THE OFFENSE-DEFENSE BALANCE in IMMUNITY by JANICE C. JUN

THE OFFENSE-DEFENSE BALANCE IN IMMUNITY

JANICE C. JUN

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Thesis Advisor: Dr. Brian A. Cobb, PhD

Department of Pathology

CASE WESTERN RESERVE UNIVERISTY

August 2016

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Janice C. Jun

candidate for the degree of Doctor of Philosophy.

Committee Chair

George Dubyak, PhD

Committee Member

Brian Cobb, PhD

Committee Member

Pamela Wearsch, PhD

Committee Member

Christine McDonald, PhD

Committee Member

Aaron Weinberg, DMD, PhD

Committee Member

Clive Hamlin, PhD

Date of Defense

April 8, 2016

*We also certify that written approval has been obtained for any proprietary material contained therein.

For my family, and Brian

Table of contents

List of Tables ………………………………………………………………………………………………………5

List of Figures ……………………………………………………………………………………………………..6

Acknowledgements ……………………………………………………………………………………………..8

List of Abbreviations ……………………………………………………………………………………………9

Abstract ……………………………………………………………………………………………………………10

Chapter 1. Against the Immunological Grain ………………..……………………………………...11

Chapter 2. Toll-like Receptor 2 Stimulation is a T cell-Intrinsic Co-Regulatory Pathway

Summary ………………………………………………………………………………………………..21

Introduction …………………………………………………………………………………………...22

Materials & Methods ……………………………………………………………………………….25

Results …………………………………………………………………………………………………...27

Figures …………………………………………………………………………………………………...35

Discussion ………………………………………………………………………………………………44

Chapter 3. Innate Immune-Directed NF-κB Signaling Requires Site-Specific NEMO Ubiquitination

Summary ………………………………………………………………………………………………..49

Introduction …………………………………………………………………………………………...50

Materials & Methods ………………………………………………………………….……………54

Results …………………………………………………………………………………………………...57

Figures ………………………………………………………………………………………………...…62

Discussion ………………………………………………………………………………………………69

Chapter 4. Generating a Novel Immunologic Framework …………………………….………72

References ……………………………………………………………………………………………...…………86

List of Tables

Chapter 2. Toll-like Receptor 2 Stimulation is a T cell-Intrinsic Co-Regulatory Pathway

Table 1: Luminex analysis of stimulated CD4+ T cells………………………………...42

List of Figures

Chapter 1. Against the Immunological Grain

Figure 1: Conflict continuum …………………………….…………………..….……….……..17

Chapter 2. Toll-like Receptor 2 Stimulation is a T cell-Intrinsic Co-Regulatory

Pathway

Figure 2: TLR2 and TCR co-Stimulation triggers synergistic CD4+ T cell-

intrinsic IL-10 production …………………………………………………………………..…...35

Figure 3: IL-10 synergy is TLR2- and dose-dependent, but independent of age

and gender …………………………………………………………………………….…...... 36

Figure 4: TLR2 co-stimulation has pleiotropic effects on T cell cytokine

production …………………………….……………..…………………………………………...... 37

Figure 5: TLR1/2 heterodimer agonists trigger IL-10 synergy ……………...…..38

Figure 6: CD4+ responder populations are non-regulatory and

antigen-experienced ………………………….…………………………………….…….…...... 39

Figure 7: TCR/TLR2 co-stimulation expands CD25+CD45Rbhi and Tem

populations ……………………….………………………………………………………...... 40

Figure 8: TCR/TLR2 co-Stimulation results in bystander suppression …....…41

Chapter 3. Innate immune-directed NF-κB signaling requires site-specific NEMO ubiquitination

Figure 9: Knockin of a non-ubiquitinatable NEMO allele causes embryonic

lethality …………………….……………………………………………………………...... 62

Figure 10: Non-ubiquitinatable NEMO knockin mice develop inflammatory

skin lesions and splenomegaly …………………….………………………….……………….63

Figure 11: Normal T cell numbers but decreased B cell numbers in the

spleens of XWTXNemoKi mice …………………….………………………………………………...64

Figure 12: Mating onto a TNFR1-deficient background rescues embryonic

lethality in the non-ubiquitinatable NEMO mice but causes steatohepatitis

and increased mortality …………………….……………………………….……………………65

Figure 13: Bone marrow derived macrophages (BMDMs) homozygous for

non-ubiquitinatable NEMO show a severe defect in NF-κB signaling in

response to the NOD2 agonist muramyl dipeptide (MDP) ………………….……..66

Figure 14: Bone marrow derived macrophages (BMDMs) homozygous for

non-ubiquitinatable NEMO show a defects in TLR4 and IL-1 responses but

normal interferon responses ……………….……………………………………………...…..67

Figure 15: NemoKi can bind to the IKKs .…………….…………………………………....68

Chapter 4. Generating a novel immunologic framework

Figure 1: Conflict continuum …………………………….…………………….……….………76

Acknowledgments

This endeavor would not have been possible without my foundation, my family. Thank you to my parents who are gracious and loving beyond words.

To Jessica – thank you for being my favorite lab sibling, for all the coffee times and non-hugs, and for always being there for me.

To the Cobb lab – Doug, Carlos, Mark, Jill, Lori – thank you for being such smart and wonderful people. You have all taught me so much, and for your incredible support and friendship, thank you.

I am so grateful for the incredible support and generosity of my entire thesis committee. Thank you all for your kindness, time, and belief in me – it means more to me than I can say.

Finally, to Brian – thank you for understanding me. I look up to you so much, and everything you have done will have effects far beyond today.

List of Abbreviations

APC: antigen presenting cell

TLR: Toll-like receptor

Tem: T effector/memory cell

NF-κB: Nuclear factor-kappa B

K63: lysine-63

TLR: Toll-like receptor

NEMO: NF-κB essential modulator

IKK: I Kappa kinase

NOD2: nucleotide oligomerization domain 2

MDP: Muramyl dipeptide

TNF: Tumor necrosis factor

TNFR1: TNF Receptor 1

AST: aspartate transferase

ALT: Alanine transferase

LDH: lactate dehydrogenase

NASH: non-alcoholic steatohepatitis

The Offense-Defense Balance in Immunity

Abstract

JANICE C. JUN

Entrenched in our conceptualization of the immune system is that it is comprised of two compartments: the innate immune system that describes the immediate, non- specific immune response to stimuli, and the adaptive immune system that provides memory, robustness, and antigen-specificity to host defense. However, growing evidence suggests significant overlap between these categories, bringing to question the functional usefulness of this dichotomy. In fact, we demonstrate here the profoundly synergistic and suppressive IL-10 response to TLR2 co-stimulation on

CD4+ T cells, confounding the traditional definitions of innate and adaptive immunity. This suggests that we can conceptualize the immune system from a different perspective that more fully encapsulates it functional capacity, which I propose is one of offensive vs defensive, rather than innate vs adaptive, immunity.

From this perspective, I identify two ubiquitination events on NEMO, a regulator of

NF-kB activation, to be strongly defensive in tandem in vivo, in contrast to its known offensive roles. By conceptualizing the immune system by an offense vs defense framework, I suggest that we can better understand immune mechanisms and phenomena in light of the fundamental purpose of the immune system, which is to maximize security.

Chapter 1: Against the Immunological Grain

Chapter 1

The adaptive vs innate immune paradigm has prevailed in immunology, shaping the way we describe immunological phenomena and imparting a certain obligation to try to “fit” novel immunologic mechanisms and components into one category or the other. While adaptive immunity describes the specific response to antigens that provides memory and robustness to host defense, innate immunity characterizes the front-line, immediate response to immune insults. One of the most canonical mediators of innate immune responses is the family of toll-like receptors, which are type I transmembrane proteins with extracellular leucine-rich regions

(LRR) thought to contribute to ligand recognition1. Toll-like receptors recognize and are activated by a breadth of motifs – such as lipopolysaccharide (LPS), flagellin, double-stranded RNA, lipoproteins – that are common to microbes and therefore signify the presence of a potential pathogen, and as such are collectively termed microbe-associated molecular patterns (MAMPs). Another (not necessarily exclusive) school of thought is that TLRs respond to indicators of injury to self, such as heat shock proteins and hyaluronan, termed danger-associated molecular patterns (DAMPs)2. The predominant outcome of TLR activation is one aimed at eliminating the insult, such as by pro-inflammatory cytokine and interferon (anti- viral) responses.

These receptors are typically functionally distinguished by their different ligand specificity and cellular localization (i.e. where TLR3, TLR7, TLR8, and TLR9 are endosomally-localized, while the remaining TLRs – TLR1, TLR2, TLR4, TLR5,

TLR6, and TLR10 – are surface-bound). TLRs have broadly varying ligand

12 specificities encompassing different pathogenic entities, such as double-stranded

RNA of viruses (TLR3), lipopolysaccharide of gram-negative bacteria (TLR4), and zymosan of fungi (TLR6). TLR activation by their respective ligands triggers signaling cascades that generally have in common the adaptor molecule myeloid differentiation primary–response protein 88 (MyD88), IL-1R-associated kinases

(IRAKs), transforming growth factor-β (TGF-β)-activated kinase (TAK1), TAK1- binding protein 1 (TAB1), TAB2, and tumor-necrosis factor (TNF)-receptor- associated factor 6 (TRAF6), culminating typically in inflammatory NF-κB responses

(MyD88-dependent) or interferon (IFN) (MyD88-independent) responses.

However, of this family, one receptor in particular is for several reasons the proverbial black sheep: TLR2. While the majority of TLRs homodimerize upon activation, TLR2 requires heterodimerization with TLR13 or TLR64, which is thought to contribute to its substantially wider ligand specificity compared to other TLRs5.

Specifically, the TLR1/2 complex is known to recognize diacyl and triacyl lipopeptides of bacteria and mycobacteria, while the TLR2/6 complex recognizes diacylated lipoproteins of mycoplasma and bacterial lipotechoic acid. Additionally, while the predominant immediate function of TLRs is to initiate effector, pro- inflammatory cascades, TLR2 is known to mount anti-inflammatory responses – specifically, TLR2 has been found to trigger IL-10 responses in dendritic cells and macrophages6,7 with significant ramifications on pathogenic potential, including that of Mycobacterium tuberculosis8, Candida albicans9, and Yersinia10.

Even more compellingly, TLR2 has been highly associated with suppressive responses not only in innate immune cells, but also on the adaptive immune side of the divide. To our lab, this association has predominantly centered on the intriguing effects of polysaccharide A (PSA), a capsular polysaccharide of the commensal bacteria Bacteroides fragilis. PSA is a TLR2 agonist capable of generating profoundly suppressive CD4+ T cells, but TLR2’s role has been limited to the antigen-presenting cell (APC) in this two-step mechanism. TLR2 activation on an APC by PSA triggers a cascade of events in the APC, including nitric oxide burst that is required to break down PSA that has concomitantly been taken up by the cell; this fragmentation of

PSA is necessary for its presentation by MHC class II molecules to CD4+ T cells, which ultimately generates the suppressive CD4+ T cell response unique so far to

PSA11,12.

The striking effect of PSA on CD4+ T cells is characterized by IL-10 production, a key immunosuppressive cytokine6,7. This IL-10-mediated immunosuppression triggered by PSA has been found to prevent disease in a multitude of animal models, including IBD13 and asthma14, sparking profound interest in further dissecting PSA’s mechanism of action. Intriguingly, recent evidence suggests that PSA is required to engage TLR2 specifically on the CD4+ T cell to promote Bacteroides fragilis colonization15. While this is interesting in the context of commensal-mediated modulation of the intestinal immune environment, this raised the fascinating possibility that an innate immune receptor (TLR2) may have a CD4+ T (i.e. adaptive) cell-intrinsic role in mediating a profoundly suppressive response.

This generated a simple, yet somewhat audacious, hypothesis that TLR2 engagement on a CD4+ T cell would generate an IL-10 response. In Chapter 2, I describe in detail the profound effect of CD4+ T cell-intrinsic TLR2 stimulation that not only generates an enormous IL-10 response, but intriguingly can have entirely different functional effects (suppressive or pro-inflammatory) depending on the presence or absence of concomitant T cell receptor (TCR) signaling, suggesting that

TLR2 may modulate a wide range of CD4+ T cell states. Even more fundamentally, however, these data add to a growing body of evidence defacing the division between innate and adaptive immunity. While the distinction between the immediate, non-specific (innate) and delayed, antigen-specific (adaptive) responses to immunogenic stimuli has been useful as broad categorizations, a growing understanding of the overlap between these categories suggests that we cannot in fact neatly separate the immune system into these compartments. Classically innate natural killer cells have been found to possess the antigen-specificity so canonical to the definition of adaptive immunity16,17, and canonically innate immune receptors have been found to be expressed and functional on the most definitively adaptive immune cells: B and T cells18–27, meaning that the overlap in the Venn diagram of innate vs adaptive immunity is expanding to the point where the functional usefulness of this dichotomy becomes moot.

In fact, we can even further illustrate the insufficiency of the innate vs adaptive dichotomy by drawing parallels between the immune system and its real- world counterpart, the military. The immune system is essentially a biological military whose purpose is to defend our body from the external environment, just as

15 the purpose of the military is to defend a country from the international environment. In this light, trying to conceptualize and fit the entire functional capacity of the immune system into innate vs adaptive components is akin to conceptualizing and fitting the entire functional capacity of the military into infantry vs Special Forces components, which is absurdly narrow in scope, merely descriptive, and too low in resolution to frame the full capabilities of either entity in a meaningful way. This suggests, then, that we should push to understand the immune system from a more meaningful perspective, which could yield profound ramifications on our understanding and ability to manipulate the immune system, and I propose that we utilize the profound homology of the immune system and the military to do.

The military and the immune system have an identical end-goal: to maximize security. By understanding how the military maximizes national security, we can apply relevant principles to the immune system to understand how it maximizes immunological security from a perspective that more deeply reflects its fundamental function that is profoundly military in nature. The functional range of the immune system mirrors that of the military, spanning from homeostasis (i.e. peace) to an inflammatory response (i.e. war) (Figure 1). Militarily, the ability to maximize security depends on achieving two principal objectives on this continuum:

1) maintain peace by understanding the factors promoting war, and if one must go to war, 2) win war. Achieving both of these objectives is, at their very root, based in the paradigm of offense vs defense. The immune system also has identical end- goals: 1) to maintain homeostasis, and if it must go to war, 2) to mount a productive

16 inflammatory response – and so I propose that the immune system’s ability to maximize security is rooted not in a dichotomy of innate vs adaptive immunity, but of offensive vs defensive immunity, and that understanding how the military applies this paradigm to maintaining peace and achieving victory in war will profoundly affect our understanding of immunity.

Figure 1.

Inflammatory Homeostasis response

1. Maintain peace 2. Win war

Adapted from Doctrine for the Armed Forces of the United States, JP-1, March 2013

By shifting our thinking from innate vs adaptive to offensive vs defensive immune components, we can immunologically translate the offense-defense theory that is used to understand the cause of war, and therefore how to avoid war

(objective 1). This theory states that there exists an offense-defense balance that is determined by “the [military] technology that is available to states,” and “that international conflict and war are more likely when offense has the advantage, while peace and cooperation more probable when defense has the advantage.”28–32

Translated, this suggests that there exists an immune offense-defense balance determined by the immune components available, and that an immune response (war) is more likely when the balance is offense-shifted, and that homeostasis (peace) is more likely when the balance is defense-shifted. We can experimentally identify the defensive vs offensive nature of immune components using the offense-defense theory, which suggests that a shift from homeostasis to inflammation due to the

17 removal of a component suggests that the component was defensive enough to tip the balance to the offense in its absence.

