic molecule of PGL and p-HBAD analogues reported. Chemical synthesis Hypervirulent: a substance that is capable of causing severe fi has provided, for the rst time, access to pure samples of these illness important biomolecules and has been crucial in probing their Pathogenesis: the cellular events and reactions and other biological and immunomodulatory activities. Recent studies pathologic mechanisms occurring in the development of have focused on the application of synthetic cell wall glycolipids disease as potential vaccine candidates for M. tb, and it is anticipated Cytokine: a cell signaling molecule that aids cell to cell that the development of a novel carbohydrate conjugate vaccine communication during immune response and stimulates the candidate for M. tb will continue to remain an active area of movement of cells toward sites of inflammation and infection interest into the future. Ultimately, access to pure samples of L-Rhamnose: a naturally occurring 6- these glycolipids and their conjugates will be essential for the realization of PGL/p-HBAD based vaccines. ■ REFERENCES ■ AUTHOR INFORMATION (1) (2016) Global Tuberculosis Report 2016,WorldHealth Organisation. Corresponding Author (2) Gandhi, N. R., Nunn, P., Dheda, K., Schaaf, H. S., Zignol, M., van *E-mail: [email protected]. Soolingen, D., Jensen, P., and Bayona, J. (2010) Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of ORCID tuberculosis. Lancet 375, 1830−1843. Eoin M. Scanlan: 0000-0001-5176-2310 (3) Shenoi, S., and Friedland, G. (2009) Extensively drug-resistant − Notes tuberculosis: a new face to an old pathogen. Annu. Rev. Med. 60, 307 fi 320. The authors declare no competing nancial interest. (4) Nachega, J. B., and Chaisson, R. E. (2003) Tuberculosis Drug Resistance: A Global Threat. Clin. Infect. Dis. 36, S24−S30. ■ ACKNOWLEDGMENTS (5) Egelund, E. F., Dupree, L., Huesgen, E., and Peloquin, C. A. (2017) The pharmacological challenges of treating tuberculosis and The authors would like to acknowledge Trinity College Dublin − for postgraduate studentships (to D.D.B. and M.L.E.L.). E.M.S. HIV coinfections. Expert Rev. Clin. Pharmacol. 10, 213 223. (6) Koul, A., Arnoult, E., Lounis, N., Guillemont, J., and Andries, K. is funded by Science Foundation Ireland (SFI) under grant (2011) The challenge of new drug discovery for tuberculosis. Nature number 15/CDA/3310. E.L. is funded by Science Foundation 469, 483−490. Ireland (SFI) under grant number 12/IA/1421 and the SFI (7) Andries, K., Verhasselt, P., Guillemont, J., Göhlmann, H. W. H., Research Centre, Advanced Materials and BioEngineering Neefs, J.-M., Winkler, H., Van Gestel, J., Timmerman, P., Zhu, M., Lee, Research (AMBER) under grant number SFI/12/RC/2278. E., Williams, P., de Chaffoy, D., Huitric, E., Hoffner, S., Cambau, E.,

1976 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

Truffot-Pernot, C., Lounis, N., and Jarlier, V. (2005) A Diarylquinoline synthesis of p-hydroxybenzoic acid derivatives. Microbes Infect. 8, Drug Active on the ATP Synthase of Mycobacterium tuberculosis. 2245−2253. Science 307, 223. (24) Cooper, A. M., Dalton, D. K., Stewart, T. A., Griffin, J. P., (8) Deoghare, S. (2013) Bedaquiline: A new drug approved for Russell, D. G., and Orme, I. M. (1993) Disseminated tuberculosis in treatment of multidrug-resistant tuberculosis. Indian J. Pharmacol. 45, interferon gamma gene-disrupted mice. J. Exp. Med. 178, 2243−2247. 536−537. (25) Flynn, J. L., Chan, J., Triebold, K. J., Dalton, D. K., Stewart, T. (9) Zumla, A., Nahid, P., and Cole, S. T. (2013) Advances in the A., and Bloom, B. R. (1993) An essential role for interferon gamma in development of new tuberculosis drugs and treatment regimens. Nat. resistance to Mycobacterium tuberculosis infection. J. Exp. Med. 178, Rev. Drug Discovery 12, 388−404. 