The Lymphotoxin/Type I Ifn Axis and Cd8+ T Cell Responses

Home , Eleanor Fish, Lymphotoxin, Lymphotoxin alpha, Lymphotoxin beta

THE LYMPHOTOXIN/TYPE I IFN AXIS AND CD8+ T CELL

RESPONSES

Dennis Ng

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Immunology

University of Toronto

The Lymphotoxin/Type I IFN axis and CD8 T Cell Responses

Dennis Ng

Doctor of Philosophy

Department of Immunology

University of Toronto

2014

Abstract

This thesis examines the role of the TNF family member Lymphotoxin-α1β2 (LTαβ) and its cognate receptor Lymphotoxin-β receptor (LTβR) in modulating DC immunogenic functions that affect the cross-priming of CD8+ T cells against protein antigen in a help-dependent manner.

In non-infection related immune responses without obvious inflammatory stimulus, dendritic cells (DC) require CD4+ T cell help to cross-prime antigen for CD8+ T cell mediated immunity.

Activated CD4+ T cells express the tumor necrosis factor (TNF) members CD40-ligand (CD40L) and LTαβ, which interacts with the TNF receptors CD40 and lymphotoxin-β receptor (LTβR) respectively on DC, significantly increasing DC immunogenic potential. This study demonstrates that LTβR signaling can directly induce Type I IFN expression in DC that affect CD8+ T cell expansion. We identified that LTβR signals through TNF receptor-associated factor (TRAF)-3 and can lead to interferon regulatory factor (IRF)-3 phosphorylation which are important signaling molecules that mediate Type I IFN production. Furthermore, using an inducible autoimmune disease model through the transfer of activated bone marrow derived DC, we found that LTβR signaling in DC is absolutely required for the proper priming of CD8 T cells and

disease onset. We found that LTβR-induced Type I IFNs affect the expression of the adhesion molecule lymphocyte function-associated antigen-1 (LFA-1) and very-late antigen 4 (VLA-4) on antigen specific CD8 T cells which directly affects their ability to infiltrate into target tissues.

iii

Acknowledgments

First, I would like to thank my supervisor Jennifer Gommerman for providing me with tremendous help and guidance throughout my entire PhD study. She is an inspiring mentor and a great friend. I am truly grateful to have this amazing opportunity to work with her. I also want to thank my supervisory committee, Tania Watts and Eleanor Fish, who have provided me with very helpful advice and suggestions on this study.

I want to acknowledge all the current and past members of the Gommerman Lab. In particular, Doug McCarthy, Leslie Summers deLuca, Lesley Ward, Natalia Pikor, Elisa Porfilio,

Olga Lucia Rojas, Georgina Galicia-Rosas, Tian Sun, Conglei Li, Bryant Boulianne and Jorg

Fritz, all of you have been wonderful colleagues and friends who have given me support and joy each and every single day.

I want to thank the DCM animal facility staff, in particular Stacy Nichols who have been terrific and dedicated in providing all the animal care. In addition, I want to thank Dionne White of the MSB flow cytometry lab for her expertise and keeping all the equipment running smoothly. I also want to thank the entire Immunology Office past and present personnel, particularly Sherry Kuhn, Anna Frey, Lynne Omoto, Kate Sedore, Ed Elsonne, and Rejeanne

Puran who provided me all the administrative help.

With great emphasis, I want to thank my parents for providing me endless support and encouragement in the pursuit of my studies. Thank you for taking care of my daily necessities so that I can focus on my research. Most of all, I want to thank my dear Gloria. Thank you for

sharing with me your joy and love. You have led me through many difficult times and have always believed in me.

