Ph.D. thesis

Development of myeloid cell-based immune profiling on microarrays

Zoltán Szittner

Supervisor: József Prechl, MD, PhD

Doctoral School of Biology Immunology Program Head: Professor Anna Erdei, DSc

Department of Immunology, Institute of Biology Eötvös Loránd University, Budapest, Hungary

2016

TABLE OF CONTENTS

List of abbreviations 3

1. INTRODUCTION 5

1.1 Immunoglobulins and their receptors 7

1.2 and complement receptors 14

1.3 Myeloid cells, monocytes and U937 16

1.4 in autoimmune diseases – rheumatoid arthritis and systemic lupus erythematosus 18

1.5 Protein microarrays 21 1.5.1 Cellular microarrays 23 1.5.2 Cellular SPR 24

2 AIMS OF THE STUDY 25

3 MATERIALS AND METHODS 26

4 RESULTS 35

4.1 Application of fluorescently labeled U937 cells in and microarrays 35 4.1.1 U937 cells are classical inflammatory monocytes based on cell surface receptor expression 36 4.1.2 U937 cells show unequal binding to hIgG subclasses compared to anti-hIgG antibody 37 4.1.3 U937 cells bind to human IgG subclasses in a dose dependent manner 39 4.1.4 Cell density as a factor determining U937 cell binding to hIgG1-4 40 4.1.5 U937 binding is mediated by IgG Fcγ part, while complement plays a modulatory role in binding to Staphylococcus aureus 41 4.1.6 Deglycosylation of IgG impairs cell binding to immune complexes 42 4.1.7 U937 cells can detect IgG containing ICs in protein microarrays 44 4.1.8 Microfluidic chambers supporting cellular microarray measurements 47

4.2 Application of U937 cell line to detect ICs in imaging SPR experiments 48 4.2.1 U937 cells bind to human IgG1, IgG3 and IgG4 but not to IgG2 48 4.2.2 U937 cells detect ICs generated by the treatment of VCP2 with RA sera 50 4.2.3 VCP2-specific IgA, IgG and IgG3 levels correlate with the U937 cell binding 52 4.2.4 VCP2 -IC detection by U937 significantly associates with RA diagnosis 56 4.2.5 IgG subclass specific serum binding response is significantly elevated in samples from RA patient donors 56

4.3 NF-κB reporter U937 58 4.3.1 LPS triggers GFP production in B12 cells in a dose and time-dependent manner 59 4.3.2 FcγR binding of coated IgG subclasses, hIgG and IVIG activates GFP production in B12 cells in a dose dependent manner while IgA and IgM trigger no activation 60

1

5 DISCUSSION 62

Summary 75

Összefoglalás 77

New scientific results 79

Acknowledgments 80

List of publications 81

Publications connected to the thesis 81

Other Publications 81

References 83

2

List of abbreviations

ACPA – Anti citrullinated peptide antigen ADCC – Antibody dependent cellular cytotoxicity ADCP – Antibody dependent cellular phagocytosis ADCR – Antibody dependent release ADRB – Antibody dependent respiratory burst AID – Activation-induced cytidine deaminase AP – Alternative pathway (of complement) APC – Antigen presenting cell BCR – B-cell receptor BSA – Bovine serum albumin CARD9 – Caspase recruitment domain-containing protein 9 CCP Cyclic citrullinated peptide CD – Cluster of differentiation CDC – Complement dependent cytotoxicity cDNA – Complementary DNA CMFDA – 5-Chloromethylfluorescein diacetate CMP – Common myeloid progenitor CP – Classical pathway (of complement) CR – Complement receptor CSR – Class switch recombination DNA – Deoxyribonucleic acid dsDNA – Double-stranded DNA ECM – Extracellular matrix EDTA – Ethylenediaminetetraacetic acid ELISA – Enzyme-linked immunosorbent assay Fab – Fragment antigen-binding FACS – Fluorescence-activated cell sorting Fc Fragment crystallizable region FcR – Fc receptor, GFP – Green fluorescent protein HCP2 – Histone citrullinated peptide 2 HIV – Human virus HSC – Hematopoietic stem cell IC – Ig – Immunoglobulin IL – Interleukin iRFP – near-infrared fluorescent protein

3

ITAM – Immunoreceptor tyrosine-based activation motif ITIM – Immunoreceptor tyrosine-based inhibitory motif LP – Lectin pathway (of complement) LPS – Lipopolysaccharide MAC – Membrane attack complex MAP – Multiple antigenic peptide MAPK – Mitogen-activated protein kinase MASP – MBL-associated serine protease MBL – Mannan-binding lectin MES – 2-(N-morpholino)ethanesulfonic acid MHC – Major histocompatibility complex NF-κB – Nuclear factor kappa-light-chain-enhancer of activated B cells NHS – Normal human serum PAD – Peptidyl arginine deiminase PBS – Phosphate buffered saline PMN – Polymorphonuclear leukocytes PRR – Pattern recognition receptor RA – Rheumatoid arthritis RAG – Recombination activating gene RE – Responsive element RFI – Relative fluorescent intensity RPMI – Roswell Park Memorial Institute medium RNA – Ribonucleic acid RU – Resonance Unit SHP-1 – Src homology region 2 domain-containing phosphatase-1 shRNA – Small hairpin RNA siRNA – Small interfering RNA SLE – Systemic lupus erythematosus SPR – Surface plasmon resonance ssDNA – single-stranded DNA TCR – T-cell receptor TLR – Toll-like receptor TNF – Tumor necrosis factor TRIS – Tris(hydroxymethyl)-aminomethane VBS – Veronal buffered saline VCP2 – Viral citrullinated peptide 2

4

1. Introduction

Humoral is mediated by various macro-molecules, mainly proteins, found in our extracellular fluids, most importantly in the blood plasma and serum. The state of determines an organism's capability to protect itself against various microorganisms, yet in some cases, as in or , can cause harm as well.

The history of humoral immunity started with the identification of serum components being responsible for antimicrobial effects. Throughout these investigations, researchers gave various names to this component of serum responsible for protection based on the experimental setting applied. Alexins, Antitoxin, Bacteriolysin, Bacterial agglutinin, Hemolysin, were all the important early observations paving the way for our current understanding of humoral immunity. These observations lead to the notion that humoral immunity is a versatile system, as for example opsonin helps the in phagocytosis, bacteriolysin lyses and antitoxin neutralizes toxin. Later, the antibody-antigen hypothesis was born and the exploration of what we know today as plasma proteome started. By biochemical methods, globulins were identified as those responsible for the anti-microbial effects and were therefore named as immunoglobulins. Today we know that next to immunoglobulins, the complement system and other soluble molecular pattern recognition molecules are important factors of humoral immunity as well. These patterns, parts of molecules hold the key in immunology. The specificity of a molecular interaction in this complex system draws the line between health and disease.

Nowadays the typical approach in diagnostics investigating antigen specific immune response focuses on the detection of one molecule binding to the antigen, despite the fact that when immune complexes form many different components: , complement components and pentraxins from the serum can react and bind to the antigen. In this typical setting monoclonal antibodies, produced in animals different from the investigated sample, specific to the given serum component may be detected through fluorescence, radioactivity or enzyme activity.

In our body these antigen-antibody binding derived immune complexes, may mediate various effector functions. The main effector functions mediated by antibodies

5 are antibody dependent cellular cytotoxicity (ADCC), antibody dependent cellular phagocytosis (ADCP), antibody dependent cytokine release (ADCR) and antibody dependent respiratory burst (ADRB). Not all antibodies are equal. While some isotypes may be better in mediating one aspect, the same antibody can be weaker in another. Therefore this thesis investigates the hypothesis that the detection of serum components alone may be further improved by detecting multiple reactions in parallel by using myeloid cells as detectors. These cells with receptors for immune complex components can be applied to detect immune complexes, and through their binding, or by detecting their inflammatory activation, activating immune complexes can be identified.

Multiplex diagnostic systems allow simultaneous detection of multiple antigen specific reactions, simply by spatially encoding the different or compounds, by fixing them at given locations of an array to form microarray or protein chip of features. Such systems show their strength in analyzing multiple immune-reactions, between the serum derived components of humoral immunity and the fixed antigens, simultaneously. These capabilities combined with fluorescent detection offer great advantage for bio- marker identification, and diagnostic purposes.

In autoimmune diseases, the that normally protects the body becomes reactive to self-antigens, leading to , local or systemic, and harms functionality of various organs and tissues depending on the target antigen. In a subgroup of these diseases auto-antibodies are generated and in many cases patients have autoantibodies against multiple autoantigens. Treatment and diagnosis of autoimmune diseases is a growing and challenging field where new approaches became necessary to improve the precision of diagnosis and the understanding of the underlying pathological mechanisms.

The diversity of humoral immunity and the effector functions triggered by the immune-complexes calls for novel methods to provide a more biological readout, compared to secondary antibodies. This work presents our results of in this field, with the emphasis on cell based diagnostics for autoimmune diseases, in a multiplex system, presenting how monocytoid U937, as a model for myeloid cells, can be applied for immune profiling.

6

1.1 Immunoglobulins and their receptors

Immunoglobulins or antibodies are glycoproteins on B-cells as parts of the Receptors and also in soluble form. The soluble form plays a central role in humoral immunity and is produced by antibody secreting cells: plasmablasts and plasma cells that differentiate from B [1]. Plasmablasts can differentiate from mature B-cells following activation by antigen. This activation can happen both in T- cell dependent (TD) and T-cell independent (TI) manner in the peripheral lymphoid organs [2]. The TI induction of B-cells can be triggered by polyvalent antigens with repetitive structures or through co-stimulation of the B-cells through other receptors recognizing pathogen associated patterns. This activation results in the antibodies of the primary response during infection; and these antibodies are mainly of IgM and have low affinity. The TD response results in secondary antibody response that constitutes of class switched, high affinity antibodies. Plasma cells will later migrate from their place of induction to various tissues guided by chemokines and adhesion molecules during the plasma-cell homing [3]. As the number niches for plasma cells is limited it maximizes the number of plasma cells and as well new plasma blasts have to compete with the plasma cells [4]. Each cell means a single clone and therefore produces antibodies with single specificity.

Antibodies consist of four polypeptide chains: two identical heavy and two identical light chains that are connected by disulfide bonds. Both chains are built from immunoglobulin domains [5], which as revealed by crystal structure, characteristically contain Beta sheet-folds [6]. Both chains take part in forming the antigen binding site, in the variable or Fab domain. The upper part of the Y shaped antibody is responsible for the bivalent and mono-specific recognition of specific molecular targets on antigens, the , by the two arms, forming the Fab or variable region (Figure 1). It is the generation of the antigen binding sites in the variable regions, at the tip of the arms, the hypervariable loops in the complementarity determining regions (CDR) that results the great diversity, determining the specificity of the antibody [7]. The antigen binding variable region of antigen receptors of B-cells are generated through somatic recombination during the V(D)J recombination. The recombination involves both the heavy and the light chains. While in case of heavy chain variable, diversity and joining gene segments are combined, the light chains lack the D segment. These gene segments can be found in multiple distinct copies in the loci encoding parts of the antibody [8].

7

The quasi stochastic recombination of segments by RAG enzymes is the primary source of antibody variation. During affinity maturation in the germinal center reaction occurs in the genes encoding the variable domain of B-cell clones, and followed by a positive selection, only the clones with the highest affinity BCR for specific antigens will be selected [9]. This process is mediated by the AID enzyme [10]. AID mediates class switch recombination (CSR) as well [11]. The lower tail part, the Fc or constant region determines the effector functions the antibody can trigger (Figure 1). The Fc and Fab regions are connected with the hinge region, that determines the flexibility of the molecule [12].

Figure 1. Schematic overview of the antibody structure. The heavy chain consists of one variable (VH) and three constant domains (CH1-3), while the light chains have one variable (VL) and one constant domain (CL). The two heavy chains of the antibodies are covalently bound by disulfide bonds (marked with -s-s-) in the hinge region formed be- tween the CH1 and CH3 domains. Light and heavy chains are held together by disulfide bonds as well. In the Antibody Binding Fragment (Fab) region the Complementary De- termining Regions (CDR) determine the antibody’s specificity while the constant frag- ment (Fc) is accountable for the effector functions. The position of carbohydrate sidechains is shown (N-Linked glycan).

The constant region of the heavy chains determines the isotypes of the antibody. In humans five antibody isotypes exist: IgA, IgD, IgE, IgG and IgM. IgA IgD and IgG heavy chains have 1 variable and 3 constant domain, while IgE and IgM have an extra constant domain in their Fc region. In case of light chains it is one variable and constant

8 domain for both types: lambda and kappa [8]. During differentiation in B cells isotype switch may occur, and the B cells initially producing IgM and IgD express other isotypes. Downstream of the VDJ genes are found the nine genes (4 IgG and 2 IgA subclasses, IgM, IgE and IgD) encoding the constant regions. CSR results in enhanced variety of the effector function of the antibodies through changing from IgM or IgD to other heavy chain [11]. During CSR the extracellular signals and the cytokine milieu determine which heavy chain will be used and only the isotype changes while the variable region remains unaffected [13]. Yet recent data compiled on the effect of constant domain on the variable region suggests that constant domains may affect the variable region function and structure [14]. Moreover class switching is a directional process, clones can only change to heavy chains encoded downstream, possibly resulting in a sequence supporting immune system to control various antigens accordingly [15]. IgG accounts for up to 75 % of all immunoglobulins in human serum, consists of four subclasses IgG1, IgG2, igG3 and IgG4, showing a decreasing abundance in this order. Each subclass plays a specific role and is raised against different antigen types. The subclasses show 95% amino acid sequence similarity. Importantly the 5% difference is in the hinge region and surface exposed residues and these have important impact on the effector function of these antibodies. Moreover IgG subclasses show allotypic variation that may further affect the triggered effector functions [16]. IgG1 is the most abundant subclass in serum and triggers CDC, ADCC and ADCP and binds to both activating and inhibitory FcγRs and targets soluble and membrane bound protein antigens. IgG2 can activate complement and phagocytosis only at large densities, shows an improved resistance to proteolytic degradation, and most commonly raised against bacterial polysaccharides. IgG3 is a strong activator of both CDC and FcγR mediated effector functions, shows a half-life significantly lower compared to other subclasses, targets protein antigens for example in viral infections. IgG4 lacks CDC activating properties can disassemble and form bispecific antibodies [17]. Most commonly production of IgG4 is the result of long-term exposure to non- infectious antigens and [16, 18, 19].

IgM is the first antibody produced by the B-cells, and also during the immune response. IgM exists in soluble form where it can be found as pentamers and hexamers, yet only pentamers contain J-chain [20].Functionally IgM plays a role as an opsonin and can strongly activate the classical complement pathway [20]. Recently Fc Receptors for

9

IgM (FcµR) have been described on lymphocytes [21] and Fcα/µR on macrophages and B-cells and were shown to trigger endocytosis [21, 22]. Natural and immune IgM are distinct molecules. Natural IgM is polyreactive and is produced without antigenic stimuli and was shown to play a role in homeostasis and immunoregulation [23]. Immune IgM is raised during the primary immune response. IgA has two subclasses IgA1 and IgA2 that differ in their hinge region. IgA is the second most abundant serum antibody and the first on mucosal surfaces [24]. While in serum they are monomers in the mucosal surfaces they are found as secretory IgA and form dimers, through J- chains, to help the secretion by pIgR [25]. IgA plays an important role in the regulation gut microbiome as well [26]. IgA can only weakly activate the complement system, still it is recognized by FcαR and can trigger ADCC, ADCP, cytokine release and respirators burst [27]. IgE levels are very low in the serum, as these antibodies can bind to FcεRI on Mast Cells and Basophil granulocytes with high affinity. IgE is the central Ig type in allergy, crosslinking two receptors through binding of an by the receptor bound IgEs results in receptor aggregation and triggers the acute allergic reactions, in immediate reaction [28]. IgE is also known to play a role in defense to parasitic infections. IgE containing immune complexes can as well bind to CD23, the low affinity FcεR [29]. IgD plays a central role in homeostasis and maturation of B-cells yet recently their role in mucosa associated immune response and antimicrobial activity have been shown [30, 31].

Immunoglobulins are glycoproteins, and N- and O- linked glycans can be found in both the Fab and Fc part of the antibodies, play important roles Ig structure. Various glycoforms exist based on the carbohydrate sidechain, and were shown to significantly affect the Ig binding to FcRs, while glycosylation of the Ig Fab part was shown to be capable of modulating antigen binding [32]. On IgG these N-linked oligosaccharides are attached to Asparagine 297. Various glycoforms were shown to contribute to autoimmune diseases as well [33]. The functions of Igs are summarized in Figure 3.

Fc receptors provide a link between adaptive and innate immunity. FcRs are membrane bound receptors specifically binding Fc part of antibodies. FcRs are accountable for various effector functions such as ADCC, ADCP, ADCR, and induction of respiratory burst triggered depending on the cellular expression, and fine specificity of the receptors in leukocytes [34]. IgA (FcαR) [35], IgE (FcεR)[36], IgG (FcγR) , IgM (FcµR)[32] and IgA/IgM (Fcα/µR)[20, 22, 35] specific FcRs were all demonstrated to

10 contribute to these mechanisms (Figure 2 and Figure 3). FcRs may also serve as transporters of the Igs, such as pIgR[37] and FcRn[38]. FcRs are mainly activating receptors, and signal through immunoreceptor tyrosine-based activation motif (ITAM), yet with one known example, through immunoreceptor tyrosine-based inhibitory motif (ITIM), they may regulate cellular functions. The signaling through these motives is initiated by the aggregation of FcRs by. Aggregation can be triggered by immune- complexes as it happens in vivo or by receptor specific antibodies. When both activating and inhibitory receptors are cross-linked their ratio determines final outcome.

Figure 2. The human Fc receptors. With the exception of FcεRII a C-type Lectin and neonatal FcR (FcRn), an MHC related molecule, all Fc receptors shown belong to the immunoglobulin superfamily. These receptors form membrane bound hetero-oligomers with each polypeptide chain marked with α, β or γ. FcγRIIIb is attached to the mem- brane through Glycosylphosphatidylinositol (GPI) anchor. Their schematic structures are shown with white blocks marking immunoreceptor tyrosine-based activation motifs (ITAM) and black in case of FcγIIb, the immunoreceptor tyrosine-based inhibitory motif (ITIM). Polymorphic variants are shown in case of FcγRIIa and FcγRIIIa. High affinity receptors are shown in the first column while the other receptors are all bind their Ig targets with low/medium affinity. Based on [39] and [40]. Most FcRs consist of immunoglobulin superfamily domain based hetero- oligomers, with a ligand binding α and accessory chains (γ, β). The most common accessory chain, the FcRγ chain can be found in complex with FcγRI and FcγRIII, FcαR and FcεRI, where the α ligand binding chains lack the ITAM motives, therefore their

11 presence is a prerequisite for FcR signaling in the case of these receptors. In the case of FcγRII the signaling motives are in the α chains. ITAM aggregation recruits and activates Syk that results in Ca2+ immobilization, phosphorylation of MAPK, and activation of the NF-κB pathway [41]. Aggregation of ITIM results in activation of SHP-1 phosphatases, which will attenuate cell activation triggered through other receptors by dephosphorylating various protein targets [42]. Recently the inhibitory ITAM signaling has been demonstrated, where by a suboptimal ligand binding to FcRs can inhibit activation of signaling initiated through heterologous receptors, in a SHP-1 mediated process [43]. For example suboptimal crosslinking of FcγRIIa can inhibit LPS triggered activation [44], and similar results were reported in case of FcαRI [45].

In humans one inhibitory, FcγIIb (CD32B), and several activating FcγRs exist: FcγRI, FcγRIIa, FcγRIIIa and FcγRIIIb. These receptors differ in their molecular structure and therefore in their affinity to bind IgG molecules in general, and as well IgG subclasses. The high affinity FcγRI can bind IgG molecules in their soluble, monomeric form, possibly due to the presence of an extra third extracellular Ig domain. The other FcγRs have lower binding affinity to IgG, and cannot bind monomeric IgG, only immune-complexes through a low affinity yet multivalent binding. Moreover polymorphic variants of FcγRIIa and FcγRIIIa have different affinities towards the IgG subclasses [46], and IgG allotypes bind with different affinity to the FcRs [16, 47]. The quality of IgG glycosylation has been demonstrated to further regulate the affinity of IgG-FcγR binding reactions[48].