2249. (10) Blondiaux, N., Moune, M., Desroses, M., Frita, R., Flipo, M., (26) Flynn, J. L., and Chan, J. (2001) Immunology of Tuberculosis. Mathys, V., Soetaert, K., Kiass, M., Delorme, V., Djaout, K., Trebosc, Annu. Rev. Immunol. 19,93−129. V., Kemmer, C., Wintjens, R., Wohlkönig, A., Antoine, R., Huot, L., (27) Cooper, A. M. (2009) Cell-mediated immune responses in Hot, D., Coscolla, M., Feldmann, J., Gagneux, S., Locht, C., Brodin, P., tuberculosis. Annu. Rev. Immunol. 27, 393−422. Gitzinger, M., Deprez,́ B., Willand, N., and Baulard, A. R. (2017) (28) O’Garra, A., Redford, P. S., McNab, F. W., Bloom, C. I., Reversion of antibiotic resistance in Mycobacterium tuberculosis by Wilkinson, R. J., and Berry, M. P. (2013) The immune response in spiroisoxazoline SMARt-420. Science 355, 1206. tuberculosis. Annu. Rev. Immunol. 31, 475−527. (11) Kaushal, D., Foreman, T. W., Gautam, U. S., Alvarez, X., (29) Jarlier, V., and Nikaido, H. (1994) Mycobacterial cell wall: Adekambi, T., Rangel-Moreno, J., Golden, N. A., Johnson, A. M., Structure and role in natural resistance to antibiotics. FEMS Microbiol. Phillips, B. L., Ahsan, M. H., Russell-Lodrigue, K. E., Doyle, L. A., Roy, Lett. 123,11−18. C. J., Didier, P. J., Blanchard, J. L., Rengarajan, J., Lackner, A. A., (30) Daffe,́ M., Reyrat, J.-M., and Avenir, G. (2008) The Khader, S. A., and Mehra, S. (2015) Mucosal vaccination with Mycobacterial Cell Envelope 1st ed., ASM Press. attenuated Mycobacterium tuberculosis induces strong central (31) Jackson, M. (2014) The Mycobacterial Cell EnvelopeLipids. memory responses and protects against tuberculosis. Nat. Commun. Cold Spring Harbor Perspect. Med. 4, a021105. 6, 8533. (32) Bai, B., Chu, C.-j., and Lowary, T. L. (2015) Lipooligosacchar- (12) Ritz, N., Hanekom, W. A., Robins-Browne, R., Britton, W. J., ides from Mycobacteria: Structure, Function, and Synthesis. Isr. J. and Curtis, N. (2008) Influence of BCG vaccine strain on the immune Chem. 55, 360−372. response and protection against tuberculosis. FEMS Microbiol. Rev. 32, (33) Schorey, J. S., and Sweet, L. (2008) The mycobacterial 821−841. glycopeptidolipids: structure, function, and their role in pathogenesis. (13) Moliva, J. I., Turner, J., and Torrelles, J. B. (2015) Prospects in Glycobiology 18, 832−841. Mycobacterium bovis Bacille Calmette et Guerin (BCG) vaccine (34) Rao, V., Fujiwara, N., Porcelli, S. A., and Glickman, M. S. (2005) diversity and delivery: why does BCG fail to protect against Mycobacterium tuberculosis controls host innate immune activation tuberculosis? Vaccine 33, 5035−5041. through cyclopropane modification of a glycolipid effector molecule. J. (14) Ponnighaus, J. M., Msosa, E., Gruer, P. J. K., Liomba, N. G., Exp. Med. 201, 535. Fine, P. E. M., Sterne, J. A. C., Wilson, R. J., Bliss, L., Jenkins, P. A., and (35) Axelrod, S., Oschkinat, H., Enders, J., Schlegel, B., Brinkmann, Lucas, S. B. (1992) Efficacy of BCG vaccine against leprosy and V., Kaufmann, S. H., Haas, A., and Schaible, U. E. (2008) Delay of tuberculosis in northern Malawi. Lancet 339, 636−639. phagosome maturation by a mycobacterial lipid is reversed by nitric (15) Aronson, N. E., Comstock, G. W., Howard, R. S., Moulton, L. oxide. Cell. Microbiol. 