TABLE OF CONTENTS

TABLE OF CONTENTS VI

LIST OF FIGURES IX

LIST OF APPENDICES XI

ABBREVIATIONS XII

CHAPTER 1 1

1.1 Overview of the immune system 2

1.2 Dendritic cells 2

1.2.1 Dendritic cell subsets 3

1.2.1a Lymphoid tissue-resident dendritic cells 3

1.2.1b Non-lymphoid tissue dendritic cells 4

1.2.1c Plasmacytoid dendritic cells 6

1.1.1d: Monocytes and monocyte-derived DC 7

1.3 Dendritic cell maturation 8

1.3.1 Microbial sensing 8

1.3.1a Toll-Like Receptors 8

1.3.1b RIG-I or RIG-I like receptors 13

1.3.1c C-type lectin receptors 13

1.3.1d NOD-like receptors 14

1.3.2 Dendritic cell migration 15

1.3.3 Co-stimulatory signals 16

1.3.3a The B7 and CD28 family 16

1.3.3b CD70 and CD27 18

1.3.3c 4-1BBL and 4-1BB 19

1.3.3d OX40L and OX40 20

1.3.4 Antigen presentation 20

1.4 T cell responses 22

1.4.1 CD4+ T cell Responses 22

1.4.2 CD8+ T cell responses 24

1.6 DC licensing 26

1.6.1 CD40 mediated licensing 26

1.6.1a CD40 signaling 27

1.6.1b The role of CD40 signaling 28

1.6.2 Lymphotoxin-α1β2 and Lymphotoxin-β receptor 29

1.6.2a LTβR signaling 30

1.6.2bThe role of LTβR signaling in immune system homeostasis 32

1.6.2c The involvement of the LT pathway in immune responses. 34

1.7 Type I IFNs 35

1.7.1 Pro-inflammatory functions of Type I IFNs 35

1.7.2 Anti-inflammatory roles of Type I Interferon 36

1.8 Summary 37

CHAPTER 2 39

2.1 Abstract 40

2.2 Introduction 40

2.3 Results 42

2.3.1 LTβR signaling cooperates with CD40-derived signals for priming of CD8+ T cells in vivo 42

2.3.2 DC-intrinsic LTβR signaling is required for optimal CD8+ T cell clonal expansion 48

2.3.3 LTβR signaling is required for full activation and cell cycle progression of Ag-specific CD8+ T cells 49

2.3.4 Stimulation of LTβR on DC results in the production of Type I IFNs 53 vii

2.3.5 LTβR-/- DC fail to support OTI proliferation in vitro but proliferation can be rescued by exogenous IFNα 59

2.4 Discussion 64

2.5 Materials and methods 68

CHAPTER 3 74

3.1 Abstract 75

3.2 Introduction 75

3.3 Results 78

3.3.1 DC-intrinsic LTβR signaling is required to induce pancreatic inflammation and glucose intolerance 78

3.3.2 Robust CD8+ T cell proliferation and up-regulation of VLA4 and LFA-1 correlate with the induction of diabetes in RIP-GP mice 80

3.3.3 Production of Type I IFNs by BMDC is required for optimal CD8+ T cell activation/expansion and diabetes induction in RIP-GP mice. 89

3.3.4 LTβR-dependent Type I IFN production requires TRAF3 92

3.3.5 Exogenous administration of IFNα restores CD8+ T cell activation and immunopathology in RIP-GP mice that received LTβR-/- BMDC. 96

3.4 Discussion 103

3.5 Materials and methods 108

CHAPTER 4 111

4.1 Summary 112

4.2 Future directions 118

REFERENCES 122

APPENDICES 162

viii

LIST OF FIGURES

Figure 1.1 Toll-like receptors (TLR) and ligands P.10

Figure 1.2 Induction of Type I IFNs by PRR during a viral infection P.12

Figure 1.3 Type I IFNs promotes T cell priming during viral infection versus soluble antigens P.25

Figure 1.4 Members of the lymphotoxin system and the interacting partners. P.31

Figure 2.1 DC-derived LTβR and CD40 signals contribute distinctly to CD8+ T cell cross-priming in vivo. P.43

Figure 2.2 Representative FACS plots of OTI T cell expansion/IFNα production P.44 following LTβR-Ig treatment.

Figure 2.3 Evaluation of DC phenotype/function in LTβR-Ig treated mice P.46

Figure 2.4 Evaluation of DC phenotype/function in LTR-/- treated mice. P.47

Figure 2.5 Expression of LTR on DC is required for optimal numbers of OTI at the P.50 peak of the immune response in vivo

Figure 2.6 Expression of LT on Ag-specific CD4+ T cells is required for optimal P.51 expansion of OTI in vivo and in vitro.

Figure 2.7 LTR signaling is required for normal CD8+ T cell activation and cell cycle completion. P.52

Figure 2.8 LTR-/- DC display altered CD86 up-regulation P.55

Figure 2.9 LTR synergizes with TLR4 to maximize Type I IFN production in BMDC P.56

Figure 2.10 LTR+/-BMDC express normal levels of co-stimulation markers before and P.57 after LPS stimulation, and are capable of producing IL-12 production in vitro.

Figure 2.11 Type I IFN production induced by LPS + OVA is attenuated in P.58 WT BMDC by LTR-Ig treatment

Figure 2.12 LTR signaling can mediate Type I IFN production independent of TLR activation P.60 in BMDC

Figure 2.13 LTR-/- BMDC do not support full OTI proliferation in vitro and this can be P.62 rescued with Type I IFNs

Figure 2.14 BMDC/OTI co-cultures promotes OTI division which is compromised P.63 in the presence of LTR-/- BMDC

Figure 3.1 Diabetes induction in RIP-GP mice by peptide loaded BMDC requires P.79 LCMV-GP specific CD4+ T cell activation.

Figure 3.2 DC-intrinsic LTβR signaling is required for the induction of diabetes P.81-82

Figure 3.3 DC-intrinsic LTβR signaling is required for optimal expansion of antigen-specific P.84-85 CD8+ T cells in the periphery and their accumulation in the pancreas

Figure 3.4 Expansion of P14 CD8+ T cells and the absolute number of GP-specific P.86 CD8+ T cell of RIP-GP mice that received WT or LTβR-/- GP-peptide loaded BMDC

Figure 3.5 Pancreas infiltrating GP-specific CD8+ T cells express high level of adhesion P.87-88 molecules VLA-4 and LFA-1

Figure 3.6 DC-intrinsic expression of IRF3 is required to prime CD8+ T cells and induce P.90-91 diabetes in RIP-GP mice.

Figure 3.7 BMDC stimulation with agonistic αLTβR P.93

Figure 3.8 TRAF3 is required for LTβR-dependent IFN-I production P.94-95

Figure 3.9 DC-intrinsic LTβR-dependent IFN-I production is required for optimal CD8+ T P.97-98 cell expansion and expression adhesion molecules VLA4 and LFA1

Figure 3.10 DC-intrinsic LTβR signaling is required for optimal expression of P.100 VLA4 and LFA1 on OVA-specific CD8+ T cells through a Type-I IFN dependent mechanism

Figure 3.11 Exogenous administration of IFN-I restores CD8+ T cell expansion and P.101 immunopathology in RIP-GP mice that received LTβR-/- BMDC

Figure 4.1 Lymphotoxin-dependent Type I IFNs (IFN-I) are required for early programming P117 of CD8+ T cells resulting in their clonal expansion, up-regulation of LFA-1 and VLA-4 and their ability to cause immunopathology in a self tissue

Figure 4.2 Splenic DC subset of LTβR-/- chimeric mice P.120

Figure 4.3 CD103+ CD11b+ DC isolated from the intestinal laminar propria of mixed WT P.121 and LTβR-/-bone marrow chimera

LIST OF APPENDICES

1) Notch2 receptor signaling controls functional differentiation of dendritic cells in P.163 the spleen and intestine.