12

Figure 3. Overview of Immunoglobulin effector functions. Igs inhibit toxin and virus receptor binding, can precipitate soluble antigens and lead to agglutination of microbi- al cells. Antibodies binding to antigens in high density may trigger activation of the classical complement pathway, that can further opsonize the targets with activation fragments and as well cause inflammation and recruit effector cells to the site of the immune response. Activation of the complement may also result in the assembly of the membrane attack complex and lead to complement dependent cytoxicity. The Fc recep- tor triggered responses are very diverse the most important receptors on immune cells are shown with listing their main FcR triggered functions. Black stars indicate expres- sion on subsets while empty stars mark inducible expression. Based on [47] and [49]. 13

1.2 Complement system and complement receptors

The complement system consists of the components of the complement activation pathway in their inactive forms, complement regulation factors, and receptors recognizing fragments of complement components produced upon the activation of the components. The chain reaction like the activation of complement system results in the assembly of the terminal or Membrane Attack Complex, which leads to the lysis of targeted foreign cells of pathogens. The activation of each complement pathway results in the proteolytic cleavage of the early components, leading to formation of convertases that will in each next step enzymatically activate the next components. Figure 4 gives a brief overview of the three pathways. While these pathways differ by their mechanism of activation, they share complement component C3 as the central molecule of activation, and from its cleavage they share and converge on the same terminal part of complement activation. The classical pathway activates through binding of C1q to complexed Ig molecules, mainly IgG1, -2 and -3 and IgM. The binding of C1q results in activation of serine protease C1r that activates C1s that has two substrates C4 and C2. Upon their cleavage the bigger fragments form a complex called C4b2a, the C3 convertase. The same C4b2a complex is generated upon activation of the lectin pathway, yet importantly its initiation is activated by the binding of Mannose Binding Lectin (MBL) or ficollins to specific carbohydrates characteristically found in the cell wall of bacteria. Here, the initial activation is carried out by the MBL associated Serine Proteases (MASPs) that will activate upon the conformational changes occurring in MBL upon substrate binding. Finally the Alternative pathway is activated by surfaces without syalic acid. Syalic acid is expressed on all nucleated cells in humans, therefore the alternative pathway will only activate on foreign compounds. To initiate AP activation the hydrolyzed form of C3, C3H2O binds Factor B, which upon activation by Factor D, forms C3(H2O)Bb, the C3 convertase of the alternative pathway. From this point the generated C3b fragments can form new complexes with Factor B; therefore this pathway provides an amplification loop for the other complement pathways as well. Recruitment of a further C3b to the C3bBb complex results in C3bBb3b complex, the C5 convertase of the alternative pathway, yet as well C3b binding to the C4b2a complex generated by the activation of the classical or the lectin pathway will result in C4b2b3b complex, that cleaves C5 into C5b, and C5b mediates the set-up of the terminal complex, while the smaller fragments C3a and C5a are released and act as

14 inflammatory, anaphylactic modulators. From the generation of C5b all pathways converge into the assembly of the terminal Membrane Attack Complex (MAC) that will lead to the lysis of the target cell. Upon the proteolytic activation of C3 the resulting C3b molecules hidden tioesther bond reveals itself and is capable of covalently binding to activating surfaces most frequently to –OH groups of proteins and carbohydrates [50, 51]. Further cleavage of the surface bound C3b results in iC3b, and later C3dg and C3d, all ligands of various complement receptors. Thus complement plays a role in opsonization and clearance of microbes, immune-complexes and apoptotic cells through induction of phagocytosis [52].

Figure 4. Overview of the complement activation. The alternative pathway (AP) is initiated by the constant hydrolysis of C3 and stabilization of the complex by factor B and activation by factor D this pathway is accountable for the amplification of the com- plement regardless of the imitation. The Lectin and Classical pathways (LP and CP) are initiated by the binding of Mannose Binding Lectin (MBL) or ficollins (F) to carbohy- drates or C1q to antibodies, through binding their serine proteases activate and cleave C2 and C4 to form C3 convertases. The activity of this complex leads to activation and cleavage of C3, opsonization by C3 fragments and assembly of the C5 convertase that will finally result cell lysis, through assembly of the Membrane Attack Complex (MAC). Anaphylatoxins C3a and C3a trigger pro-inflammatory response and chemotaxis, while fragments of C3b are important mediators of phagocytosis. Based on [51].

In humans four classical receptors for these complement fragments are known: CR1 binds C3b, and is expressed widely, importantly on Red Blood Cells that help the clearance of immune-complexes through binding them with CR1 and transporting the ICs to tissue macrophages of the spleen and liver. CR2 is the receptor for C3d and 15

C3dg, expressed on B- cells, part of the co-receptor complex. CR3(CD11b/CD18) and CR4(CD11c/CD18) are integrins and show widespread expression among PMNs, and as well expressed by monocytes and macrophages, and help phagocytosis and leukocyte adhesion [53]. A further phagocytic receptors: CRIg have been demonstrated on tissue macrophages [54] and receptors for C1q have been suggested, yet their role is not yet clear [55]. The smaller complement fragments: C5a and C3a trigger inflammatory and chemotactic activation in almost all leukocytes, and epithelial cells, through C5aR and C3aR, G-protein coupled, 7-span transmembrane receptors [56]. The complement system may as well cause harm to host tissues and organs, yet numerous inhibitors membrane-bound and soluble exist to prevent and regulate uncontrolled activation. These regulators of complement activation may act on the level of convertases, by inhibiting convertase activity and thus further cleavage, and as well by inhibiting the polymerization of C9 thus preventing the assembly of MAC [57].

1.3 Myeloid cells, monocytes and U937

All blood cells develop and differentiate from Hematopoietic Stem Cells (HCSs). Two main arms of HSC differentiation are established: with a simplified approach adaptive immune cells arise from the Common Lymphoid Progenitors and innate immune cells from the Common Myeloid Progenitors (CMP), and both differentiate from the HSCs in the bone marrow, yet may partly overlap in some cases [58]. Between HSC and the terminally differentiated cells stages of cell differentiation are identified based on cell surface marker expression profiles and controlled by both transcription factors and epigenetic regulators [59, 60].

Myeloid cells develop In the Red Bone Marrow from CMPs. Monocytes and granulocytes develop from the granulocyte progenitors [58]. These cell types and macrophages are known to cooperate during the course of inflammation in many aspects [61]. While generally granulocytes are short-lived and terminally differentiated, monocytes that enter the circulation show great plasticity and can undergo further differentiation after extravasation to inflamed tissues [62]. Recently it became clear that tissue resident macrophages do not exclusively derive from monocytes, as some populations for example Kuppfer cells and microglia were found

16 have self-renewal capacity, and derive from yolk sac and derived embryonic progenitors [63, 64].

Subsets of human blood monocytes are established based on their expression of CD14, the LPS coreceptor and FcR CD16: the CD14++, CD16– population, called classical (inflammatory) monocytes, the CD14–, CD16++ non-classical monocytes and the CD14+, CD16+ intermediate monocytes [65]. Classical monocytes are recruited to the site of inflammation, produce inflammatory , coagulation and complement factors, show phagocytic and superoxide producing activity, serve as APCs, and as well help the resolution of inflammation [62]. Once they reach the site of inflammation, monocytes differentiate into monocyte derived macrophages or dendritic cells [66]. The non-classical population is also called patrolling monocytes, as they adhere to endothelium and provide surveillance for blood vessel integrity. Through their expression of FcRs, CRs and TLRs and other PRRs these cells can react to various conditions through cytokine production. cytotoxic and phagocytic activity, migration, adhesion, differentiation into macrophage, growth and induction of apoptosis (Figure 5) [62]. The role of FcγRs has been investigated in monocytes in the aspect of the inflammatory pathway NF-κB activation as well [67], with the focus on FcγRIIa and FcγRI. These results showed that these receptors can activate the NF-κB pathway and that the activation is independent of phagocytosis [68]. Yet the signaling pathway and the separate role of these receptors remained unclear. Differences between FcγRIIa and FcγRI ITAMs have been shown and FcγRIIa turned out to be a weaker activator of cytokine production and it was also shown that it can use the FcRγ chain for signaling [69]. Moreover while FcγRIIa has to be in lipid rafts for ITAM phosphorylation [70], thus requires inside out signaling triggered by the activation of the cell [71]. Raft association was as well shown to be necessary for NF-κB activation by FcγRIIa [72]. FcγRI is associated with the lipid rafts without activation[73]. The FcRγ chain mediated signaling of FcγRIII and FcγRI though CARD9 was shown to result in NF-κB activation in myeloid cells [74], thus presumably this pathway is also accountable for the NF-κB activation in the monocytes. The dominance of FcγRI in triggering inflammatory response has been shown in monocytes [75].

17

Figure 5. Overview of the activating cell-surface innate immune receptors, their signaling and functions. Receptors are grouped based on their signaling. TLR-MyD88, FcR y chain, DAP12 mediated receptors, and ITAM mediated receptors are shown to emphasize the correspondence between these pathways. Src Family kinases (SFK) initi- ate the signaling by phosphorylating the tyrosines in the ITAMs, thus activating Syk, a central kinase that activates kinases activating lipid signaling(PLCγ, PI3-K), scaffold- ing proteins(SLP76, LAT), CARD9, and G proteins(Ras, RhoGTPase). These pathways control the Ca2+ signal activation of downstream kinases that will orchestrate the cel- lular response. Based on [76], [77] and [78]

The monocytoid cell line U937 [79] was originally established from a histiocytic lymphoma. U937 cells have been widely applied as a model for monocyte in both developmental and functional studies, with currently over 10000 journal articles the literature. U937 express FcγRI, FcεRI [80], the R131 variant of FcγRIIa [81] while FcγIIb is not expressed on this cell line [75]. The LPS mediated NF-κB activation have been demonstrated in U937 on RNA level and was shown to affect more than 300 genes [82].

1.4 Autoantibodies in autoimmune diseases – rheumatoid arthritis and systemic lupus erythematosus

In a subgroup of autoimmune diseases, despite multiple checkpoints during antibody selection process of B-cells, self-antigen specific, autoantibodies are produced by plasma cells [83]. Both genetic [84] and environmental factors: chemicals [85], air pollution [86], smoking [87] and UV radiation [88] may play a role in the pathophysiology of these diseases. Viral infections [89] can also trigger autoimmune

18 response. The etiology of these diseases is not yet completely understood. The diagnosis of these conditions is just as challenging as their treatment and new biomarkers for predictions for disease outcome, and activity and as well for efficiency of possible interventions are desired [90]. As in both SLE and RA autoantibodies against multiple autoantigens are present these parameters are involved in the diagnostic criteria of these diseases [91] and [92], respectively. In our work we tested and compared serum samples from healthy and RA or SLE patient donors.

RA is an autoimmune inflammatory disease affecting synovial joints. Starting with inflammation of the synovium it can eventually lead to irreversible destruction of the surrounding bones and cartilages. RA affects 1% of the population, and the woman to men ratio is around 2.5:1. High levels of inflammatory cytokines, such as TNF-alpha, IL-1 and IL-6 and autoantibodies against various autoantigens can be found in RA patients [93]. Testing for two serological markers: Rheumatoid factors and Anti- Citrullinated protein antibodies is used for the diagnosis of RA [94]. RFs are antibodies (IgA and IgM, next to IgG) recognizing antibodies mainly of the IgG class. In diagnostics their use is limited due to relatively low sensitivity next to high specificity [95]. RFs may be found in other cases: other autoimmune or infectious diseases and even in elderly healthy samples as well [96]. Moreover elevated RF levels can predict a higher risk of later onset of RA [97]. Both IgA and IgM RFs were reported to enhance the ACPA mediated immune response, with further increasing the production of inflammatory cytokines [98]. Citrullination is a post-translational modification where through deimination, arginine is converted into citrulline, in a process mediated by polypeptydil arginine deiminase enzymes. Members of the PAD enzyme family are expressed in various tissues and have tissue-specific roles [99]. In RA patients' synovial tissue PAD2 and PAD4 were found to be expressed, and were suggested to have a pathogenic role via citrullination of fibrin [100]. ACPA is raised and targeted against citrullinated protein targets. In the joints therefore as long as the high ACPA levels are present, immune-complexes will constantly form and cause chronic inflammation. This environment was shown to induce the differentiation of monocytes to osteoclasts and cause irreversible destruction of the surrounding bones. Therefore early recognition and diagnosis of RA is important to prevent abiding damage and ACPA positivity reportedly precede the onset of RA [101].

19

SLE is considered as the archetypal multi-system autoimmune disease, where breakdown of and thus an impaired B-cell selection combined with insufficient clearance of apoptotic cells and immune complexes is suggested as the underlying etiology in a process where the involvement of T cells, FDCs, DCs, moreover monocytes and macrophages was shown [102]. This disease affects many organs and tissues and the clinical picture is very heterogeneous, as well as the clinical course where flares and remissions are unpredictable. SLE is multi-factorial, as genetic [103] and environmental factors are involved. SLE serologically is among the most diverse autoimmune diseases with more than hundred auto-antibodies targeting different auto-antigens identified. Anti-nuclear antibodies were the first to be identified, these antibodies recognize for example DNA, histones and other nuclear antigens such as Sm antigen [104]. Other clinically applied tests determine levels of anti-C1q antibody [105]. Under normal immune homeostasis immune complexes are carried to the liver's Kuppfer cells by the circulating red blood cells [106]. Red blood cells can bind immune complexes that activated the complement system, through CR1 [107]. The IC burden and decreased expression and mutations in components of the complement system for example CR1 [108] and C1q, may further enhance the impaired IC clearance. Therefore mutation affecting these genes are considered as risk factors in development of SLE [109, 110] just as complement deficiencies [111]. Genetic association studies revealed further genes as susceptibility factors in SLE, and show that SLE follows polygenic heredity [110]. The increased production of immune-complex leads to their deposition in various organs and tissues thus to increased complement consumption further amplifying inflammation through recruitment of various cell types: monocytes and macrophages, neutrophil granulocytes, dendritic cells [102]. Phagocytosis of ICs by monocytes and macrophages were shown to be decreased in SLE patients, due to abnormal FcγR expression and results in prolonged activation by the ICs [112]. The activation of these cells further expands the inflammation and probably the number of autoantigens involved, through triggering adaptive immune response [113]. Different sub-types of the disease are established based on the affected organs, with the renal and neuropsychiatric lupus being the most characterized. In severe cases this condition can be life-threatening for example through renal failure [114]. More women are involved, with a women the men ratio being around 9:1, and involvement of sex hormones have also been demonstrated [115]. There is no single test for clear diagnosis of SLE

20 therefore patients are often misdiagnosed. It has been shown that autoantibodies just as in case of RA may be present long before the clinical diagnosis [116].

1.5 Protein microarrays

Microarrays allow simultaneous detection of multiple molecular binding events in a single reaction space, through spatial separation and coding of the reaction partners. This setting results in a microarray of compounds that are to be investigated. In a simple example our protein microarray consists of various antigens and control materials in a 9x9 matrix, with each antigen printed in triplicates, allowing us to investigate binding to 27 different materials. These microarrays are printed with the help of a robotic microarray printers onto glass slides covered with various substrates, detailed below. In the next step the investigated serum sample is added to the array, and binding takes place. To see the various binding events secondary, fluorescently labeled antibodies are applied with specific binding to the molecule we would like to determine such IgG or IgM specific to the printed antigen. The glass slide is then scanned by fluorescent scanner and the binding events are revealed.

The concept of multi-analyte immunoassays [117], the advantage of their combination with fluorescent labeling[118] and later the concept of “micro spots” [119] were proposed and established by Roger Ekins . Multi-analyte screening for clinical samples was first presented in 1989 in an enzyme linked immunoassay based system applying test cards with multiple test areas [120]. Later in 1992 fluorescence based multi-analyte detection was introduced [121]. The breakthrough arrived when the microarray technique was first introduced in the field of DNA hybridization development techniques. These investigations showed that the microarray format supports the parallel determination of multiple binding events from biological samples. These works applied fluorescent labeling of ligands and spatial setting to separate various reaction partners. The great advantage of the high density arrays, to measure multiple reactions from a tiny sample volume was finally demonstrated [122]. On the way to protein microarrays from their DNA counterparts, the next step was to print cDNA containing vector libraries onto the microarrays, to transfect bacteria and express various genes in them on the microarray [123] and was applied for gene expression [124] and antibody screening [125].

21

Finally the first printed protein microarray was presented in 2000 and was applied for screening protein-protein interactions, identifying kinase and small molecule targets [126]. This technique also supports screening and exploration of biomarkers and with multiple fluorescently labeled secondary antibodies. This versatility has been presented by a wide scale of applications to detect molecular interactions [127]. Importantly, antigen microarrays have been applied for the diagnosis of cancer, various autoimmune diseases, infections, allergy and to follow the efficacy of vaccination from blood samples [128]. Previously our group showed the detection of complement fragments on antigen microarrays [129].

Various experimental settings exist: the example mentioned above in the beginning of this section, depicts the forward setting; in this case the investigated sample incubated with the printed array of characterized ligands and interaction partners. The investigated samples can be printed as well and then in a single step with labeled ligands, antigens or antibodies can be detected after binding in the reverse setup. Also different sandwich settings can be applied to increase specificity of the detection, just as in the case of ELISA experiments [130].

Various immobilization techniques and coating materials have been developed. The solid support of most microarrays is glass microscopy slide. First of all, the glass slide itself can be activated and the generated epoxy aldehyde or amine groups can be used to covalently bind molecules through various functional groups. This approach results in a 2D microarray, and following blocking of the active groups the slides no longer can bind molecules and therefore offers low background. On other hand these coatings have a lower binding capacity compared to the next group, the 3D microarrays. In the case of 3D coating, for example nitrocellulose or polyacrylamide gel binds printed materials through non-specific adsorption [131]. While the high binding capacity in this case may partly compensate for it, the orientation of the bound materials is random. To overcome this issue, streptavidin coats can bind biotinylated molecules in a fixed orientation, or given orientation can be achieved by DNA-directed immobilization [132].

22

1.5.1 Cellular microarrays

Four main tracks of investigation established the field of cellular microarrays in the last 15 years. First the technique was introduced by Belov et al in 2001 to investigate how cell surface antigen specific antibodies bind various cell types to yield a CD antigen based phenotyping of leukocytes isolated from the blood of leukemia patient or healthy donors, and cell lines [133]. At first 60 CD antigen specific antibodies, then later 82 were printed onto nitrocellulose covered slides [134].The same technique can be applied for blood phenotyping [135] and neural stem cell capture as demonstrated [136]. Wu et al. analyzed T-cells based on their binding to an array of CD antigen specific antigens, and found difference between binding profile of T-cells from HIV + and HIV - donors [137]. Later the technique was applied establish a glycome profiling platform by applying different lectins on the microarrays, and tested different cell lines [138].

Soen et al opened the of analysis antigen specific T-cells' TCR clonality based on binding to MHC-peptide complex tetramers on microarrays in 2003 and showed how these arrays can be used for detection and identification of specific MHC-peptide complex reactive cells [139], and demonstrated that activation of the cells can be tested with the platform through application of Ca2+ sensing fluorescent probes. Since then the MHC-peptide method was further improved to investigate cytokine production of primary, in vitro primed T-cells. By printing costimulatory molecules together with the MHC-peptide complex further functional assays were developed [140], moreover the technique was combined with printing cytokine capture antibodies to the same spots to in situ detect production of IL-4, TNF-a and granzymes by various T-cell clones[141].

Another track of investigations, the transfected-cell microarrays, showed how basically printing a mixture of various proteins supporting cell attachment (ECM, Collagen, fibronectin, fibronectin) with cDNA (and as well siRNA and shRNA) to chips can be applied for on-site transfection of cells, to investigate which cDNAs affect the cellular property of interest [142]. These cell arrays are useful to investigate gene functions, with spatially encoding the DNA transfected into the cells [143].For broader view on the topic please take a look at our review.

23

1.5.2 Cellular SPR

Surface Plasmon Resonance (SPR) measurements can detect molecular interactions in a thin area over the surface of a thin layer of noble metal, the evanescent field. Since the first Biacore instrument was sold in 1990 [144], the versatility and power of the technique had been demonstrated by many applications [145]. A great advantage of the technique is that it is label-free: no modifications of the tested molecules are required for the measurements, thus molecules in their native state can be investigated. Aside from the typical measurements such as determination of affinity, specificity and binding kinetics, different applications in the field of SPR technology involving living eukaryotic cells had been introduced since 1999 when surface plasmon resonance microscopy was developed to monitor cell substrate interactions [146]. By their experimental setting in which cells are tested, two main approaches seem to trace out: in the first, where cell – substratum interactions are investigated, cellular ligands (e.g. cell surface antigen specific antibodies) are coupled covalently onto the sensor and cells are used as analytes. This approach was shown in detection of breast cancer cell lines [147] and T- and B-lymphocytes [148] .Later the technique was applied for typing of Red Blood Cells [149] and platelets [150] as well .

The second approach makes a further step: adherent cells are captured by antibodies or ECM proteins to form a monolayer on the sensor surface, then during incubation with any compound of interest morphology and apoptosis can be determined[151]. The technique has been applied to follow mast cell activation in response to allergens following cell sensitization by antibodies, and deducted from the angular shift change, the process can be followed in real-time [152] thus not only binding through specific ligand-analyte interactions, but spreading, migration and changing membrane characteristics can be monitored as well [153]. SPR was applied to detect antigen specific antibody production during on-chip antibody secretion by B-cell hybridomas [154].

24

2 Aims of the study

The aim of this study was to investigate how cells can be applied to detect immune-complexes. We hypothesized that cellular detection may hold advantages in this field, as myeloid cells with their set of receptors are the main effector cells reacting with immune complexes in vivo. As a model we picked U937 monocytoid cells and tested how components of immune complexes determine the cellular adhesion and how this adhesion differentiates healthy and autoimmune donors. Moreover we planned to modify the U937 cells to apply them as detectors for NF-κB translocation mediated inflammation. Finally we tested the adhesion based model for immune complex and IgG detection in a label-free manner through applying imaging SPR.

Our specific goals were as follows:

1. to compare U937 cell adhesion to ICs, and find correlations between antigen specific antibody isotypes and cell adhesion

2. to show that FcR based detection is susceptible to IgG glycosylation, removal of the core glycan will lower the cell binding to ICs

3. to generate a reporter U937 cell line to detect inflammatory activation of these cells, through stable transfection with an NF-κB reporter element GFP plasmid, selection and cloning of reporter cell line

4. to test the cell NF-κB reporter cell line activation induced GFP production by immunoglobulins and LPS

5. to demonstrate the application of u937 cells in label-free detection of immune complexes and immunoglobulins with imaging SPR

25

3 Materials and methods 3.1 Cell culture

Human histiocytic lymphoma cell line U-937 (ATCC CRL-1593.2) was cultured in RPMI-1640 medium (Gibco) supplemented with 10% FCS (complete medium), 2 mM glutamine, penicillin (100 U/ml) and streptomycin (100 μg/ml) and was maintained at 37°C in a humidified atmosphere of 5% carbon dioxide. Cell density was maintained between 5*105 – 1.5*106 /ml by adding fresh medium every 3-4 days.