In Chapter 3, I use the offense-defense theory to come to a new understanding of two ubiquitination events on NEMO, best known for mediating pro-inflammatory, “offensive” responses via NF-kB. NF-kB, a transcription factor that has incredibly diverse roles in biology, is involved in multiple contexts of inflammation and is well known for its regulation and induction of pro- inflammatory responses in response to diverse receptor activation. While NF-kB is basally inhibited in the cytoplasm by IkB proteins, activating stimuli trigger the degradation of IkB, the release of NF-kB to the nucleus, and the activation of a transcriptional program that is predominantly pro-inflammatory downstream of cell surface receptors such as TNF-R1, TLRs, or cytoplasmic counterparts like NOD2 that respond to different bacterial patterns (i.e. muramyl-dipeptide derived from the bacterial cell wall)33,34. Signals from many of these receptors converge on an intermediary hub, the IKK complex, which controls the degradation of IkB that releases NF-kB for activity, suggesting that the regulation of the IKK complex is an important event in responses to inflammatory stimuli35.

The IKK complex is tripartite, consisting of two kinases (IKKα and IKKβ) and a non-catalytic subunit, IKKγ/NEMO. Post-translational modifications such as ubiquitination are a precise mechanism by which to fine-tune signaling responses, and two ubiquitination sites on NEMO (K285, K399) have been found to mediate inflammatory signals downstream of stimuli such as TLR4 and NOD2 activation. The

18 loss of either ubiquitination site is associated with a loss of NF-kB activation and productive inflammatory responses36–40, suggesting that these ubiquitination events are important to mounting NF-kB-driven inflammatory, or in other words, offensive responses. In Chapter 3, we test the in vivo effect of loss-of-ubiquitination at NEMO

K285/K399, and find, according to the offense-defense balance, that these ubiquitination events are in fact profoundly defensive in nature at the whole animal

(mouse) level. This sheds new significance to a molecular event most appreciated so far for its ability to mediate inflammatory, “offensive” NF-kB responses both in vitro and in vivo36–38.

In Chapter 4, I explore deeper ramifications of the offense-defense paradigm in immunity beyond simply identifying offensive vs defensive components, illustrating the value that can come of applying concepts from two entities that are normally very much disconnected from one another in the real world – immunology vs the military – and yet are so profoundly relevant to one another.

Chapter 2: Toll-like Receptor 2 Stimulation is a T cell-Intrinsic Co-Regulatory Pathway

Ch. 2 Summary

Here, we describe a striking example of the insufficiency of the adaptive vs innate immune dichotomy, where one of the most classical and well-established mediators of innate immune responses (toll-like receptors) are expressed and functional on one of the most canonical adaptive immune cells (T cells). Specifically, we show that

TLR2 co-stimulation on CD4+ T cells triggers robust, synergistic, and functionally suppressive IL-10 response that intriguingly is independent of classical regulatory T cells, which are typically most associated with suppressive IL-10 responses. That

TLR2 may have an even more complex effect on CD4 T cells than simply an IL-10- mediated suppressive response is suggested by its striking ability to promote pro- inflammatory responses (i.e. IFNγ) in bystander T cells in the absence of concomitant TCR signaling. Even further, we show that besides TLR2, TLR9 co- stimulation also triggers a synergistic IL-10 response in CD4 T cells, illustrating an innate immune receptor-mediated phenomenon in canonically adaptive T cells that further degrades the functional legitimacy of the innate vs adaptive divide.

Chapter 2

Introduction

The prototypical innate immune receptor family is the Toll-like receptors

(TLRs). These cell surface glycoproteins recognize molecular ‘patterns’ ranging from lipopolysaccharide and peptidoglycan to host-encoded heat-shock proteins, and are highly expressed by cells within the myeloid lineage. For nearly two decades, TLRs have been the subject of intense study for their role in ‘pattern recognition’ and the induction of the inflammatory response by neutrophils, macrophages, and other leukocytes 41,42. It is also well-appreciated that TLR stimulation can have potent, albeit indirect, effects on the downstream adaptive response through the promotion of cytokine, chemokine, and other mediator secretion from activated leukocytes. As such, the impact of TLR signaling upon the adaptive response is driven by the intrinsic APC and local leukocyte response.

In contrast, T cell activation occurs through the T cell receptor (TCR) and the

Lck-dependent proximal signaling complex upon specific recognition of its cognate antigen within the context of MHC molecules on opposing antigen presenting cells

(APCs)43. Amplification and suppression of that response is partly achieved through many co-stimulatory and co-regulatory molecules, such as the canonical CD28-

CD80/86 and CTLA4-CD80/86 pathways44, respectively. The response is also modulated by the cytokine milieu, which is partly a reflection of TLR stimulation in

APCs and other neighboring cells. For example, IL-12 from stimulated innate immune cells is well-known to promote Th1-type CD4+ T cell skewing45–47.

Despite the apparent separation of TLR and TCR stimulation among immune system cells, growing evidence suggests that TLRs are not limited to “innate” leukocytes and APCs. More specifically, several TLRs have been shown to be expressed in both mouse and human CD4+ T cells24,48–52, raising critical questions about the T cell-intrinsic role these receptors play in mounting an immune response and the maintenance of homeostasis.

Although members of the TLR family share many facets of their downstream signaling cascades, TLR2 appears to be somewhat distinct in its association with anti-inflammatory, suppressive responses. In fact, TLR2 engagement in both macrophages and dendritic cells has been found to mediate IL-10 production, a cytokine strongly associated with a regulatory response6,7. The result of such stimulation has been shown to suppress the immune system53, which holds serious implications for host defenses against pathogens such as Mycobacterium tuberculosis8, Candida albicans9, and pathogenic Yersinia species10. Moreover, the relationship between TLR2 and suppressive immune outcomes is further underscored by studies linking TLR2 stimulation with CD4+ regulatory T cells

(Tregs). While robust TLR2 expression has been demonstrated in Tregs24, the 50% reduction in Treg numbers in TLR2 knockout mice9 solidifies a specific role for

TLR2 in Tregs, although whether this role is intrinsic or extrinsic remains unclear.

Initial work exploring the role of TLR2 in Treg modulation suggested that

TLR2 induces Treg proliferation while reducing their suppressive capacity54,55; however, this contradicts follow up studies showing that TLR2 promotes Treg

23 survival without altering their suppressive capabilities56. In fact, an endogenous

TLR2 ligand has been shown to enhance Treg function57, and this correlation is supported by an in vivo study showing functionally significant TLR2-driven Treg expansion in an ovalbumin-based acute asthma model20. Yet despite these reports, little is known about the relationship between T cell-intrinsic TLR2 stimulation and

IL-10 production, T cell differentiation, or TCR stimulation within responding T cells.

In this study, we report that TLR2 stimulation is a CD4+ T cell-intrinsic co- regulatory pathway that synergistically induces the production of IL-10 when in combination with TCR signaling. In contrast to previous work focused on traditional

CD25+ and FoxP3+ Tregs, we found that TCR and TLR2 co-stimulation occurs preferentially in CD4+CD25- and CD4+FoxP3- T cell subsets, while promoting the expansion of CD4+CD25+CD62L-CD44+CD45Rbhi effector/memory T cells (Tem) which produce high concentrations of IL-10. While TLR2 co-activation also led to synergistic secretion of IL-6, CXCL-10, and IL-13, the net impact of TCR and TLR2 co- stimulation was suppression of bystander T cells that was IL-10-dependent and independent of FoxP3. Our findings show that TLR2 represents a new class of co- regulatory molecules expressed by CD4+ T cells which stimulates a unique, cell- intrinsic differentiation and cellular response program.

Materials and Methods

Animals. WT C57Bl/6 mice, IL-10-IRES-GFP reporter, FoxP3-IRES-RFP reporter, and IL-10 knockout mice were purchased from Jackson labs and housed in specific pathogen free conditions per guidelines established by the Institutional Animal Care and Use Committee of Case Western Reserve University (Cleveland, OH, USA). T cells from TLR2 knockout mice were a kind gift of Dr. Clifford V. Harding

(Department of Pathology, CWRU).

Primary cell isolation. Splenic CD4+ T cells were purified using positive selection by CD4-conjugated magnetic beads and MACS sorting (Miltenyi Biotec). Flow sorting was performed on a FACSAria (BD Biosciences) in the Flow Cytometry Core of the

Department of Pathology using fluorophore-conjugated antibodies (eBioscience).

Cell stimulation assays. For all ex vivo assays, 100,000 purified CD4+ T cells were cultured in round bottom 96-well tissue culture plates (Corning Inc.) in RPMI 1640

(Life Technologies) supplemented with 5% serum, in the presence or absence of plate-bound αCD3 (eBioscience) and/or Pam3Csk4 (InvivoGen), both at 2µg/ml unless otherwise indicated. All TLR agonists were purchased from InvivoGen, with the exception of LpqH, a kind gift from Dr. Clifford V. Harding (Department of

Pathology, CWRU), and used at the concentrations indicated. All in vitro cultures were incubated for 3 days unless otherwise indicated.

Flow Cytometry. For all flow cytometric analyses, cells were stained with the indicated antibodies (eBioscience) on ice for 30 minutes in FACS buffer (PBS, pH 7.2,

2% fetal bovine serum, 0.05% sodium azide) and washed twice with fresh FACS buffer. Sterile flow sorting was performed on a FACSAria, and standard flow analysis was performed on an Accuri C6 flow cytometer (BD Biosciences). Cytometric data were analyzed using FCS Express software v.3 (De Novo Software).

Cytokine Quantitation. Routine cytokine analysis was performed by standard sandwich ELISA according to the manufacturer’s protocols (BioLegend). Signal quantitation was performed on a Victor3V multilabeled plate reader (Perkin-Elmer).

Multiplex cytokine/chemokine analysis was performed by Luminex using a pre- designed cohort of 32 analytes and a separate TGFβ array at Eve Technologies.

Gene Expression. mRNA transcripts were isolated from resting CD4+ T cells, and then quantified were quantified using the GeneChip Mouse Transcriptome Assay 1.0

(Affymetrix) at the Case Center for Proteomics.

Data analysis. All data are expressed as mean ± standard deviation (SD), and 1- way ANOVA was performed for all pair-wise statistical analyses using GraphPad

Prism (version 5.0).

Results

TLR2 and TCR Co-Stimulation Triggers Synergistic CD4+ T cell-Intrinsic IL-10

Production

In order to define the role of TLR2 stimulation in T cell responses, CD4+ T cells were harvested from IL-10-IRES-eGFP reporter mice spleens by positive magnetic bead sorting, and stimulated with plate-bound αCD3 (2µg/ml), Pam3Csk4 (P3C; 2µg/ml), or both for 3 days and compared to unstimulated controls. eGFP-positive cells were then quantified by flow cytometry. By measuring IL-10 expression as the percentage of GFP+ cells, we found modest increases in IL-10 expression in response to αCD3 or

P3C alone (16.5% and 7.4%, respectively), but a dramatic synergistic increase to over 40% in response to receptor co-stimulation (Fig. 2A and 1B; p<0.0001; n=3).

ELISA analysis of the culture media further confirmed that TLR2 co-stimulation induced synergistic IL-10 secretion compared to the response to either agonist alone (Fig. 2C; p<0.0001; n=3). Finally, we also compared the effect of P3C with

αCD3 alone or in combination with αCD28 (1µg/ml). No difference in the IL-10 response was seen in the presence or absence of the αCD28 antibody (Fig. 2D; p<0.0001; n=3).

Given the robust response of non-T cell leukocytes to P3C and the potential for confounding effects of such cells contaminating the experiments, we measured the purity of the T cells. We found that on average, magnetic separation generated

91% T cell purity (Fig. 2E), raising the possibility that the TLR2 and IL-10 effect was not T cell intrinsic. To address this concern, we performed flow cytometry-based

27 sorting (FACS) to maximally isolate CD4+ T cells from the spleens of WT C57Bl/6 mice. The purity of the resulting population was found to be over 99% (Fig. 2F).

More importantly, the pattern of IL-10 synergy in this ultra-pure population (Fig.

2G; p<0.0001; n=3) was indistinguishable from that seen in magnetic bead-sorted cells (Fig. 2C), confirming that this effect is T cell intrinsic.

IL-10 Synergy is TLR2- and Dose-Dependent, but Independent of Age and

Gender

Although P3C is not known to stimulate pattern recognition receptors other than

TLR2, we sought to confirm that P3C co-stimulation of CD4+ T cells was mediated by

T cell-expressed TLR2. The response of WT and TLR2-/- CD4+ T cells to αCD3, P3C or both was measured as before. We found that the modest IL-10 response to αCD3 alone remains intact in both WT and TLR2-/- T cells, yet the synergistic impact of

P3C co-stimulation was completely eliminated in the TLR2-/- T cells (Fig. 3A; n=3).

With the T cell P3C receptor validated, we determined the kinetics, dose- response, age-dependence, and gender-dependence of P3C activity. We stimulated splenic CD4+ T cells from the IL-10 reporter mouse and measured the response by

GFP quantitation using flow cytometry at days 1 through 4 post-stimulation. IL-10 synergy and absolute response was greatest at 3 days post-stimulation, with a loss of synergy by day 4 (Fig. 3B; n=3).

In order to define the dose-dependence of the IL-10 response to either

28 agonist, we stimulated the cells using varied P3C or αCD3 concentrations. The titration of P3C (10ng/ml to 10µg/ml) in the presence of constant αCD3 (2µg/ml) revealed exquisite sensitivity of the IL-10 response to TLR2 agonist doses (Fig. 2C; n=3). Titrating αCD3 (50ng/ml to 2µg/ml) in the presence of constant P3C (2µg/ml) also showed dose dependence which started at 1µg/ml of TCR stimulus (Fig. 3D; n=3).

Since immunity and susceptibility to autoimmune disease is well-known to be influenced by age and gender, we analyzed the IL-10 response in CD4+ T cells from both male and female mice within an age range from 7 to 18 weeks. While female and older animal IL-10 responses were consistently higher than those of the males and young mice respectively, the magnitude of IL-10 synergy was independent of gender or age (Fig. 3E-2H; n=3).

TLR2 Co-Stimulation has Pleiotropic Effects on T cell Cytokine Production

We have demonstrated that TLR2/TCR co-stimulation triggers a dramatic and synergistic IL-10 response in CD4+ T cells; however, the impact on many other key cytokines and chemokines was unknown. To obtain a more comprehensive analysis of TLR2 stimulation in αCD3-activated T cells, we used Luminex analysis to quantify

35 secreted targets from both WT and TLR2-/- T cells (Table 1; all n=3). By Luminex, the IL-10 response measured by both GFP reporter (Fig. 2B) and standard ELISA

(Fig. 2C) was recapitulated (Fig. 4A). Interestingly, only 3 additional cytokines within the panel were upregulated synergistically in response to TLR2 co-

29 stimulation: IL-6, CXCL-10, and IL-13 (Figs. 4B-4D), of which IL-10 was produced at the highest concentration, corresponding to over 3 times the amount of IL-6 and

CXCL-10, and 10 times the amount of IL-13. With the exception of CXCL-10, which retained a modest increase upon P3C co-stimulation, all synergistic P3C responses were completely lost in TLR2-/- cells.