2) Notch2-dependent classical dendritic cells orchestrate intestinal immunity to P.175 attaching-and-effacing bacterial pathogens.

ABBREVIATIONS

DC Dendritic Cell

APC Antigen Presenting Cell

CTL Cytotoxic T Lymphocyte

PRR Pattern Recognition Receptor

CMP Common Myeloid Progenitor

CLP Common Lymphoid Precursor

SLO Secondary Lymphoid Organ

PALS Periarteriolar Lymphoid Sheath

MZ Marginal Zone

HSV Herpes simplex Virus

ESAM Endothelial cell-Specific Adhesion Molecule

CCR CC-Chemokine Receptor

LN Lymph Node

IL Interleukin

TGF Transforming Growth Factor

IFN Interferon

Siglec-H Sialic acid binding Ig-like Lectin H mPDCA-1 Mouse Plasmacytoid Dendritic Cell Antigen-1

PAMP Pathogen Associated Molecular Pattern

DAMP Danger Associated Molecular Pattern

TLR Toll-Like-Receptor

RIG-I Retinoic acid Inducible Gene - I

CLR C-type Lectin Receptor

NLR NOD-Like Receptor xii

TIR Toll-Interleukin-1 Receptor

LPS Lipopolysaccharide

IRAK Interleukin-1 Receptor-Associated Kinase

TRAF Tumor Necrosis Factor Receptor Factor

K Lysine

TAK Transforming growth factor-β-Activated protein Kinase

NF Nuclear Factor

MAPK Mitogen activated protein Kinase

TBK1 Tank-binding Kinase-1

IKKε IκB Kinase-ε

IRF3 Interferon Response Factor-3

MDA5 Melanoma Differentiation Antigen-5

CARD Caspase Recruitment Domain

MAVS Mitochondrial Antiviral Signaling

ITAM Immunoreceptor Tyrosine-based Activation Motif

NOD Nucleotide binding and Oligomerization Domain

NOD mice Non-obese diabetes

LRR Leucine-Rich Repeat

NAIP NLRs with Apoptosis Inhibitory Protein

TCR T cell Receptor

FRC Fibroblast Reticular Cells

IDO Indoleamine 2,3-dioxygenase

TH T Helper cells

TFH T Folicullar Helper cells

ITIM Immunoreceptor Tyrosine-based Inhibition Motif

xiii

ITSM Immunoreceptor Tyrosine-based Switch Motif

GP34 Glycoprotein 34

CTL Cytotoxic T Lymphocyte

ER Endoplasmic Reticulum

LCMV Lymphocytic Choriomeningitis Virus

VSV Vesicular Stomatitis Virus

LTαβ Lymphotoxin-α1β2

LTβR Lymphotoxin-β Receptor cIAP Cellular Inhibitor of Apoptosis

NIK NFκB-inducing Kinase

Eomes Eomesodermin

LIGHT Lymphotoxin-like, exhibits inducible expression, and competes with herpes simplex virus glycoprotein D for HVEM, a receptor expressed by T lymphocytes

HVEM Herpesvirus Entry Mediator

BTLA B and T Lymphocyte Attenuator

LTi Lymphoid Tissue inducer cell

HEV High Endothelial Venules mCMV Murine Cytomegalovirus

LTβR-Ig Pharmacological inhibitor, LTβR fusion protein with human Ig

EAE Experimental Autoimmune Encephalomyelitis

IFNAR IFN-α Receptor

TNF Tumor Necrosis Factor

OVA Ovalbumin

OTI T cells OVA-specific CD8 T cells

OTII T cells OVA-specific CD4 T cells

xiv

DT Diphtheria Toxin

AT Adoptive Transfer

GFP Green Fluorescence Protein

BMDC Bone Marrow derived Dendritic Cell

TNFRSF TNF Receptor Super Family

TNFSF TNF Super Family

LCMV Lymphocytic Choriomeningitis Virus

LCMV-GP LCMV-glycoprotein

VLA-4 Very-late Antigen 4

LFA-1 Lymphocyte Function-associated Antigen 1

ICAM Intercellular Adhesion Molecule

Ag Antigen

xv 1

CHAPTER 1

(Some sections were taken and modified from a published review1)

______

1Ng D, Gommerman JL.Front Immunol. 2013 Apr 22;4:94. doi: 10.3389/fimmu.2013.00094. The Regulation of Immune Responses by DC Derived Type I IFNs.

1.1 Overview of the immune system

In order for the human body to ward off infections, we rely on our immune system to detect, recognize and mount an appropriate immune response in order to clear the invading pathogen. Our immune system can be divided into the innate and the adaptive immune system.

Innate immune responses, orchestrated by innate immune cells such as neutrophils, macrophages and dendritic cells (DC), represent the first line of defense against acute infections. Innate immune cells can phagocytose and kill pathogens to control infections and at the same time secrete pro-inflammatory cytokines for further immune cell recruitment to the site of infection.