3.2 Flow cytometry

Cell surface receptors were determined by applying the following antibodies: anti-human CD11b-PE (Dako; 2LPM19c; R0841), anti-human CD32-A647 (Biolegend; FUN-2; 303212), anti-human CD16-PE (Pharmingen; B73.1; 561313), anti-human CD64-FITC (Pharmingen; 10.1; 560970), anti-human CD35-FITC (Pharmingen; E11; 555452), anti-human CD14-APC (Santa Cruz; 61D3; sc-52457) and antihuman CD11c- A647 (Serotec; BU15; MCA2087) antibodies were used, while mouse IgG1-PE (Immunotools; PPV-06; 21275514), mouse IgG1-FITC (Dako; PPV-06; DAKGO1), mouse IgG1- A647 (Serotec; W3/25; MCA928A647) and mouse IgG2b-A647 (Biolegend; MPC-11; 400330) were applied as isotype controls. 3*105 cells per stain were centrifuged (250g, 7 minutes) then washed in FACS buffer, after centrifuging again the pellets were loosen and the antibodies were added and the tubes were gently mixed. This was followed by incubation for 20 miuntes on ice, covered from light. Finally cells were washed in FACS buffer again, before the measurement by FACSCalibur (Becton Dickinson). FCS Express 3 (De novo) was used to analyze the data; living cells were gated based on forward and side scatter.

3.3 Cell staining

Prior to use on microarrays U937 cells were stained with 4 µM celltracker Green CMFDA (5-Chloromethylfluorescein Diacetate, Life Technologies) according to the manufacturer’s protocol. Briefly cells were centrifuged (250g, 7 minutes), then RPMI

26 medium without FCS was added to result in 4*106 cells/ml concentration, and CMFDA was added. Cells were incubated on shaker covered from light, at 37°C for 30 minutes then centrifuged again, and resuspended at 106 cells/ml in complete medium and then were incubated again at 37°C for 30 minutes on shaker. Finally cells were centrifuged again before use and were resuspended at the desired concentration in complete medium.

3.4 Printing of protein microarrays nitrocellulose

Antibodies and antigens were printed onto homemade nitrocellulose-covered glass slides or FAST slides (Whatman) using a BioOdyssey Calligrapher miniarrayer (Bio-Rad). All materials were diluted in PBS. Immunoglobulin subclasses were printed 0.5 mg/ml, BSA 1 mg/ml, Staphylococcus aureus (all from Sigma) suspension in triplicates for experiments investigating binding and comparison with C3 and IgG signals, and effect of agitation and cell number. To determine dose dependence a five- fold serial dilution of four steps was printed starting at 0.5mg/ml from every subclass. Tetanus toxin was printed at 0.5 mg/ml in sixplicates. For SLE chips the following antigens were printed in sixplicates Ro (SSA) (Arotec; 0.374 mg/ml), dsDNA and ssDNA (Sigma; 0.2 mg/ml), C1q (Sigma; 1 mg/ml), pG (Sigma; 0.25 mg/ml) and human IgG (hIgG) (Sigma; 0.5 mg/ml). All dilutions were made PBS. HCP2 and VCP2 antigens were diluted in MilliQ water to 4 mg/ml and were printed by sciFlexarrayer S11 (Scienion AG, sciArraying service) in triplicates, then stored at 4°C in sealed bags.

3.5 Cellular microarray

Before use microarrays were rehydrated with PBS for 15 min. PBS and VBS (Veronal buffered saline) based buffers applied were as follows: to follow complement activation C-VBS [2.5 mM Ca2+ , 0.7 mM Mg2+, 0.05% Tween 20, 5% BSA (Bovine Serum Albumin), VBS] while to inhibit complement in E-VBS [25 mM EDTA, 5% BSA, 0.05% Tween 20, VBS ], following serum treatment washing buffer [0.05% Tween 20, PBS] was used, when cell were applied subarrays were also washed with PBS and empty RPMI-1640 medium. Secondary antibodies were diluted in dilution buffer [0.05% Tween 20, 5% BSA, 0.05% azide, PBS]. Serum treatment was applied for

27

1h, 37 °C on shaker with gentle agitation. To see the effect of serum incubation human serum diluted five times in C-VBS was used and cell or secondary antibodies were applied for detection, here IgG was detected with anti-human IgG-FITC (Invitrogen, 1:500 in dilution buffer) and C3 was detected with anti-human C3-Alexa- 647 (Cappel, 1:5000 in dilution buffer). The same anti-human IgG-FITC (Invitrogen, 1:500 in dilution buffer) was used in the dose dependence experiments. When no serum was applied empty serum dilution buffer was used, and arrays were washed with washing buffer. For cellular detection CMFDA stained U937 cells were added, with 106 cells in 100 µl RPMI + 10% FCS to each subarray and microarrays were incubated for 1h at 37°C. This was followed by a gentle wash by pipetting with PBS, microarray chambers were then disassembled and microarrays were dried under hood before scanning with Axon GenePix 4300A (Molecular Devices). This same protocol was used for all other experiments, unless otherwise stated (cell number, agitation).

To test the effect of masking the Fc part of the printed antibodies 100 µl of goat antihuman IgG-F(ab’)2 (Jackson) 1:100 in dilution buffer was incubated for 30 min on shaker prior to applying the cells, while control subarrays were incubated with dilution buffer only. In case of Staphylococcus aureus prior to masking the subarrays were incubated with 20% serum diluted 1:5 in E-VBS or C-VBS for 1 h 37°C in humidified chamber. For deglycosylation, following serum treatment with 1:125 diluted serum in E-VBS, for 1 hour, at 37°C in humid chamber on shaker, subarrays were washed with washing buffer, then with PBS and 80 U EndoS (IgGZero from Genovis) in 100 µl PBS or this buffer only were added to the subarrays. Slides were incubated for 24 h 37uC in humidified chamber. After washing anti-human IgGDyL649 (Jackson, 1:2500 diluted in diluting buffer) or cells were added. To detect SLE immune complexes serum samples were diluted 1:5 in E-VBS, here anti-human IgG-DL649 (Jackson, diluted 1:2500 in dilution buffer) and anti-human IgM-Cy3 (Jackson, diluted 1:3500 in dilution buffer) were used to detect antigen bound IgG and IgM, respectively.

3.6 Serum samples

SLE serum samples were taken at University of Pécs The study was approved by the national Scientific and Research Ethics Board (reference number 25563-0/2010- 1018EKU); written informed consent was obtained from each participant. Serum 28 samples were obtained by venipuncture and stored at -70°C until use. Control serum samples of subjects without known autoimmune conditions were selected from the repository of the Drug Research Center. SLE patients fulfilled the international criteria.

Serum samples from RA patients and control serum samples were obtained by venipuncture at University of Pécs and at University of Pisa and were stored at -20°C until use and tested with anti-CCP2 ELISA (InovaDiagnostics). The national ethics committee of Hungary and Italy gave their approval for conducting study with the following contract numbers, respectively: 24973-1/2012/EKU (658/PI/2012.), 45066/2012. RA patients fulfilled the international rheumatoid arthritis classification criteria of RA diagnosis. Heat-inactivation of serum samples was carried out at 56°C for 30 minutes.

3.7 Analysis of cellular microarray measurements

Scanned microarrays were analyzed with GenePix 7 pro (Axon) software. Further analysis was carried out using MS Excel. Faulty spots and subarrays were removed from analyses. On slides developed by secondary antibodies ‘find circular features’ option was used in GenePix 7 pro, then signal intensities were calculated by subtracting background from medians of the parallel signal intensities. For slides developed with cells, ‘find irregular features’ option was used in GenePix 7 pro, then the background median value of each feature was multiplied by the number of pixels of each feature and the resulting value was subtracted from the features total intensity value. In both cases medians of parallel features were calculated and were used for further statistical analyses. Fluorescence intensity data were normalized for interassay comparison following incubation of SLE-specific antigens with SLE and control sera. Assuming saturation on certain features (on printed human IgG in the case of detection by antihuman IgG and U937 cells, and on protein G in the case of detection by anti- human IgM) normalization factors that reflect alteration from the grand mean on these features were derived. All respective signals were adjusted using these normalization factors, resulting in identical normalized signals on the reference features.

29

3.8 Antibody detection in protein microarrays

Dried arrays were rinsed in PBS for 15 minutes before use, then 5-fold diluted serum, in 25mM EDTA, 0.05% Tween 20 and 5% BSA complemented PBS buffer was incubated at 37°C for 1 hour. Serum treated slides were washed with PBS containing 0.05% Tween 20 (PBS-T), then incubated in the mixture of 1:2500 diluted DyLight 488-conjugated F(ab’)2 fragment goat anti-human IgM (Jackson ImmunoResearch) and 1:2500 diluted DyLight 649-conjugated F(ab’)2 fragment goat anti-human IgG (Jackson ImmunoResearch) or 1:1000 diluted Cy5-conjugated F(ab’)2 fragment goat anti-human IgA (Jackson ImmunoResearch). Labeling with fluorescent antibodies was carried out at room temperature for 30 minutes in PBS containing 5% BSA and 0.05% Tween 20. After washing in PBS-T, arrays were dried and scanned by FLAIR Fluorescent Array Imaging Reader (Sensovation). Data were analyzed by Array-Reader software by Sensovation, Radolfzell, Germany. Signal intensities were calculated by subtracting local background from medians of the three parallel signal intensities. Negative signals values were clamped to arbitrary value 1.

3.9 HCP2 and VCP2 IgG subclass specific ELISA

To determine specific IgG subclasses levels HCP2 and VCP2 were coated in 96 well ELISA plates overnight, 4°C, with 5 μg/ml HCP2 and VCP2, diluted in carbonate- bicarbonate buffer pH 9.6 or PBS, respectively. After washing with PBS + 0.05% Tween-20 wells were blocked with PBS + 3% BSA for 1h room temperature (RT), samples were diluted 1:50 in PBS with 1% BSA and 0.05% Tween-20 and were incubated for 3h at on shaker, RT. Subclass specific biotinylated mouse anti-human antibodies (ahIgG1-b and 2b (BD Pharmingen), ahIgG3-b and 4-b (Southern Biotech)) were diluted 1:8000 in PBS-Tween and were incubated for 1h, RT. Biotinylated antibodies were detected with Streptavidin-HRPO (Sigma), diluted 1:5000 in PBS-

Tween, 1h, RT. After washing 1% TMB with 0.2% H2O2 in TMB buffer was added to each well. After 10 minutes 2N H2SO4 was added to each well to stop the reaction. Optical Density was recorded at 450 nm. Control measurements without serum samples were used to determine the background values, background subtracted values were analyzed.

30

3.10 Imaging Surface Plasmon Resonance Measurements

Measurement of U937 binding to IgG subclasses and on-sensor formed complexes was determined with IBIS MX96 (Ibis Technologies, Enschede, The Netherlands). Human IgG subclasses (Sigma) were coupled in 10 mM acetate buffer, pH 5.4 in replicates to pre-activated Easy2Spot P-type chips (Ssens, Enschede, The Netherlands) using a Continuous Flow Microspotter (Wasatch Microfluidics, Salt Lake City, USA). 100 nM BSA in 10 mM sodium acetate buffer pH 4, and acetate buffer were applied as negative controls. Multiple Antigenic Peptides (MAPs): Histone derived HCP2 [30], and Epstein-Barr virus derived VCP2 [31] (from Toscana Biomarkers Srl) were spotted at 10 mM and 1 mM spots in replicates in 10 mM acetate buffer, pH 5 and 10 mM MES buffer, pH 6.0 respectively with IgG1 and 3 in acetate buffer, pH 5.4 as positive controls. After coupling the sensor was blocked with 1 mg/ml BSA in 10 mM sodium acetate buffer, pH 4 + 0,075% Tween-80 and then with 100 mM ethanolamine, pH 8 with 0.075% Tween-80. 50 nM anti-kappa-chain specific antibody (mouse anti- human Kappa UNLB, Southern Biotech, clone SB81a) was used to determine equal spotting concentration of the human IgG subclasses, identified as 108 nM for IgG1, 125 nM for IgG2, 100 nM for IgG3 and 148 nM for IgG4. As system and sample buffer PBS + 0.1% BSA (PBS-BSA) was used. All materials including cells and serum samples were diluted in this buffer. For regeneration in between measurements 200 nM H3PO4 and 0.5% SDS was used in two cycles to regenerate the sensor surface. The temperature was set to 25°C. To test serum reactivity to the coupled MAPs, and the binding of U937-myeloid cell to the on-chip formed immune complexes first the baseline (0-300 sec) was set with system buffer. Then 100 μl 1:10 diluted heat-inactivated serum was injected and incubated for 15 minutes under back-and-forth flow in the sample association phase. In the dissociation phase with system buffer, under back and forth flow for 10 minutes, the on-sensor formed complexes were stabilized by removing weakly bound serum components. After dissociation U937 cells (200 μl of 2*106/ml in PBS-BSA) were injected onto the sensor surface and were incubated for 15 minutes without flow, thus the cells could sedimentate recognize and bind to their activating ligands. Throughout the measurements changes in angular-shift were recorded to all spots in real time, monitoring both reactivity of the sera and cells, in the serum association and the cell association phases of the measurement, respectively. Of each phase the last ten seconds were analyzed, medians of replicate values were calculated 31 and reference responses measured in the same cycle were subtracted. The data shows the mean of replicate resonance unit values measured in the last 10 seconds of each phase.

3.11 Statistics

Spearman rank correlations were calculated to compare IgG, IgM and U937 signals using Statistica AGA software. Mann-Whitney U test was used for assessing significance of differences between EndoS treated and untreated control values obtained by detection with U937cells and anti-human IgG and to compare the ability of IgG, IgM and U937 signals on various antigens to separate samples from SLE patients and healthy controls. Statistical differences between cell and antibody binding to IgG subclasses with and without serum treatment, and the effect of Fc region blocking on the adhesion of cells to IgG subclasses and Staphylococci, were assessed by pairwise comparisons of relevant groups using permutation tests. Values from the groups to be compared were randomly reassigned into two groups and the differences between the group means were calculated. The distribution of 5000 randomizations was drawn and the two-tailed p value corresponding to the real sample assignments was determined. The arithmetic mean of 50 such p values was accepted as the probability of a-error. Prior to permutation tests the Kruskal-Wallis analysis was performed to test if at least one group is different from the others. Values of p < 0.05 were considered significant.

To analyze SPR measurements Spearman rank correlations were calculated to compare IgG, IgA, IgM and IgG1-4 and U937 and serum association signals on MAPs. Statistical differences in serum binding and cell adhesion during and following incubation of IgG subclasses and MAPs with healthy and RA sera were assessed by pairwise comparisons of relevant groups using Mann-Whitney U test. To compare the results obtained by cell adhesion to MAPs following serum treatment and results of CCP2 tests and RA diagnosis by physician, Fisher’s exact test was used. Statistical tests were considered significant as follows: non-significant (n.s.) - p ≥ 0.05, * - p < 0.05, ** - p < 0.01 and *** - p < 0.001.

32

3.12 Generation and cloning of U937 transfectant B12

U937 cells were transfected as described by [155]. Briefly 10 µg pNFkB- PtGFP.2 (Xactagen, Shoreline, WA, USA) plasmid was added to 2*107 cells in 400 µl RPMI-1640, without supplements, in electroporation cuvette and the cells were electroporated at 960 µF and 0.17 kV. After selection with 1 mg/ml G418(Geneticin, Gibco), cells were plated and resistant clones were isolated on the basis GFP production triggered by LPS. Similarly iRFP (Addgene, Plasmid #31857) were transfected into U937. Following successfully cloning the pNFkB-PtGFP.2 positive cells, cells from the clone B12, were transfected with iRFP plasmid to generate double positive cells. Positive cells were sorted with FACS ARIA III (Becton Dickinson) based on expression of the constitutively expressed iRFP, or GFP upon activation with LPS.

3.13 In vitro activation of B12 cells by Immunoglobulins and LPS

To investigate the NF-κB reporter properties of B12 cells, cells were plated into tissue culture plates at 1*106 cells / ml resuspended in fresh medium. Various concentrations of LPS were added to the medium of the plated cells directly after adding them to the wells. Cells were incubated overnight at 37°C in a humidified atmosphere of 5% carbon dioxide. After incubation cells were removed from the wells to polystyrene FACS tubes, centrifuged and resuspended in FACS buffer and were analyzed with FACSCalibur (Becton Dickinson). Basically the same process was repeated for measurements of time dependence then with a fixed 10 µg/ml LPS concentration added at various time points, to finish at the same time to measure all at once.

IgG subclasses and IgA, IgM, hIgG (all from Sigma) and IVIG(Endobulin, Immuno Ag) were coated into wells of 96 well tissue culture plate separately at various concentrations all diluted in sterile filtered PBS, by overnight incubation at 4°C. Coating solutions were removed, and wells were washed with PBS and then with RPMI+10% FCS and cells were seeded just as in the previous measurements. To investigate the effect of masking IgG Fc part coated IgG subclasses were incubated with goat anti-human IgG-F(ab’)2 (Jackson) prior to addition of the cells. Cells were applied as in the previous measurements. The effect of IgG in solution was investigated by

33 adding the IgG subclasses to the culture medium , containing the cells after seeding them; to result in a 10 µg/ml solution in the wells otherwise the protocol was as previously described.

3.14 Testing microfluidic chambers

Tetanus toxin (0.2 mg/ml), Diphtheria toxin (0.5 mg/ml), Staphylococcus aureus (10 w/v %), human IgG3 (0.5 mg/ml), BSA (0.5 mg/ml), human IgG3 (0.5 mg/ml) (all from Sigma) and Measles virus (0.5 mg/ml, AbD Serotec) were printed in seven replicates to microarrays onto Hydrogel slides. After printing slides were incubated overnight RT, in humid chamber. Slides were blocked in 100 mM Tris for 1h, washed in PBS and rinsed in MilliQ Water and dried. PDMS blocks were oxygen plasma treated and mounted onto the microarrays to form the chambers. After PBS wash serum diluted 1:5 in 2.5 mM Ca2+ , 0.7 mM Mg2+ + 0.05% Tween 20, 5% BSA was incubated in humid chamber at 37°C followed by wash CMFDA stained or iRFP+ cells were resuspended from pellet and were added to the chamber to completely fill it(at high density). Following incubation in humid chamber at 37°C for 1h chambers were washed and cell were fixed with 1% Paraformaldehyde. Finally mask was removed and the slide was dried then scanned with Axon GenePix 4300A (Molecular Devices). Analysis was performed by GenePix 7 pro; here pixels containing cells were automatically counted using the find irregular feature threshold set to include all fluorescent cells.

34

4 Results

The results presented here follow our two accepted and one submitted research article in the field with the aim to prove that cellular detection of immune complexes have many advantages over classical secondary antibody based detection methods. We present our results on nitrocellulose based microarrays and imaging SPR, where both techniques are based on the adhesion of U937 cells to their receptor targets. And finally we show the characterization of the U937 cell line we generated by transfection of the cell line with a plasmid encoding NF-κB responsive element-GFP construct to present a cell line available to monitor NF-κB transcription factor driven activation of inflammatory pathways in response to immunoglobulins.

The results presented here are all my own work. Of important note subchapters 4.1.1 – 4.1.5 were already presented in my master thesis, therefore we cannot take them as new scientific results, yet the interpretation and full discussion of the data were conducted later. Also to give a complete picture I include these results as well.

4.1 Application of fluorescently labeled U937 cells in antigen and antibody microarrays

To add an extra dimension to protein microarray detection we first characterized the receptor expression of U937 cells with regard to receptors potentially recognizing IC components. We characterized the binding of these cells to IgG subclasses and Staphylococcus aureus, and compared these results, to those determined by classical secondary antibodies. We tested the dose dependence, and the effect of incubation on the cell adhesion to IgG subclasses. We blocked the Fc part of the antibodies to verify that this process is dependent on the FcγRs, and investigated the effect of complement binding on cell binding to Staphylococcus aureus. To show that the FcγR triggered cell adhesion is dependent on the glycosylation of the IgG Fc parts we removed the carbohydrate side-chains on-chip from antigen specific IgGs to investigate its role in the binding. Finally we tested the cells’ applicability in IC detection in an SLE model, following incubation of sera from healthy and SLE patient donors, and compared the cell binding results to those determined by secondary antibodies.

35

4.1.1 U937 cells are classical inflammatory monocytes based on cell surface receptor expression

We determined the FcR and CRs expression on U937 cells by flow cytometry (Figure 6), to see the expression of receptors that could mediate binding to IgG and complement containing immune complexes. We tested the binding of receptor specific antibodies and compared the results to isotype controls as negative controls for aspecific or FcR mediated binding of the antibodies. The antibody panel was as follows: CD16 (FcγRIII), CD32 (FcγRII), CD64 (FcγRI), CD35 (CR1), CD11b (CR3) and CD11c (CR4), and we found that from the FcRs CD32 and CD64 were expressed, but not CD16. Of note the CD32 specific antibody recognizes both CD32A and CD32B therefore we could not comment on intracellular signaling through CD32.

Figure 6. Cell surface receptor expression of U937 cells. Representative histograms of flow cytometric characterization of U937 cells with (A) fluorescently labelled anti- bodies recognizing FcγRs: CD16, CD32, CD64 and complement receptors: CD35, CD11b, CD11c and LPS coreceptor CD14 and (B) with vital staining of U937 cells with CMFDA. Background autofluorescence showed with gray fill, and isotype controls with dashed lines. 36

CR1, CR3 and CR4, receptors for complement fragments, were all expressed on U937. We also tested the U937 expression of the LPS coreceptor CD14, an important marker of monocyte classification, and found that CD14 was expressed on U937 (Figure 6A). Although classification of blood monocyte subsets is always relative, based on the receptor expression profile CD14+, CD16- we identified U937 cells as classical monocytes. Finally as our plan was to apply this cell line in microarray experiments, therefore we also tested Cytotracker green (CMFDA), a fluorescent vital dye, and found that it can homogenously stain the cells (Figure 6B).