The remaining 31 cytokines and chemokines fell into 3 distinct categories

(Table 1; all n=3). The majority of these 31 molecules were unresponsive to any stimulation conditions, represented by IL-1β (Fig. 4E) and all three TGF-β isoforms

(Fig. 4F). A second pattern, represented by MIP-1α, was characterized by a significant response to αCD3 alone and no change with the addition of P3C or TLR2 expression (Fig. 4G). A few molecules, represented by MIP-1β, were modestly upregulated (<2 fold) in response to TLR2 co-stimulation (Fig. 4H). Finally, MIP-2 and IL-2 were outliers. MIP-2 was primarily induced, albeit to low concentrations, by P3C alone compared to any other stimulatory condition, while IL-2 was increased by TCR-only (αCD3) stimulation in T cells that express TLR2 (Fig. 4J).

TLR1/2 Heterodimer Agonists Trigger IL-10 Synergy

TLR2 is merely one of a family of TLR molecules potentially expressed by a cell. In addition, TLR2 is known to exist in three forms – TLR2 homodimers, and

TLR2/TLR1 and TLR2/TLR6 heterodimers. For a more complete analysis of TLR effects on TCR stimulation of CD4+ T cells, we used three different TLR2 agonists with specificity to TLR1/2 (P3C, LpqH) and TLR2/6 (lipoteichoic acid), along with

TLR3 (PolyI:C), TLR4 (LPS), TLR5 (flagellin), and TLR9 (CpG) agonists, both alone and in combination with αCD3 (2µg/ml) in TLR agonist titration assays with IL-10 production as the response indicator (all n=3). We found that the synergistic IL-10 response was triggered only with TLR1/2 agonists (P3C, Figs. 2-4; LpqH, Fig. 5B), whereas all other TLR agonists failed to increase IL-10 secretion over αCD3 stimulation alone (Fig. 5A, 5C-5F).

Using transcriptomic microarray analysis, we also quantified the expression level of all TLRs functionally tested. Total signal obtained for each receptor is shown in Fig. 5G, and normalized data using the canonical T cell marker and key signaling factor Lck in Fig. 5H. TLR1 expression is comparable to Lck levels and is the most highly transcribed TLR gene, whereas TLR2, TLR5, and TLR9 are transcribed at about 60% that of Lck. Finally, TLR3, TLR4, and TLR6 were found at approximately

40% of the Lck level. These data indicate that TLR2 would predominantly exist as

TLR2/TLR1 heterodimers, correlating to the dominance of TLR2/TLR1 agonists in promoting IL-10 secretion.

CD4+ Responder Populations are Non-Regulatory and Antigen-Experienced

Much of the focus in studying the role of TLR2 in CD4+ T cells has been on Treg cells.

In order to determine the identity of the T cell population responding to TCR/TLR2 co-stimulation in our experiments, we used FACS to partition splenic CD4+ cells from WT and FoxP3-RFP reporter mice into Treg subsets, including CD4+CD25+ and

CD4+FoxP3+ populations. We found that only CD4+CD25- and CD4+FoxP3- non-

31 regulatory T cells mirrored the response of bulk CD4+ T cells (Fig. 6A-6B; n=3), and that the regulatory CD25+ and FoxP3+ populations failed to respond at all. We also isolated naïve (CD62L+CD44-) and Tem (CD62L-CD44+) cells and stimulated them as before. Naïve CD4+ T cells failed to produce IL-10, whereas the Tem cells retained

IL-10 synergy (Fig. 6C; n=3).

TCR/TLR2 Co-Stimulation Expands CD25+CD45Rbhi and Tem Populations

We have shown that antigen experienced, non-regulatory T cells respond synergistically to TCR/TLR2 co-simulation, but it was still unknown whether co- stimulation might impact subset differentiation downstream. Thus, we stimulated bulk CD4+ T cells as before and then used flow cytometry to quantify surface markers associated with T cell activation and inflammatory status (CD25 and

CD45Rb; Fig. 7A) and memory T cell subsets (CD62L and CD44; Fig. 6B) after 3 days.

Most (>65%) unstimulated T cells were CD25-CD45Rbhi, similar to the profile of cells stimulated with P3C alone (>74%). αCD3 stimulation alone induced a large shift to an activated CD25+CD45Rbhi population (>83%), which increased to over

93% in response to added TLR2 co-stimulation (Fig. 7A). CD62L/CD44 profiling revealed a shift towards memory T cells (CD44+) in response to TCR/TLR2 co- stimulation, encompassing both central (CD62L+CD44+) and effector (CD62L-CD44+) memory compartments, with the majority being Tem (Fig. 7B). This response was indistinguishable from the αCD3 response alone.

TCR/TLR2 Co-Stimulation Results in Bystander Suppression

Given the total response by T cells upon TCR/TLR2 co-stimulation that includes not only high concentrations of inhibitory (IL-10) but also pro-inflammatory (e.g., IFNγ) cytokines (Table 1), we sought to determine the net impact of the resulting cytokine milieu upon bystander T cell activation. Using IFN production by freshly isolated recipient CD4+ T cells that were not exposed P3C, we used a supernatant transfer assay (diagrammed in Fig. 8A) to test the functional significance of the TLR2 co- stimulation response. To generate culture supernatants, we stimulated WT or IL-10-

/- CD4+ T cells with media only, αCD3 alone (2µg/ml), P3C alone (2µg/ml), or both agonists (each at 2µg/ml) for 1 day. Cells were rinsed thoroughly with PBS at 24 hours to remove P3C and allowed to incubate for another 3 days before harvesting the spent media, which was analyzed for IL-10 by ELISA (Fig. 8B; n=3) and transferring them to freshly harvested WT CD4+ T cells (recipients) in wells coated with 2µg/ml αCD3. To measure the suppressive capacity of the transferred supernatants, we compared the ratio of IFN produced by recipient cells in the presence of supernatant to that produced by recipient cells in the absence of transferred supernatant, which was set to 100% (Fig. 8C, dotted line) after 3 days.

Supernatants from unstimulated and αCD3-only stimulated cells did not appreciably change the IFN response from recipient cells; however, the P3C alone supernatant increased IFNγ to over 200% of the control while the TCR/TLR2 co-stimulated supernatant inhibited IFN production by the recipient cells to 50% of baseline (Fig.

8C; n=3). We also found that the IFNγ suppression by TCR/TLR2 supernatants was

33 lost when the supernatants were generated from IL-10-/- T cells (Fig. 8D; n=3), indicating that the bystander suppression of the recipient cells was dependent upon

IL-10 production. Intriguingly, in the absence of IL-10, all of the culture supernatants from stimulated T cells promoted IFNγ, suggesting that IL-10 is functionally dominant under WT conditions.

Figures

Figure 2

Figure 2. Splenic CD4+ T cells from an IL-10 (IRES-GFP) reporter were stimulated with TCR (αCD3) and TLR2 receptor agonists

(Pam3Csk4, P3C) independently or in combination at 2µg/ml each. IL-10 expression was measured on day 3 by flow cytometry as %

GFP positive. Representative histograms (A) and replicates (B) show synergistic IL-10 response to simultaneous TCR/TLR2 co- stimulation above either stimulus independently (n=3; p<0.0005). (C) WT C57Bl/6 splenic CD4+ T cells were cultured as above and

IL-10 quantified by ELISA on day 3, showing synergistic IL-10 secretion (n=3;p<0.0001). (D) WT CD4+ T cells cultured with αCD3 and/or P3C as before, with and without αCD28. IL-10 synergy is not significantly impacted by the addition of CD28 stimulation. (E)

Magnetic bead-isolated CD4+ T cells used in panels A-D by flow cytometry showed 91.2% purity. (F) Sterile flow cytometry-sorted

CD4+ T cells showed 99.2% purity. (G) IL-10 secretion of ultra-pure flow sorted CD4+ T cells (see panel F) stimulated as before, showing indistinguishable synergy (n=3;p<0.0001) from magnetic bead-isolated T cells (see panel C).

Figure 3

Figure 3. (A) CD4+ T cells from WT (gray) and TLR2-/- (white) mice were stimulated as indicated for 3 days and IL-10 quantified by

ELISA showing ablation of IL-10 synergy without TLR2. (B) IL-10-GFP reporter CD4+ T cells were used to test kinetics of the IL-10 response over 4 days. Peak synergy was seen on day 3. (C) IL-10-GFP reporter CD4+ T cells were stimulated with αCD3 (2µg/ml) and varying concentrations of P3C for 3 days and analyzed by flow cytometry, revealing strong dose-dependence on P3C. (D) WT CD4+ T cells were stimulated with P3C (2µg/ml) and varying concentrations of αCD3 for 3 days and analyzed by ELISA, also showing strong dose-dependence on αCD3. (E-H) WT CD4+ T cells were isolated from males and females of varying ages ranging from 7 weeks to 18 weeks and stimulated as before for 3 days. Overall IL-10 amplitude increased over the first several weeks, and female mice produced more IL-10 than their male counterparts, however, the degree of synergy was unchanged across age and gender.

Figure 4

Figure 4. Multianalyte analysis was performed using Luminex on 35 cytokines and chemokines on WT CD4+ T cells stimulated as before. TLR2-dependent synergistic increases in IL-10 (A), IL-6 (B), CXCL-10 (C), and IL-13 (D) were seen in TCR/TLR2 co- stimulated cells. In contrast, IL-1β (E) and TGFβ (F) showed no response to any stimulus. MIP-1α (G) was strongly induced by αCD3, but TLR2 stimulation failed to change the level of production, whereas MIP-1β showed modest synergy, albeit less than 2 fold (H).

MIP-2 showed strong responsiveness to P3C alone (I), and IL-2 was highest in αCD3-stimulated cells from WT mice (J).

Figure 5

Figure 5. Co-stimulation of WT CD4+ T cells with different TLR agonists at varying concentrations was performed for 3 days and IL-

10 quantified by ELISA. The TLR2/6-dependent agonist lipoteichoic acid (LTA) failed to increase IL-10 (A), whereas the TLR2/1- dependent LpqH (B) and P3C (not shown) showed IL-10 synergy. Similar to LTA, TLR3-dependent Poly I:C (C), TLR4-dependent

LPS (D), and TLR5-dependent flagellin (E) all failed to increase IL-10. As seen with P3C and LpqH, the TLR9-dependent CpG also induced robust IL-10 synergy (F). mRNA transcript levels of the TLR genes from freshly isolated unstimulated WT CD4+ T cells was measured by microarray. Raw fluorescence values (G) and normalized expression levels compared to the canonical T cell marker Lck

(H) are shown. TLR1, expressed at essentially the same level as Lck, is the most highly expressed TLR gene.

Figure 6

Figure 6. In order to identify the TCR/TLR2 responding populations, cells from WT or FoxP3-reporter mice were purified by FACS into varied populations, stimulated as before for 3 days and analyzed by IL-10 ELISA. Neither CD25+ (A), regulatory FoxP3+ (B), nor naïve CD62L+CD44- (C) CD4+ T cells produced IL-10 in response to stimulation, whereas CD25-, FoxP3- and CD62L-CD44+ antigen experienced CD4+ T cells showed IL-10 synergy in response to TCR/TLR2 co-stimulation.

Figure 7

Figure 7. To further characterize the phenotype of the T cells after stimulation, we measured expression of CD25 and CD45Rb (A) and CD62L and CD44 (B) by flow cytometry of cells after 3 days of stimulation. P3C-stimulated cells are most similar to T cells without stimulation (No Stim), while αCD3 induced a robust shift to CD25+CD45Rhi (85.4%) which was increased even further to

94.5% in response to P3C. Likewise, the majority of cells following stimulation with αCD3 were CD62L-CD44+, with P3C co- stimulation having little impact on this pattern.

Figure 8

Figure 8. (A) WT and IL-10-/- CD4+ T cells were stimulated at day 0 as before. After 24 hours, cells were washed to remove soluble

agonists, and then allowed to incubate for an additional 3 days. Conditioned media supernatants, which were analyzed for IL-10 by

ELISA (B), were then transferred to fresh recipient CD4+ T cells stimulated by αCD3 on day 4. On day 7, the IFN response of the

recipient cells was quantified. Normalized to the response without transferred supernatant (100%; dotted line) and subtracting IFNγ concentrations in the donor supernatants, we found that supernatants from TCR/TLR2 co-stimulated WT T cells inhibited the recipient

IFNγ response by 50%, whereas supernatants from TLR2-only stimulated T cells augmented recipient IFNγ production by over 2 fold

(C). In contrast, supernatants generated from all stimulated IL-10-/- donor T cells promoted IFNγ by the recipients by nearly 2 fold (D).

Tables

Table 1: Luminex analysis of stimulated CD4+ T cells. MACS-sorted CD4+ T cells were stimulated as indicated for 72 hours. Conditioned media was harvested and analysis by 32-plex cytokine/chemokine Luminex at Eve Technologies. Analytes were grouped by the pattern of response by the co-stimulated cells. (n=3 per stimulation)

Average pg/ml ± SD (n.d. = not detectable) Cytokine No stim αCD3 P3C αCD3+P3C

Synergistic

IL-10 1.92±0.16 384.72±44.5 20.53±1.30 2623.65±72.79

IL-6 n.d. 63.47±2.45 31.32±0.86 801.95±38.79

CXCL-10 0.84±0.30 149.84±12.77 1.44±0.05 720.76±94.24

IL-13 n.d. 36.17±10.85 n.d. 273.53±41.62

Baseline

IL-1β n.d. 3.27±0.86 1.82±0.84 1.98±1.23

TGFβ1 302.55±17.85 564.87±4.55 314.82±15.24 563.40±25.87

TGFβ2 182.1±1.03 198.97±0.71 152.36±21.42 206.99±3.565

TGFβ3 7.00±4.56 20.22±1.80 4.14±4.14 3.04±3.04

Eotaxin 0.50±0.28 1.19±0.29 n.d. 1.25±0.54

G-CSF 0.64±0.31 0.80±0.47 22.51±4.42 18.13±5.45

IL-1α n.d. 0.19±0.19 n.d. n.d.

IL-5 n.d. 102.88±24.12 n.d. 101.81±0.01

IL-7 n.d. n.d. n.d. n.d.

IL-9 7.04±0.59 22.30±3.87 6.75±0.30 27.56±0.78

IL-12(p40) n.d. n.d. n.d. n.d.