In contrast, T and B lymphocytes mediate adaptive immune responses for the clearance of pathogens or tumors, and they are able to generate memory and recall response against previous attacks, leading to rapid and efficient immunity against subsequent pathogen encounters. In order to trigger the adaptive immune system, antigen presenting cells (APC) are essential for antigen processing and presentation for T cell priming. DC in particular are the most efficient APC, with the unique capacity to generate cytotoxic T lymphocytes (CTL) response against a pathogen, thus making DC the key immune cells that bridge the adaptive and innate components of the immune system.

1.2 Dendritic cells

DC were first identified in the late 1970s by Ralph Steinmen and Zanvil Cohn2,3. Their discovery has since contributed to countless breakthroughs and greatly advanced the field of immunology. DC play various roles during an immune reaction. They are armed with pattern recognition receptors (PRR) that can detect different microbial-associated molecules and mount the appropriate response by producing either inflammatory or anti-inflammatory cytokines that

3 promote or suppress an immune reaction. In addition, they are capable of regulating the adaptive immune system by directly interacting with T and B cells for their activation or suppression.

Therefore, the immunogenic status of DC can often dictate the outcome of disease pathology.

1.2.1 Dendritic cell subsets

DC represent a broad group of cells that share similar phenotypic features and functions.

DC derived from pre-DC that exert classical DC functions are regarded as conventional DC, while blood derived DC differentiated from the monocyte origin, including plasmacytoid DC are considered non-conventional DC. All splenic and thymic-derived DC are generated from either common myeloid progenitors (CMP) or common lymphoid progenitors (CLP). The CMP precursor cells can differentiate into macrophage-dendritic cell precursor cells (MDP) that give rise to conventional DC and macrophages. Some evidence has suggested that CLP can also directly differentiate into conventional DC4, whereas other studies however showed that CLP are essential for the generation of plasmacytoid DC5. Recent data derived from the Immunological

Genome Project further refined the DC lineage based on transcriptional programming6. Notably,

Zbtb46 has been identified as the key transcription factor that promotes the differentiation of majority of the conventional DC7, except for the CD103+ CD8α+ DC which requires BATF3 and

Notch2 for their final stage of development8. Conventional DC represents heterogeneous populations of DC that are further defined by their distribution, expression of surface markers, cytokine production and immunogenic potential.

1.2.1a Lymphoid tissue-resident dendritic cells

Lymphoid tissue-resident DC are found mainly within the secondary lymphoid organs

(SLO) such as the paracortex of the lymph nodes, periarteriolar lymphoid sheath (PALS) of the

4 spleen and Peyer's Patches9. Lymphoid tissue-resident DC can be further categorized based on the expression of CD8α and CD11b. CD8α+ DC are CD11b-, and their expression of CD8α protein is distinct from the CD8αβ expression by CD8+ T cells10. CD8α+ DC require Flt3 ligand for proliferation, and Flt3-L deficient mice show a substantial loss of the CD8α+ DC population11. In contrast, CD11b+ DC lacks CD8α expression. This subset of DC are mainly found within the subcapsular sinus in the LN, or within the marginal zone (MZ) and the red pulp of the spleen9. CD11b+ DC represent the largest subset of DC situated in SLO, and they display differential CD4 expression with no distinctive difference in function based on CD4 expression.

A recent study has shown that CD11b+ DC can be further divided into two populations in the spleen that is defined by the expression of the endothelial cell-specific adhesion molecule

(ESAM). ESAMhi CD11b+ DC require Notch2 signaling for proliferation and they appear to be derived from DC-restricted precursors whereas ESAMloCD11b+ DC are thought to be derived from circulating monocytes12. It is unclear whether these two subsets of DC are functionally distinct, but many studies have compared the immune function of CD8α+ DC versus CD11b+

DC. Some reports have suggested that CD8α+ DC are specialized in priming CD8+ T cells particularly against viruses13,14, while CD11b+ DC are efficient at interacting with CD4+ T cells9.

Nevertheless, other studies have also shown that all subsets of lymphoid tissue resident DC are capable of priming CD4+ or CD8+ T cell responses depending on the adjuvant and cytokine milieu15–17, suggesting functional flexibility between the different DC subsets.

1.2.1b Non-lymphoid tissue dendritic cells

Non-lymphoid tissue DC are scattered throughout the body including the skin, respiratory system and intestinal tract. They are generally defined by differential expression of CD103 and

CD11b18. DC that reside in the skin can include the Dermal DC (langer- CD103- or langer+

CD103+) and Langerhans cells (langerin+CD103+). Dermal DC and Langerhans cells are different from other DC in terms of their unique ontogeny and slow turnover rate19,20. Both subsets of dermal DC are capable of activating T cells against virus or parasitic infection21–23, while Langerhans cells seem to also be capable of antigen presentation in vitro24, recent studies have shown that they are largely responsible for dampening immune responses and prevent skin hypersensitivity to antigen25,26.

In the lung, CD103+ CD11b- DC are located in the intraepithelial tissues and CD103-

CD11b+ DC are found within the lamina propria of the conducting airways. Upon activation,

CD103+ DC have the capacity to migrate to the draining LN upon activation, and are capable of producing IL-12. Hence, they represent the major subset of DC that cross-present antigen to

CD8+ T cells during pulmonary infections27. Specifically, CD103+ DC have been shown to interact with antigen specific CD8+ T cell for the generation of CTLs while CD11b+ DC induce a memory CD8+ T cell phenotype during influenza infection in mice28,29, and the mechanisms of antigen presentation, including cross-presentation, are described in a later section.