10, 1530−1545. H., Rhoades, E. R., Harrison, L. H., and Santosham, M. (2004) Long- (36) Rajni, Rao, N., and Meena, L. S. (2011) Biosynthesis and term Efficacy of BCG Vaccine in American Indians and Alaska Natives. Virulent Behavior of Lipids Produced by Mycobacterium tuberculosis: J. Am. Med. Assoc 291, 2086. LAM and Cord Factor: An Overview. Biotechnol. Res. Int. 2011, (16) Parida, S. K., Axelsson-Robertson, R., Rao, M. V., Singh, N., 274693. Master, I., Lutckii, A., Keshavjee, S., Andersson, J., Zumla, A., and (37) Mahon, R. N., Rojas, R. E., Fulton, S. A., Franko, J. L., Harding, Maeurer, M. (2015) Totally drug-resistant tuberculosis and adjunct C. V., and Boom, W. H. (2009) Mycobacterium tuberculosis cell wall therapies. J. Intern. Med. 277, 388−405. glycolipids directly inhibit CD4+ T-cell activation by interfering with (17) Orme, I. M. (2013) Vaccine development for tuberculosis: proximal T-cell-receptor signaling. Infect. Immun. 77, 4574−4583. current progress. Drugs 73, 1015−1024. (38) Blaauw, G. J., and Appelmelk, B. J. (2006) Protein-Carbohydrate (18) Tang, J., Yam, W. C., and Chen, Z. (2016) Mycobacterium Interactions in Infectious Diseases, Royal Society of Chemistry. tuberculosis infection and vaccine development. Tuberculosis 98,30− (39) Daffe,́ M., Lacave, C., Laneelle,́ M.-A., and Laneelle,́ G. (1987) 41. Structure of the major triglycosyl phenol-phthiocerol of Mycobacte- (19) Young, D., and Dye, C. (2006) The development and impact of rium tuberculosis (strain Canetti). Eur. J. Biochem. 167, 155−160. tuberculosis vaccines. Cell 124, 683−687. (40) Rousseau, C., Winter, N., Pivert, E., Bordat, Y., Neyrolles, O., (20) Richardson, E. T., Shukla, S., Sweet, D. R., Wearsch, P. A., Ave,́ P., Huerre, M., Gicquel, B., and Jackson, M. (2004) Production of Tsichlis, P. N., Boom, W. H., and Harding, C. V. (2015) Toll-Like phthiocerol dimycocerosates protects Mycobacterium tuberculosis Receptor 2-Dependent Extracellular Signal-Regulated Kinase Signaling from the cidal activity of reactive nitrogen intermediates produced in Mycobacterium tuberculosis-Infected Macrophages Drives Anti- by macrophages and modulates the early immune response to Inflammatory Responses and Inhibits Th1 Polarization of Responding infection. Cell. Microbiol. 6, 277−287. T Cells. Infect. Immun. 83, 2242−2254. (41) Camacho, L. R., Ensergueix, D., Perez, E., Gicquel, B., and (21) Kaufmann, S. H. E. (2001) How can immunology contribute to Guilhot, C. (1999) Identification of a virulence gene cluster of the control of tuberculosis? Nat. Rev. Immunol. 1,20−30. Mycobacterium tuberculosis by signature-tagged transposon muta- (22) Cambier, C. J., Takaki, K. K., Larson, R. P., Hernandez, R. E., genesis. Mol. Microbiol. 34, 257−267. Tobin, D. M., Urdahl, K. B., Cosma, C. L., and Ramakrishnan, L. (42) Cox, J. S., Chen, B., McNeil, M., and Jacobs, W. R. (1999) (2013) Mycobacteria manipulate macrophage recruitment through Complex lipid determines tissue-specific replication of Mycobacterium coordinated use of membrane lipids. Nature 505, 218−222. tuberculosis in mice. Nature 402,79−83. (23) Stadthagen, G., Jackson, M., Charles, P., Boudou, F., Barilone, (43) Huet, G., Constant, P., Malaga, W., Laneelle, M. A., Kremer, K., N., Huerre, M., Constant, P., Liav, A., Bottova, I., Nigou, J., Brando, T., van Soolingen, D., Daffe, M., and Guilhot, C. (2009) A lipid profile Puzo, G., Daffe,́ M., Benjamin, P., Coade, S., Buxton, R. S., Tascon, R. typifies the Beijing strains of Mycobacterium tuberculosis: identi- E., Rae, A., Robertson, B. D., Lowrie, D. B., Young, D. B., Gicquel, B., fication of a mutation responsible for a modification of the structures and Griffin, R. (2006) Comparative investigation of the pathogenicity of phthiocerol dimycocerosates and phenolic glycolipids. J. Biol. Chem. of three Mycobacterium tuberculosis mutants defective in the 284, 27101−27113.