4.1.2 U937 cells show unequal binding to hIgG subclasses compared to anti-hIgG antibody

The next step was to see if cells can bind to the printed ligands of their receptors in protein microarray experiments, with and without serum treatment of the subarrays. We printed IgG subclasses onto nitrocellulose covered slides, with Staphylococcus aureus as a positive control, as following serum treatment Staphylococcus aureus cell wall components are known to bind IgG non-specifically and as well activate the complement cascade. First we determined the binding profile of anti-human IgG and the U937 cells to the IgG subclasses without serum treatment. Following detection with the anti-human IgG we found that this secondary antibody binds the subclasses IgG1, IgG3 and IgG4 equally while showed increased binding to IgG2, despite the equal concentrations of printed hIgG subclasses. Only a slight background binding was observable on the Staphylococcus aureus spots. We then tested the binding of U937 cells to the same printed materials. Most importantly we saw that cells can bind to the printed materials on the microarray, and we found that U937 cells show the strongest binding to IgG3, and as well bind to IgG1 and IgG4, and show no binding to IgG2 and as such, significantly distinguished from the other IgG subclasses (Figure 7A).

We also tested the same microarrays with serum treatment; here we incubated 1:5 diluted sera on the microarray at 37 °C, to enable complement activation, prior to detection with complement C3 fragment specific antibodies or U937 cells. As expected we found that IgG4 shows a significantly weaker complement activating property compared to IgG1, IgG2 and IgG3. Following serum incubation the Staphylococcus aureus spots were also stained, due to complement and antibody fixing properties of its cell wall. The serum incubation of Staphylococcus aureus as well resulted in cell 37 binding probably due to same reasons, and also showed that cells are also capable of binding to serum derived molecules in our protein microarray system. Importantly the serum incubation, possibly due to the complement activation, also changed the binding profile of the cells to the IgG subclasses. Binding to IgG2 still showed the lowest binding, significantly lower compared to all other subclasses and as well the U937 cell binding to IgG1 significantly decreased compared to IgG3 and IgG4 (Figure 7B). All together these results showed us that cell binding differentiates IgG subclasses, in a manner suggesting that the U937 binding is indeed mediated by receptors that can bind ICs, FcγRs and CRs, based on differences between the printed molecules that cannot be detected by a single IgG specific antibody. Here we characterized cell binding based on the total fluorescent intensity of the bound cells.

Figure 7. Comparison of IgG subclass detection by second- ary antibodies and U937 cells. Part (A) shows detection IgG subclasses (IgG1-4) and Staph- ylococcus aureus (Staph.) with anti-human IgG(left) or by cell binding (right) without serum treatment, with representative fluorescent images of the cell binding taken at 5 µm resolu- tion, where even single cells are recognizable. The spot diameter is approximately 300 µm. (B) shows detection of IgG sub- classes following serum treat- ment detected by anti-human C3 specific antibody (left) and U937 cells(right), with the rep- resentative images below. Each spot in the diagram represents median values of the triplicates in each subarray. Following Kruskal–Wallis test, permuta- tion test was performed to as- sess significance of differences. Values of p<0.05 were consid- ered significant and were indi- cated as follows: *p<0.05; **p<0.01; ***p<0.001. RFI – Relative fluorescence intensity.

38

4.1.3 U937 cells bind to human IgG subclasses in a dose dependent manner

We next assessed U937 cell binding on microarrays with four steps of five-fold serial dilution of IgG subclasses printed onto the nitrocellulose based microarrays, starting at 0,5 mg/ml. We compared results of U937 cell binding to detection by anti- human IgG (Figure 8). Our results confirmed our previous impression of cell binding and further strengthened IgG3 as the most potent activator of cell binding, over IgG1 and IgG4. These results also confirmed that cell binding is dependent on the printed antibody concentration that results in the density of the printed antibody in the microarray spots. This strong binding to IgG3 was also observable as the binding curve on the first 3 dilution points showed a plateau and the binding decreased only at the fourth dilution point with 0,004mg/ml printed concentration. Of note U937 cells were more sensitive only in the case of detecting of IgG3 compared to the applied secondary anti-human IgG antibody. In case of IgG1 and IgG4 a decreased cell binding started already at the second dilution point, while binding to IgG2 remained very low.

Figure 8. U937 cell binding is determined by the density of the printed IgG sub- classes. U937 cell binding was determined on printed serial dilution of human IgG sub- classes compared to detection by anti-human IgG. Relative binding was calculated by taking the highest measured value for both detection methods as 100 % and other val- ues were expressed as their percentage to yield relative binding. Points on the diagram represent mean of four triplicates’ median with error bars representing coefficient of variation.

39

4.1.4 Cell density as a factor determining U937 cell binding to hIgG1-4

To further investigate factors influencing cell binding in protein microarray measurements we looked at role of cell number and shaking during the 60 minutes incubation of cells on the microarrays (Figure 9). Once again our results pointed out the importance of IgG3 during incubation with agitation. When cell numbers were limited to 105 cells per subarray this dominance was even more pronounced, as cell binding was only observable on IgG3. At 106 cells used, a binding weaker than that to IgG3 was found in the case of IgG1 and IgG4. Surprisingly without agitation IgG1 equaled the level of cell binding on IgG3, and as well spots of IgG4 and even IgG2 showed increased cell binding, yet cells bound to spots printed with negative control BSA. These results suggest that CD64, the high affinity receptor for IgG determines the binding of cells to IgG subclasses during agitation, while without agitation CD32 can as well take part. Also we cannot exclude further receptors, for example Type II FcRs in this interaction as the above mentioned FcγRs have a very low affinity, compared to the other subclasses, to IgG2.

Figure 9. U937 cells show preference of IgG3 over the other IgG subclasses. Shak- ing the microarray during incubation with the U937 cells directs cell binding towards the preferred targets, presumably offering a binding with higher affinity and/or by trig- gering cell activation and thus results a firm binding. This effect was even more pro- nounced with limiting the cell number. Without agitation the cells bound to IgG1 and IgG3 comparably next to a lower binding to IgG4, weak binding was found even to IgG2. Columns represent medians of RFI (Relative Fluorescence Intensity) values of four triplicates’ median, error bars indicate Standard Deviation.

40

4.1.5 U937 binding is mediated by IgG Fcγ part, while complement plays a modulatory role in binding to Staphylococcus aureus

To verify that the binding we observed was mediated by the Fc part of the IgG subclasses, prior to adding the cells to the subarrays, we incubated the printed antibodies with Goat F(ab’)2 Fragment of Anti-Human Fcγ chain specific antibody to block these parts of the antibodies (Fc Block). We hypothesized that these antibodies would bind to and thus mask the Fc part of the printed antibodies and thus decrease the binding of cells to the IgG subclasses. We found that indeed the Fc block reduced cell binding significantly to the IgG subclasses (Figure 10). To test the same principle with immune complexes we also performed Fc block following serum treatment of printed Staphylococcus aureus, yet in this case we also investigated the effect of EDTA, to see the effect of inhibiting complement activation.

Figure 10. Masking IgG Fc parts results in decreased cell binding. (A)Comparison of U937 cell binding to IgG subclasses, with and without masking the IgG subclasses prior to incubation with cells by F(ab’)2 Fragment of Goat Anti-Human Fcγ (Fc Block). (B) Comparison of cell binding to Staphylococcus aureus spots following serum treatment with or without Fc Block, in combination with the effect of complement inhibition by diluting the serum samples in EDTA containing buffer. Each marker in the diagram represents median values of the triplicates in each subarray. Permutation test was performed to assess significance of differences. Prior to permutation test Kruskal – Wallis test was performed when comparing the effect of the different treatments in case of Staphylococcus spots. Values of p<0.05 were considered significant and were indicated as follows: *p<0.05; **p<0.01; ***p<0.001. RFI – Relative Fluorescence Intensity

41

The roughly 100-fold decrease in RFI signals in case of IgG1, IgG3 IgG4 and serum treated Staphylococcus spots, indicate an almost complete block of adhesion. The inhibition of complement activation by EDTA in combination with the Fc block strengthened the decrease in cell binding to Staphylococcus spots. These results suggest that the cell binding we see is mediated by the FcγRs on U937 cells and that as well complement may also contribute as expected, yet to a much lower extent than the IgG Fc part, as alone EDTA did not reduced U937 binding to printed and serum treated Staphylococcus aureus spots.

4.1.6 Deglycosylation of IgG impairs cell binding to immune complexes

The N-linked glycosylation of IgG plays an important role in IgG FcγR interaction. To see the role of these carbohydrate side-chains in our cell binding experiments our plan was to deglycosylate IgG to test the effect of their removal in cell binding. In these experiments we printed tetanus toxin on the microarrays, and first screened serum samples for tetanus toxin specific IgG reactivity. In the next step we applied EndoS endoglycosydase (IgGZero) after serum treatment. This enzyme is an IgG specific endoglycosidase, and can remove the core glycan attached to the asparagine 297 of IgG by hydrolysis. We compared the effect of EndoS treatment by detecting serum reactivity based anti-human IgG and cell binding (Figure 11). Our results show that while EndoS treatment resulted in a significantly decreased cell binding, the detection by anti-human IgG was not affected. These results strengthened our view that cell binding to ICs can detect antibodies through FcγRs, and by sensing glycosylation, these measurements may improve the standard secondary antibody based detection methods in immune profiling.

42

Figure 11. The removal of IgG N-linked core glycan decreases U937 cell binding to immune complexes Following serum treatment of tetanus toxin spots printed onto nitrocellulose microarray, with a serum sample containing tetanus toxin specific IgG, we applied EndoS to hydrolyze the asparagine 297 attached core glycan from the antibody or buffer only treatment. The immune complexes were detected by U937 cells or anti-human IgG (ahIgG). Columns represent the medians of three subarrays’ RFI values. Untreated values were taken as 100% and values obtained following EndoS treatment were expressed as their percentage. Error bars indicate coefficient of variation between the subarrays. Mann-Whitney U test was performed, *** indicates p<0.001, n.s. – non-significant.

43

4.1.7 U937 cells can detect IgG containing ICs in protein microarrays

We saw that U937 cells can detect IgG containing immune complexes and also that they differentiate between IgG subclasses and as well their binding is affected by the N-linked glycan of IgG. Next we tested U937 binding to ICs formed on-chip, following incubation of healthy and SLE patient donor serum samples with autoantigens applied in SLE diagnosis. We tested serum samples from 16 SLE patients and 16 healthy donors on microarrays printed with SLE-specific autoantigen: C1q, dsDNA, Ro and ssDNA, next to IgG as positive control. We measured the IgG and IgM content of the formed ICs and as well detected U937 binding in separate measurements, and compared results of the healthy and SLE group with Mann-Whitney U tests (Figure 12). In case of all autoantigens we found significantly elevated U937 binding following treatment with SLE samples compared to the healthy group. We also compared the obtained results by correlation tests. Anti-human IgG and U937 cell binding showed significant correlation on all antigens, and IgG-IgM comparisons resulted in significant correlations as well, yet indicated weaker correlations. No correlation was found between cell binding and IgM signals (Figure 13 and Table 1).

Figure 12. Distribution C1q, dsDNA, Ro and ssDNA specific IgG, IgM and U937 cell binding of healthy and SLE patient donors. Each marker represents median values of sixplicates. Mann-Whitney U test was performed to assess statistical significance of differences. Values of p<0.05 were considered significant and were indicated as follows: *p<0.05; **p<0.01; ***p<0.001. (nSLE=16, ncontrol=16)

44

Figure 13. Correlation scatterplots of U937 cell binding and anti-human IgG or anti-human IgM on ssDNA, C1q, dsDNA and Ro following serum treatment. Relative fluorescence intestines were obtained by anti-human IgG or IgM and compared and U937 binding signals on the SLE autoantigens measured following incubation with sera from healthy or SLE patient donor.

45

Antigen Comparison R value Antigen Comparison R value

IgG - U937 0.7262 *** IgG - U937 0.3717 *

ssDNA IgM - U937 0.1873 C1q IgM - U937 -0.1957

IgG - IgM 0.4054 * IgG - IgM 0.2885

IgG - U937 0.6793 *** IgG - U937 0.6092 ***

Ro(SSA) IgM - U937 0.3215 dsDNA IgM - U937 0.1301

IgG - IgM 0.5447 ** IgG - IgM 0.4688 **

Table 1. Spearman rank correlation of results obtained by measurement of U937 cell, anti-human IgG and anti-human IgM binding on ssDNA, C1q, dsDNA and Ro following serum treatment of antigen microarrays. Significance of Spearman rank correlations were indicated as follows: *p<0.05;**p<0.01; ***p<0.001.

46

4.1.8 Microfluidic chambers supporting cellular microarray measurements

These previous results showed us that cellular detection offer many advantage, yet the technology needs further improvements to be widely applicable. In order to do so our group started into the development of microfluidic chambers to support cellular microarray measurements in cooperation with the BioMEMS group of Péter Fürjes at MTA-MFA-KFKI. I had the chance to test application of the prototype and briefly I present my experience with it in the figure below.

Figure 14. PDMS based microfluidic chambers for cellular protein microarray experiments. The microfluidic chambers (A) were fabricated by soft lithography in Polydimethylsiloxane (PDMS) using SU-8 epoxy based photoresist as molding replica by our collaboration partners. Immunoglobulin subclasses and antigens were spotted and coupled to N-hydroxysuccinimide covalently through amine reactive binding onto hydrogel-covered microarray slides, which were blocked afterwards. The chambers having height of 20 µm were precisely aligned to the positions of the microarray subarrays (B). In this system capillary forces are used to draw fluid into the microstructure. Exchange and removal of the different solutions and suspensions is carried out by applying filter paper to entrance at one side while replacing the solutions on the other side (C). We tested these chambers in cell binding experiments without serum to IgG subclasses (D and E Left panel) and to detect immune reactivity from 1:5 diluted serum samples to microbial antigens (E).

47

4.2 Application of U937 cell line to detect ICs in imaging SPR experiments

Through collaboration we had the chance to set up a new approach to detect cellular adhesion to ICs. Here the principles of an SPR measurement were applied, thus the binding of the cells can be detected without any labeling. Our goal was to verify our previous findings regarding U937 binding to IgG subclasses and immune complexes. In this proof of concept study we tested rheumatoid arthritis specific multiple antigenic peptides with healthy and RA patient samples. To yield further insight into cell binding we compared our results with microarray and ELISA experiments and looked for correlations between cell and serum binding with serological measurements. Cell binding in SPR measurements have been demonstrated before but not to ligands deposited on the sensor surface form biological samples.

4.2.1 U937 cells bind to human IgG1, IgG3 and IgG4 but not to IgG2

Similarly to the nitrocellulose based measurements our first step was to see the cell binding to the IgG subclasses. Here the sensor can be spotted covalently with up to 48 different ligands allowing the simultaneous detection of for example serum responses to antigens and cells to various ligands. The first step was to yield equal coupling of the IgG subclasses on the sensor surface. Starting from equimolar concentrations we determined the coupling efficiency of the subclasses, based on recognition by anti-human kappa chain specific antibody, as all subclasses we applied bear the kappa light chain. With the determined concentrations we coupled IgG subclasses to result in 100 response unit, when detected with anti-human kappa chain specific antibody and we saw no significant difference. As expected the IgG subclasses at tenfold dilution resulted in decreased binding, especially in case of IgG2, suggesting different binding configuration for this subclass at lower densities. The 1RU spots of IgG subclasses resulted in no binding by the anti-human kappa chain specific antibody. We then tested the cell binding to similarly spotted sensors and observed cell binding. During the incubation, the cells sedimantate on the sensor surface, thus the signals we measure derive from their active interactions with the coupled IgG subclasses, as no cell binding response was observed on empty or BSA coupled spots. Indicating that by this method the real-time interaction and dynamics of cell binding can be measured through changes in the angle-shift.

48

The incubation of U937 cells on the sensor array resulted in specific binding of the cells to spots with coupled IgG1, IgG3 and IgG4 when compared to BSA, that we followed based on the measured angle shift. IgG3 activated cell binding most effectively, as spots with coupled IgG3 diluted 10 and 100 fold resulted similar cell binding response to the non-diluted spots. IgG3 was followed by IgG1 and IgG4, while we observed very weak binding triggered by IgG2, just above BSA. On diluted spots of IgG1 and IgG4 the signals decreased in agreement with the lower coupling concentrations. In these measurements U937 cells detected the activating IgG subclasses even at the lowest coupled concentrations (Figure 15). During these measurements we also observed that cell binding to response declined through the repeated measurement cycles on the same sensor, while within a single cycle measurements were comparable, therefore we only used each sensor up to 14 cycles (Figure 16).

Figure 15. U937 cells binding to hIgG subclasses in imaging SPR measurements. U937 cells bind IgG1, IgG3 and IgG4, but not IgG2. (A) Equal coupling of IgG subclasses, determined with kappa chain-specific antibody. (B) Representative sensogram of IgG subclasses detection by U937 cells. (C) Cell binding response expressed as percentage of the highest signal in each cycle. Data in (A, C) shows the mean of three replicate runs with three spots per run. Resonance unit values evaluated shown were recorded in the last 10 seconds of the cell binding curves, marked with the dotted frame in (B). Columns represent means with error bars showing standard deviation. RU - Resonance unit

49

Figure 16. U937 cell binding to IgG1 and IgG3 in 22 cycles. Cell binding response results on triplicates of 100 nM spots of IgG1 and IgG3 was recorded in 22 cycles on the same sensor after each other. Each symbol shows results of a single measurement in a cycle, and bars mark the mean of the triplicates. Response expressed in resonance units (RU).

4.2.2 U937 cells detect ICs generated by the treatment of VCP2 with RA sera

The next step was to see if the U937 cells are capable of binding to on-chip generated ICs. First we had to set up serum binding to RA specific Multiple Antigenic Peptides. Here we used Histone Citrullinated Peptide 2 (HCP2) and Viral Citrullinated Peptide 2 (VCP2) as antigens, and incubated these peptides with 1:20 diluted heat- inactivated sera form Normal Healthy Serum (NHS) samples and samples from RA patient donors. Next to these peptides we also spotted hIgG1 and hIgG3 as positive controls spotted at 10 nM coupling concentrations, and PBS and BSA as negative controls. Serum samples were introduced the to the flow chamber, with the spotted sensor on the bottom, and incubation of the sera resulted in association of serum components to the spotted ligands. To remove those components that bound weakly to the spots, we applied a dissociation phase where empty buffer was added to the flow chamber where weakly bound serum components could dissociate. These association and dissociation curves showed typical antibody-antigen binding characteristics (Figure 17A). As we tested the serum samples we saw significantly elevated binding to MAPs when RA samples were incubated, on both peptides compared to the NHS group (Figure 17 D). Yet in case of HCP2 some NHS samples showed serum reactivity towards HCP2. We also saw elevated serum response towards IgG1 from RA samples but not in case of IgG3 (Figure 17C). As here we applied heat inactivated serum samples, complement deposition could not be the cause of this elevated binding and Rheumatoid Factors offered a more likely explanation. The next step to introduce the cells to the flow chamber and see their interaction with the on-chip formed ICs (Figure 17B).

50

Figure 17. U937 cell binding response detects VCP2 based ICs. (A) Representative binding curves recorded during the measurement, showing phases of the measurement in case of an NHS and an RA sample. (B) Schematic illustration of the measurement phases. (C) Serum response and (E) cell binding response measured on IgG1 and IgG3 (D) Serum response towards HCP2 and VCP2 and (F) cell binding response to the ICs. (G) Correlation scatterplots of serum and cell response measured VCP2 and HCP2. Statistical significance was calculated by Mann–Whitney U-test, Symbols mark show mean of triplicates. (C-F) and Spearman’s rank correlation coefficients (r) were calculated (G). Statistical significance was indicated as follows: *p<0.05;**p<0.01; ***p<0.001. Each symbol marks the mean of the replicate spots minus the PBS mean. nNHS=5; nRA=18. 51

We saw no significant difference in cell binding to IgG1 and IgG3 between the two groups (Figure 17E). While cell binding to HCP2 based ICs did not resulted in significant difference between the groups, the cell binding to VCP2 was significantly higher in the RA group (Figure 17F). Finally we looked at the correlation between serum and cell binding response and found no correlation in case of HCP2, while in case of VCP2 we found a significant moderate positive correlation(Figure 17G).

4.2.3 VCP2-specific IgA, IgG and IgG3 levels correlate with the U937 cell binding

To further analyze these RA serum samples we measured the HCP2 and VCP2 specific IgA, IgG and IgM composition by antigen microarray and the antigen specific IgG subclass composition by ELISA. We looked for correlations between the results of these measurements and the measured cell binding response to the complexes generated with same antigens and serum samples. We did not found any significant correlation in the case of HCP2 neither with the serum nor with the cell binding response. In the case of VCP2 we found significant positive correlations between serum response and antigen specific IgG and IgA levels and as well with the IgG1, IgG3 and IgG4 levels. Cell response results on VCP2 showed correlations with the measured antigen specific IgG, IgA and IgG3 levels (Figure 18-19). Moreover we compared the VCP2 and HCP2 specific IgG subclass levels with each other separately for the two antigens and found strong significant correlations in most of the comparisons (Table 2). We repeated this analysis with IgG, IgA and IgM levels and found strong significant correlations in all comparisons (Table 3).