Similar to pulmonary and dermal DC, intestinal DC are also broadly categorized into the

CD103+ and CD103- populations. The CD103+ subset expresses the CC-chemokine receptor

(CCR)-7, and they have the capacity to migrate between the intestinal lamina propria and the mesenteric lymph nodes (LN)30,31. This subset of gut-resident DC have been shown to produce anti-inflammatory mediators within the intestinal microenvironment by secretion of cytokines such as Interleukin (IL)-10 and Transforming Growth Factor (TGF)-β, and promote the development of Foxp3+ regulatory T cells to maintain a balance between the gut microflora and host immune responses32. In addition, recent evidence shows that CD103+ DC found within isolated lymphoid follicles are essential for the production of IL-23 in response to intestinal

6 bacteria infection, a key cytokine required for the activation of innate lymphoid cells and pathogen killing8. Conversely, the CD103- CD11b+ population of intestinal DC play an important role in T cell differentiation into IL-17 and interferon (IFN)-γ producing effector T cells that are important for defense against intestinal pathogens33.

1.2.1c Plasmacytoid dendritic cells

In contrast to conventional DC, pDC are specialized in the production and secretion of

Type I IFNs rather than antigen presentation. They circulate within the blood compartment and lymphoid organs, playing a critical role as part of the first line of defense against various infections and contribute to the debilitating autoimmune disease systemic lupus erythematosus34.

Early studies suggested that pDC are of lymphoid origin, where pDC express lymphocyte specific gene transcripts that include the pre-T cell receptor-α, λ5 and Spi-B35,36. However, recent evidence has demonstrated that both common lymphoid progenitors (CLPs) and common myeloid progenitors (CMPs) can differentiate into pDC in vitro37. Like the lymphoid tissue DC, pDC rely on Flt3-L for development and proliferation. Mice deficient in Flt3-L have a significantly reduced pDC population while Flt3-L transgenic mice showed increase pDC numbers38. Mouse pDC can be identified by their expression of cell surface antigens B220 and

Ly6C, and they express low levels of CD11c and MHC-II. Given that all these surface markers are commonly expressed on other immune cells, it makes the identification and characterization of pDC difficult. Recent efforts into better characterizing pDC in mouse has led to the development of specific antibodies that recognize the mouse plasmacytoid dendritic cell antigen-

1 (mPDCA-1) and surface molecule sialic acid binding Ig-like lectin H (Siglec-H) which are uniquely expressed by pDC39,40.

In addition to inflammatory responses and production of Type I IFNs, other studies have demonstrated immunoregulatory functions ascribed to pDC. Specifically, in the mouse model of

Multiple Sclerosis, depletion of pDC results in overt inflammation with increased effector T cell accumulation in the brain and spinal cord41. Similarly, mice infected with Chlamydia pneumoniae showed that depletion of pDC led to increased secretion of inflammatory cytokines and significant loss of regulatory T cells in the lung and draining lymph nodes42. Whether pDC can directly regulate T cell responses remains unclear, however as discussed in a later section,

Type I IFNs, the major cytokine produced by pDC, can exert both inflammatory and anti- inflammatory functions that may account for such functional dichotomy exerted by pDC.

1.1.1d: Monocytes and monocyte-derived DC

Monocytes are derived from MDPs that circulate the blood, their development and differentiation into monocyte-derived DC is independent of the expression the DC lineage specific Zbtb46 transcription factor43, but instead rely on PU.1 for their development44. During an infection, monocytes with high expression of the chemokine receptor Ly6C are recruited to the site of infection, where local inflammation can trigger their differentiation into macrophages and DC45. Monocyte-derived DC can be found in various locations of the body that are often exposed to the environment such as the lung, skin, intestines and the spleen. Similar to conventional DC, they are capable of capturing antigens, migrate to the draining LN and present antigen to CD4+ and CD8+ T cells. In addition, they also capable of expressing anti-microbicidal molecules such as iNOS and TNF which are important for the clearance of pathogens independent of their antigen presenting cell function46.

1.3 Dendritic cell maturation

In the steady state, DC exist in a heterogeneous status of maturation due to the constant sampling of self or foreign antigens. They govern immunogenic versus tolerogenic immune responses through the different cytokines that they produce. Armed with a set of pattern recognition receptors (PRR - described below) that allow DC to detect pathogen- or danger- associated molecular patterns (PAMPs or DAMPs), DC can discriminate between foreign versus self-antigen based on the co-existence of these PRR-mediated signals, resulting in induction of a specific cytokine milieu that accelerate or suppress DC maturation to provide an appropriate response.

1.3.1 Microbial sensing

Our immune system is in a constant evolutionary battle with pathogens that has played out through millennia. In order to combat infections, our body relies on detection tools against a plethora of micro-organisms. The major classes of PRR include Toll-Like-Receptors (TLRs),

Retinoic acid Inducible Gene – I (RIG-I) receptors, C-type lectin receptors (CLRs), and NOD- like receptors (NLRs).

1.3.1a Toll-Like Receptors

Thus far, 13 mammalian TLRs have been identified (10 in human and 12 in mice) and each

TLR recognizes a specific class of microbial PAMPs that trigger a distinctive set of immune responses47. TLRs belong to the IL-1 receptor family, and all of them contain a common cytoplasmic domain referred to as the Toll-interleukin-1 receptor (TIR) domain. Upon activation, the TIR domain recruits TIR-associated adapter molecules including MyD88, TRIF, TRAM and

TIRAP that mediate various downstream signaling pathways. TLRs are mainly expressed in

9 immune cells such as neutrophils, macrophages, DC, and some lymphocytes, while non-immune cells such as fibroblasts and intestinal epithelial cells express a more restricted sets of TLRs48,49.