1977 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews

(44) Malaga, W., Constant, P., Euphrasie, D., Cataldi, A., Daffe,́ M., (60) Perez,́ E., Constant, P., Lemassu, A., Laval, F., Daffe,́ M., and Reyrat, J.-M., and Guilhot, C. (2008) Deciphering the Genetic Bases of Guilhot, C. (2004) Characterization of Three Glycosyltransferases the Structural Diversity of Phenolic Glycolipids in Strains of the Involved in the Biosynthesis of the Phenolic Glycolipid Antigens from Mycobacterium tuberculosis Complex. J. Biol. Chem. 283, 15177− the Mycobacterium tuberculosis Complex. J. Biol. Chem. 279, 42574− 15184. 42583. (45) Daffe,́ M., Laneelle,́ M.-A., Lacave, C., and Laneelle,́ G. (1988) (61) Cole, S. T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Monoglycosyldiacylphenol-phthiocerol of Mycobacterium tuberculosis Harris, D., Gordon, S. V., Eiglmeier, K., Gas, S., Barry, C. E., Tekaia, F., and Mycobacterium bovis. Biochim. Biophys. Acta, Lipids Lipid Metab. Badcock, K., Basham, D., Brown, D., Chillingworth, T., Connor, R., 958, 443−449. Davies, R., Devlin, K., Feltwell, T., Gentles, S., Hamlin, N., Holroyd, S., (46) Constant, P., Perez, E., Malaga, W., Laneelle,́ M.-A., Saurel, O., Hornsby, T., Jagels, K., Krogh, A., McLean, J., Moule, S., Murphy, L., Daffe,́ M., and Guilhot, C. (2002) Role of the pks15/1 Gene in the Oliver, K., Osborne, J., Quail, M. A., Rajandream, M. A., Rogers, J., Biosynthesis of Phenolglycolipids in the Mycobacterium tuberculosis Rutter, S., Seeger, K., Skelton, J., Squares, R., Squares, S., Sulston, J. E., − Complex. J. Biol. Chem. 277, 38148 38158. Taylor, K., Whitehead, S., and Barrell, B. G. (1998) Deciphering the ́ ́ (47) Stadthagen, G., Kordulakova, J., Griffin, R., Constant, P., biology of Mycobacterium tuberculosis from the complete genome ́ ́ Bottova, I., Barilone, N., Gicquel, B., Daffe, M., and Jackson, M. (2005) sequence. Nature 393, 537−544. p-Hydroxybenzoic Acid Synthesis in Mycobacterium tuberculosis. J. (62) Ferreras, J. A., Stirrett, K. L., Lu, X., Ryu, J.-S., Soll, C. E., Tan, − Biol. Chem. 280, 40699 40706. D. S., and Quadri, L. E. N. (2008) Mycobacterial Phenolic Glycolipid (48) Ng, V., Zanazzi, G., Timpl, R., Talts, J. F., Salzer, J. L., Brennan, Virulence Factor Biosynthesis: Mechanism and Small-Molecule P. J., and Rambukkana, A. (2000) Role of the Cell Wall Phenolic Inhibition of Polyketide Chain Initiation. Chem. Biol. 15,51−61. Glycolipid-1 in the Peripheral Nerve Predilection of Mycobacterium (63) Chavadi, S. S., Edupuganti, U. R., Vergnolle, O., Fatima, I., − leprae. Cell 103, 511 524. Singh, S. M., Soll, C. E., and Quadri, L. E. N. (2011) Inactivation of (49) Yu, J., Tran, V., Li, M., Huang, X., Niu, C., Wang, D., Zhu, J., tesA Reduces Cell Wall Lipid Production and Increases Drug Wang,J.,Gao,Q.,andLiu,J.(2012)BothPhthiocerol Susceptibility in Mycobacteria. J. Biol. Chem. 286, 24616−24625. Dimycocerosates and Phenolic Glycolipids Are Required for Virulence − (64) Azad, A. K., Sirakova, T. D., Fernandes, N. D., and Kolattukudy, of Mycobacterium marinum. Infect. Immun. 