52

Figure 18. VCP2 specific IgA and IgG levels show correlation with cell and serum binding response. Correlation scatterplots of serum (A) and cell binding (B) response on HCP2 and VCP2 with specific IgG, IgM and IgA levels of RA serum samples. Spearman’s rank correlation coefficients (r) were calculated). Statistical significance is indicated as follows: *p<0.05;**p<0.01; ***p<0.001. RU - resonance unit, AU - Arbitrary Unit, n.s. - non significant. Cell response on MAPs was expressed in ratio of cell response on hIgG1 10mM spots in the same cycle.

53

Figure 19. VCP2 specific IgG3 levels show correlation with both cell and serum binding response. Correlation scatterplots of serum (A) and cell binding (B) response on HCP2 and VCP2 with specific IgG subclass (IgG1-4) levels of RA serum samples. Spearman’s rank -correlation coefficients (r) were calculated). Statistical significance is indicated as follows: *p<0.05;**p<0.01; ***p<0.001. RU - resonance unit, AU - Arbitrary Unit, n.s. - non significant. Cell response on MAPs was expressed in ratio of cell response on hIgG1 10mM spots in the same cycle.

54

Table 2. MAP specific IgG isotypes show strong correlations in RA samples. Spearman rank correlation of VCP2 and HCP2 specific IgG subclasses in RA serum samples. Correlation coefficient (r) and p-value are shown in each cell, for the comparison indicated by the cell coordinates, for HCP2 (light grey) and VCP2 separately (dark grey). Correlations were considered significant at P<0.05.

Table 3. MAP specific Ig isotypes show strong correlations in RA samples. Spearman rank correlation of VCP2 and HCP2 specific IgG, IgA and IgM levels in RA serum samples. Correlation coefficient (r) and p-value are shown in each cell, for the comparison indicated by the cell coordinates, for HCP2 (light grey) and VCP2 separately (dark grey). Correlations were considered significant at P<0.05.

55

4.2.4 VCP2 -IC detection by U937 significantly associates with RA diagnosis

To see how cell binding in our iSPR measurements performs in diagnostic purposes we compared the cell binding results with results of CCP2 positivity tests and RA criteria based diagnosis of the donors. We categorized cell response results into positive and negative subsets by applying the highest cell binding response measured in a healthy sample as a cut-off value, resulting in VCP2/HCP2 U937 cell response positive and negative groups. We compared this differentiation to the subsets established by CCP2 ELISA and the RA Criteria based diagnosis. And summarized our data into a contingency table and to evaluate the associations we applied Fischer’s exact text (Table 5). U937 cell binding response on HCP2 following serum treatment showed no significant association, while comparisons with VCP2 resulted in significant association with CCP2 positivity and RA diagnosis as well. All together the VCP2 based measurement showed promising

Table 5. U937 cell binding response on VCP2 associates with CCP positivity and criteria based diagnosis of RA. Contingency table summarizing the associations between VCP2 and HCP2 based immune complex detection by U937 cells, CCP2 positivity and criteria based RA diagnosis positivity. Statistical significance was calculated by Fisher’s exact test. n.s. – non-significant

4.2.5 IgG subclass specific serum binding response is significantly elevated in samples from RA patient donors

We started our measurements with the goal to also compare serum binding responses following and without heat–inactivation (native), to investigate the role of complement. Early on we disapproved these measurements, as the complement

56 fragments can bind covalently to any spotted material and thus can influence the cell binding in every following measurement cycle. We later saw that even after heat- inactivation, the serum association response in the serum measurement cycles resulted in elevated binding from RA samples to IgG1 but not IgG3, thus we concluded that what we see is not the complement activation and covalent fragment binding that abolished serum response to IgG3. This thinking encouraged us to reevaluate our results where we applied heat-inactivated and native samples on a full panel of IgG subclasses (Figure 20). In these measurements IgG subclasses were spotted at 10 nM coupling concentration and heat inactivated and native serum samples from RA and healthy donors were incubated of the sensor surface. Here we observed significantly higher serum binding response from serum samples of RA donors to IgG1 IgG2 and IgG4, while no such binding occurred on IgG3.

Figure 20. RA samples show increased serum response towards IgG1, IgG2 and IgG4 and not to IgG3. Boxplots show serum binding response from NHS and RA serum samples are compared with and without heat-inactivation of the samples. Welch’s t-test was performed to assess statistical significance, p<0.05 was considered significant. (nNHS=12, nRA=20)

57

4.3 NF-κB reporter U937

We saw that cells bind to surface bound IgG or ICs. Yet we also sought to monitor the inflammatory activation of the cells to see how it relates to the binding of cells to IgG or ICs. As monocytes play an important role in inflammation we aimed to detect the NF-κB transcription factor translocation triggered activation of U937. Our plan was to stably transfect U937 cells with an NF-κB reporter vector construct. This vector contains Green Fluorescent Protein gene under the control of NF-κB Responsive Elements. As NF-κB translocation results in the activation of downstream genes the transfected U937 cell line would express GFP upon activation of this pathway. We stably transfected U937 with this vector by G-418 selection, cloned the positively selected cells, and performed basic characterization of their activation by LPS and human immunoglobulins. In the results below the pNF-κB-RE-GFP+U937 clone B12 was applied. Our original plan was to generate cells that express constitutively another fluorescent protein as well, that could be applied on microarrays to simultaneously detect binding and NF-κB activation. To achieve this we transfected B12 cell with a vector encoding infra-Red Fluorescent Protein and generated and cloned piRFP+pNF- κB-RE-GFP+U937 double positive cells (Figure 21). Unfortunately due to decreasing viability of the cells on microarrays during the long incubation and unavoidable contamination we could not use these cells as we originally planned. We also generated single positive piRFP+ clones.

Figure 21. Flow cytometry scatterplots of the generated clones after incubation with LPS 58

4.3.1 LPS triggers GFP production in B12 cells in a dose and time-dependent manner

To see if the selected B12 clone is indeed capable GFP production in response translocation of NF-κB we tested their activation by incubation with LPS added from five steps of a ten-fold serial dilution, starting at 100 µg/ml. We incubated the cells overnight and analyzed them by flow cytometry. Here we saw that the B12 cells can produce GFP in response to signals activating NF-κB translocation, as it can be detect by their increased fluorescence. The cells reached their maximal fluorescence at 10 µg/ml LPS, while incubation with 0.1 µg/ml LPS was similar to the buffer only controls. Activation with 1 µg/ml LPS was sufficient to trigger GFP production in B12 cells (Figure 22A). As we saw that activation by 10 µg/ml LPS can trigger maximal GFP production, we applied this concentration in the next step to determine the time dependency of this process. Here we saw that B12 cells produce GFP already after 4 hours of incubation and reached the maximum after 8 hours (Figure 22B).

Figure 22. Activation of B12 cells by LPS triggers time and dose dependent activation of NF-κB. Representative histograms of the flow cytometry analysis. (A) B12 cells show LPS dose dependent production of GFP through NF-κB activation. (B) NF- κB activation triggered by LPS can be detected after 4 hour incubation in B12 cells and reaches its maximum after 8 hours. NF-κB activation was followed by detecting GFP production.

59

4.3.2 FcγR binding of coated IgG subclasses, hIgG and IVIG activates GFP production in B12 cells in a dose dependent manner while IgA and IgM trigger no activation

As we saw that B12 cells produce GFP upon activation of the NF-κB pathway we moved on to investigate if the activation can be triggered by incubation of the cells on coated immunoglobulins (Figure 23). We applied four step 10-fold serial dilutions of IgG1-4, IgA, IgM, IVIG and hIgG as coat in 96 well tissue culture plates. Incubation on IgG1, IgG3, IgG4, hIgG and IVIG all resulted in NF-κB activation, while the other stimulations, IgG2, IgA and IgM did not, based on the measured GFP expression by flow cytometry. IgG3 and IgG4 were the most effective in triggering this response, here the 10 µg/ml coating concentrations resulted in a similar GFP production to that of the 100 µg/ml coat and as well only at this two case were the 1 µg/ml effective enough to trigger GFP production. Surprisingly IgG1 showed a weak response even at the highest coat concentration that was comparable to the effect of the 1 µg/ml IgG4 coat. IVIG and hIgG showed very similar effects, and could activate the GFP as effectively as IgG3 at the highest concentration yet at the 10 µg/ml the response already decreased.

Figure 23. NF-κB activation by human Immunoglobulins in B12 cells. Representative histograms of flow cytometry analysis show activation of GFP production triggered by overnight incubation on immunoglobulin coat with the concentrations indicated in the legend. NF-κB activation was followed by detecting GFP. 60

To verify that the cell activation was initiated by the FcγR aggregation on the cell surface, B12 cells were incubated with 10 µg/ml doses of IgG subclasses, yet now the IgG subclasses were added into cell culture media. None of the subclasses showed stimulation based on flow cytometry measurements (Figure 24). We also tested if masking the Fc fragments of the IgG subclasses could inhibit the activation, as did previously in our experiments regarding cell binding. Our results show that accessibility of the Fc fragments is essential in triggering NF-κB response by surface bound IgG (Figure 25).

Figure 24. Soluble IgG does not trigger NF-κB activation in B12 cells. Representative histograms of flow cytometry analysis show that IgG added in solution did not trigger NF-κB activation in B12 cells.

Figure 25. Masking of Fc fragment of IgG subclasses inhibits NF-κB activation in B12 cells. Representative histograms of flow cytometry analysis show that by masking the surface bound IgG subclasses with human IgG specific F (ab’) 2 antibody fragments, the NF-κB translocation triggered through FcγRs can be inhibited in B12 cells.

61

5 Discussion

Detection of antigen specific immunoglobulins is a great challenge in diagnosis of antibody mediated autoimmune diseases. Current approaches for the detection of antigen specific antibodies simply detect a single or few antigen specific antibody isotypes. With this approach the antibody effector functions remain elusive. To present a new approach we aimed to develop new techniques to detect cellular activation on antibodies and immune complexes. For our investigations, we picked U937 cell line as a model for monocytes.

First we characterized the U937 cell line in FACS measurements to determine the expression of receptors capable of recognizing immune complexes in addition to the important monocyte marker CD14, the coreceptor for LPS. Our results regarding expression of CD64 and CD32, and the lack of CD16 FcγRs were in agreement with the results in the literature [80]. We found that CD35 [156], CD11b and CD11c is also expressed, that was also known just as the expression of CD14 [157]. As our plan was to apply these cells in antigen microarray we had to find a reliable, vital fluorescent dye to detect cell adhesion to various compounds on the microarray. We choose to apply cytotracker green, and found this dye to result in a strong fluorescence we can easily detect the cells. In our first set of antigen microarray experiments we showed that cells can bind to immunoglobulins printed onto nitrocellulose, and in addition that they don’t bind to control Staphylococcus aureus spots. Here we found that in agreement with data published on binding affinity of the FcγRs expressed, the U937 cells differentiate between the IgG subclasses [158]. We found IgG3 as the most potent activator of cell binding, followed by the IgG1 and IgG4, while IgG2 triggered no adhesion of U937 cells. We found this observation very promising, as this already holds great promise for cellular detections, considering that the common antibody detections applied do not differentiate between the subclasses, and one can see from these results that treating IgG2 and IgG3 simply as “IgG” may mislead investigations and diagnosis.

The binding of phagocytes has been investigated in a similar approach, where neutrophil granulocyte binding was shown to coated IgG subclasses, and in agreement with our results these cells showed binding to IgG1 and IgG3, and showed very low binding of IgG2. IgG4 triggered no binding, yet the neutrophils applied in these experiments do not express FcγRI [159], which was shown to be major receptor for IgG4. In fact while the difference between high affinity FcγRI and the low affinity 62 receptors with the highest affinity (FcγRII or FcγRIII) is approximately 100-fold, in the case of IgG4 it is only around 10 fold in case of IgG1 and IgG3, as was determined by IgG binding to recombinant FcRs in SPR measurements. This could have resulted in the low binding to IgG4. Also it has to be noted that the studies determining binding of IgG to FcγRs applied multimeric IgG generated with anti-human IgG F(ab')2 fragments [46] or BSA-TNP based immune complexes. These measurements showed strikingly different binding profiles for the subclasses, as ICs containing the IgG subclasses all bound equally to FcγRI in the BSA-TNP based model. [160]. Therefore it is important to add that the low affinity of FcγRI does not necessarily mean the lack of binding, questioning the accentuated importance of binding affinity in FcR-Ig interactions. We also saw in our experiments with stationary incubation on IgG subclasses, that even IgG2 can as well trigger binding. Moreover both studies comparing all FcγRs' IgG subclass binding used a system where the FcγRs were expressed in CHO cells [46, 160]. Therefore the activation triggered in these cells cannot be taken into account, while our results suggest that the binding as an active process and mimics phagocytosis.

When we incubated the same antibody microarray prior to addition of cells with control serum samples to provide a source of immunoglobulins and complement proteins, the adhesion profiles changed. First of all the non-specific binding antibodies and complement to Staphylococcus aureus resulted in cell adhesion to these features. The non-specific binding of antibodies through Fc part of the antibodies to Staphylococcus aureus is due to the SpA and Sbi compounds found in the cell wall of this bacteria [161], and all three pathways of complement may as well be activated by Staphylococcus aureus [162-164]. More surprisingly the cellular adhesion to the IgG subclasses changed as well following serum treatment. Our results show that in agreement with previous observations; IgG4 activated the complemented based on C3 fragment deposition, to a much lesser extent when compared to the other IgG subclasses [165]. One would expect the complement to strengthen the cellular binding as this way more activating receptors could contribute to cell adhesion. Yet we only saw this amelioration in case of IgG2, where without serum no cell binding was detected, the binding of complement to IgG2 helped U937 spots. On the other hand we saw decreased binding to IgG1 while to IgG3 the binding remained on the same level. IgG4 was the most surprising as cell binding levels on these spots equaled the level of IgG3. We accounted these changes for complement deposition and thus impaired accessibility

63 of the IgG molecules by FcγRs. This masking effect of complement deposition has been shown to decrease respiratory burst activity of neutrophil granulocytes[166] and as well ADCC function of NK cells[167] as complement fragment deposition was shown to be capable of concealing IgG and C1q [168].

Figure 26. Overview of molecular interaction sites on IgG. (1) Rheumatoid Factors,

FcRn, protein G and protein A share binding sites on IgG in the region between CH2 and CH3 domains [169]. (2) MBL can bind to the exposed mannose in agalactosyl N- linked glycans (3) C1q binds to the outer sides of CH2 domain close to the hinge region.

(4) FcγRs recognize the interface of the inner side of CH2 and the hinge region [170].

(5) Covalent binding of C3b fragments was first shown in the CH1 domain, later C3b binding to other regions of IgG was demonstrated, showing that C3b may bind to other domains as well [171]. (6) Bacterial protein L binds to the VL domain of IgG [172]. (7)

The antigen binding sites of IgG are located at the interface of VH and VL domains.

The cell binding activity triggered by the different printed IgG subclasses suggested that the binding we see is due to recognition of IgGs by the U937 cells FcγRs. The preference of IgG3 and the similar biding to IgG1 and IgG4 was strengthen by our

64 experiments showing the dose, cell number and agitation dependency of the cell binding. To show that it is the Fc part of the IgG molecules that is responsible for the binding we masked their fc parts with fab fragments recognizing hIgG-Fc. This treatment completely abolished the cell binding to the printed IgGs, thus we concluded that the binding we see here is FcγR mediated. As the glycosylation is known to influence the FcR-Ig interactions [32], we next investigated the role of carbohydrate side chains. We found in agreement with the literature the EndoS treatment of the antigen specific antibodies decreased the cell binding. EndoS an endoglycosidase derived from Streptococcus pyogenes [173], hydrolyses the core glycan off the antibodies and therefore weakens the FcγR IgG interactions which in our case results in decreased binding to immune complexes. Glycosylation of the antibodies is important aspect that may affect IgG-FcγR interactions. Basically the N-linked glycosylation of IgG serves structural roles as was shown to correlate with the structural integrity of IgG Fc part, especially in the lower hinge region, the binding site for FcγRs [174, 175]. Recently the binding of IgG1 was investigated to FcγRI, FcγRIIa, FcγRIIb and FcγRIIIa in SPR measurements and the binding activity of glycovariants (Figure 27.) was shown to differ compared to the wild type control (mixture of glycovariants) in case of various tested glycovariants. While deglycosylation completely abolished the binding of IgG1 to FcγRIIa, FcγRIIb and FcγRIIIa, on FcγRI the binding partly remained (60%), showing similarity with our findings. In general the binding to FcγRI was only slightly affected by the various glycan structures, similarly to FcγRIIIa. FcγRIIa showed decreased binding of hypogalactosylated (G0, G1), and slightly increased binding of the galatosylated (G2) variants, while FcγRIIb showed only the increased binding of the G2 variants [176]. The aforementioned BSA-TNP based IC model also investigated the role of EndoS treatment and showed similar results. The core fucose was not investigated in this study yet afucosylated antibodies were suggested to play an inflammatory role in various autoimmune diseases [177].

65

Figure 27. Schematic structure of the Asn297 N-linked Glycan and the glycovariants. The core glycan heptamer is shown in the bracket. Hydrolysis activity on endoglycosidase EndoS is shown. The basic glycoforms number of Galactose residues in the glycan (G0, G1 and G2). GlcNAc - N-acetylglucosamine, Fuc - fucose; Man - mannose, Gal -galactose, Neu5Ac - N-acetyl neuraminic acid. Based on [175].

To further test the cell binding we tested serum samples from healthy and SLE patient donors on a panel of SLE autoantigens. C1q, dsDNA, Sm and ssDNA are all applied in SLE diagnostic tests as autoantigens [178]. We tested immune complexes with anti-human IgG and IgM and as well with U937 cells and found that almost all antigens with any detection method was capable of significantly discriminating the healthy and the SLE group. We also investigated the correlations and pairwise comparisons showed that the U937 cells correlated strongly in most cases with IgG, and not with IgM, despite the correlation between IgG and IgM. Our results showed that cells with Fc receptors and CRs can be applied in multiplex microarray measurements for the detection of immunological reactivity following serum treatment of various antigens. These results opened new directions. First of all we aimed to develop an experimental system to support the cell binding experiment standardization, and after unsuccessful endeavors we turned to the application microfluidic chambers, where through capillarity, movement of fluids is under strict control. Yet in this thesis we go on to cellular SPR based immune-complex detection and NF-κB reporter cells.

To verify our findings we further explored cellular binding yet in the next experiments in a label-free multiplex system, the imaging SPR. Therefore the principles applied here remained the same as before. The main difference was the detection, as iSPR simply detects the mass accumulation in real time in a thin volume of the reaction chamber just above the surface on which the investigated substrates are coupled. Label-

66 free, SPR based techniques are still in their infancy for cellular applications. Moreover to our understanding the technique has not been applied to detect immune-complexes before. First we looked at the cell binding to equally coupled IgG subclass samples. To do so we set the coupling concentration of each subclass to yield equal resonance units when detected by anti-human kappa chain specific antibody, as all coupled human IgG subclasses contain a kappa light chain. After verifying equal coupling we looked at binding of U937 cells on the coupled IgG subclass spots, and once again we observed the dominance of IgG3 over IgG1 and IgG4, that showed close to equal properties in triggering cellular binding. Moreover compared to anti-human IgG the cells showed great sensitivity and detected the coupled antibodies at a low coupling concentration yet from these measurements it became obvious that with every cycle the binding in general decreases. This could be due to decreased viability of the cells during the long suboptimal incubation, as the cells were loaded into 96 well plates, and sealed for every cycle in separate wells. Therefore the cells in for example 10th cycle were already incubated for 200 minutes prior to the measurement at high density at 25 °C that could have easily changed their binding characteristics and viability. To compensate for this effect we decided to normalize immune-complex measurements with cell binding results measured on coupled immunoglobulins in the same cycle.

The next step was to test the binding of serum components to RA diagnostic VCP2 [179] and HCP2 [180] peptides. Both VCP2 and HCP2 are multi antigenic peptides (MAPs) developed by Miglorini and Rovero. MAPs are simply stand for a lysine based dendrimer, where the antigenic peptides are covalently bound onto the amine groups in the side chains of lysine. While VCP2 is an Epstein Barr virus derived peptide, HCP2 is based on histone sequences. The testing of autoimmune sera in SPR based system have been demonstrated earlier applying citrullinated peptides and proteins and their arginine containing versions as controls [181], and as well for VCP2 [182]. In agreement with these studies, when we compared healthy an RA patient samples we observed binding of serum components to the MAPs. We found that both peptides offer a significant discrimination between the healthy and the RA group. Moreover we saw significant serum binding response on IgG1 when incubated with RA serum samples. As we used heat inactivated serum samples, we can exclude that the complement activation caused the serum binding. Rheumatoid factor binding offers the most probable explanation. Earlier studies investigating IgM RF specificity support this

67 explanation as IgG1 specific IgM RF can be found frequently in RA serum samples while the occurrence of IgG3 specific IgM RF is minimal [183, 184]. Our results regarding binding of heat-inactivated and native sera to spotted IgG subclasses showed good agreement with this data. Yet other investigators found substantial IgG3 reactivity in serum RF of IgG isotype, however only the level of IgG1 and IgG2 specific RF was elevated in RA samples compared to healthy samples [185]. As of note the serum RF and that from synovium shows significant difference in reactivity [186], to IgG3 as well, as the synovium derived monoclonal RFs were shown to bind IgG3 in an dependent manner [187].