IL-12(p70) 0.32±0.32 3.06±0.29 n.d. 4.27±0.63

IL-15 0.36±0.28 1.35±0.15 0.85±0.49 5.49±0.31

KC 3.15±0.03 3.89±0.45 20.57±2.88 15.74±1.54

LIX 8.04±8.04 22.09±8.32 n.d. 54.46±46.86

MCP-1 1.60±1.60 6.73±5.94 3.41±3.41 14.87±2.21

M-CSF n.d. 2.46±1.33 n.d. 2.72±0.36

MIG n.d. 4.12±1.46 n.d. 3.55±0.36

RANTES 3.15±0.19 18.48±0.94 2.73±0.24 10.86±1.00

VEGF 0.18±0.04 30.70±1.97 0.73±0.03 70.55±2.76

Mild Co-Stim by P3C

MIP-1β 24.33±15.54 6020.27±98.26 57.18±9.19 8875.83±11.18

IFNγ 0.94±0.94 3352.25±532.28 n.d. 4908.02±0.36

LIF 1.58±0.14 417.05±32.95 2.56±0.37 675.70±29.42

TNFα n.d. 75.30±0.52 7.03±0.31 132.42±3.8

No Co-stim by P3C

MIP-1a 34.31±7.49 3101.73±14.87 57.29±11.10 3251.69±79.33

GM-CSF 2.61±2.61 1080.17±20.91 n.d. 1027.34±55.80

IL-4 0.71±0.31 409.69±100.39 0.29±0.12 465.80±8.90

IL-17 n.d. 126.84±26.05 18.24±6.21 172.08±28.10

Outliers

MIP-2 5.59±5.31 15.02±0.99 114.93±7.03 48.48±6.00

IL-2 6.65±0.09 581.87±26.22 16.08±0.98 172.29±5.50

Discussion

While TLRs have been primarily studied in innate-associated leukocyte activation, growing evidence suggests that these receptors are expressed in CD4+ T cells and intrinsically alter their behavior upon stimulation. TLR2 is of particular interest because of its reported and apparently unique ability to modulate the activity of Treg cells, which are key components in the maintenance of homeostasis and health. In this study, we demonstrate that simultaneous stimulation of TLR2 and TCR produces a profound increase in IL-10 production that is able to inhibit bystander T cell activation. In contrast to previous reports, we discovered that this effect occurs only in CD44+FoxP3- T cells, and leads to the expansion of a

CD4+CD25+CD62L-CD44+CD45Rbhi Tem population that is not typically associated with suppressive activity. These findings suggest that TLR2 represents a novel class of co-regulatory molecules on antigen-experienced, non-Treg cells which alters canonical TCR stimulation by enhancing IL-10 secretion.

IL-10 is one of the key mediators of immune regulation, a role essential to maintaining immune homeostasis. Loss or disruption of this regulatory response results in multiple inflammatory and autoimmune diseases, illustrating the importance of immune balance for overall health. Conversely, induction of the T cell- driven IL-10 response is well-known to mitigate diseases in mice ranging from asthma14,58 and IBD13 to multiple sclerosis59,60. Following this success, translational efforts have focused on autologous cell transfer immunotherapy in which IL-10- producing Tregs are isolated, expanded ex vivo, and then re-introduced into the original donor. Such an approach has shown great promise in mouse models61–63,

44 leading a burgeoning effort to develop this method for clinical use in humans64,65 for diseases including Type I diabetes66, graft-versus-host disease67, and Crohn’s

Disease68. Unfortunately, Tregs comprise a small percentage of total CD4+ T cells

(~5%) and are refractory to continuous in vitro stimulation, which is also associated with down-regulation of FoxP369. Our findings suggest an alternate approach which induced over 50% of a bulk CD4+ population of T cells to make high inhibitory concentrations of IL-10. The method does not depend on scarce Tregs, but instead on abundant FoxP3-CD25- T cells, thus bypassing the technical limitations that underlie current autologous transfer techniques.

From a homeostatic point of view and given the FoxP3 independence of this response, these data also raise the possibility that FoxP3- Treg subsets, such as Tr1 cells, may utilize TLR2 for stimulation and suppression. This is consistent with observations with well-characterized commensal flora like Bacteroides fragilis, which carry capsular polysaccharides that not only activate suppressive CD4+FoxP3-

T cells14,58 via αβTCR recognition12,70, but are also potent TLR2 agonists11. Indeed, colonization of B. fragilis appears to be directly impacted by TLR2 expression15. The identification of TLR2 as a T cell-intrinsic co-regulatory signaling receptor could hold profound new insight into the interaction and homeostatic mechanisms between the adaptive immune system and the microbiome.

It is equally important to note that while TLR2 activation concomitant with

TCR signaling generated IL-10 and bystander suppression activity in T cells, TLR2 stimulation alone in the absence of TCR engagement enhanced the IFNγ response of

45 bystander T cells. The mediator of this effect remains unknown, yet this dual-edged nature of TLR2 stimulation fits well into the broad literature on TLR2. Like other

TLRs, TLR2 stimulation has been described as both pro- and anti-inflammatory, and our findings reveal the context-dependent outcome of TLR2 stimulation in CD4+ T cells. Moreover, this holds profound implications for using TLR agonists to manipulate the balance between effector and suppressive mechanisms, such as in cancer therapeutics designed to generate tumor cytotoxicity. The logic is that CD4+ T cells comprise a large proportion of tumor infiltrating lymphocytes71, and it is well- recognized that tumor antigen-specific CD4+ T cells play a significant role in tumor immunology72–74. In light of our results, efforts in using TLR2 agonists to generate T cell-dependent anti-tumor responses75,76 need to consider that TLR2 stimulation can have an entirely opposite effect depending on whether the TCR is engaged or not.

Finally, mRNA transcript levels of the TLR family in CD4+ T cells indicate that the TLR2/TLR1 heterodimer should predominate, assuming stochastic association characteristics. TLR1 levels are higher than TLR2 and much higher than TLR6, suggesting that most TLR2 will exist in TLR2/TLR1 form. This is consistent with the robust response to TLR2/TLR1-biased agonists like P3C and LpqH, and the lack of

TLR2/TLR6-biased LTA responsiveness. Beyond the TLR2 receptor heterodimer variants, TLR9 was the only other receptor to initiate IL-10 synergy. Since the TLR9 transcript concentration was similar to TLR5, which failed to induce IL-10, the data supports the notion that both TLR2/TLR1 and TLR9 share a unique ability to act as co-regulatory receptors in T cells. It remains unclear whether TLR2 homodimers or

TLR2/TLR6 heterodimers behave as the TLR2/TLR1 heterodimer in this context due to the lower expression of both TLR2 and TLR6 compared to TLR1.

Our results demonstrate that TLR2/TLR1 heterodimer is a co-regulatory receptor on CD4+ T cells that synergistically induces IL-10 production and bystander suppressive activity in a FoxP3-independent fashion. Moreover, our findings point to the context-dependent nature of this response, such that in the absence of TCR stimulation, T cell intrinsic TLR2/TLR1 stimulation instead promotes IFNγ in adjacent T cells. The functional interactions between TLR and TCR ligation within T cells highlights a novel paradigm of immune activation and regulation that further degrades the division of innate and adaptive immunity while enhancing our understanding of key immune mechanisms such as the microbiome’s influence over homeostasis.

Chapter 3: Innate immune-directed NF-κB signaling requires site-specific

NEMO ubiquitination

Ch. 3 Summary

In this chapter, we illustrate the direct relevance of the offense-defense balance in understanding molecular phenomena in the lab. Here, we explore specific ubiquitination events on NEMO, a scaffolding component of the IKK complex. These post-translational events on NEMO are by all accounts reputed to be “offensive” by their ability to mediate inflammatory NF-kB responses. However, we show here by the offense-defense balance (where the shift from homeostasis to an inflammatory response in response to a removing a component from the system indicates that that component was defensive) that these ubiquitination events on NEMO are also profoundly defensive, given the severe basal inflammation that occurs in their absence. This illustrates the novel understanding of molecular/cellular components that underlie dysregulated inflammatory responses from the perspective of offense vs defense.

Chapter 3

Introduction

Non-traditional ubiquitination plays a critical role in innate immune signaling, and in no innate immune signaling pathway is this more well-described than in the NF-κB pathway. In response to a variety of inflammatory agonists, E3 ubiquitin ligases such as TRAF2, TRAF6, cIAP1/2 and LUBAC are activated and induce both K63-linked and linear ubiquitination of key signaling proteins such as

RIP1, RIP2, TAK1 and NEMO. These polyubiquitin chains then serve as binding surfaces to link signal transduction proteins and initiate and sustain an NF-κB response. This theme is repeated across multiple signaling pathways that culminate in NF-κB activation, including the TNF, TLR and NOD signaling pathways, and is required for physiologic inflammation and the avoidance of autoinflammatory disease (all reviewed in 77,33). Ultimately, these ubiquitin signaling cascades converge upon the IKK signalosome consisting, in its most rudimentary form, of

IKKα and IKKβ both bound to the scaffolding protein, NEMO. Within this central NF-

κB signaling hub, NEMO binds ubiquitin directly. It binds both K63-linked and linear ubiquitin chains to allow activation of the IKKs and NF-κB nuclear translocation with subsequent gene expression78–85.

Lost in the fact that NEMO’s ubiquitin binding is a central feature of inflammatory signal coordination is that NEMO, itself, is both K63 and linearly polyubiquitinated37,86,87. In a prominent, disease-related example, activation of the

Crohn’s disease susceptibility protein, NOD2, causes ubiquitination of NEMO at

Lysine-285 (K285). Loss-of-function Crohn’s disease-associated polymorphisms do not allow this ubiquitination to occur, and this loss of this ubiquitination correlates with a substantial decrease in NF-κB signaling in response to NOD2 activation37,38,88.

While NOD2 allows recognition of intracellular bacterial invasion by sensing a breakdown product of peptidoglycan called muramyl dipeptide (MDP)89,90, extracellular innate immune sensors also require NEMO ubiquitination for signaling.

LPS induces ubiquitination of NEMO at a site presumed to be K399, and if this major

LPS-induced NEMO ubiquitination site is mutated in vivo, mice fail to achieve a productive cytokine response and become hyper-resistant in LPS-sepsis models36.

Thus, while NEMO’s own ubiquitin binding capabilities have received much more attention, NEMO’s ubiquitination appears to be equally important for optimal NF-κB signaling in response to inflammatory agonists.

Like K399 ubiquitination on NEMO, NEMO ubiquitination at K285 also appears to be crucial for innate immune responses37,39,91,40. It is required for effective viral responses in tissue culture systems92, and our lab’s own work has found that, in tissue culture and in reconstitution systems, K285 ubiquitination is crucial for NOD2 signaling37,38. To date, though, while a major innate immune ubiquitination site (mouse K392 – human K399) has been mutated in the whole mouse36, the other major innate immune ubiquitination site (K285) has not been mutated in vivo. While K285 ubiquitination is crucial to NOD2 signaling in vitro and is thus suggested to be important in pathophysiologic disease states in which NOD2 signaling is dysregulated, its role in NOD2 signaling in non- overexpression/reconstitution systems remains speculative. The physiologic effect

51 of this key NEMO ubiquitination site is unknown, and the consequence of the combined lack of ubiquitination of these two major innate immune ubiquitination sites is unknown.

Given the questions surrounding NEMO ubiquitination, we generated a mouse that contained a humanized NEMO cDNA with K285 and K399 mutated to

Arginines. In previous tissue culture studies, we found that NEMO with this

K285RK399R mutation could bind to the IKKs normally, but could not signal properly in reconstituted MEF experiments38. In the current in vivo study, we find the lack of ubiquitination on K285 and K399 causes embryonic lethality in mice.

Heterozygous knockin mice develop severe skin ulcerations and inflammation that correlates with enlarged spleens, decreased splenic B cells and increased splenic

Gr1+CD11b+ cells. This embryonic lethality and inflammatory phenotype can be genetically complemented by mating these mice with TNFR1-/- null mice, suggesting a genetic interaction between the TNF pathway and NEMO ubiquitination. Despite the fact that non-ubiquitinatable NEMO no longer caused embryonic lethality, the

TNFR1-/-XNemoKiY mice show increased mortality and evidence of severe steatohepatitis. Lastly, the TNFR1-/-XNemoKiY mice are severely defective in NOD2,

TLR4 and IL-1 signaling, suggesting that ubiquitination of NEMO is essential for optimal innate immune-induced cytokine production. In sum, this work provides evidence that ubiquitination of NEMO is essential for proper functioning of the NF-

κB pathway as lack of ubiquitinated NEMO largely phenocopies NEMO-null mice. It provides evidence of complementarity of TNF signaling with NEMO ubiquitination and establishes the in vivo role of NEMO ubiquitination in innate immune signaling.

Thus, while NEMO has received much attention for its role as a ubiquitin binding protein, its own ubiquitination is also crucial for NF-κB function.

Materials and Methods

Generation of NEMO knockin mouse: These mice were generated under contract from Ingenious Targeting Labs (Stony Brook, NY). The targeting vector shown in

Figure 1A was designed to knock a cDNA containing a full length myc-tagged NEMO containing K285R and K399R into exons 2-5 of the endogenous NEMO allele. After electroporation into ES cells BA1 Hybrid;129svev/C57Bl6N and generation of stably integrated clones as evidence by both PCR and Southern blotting, these ES cells were injected into C57Bl/6 blastocysts. 10 chimeric mice were derived. All 6 of the males showed 80-90% chimerism and 2 of the females showed 80-90% chimerism.

The chimeric mice were mated and germline transmission of the knocked-in Nemo gene was confirmed by PCR. These mice were twice backcrossed onto C57BL/6 FLT mice to excise the NEO cassette and were further backcrossed 8 additional times times onto C57BL/6 before experimentation was performed. TNFR1-/- mice were obtained from Jackson Labs (Bar Harbor, ME).

Antibodies, reagents and chemicals: Myc, NEMO (rbt), IKKα, IKKβ, pIκBα, IκBα, phospho-ERK, ERK, phospho-p105, phospho-p38 and p38 were obtained from Cell

Signaling Technologies (Danvers, MA). Anti-Actin was obtained from Santa Cruz

Biotechnologies (Santa Cruz, CA). Anti-NEMO (monoclonal mouse clone C73-764) was obtained from BD Biosciences. MDP was obtained from Bachem. Ultra-pure

LPS was obtained from Invivogen. IL-1 was obtained from R&D systems. All antibodies used in flow cytometry were obtained from BioLegend.

BMDM generation, western blotting and liver function tests: Bone marrow- derived macrophages (BMDMs) from age and sex-matched mice were generated by culturing bone marrow from the indicated mouse for 7 days in 10% DMEM with

25% conditioned Ladmac medium (gift from Clifford Harding, CWRU). Media was changed to 10% FBS–DMEM overnight before use. Western blotting was performed as previously described. Liver function tests were performed on whole blood through the mouse physiology core facility at UC Davis.

Flow cytometry: Spleens were harvested from mice 2 months of age. For each experiment, age and sex-matched mice were used. Splenocytes were resuspended in

RPMI with 10% serum and passed through a 70um cell strainer before centrifugation at 1200rpm for 5 mins. Red blood cells were lysed and cells were adjusted to a concentration of 20 x 106 cells/ml. 50uL of cells (1 x 106 cells) were used for each staining condition. FcBlock was diluted according to the recommended concentration and 25ul was added to the cells for 5 minutes on ice to block non-specific binding. FcBlock, FITC anti-CD11b, APC anti-Gr1, PE anti-CD11c,

FITC anti-CD3, PE anti-CD19, FITC anti-CD4, PE anti-CD8, PE anti-Ly6G were all from eBioscience. Antibody cocktails were made fresh for each staining condition and 25ul of this was added to the cells (for a total staining volume of 100uL). Cells were stained for 1 hour on ice, washed twice with PBS and resuspended in 300uL

PBS. Data was acquired using an A6 Accuri Flow cytometer (Invitrogen) and analyzed using the CFlow software (Invitrogen).

qRT-PCR and statistical analysis: RNA was isolated and reverse transcribed using a Quantitect RT kit (Qiagen). Primer sequences are listed in Tigno-Aranjuez et al.,

2013. Sybr green was obtained from Bio-Rad, and the real-time PCRs were carried out with a CFX96 C1000 Real-Time Thermal Cycler from Bio-Rad. RT-PCR data are presented as the mean ± the standard error of the mean. RT-PCR experiments were performed in duplicate and repeated twice. Results of representative experiments are shown. The significance of the comparisons shown was assessed by Student's two-tailed t test or a Chi-squared test with the cutoff for significance set at P ≤ 0.05.