The majority of TLRs are found on the cell membrane with the exception of TLR-3, 7, 8 and 9, which are expressed within intracellular endosomes and are mainly associated with the detection of viral nucleic acids, triggering the production of Type I IFNs50,51. The sub-cellular compartmentalization of these TLRs allows the recognition of self versus non-self pathogen- associated nucleic acids due to differential mechanisms of antigen up-take, allowing non-self antigen to trigger compartmentalized TLR activation52. In contrast, TLR-1, 2, 4, 5, 6 and 11 are all transmembrane receptors that can readily recognize a wide range of microbial components such as lippopolysaccharide (LPS) by TLR4, bacterial cell wall elements that include triacyl or diacyl lipoprotein by the heterodimerized TLR2/TLR1 or TLR2/TLR6 respectively, and bacterial flagellin by TLR553. Specific TLR activation leads to recruitment of specific TIR-associated adapter molecules, thereby triggering unique signaling pathways that induce the transcription of specific gene profiles in response to different types of pathogens (Figure 1.1).

Among the TIR-associated adapters, MyD88 has been well characterized as the key molecule required for all TLR signaling, with the exception of TLR3 which requires Trif for Type I IFN induction. Upon activation, MyD88 recruits the Interleukin-1 receptor-associated kinase (IRAK)-

1 and IRAK-4 for their phosphorylation, resulting in the recruitment of tumor necrosis factor receptor factor (TRAF)-6. TRAF6 acts as the ubiquitin E3 ligase that mediates lysine (K)-63 ubiquitination of the transforming growth factor-β-activated protein kinase 1 (TAK1) to promote the recruitment of TAK-binding proteins (TAB)-1, TAB-2 and TAB-3. The activated TAK1 complex then further mediates the activation of classical nuclear factor (NF)-κB and mitogen activated protein kinase (MAPK) pathways which are important for the induction of various

Figure 1.1 Toll-like receptors (TLR) and ligands

TLR can recognize various PAMPs or DAMPs and trigger different inflammatory responses. TLR5, TLR4, TLR11 as well as the heterodimers of TLR2/TLR1 and TLR2/TLR6 are cell surface receptors, whereas TLR3, TLR7/8 and TLR9 are localized to the endosomes. TLR signaling is initiated by ligand-induced dimerization of receptors. Through the TIR domain, TLR can recruit TIR domain containing adaptor proteins that mediate downstream signaling. All TLR with the exception of TLR3 mediate signals through MyD88 that leads to the production of inflammatory cytokines such as TNF-α. TLR3 and TLR4 can trigger Type I IFN induction through TRIF. In plasmacytoid DC, TLR7/8 and TLR9 signal through IRF7 activation that regulates a robust Type I IFN response.

11 inflammatory cytokines such as IL-12 and TNF-α54. In response to viral infection, conventional

DC are capable of producing moderate levels of Type I IFNs primarily through TLR3, which is independent of MyD88 signaling. TLR3 is expressed on conventional DCs but not on pDCs, and it is the only TLR that exclusively recruits Trif to mediate signaling. Upon activation, Trif binds to TRAF3 and TRAF6. TRAF6 is required for the recruitment of RIP1 that induces NFκB activation. Conversely, TRAF3 becomes K63-ubiquitinated, and it is essential for the recruitment of Tank-binding Kinase-1 (TBK1), IκB Kinase-ε (IKKε) and interferon response factor-3 (IRF3), ultimately leading to IRF3 phosphorylation55–57. Phosphorylated IRF3 subsequently homodimerizes and translocates into the nucleus. Together IRF3 and NFκB form the transcription factors required for the expression of IFN genes (Figure 1.2).

In contrast to conventional DC, pDC respond to viral infections through high expression of

TLR7 and TLR9, leading to the production of large amounts of Type I IFNs. Even though the signaling pathway downstream of TLR7, 8 and 9 in pDC is mediated through MyD8858, the signaling is unique in terms of of Type I IFN induction. Upon TLR7/8 and -9 activation, MyD88 recruits IRAK4 and IRAK1 to the receptors, and through interaction of their death domains,

IRAK1 becomes phosphorylated. Activated IRAK1 further recruits and activates IRF-7, resulting in the production of Type I IFNs 59,60. Even though conventional DCs also express TLR7 and 9, their activation predominantly initiates a DC maturation program resulting in the production of

IL-1261,62. Hence, in response to viral infection, pDCs are specialized in producing a copious amount of Type I IFNs that trigger a systemic antiviral state.