The next step was to see how U937 cells bind to the on-chip formed immune- complexes. After inserting the sensor with the coupled antigens and immunoglobulins the measurement had three steps: first the serum association took place, here various molecules from the sera reacted with the substrates, this was followed by a dissociation phase to remove weakly binding molecules so that only those forming the immune- complexes remain, and finally the U937 cells were introduced to the reaction chamber and the cells bound to their targets. It is important to emphasize that as the sensor is situated in the bottom of the reaction chamber and that during the cellular binding the flow is off thus all cells sedimantate onto the sensor surface. This suggests that with this setup we only see the cells that actively bind to the coupled molecules. From binding response curves of the three phases we analyzed the end of the serum association and the cell association phases to extract the binding response results. Here we found significant low positive correlation between serum binding and the following cell binding in case of VCP2 as expected. In case of HCP2 the cell binding lost discriminating power, and no significant correlation was found between the serum and the cell binding. To resolve this, we looked into the antigen specific antibody composition of the RA samples. With ELISA experiments we determined the IgG antigen specific IgG subclass levels in the samples and with antigen microarray the antigen specific IgG, IgA and IgM levels for VCP2 and HCP2. We looked for correlations between these results and the cellular binding. Significant moderate positive correlations were found between VCP2 specific IgG, IgA and IgG3 and the cellular binding, yet none in case of HCP2. We found it controversial that HCP2 separates healthy and RA samples based on serum response but not by cell response. Moreover no correlations were found between serum response and any other measured

68 parameter in case of HCP2. As an explanation we could only think about the role of high concentrations of low affinity IgM in the samples that we probably removed by washing in the microarray experiments, and through the dissociation in the iSPR measurements, yet during the serum association these antibodies showed up. IgM might indicate ongoing immune response triggered by constant exposure to the autoantigen[188] and as well could be caused by natural low affinity antibodies [189]. In case of VCP2 IgG1, IgG3, IgG4, IgG, and IgA all showed significant positive correlations with the measured serum binding response, and IgA, IgG and IgG3 with cell binding response on the formed complexes. The presence of autoantigen specific IgG subclasses has been investigated earlier in RA. Starting with Anti–keratin antibodies and later studies regarding ACPA isotypes showed IgG1 to be the most common isotype, followed by IgG4 and IgG3, but no contribution of IgG2 was reported [190]. Moreover we found significant strong correlation between levels of IgG, IgA and IgM and as well between the IgG ACPA subclasses suggesting that there is no single isotype associated with RA. However, a study found that the presence of five or more ACPA isotypes is associated with accelerated disease progression [191]. ACPA may occur as well in samples from healthy donors, most importantly it was found in samples of healthy relatives of RA patients; however it is restricted to IgG1 and IgA [192]. RF, ACPA and anti carbamylated antibodies were shown to be present in serum samples before the disease onset, therefore testing for these antibodies is of great importance to prevent and stop progression of the disease [101, 193]. ACPA IgA, IgM and IgG were all shown to bear clinical relevance [191]. Finally we found that cell response on VCP2 following serum incubation considerably match with CCP2 ELISA based and international RA criteria based diagnosis, demonstrating the potential use of this proof of concept approach.

Monocytes have been shown to play multiple roles in RA, as producers of inflammatory cytokines [194], and as osteoclast precursors upon activation by ACPA [195], and by their interplay with CD4+ T-cells, another cell type suggested to be central in RA [196]. Therefore understanding their activation and differentiation is of great importance. To investigate the pro-inflammatory activating potential of immune complexes we generated an NF-κB reporter cell line where upon translocation of NF-κB family member molecules from the cytoplasm to the nucleus, they bind to NF-κB Responsive Element in the DNA and activate transcription of the downstream genes. In

69 this reporter construct this GFP is encoded, therefore upon NF-κB activation we can measure the GFP production with fluorescent methods, this approach enables the monitoring of de novo protein synthesis. Similar constructs were applied earlier, yet in those luciferase enzyme were encoded downstream of the REs. NF-κB was originally described in relation to B-cell antibody production, yet since then its importance had been demonstrated in inflammatory cytokine production where it is considered to play a key role [197]. Activation of NF-κB pathway through PRRs especially TLRs binding microbial products[198] serves as an important switch to alert innate immune reactions. The activation of NF-κB pathway by LPS in monocytes and U937 cells has been demonstrated earlier [199, 200]. CD14 and TLR4, both expressed on U937 [201], are required for the activation [202]. We demonstrated that upon incubation with LPS the transfected cells produce GFP in a time and dose dependent manner, to verify mechanism of the reporter cell line.

In our work we generated and characterized basic effects of Ig isotypes on of NF-κB pathway activation. Surprisingly we found an activation profile different from that of cellular binding experiments, most importantly IgG4 showed an activating potential comparable to IgG3, while IgG1 only weakly activated GFP production. IgG2 triggered no activation at all, here it has to be noted that the FcγRIIa variant R131 expressed on U937 cells [75] has a very low affinity to IgG2 [27], and it is yet to answer whether the FcγRIIa H131 the major receptor for IgG2 would trigger activation . In agreement with our results IgG1, the most frequently applied isotype among therapeutic antibodies, does not trigger cytokine production and its mechanism of action described suggest the involvement of ADCC and CDC as effector mechanisms [203]. The effect of IgG4 is more surprising, as until recently has been described as a neutralizing antibody, disassembly of the IgG4 results in monovalent half antibodies, which therefore does not trigger effector functions. Recombination of the IgG4 through Fab arm exchange with other IgG4 molecules is also possible and results in bi-specific antibodies that can trigger FcR mediated effector functions [204]. However in IgG4 related disease the local production of IgG4 by plasma cells has been demonstrated, suggesting an inflammatory role for this isotype [205]. To verify that the activation is Fc dependent and the multimerization of IgG molecules, in our case by coupling them onto a surface, is necessary for the activation we showed that incubation with human IgG Fc specific F(ab’)2 antibodies the GFP production is inhibited, and as well the soluble

70 monomers of IgG subclasses did not activated the reporter cells. We also investigated the possible role of IgM, a receptor for IgM was reported to be expressed on monocytes, and described to play role in phagocytosis, yet U937 does not express this receptor. Therefore no activation was triggered by IgM just as in case of colostral IgA, that is the dimer form of IgA connected with the J-chain for transport purposes. Later other tests were performed by our group that also showed a role for IgA to enhance the response in combination with IgG, this activation most probably involves inside-out signaling of FcαRI, the main receptor for IgA on U937. And finally these cells were tested on ACPA based immune complexes and showed activation in sera containing both IgG and IgA ACPA. As these experiments were performed by my colleague Csilla Kecse-Nagy I set aside from their detailed presentation.

The NF-κB activating properties of immune complexes or crosslinking of the FcRs were demonstrated earlier. The FcR mediated activation of NF-κB pathway has also been described in monocytes through FcγRI and FcγRII [41, 68]. Various tests have been developed to investigate the cytokine production of monocytes triggered by ACPA based immune-complexes. In these experiments citrullinated peptides were coated, and incubated with serum samples from healthy and RA patient donors, and then monocytes or macrophages were incubated on these immune-complexes. Pro-inflammatory cytokines were measured upon activation, and monocytes were shown to produce TNF- a, IL-1, IL-6, important inflammatory markers characteristic of RA, that are known as to be under the control of NF-κB translocation [206, 207]. While many of these works suggest the cooperation TLR and FcRs [208], it has been demonstrated that immunoglobulins can trigger TNF-a production without the citrullinated peptides [209]. These works established the role of FcR mediated inflammatory activation in RA. It was important to show that even monocytes from healthy donors produce the same amount TNF-a when activated by the ACPA immune-complexes [206], thus although copy-number variation of FcRs may be associated with RA [210], it is the autoantibodies that trigger the inflammation.

71

Figure 28. The possible role of monocytes in immune complex mediated autoimmune diseases. Once the self-immune tolerance is broken, for example by molecular mimicry the formation of ICs will recruit monocytes through chemotaxis. Due to the impaired IC clearance first frustrated phagocytosis of macrophages will cause tissue damage and further recruitment of effector cells later monocytes can as well differentiate into imDCs, possibly enhancing both macrophage and antibody secreting cell functions through various t helper populations. Due to the highly inflammatory milieu bystander activation of self-reactive clones and spreading may occur and increase the number of targeted autoantigens. Based on [211] and[212]. mo – monocyte, M – Macrophage, APC – Antigen Presenting Cell, T – T-cell, B – B-cell, Th1, Th2 - T helper 1 and 2

72

Furthermore FcγRI expression on monocytes increases during systemic inflammation in SLE [213], and was shown to play an inflammatory role in humanized mice models [214]. Yet only speculation we suspect that this receptor contributes to the NF-κB activation we saw in B12 cells. Interestingly FcγRI was as well reported to play a central role in monocyte differentiation to immature dendritic cells (imDC). Similarly to our approach the effect of surface immobilized hIgG was investigated and after 2 days of incubation the increase of DC specific markers (CD80, CD86, and CD1b) was shown in CD14+CD16-CD64+ monocytes, and as well these imDCs T-cell activating property was shown[212]. This study indicates what may happen to monocytes in immune complex autoimmune diseases after such stimulation, suggesting an interesting contribution in triggering and sustaining the autoimmune response (Figure 28.).

Altogether our research investigated testing immunoglobulin effector functions by detecting cellular adhesion and inflammatory, NF-κB driven activation of monocytoid cell line U937. We consider these measurements to model frustrated phagocytosis during type 3 hypersensitivity reaction where due to impaired clearance or continuous generation of immune complexes results in inflammation and tissue damage. We demonstrated that cell binding and NF-κB activation are both mediated through FcγR, we found that these two processes only partially overlap. IgG3 triggered a strong response in both tests. IgG1 showed ability to trigger binding, yet was much weaker in triggering inflammatory activation. IgG4 activated cell binding well, and elicited the strongest NF-κB activation. Considering their ratios in serum, these results suggest that IgG1, the most abundant subclass targeting basically protein antigens and triggers phagocytosis through FcγR and by activating the complement system. IgG3 makes up about 6% of the total IgG and as our results show activates complement, cell binding and has strong inflammatory potential, While IgG4 is makes the 4% of all IgG, does not activate the complement, and may even disassemble in the sera, triggers considerable cell binding and has the strongest inflammatory potential. Moreover IgG4 can make bispecific antibodies. The lack of complement activation makes this antibody even more dependent on the FcγRs and thus high local concentration may trigger FcR dependent response. Yet this is only speculation it may make sense that IgG3 and IgG4 with such a great inflammatory potential compared to IgG1 make up only 10% of all IgGs. Unfortunately, considering the aforementioned reasons we cannot really speculate on IgG2. Yet we saw that even IgG2 can activate cell binding in stationary measurements,

73 suggesting that there are scenarios where these antibodies can as well trigger cell binding.

We consider cellular detection as a method with great potential. Our experiments show that these functional assays have many advantages and add a biological readout to antibody and immune-complex testing. Susceptibility to glycosylation, subclass discrimination, and potential incorporation of all extracellular signals recognizable by the cells could improve what we now simply describe for example as units or IgG levels in samples. Application of primary cells could lead to personalized detection that would even further refine these methods. Not only diagnostics, but development of therapeutic antibodies as well desires deeper understanding, and personalized testing of antibody effector functions.

74

Summary

Detection of antigen specific humoral immune response from biological samples is of great interest. Current methods for this purpose apply secondary antibodies produced by immunization of various animals. Looking at the complexity of humoral immune response one can see that through detection of each of its components alone would require many secondary antibodies and on the other hand even with those results determining the profile the next task would be to decipher what the given combination means for an organism in an immunological sense.

To provide a new approach our work aimed to detect immune-complexes with monocytoid cell line U937. These cells express receptors for immunoglobulins and complement components as well and therefore seemed to be an ideal model for cellular immune-complex detection. First we tested the U937cell binding on nitrocellulose based microarrays, and labeled the cells with Cytotracker green fluorescent dye. We showed that U937 cells can differentiate between human IgG subclasses in agreement with the expressed Fc receptors affinity towards these molecules. We also showed that the deposition of complement fragments play an important modulatory role in cell adhesion. According to our results the binding of U937 cells depends on the density of the printed Immunoglobulin subclasses. Our results showed that IgG3 is the most prominent activator of cell U937 cell binding. Yet the picture changed in stationary measurements where IgG1 was just as effective in triggering cell binding as IgG3 and even IgG2 activated it. We also looked at binding to serum treated Staphylococcus aureus as a model for immune-complexes and showed that cell can detect these complexes as well. We verified that the binding is mediated by the antibodies Fc-part by masking them with IgG specific F(ab’)2 fragments and as well by hydrolyzing the N- linked glycan of IgG with EndoS, as we saw that both treatments results in decreased cell binding. We applied the cells in systemic lupus erythematosus autoantigen microarrays and tested the cell compared the cell binding to signals obtained by detection of IgG and IgM following incubation of the microarray with control and SLE patient donors serum samples. We found that cellular signals can differentiate the healthy and the SLE group based on binding to on-microarray generated immune- complexes and that cell binding correlates with IgG signals. These results showed us that cells can be applied for this purpose. We went on to improve applicability by applying microfluidic chambers.

75

We also tested another platform, imaging SPR for the same purpose by applying U937 cells, and verified our results regarding cell binding to IgG subclasses. We tested this approach with rheumatoid arthritis specific peptides, and showed for the first time as a proof of concept that cells can recognize on-sensor surface deposited molecules in SPR measurements. We also found that cell binding correlates well with antigen specific IgG, IgG3 and IgA in the RA samples. While HCP2 only triggered serum binding in our measurements, the deposited antigen specific molecules triggered U937 cell binding as well. When compared to physician based RA diagnosis and RA diagnostic CCP2 ELISA U937 cell binding showed good agreement with both.

Finally to test the inflammatory activation we transfected U937 with a plasmid encoding an NF-κB Responsive Element with Green Fluorescent Protein in a downstream position, thus coupling inflammatory activation to fluorescently detectable protein production. After cloning the generated cell line we showed the time and dose dependent response to Lipopolysaccharide, well characterized inducer of the NF-κB pathway. We also tested the inflammatory activating potential of IgG subclasses with these cells and found that IgG3 and IgG4 were equally potent activators of this pathway; IgG1 was less effective as IgG2 triggered no activation, just as IgM and IgA. We also showed that activation is Fc part triggered as masking the coupled antibodies or adding them in solution did not trigger the same activation.

Altogether our results suggest that cellular detection in many ways can add to the conventional immune-complex detecting methods

76

Összefoglalás

Az antigén specifikus humorális immunválasz biológiai mintákból történő meghatározását nagy érdeklődés övezi. Az erre a célra kidolgozott jelenleg széleskörűen alkalmazott eljárások állatok immunizálásából származó másodlagos ellenanyagokon alapulnak. A humorális immunválaszban résztvevő molekulák komplexitása és azok hatásainak együttes vizsgálata tehát csak nagyszámú másodlagos ellenanyag segítségével volna pontosan megállapítható, és a teljes profil birtokában is komoly nehézséget jelenthetne annak értelmezése egy szervezet szempontjából.

Egy új megközelítést megcélozva, a monocitoid U937 sejtvonallal vizsgáltuk hogyan lehet sejtekkel detektálni a humorális immunválasz során létrejövő immun komplexeket. Az U937 sejtek egyaránt kifejeznek komplement és immunglobulin receptorokat, ezért úgy gondoltuk jó modelljei lehetnek a sejtes alapú immun komplex detektálásnak. Elsőként nitrocellulóz alapú fehérje mikro-mátrixokon teszteltük a fluoreszcensen, Cytotracker vitális festékkel jelölt sejteket. Ezekben a kísérletekben bemutattuk, hogy az U937 sejtek az általuk kifejezett FcγR-ok affinitásbeli különbségeivel összhangban megkülönböztetik a kinyomatott immunglobulin alosztályokat. Kimutattuk azt is, hogy a kikötődő komplement fragmentumok ezt a kikötődést befolyásolhatjuk mind pozitív mind negatív irányban. Eredményeink alapján a sejtek kikötődése függ a kinyomatott ellenanyag pontok sűrűségétől, ezek alapján kimondhattuk, hogy az U937 sejtek kikötődését leginkább az IgG3 aktiválja. További kísérleteink azonban megmutatták, hogy mindez további körülmények függvénye, az egyszerűen ülepített inkubáció során a sejtek az IgG1-el beborított pontokhoz az IgG3- hoz hasonlóan kötődtek ki továbbá ilyen körülmények között az IgG2-höz is kikötődtek. Megvizsgáltuk a sejtek kikötődést Staphylococcus aureus–hoz szérum inkubációt követően és láthattuk, hogy a sejtek kikötődtek ezekhez az immun komplexekhez. Igazoltuk, hogy a sejtek kikötődését az ellenanyagok Fc része közvetíti, hiszen IgG specifikus F(ab’)2 ellenanyag fragmentumokkal azt meg tudtuk gátolni. Továbbá az IgG ellenanyagok FcR felismerésben kulcsfontosságú szénhidrát oldalláncok EndoS kezeléssel való hidrolízisével bizonyítottuk, hogy a kikötődésben az FcγR-ok játszanak szerepet. Az U937 sejtekkel való immun-komplex detektálást SLE specifikus antigéneket hordozó mikro-mátrixon is kipróbáltuk, összehasonlítva egészséges és SLE- s donorokból származó szérumokkal történő inkubációt követően a sejtekkel, anti-

77 humán IgM és anti-humán IgG ellenanyagokkal kapott jeleket. Eredményeink alapján a sejtek elkülönítették a SLE-s és az egészséges csoportot, valamint megállapítottuk, hogy a sejtes detektálás jó korrelációt mutat az anti-humán IgG-vel történő detektálással. Eredményeink alapján tehát az U937 sejtek kikötődése alkalmas immun-komplexek detektálására. A módszer további fejlesztéséhez mikrofluidikai kamra rendszerek beállításába kezdtünk.

Az U937 sejtek kikötődését megvizsgáltuk egy felületi plazmon rezonancián alapuló méréssorozatban is és igazoltuk korábbi megfigyeléseinket az U937 sejtek IgG alosztályokhoz való kikötődését illetően. Az immun komplexekhez történő kikötődést ebben az esetben rheumatoid arthritis specifikus peptidek segítségével vizsgáltuk, és egy megvalósíthatósági tanulmányban bemutattuk, hogy az SPR mérések során a szenzorhoz kovalensen kapcsolt antigénekhez a szérum inkubáció során kikötődő szérum komponenseket a sejtek felismerik és azokhoz aktívan kikötődnek. Eredményeink alapján a sejtek kikötődése korrelációt mutat a szérummintákban található antigén specifikus IgA, IgG és IgG3 szintekkel. Végül az RA specifikus antigéneken mért eredményeinket összehasonlítva klinikai RA diagnózissal és CCP2 ELISA eredményekkel jó egyezést kaptunk.

Végül, hogy a sejtek gyulladási folyamatokban való szerepét vizsgáljuk, az U937 sejteket transzfektáltuk egy NF-κB reszponzív elemet követően zöld fluoreszcens fehérjét (GFP) kódoló plazmiddal, így csatolva a sejtek gyulladásos aktivációját egy fluoreszcens technikákkal detektálható fehérje termeléséhez. A sejtvonal szelekcióját és klónozást követően kimutattuk, hogy a létrehozott klón LPS hatására dózis- és időfüggő módon termeli a GFP-t. Megvizsgáltuk továbbá a szilárd hordozóra kikötött IgG alosztályok hatását a GFP termelésre, és eredményeink alapján az IgG3 és IgG4 alosztályok váltják ki az NF-κB útvonal aktivációját legnagyobb mértékben míg IgG1 gyengén, IgG2 pedig nem váltja azt ki. Sem az oldatban adott hasonló dózisú sem pedig az IgG specifikus F(ab’)2 ellenanyaggal kezelt IgG felszín nem váltottak ki aktivációt.

Összességében eredményeink azt mutatják, hogy a sejtes detektálás sok szempontból kiegészítheti a jelenleg az antigén specifikus humorális immunválasz jellemzésére szolgáló módszereket.

78

New scientific results

1. We showed that U937 cell binding to IgG is susceptible to IgG glycosylation, removal of the core glycan decreased the cell binding to ICs

2. We found that U937 cell binding correlates with the IgG and not the IgM content of the immune complexes

3. We generated a NF-κB responsive element GFP reporter U937 cell line through stable transfection, selection and cloning of the reporter cell line

4. We found that the NF-κB reporter cell line’s GFP production can be induced by LPS IgG3 and IgG4 most potently

5. We demonstrated the application of U937 cells in label-free detection of immune complexes and immunoglobulins with imaging SPR and introduced cell binding measurements to on-chip deposited immune-complexes in SPR

79

Acknowledgments

I would like to thank for Prof. Anna Erdei for letting me join the Immunology Program of the Biology Doctoral School at ELTE. And as well for her huge work as the Head of Department, to keep things going at the department during most of my years here.

I have to express my greatest gratitude to my supervisor Dr. József Prechl, for his end- less patience and support, for introducing me to the field I really enjoy, and for helping and pushing me all the way.

I thank Dr.Krisztián Papp for his practical help and teaching me the microarrays, statis- tics, reagents, R and may other things throughout the years.

I would like to thank Melinda Herbáth, Noémi Sándor, Mariann Kremlitzka, Eszter Lomen, Csilla Kecse-Nagy, Anita Orosz and Arthur Bentlage for their practical help in the lab.

I am grateful to all my colleagues whom I had the chance to collaborate with for their work, help, support and inspiring discussions.

I would like to thank Márta Pásztor, Gizella Gősi Sándorné and Árpád Mikesy for their technical assistance and all my colleagues at the Department of Immunology for form- ing such a great community.

I am grateful for my family and my friends for their patience and support.

80

List of publications

Publications connected to the thesis

Kecse-Nagy C*, Szittner Z*, Papp K, Hegyi Z, Rovero P, Migliorini P, Lóránd V, Homolya L, Prechl J. Characterization of NF-κB reporter U937 cells and their application for the detection of inflammatory immune-complexes. Under revision

Szittner Z, Bentlage AE, Rovero P, Migliorini P, Lóránd V, Prechl J, Vidarsson G. Label- free detection of immune complexes with myeloid cells Clin Exp Immunol. 2016 Mar 8.