Results

To determine the in vivo role of two major innate immune NEMO ubiquitination sites, we generated a targeting construct that contained a myc-tagged human NEMO cDNA containing lysines 285 and 399 mutated to arginine. Our previous studies found that this N-terminally myc-tagged mutant NEMO bound the

IKKs identically to untagged NEMO and myc-tagged WT NEMO was able to reconstitute NF- κB signaling in NEMO-null MEFs identically to untagged NEMO37,38.

This cDNA was knocked into the endogenous NEMO locus on the X Chromosome such that it replaced Exons 2-5 (Figure 9A). After establishing germline transmission, NEMO knockin mice were mated with FLP recombinase mice to remove the NEO cassette, and were then backcrossed 10 times onto C57BL/6 to establish the XWTXNemoKi line. Evidence of protein expression (both with NEMO antibodies and via the myc tag) and evidence of germline transmission is shown in

Figure 8B. Matings aimed at generating male NEMO knockin mice showed that despite generating over 200 mice, none were XNemoKiY suggesting embryonic lethality (Figure 9C). The female heterozygous mice displayed difficult fecundancy as times between litters varied between 27 and 90 days and did not display consistency within a single mother or across mothers.

In addition to fertility difficulties, female heterozygous mice developed ulcerating skin lesions (Figure 10A). Histologically, these lesions showed hyperkeratosis and parakeratosis with elongated Rete Ridges (Figure 10A) suggestive of a psoriasis-like histology. qRT-PCR analysis of lesional skin showed elevated cytokines in the heterozygous NEMO knockin mice compared to WT

57 littermate controls (Figure 10B). This skin inflammation correlated with the appearance of a greatly enlarged spleen (Figure 10C). Quantitatively, the splenic enlargement was due to an abundance of Gr1+CD11b+ cells (Figure 10C), suggesting an inflammatory phenotype in these mice. Splenic T cell counts were similar between WT and heterozygous NEMO knockin mice while splenic B cells were decreased in the NEMO knockin mice (Figure 11). These findings are consistent with the requirement for functional NEMO in B cell development93,94. The results in the heterozygous NEMO knockin female mice closely resemble those of the heterozygous female NEMO knockout mice. These heterozygous complete knockout mice develop alopecia and inflammatory skin lesions, albeit in a much more severe manner and a much earlier age95–97 . This fact, coupled with the embryonic lethality seen with both a null allele and the K285RK399R allele, suggests that the

K285RK399R NEMO knockin allele is loss of function and further suggests that loss of these two ubiquitination sites resembles a null NEMO allele.

Due to the reproductive difficulties in the heterozygous NEMO knockin mice and because X-inactivation essentially causes chimerism in the heterozygous NEMO knockin mice, rather than performing timed matings in a line with extremely irregular estrous and birth cycles, we took an alternative approach to determine the cause of embryonic lethality. NEMO-null mice die at E12-E13 because of massive liver apoptosis due to an imbalance in TNF signaling favoring JNK-mediated apoptosis over NF-κB-induced survival95–97. Given this, we sought to determine if

TNF receptor loss would complement the embryonic lethality in the NEMO knockin mice. To this end, we mated TNFR1-/- mice to the heterozygous NEMO knockin mice

58 to generate TNFR1-/-XNemoKiY mice. While these mice were generated at less-than-

Mendelian ratios, TNFR1-/-XNemoKiY were viable, suggesting that TNFR1 loss can complement the nonubiquitinatable NEMO phenotype (Figure 12A). Both the skin lesions and hypercellularity of the spleens were not present in the TNFR-/-XNemoKiY mice, further suggesting that TNF signaling is responsible for the skin inflammation in the XWTXNemoKi mice and consistent with the results showing that the skin defect in

NEMO-null mice can be reversed by mating onto a TNFR1-null background (Nenci et al., 2006). The TNFR1-/-XNemoKiY were not entirely healthy however, as they showed increased mortality relative to either WT or TNFR1-/- mice (Figure 12B). Whole necropsy showed normal development and histology of organs with the exception of a substantially diseased liver. Histologically, the liver showed evidence of macro- and microvesicular steatosis with centrilobular and Zone 3 hepatitis with moderate fibrosis with regenerative nodules (Figure 12C). Hepatocytes showed dysplasia with prominent nucleolar vacuoles (Figure 12C). Liver function tests showed greatly elevated AST, ALT, LDH and total bilirubin suggesting hepatocyte cell death and acute hepatitis (Figure 12D). Together, these features closely resemble non- alcoholic steatohepatitis (NASH) and also closely resemble the phenotype of the hepatocyte-specific NEMO knockout mouse98–100, again suggesting that the nonubiquitinatable NEMO is a loss-of-function allele.

The generation of TNFR1-/-XNemoKiY mice also allowed us to determine the role that ubiquitination of NEMO plays in the NOD2 signaling pathway. The biochemical mapping and signaling work had previously been done in overexpression and reconstitution systems37,38, and the role that NEMO

59 ubiquitination plays in NOD2 signaling in vivo has been unclear. The generation of this mouse allowed us to test the role of K285 and K399 ubiquitination in NOD2- driven innate immune signaling. To this end, bone marrow-derived macrophages

(BMDMs) were generated from WT, TNFR1-/- and TNFR1-/-XNemoKiY mice and treated with the NOD2 agonist, MDP. Time courses showed that while p38 signaling was intact, MDP-induced NFkB signaling and ERK signaling were greatly diminished

(Figure 13A). Because ERK signaling requires the IKKs to phosphorylate p105101, these features suggest a proximal defect at the level of the IKK signalosome.

Importantly, NOD2-induction of NEMO ubiquitination was lost in the BMDMs derived from the TNFR1-/-XNemoKiY mice (Figure 13B) while binding to the IKKs was retained (Figure 15). This suggests that the replacement of these two lysine residues specifically affects ubiquitination, but leaves the general folding of NEMO intact. To then further verify this signaling defect, qRT-PCR of genes induced by

MDP was performed in these BMDMs. In all cases, the TNFR1-/-XNemoKiY BMDMs showed greatly decreased MDP-induced gene expression when stimulated by MDP

(Figure 13C). In sum, these features suggest a broad defect in NOD2 signaling when

NEMO cannot be ubiquitinated and suggest that NEMO ubiquitination is central to

NOD2 signaling.

To then further determine the role that NEMO ubiquitination plays in generating and propagating innate immune signaling, TNFR1-/-XNemoKiY BMDMs were stimulated with ultra-pure LPS. While initial levels IκBα decreased similarly

(see lanes 2 and 7 in Figure 14A), BMDMs from TNFR1-/-XNemoKiY were not able to sustain an NF-κB response as indicated by their inability to generate

60 phosphorylated IκBα at later time points of stimulation (Figure 14A). This was recapitulated by a marked inability to generate LPS-driven cytokines (Figure 14B).

Lastly, the ability of two additional innate immune stimuli to generate cytokine responses in these macrophages were determined. Limiting doses of IL-1 showed a defect in cytokine release, while IFNγ showed no specific defect. This last point is an important control. Cytokine responses to IFNγ are dependent on JAK-STAT pathways and less so on NF-kB102. The fact that the IFNγ response is intact suggests that there isn’t a broad immunologic signaling defect in these cells. Collectively, these findings suggest that NEMO ubiquitination is crucial for NF-κB-directed cytokine responses.

Figures

Figure 9

Figure 9: Knockin of a non-ubiquitinatable NEMO allele causes embryonic lethality.

A. Schematic of targeting construct used to generate a non-ubiquitinatable NEMO mouse. Exons 2-5 were replaced with a cDNA encoding NEMO in which lysines 285 and 399 were replaced with arginines such that ubiquitination cannot occur. The construct was tagged on the N-terminus with a myc tag such that it could be differentiated from endogenous NEMO. Additionally, the NEO cassette was flanked by FRT sites such that when generated, the mouse could be mated with a Flippase mouse to remove the NEO cassette.

B. Stably transfected ES clones were assayed for non-ubiquitinatable NEMO expression. An anti-myc western blot showed expression of the non-ubiquitinatable myc gene while an anti-NEMO blot showed relative expression of endogenous from knocked-in NEMO in the ES cells (upper panels). The lower panel shows germline transmission in heterozygous female mice.

C. Mendelian ratios of matings of WT male BL/6 mice with heterozygous female NEMO knockin mice. NEMO is on the X- chromosome and of over 200 mice generated, no NEMO knockin males were observed.

Figure 10

Figure 10: Non-ubiquitinatable NEMO knockin mice develop inflammatory skin lesions and splenomegaly.

A. Gross photograph of representative 8 month old non-ubiquitinatable NEMO heterozygous female compared with a WT mouse. The non-ubiquitinatable NEMO heterozygous mice develop ulcerated and plaque like lesions distributed throughout the body. Histologically, the plaque-like skin lesions show hyperkeratosis, parakeratosis, keratin pearls and a hyperproliferative epidermis with Rete ridge like protrusions into the dermis. Non-lesional skin from the non-ubiquitinatable heterozygous NEMO knockin mice show little histological change relative to WT mice.

B. qRT-PCR analysis of cytokines from skin of WT littermate or from lesional or non-lesional skin from heterozygote non- ubiquitinatable NEMO knockin mice (N=3). Cytokines are significantly elevated in both the lesional and non-lesional skin from the non-ubiquitinatable NEMO heterozygous mice. *=P<0.05, ***=P<0.001.

C. Representative gross picture showing the enlarged spleens present in the heterozygous non-ubiquitinatable NEMO female mice relative to WT female littermate control mice. Flow cytometry gating on the different subsets of splenic immune cells + + reveals that the percentage of Gr1 CD11b cells is greatly increased, suggesting an inflammatory phenotype in the heterozygous non-ubiquitinatable NEMO mice.

Figure 11

Figure 11: Normal T cell numbers but decreased B cell numbers in the spleens of XWTXNemoKi mice.

A. Flow cytometry with CD3 and CD19 antibodies shows that while CD19-CD3+ cells (T cells) are similar in number, C19+CD3- (B cells) are greatly diminished in the XWTXNemoKi mice.

B. Graphical representation of fluorescent intensity of CD19 and CD3 staining again showing diminished B cells in the XWTXNemoKi mice.

C. Absolute number of splenic B and T cells in the WT and XWTXNemoKi mice. Similar results were found in a total of 3 mice from each genotype.

Figure 12

Figure 12: Mating onto a TNFR1-deficient background rescues embryonic lethality in the non-ubiquitinatable NEMO mice but causes steatohepatitis and increased mortality.

A. C57BL/6 TNFR1-/- mice were mated with TNF+/+XWTXNemoKi mice and resulting TNFR+/-XWTXNemoki mice were mated with TNFR-/-XWTY mice to obtain TNFR1-/-XNemoKiY mice. Mendelian ratios of this mating are shown. Where in a WT TNFR1 background, we did not obtain pure NEMO knockin mice, when TNFR1 was deleted, male non-ubiquitinatable NEMO knockin mice were obtained, suggesting that TNFR1 deletion complements the non-ubiquitinatable NEMO defect.

B. Kaplan-Maier curves showing significantly decreased lifespan in the TNFR1-/-XNemoKiY and the TNFR1-/-XNemoKiXNemoKi mice.

C. Histologic evidence of steatohepatitis in the TNFR1-/-XNemoKiY mice including macro and microvesicular steatosis, inflammatory infiltrates most prominent in zone 3 of the liver and prominent nucleolar vacuolization (lower panel, nucleolar vacuolization is shown by arrows). These are all features of non-alcoholic steatohepatitis (NASH).

D. Elevated liver function tests in the TNFR1-/-XNemoKiY mice indicating significant hepatocyte injury. *=P<0.05, **=P<0.01, ***=P<0.001

Figure 13

Figure 13: Bone marrow derived macrophages (BMDMs) homozygous for non-ubiquitinatable NEMO show a severe defect in NF-κB signaling in response to the NOD2 agonist muramyl dipeptide (MDP).

A. Primary BMDMs were generated from TNFR1-/-XWTY, TNFR1-/-XNemoKiY or TNFR1+/+XWTY mice and were treated with 10 ug/mL of MDP for the indicated time period. BMDMs containing non-ubiquitinatable NEMO showed an inability to phosphorylated two IKK substrates, IκBα and p105, in response to MDP. The inability to phosphorylated p105 further leads to an inability to activate p44/p42 ERK in response to MDP.

B. Primary BMDMs were generated from TNFR1-/-XWTY or TNFR1-/-XNemoKiY mice and were treated with 10 ug/mL of MDP for the indicated time period. Lysates were generated in denaturing conditions (boiled with 1% SDS). After cooling, the SDS was diluted to 0.1%and NEMO immunoprecipitation was performed. Western blotting showed that while WT NEMO could be inducibly ubiquitinated, knockin NEMO could not.

C. To quantitate the signaling effect and to determine the role of NEMO ubiquitination in MDP-induced gene expression, the indicated BMDMs were treated with 10 mg/mL MDP for 4 hours before RNA was harvested and subjected to qRT-PCR. In all cases studied, the BMDMs expressing non-ubiquitinatable NEMO showed greatly decreased MDP-induced gene expression. *=P<0.05, **=P<0.01, ***=P<0.001

Figure 14

Figure 14: Bone marrow derived macrophages (BMDMs) homozygous for non-ubiquitinatable NEMO show a defects in TLR4 and IL-1 responses but normal interferon responses.

A. Primary BMDMs were generated from TNFR1-/-XWTY, TNFR1-/-XNemoKiY or TNFR1+/+XWTY mice and were treated with 25 ug/mL of MDP for the indicated time period. While initial levels of total IκBα dropped in all three genetic lines, only BMDMs with WT NEMO could sustain an NFkB response as indicated by phospho-IκBα and phospho-p105 blots. Erk signaling was likelwise substantially diminished.

B. To quantify and verify these signaling results, the indicated BMDMs were treated with 3 ng/mL for 4 hours. RNA was harvested and subjected to qRT-PCR. While all three genetic lines showed an increase in KC or TNF-α production, this was severely decreased in the BMDMs from TNFR1-/-XNemoKiY mice. *=P<0.05, **=P<0.01

C. The indicated BMDMs were treated with IL-1 (5 ng/mL) for 4 hours before cells were harvested and subject to qRT-PCR. IL-1-induced KC and IL-6 expression were significantly impaired in the TNFR1-/-XNemoKiY BMDMs. *=P<0.05

D. The indicated BMDMs were treated with IFN-γ (2 ng/mL) for 4 hours before cells were harvested and subject to qRT-PCR. There were no significant changes in gene expression between any of the cell lines.