Figure 1.2 Induction of Type I IFNs by PRR during a viral infection

TLR 3, 7 and 9 are mainly expressed within the endosomes of innate immune cells. Virus or virus-infected cells are taken up by macrophages or DCs, and the viral nucleic molecules are exposed upon endosomal acidification. Activation of TLR7 and 9 requires signaling through MyD88 and recruitment of IRAK4, IRAK1 and IRF7. IRF7 becomes phosphorylated and translocates into the nucleus upon dimerization resulting in transcription of Type I IFN genes. TLR3 signals exclusively through Trif which binds TRAF6 and recruits RIP1 for NF-κB activation. Trif also binds TRAF3 leading to TRAF3 K63-linked self-ubiquintination, facilitating the recruitment of TBK1, IKKε and IRF3 for IRF3 phosphorylation. Phosphorylated IRF3 homodimerizes and translocates into the nucleus for transcription of Type I IFN genes. RIG-I or RIG-I like receptors are expressed in all nucleated cells, and they recognizes viral RNA found in the cytoplasm. Upon activation, RIG-I recruits MAVS through the CARD domain interaction, and, analogous to TRIF, MAVS further binds IKKε, TBK1 and IRF3 to promote IRF3-activation and Type I IFN expression. (Figure adapted from Ng et al1)

1.3.1b RIG-I or RIG-I like receptors

While TLRs are primarily expressed on innate immune cells, RIG-I are ubiquitously expressed in the cytoplasm of all nucleated cells. Instead of actively sensing viral particles, RIG-

I are triggered when cells become infected. RIG-I and the RIG-I like receptors, such as the melanoma differentiation antigen 5 (MDA5) belong to a family of DExD/H box RNA helicases.

The N-terminal region of RIG-I is characterized by two caspase recruitment domains (CARD), and the C-terminal region contains RNA helicase activity 63. RIG-I recognizes double stranded

RNA by the RNA helicase domain, and through a CARD-CARD interaction, RIG-I recruits the

CARD-containing adaptor mitochondrial antiviral signaling protein (MAVS - also known as

VISA, IPS-1 or CARDIF) to mediate downstream events 63. The signaling pathway activated by

MAVS remains largely unclear, but studies have shown that MAVS is required for NFκB activation and it interacts with TBK1 and IKKε to form a "signalosome" that phosphorylates

IRF3 and IRF7 for Type I IFN production (Figure 1.2)64,65. In vivo models have shown that RIG-

I-deficient mice, despite having intact TLR signaling, succumb to infection by vesicular stomatitis virus, Newcastle disease virus and Sendai virus (Figure 1.2)66. TLRs and RIG-I signaling complement each other to provide complete coverage across various types of viruses, and both detection systems are geared toward rapid production of Type I IFNs leading to a systemic antiviral state and the control of viral infection.

1.3.1c C-type lectin receptors

CLRs are expressed on innate immune cells such as macrophages, and they recognize various pathogens based on their carbohydrate components such as mannose, fucose or glucan carbohydrate structures, through one or more carbohydrate recognition domain that the CLR family shares in common67. CLR can be expressed within the cytoplasm as soluble receptor or on

14 the cell surface as a transmembrane protein. In general CLRs can be divided into two groups:

CLR I of the mannose receptor family and CLR II which belongs to the asialoglycoprotein receptor family. CLR activation primarily leads to antigen internalization into the endosome for degradation. CLR signaling is complex and can cross-talk with other PRR signaling pathways.

Numerous studies have shown that CLR can either signal through the immunoreceptor tyrosine- based activation motif (ITAM)-containing adaptor molecules such as DAP1268 or by activation of protein kinases and phosphatases through direct protein-protein interaction of the CLR cytoplamic tails69,70.

1.3.1d NOD-like receptors

NLRs are a family of intracellular receptors that can detect pathogens through recognition and binding of PAMPs or DAMPs. There are 23 NLR members in humans, and 34 in mice. All

NLRs feature an N-terminal CARD domain, a central nucleotide binding and oligomerization domain (NOD), and a C-terminal leucine-rich repeat (LRR) domain71. The ligand, signaling and function of many of the NLR members have yet to be identified and characterized. This large number of receptors is further subdivided into three major families which include NODs, pyrin containing NLRs (NLRP) and NLRs with apoptosis inhibitory protein (NAIP). NOD1 and

NOD2 were the earliest-identified NLRs. They recognize different components of peptidoglycan on bacteria, triggering classical NFκB and MAPKs activation resulting in the induction of pro- inflammatory cytokines upon activation72. Furthermore, a recent study showed that systemic

NOD1 activation can lead to marked chemokine production in the spleen that results in B cell recruitment for activation, suggesting a unique and novel role for NOD1 signaling that facilitates

B cell innate immunity73.

Distinct from other NLRs, NLRPs are unique in their capacity to induce and regulate the caspase-1 inflammasome through the N-terminal pyrin domain. In particular NALP1, NLRP3 and NLRC4 can trigger caspase-1 activation for the cleavage of pro-IL-1β and pro-IL-18 into their active forms, leading to a different type of programmed cell death termed pyroptosis, which is distinct from apoptosis and necrosis74,75. Pyroptosis can induce rapid cell death in infected host cells as a result of IL-1β and IL-18 mediated formation of pores in the plasma membrane, leading to the release of mediators that amplify inflammation. This mechanism of response has been found to be uniquely applied against specific bacterial pathogens such as Salmonella76,

Shigella77, Listeria78 and Anthrax79.