Szittner Z, Papp K, Sándor N, Bajtay Z, Prechl J. Application of fluorescent monocytes for probing immune complexes on antigen microarrays. PLoS One. 2013 Sep 5;8(9):e72401 Papp K, Szittner Z, Prechl J. Life on a microarray: assessing live cell functions in a microarray format. Cell Mol Life Sci. 2012 Mar 4

Other Publications

Prechl J, Papp K, Hérincs Z, Péterfy H, Lóránd V, Szittner Z, Estonba A, Rovero P, Paolini I, Del Amo J, Uribarri M, Alcaro MC, Ruiz-Larrañaga O, Migliorini P, Czirják Serological and Genetic Evidence for Altered Complement System Functionality in Systemic Lupus Erythematosus: Findings of the GAPAID Consortium. PLoS One. 2016 Mar 7;11(3):e0150685.

Szarka E, Babos F, Magyar A, Huber K, Szittner Z, Papp K, Prechl J, Pozsgay J, Nagy G, Rojkovich B, Gáti T, Kelemen J, Baka Z, Brózik M, Pazár B, Poór G, Hudecz F, Sármay G. Recognition of new citrulline containing peptide epitopes by autoantibodies produced in vivo and in vitro by B cells of Rheumatoid arthritis patients. Immunology. 2013 Oct 1

Papp K, Végh P, Hóbor R, Szittner Z, Vokó Z, Podani J, Czirják L, Prechl J. Immune complex signatures of patients with active and inactive SLE revealed by multiplex protein binding analysis on antigen microarrays. PLoS One. 2012;7(9):e44824

Prechl J, Szittner Z, Papp K. Complementing antibody profiles: Assessing antibody function on antigen microarrays. Immunol Lett. 2012 Mar 30;143(1):101-5.

Szekeres Z, Herbáth M, Szittner Z, Papp K, Erdei A, Prechl J. Modulation of the humoral immune response by targeting CD40 and FcγRII/III; delivery of soluble but not particulate antigen to CD40 enhances antibody responses with a Th1 bias. Mol Immunol. 2011 Oct;49(1-2):155-62.

81

Szekeres Z, Herbáth M, Angyal A, Szittner Z, Virág V, Balogh P, Erdei A, Prechl J. Modulation of immune response by combined targeting of complement receptors and low-affinity Fcgamma receptors. Immunol Lett. 2010 May 4;130(1-2):66-73.

Abstracts at international conferences

Szittner Z, Bentlage AEH, Alcaro C, Prechl J, Vidarsson G. Label-free detection of immune complexes with myeloid cells. Controversies in Rheumatology and Autoimmunity, 12-14 March 2015, Sorrento, Italy

Szittner Z, Papp K, Sándor N, Holczer E, Herbáth M , Kecse-Nagy Cs, Fürjes P, Prechl J. Self-driven microfluidic chambers for a protein microarray cell-binding assay. EMBL Conference Microfluidics, 23-25 July 2014, Heidelberg, Germany

Szittner Z, Papp K, Sándor N, Bajtay Z, Prechl J. Application of monocytes on antigen microarrays - detecting inflammatory activation. IMPULSE conference an EFIS-EJI Symposium, 31 August-3 September 2013, Mátrahaza, Hungary

Szittner Z, Papp K, Sándor N, Bajtay Z, Prechl J. Detecting autoantibodies by myeloid cells: a direct way to assess effector functions. Controversies in Rheumatology & Autoimmunity, 4-6 April, 2013, Budapest, Hungary

Szittner Z, Szarka E, Pozsgay J , Papp K, Prechl J, Babos F, Magyar A, Hudecz F, Nagy G, Sármay G. Application of a microarray platform in rheumatoid arthritis for the detection of anti-citrullinated peptide antibodies and rheumatoid factors. 46th Annual Scientific Meeting of ESCI in Budapest, 22-24 March 2012, Budapest, Hungary

Szittner Z, Papp K, Bajtay Z, Prechl J. Application of fluorescent monocytes for probing antibody profiles on antigen microarrays. Impulse Conf./EFIS-EJI symposium, 3-6 September 2011, Visegrád, Hungary

82

References

1. Nutt, S.L., et al., The generation of antibody-secreting plasma cells. Nat Rev Immunol, 2015. 15(3): p. 160-71. 2. Xu, W. and J. Banchereau, The antigen presenting cells instruct plasma cell differentiation. Front Immunol, 2014. 4: p. 504. 3. Kunkel, E.J. and E.C. Butcher, Plasma-cell homing. Nat Rev Immunol, 2003. 3(10): p. 822-9. 4. Radbruch, A., et al., Competence and competition: the challenge of becoming a long-lived plasma cell. Nat Rev Immunol, 2006. 6(10): p. 741-50. 5. Williams, A.F. and A.N. Barclay, The immunoglobulin superfamily-- domains for cell surface recognition. Annu Rev Immunol, 1988. 6: p. 381-405. 6. Steipe, B., A. Pluckthun, and R. Huber, Refined crystal structure of a recombinant immunoglobulin domain and a complementarity- determining region 1-grafted mutant. J Mol Biol, 1992. 225(3): p. 739- 53. 7. Sela-Culang, I., V. Kunik, and Y. Ofran, The structural basis of antibody- antigen recognition. Front Immunol, 2013. 4: p. 302. 8. Schroeder, H.W., Jr. and L. Cavacini, Structure and function of immunoglobulins. J Allergy Clin Immunol, 2010. 125(2 Suppl 2): p. S41-52. 9. Victora, G.D. and M.C. Nussenzweig, Germinal centers. Annu Rev Immunol, 2012. 30: p. 429-57. 10. de Villartay, J.P., A. Fischer, and A. Durandy, The mechanisms of immune diversification and their disorders. Nat Rev Immunol, 2003. 3(12): p. 962-72. 11. Matthews, A.J., et al., Regulation of immunoglobulin class-switch recombination: choreography of noncoding transcription, targeted DNA deamination, and long-range DNA repair. Adv Immunol, 2014. 122: p. 1-57. 12. Pumphrey, R., Computer models of the human immunoglobulins shape and segmental flexibility. Immunol Today, 1986. 7(6): p. 174-8. 13. Moens, L. and S.G. Tangye, Cytokine-Mediated Regulation of Plasma Cell Generation: IL-21 Takes Center Stage. Front Immunol, 2014. 5: p. 65. 14. Janda, A., et al., Ig Constant Region Effects on Variable Region Structure and Function. Front Microbiol, 2016. 7: p. 22. 15. Collins, A.M. and K.J. Jackson, A Temporal Model of Human IgE and IgG Antibody Function. Front Immunol, 2013. 4: p. 235. 16. Vidarsson, G., G. Dekkers, and T. Rispens, IgG subclasses and allotypes: from structure to effector functions. Front Immunol, 2014. 5: p. 520. 17. van der Neut Kolfschoten, M., et al., Anti-inflammatory activity of human IgG4 antibodies by dynamic Fab arm exchange. Science, 2007. 317(5844): p. 1554-7.

83

18. Brezski, R.J. and G. Georgiou, Immunoglobulin isotype knowledge and application to Fc engineering. Curr Opin Immunol, 2016. 40: p. 62-69. 19. Irani, V., et al., Molecular properties of human IgG subclasses and their implications for designing therapeutic monoclonal antibodies against infectious diseases. Mol Immunol, 2015. 67(2 Pt A): p. 171-82. 20. Klimovich, V.B., IgM and its receptors: structural and functional aspects. Biochemistry (Mosc), 2011. 76(5): p. 534-49. 21. Kubagawa, H., et al., Identity of the elusive IgM Fc receptor (FcmuR) in humans. J Exp Med, 2009. 206(12): p. 2779-93. 22. Shibuya, A., et al., Fc alpha/mu receptor mediates endocytosis of IgM- coated microbes. Nat Immunol, 2000. 1(5): p. 441-6. 23. Vas, J., C. Gronwall, and G.J. Silverman, Fundamental roles of the innate- like repertoire of natural antibodies in immune homeostasis. Front Immunol, 2013. 4: p. 4. 24. Leusen, J.H., IgA as therapeutic antibody. Mol Immunol, 2015. 68(1): p. 35-9. 25. Macpherson, A.J., M.B. Geuking, and K.D. McCoy, Homeland security: IgA immunity at the frontiers of the body. Trends Immunol, 2012. 33(4): p. 160-7. 26. Gutzeit, C., G. Magri, and A. Cerutti, Intestinal IgA production and its role in host-microbe interaction. Immunol Rev, 2014. 260(1): p. 76-85. 27. Bakema, J.E. and M. van Egmond, The human immunoglobulin A Fc receptor FcalphaRI: a multifaceted regulator of mucosal immunity. Mucosal Immunol, 2011. 4(6): p. 612-24. 28. Oettgen, H.C. and O.T. Burton, IgE and Mast Cells: The Endogenous Adjuvant. Adv Immunol, 2015. 127: p. 203-56. 29. Wu, L.C. and A.A. Zarrin, The production and regulation of IgE by the immune system. Nat Rev Immunol, 2014. 14(4): p. 247-59. 30. Geisberger, R., M. Lamers, and G. Achatz, The riddle of the dual expression of IgM and IgD. Immunology, 2006. 118(4): p. 429-37. 31. Chen, K., et al., Immunoglobulin D enhances immune surveillance by activating antimicrobial, proinflammatory and B cell-stimulating programs in basophils. Nat Immunol, 2009. 10(8): p. 889-98. 32. Arnold, J.N., et al., The impact of glycosylation on the biological function and structure of human immunoglobulins. Annu Rev Immunol, 2007. 25: p. 21-50. 33. Quast, I. and J.D. Lunemann, Fc glycan-modulated immunoglobulin G effector functions. J Clin Immunol, 2014. 34 Suppl 1: p. S51-5. 34. Raghavan, M. and P.J. Bjorkman, Fc receptors and their interactions with immunoglobulins. Annu Rev Cell Dev Biol, 1996. 12: p. 181-220. 35. Monteiro, R.C. and J.G. Van De Winkel, IgA Fc receptors. Annu Rev Immunol, 2003. 21: p. 177-204. 36. Bruhns, P., S. Fremont, and M. Daeron, Regulation of allergy by Fc receptors. Curr Opin Immunol, 2005. 17(6): p. 662-9. 37. Phalipon, A. and B. Corthesy, Novel functions of the polymeric Ig

84

receptor: well beyond transport of immunoglobulins. Trends Immunol, 2003. 24(2): p. 55-8. 38. Pyzik, M., et al., FcRn: The Architect Behind the Immune and Nonimmune Functions of IgG and Albumin. J Immunol, 2015. 194(10): p. 4595-603. 39. Boross, P., et al., Fc Receptors, in eLS2001, John Wiley & Sons, Ltd. 40. Nimmerjahn, F., S. Gordan, and A. Lux, FcgammaR dependent mechanisms of cytotoxic, agonistic, and neutralizing antibody activities. Trends Immunol, 2015. 36(6): p. 325-36. 41. Sanchez-Mejorada, G. and C. Rosales, Fcgamma receptor-mediated mitogen-activated protein kinase activation in monocytes is independent of Ras. J Biol Chem, 1998. 273(42): p. 27610-9. 42. Kant, A.M., et al., SHP-1 regulates Fcgamma receptor-mediated phagocytosis and the activation of RAC. Blood, 2002. 100(5): p. 1852- 9. 43. Blank, U., et al., Inhibitory ITAMs as novel regulators of immunity. Immunol Rev, 2009. 232(1): p. 59-71. 44. Ben Mkaddem, S., et al., Shifting FcgammaRIIA-ITAM from activation to inhibitory configuration ameliorates arthritis. J Clin Invest, 2014. 124(9): p. 3945-59. 45. Pasquier, B., et al., Identification of FcalphaRI as an inhibitory receptor that controls inflammation: dual role of FcRgamma ITAM. Immunity, 2005. 22(1): p. 31-42. 46. Bruhns, P., et al., Specificity and affinity of human Fcgamma receptors and their polymorphic variants for human IgG subclasses. Blood, 2009. 113(16): p. 3716-25. 47. Bruhns, P. and F. Jonsson, Mouse and human FcR effector functions. Immunol Rev, 2015. 268(1): p. 25-51. 48. Hayes, J.M., et al., Glycosylation and Fc receptors. Curr Top Microbiol Immunol, 2014. 382: p. 165-99. 49. Erdei, A., Immunológia2012, Budapest: Medicina Könyvkiadó Zrt. 50. Merle, N.S., et al., Complement System Part I - Molecular Mechanisms of Activation and Regulation. Front Immunol, 2015. 6: p. 262. 51. Lambris, J.D., D. Ricklin, and B.V. Geisbrecht, Complement evasion by human pathogens. Nat Rev Microbiol, 2008. 6(2): p. 132-42. 52. Merle, N.S., et al., Complement System Part II: Role in Immunity. Front Immunol, 2015. 6: p. 257. 53. Holers, V.M., Complement and its receptors: new insights into human disease. Annu Rev Immunol, 2014. 32: p. 433-59. 54. Helmy, K.Y., et al., CRIg: a macrophage complement receptor required for phagocytosis of circulating pathogens. Cell, 2006. 124(5): p. 915- 27. 55. Ricklin, D., et al., Complement: a key system for immune surveillance and homeostasis. Nat Immunol, 2010. 11(9): p. 785-97. 56. Puri, T.S. and R.J. Quigg, The many effects of complement C3- and C5-

85

binding proteins in renal injury. Semin Nephrol, 2007. 27(3): p. 321- 37. 57. Zipfel, P.F. and C. Skerka, Complement regulators and inhibitory proteins. Nat Rev Immunol, 2009. 9(10): p. 729-40. 58. Orkin, S.H. and L.I. Zon, Hematopoiesis: an evolving paradigm for stem cell biology. Cell, 2008. 132(4): p. 631-44. 59. Doulatov, S., et al., Hematopoiesis: a human perspective. Cell Stem Cell, 2012. 10(2): p. 120-36. 60. Alvarez-Errico, D., et al., Epigenetic control of myeloid cell differentiation, identity and function. Nat Rev Immunol, 2015. 15(1): p. 7-17. 61. Soehnlein, O. and L. Lindbom, Phagocyte partnership during the onset and resolution of inflammation. Nat Rev Immunol, 2010. 10(6): p. 427- 39. 62. Kantari, C., M. Pederzoli-Ribeil, and V. Witko-Sarsat, The role of neutrophils and monocytes in innate immunity. Contrib Microbiol, 2008. 15: p. 118-46. 63. Haldar, M. and K.M. Murphy, Origin, development, and homeostasis of tissue-resident macrophages. Immunol Rev, 2014. 262(1): p. 25-35. 64. Sieweke, M.H. and J.E. Allen, Beyond stem cells: self-renewal of differentiated macrophages. Science, 2013. 342(6161): p. 1242974. 65. Ziegler-Heitbrock, L., et al., Nomenclature of monocytes and dendritic cells in blood. Blood, 2010. 116(16): p. e74-80. 66. Ginhoux, F. and S. Jung, Monocytes and macrophages: developmental pathways and tissue homeostasis. Nat Rev Immunol, 2014. 14(6): p. 392-404. 67. Tak, P.P. and G.S. Firestein, NF-kappaB: a key role in inflammatory diseases. J Clin Invest, 2001. 107(1): p. 7-11. 68. Garcia-Garcia, E., G. Sanchez-Mejorada, and C. Rosales, Phosphatidylinositol 3-kinase and ERK are required for NF-kappaB activation but not for phagocytosis. J Leukoc Biol, 2001. 70(4): p. 649- 58. 69. Van den Herik-Oudijk, I.E., et al., Functional differences between two Fc receptor ITAM signaling motifs. Blood, 1995. 86(9): p. 3302-7. 70. Kwiatkowska, K., J. Frey, and A. Sobota, Phosphorylation of FcgammaRIIA is required for the receptor-induced actin rearrangement and capping: the role of membrane rafts. J Cell Sci, 2003. 116(Pt 3): p. 537-50. 71. Brandsma, A.M., et al., Fc receptor inside-out signaling and possible impact on antibody therapy. Immunol Rev, 2015. 268(1): p. 74-87. 72. Garcia-Garcia, E., E.J. Brown, and C. Rosales, Transmembrane mutations to FcgammaRIIA alter its association with lipid rafts: implications for receptor signaling. J Immunol, 2007. 178(5): p. 3048- 58. 73. Beekman, J.M., et al., FcgammaRI (CD64) resides constitutively in lipid

86

rafts. Immunol Lett, 2008. 116(2): p. 149-55. 74. Hara, H., et al., The adaptor protein CARD9 is essential for the activation of myeloid cells through ITAM-associated and Toll-like receptors. Nat Immunol, 2007. 8(6): p. 619-29. 75. Dai, X., et al., Differential signal transduction, membrane trafficking, and immune effector functions mediated by FcgammaRI versus FcgammaRIIa. Blood, 2009. 114(2): p. 318-27. 76. Lowell, C.A., Src-family and Syk kinases in activating and inhibitory pathways in innate immune cells: signaling cross talk. Cold Spring Harb Perspect Biol, 2011. 3(3). 77. Goley, E.D. and M.D. Welch, The ARP2/3 complex: an actin nucleator comes of age. Nat Rev Mol Cell Biol, 2006. 7(10): p. 713-26. 78. Cheng, G., et al., Nox1-dependent reactive oxygen generation is regulated by Rac1. J Biol Chem, 2006. 281(26): p. 17718-26. 79. Sundstrom, C. and K. Nilsson, Establishment and characterization of a human histiocytic lymphoma cell line (U-937). Int J Cancer, 1976. 17(5): p. 565-77. 80. Boltz-Nitulescu, G., et al., Modulation of IgA, IgE, and IgG Fc receptor expression on human mononuclear phagocytes by 1 alpha,25- dihydroxyvitamin D3 and cytokines. J Leukoc Biol, 1995. 58(2): p. 256- 62. 81. Valenzuela, N.M., et al., Monocyte recruitment by HLA IgG-activated endothelium: the relationship between IgG subclass and FcgammaRIIa polymorphisms. Am J Transplant, 2015. 15(6): p. 1502-18. 82. Schreiber, J., et al., Coordinated binding of NF-kappaB family members in the response of human cells to lipopolysaccharide. Proc Natl Acad Sci U S A, 2006. 103(15): p. 5899-904. 83. Suurmond, J. and B. Diamond, Autoantibodies in systemic autoimmune diseases: specificity and pathogenicity. J Clin Invest, 2015. 125(6): p. 2194-202. 84. Delgado-Vega, A., et al., Recent findings on genetics of systemic autoimmune diseases. Curr Opin Immunol, 2010. 22(6): p. 698-705. 85. Xiao, X. and C. Chang, Diagnosis and classification of drug-induced autoimmunity (DIA). J Autoimmun, 2014. 48-49: p. 66-72. 86. Farhat, S.C., et al., Air pollution in autoimmune rheumatic diseases: a review. Autoimmun Rev, 2011. 11(1): p. 14-21. 87. Perricone, C., et al., Smoke and autoimmunity: The fire behind the disease. Autoimmun Rev, 2016. 15(4): p. 354-74. 88. Wahren-Herlenius, M. and T. Dorner, Immunopathogenic mechanisms of systemic autoimmune disease. Lancet, 2013. 382(9894): p. 819-31. 89. Draborg, A.H., K. Duus, and G. Houen, Epstein-Barr virus in systemic autoimmune diseases. Clin Dev Immunol, 2013. 2013: p. 535738. 90. Tektonidou, M.G. and M.M. Ward, Validation of new biomarkers in systemic autoimmune diseases. Nat Rev Rheumatol, 2011. 7(12): p. 708-17.

87

91. Yu, C., M.E. Gershwin, and C. Chang, Diagnostic criteria for systemic lupus erythematosus: a critical review. J Autoimmun, 2014. 48-49: p. 10-3. 92. Kourilovitch, M., C. Galarza-Maldonado, and E. Ortiz-Prado, Diagnosis and classification of rheumatoid arthritis. J Autoimmun, 2014. 48-49: p. 26-30. 93. Firestein, G.S., Evolving concepts of rheumatoid arthritis. Nature, 2003. 423(6937): p. 356-61. 94. Aletaha, D., et al., 2010 rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Ann Rheum Dis, 2010. 69(9): p. 1580-8. 95. Miller, A., et al., Is rheumatoid factor useful in primary care? A retrospective cross-sectional study. Clin Rheumatol, 2013. 32(7): p. 1089-93. 96. Ryu, H.J., et al., The diagnostic utilities of anti-agalactosyl IgG antibodies, anti-cyclic citrullinated peptide antibodies, and rheumatoid factors in rheumatoid arthritis. Rheumatol Int, 2011. 31(3): p. 315-9. 97. Nielsen, S.F., et al., Elevated rheumatoid factor and long term risk of rheumatoid arthritis: a prospective cohort study. BMJ, 2012. 345: p. e5244. 98. Anquetil, F., et al., IgM and IgA rheumatoid factors purified from rheumatoid arthritis sera boost the Fc receptor- and complement- dependent effector functions of the disease-specific anti-citrullinated protein autoantibodies. J Immunol, 2015. 194(8): p. 3664-74. 99. Gyorgy, B., et al., Citrullination: a posttranslational modification in health and disease. Int J Biochem Cell Biol, 2006. 38(10): p. 1662-77. 100. Foulquier, C., et al., Peptidyl arginine deiminase type 2 (PAD-2) and PAD-4 but not PAD-1, PAD-3, and PAD-6 are expressed in rheumatoid arthritis synovium in close association with tissue inflammation. Arthritis Rheum, 2007. 56(11): p. 3541-53. 101. Kokkonen, H., et al., Antibodies of IgG, IgA and IgM isotypes against cyclic citrullinated peptide precede the development of rheumatoid arthritis. Arthritis Res Ther, 2011. 13(1): p. R13. 102. Munoz, L.E., et al., Autoimmunity and chronic inflammation - two clearance-related steps in the etiopathogenesis of SLE. Autoimmun Rev, 2010. 10(1): p. 38-42. 103. Rhodes, B. and T.J. Vyse, The genetics of SLE: an update in the light of genome-wide association studies. Rheumatology (Oxford), 2008. 47(11): p. 1603-11. 104. Yaniv, G., et al., A volcanic explosion of autoantibodies in systemic lupus erythematosus: a diversity of 180 different antibodies found in SLE patients. Autoimmun Rev, 2015. 14(1): p. 75-9. 105. Beurskens, F.J., R.A. van Schaarenburg, and L.A. Trouw, C1q, antibodies and anti-C1q autoantibodies. Mol Immunol, 2015. 68(1): p. 6-13.