80, 1381 1389. P. E. (1997) Gene Knockout Reveals a Novel Gene Cluster for the (50) Robinson, N., Kolter, T., Wolke, M., Rybniker, J., Hartmann, P., Synthesis of a Class of Cell Wall Lipids Unique to Pathogenic and Plum, G. (2008) Mycobacterial Phenolic Glycolipid Inhibits Mycobacteria. J. Biol. Chem. 272, 16741−16745. Phagosome Maturation and Subverts the Pro-inflammatory Cytokine − (65) Azad, A. K., Sirakova, T. D., Rogers, L. M., and Kolattukudy, P. Response. Traffic 9, 1936 1947. E. (1996) Targeted replacement of the mycocerosic acid synthase gene (51) van Soolingen, D., Qian, L., de Haas, P. E., Douglas, J. T., in Mycobacterium bovis BCG produces a mutant that lacks mycosides. Traore, H., Portaels, F., Qing, H. Z., Enkhsaikan, D., Nymadawa, P., Proc. Natl. Acad. Sci. U. S. A. 93, 4787−4792. and van Embden, J. D. (1995) Predominance of a single genotype of (66) Rainwater, D. L., and Kolattukudy, P. E. (1985) Fatty acid Mycobacterium tuberculosis in countries of east Asia. J. Clin. Microbiol. biosynthesis in Mycobacterium tuberculosis var. bovis Bacillus 33, 3234−3238. Calmette-Guerin.́ J. Biol. Chem. 260, 616−623. (52) Glynn, J. R., Whiteley, J., Bifani, P. J., Kremer, K., and van (67) Fitzmaurice, A. M., and Kolattukudy, P. E. (1997) Open reading Soolingen, D. (2002) Worldwide Occurrence of Beijing/W Strains of frame 3, which is adjacent to the mycocerosic acid synthase gene, is Mycobacterium tuberculosis: A Systematic Review. Emerging Infect. − expressed as an acyl coenzyme A synthase in Mycobacterium bovis Dis. 8, 843 849. − (53) Reed, M. B., Domenech, P., Manca, C., Su, H., Barczak, A. K., BCG. J. Bacteriol. 179, 2608 2615. Kreiswirth, B. N., Kaplan, G., and Barry, C. E. (2004) A glycolipid of (68) Mathur, M., and Kolattukudy, P. E. (1992) Molecular cloning hypervirulent tuberculosis strains that inhibits the innate immune and sequencing of the gene for mycocerosic acid synthase, a novel fatty − acid elongating multifunctional enzyme, from Mycobacterium tuber- response. Nature 431,84 87. − (54) Manca, C., Reed, M. B., Freeman, S., Mathema, B., Kreiswirth, culosis var. bovis Bacillus Calmette-Guerin. J. Biol. Chem. 267, 19388 B., Barry, C. E., and Kaplan, G. (2004) Differential Monocyte 19395. Activation Underlies Strain-Specific Mycobacterium tuberculosis (69) Rainwater, D. L., and Kolattukudy, P. E. (1983) Synthesis of Pathogenesis. Infect. Immun. 72, 5511−5514. mycocerosic acids from methylmalonyl coenzyme A by cell-free (55) Tsenova, L., Ellison, E., Harbacheuski, R., Moreira, A. L., extracts of Mycobacterium tuberculosis var. bovis BCG. J. Biol. Chem. − Kurepina, N., Reed, M. B., Mathema, B., Barry Iii, C. E., and Kaplan, G. 258, 2979 2985. (2005) Virulence of Selected Mycobacterium tuberculosis Clinical (70) Fitzmaurice, A. M., and Kolattukudy, P. E. (1998) An Acyl-CoA Isolates in the Rabbit Model of Meningitis Is Dependent on Phenolic Synthase (acoas) Gene Adjacent to the Mycocerosic Acid Synthase Glycolipid Produced by the Bacilli. J. Infect. Dis. 192,98−106. (mas) Locus Is Necessary for Mycocerosyl Lipid Synthesis in (56) Reed, M. B., Gagneux, S., DeRiemer, K., Small, P. M., and Barry, Mycobacterium tuberculosisvar. bovis BCG. J. Biol. Chem. 273, − C. E. (2007) The W-Beijing Lineage of Mycobacterium tuberculosis 8033 8039. Overproduces Triglycerides and Has the DosR Dormancy Regulon (71) Trivedi, O. A., Arora, P., Vats, A., Ansari, M. Z., Tickoo, R., Constitutively Upregulated. J. Bacteriol. 