Figure 15

Figure 15: NemoKi can bind to the IKKs. BMDMs from TNFR1-/-XWTY or TNFR1-/-XNemoKiY mice were treated with 10 ug/mL MDP for the indicated time period. Cells were then harvested in lysis buffer and immunoprecipitated with an anti-NEMO monoclonal antibody. Western blotting showed that NEMO bound the IKKs, suggesting that at least the N-terminal region of NEMO (IKK interaction domain) was properly folded.

Discussion

In all, these findings highlight the importance of NEMO ubiquitination in NF-

κB signaling. Genetically, NEMO ubiquitination is required for a functioning IKK signalosome as evidenced by the fact that a NEMO allele that lacks the ability to be ubiquitinated at K285 and K399 largely phenocopies NEMO loss. Both NEMO-null and NEMO K285RK399R mice show embryonic lethality due to TNF signaling and in the female heterozygous state, both develop inflammatory skin lesions95–97. Thus, in addition to helping to coordinate NF-κB signaling through the binding of ubiquitinated proteins, NEMO itself must be ubiquitinated in order to properly function in the NF- κB pathway. Failure to do so not only causes embryonic lethality and steatohepatitis, but also causes dysfunction of a number of innate immune signaling systems.

Productive inflammation is important not only for the body to fight infection and damage, but also to avoid inflammatory disease. As the NF-kB pathway represents one of the key inflammatory signaling pathways in the body, it is not surprising that there are numerous checks and balances to optimize signaling. One key point is that despite the fact that LPS-induced cytokine release is significantly compromised in the TNFR1-/-XNemoKiY BMDMs, it is not completely abrogated. There is still a 10-20 fold increase in cytokine release in these BMDMs (Figure 6B). When compared to WT BMDMs, this is substantially lower, however, the increase is still measurable. The same is true for IL-1 and MDP, albeit at a lower total induction level (Figures 6C and 7C). One interpretation of this finding is that the inability to

69 induce site-specific NEMO ubiquitination is not absolutely required for signaling. It is only required for maximal signaling. Another interpretation, suggested by the equivalent initial decrease in IκBα levels upon LPS exposure with a failure to initiate a later wave of phosphorylation of IκBα (Figure 7A), is that NEMO ubiquitination might be required to sustain NF- κ B signaling.

An additional area of interest lies in the fact that TNF receptor loss partially complements the NEMO knockin embryonic lethality. The NEMO knockin females fertility difficulties and irregular breeding precluded us from determining the cause of embryonic lethality, however, the fact that mating onto a TNFR1-/- background complemented the lethality suggests that like the NEMO-null mice, the NEMO knockin mice die of TNF-induced liver apoptosis. This is interesting for two reasons.

First, using reconstituted NEMO-null MEFs, we have previously shown that NEMO- null cells reconstituted with K285R K399R NEMO via retroviral transfer show intact

TNF signaling38. The differences between the in vitro system and the in vivo system could be due to developmental timing, cell type specificity or simple artifact due to overexpression. Secondly, the fact that TNFR1 loss complements the embryonic lethality suggests a genetic interaction. K399 ubiquitination was initially shown to be TNF-inducible by mutagenesis and reconstitution studies and further shown to be required for optimal TNF-induced NF-kB activity in tissue culture systems86. In contrast, despite the fact that K285 has been shown repeatedly to be directly ubiquitinated in mass spectrometry experiments by numerous groups37,39,91,40, K285 ubiquitination has not yet been linked in vivo to TNF signaling. While a K285R- only mouse has not been generated, the fact that the described K399R (K391R mouse)

70 does not show embryonic lethality, coupled with the fact that the K285RK399R knockin mouse’s embryonic lethality can be complemented by TNFR1 loss, implies that K285 ubiquitination may be required for optimal TNF family member signaling.

Future work will address this possibility both genetically and biochemically.

In summary, while ubiquitin chain binding by NEMO has been shown to be important in physiologic NF- κB signaling, this work shows that site-specific ubiquitination of NEMO is required for physiologic NF- κB signaling as well.

Chapter 4: Generating a novel immunologic framework

Chapter 4

Central to our current view of immunity is the separation between innate and adaptive immunity, which represent two broadly distinct mechanisms of mounting an immune response. An innate immune response is the immediate, non- specific response to microbe-associated molecular patterns (MAMPs) that signify the presence of a potential pathogen (i.e. lipopolysaccharide (LPS) or double- stranded RNA), or to danger-associated molecular patterns (DAMPs) such as heat- shock proteins or hyaluron that indicate injury to self. The non-specificity and immediacy of innate immune responses is mediated by germline-encoded receptors on front-line cells such as macrophages, dendritic cells and neutrophils, of which toll-like receptors are among the most studied and understood. On the other hand, the adaptive response, shaped by the initial innate immune response, provides antigen specificity and memory to the immune response and is mediated by antigen- specific receptors on B and T cells. Growing evidence suggests, however, that we cannot simply demarcate innate from adaptive immunity, given evidence of canonically adaptive characteristics in innate immune cells (e.g. antigen-specificity in natural killer cells), and canonically innate components in adaptive immune cells

(e.g. TLR expression and function in B and T cells). In fact, we demonstrate here a profound, synergistic IL-10 response in CD4+ T cells by T cell-intrinsic TLR2 co- stimulation, which is strikingly independent of classical regulatory CD25+ or

FoxP3+ cells. Further adding to the functional complexity and intriguing nature of

TLR2 stimulation in this adaptive immune context is IL-10-mediated bystander suppression in the presence of concomitant TCR engagement, and a complete

73 reversal of this suppression in its absence, demonstrated by a 100% increase in the pro-inflammatory cytokine IFNγ.

While the bi-directionality of traditional innate and adaptive elements already suggests that the innate vs adaptive paradigm lacks depth of meaning, we can further illustrate its insufficiency by applying the analogy of the military – in which case, conceptualizing and trying to fit the entire functional capacity of the immune system into innate vs adaptive categories is analogous to conceptualizing and trying to fit the entire functional capacity of the military into infantry vs special forces categories. This framework is not only incredibly narrow and lacking in resolution, but is also simply a description of two broadly different mechanisms of mounting a response that does not provide insight into the root purpose and fundamental nature of the immune system – meaning that the innate vs adaptive paradigm is remarkably insufficient in meaningfully framing the functional capacity of the immune system.

The question is, then, how to conceptualize a framework that better encapsulates the immune system in light of its fundamental purpose. I suggest that we use the profound parallels between the immune system and military to do so, given that the functional nature of the immune system is that of a biological military, charged with interacting with and defending us from a complex environment. The search for an appropriate framework for the immune system is, in other words, a search for a description of the form that is elemental to and frames the fullest continuum of its function. If the immune system and the military are functionally

74 homologous, then so too can we assume will the basic tenets of their form. So, to develop a novel framework by which to conceptualize the full functional capacity of the immune system, I propose that we apply the form that underlies military function to the immune system, which will provide deeper insight into immune function in a way that, in a sense, has already been experimentally validated in the real world.

While the homology of the immune system and military is intuitively evident in the language ubiquitously used to describe immune functions – “defense”,

“battle”, “combat”, “war” – this is often a superficial appreciation of their parallels.

More specifically, the immune system and military have an identical purpose: maximizing security. In the real world, maximizing national security depends on achieving two principal objectives on the continuum between peace and war on which every nation lies: 1) maintain peace, and if one must go to war, 2) win war

(Figure 1). Translating this to immunity, then, this suggests that maximal immunological security also depends on achieving two principal objectives on the continuum between homeostasis (peace) and an inflammatory response (war): 1) to maintain homeostasis, and if it must respond immunologically, 2) to mount a productive inflammatory response. Given these parallels, I suggest we can better understand how the immune system attains its objectives by understanding how nations do so, namely, through international security theory (objective 1) and warfare strategy (objective 2).

Figure 1.

Inflammatory Homeostasis response

1. Maintain peace 2. Win war

Adapted from Doctrine for the Armed Forces of the United States, JP-1, March 2013

War is the means by which nations, rival gangs, or as proposed here, the immune system, impose control and influence over enemy and unwanted/dysregulated forces and entities. Winning war (objective 2) requires defeating the enemy’s armed forces and war-making capacities, reflecting well- known immunological mechanisms such as phagocytosis, release of pro- inflammatory cytokines and anti-microbial factors (i.e. defensins), increased body temperature, iron sequestration, and antibody-mediated opsonization. Lost in this textbook list of immune strategies is the most fundamental element of warfare – which is the existence and optimal interplay of offensive and defensive elements.

That the immune system is dubbed a “defense” system obscures the fact that intrinsic to its function is a duality of both offensive and defensive capabilities that is absolutely elemental to waging successful warfare103,104, or in other words, mounting a productive immune response. Defense, at its most fundamental level, is defined by characteristics such as firepower, communication, barrier, border protection, and surveillance – reflecting immune components such as antimicrobial factors, cytokines, epithelium, cells at organ/environment interfaces (e.g. alveolar macrophages), and central memory T cells. On the other hand, offense, at its most elemental level, is defined by characteristics such as lethality, mobility, and

76 concentration, reflecting immune components such as pro-inflammatory cytokines, effector memory T cells, and the chemotactic mobilization of cells to a site of injury.

An offense vs defense perspective, then, more deeply illuminates the fundamental nature of immune responses, which is that of warfare against enemy and unwanted forces, and may provide strategic advantages in developing therapeutic mechanisms of modulating immune responses.

On the other hand, preventing going to war in the first place (objective 1) requires understanding the factors that lead to war. In national security, this requires developing appropriate theories that model the complex relationship of a nation against the international environment against which it has to protect itself.

For the past half century, a dominant theory used to understand the factors promoting war has been defensive neorealism, credited with influencing arms control negotiations, military operations, and foreign policy. The overarching premise of defensive neorealism is that nations (i.e. “separate entities that must independently procure security for themselves”) exist in a state of anarchy with one another, in that there is no “superpower” or “world government” that collectively governs all nations. Further, defensive neorealism specifically states that nations in an international environment are most secure in a defensive state, and that it is not in their best interest to pre-emptively mount aggressive, offensive responses, unless encroached upon first105–107. Intriguingly, this model of the relationship of nations in the international environment reflects the relationship of one’s immune system in the external environment – where the immune system and environment a) exist in a state of anarchy without an overarching force that co-regulates the two, and 2)

77 where the immune system is intrinsically most secure in a defensive state, in which it is not in the system’s best interest to aggressively attack and overcome the environment unless encroached upon first.

If defensive neorealism models the state-of-being of nations in an international environment, how does it specifically explain why nations move away from peace towards a state of war? It does so by the offense-defense theory, defined as the balance of the offensive and defensive capabilities of the military technology available to nations, in which a defense-favored balance promotes peace, and an offense-favored balance promotes war29,30,108. Translating this to immunity, this suggests, then, that the makeup and balance of immune offensive and defensive capabilities may have profound ramifications on the immune system’s tendency towards homeostasis or inflammation.

If the offense vs defense paradigm underscores security – encompassing the tendency towards war and warfare itself – then this warrants interpreting immunological components and phenomena from an offense vs defense perspective.

The most basic definition of defense is reflected in the epithelial barrier, alveolar macrophages that mediate border protection at the lung/environment interface, and lymph node-localized central memory T cells that surveil incoming antigens that lack immediate effector function. This is in contrast to cells such as the mobile and immediately effective, effector (i.e. offensive) memory T cells. However, the immune system and its proposed offensive and defensive capabilities are mediated not only by cellular players, but also by underlying molecular events – and like

78 military technology, just because a component is reputed to be predominantly defensive or offensive does not mean that it does not have elements of the other. For instance, the alveolar macrophages that are defensive by virtue of their surveillance of the lung-environment border also have specific offensive mechanisms (i.e. phagocytosis and oxidative burst) to fully effect their protective function.

The ubiquitous transcription factor, NF-kB, is easily relegated to the

“offensive” category for its well-reputed ability to transcriptionally promote pro- inflammatory responses necessary to eradicate pathogens and other insults. While many different receptors such as TLRs, TNFα receptor, and NODs activate NF-κB, these signals first converge at a hub, the IKK complex, consisting of two serine- threonine kinases (IKKα and IKKβ) and a scaffolding protein, NEMO. Given the centrality of the IKK complex in mediating pro-inflammatory consequences of NF-

κB, there is significant interest in understanding the molecular regulation of the IKK complex, and recently, ubiquitination of NEMO on at least two lysines (positions 285 and 399) was found to be associated with pro-inflammatory NF-kB responses downstream of diverse receptor activation37,86,87. However, what is easily overlooked in this focus on the offensive capabilities of NF-kB is the possibility of defensive capabilities, which would have new ramifications on the meaning of molecular events at the IKK complex. We can, in fact, experimentally test the nature of immune components using the offense-defense balance, which states that if the removal of a component changes the immune state from homeostasis to inflammation (either to an exogenous stimulus, or against self in the absence of any stimulus), then that component was defensive in nature, enough to shift the balance

79 to the offense in its absence. In fact, removing the ubiquitination events at K285 and

K399 on NEMO results in a dramatic inflammatory phenotype in the mouse, including severe skin inflammation, splenomegaly, and steatohepatitis, suggesting that these post-translational modifications, otherwise known to impart offensive capabilities to NEMO, are in fact also robustly defensive and necessary for homeostasis in multiple immunological capacities.

Beyond interpreting known immune components from the offense vs defense perspective, this paradigm can be used to illuminate deeper insights into immunity.

Perhaps the most obvious inference is the significance of the immunological offense- defense balance in key immune states, whether in health or disease. That homeostasis is promoted by a sufficiently defense-shifted balance suggests that this is also the case in commensalism, a specific type of homeostasis between microbial species and host tissues – and given our scant understanding of how commensalism occurs, perhaps this interpretation provides additional mechanistic clues that focus our attention specifically towards defensive characteristics of host and microbe in each commensal context (e.g. suppressive cytokine production, upregulation of suppressive surface molecules, decreased mobilization, to name an obvious few). On the other hand, an offense-shifted balance promotes inflammation not only in health, but also in disease (i.e. due to an insufficient defense), misdirected towards self as in autoimmunity – suggesting that bolstering the deficient defense underlying an offense-heavy balance may mitigate this tendency towards immune war.

Similarly, immunodeficiencies can generally be described as a defective ability to

80 shift to an offense-heavy balance as needed, due typically to defective offensive immune components (e.g. B and T cell and phagocytic deficiencies).

While the offense vs defense paradigm illuminates immune reactions in their most fundamental light (i.e. as immune warfare), this perspective may be of particular value in the development of therapeutics, which is simply an extrinsic means of modulating immune reactions. Interpreting disease mechanisms, such as cancer, through a lens of offense vs defense can provide a strategic benefit that would be impossible from an innate vs adaptive perspective. While the mainstay of cancer therapy remains chemotherapy, radiation, and surgery, all with significant limitations, what could benefit the growing focus on targeted therapies is a strategic understanding of cancer from the perspective of waging war against it, possible only through the fundamental elements of offense vs defense, and not simply a mechanistic understanding. While metastasis is the predominant cause of death, and while it is clearly important to understand this offensive (i.e. mobile, aggressive, lethal) element of cancer progression, the ability of the tumor to survive in the first place in a host environment otherwise promoting its eradication (i.e. the hallmark tenacity of cancer) is in fact defense-dominant (i.e. by immune evasion). In warfare, it is strategically ideal to attack the enemy’s strength. If defense is cancer’s primary strength, this suggests it may be worthwhile, given a choice, to focus therapeutic efforts on defensive elements (e.g. communication between tumor and host cells, recruitment of elements necessary for tumor survival such as suppressive immune cells and tumor-associated macrophages, firepower such as cytokines promoting the pro-tumor nature of infiltrating immune cells, its mechanisms of adaptation and

81 evasion, etc.), rather than on classical, but more offensive mechanisms of cancer such as angiogenesis and metastasis.