88

106. Kavai, M., Immune complex clearance by complement receptor type 1 in SLE. Autoimmun Rev, 2008. 8(2): p. 160-4. 107. Emlen, W., V. Carl, and G. Burdick, Mechanism of transfer of immune complexes from red blood cell CR1 to monocytes. Clin Exp Immunol, 1992. 89(1): p. 8-17. 108. Arora, V., et al., Reduced complement receptor 1 (CR1, CD35) transcription in systemic lupus erythematosus. Mol Immunol, 2004. 41(4): p. 449-56. 109. Walport, M.J., Complement and systemic lupus erythematosus. Arthritis Res, 2002. 4 Suppl 3: p. S279-93. 110. Lintner, K.E., et al., Early Components of the Complement Classical Activation Pathway in Human Systemic Autoimmune Diseases. Front Immunol, 2016. 7: p. 36. 111. Bryan, A.R. and E.Y. Wu, Complement deficiencies in systemic lupus erythematosus. Curr Allergy Asthma Rep, 2014. 14(7): p. 448. 112. Kavai, M. and G. Szegedi, Immune complex clearance by monocytes and macrophages in systemic lupus erythematosus. Autoimmun Rev, 2007. 6(7): p. 497-502. 113. Mok, C.C. and C.S. Lau, Pathogenesis of systemic lupus erythematosus. J Clin Pathol, 2003. 56(7): p. 481-90. 114. Mak, A., D.A. Isenberg, and C.S. Lau, Global trends, potential mechanisms and early detection of organ damage in SLE. Nat Rev Rheumatol, 2013. 9(5): p. 301-10. 115. Weckerle, C.E. and T.B. Niewold, The unexplained female predominance of systemic lupus erythematosus: clues from genetic and cytokine studies. Clin Rev Allergy Immunol, 2011. 40(1): p. 42-9. 116. Arbuckle, M.R., et al., Development of autoantibodies before the clinical onset of systemic lupus erythematosus. N Engl J Med, 2003. 349(16): p. 1526-33. 117. Ekins, R.P., Multi-analyte immunoassay. J Pharm Biomed Anal, 1989. 7(2): p. 155-68. 118. Ekins, R., F. Chu, and J. Micallef, High specific activity chemiluminescent and fluorescent markers: their potential application to high sensitivity and 'multi-analyte' immunoassays. J Biolumin Chemilumin, 1989. 4(1): p. 59-78. 119. Ekins, R.P. and F.W. Chu, Multianalyte microspot immunoassay-- microanalytical "compact disk" of the future. Clin Chem, 1991. 37(11): p. 1955-67. 120. Donohue, J., et al., Enzyme immunoassay system for panel testing. Clin Chem, 1989. 35(9): p. 1874-7. 121. Kakabakos, S.E., T.K. Christopoulos, and E.P. Diamandis, Multianalyte immunoassay based on spatially distinct fluorescent areas quantified by laser-excited solid-phase time-resolved fluorometry. Clin Chem, 1992. 38(3): p. 338-42. 122. Schena, M., et al., Quantitative monitoring of gene expression patterns

89

with a complementary DNA microarray. Science, 1995. 270(5235): p. 467-70. 123. Bussow, K., et al., A method for global protein expression and antibody screening on high-density filters of an arrayed cDNA library. Nucleic Acids Res, 1998. 26(21): p. 5007-8. 124. Bussow, K., et al., A human cDNA library for high-throughput protein expression screening. Genomics, 2000. 65(1): p. 1-8. 125. de Wildt, R.M., et al., Antibody arrays for high-throughput screening of antibody-antigen interactions. Nat Biotechnol, 2000. 18(9): p. 989-94. 126. MacBeath, G. and S.L. Schreiber, Printing proteins as microarrays for high-throughput function determination. Science, 2000. 289(5485): p. 1760-3. 127. Tao, S.C., C.S. Chen, and H. Zhu, Applications of protein microarray technology. Comb Chem High Throughput Screen, 2007. 10(8): p. 706- 18. 128. Robinson, W.H., Antigen arrays for antibody profiling. Curr Opin Chem Biol, 2006. 10(1): p. 67-72. 129. Papp, K., et al., On-chip complement activation adds an extra dimension to antigen microarrays. Mol Cell Proteomics, 2007. 6(1): p. 133-40. 130. Melinda Herbáth, et al., Exploiting fluorescence for multiplex immunoassays on protein microarrays. Methods and Applications in Fluorescence, 2014. 2(3): p. 032001. 131. Balboni, I., et al., Evaluation of microarray surfaces and arraying parameters for profiling. Proteomics, 2008. 8(17): p. 3443-9. 132. Wacker, R., H. Schroder, and C.M. Niemeyer, Performance of antibody microarrays fabricated by either DNA-directed immobilization, direct spotting, or streptavidin-biotin attachment: a comparative study. Anal Biochem, 2004. 330(2): p. 281-7. 133. Belov, L., et al., Immunophenotyping of leukemias using a cluster of differentiation antibody microarray. Cancer Res, 2001. 61(11): p. 4483-9. 134. Belov, L., et al., Screening microarrays of novel monoclonal antibodies for binding to T-, B- and myeloid leukaemia cells. J Immunol Methods, 2005. 305(1): p. 10-9. 135. Campbell, C.J., et al., Cell interaction microarray for blood phenotyping. Anal Chem, 2006. 78(6): p. 1930-8. 136. Ko, I.K., K. Kato, and H. Iwata, Parallel analysis of multiple surface markers expressed on rat neural stem cells using antibody microarrays. Biomaterials, 2005. 26(23): p. 4882-91. 137. Wu, J.Q., et al., Antibody microarray analysis of cell surface antigens on CD4+ and CD8+ T cells from HIV+ individuals correlates with disease stages. Retrovirology, 2007. 4: p. 83. 138. Tateno, H., et al., A novel strategy for mammalian cell surface glycome profiling using lectin microarray. Glycobiology, 2007. 17(10): p. 1138-

90

46. 139. Soen, Y., et al., Detection and characterization of cellular immune responses using peptide-MHC microarrays. PLoS Biol, 2003. 1(3): p. E65. 140. Ge, X., et al., Peptide-MHC cellular microarray with innovative data analysis system for simultaneously detecting multiple CD4 T-cell responses. PLoS One, 2010. 5(6): p. e11355. 141. Stone, J.D., W.E. Demkowicz, Jr., and L.J. Stern, HLA-restricted epitope identification and detection of functional responses by using MHC- peptide and costimulatory microarrays. Proc Natl Acad Sci U S A, 2005. 102(10): p. 3744-9. 142. Ziauddin, J. and D.M. Sabatini, Microarrays of cells expressing defined cDNAs. Nature, 2001. 411(6833): p. 107-10. 143. Hook, A.L., H. Thissen, and N.H. Voelcker, Surface manipulation of biomolecules for cell microarray applications. Trends Biotechnol, 2006. 24(10): p. 471-7. 144. Liedberg, B., C. Nylander, and I. Lundstrom, Biosensing with surface plasmon resonance--how it all started. Biosens Bioelectron, 1995. 10(8): p. i-ix. 145. Rich, R.L. and D.G. Myszka, Advances in surface plasmon resonance biosensor analysis. Curr Opin Biotechnol, 2000. 11(1): p. 54-61. 146. Giebel, K., et al., Imaging of cell/substrate contacts of living cells with surface plasmon resonance microscopy. Biophys J, 1999. 76(1 Pt 1): p. 509-16. 147. Stojanovic, I., R.B. Schasfoort, and L.W. Terstappen, Analysis of cell surface antigens by Surface Plasmon Resonance imaging. Biosens Bioelectron, 2014. 52: p. 36-43. 148. Suraniti, E., et al., Real-time detection of lymphocytes binding on an antibody chip using SPR imaging. Lab Chip, 2007. 7(9): p. 1206-8. 149. Schasfoort, R.B., et al., Label-free cell profiling. Anal Biochem, 2013. 439(1): p. 4-6. 150. Kapur, R., et al., C-reactive protein enhances IgG-mediated phagocyte responses and thrombocytopenia. Blood, 2015. 125(11): p. 1793-802. 151. Maltais, J.S., et al., Label-free monitoring of apoptosis by surface plasmon resonance detection of morphological changes. Apoptosis, 2012. 17(8): p. 916-25. 152. Yanase, Y., et al., Application of SPR imaging sensor for detection of individual living cell reactions and clinical diagnosis of type I allergy. Allergol Int, 2013. 62(2): p. 163-9. 153. Peterson, A.W., et al., Using surface plasmon resonance imaging to probe dynamic interactions between cells and extracellular matrix. Cytometry A, 2010. 77(9): p. 895-903. 154. Milgram, S., et al., On chip real time monitoring of B-cells hybridoma secretion of immunoglobulin. Biosens Bioelectron, 2011. 26(5): p. 2728-32.

91

155. Ward, J.O., et al., The acute promyelocytic leukemia-associated protein, promyelocytic leukemia zinc finger, regulates 1,25-dihydroxyvitamin D(3)-induced monocytic differentiation of U937 cells through a physical interaction with vitamin D(3) receptor. Blood, 2001. 98(12): p. 3290- 300. 156. Spittler, A., et al., IL-10 augments CD23 expression on U937 cells and down-regulates IL-4-driven CD23 expression on cultured human blood monocytes: effects of IL-10 and other cytokines on cell phenotype and phagocytosis. Immunology, 1995. 85(2): p. 311-7. 157. Spittler, A., et al., Effects of 1 alpha,25-dihydroxyvitamin D3 and cytokines on the expression of MHC antigens, complement receptors and other antigens on human blood monocytes and U937 cells: role in cell differentiation, activation and phagocytosis. Immunology, 1997. 90(2): p. 286-93. 158. Hogarth, P.M. and G.A. Pietersz, Fc receptor-targeted therapies for the treatment of inflammation, cancer and beyond. Nat Rev Drug Discov, 2012. 11(4): p. 311-31. 159. Pankhurst, T., et al., Immunoglobulin subclass determines ability of immunoglobulin (Ig)G to capture and activate neutrophils presented as normal human IgG or disease-associated anti-neutrophil cytoplasm antibody (ANCA)-IgG. Clin Exp Immunol, 2011. 164(2): p. 218-26. 160. Lux, A., et al., Impact of immune complex size and glycosylation on IgG binding to human FcgammaRs. J Immunol, 2013. 190(8): p. 4315-23. 161. Atkins, K.L., et al., S. aureus IgG-binding proteins SpA and Sbi: host specificity and mechanisms of immune complex formation. Mol Immunol, 2008. 45(6): p. 1600-11. 162. Wilkinson, B.J., et al., Activation of complement by cell surface components of Staphylococcus aureus. Infect Immun, 1978. 20(2): p. 388-92. 163. Neth, O., et al., Enhancement of complement activation and opsonophagocytosis by complexes of mannose-binding lectin with mannose-binding lectin-associated serine protease after binding to Staphylococcus aureus. J Immunol, 2002. 169(8): p. 4430-6. 164. Verbrugh, H.A., et al., The role of Staphylococcus aureus cell-wall peptidoglycan, teichoic acid and protein A in the processes of complement activation and opsonization. Immunology, 1979. 37(3): p. 615-21. 165. Seino, J., et al., Activation of human complement by mouse and mouse/human chimeric monoclonal antibodies. Clin Exp Immunol, 1993. 94(2): p. 291-6. 166. Chatham, W.W. and W.D. Blackburn, Jr., Fixation of C3 to IgG attenuates neutrophil HOCl generation and collagenase activation. J Immunol, 1993. 151(2): p. 949-58. 167. Wang, S.Y., et al., NK-cell activation and antibody-dependent cellular cytotoxicity induced by rituximab-coated target cells is inhibited by the

92

C3b component of complement. Blood, 2008. 111(3): p. 1456-63. 168. Nilsson, U.R., Deposition of C3b/iC3b leads to the concealment of antigens, immunoglobulins and bound C1q in complement-activating immune complexes. Mol Immunol, 2001. 38(2-3): p. 151-60. 169. Roopenian, D.C. and S. Akilesh, FcRn: the neonatal Fc receptor comes of age. Nat Rev Immunol, 2007. 7(9): p. 715-25. 170. Jefferis, R., J. Lund, and J.D. Pound, IgG-Fc-mediated effector functions: molecular definition of interaction sites for effector ligands and the role of glycosylation. Immunol Rev, 1998. 163: p. 59-76. 171. Vivanco, F., et al., The covalent interaction of C3 with IgG immune complexes. Mol Immunol, 1999. 36(13-14): p. 843-52. 172. Wikstrom, M., et al., Mapping of the immunoglobulin light chain- binding site of protein L. J Mol Biol, 1995. 250(2): p. 128-33. 173. Allhorn, M., A. Olsen, and M. Collin, EndoS from Streptococcus pyogenes is hydrolyzed by the cysteine proteinase SpeB and requires glutamic acid 235 and tryptophans for IgG glycan-hydrolyzing activity. BMC Microbiol, 2008. 8: p. 3. 174. Krapp, S., et al., Structural analysis of human IgG-Fc glycoforms reveals a correlation between glycosylation and structural integrity. J Mol Biol, 2003. 325(5): p. 979-89. 175. Jefferis, R., Isotype and glycoform selection for antibody therapeutics. Arch Biochem Biophys, 2012. 526(2): p. 159-66. 176. Dashivets, T., et al., Multi-Angle Effector Function Analysis of Human Monoclonal IgG Glycovariants. PLoS One, 2015. 10(12): p. e0143520. 177. Goulabchand, R., et al., Impact of autoantibody glycosylation in autoimmune diseases. Autoimmun Rev, 2014. 13(7): p. 742-50. 178. Reveille, J.D., Predictive value of autoantibodies for activity of systemic lupus erythematosus. Lupus, 2004. 13(5): p. 290-7. 179. Pratesi, F., et al., Antibodies to a new viral citrullinated peptide, VCP2: fine specificity and correlation with anti-cyclic citrullinated peptide (CCP) and anti-VCP1 antibodies. Clin Exp Immunol, 2011. 164(3): p. 337-45. 180. Pratesi, F., et al., Antibodies from patients with rheumatoid arthritis target citrullinated histone 4 contained in neutrophils extracellular traps. Ann Rheum Dis, 2014. 73(7): p. 1414-22. 181. Sokolove, J., et al., Immune complexes containing citrullinated fibrinogen costimulate macrophages via Toll-like receptor 4 and Fcgamma receptor. Arthritis Rheum, 2011. 63(1): p. 53-62. 182. Rossi, G., et al., Biosensor analysis of anti-citrullinated protein/peptide antibody affinity. Anal Biochem, 2014. 465: p. 96-101. 183. Mannik, M., F.A. Nardella, and E.H. Sasso, Rheumatoid factors in immune complexes of patients with rheumatoid arthritis. Springer Semin Immunopathol, 1988. 10(2-3): p. 215-30. 184. Robbins, D.L., et al., Differential reactivity of rheumatoid synovial cells and serum rheumatoid factors to human immunoglobulin G subclasses 1

93

and 3 and their CH3 domains in rheumatoid arthritis. Arthritis Rheum, 1987. 30(5): p. 489-97. 185. Kanmert, D., et al., IgG rheumatoid factors against the four human Fc- gamma subclasses in early rheumatoid arthritis (the Swedish TIRA project). Scand J Immunol, 2012. 75(1): p. 115-9. 186. Wong, A., et al., IgG3 reactive rheumatoid factor in rheumatoid arthritis: etiologic and pathogenic considerations. Autoimmunity, 1994. 19(3): p. 199-210. 187. Robbins, D.L., et al., Allotypic dependency of the specificity and avidity of human monoclonal IgM rheumatoid factors derived from rheumatoid synovial cells. Arthritis Rheum, 1993. 36(3): p. 389-93. 188. Verpoort, K.N., et al., Isotype distribution of anti-cyclic citrullinated peptide antibodies in undifferentiated arthritis and rheumatoid arthritis reflects an ongoing immune response. Arthritis Rheum, 2006. 54(12): p. 3799-808. 189. Boes, M., Role of natural and immune IgM antibodies in immune responses. Mol Immunol, 2000. 37(18): p. 1141-9. 190. Pratesi, F., et al., Fingerprinting of anti-citrullinated protein antibodies (ACPA): specificity, isotypes and subclasses. Lupus, 2015. 24(4-5): p. 433-41. 191. van der Woude, D., et al., The ACPA isotype profile reflects long-term radiographic progression in rheumatoid arthritis. Ann Rheum Dis, 2010. 69(6): p. 1110-6. 192. Ioan-Facsinay, A., et al., Marked differences in fine specificity and isotype usage of the anti-citrullinated protein antibody in health and disease. Arthritis Rheum, 2008. 58(10): p. 3000-8. 193. Brink, M., et al., Rheumatoid factor isotypes in relation to antibodies against citrullinated peptides and carbamylated proteins before the onset of rheumatoid arthritis. Arthritis Res Ther, 2016. 18(1): p. 43. 194. Davignon, J.L., et al., Targeting monocytes/macrophages in the treatment of rheumatoid arthritis. Rheumatology (Oxford), 2013. 52(4): p. 590-8. 195. Harre, U., et al., Induction of osteoclastogenesis and bone loss by human autoantibodies against citrullinated vimentin. J Clin Invest, 2012. 122(5): p. 1791-802. 196. Roberts, C.A., A.K. Dickinson, and L.S. Taams, The Interplay Between Monocytes/Macrophages and CD4(+) T Cell Subsets in Rheumatoid Arthritis. Front Immunol, 2015. 6: p. 571. 197. Li, Q. and I.M. Verma, NF-kappaB regulation in the immune system. Nat Rev Immunol, 2002. 2(10): p. 725-34. 198. Takeuchi, O. and S. Akira, Pattern recognition receptors and inflammation. Cell, 2010. 140(6): p. 805-20. 199. Legrand-Poels, S., et al., Activation of the transcription factor NF- kappaB in lipopolysaccharide-stimulated U937 cells. Biochem Pharmacol, 1997. 53(3): p. 339-46.

94

200. Huang, G., et al., NF-kappaB plays a key role in inducing CD274 expression in human monocytes after lipopolysaccharide treatment. PLoS One, 2013. 8(4): p. e61602. 201. Wang, R., J. Stephens, and M.J. Lacy, Characterization of HTA125 with specificity for human TLR4. Hybrid Hybridomics, 2003. 22(6): p. 357-65. 202. Guha, M. and N. Mackman, LPS induction of gene expression in human monocytes. Cell Signal, 2001. 13(2): p. 85-94. 203. Chan, A.C. and P.J. Carter, Therapeutic antibodies for autoimmunity and inflammation. Nat Rev Immunol, 2010. 10(5): p. 301-16. 204. Davies, A.M. and B.J. Sutton, Human IgG4: a structural perspective. Immunol Rev, 2015. 268(1): p. 139-59. 205. Yamamoto, M., H. Takahashi, and Y. Shinomura, Mechanisms and assessment of IgG4-related disease: lessons for the rheumatologist. Nat Rev Rheumatol, 2014. 10(3): p. 148-59. 206. Laurent, L., et al., Fcgamma receptor profile of monocytes and macrophages from rheumatoid arthritis patients and their response to immune complexes formed with autoantibodies to citrullinated proteins. Ann Rheum Dis, 2011. 70(6): p. 1052-9. 207. Clavel, C., et al., Induction of macrophage secretion of tumor necrosis factor alpha through Fcgamma receptor IIa engagement by rheumatoid arthritis-specific autoantibodies to citrullinated proteins complexed with fibrinogen. Arthritis Rheum, 2008. 58(3): p. 678-88. 208. Vogelpoel, L.T., et al., Control of cytokine production by human fc gamma receptors: implications for pathogen defense and autoimmunity. Front Immunol, 2015. 6: p. 79. 209. Magnusson, S.E., et al., Dysregulated Fc receptor function in active rheumatoid arthritis. Immunol Lett, 2014. 162(1 Pt A): p. 200-6. 210. Hargreaves, C.E., et al., Fcgamma receptors: genetic variation, function, and disease. Immunol Rev, 2015. 268(1): p. 6-24. 211. Sfriso, P., et al., Infections and autoimmunity: the multifaceted relationship. J Leukoc Biol, 2010. 87(3): p. 385-95. 212. Tanaka, M., et al., Activation of Fc gamma RI on monocytes triggers differentiation into immature dendritic cells that induce autoreactive T cell responses. J Immunol, 2009. 183(4): p. 2349-55. 213. Li, Y., et al., Increased expression of FcgammaRI/CD64 on circulating monocytes parallels ongoing inflammation and nephritis in lupus. Arthritis Res Ther, 2009. 11(1): p. R6. 214. Mancardi, D.A., et al., The high-affinity human IgG receptor FcgammaRI (CD64) promotes IgG-mediated inflammation, anaphylaxis, and antitumor immunotherapy. Blood, 2013. 121(9): p. 1563-73.

95