189, 2583−2589. Sridharan, V., Mohanty, D., and Gokhale, R. S. (2005) Dissecting the (57) Chen, J. M., Islam, S. T., Ren, H., and Liu, J. (2007) Differential Mechanism and Assembly of a Complex Virulence Mycobacterial productions of lipid virulence factors among BCG vaccine strains and Lipid. Mol. Cell 17, 631−643. implications on BCG safety. Vaccine 25, 8114−8122. (72) Chavadi, S. S., Onwueme, K. C., Edupuganti, U. R., Jerome, J., (58) Sinsimer, D., Huet, G., Manca, C., Tsenova, L., Koo, M.-S., Chatterjee, D., Soll, C. E., and Quadri, L. E. N. (2012) The Kurepina, N., Kana, B., Mathema, B., Marras, S. A. E., Kreiswirth, B. mycobacterial acyltransferase PapA5 is required for biosynthesis of cell N., Guilhot, C., and Kaplan, G. (2008) The Phenolic Glycolipid of wall-associated phenolic glycolipids. Microbiology (London, U. K.) 158, Mycobacterium tuberculosis Differentially Modulates the Early Host 1379−1387. Cytokine Response but Does Not in Itself Confer Hypervirulence. (73) Simeone, R., Huet, G., Constant, P., Malaga, W., Lemassu, A., Infect. Immun. 76, 3027−3036. Laval, F., Daffe,́ M., Guilhot, C., and Chalut, C. (2013) Functional (59) Perez,́ E., Constant, P., Laval, F., Lemassu, A., Laneelle,́ M. A., Characterisation of Three O-methyltransferases Involved in the Daffe,́ M., and Guilhot, C. (2004) Molecular Dissection of the Role of Biosynthesis of Phenolglycolipids in Mycobacterium tuberculosis. Two Methyltransferases in the Biosynthesis of Phenolglycolipids and PLoS One 8, e58954. Phthiocerol Dimycoserosate in the Mycobacterium tuberculosis (74) Camacho, L. R., Constant, P., Raynaud, C., Laneelle,́ M.-A., Complex. J. Biol. Chem. 279, 42584−42592. Triccas, J. A., Gicquel, B., Daffe,́ M., and Guilhot, C. (2001) Analysis of

1978 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979 ACS Chemical Biology Reviews the Phthiocerol Dimycocerosate Locus of Mycobacterium tuber- Ester p-HBAD-I, Virulence Markers of Mycobacterium tuberculosis. culosis. J. Biol. Chem. 276, 19845−19854. Eur. J. Biochem. 2013, 4642−4654. (75) Li, W., Upadhyay, A., Fontes, F. L., North, E. J., Wang, Y., Crans, (91) Bourke, J., Brereton, C. F., Gordon, S. V., Lavelle, E. C., and D. C., Grzegorzewicz, A. E., Jones, V., Franzblau, S. G., Lee, R. E., Scanlan, E. M. (2014) The synthesis and biological evaluation of Crick, D. C., and Jackson, M. (2014) Novel Insights into the mycobacterial p-hydroxybenzoic acid derivatives. Org. Biomol. Chem. Mechanism of Inhibition of MmpL3, a Target of Multiple 12, 1114−1123. Pharmacophores in Mycobacterium tuberculosis. Antimicrob. Agents (92) Elsaidi, H. R., and Lowary, T. L. (2014) Inhibition of cytokine Chemother. 58, 6413−6423. release by mycobacterium tuberculosis phenolic glycolipid analogues. (76) Jain, M., and Cox, J. S. (2005) Interaction between Polyketide ChemBioChem 15, 1176−1182. Synthase and Transporter Suggests Coupled Synthesis and Export of (93) Arbues, A., Malaga, W., Constant, P., Guilhot, C., Prandi, J., and Virulence Lipid in M. tuberculosis. PLoS Pathog. 