The offense vs defense paradigm to the immune system not only provides a new perspective on immune responses that more accurately reflects its nature as a type of warfare, but also describes an immune offense-defense balance that promotes different states, whether of homeostasis or inflammation. However, by interpreting the model from the other direction – that the immunological state reflects the offense-defense balance of the system – we arrive at a significantly more profound inference. The concept that the immune response is a reflection of the underlying offense-defense balance may be most clearly illustrated by the “simple” inflammatory response (e.g. mediated by neutrophils and macrophages) targeted against the influx of microbes through a cut in the skin. This inflammatory response reflects an offense-shifted balance in that immune context due to the loss of the heavily defensive epithelial barrier (cut skin); the same makeup of microbes and exogenous stimuli in the context of a defense-shifted balance (i.e. intact skin) would not trigger such an inflammatory response.

What this illustrates is a principle in which an immune response (whether inflammation or homeostasis) is in fact a reflection of the state of self (i.e. the offense-defense balance). This is perhaps even more illustrative with an example at the macroscopic scale, where the response of two different people to the same stimulus can be entirely different, reflecting not the nature of the stimulus, but instead the state of self of the person. This principle, then, has profound

82 implications on what an immune response to a stimulus actually means. To date, our understanding of the immune response to exogenous stimuli has been deeply embedded in the concepts of MAMPs (microbe-associated molecular patterns) and

DAMPs (danger-associated molecular patterns)109,2. We often take these models of triggering immune responses (in which the immune system responds to specific molecular indicators) for granted and don’t appreciate that the concept of why an immune response is initiated at all is still poorly understood. Attempts to understand the phenomena of discriminatory immune responses have primarily focused on the stimulus side of the equation – for instance, the DAMP theory (in which the immune system responds to specific indicators of damage-to-self) is used to explain why the immune system rejects foreign substances such as grafts (in which surgical and/or ischemic damage is unavoidable), and yet is non-responsive to foreign entities that do not mediate damage, such as fetuses.

The importance of truly understanding the nature of an immune response is reflected in the influence this has on the way we design therapies – for instance,

Polly Matzinger suggests that to “induce acceptance of transplants without lifelong immunosuppression, we should mimic the body’s own way of inducing tolerance, i.e., by blocking the endogenous alarm and/or costimulatory signals.”2 This linear, one-way immunological view of nature of stimulus  immune response is further reflected in more recent hypotheses that an immune response requires certain combinations of MAMPs and not just the existence of a single MAMP, or that living and dying bacteria produce distinct MAMPs (i.e. “MAMPs-per vita” vs “MAMPs- postmortem”) indicating different levels of urgency of an immune response, or that

83 the cellular locations of MAMPs differentially influence the immune response110,111.

In other words, the attempt to explain why the immune system mounts a response has been hyper-focused on decrypting the nature of the stimulus. What if, on the other hand, we are missing a piece of the puzzle and the immune response is in fact the outcome of a two-way street that is not simply a consequence of the nature of the stimulus, whether they be MAMPs or DAMPs, in combination or not? What if intrinsic to the immune system’s decision to respond (whether at the level of a cell or at the whole organism) is not only the nature of the stimulus, but also the nature of the immune state-of-self, such as its offense-defense balance? If so, the reason why the immune system responds to grafts and not to fetuses may be due not simply to the presence or absence of danger molecules, but to a fundamentally different state-of-immune-self at the site of engraftment compared to the uterine environment. Extrapolating one step further, this may insinuate the reason for certain disease states. What if, genetic and environmental factors being equal, the existence and progression of cancer is due not only to its own intrinsic, insidious mechanisms, but to the state-of-self of the host? What if a disease state is a reflection not just of a particular set of molecular and cellular aberrations, but of a certain weakness in the host necessary for the disease to take hold? Perhaps, just as important as developing effective therapeutics targeted at specific disease mechanisms is understanding how to maximize and maintain an optimal state-of- self of the host.

By immunologically translating how nations achieve security in the real world, we can come to a different and deeper understanding of immunity whose

84 function it is to achieve host security. Defensive neorealism and its intrinsic offense- defense theory are used to understand the likelihood of war – not only that of nations, tribes, and gangs, but also now of the immune system. Additionally, warfare, or in our case, immune responses, is a complicated affair most fundamentally based on the interplay of offense vs defense. This suggests that immunity is at its root a matter of offensive and defensive entities, the makeup of which would have profound ramifications on the immune system’s state of being on the continuum between homeostasis and inflammation – or in other words, immune security. Inferring from these real-world models leads us to novel reflections on the nature of immunity – whether in health, disease, or as a target of extrinsic, therapeutic measures. It also illustrates the value that comes of building bridges between related and yet normally entirely disparate entities – the military and immunology – and applying relevant principles to one another.

References

1. Ulevitch, A. A. and R. J. Toll-like receptors in the induction of the innate immune response. Nature 406, 782-7, (2000).

2. Matzinger, P. The danger model: a renewed sense of self. Science 296, 301– 305 (2002).

3. Sun Jin, M. et al. Crystal Structure of the TLR1-TLR2 Heterodimer Induced by Binding of a Tri-Acylated Lipopeptide. doi:10.1016/j.cell.2007.09.008

4. Kang, J. Y. et al. Recognition of Lipopeptide Patterns by Toll-like Receptor 2- Toll-like Receptor 6 Heterodimer. Immunity 31, 873–884 5. Farhat, K. et al. Heterodimerization of TLR2 with TLR1 or TLR6 expands the ligand spectrum but does not lead to differential signaling. doi:10.1189/jlb.0807586 6. D’Andrea A, Aste-Amezaga M, Valiante NM, Ma X, Kubin M, T. G. Interleukin 10 (IL-10) inhibits human lymphocyte interferon gamma-production by suppressing natural killer cell stimulatory factor/IL-12 synthesis in accessory cells. J Exp Med 178(3):104, (1993).

7. Mege JL, Meghari S, Honstettre A, Capo C, R. D. The two faces of interleukin 10 in human infectious diseases. Lancet Infect Dis. 6(9):557-6, (2006).

8. Jang S, Uematsu S, Akira S, S. P. IL-6 and IL-10 induction from dendritic cells in response to Mycobacterium tuberculosis is predominantly dependent on TLR2-mediated recognition. J Immunol. 173(5):339, (2004).

9. Kullberg, B. J. et al. Toll-Like Receptor 2 Suppresses Immunity Toll-Like Receptor 2 Suppresses Immunity against Candida albicans through Induction of IL-10 and Regulatory T Cells. J Immunol Ref. J. Immunol. Clevel. Heal. Sci. Libr. 172, 3712–3718 (2004).

10. Sing A, Reithmeier-Rost D, Granfors K, Hill J, Roggenkamp A, H. J. A hypervariable N-terminal region of Yersinia LcrV determines Toll-like receptor 2-mediated IL-10 induction and mouse virulence. Proc Natl Acad Sci U S A. 102(44):16, (2005).

11. Wang Q, McLoughlin RM, Cobb BA, Charrel-Dennis M, Zaleski KJ, Golenbock D, Tzianabos AO, K. D. A bacterial carbohydrate links innate and adaptive responses through Toll-like receptor 2. J Exp Med 203(13):28, (2006). 12. Cobb, B. A., Wang, Q., Tzianabos, A. O. & Kasper, D. L. Polysaccharide Processing and Presentation by the MHCII Pathway. Cell 117, 677–687 (2004).

13. Mazmanian, S. K., Round, J. L. & Kasper, D. L. ARTICLES A microbial symbiosis factor prevents intestinal inflammatory disease. doi:10.1038/nature07008

14. Johnson, J. L., Jones, M. B. & Cobb, B. A. Bacterial capsular polysaccharide prevents the onset of asthma through T-cell activation. doi:10.1093/glycob/cwu117

15. Round JL, Lee SM, Li J, Tran G, Jabri B, Chatila TA, M. S. The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science. 332(6032):, (2011).

16. Silke Paust, Balimkiz Senman, and U. H. von A. Adaptive Immune Responses Mediated by Natural Killer Cells. Immunol Rev 2011,

17. Joseph C. Sun, J. N. B. & L. L. L. Adaptive immune features of natural killer cells. Nature 457, 557-5, (2009).

18. Suzuki, N. et al. Effector Functions Cutting Edge: TLR2 Directly Triggers Th1. J Immunol Ref. J. Immunol. Clevel. Heal. Sci. Libr. Oct. Clevel. Heal. Sci. Libr. 178, 6715–6719 (2007).

19. Lee, S.-M., Joo, Y.-D. & Seo, S.-K. Expression and Function of TLR2 on CD4 Versus CD8 T Cells. Immune Netw. (2009). doi:10.4110/in.2009.9.4.127 20. Nawijn, M. C. et al. TLR-2 Activation Induces Regulatory T Cells and Long- Term Suppression of Asthma Manifestations in Mice. (2013). doi:10.1371/journal.pone.0055307

21. Reba, S. M. et al. TLR2 engagement on CD4 + T cells enhances effector functions and protective responses to Mycobacterium tuberculosis. Eur. J. Immunol 44, 1410–1421 (2014). 22. Akhade, A. S. & Qadri, A. T-cell receptor activation of human CD4(+) T cells shifts the innate TLR response from CXCL8(hi) IFN-γ(null) to CXCL8(lo) IFN- γ(hi). Eur. J. Immunol. 45, 2628–37 (2015). 23. Crellin, N. K. et al. Human CD4+ T cells express TLR5 and its ligand flagellin enhances the suppressive capacity and expression of FOXP3 in CD4+CD25+ T regulatory cells. J. Immunol. 175, 8051–8059 (2005).

24. Caramalho, I. et al. Regulatory T Cells Selectively Express Toll-like Receptors and Are Activated by Lipopolysaccharide. J. Exp. Med. J. Exp. Med 02403900, 403–411 (2003).

25. Hou, Z. H. and B. TLR signaling in B-cell development and activation. Cell. Mol. Immunol. 10, 103–10, (2013). 26. Gururajan M, Jacob J, P. B. Toll-like receptor expression and responsiveness of distinct murine splenic and mucosal B-cell subsets. PLoS One 2(9):e863.,

(2007).

27. Meyer-Bahlburg A1, Bandaranayake AD, Andrews SF, R. D. Reduced c-myc expression levels limit follicular mature B cell cycling in response to TLR signals. J Immunol 182(7):406, (2009).

28. Lynn-Jones, S. M. Lynn-Jones - Offense-Defense Theory and Its Critics.pdf. Security Studies 4, 660–691 (1995). 29. Glaser, C. L. & Kaufmann, C. What is the offense-defense balance and can we measure it ?( Offense , Defense , and International Politics ) What is the offense-defense balance and can we measure it ?( Offense , Defense , and International Politics ). 44, 1–23 (1998).

30. Nilsson, M. Offense-Defense Balance, War Duration, and the Security Dilemma. J. Conflict Resolut. 56, 467–489 (2012).

31. Keir A. Lieber. Grasping the Technological Peace. Int. Secur. Vol 25, (2000). 32. Levy, J. S. The Offensive/Defensive Balance of Military Technology: A Theoretical and Historical Analysis. Int. Stud. Q. Vol. 28, (1984).

33. Hayden MS, G. S. NF-κB, the first quarter-century: remarkable progress and outstanding questions. Genes Dev 26(3):203-, (2012).

34. Hayden, M. S. & Ghosh, S. Signaling to NF-kappaB. Genes Dev. 18, 2195–224 (2004).

35. Israël, A. The IKK complex, a central regulator of NF-kappaB activation. Cold Spring Harb. Perspect. Biol. 2, a000158 (2010). 36. Ni CY, Wu ZH, Florence WC, Parekh VV, Arrate MP, Pierce S, Schweitzer B, Van Kaer L, Joyce S, Miyamoto S, Ballard DW, O. E. Cutting edge: K63-linked polyubiquitination of NEMO modulates TLR signaling and inflammation in vivo. J Immunol 180(11):71, (2008).

37. Abbott, D. W., Wilkins, A., Asara, J. M. & Cantley, L. C. The Crohn ’ s Disease Protein , NOD2 , Requires RIP2 in Order to Induce Ubiquitinylation of a Novel Site on NEMO. 14, 2217–2227 (2004).

38. Abbott, D. W. et al. Coordinated regulation of Toll-like receptor and NOD2 signaling by K63-linked polyubiquitin chains. Mol. Cell. Biol. 27, 6012–25 (2007).

39. Hinz M, Stilmann M, Arslan SÇ, Khanna KK, Dittmar G, S. C. A cytoplasmic ATM-TRAF6-cIAP1 module links nuclear DNA damage signaling to ubiquitin- mediated NF-κB activation. Mol Cell. 40(1):63-7, (2010).

40. Niu J, Shi Y, Iwai K, W. Z. LUBAC regulates NF-κB activation upon genotoxic stress by promoting linear ubiquitination of NEMO. EMBO J 30(18):374,

(2011).

41. Janeway, C. A. & Medzhitov, R. I NNATE I MMUNE R ECOGNITION. Annu. Rev. Immunol. 20, 197–216 (2002). 42. Medzhitov, R. Toll-like receptors and innate immunity. Nat. Rev. Immunol. 1, 135–145 (2001). 43. Huppa JB, D. M. T-cell-antigen recognition and the immunological synapse. Nat Rev Immunol. 3(12):973-,

44. Lenschow, D. J., Walunas, T. L. & Bluestone, J. A. CD28/B7 SYSTEM OF T CELL COSTIMULATION. Annu. Rev. Immunol 14, 233–58 (1996).

45. CS Hsieh, SE Macatonia, CS Tripp, SF Wolf, A O’Garra, K. M. Development of TH1 CD4+ T cells through IL-12 produced by Listeria-induced macrophages. Science (80-. ). Vol. 260, , (1996).

46. Macatonia SE, Hosken NA, Litton M, Vieira P, Hsieh CS, Culpepper JA, Wysocka M, Trinchieri G, Murphy KM, O. A. Dendritic cells produce IL-12 and direct the development of Th1 cells from naive CD4+ T cells. J Immunol. 154(10):50, (1995). 47. Manetti R, Parronchi P, Giudizi MG, Piccinni MP, Maggi E, Trinchieri G, R. S. Natural killer cell stimulatory factor (interleukin 12 [IL-12]) induces T helper type 1 (Th1)-specific immune responses and inhibits the development of IL-4- producing Th cells. J Exp Med. 177(4):119,

48. Turka, L. A., Gelman, A. E., Zhang, J. & Choi, Y. T Cell Survival + Activated CD4 Toll-Like Receptor Ligands Directly Promote Toll-Like Receptor Ligands Directly Promote Activated CD4 T Cell Survival. J Immunol J. Immunol. Clevel. Heal. Sci. Libr. 172, 6065–6073 (2004).