1,12−19. Astarie-Dequeker, C. (2016) Trisaccharides of Phenolic Glycolipids (77) Sulzenbacher, G., Canaan, S., Bordat, Y., Neyrolles, O., Confer Advantages to Pathogenic Mycobacteria through Manipulation Stadthagen, G., Roig-Zamboni, V., Rauzier, J., Maurin, D., Laval, F., of Host-Cell Pattern-Recognition Receptors. ACS Chem. Biol. 11, Daffe,́ M., Cambillau, C., Gicquel, B., Bourne, Y., and Jackson, M. 2865−2875. (2006) LppX is a lipoprotein required for the translocation of (94) Schmidt, R. R., and Toepfer, A. (1991) Glycosylation with phthiocerol dimycocerosates to the surface of Mycobacterium highly reactive glycosyl donors: efficiency of the inverse procedure. tuberculosis. EMBO J. 25, 1436−1444. Tetrahedron Lett. 32, 3353−3356. (78) Nigudkar, S. S., and Demchenko, A. V. (2015) Stereocontrolled (95) Jones, C. (2005) Vaccines based on the cell surface 1,2-cis glycosylation as the driving force of progress in synthetic carbohydrates. An. Acad. Bras. Cienc. 77, 293−324. carbohydrate chemistry. Chem. Sci. 6, 2687−2704. (96) Peri, F. (2013) Clustered carbohydrates in synthetic vaccines. (79) Fujiwara, T., Hunter, S. W., Cho, S. N., Aspinall, G. O., and Chem. Soc. Rev. 42, 4543−4556. Brennan, P. J. (1984) Chemical synthesis and serology of (97) Anderson, P., Pichichero, M. E., and Insel, R. A. (1985) and trisaccharides of phenolic glycolipid antigens from the leprosy Immunogens consisting of oligosaccharides from the capsule of bacillus and preparation of a disaccharide protein conjugate for Haemophilus influenzae type b coupled to diphtheria toxoid or the − serodiagnosis of leprosy. Infect. Immun. 43, 245−252. toxin protein CRM197. J. Clin. Invest. 76,52 59. (80) Chatterjee, D., Cho, S.-N., Stewart, C., Douglas, J. T., Fujiwara, (98) Darkes, M. J. M., and Plosker, G. L. (2002) Pneumococcal T., and Brennan, P. J. (1988) Synthesis and Immunoreactivity of Conjugate Vaccine A Review of its Use in the Prevention of Neoglycoproteins containing the trisaccharide unit of phenolic Streptococcus pneumoniae Infection. Paediatr. Drugs 4, 609. glycolipid I of Mycobacterium leprae. Carbohydr. Res. 183, 241−260. (99) Balmer, P., Borrow, R., and Miller, E. (2002) Impact of (81) Fujiwara, T., Aspinall, G. O., Hunter, S. W., and Brennan, P. T. meningococcal C conjugate vaccine in the UK. J. Med. Microbiol. 51, − (1987) Chemical synthesis of the trisaccharide unit of the species- 717 722. specific phenolic glycolipid from mycobacterium leprae. Carbohydr. (100) Bovier, P. (2008) Epaxal: a virosomal vaccine to prevent − Res. 163,41−52. hepatitis A infection. Expert Rev. Vaccines 7, 1141 1150. (82) Wu, X., Mariño-Albernas, J. R., Auzanneau, F. I., Verez- (101) Safari, D., Marradi, M., Chiodo, F., Th Dekker, H. A., Shan, Y., Bencomo, V., and Pinto, B. M. (1998) Synthesis and NMR analysis of Adamo, R., Oscarson, S., Rijkers, G. T., Lahmann, M., Kamerling, J. P., 13C-labeled oligosaccharides corresponding to the major glycolipid Penades, S., and Snippe, H. (2012) Gold nanoparticles as carriers for a − synthetic Streptococcus pneumoniae type 14 conjugate vaccine. from Mycobacterium leprae. Carbohydr. Res. 306, 493 503. − (83) Gurjar, M. K., and Reddy, K. R. (1992) Synthesis of di- and tri- Nanomedicine (London, U. K.) 7, 651 662. saccharide segments of a phenolic glycolipid of Mycobacterium (102) Deng, S., Bai, L., Reboulet, R., Matthew, R., Engler, D. A., kansasii. Carbohydr. Res. 226, 233−238. Teyton, L., Bendelac, A., and Savage, P. B. (2014) A peptide-free, (84) Elsaidi, H. R. H., and Lowary, T. L. (2015) Effect of phenolic liposome-based oligosaccharide vaccine, adjuvanted with a natural killer T cell antigen, generates robust antibody responses in vivo. glycolipids from Mycobacterium kansasii on proinflammatory cytokine − release. A structure−activity relationship study. Chem. Sci. 6, 3161− Chem. Sci. 5, 1437 1441. 3172. (103) Sun, X., Pan, Q., Yuan, C., Wang, Q., Tang, X. L., Ding, K., (85) Elsaidi, H. R. H., Barreda, D. R., Cairo, C. W., and Lowary, T. L. Zhou, X., and Zhang, X. L. (2016) A Single ssDNA Aptamer Binding (2013) Mycobacterial Phenolic Glycolipids with a Simplified Lipid to Mannose-Capped Lipoarabinomannan of Bacillus Calmette-Guerin Enhances Immunoprotective Effect against Tuberculosis. J. Am. Chem. Aglycone Modulate Cytokine Levels through Toll-Like Receptor 2. − ChemBioChem 14, 2153−2159. Soc. 138, 11680 11689. (86) Veeneman, G. H., Van Leeuwen, S. H., Zuurmond, H., and Van (104) Tam, P.-H., and Lowary, T. L. (2010) Mycobacterial lipoarabinomannan fragments as haptens for potential anti-tuber- Boom, J. H. (1990) Synthesis of Carbohydrate-Antigenic Structures of − Mycobacterium Tuberculosisusing an Iodonium Ion Promoted culosis vaccines. Carbohydr. Chem. 36,38 63. − (105) Patil, P. S., Zulueta, M. M. L., and Hung, S.-c. (2014) Synthesis Glycosidation Approach. J. Carbohydr. Chem. 9, 783 796. − (87) Fujiwara, T. (1991) Synthesis of the trisaccharide-protein of Phosphatidylinositol Mannosides. J. Chin. Chem. Soc. 61, 151 162. (106) Meng, X., Ji, C., Su, C., Shen, D., Li, Y., Dong, P., Yuan, D., conjugate of the phenolic glycolipid of Mycobacterium tuberculosis for Yang, M., Bai, S., Meng, D., Fan, Z., Yang, Y., Yu, P., and Zhu, T. the serodiagnosis of tuberculosis. Agric. Biol. Chem. 55, 2123−2128. (2017) Synthesis and immunogenicity of PG-tb1 monovalent (88) Gurjar, M. K., and Viswanadham, G. (1991) Communication: glycoconjugate. Eur. J. Med. Chem. 134, 140−146. Synthesis of Me α-L-Rhap-(1→3)-2-O-Me-α-L-Rhap And Me 2,3,4- Tri-O-Me-α-L-Fucp-(1→3)-α-L-Rhap-(1→3)-2-O-Me-α-L-Rhap: Oli- gosaccha-Ride Segments of Phenolic Glycolipids in Mycobacterium Bovis Bcg and Tuberculosis Strain Canetti. J. Carbohydr. Chem. 10, 481−485. (89) Barroso, S., Castelli, R., Baggelaar, M. P., Geerdink, D., ter Horst, B., Casas-Arce, E., Overkleeft, H. S., van der Marel, G. A., Codee,́ J. D. C., and Minnaard, A. J. (2012) Total Synthesis of the Triglycosyl Phenolic Glycolipid PGL-tb1 from Mycobacterium tuberculosis. Angew. Chem., Int. Ed. 51, 11774−11777. (90) Barroso, S., Geerdink, D., ter Horst, B., Casas-Arce, E., and Minnaard, A. J. (2013) Total Synthesis of the Phenolic Glycolipid Mycoside B and the Glycosylated p-Hydroxybenzoic Acid Methyl

1979 DOI: 10.1021/acschembio.7b00394 ACS Chem. Biol. 2017, 12, 1969−1979