49. Jeannin, D. et al. Direct Stimulation of Human T Cells via Direct Stimulation of Human T Cells via TLR5 and TLR7/8: Flagellin and R-848 Up-Regulate Proliferation and IFN-␥ Production by Memory CD4 ␥ T Cells 1. J Immunol Ref. J. Immunol. Clevel. Heal. Sci. Libr. 175, 1551–1557 (2005).

50. Platzbecker U, Stoehlmacher J, Pabst C, Goekkurt E, Oelschlägel U, Schirutschke H, Hölig K, Theuser C, Mogck U, Ehninger G, B. M. Induction of Toll-like receptor 2 and 4 expression on CD4+ and CD8+ T cells in G-CSF- mobilized unrelated peripheral blood stem cell grafts during leukapheresis: impact on patient outcome. Leukemia 22(7):1438, 51. Reynolds AD, Stone DK, Hutter JA, Benner EJ, Mosley RL, G. H. Regulatory T cells attenuate Th17 cell-mediated nigrostriatal dopaminergic neurodegeneration in a model of Parkinson’s disease. J Immunol. 184(5):226,

52. González-Navajas JM, Fine S, Law J, Datta SK, Nguyen KP, Yu M, Corr M,

Katakura K, Eckman L, Lee J, R. E. TLR4 signaling in effector CD4+ T cells regulates TCR activation and experimental colitis in mice. J Clin Invest. 120(2):570, (2010).

53. Dillon S, Agrawal A, Van Dyke T, Landreth G, McCauley L, Koh A, Maliszewski C, Akira S, P. B. A Toll-like receptor 2 ligand stimulates Th2 responses in vivo, via induction of extracellular signal-regulated kinase mitogen-activated protein kinase and c-Fos in dendritic cells. J Immunol. 172(8):473, (2004). 54. Liu, H., Komai-Koma, M., Xu, D., Liew, F. Y. & Moncada, S. Toll-like receptor 2 signaling modulates the functions of CD4. 55. Sutmuller RP, den Brok MH, Kramer M, Bennink EJ, Toonen LW, Kullberg BJ, Joosten LA, Akira S, Netea MG, A. G. Toll-like receptor 2 controls expansion and function of regulatory T cells. J Clin Invest. 116(2):485, (2006). 56. Qian Chen, S., Davidson, T. S., Huter, E. N., Chen, Q. & Shevach, E. M. Cells, but Promotes Their Survival Suppressor Function of Mouse Regulatory T Engagement of TLR2 Does not Reverse the Engagement of TLR2 Does not Reverse the Suppressor Function of Mouse Regulatory T Cells, but Promotes Their Survival 1. J Immunol Ref. J. Immunol. Clevel. Heal. Sci. Libr. (2014). doi:10.4049/jimmunol.0901465

57. Zanin - Zhorov, A. & Cahalon, L. Heat shock protein 60 enhances CD4 + CD25 + regulatory T cell function via innate TLR2 signaling. 116, (2022).

58. Kreisman, L. S. C. & Cobb, B. A. Glycoantigens Induce Human Peripheral Tr1 Cell Differentiation with Gut-homing Specialization *. (2011). doi:10.1074/jbc.M110.206011

59. Ochoa-Rep, J. et al. A polysaccharide from the human commensal Bacteroides fragilis protects against CNS demyelinating disease. Mucosal Immunol. 3, (2010).

60. Ochoa-Repáraz J, Mielcarz DW, Ditrio LE, Burroughs AR, Begum-Haque S, Dasgupta S, Kasper DL, K. L. Central nervous system demyelinating disease protection by the human commensal Bacteroides fragilis depends on polysaccharide A expression. J Immunol. 185(7):410, (2010). 61. Lapierre P, Béland K, Yang R, A. F. Adoptive transfer of ex vivo expanded regulatory T cells in an autoimmune hepatitis murine model restores peripheral tolerance. Hepatology 57(1):217-, (2013). 62. Sarkar D1, Biswas M1, Liao G2, Seay HR3, Perrin GQ1, Markusic DM1, Hoffman BE1, Brusko TM3, Terhorst C2, H. R. Ex Vivo Expanded Autologous Polyclonal Regulatory T Cells Suppress Inhibitor Formation in Hemophilia. Mol Ther Methods Clin Dev 1. pii: 14, (2014). 63. Tanaka H, Zhang W, Yang GX, Ando Y, Tomiyama T, Tsuneyama K, Leung P,

Coppel RL, Ansari AA, Lian ZX, Ridgway WM, Joh T, G. M. Successful immunotherapy of autoimmune cholangitis by adoptive transfer of forkhead box protein 3(+) regulatory T cells. Clin Exp Immunol. 178(2):253, (2014).

64. Bacchetta R1, Lucarelli B1, Sartirana C1, Gregori S1, Lupo Stanghellini MT2, Miqueu P3, Tomiuk S4, Hernandez-Fuentes M5, Gianolini ME1, Greco R2, Bernardi M2, Zappone E2, Rossini S6, Janssen U4, Ambrosi A7, Salomoni M8, Peccatori J2, Ciceri F2, R. M. Immunological Outcome in Haploidentical-HSC Transplanted Patients Treated with IL-10-Anergized Donor T Cells. Front Immunol 5:16., (2014).

65. Edinger M, H. P. Regulatory T cells in stem cell transplantation: strategies and first clinical experiences. Curr Opin Immunol 23(5):679-, (2011).

66. Marek-Trzonkowska N1, Mysliwiec M, Dobyszuk A, Grabowska M, Techmanska I, Juscinska J, Wujtewicz MA, Witkowski P, Mlynarski W, Balcerska A, Mysliwska J, T. P. Administration of CD4+CD25highCD127- regulatory T cells preserves β-cell function in type 1 diabetes in children. Diabetes Care 35(9):1817, (2012). 67. Trzonkowski P, Bieniaszewska M, Juścińska J, Dobyszuk A, Krzystyniak A, Marek N, Myśliwska J, H. A. First-in-man clinical results of the treatment of patients with graft versus host disease with human ex vivo expanded CD4+CD25+CD127- T regulatory cells. Clin Immunol 133(1):22-, (2009). 68. Desreumaux P1, Foussat A, Allez M, Beaugerie L, Hébuterne X, Bouhnik Y, Nachury M, Brun V, Bastian H, Belmonte N, Ticchioni M, Duchange A, Morel- Mandrino P, Neveu V, Clerget-Chossat N, Forte M, C. J. Safety and efficacy of antigen-specific regulatory T-cell therapy for patients with refractory Crohn’s disease. Gastroenterology 143(5):120, (2012).

69. Hoffmann P, Boeld TJ, Eder R, Huehn J, Floess S, Wieczorek G, Olek S, Dietmaier W, Andreesen R, E. M. Loss of FOXP3 expression in natural human CD4+CD25+ regulatory T cells upon repetitive in vitro stimulation. Eur J Immunol. 39(4):1088, (2009). 70. Johnson, J. L., Jones, M. B. & Cobb, B. A. Polysaccharide A from the Capsule of Bacteroides fragilis Induces Clonal CD4 ␥ T Cell Expansion *. (2014). doi:10.1074/jbc.M114.621771

71. Protti, M. P., Monte, L. D. & Lullo, G. D. Tumor antigen-specific CD4+ T cells in cancer immunity: From antigen identification to tumor prognosis and development of therapeutic strategies. Tissue Antigens 83, 237–246 (2014). 72. Pardoll DM, T. S. The role of CD4+ T cell responses in antitumor immunity. Curr Opin Immunol 10(5):588-, (1998). 73. SL., T. MHC class II restricted tumor antigens and the role of CD4+ T cells in

cancer immunotherapy. Curr Opin Immunol. 6(5):741-5, (1994).

74. Matsuzaki, J. et al. Direct tumor recognition by a human CD4(+) T-cell subset potently mediates tumor growth inhibition and orchestrates anti-tumor immune responses. Sci. Rep. 5, 14896 (2015).

75. Zhang, Y. et al. TLR1/TLR2 agonist induces tumor regression by reciprocal modulation of effector and regulatory T cells. J. Immunol. 186, 1963–9 (2011). 76. Asprodites N, Zheng L, Geng D, Velasco-Gonzalez C, Sanchez-Perez L, D. E. Engagement of Toll-like receptor-2 on cytotoxic T-lymphocytes occurs in vivo and augments antitumor activity. FASEB J 22(10):362, (2008). 77. ZJ, C. Ubiquitination in signaling to and activation of IKK. Immunol Rev. 246(1):95-, (2012).

78. Wu, C.-J., Conze, D. B., Li, T., Srinivasula, S. M. & Ashwell, J. D. Sensing of Lys 63-linked polyubiquitination by NEMO is a key event in NF-kappaB activation [corrected]. Nat. Cell Biol. 8, 398–406 (2006). 79. Cordier, F. et al. The zinc finger of NEMO is a functional ubiquitin-binding domain. J. Biol. Chem. 284, 2902–7 (2009). 80. Laplantine, E. et al. NEMO specifically recognizes K63-linked poly-ubiquitin chains through a new bipartite ubiquitin-binding domain. EMBO J. 28, 2885– 95 (2009). 81. Lo, Y.-C. et al. Structural basis for recognition of diubiquitins by NEMO. Mol. Cell 33, 602–15 (2009).

82. Rahighi, S. et al. Specific recognition of linear ubiquitin chains by NEMO is important for NF-kappaB activation. Cell 136, 1098–109 (2009).

83. Hadian, K. et al. NF-κB essential modulator (NEMO) interaction with linear and lys-63 ubiquitin chains contributes to NF-κB activation. J. Biol. Chem. 286, 26107–17 (2011).

84. Ikeda F, Rahighi S, Wakatsuki S, D. I. Selective binding of linear ubiquitin chains to NEMO in NF-kappaB activation. Adv Exp Med Biol 691:107-14, (2011).

85. Kensche, T. et al. Analysis of nuclear factor-κB (NF-κB) essential modulator (NEMO) binding to linear and lysine-linked ubiquitin chains and its role in the activation of NF-κB. J. Biol. Chem. 287, 23626–34 (2012).

86. Zhou, H. et al. Bcl10 activates the NF- k B pathway through ubiquitination of NEMO. 167–171 (2004). doi:10.1038/nature02269.1.

87. Tokunaga, F. et al. Involvement of linear polyubiquitylation of NEMO in NF ‑ κB activation. 11, (2009).

88. Chen, G., Shaw, M. H., Kim, Y.-G. & Nuñez, G. NOD-like receptors: role in innate immunity and inflammatory disease. Annu. Rev. Pathol. 4, 365–98 (2009).

89. Girardin, S. E. et al. Nod2 is a general sensor of peptidoglycan through muramyl dipeptide (MDP) detection. J. Biol. Chem. 278, 8869–72 (2003).

90. Inohara, N. et al. Host recognition of bacterial muramyl dipeptide mediated through NOD2. Implications for Crohn’s disease. J. Biol. Chem. 278, 5509–12 (2003).

91. Kim W, Bennett EJ, Huttlin EL, Guo A, Li J, Possemato A, Sowa ME, Rad R, Rush J, Comb MJ, Harper JW, G. S. Systematic and quantitative assessment of the ubiquitin-modified proteome. Mol Cell. 44(2):325-, (2011).

92. Zhao T, Yang L, Sun Q, Arguello M, Ballard DW, Hiscott J, L. R. The NEMO adaptor bridges the nuclear factor-kappaB and interferon regulatory factor signaling pathways. Nat Immunol. 8(6):592-6, (2007).

93. Kim, S., La Motte-Mohs, R. N. a, Rudolph, D., Zuniga-Pflucker, J. C. & Mak, T. W. The role of nuclear factor-kappaB essential modulator (NEMO) in B cell development and survival. Proc. Natl. Acad. Sci. U. S. A. 100, 1203–8 (2003).

94. Derudder E, Cadera EJ, Vahl JC, Wang J, Fox CJ, Zha S, van Loo G, Pasparakis M, Schlissel MS, Schmidt-Supprian M, R. K. Development of immunoglobulin lambda-chain-positive B cells, but not editing of immunoglobulin kappa-chain, depends on NF-kappaB signals. Nat Immunol 10(6):647-, (2009).

95. Makris, C. et al. Female mice heterozygous for IKK gamma/NEMO deficiencies develop a dermatopathy similar to the human X-linked disorder incontinentia pigmenti. Mol. Cell 5, 969–79 (2000). 96. Rudolph, D. et al. Severe liver degeneration and lack of NF- ␥ B activation in NEMO / IKK ␥ -deficient mice. 854–862 (2000). doi:10.1101/gad.14.7.854

97. Schmidt-Supprian, M. et al. NEMO/IKK gamma-deficient mice model incontinentia pigmenti. Mol. Cell 5, 981–92 (2000).

98. Beraza N, Lüdde T, Assmus U, Roskams T, Vander Borght S, T. C. Hepatocyte- specific IKK gamma/NEMO expression determines the degree of liver injury. Gastroenterology 132(7):250, (2007).

99. Luedde T, Beraza N, Kotsikoris V, van Loo G, Nenci A, De Vos R, Roskams T, Trautwein C, P. M. Deletion of NEMO/IKKgamma in liver parenchymal cells causes steatohepatitis and hepatocellular carcinoma. Cancer Cell 11(2):119-, (2007).

100. Beraza N, Malato Y, Sander LE, Al-Masaoudi M, Freimuth J, Riethmacher D, Gores GJ, Roskams T, Liedtke C, T. C. Hepatocyte-specific NEMO deletion promotes NK/NKT cell- and TRAIL-dependent liver damage. J Exp Med

206(8):172, (2009).

101. Yang HT, Papoutsopoulou S, Belich M, Brender C, Janzen J, Gantke T, Handley M, L. S. Coordinate regulation of TPL-2 and NF-κB signaling in macrophages by NF-κB1 p105. Mol Cell Biol. ;32(17):34, (2012).

102. Stark GR, D. J. J. The JAK-STAT pathway at twenty. Immunity. 36(4):503-, (2012). 103. Department of Defense. Joint Publication 1: Doctrine for the Armed Forces of the United States. Jt. Publ. 172 (2013). 104. Offense and Defense. Army Doctrin. Ref. Publ. 3-90, (2012).

105. Waltz, K. N. Realist Thought and Neorealist Theory. Journal of International Affairs 44, 21–37 (1990).

106. Waltz, K. N. The Origins of War in Neorealist Theory. J. Interdiscip. Hist. Vol. 18, (1988). 107. Taliaferro, J. W. Security Seeking under Anarchy: Defensive Realism Revisited. Int. Secur. 25, 128–161 (2001). 108. Levy, J. S. The offensive/defensive balance of military technology: a theoretical and historical analysis. Int. Stud. Q. 28, 219–238 (1984).

109. Medzhitov, R. & Janeway, C. A. Decoding the patterns of self and nonself by the innate immune system. Science 296, 298–300 (2002).

110. Vance RE, Isberg RR, P. D. Patterns of pathogenesis: discrimination of pathogenic and nonpathogenic microbes by the innate immune system. Cell Host Microbe 6(1):10-21, (2009).

111. Lazzaro, B. P. & Rolff, J. Danger, microbes, and homeostasis. Science (80-. ). 332, 43–45 (2011).