1.3.2 Dendritic cell migration

In the steady state, DC in the periphery are continuously trafficking between the blood, interstitial space and SLO. During an infection, however, tissue-intrinsic DC activated by the aforementioned immature DC in the periphery rely on chemokine receptors such as CCR2,

CCR5 and CCR6 to home to peripheral tissues80, and they express adhesion molecules such as E- cadherin that allow DC to anchor themselves with neighboring tissue cells81. However, upon infection, PRR-derived signals will promote the rapid migration (within hours) of tissue-resident

DC via the afferent lymph from the tissue parenchyma into SLO 82. Specifically, PRR engagement leads to morphological changes in DC that result in down-regulation of the tissue specific chemokine receptors and adhesion molecules, allowing DC to detach themselves from the site of infection/inflammation. Activated DC further up-regulate the chemokine receptors

CCR7 and CXCR4 which bind to CCL21 and CXCL12 respectively. These chemokines are highly expressed in the draining LN83–85, thereby promoting DC migration into the SLO through chemotaxis. Lymphoid tissue-resident DC are localized at the cortical ridge of the LN and at the

16 outer area of the PALS in the spleen. They can sample and ingest antigen that enters the SLO through the afferent lymph. Upon activation, they also migrate deeper into the paracortex of the

LN, or into the PALS where T cells reside. Within the paracortex and the T cell zone, DC and T cells travel across an extracellular matrix network that is constitutively produced by splenic stromal cells called fibroblast reticular cells (FRC). In order for DC to locate and interact with the corresponding T cell expressing the cognate T cell receptor (TCR), DC produce chemoattractants such as CCL21, CCL3 and CCL4 to direct T cell migration into their vicinity86.

1.3.3 Co-stimulatory signals

DC activation is characterized by phenotypic changes and substantial increase of DC immunogenic potential in priming T cell responses. This is achieved in part through up- regulation of co-stimulatory molecules such as the B7 family, and multiple TNFSF members that include 41BB-L and OX40L, and later CD70, generally as a result of PRR activation. In the absence of co-stimulation, triggering of the T cell receptor can result in tolerance.

1.3.3a The B7 and CD28 family

CD28 is the major co-stimulatory receptor for T cell activation. B7-1 and B7-2 molecules, also known as CD80 and CD86, are the best characterized co-stimulatory signals expressed by

APCs that bind the CD28 receptor on T cells. CD80 and CD86 also bind CTLA4, a receptor that is upregulated later during T cell activation that mediates negative signals (see next section).

Both CD80 and CD86 molecules are expressed at low levels on immature DC and macrophages, and are highly up-regulated upon activation. Hence, the expression levels of CD80 and CD86 alone are often sufficient to determine DC activation status.

CD28 is a type I transmembrane glycoprotein. Its cross-linking significantly lowers the threshold of TCR activation through amplifying TCR-induced activation of transcription factors such as AP1, NFAT and NFκB87. In addition, CD28 signaling has also been shown to induce the

88 up-regulation of anti-apoptotic proteins Bcl-XL and Bcl-2 to promote T cell survival . CD28 can bind to and activate phosphoinositide 3-kinase (PI3K) through its SH2 domain. PI3K activation results in the further recruitment of AKT for its phosphorylation, leading to cell cycle progression and facilitating cell metabolism by increasing glucose uptake to allow T cell expansion89.

CTLA-4 is another member of the CD28 family that also binds B7-H1 and B7-H2, however, as opposed to the CD28 receptor, CTLA-4 serves a negative regulatory role in T cell responses. CTLA-4 is mainly expressed on activated T cells to block further co-stimulatory signals on APCs to suppress overt activation and curtail T cell responses. Early work has shown that CTLA-4 can inhibit IL-2 production in T cells and reverse signals in DC through the B7 molecules to initiate immune suppressive responses, leading to IFN-γ secretion and production of indoleamine 2,3-dioxygenase (IDO) that further inhibits T cell proliferation90, but more recent evidence indicates that CTLA-4 can inhibit CD28 co-stimulation by capturing CD28 shared ligands B7.1 and B7.2 through trans-endocyotsis which negatively regulates T cell responses91.

ICOS-L and PD-L1 are two members of the B7 family that also contribute significantly to

T cell activation. ICOS-L binds the receptor ICOS, a CD28 homolog, expressed on activated T cells. ICOS activation can trigger PI3K activation which results in AKT phosphorylation that drives T cell metabolism. ICOS signaling has been associated with T Helper (TH)-2 cell differentiation since T cells expressing high level of ICOS mainly secrete TH-2 relevant cytokines such as IL-4, IL-5, IL-10 and IL-13 in Schistosoma mansoni infection and in asthma

18 models92,93. However, other studies have also shown that ICOS signaling can trigger TH-1 and

TH17 responses leading to activation of colonic intraepithelial lymphocytes and production of

IFN-γ in certain inflammatory bowel diseases94. In addition, recent studies have shown that

ICOS signaling can lead to the expression of Bcl6, a key transcription factor that initiates T folicullar helper cell (TFH) differentiation95.

In contrast to other CD28 family members, PD-1 is structurally different in that PD-1 lacks the membrane proximal cysteine that allows other members of the CD28 receptors to homodimerize. Instead, PD-1 contains an immunoreceptor tyrosine-based inhibition motif

(ITIM) and an immunoreceptor tyrosine-based switch motif (ITSM) within the cytoplasmic domain that regulates PD-1 function96. PD-1 binds two ligands, PD-L1 and PD-L2. PD-L1 is the major ligand expressed on various cell types including B cells, T cells and DCs97, as well as non- hematopoietic cells such as endothelial cells98. With regards to PD-1 function, most studies agree that PD-1 exerts a negative signal resulting in cell cycle arrest in both T and B cells through the blocking of CD28-mediated activation of PI3K99,100, however another study has shown that PD-1 signaling can stimulate T cell proliferation whereby PD-L1-Ig fusion protein enhances T cell expansion and effector cytokine production in vitro101. PD-1 signals through the ITSM motif and is capable of interacting with different phosphatases that may contribute to its bi-directional signaling. In addition, PD-L1 has also been shown to directly bind and interact with B7.1, inhibiting T cell proliferation and cytokine production102.

1.3.3b CD70 and CD27

CD70 belongs to the TNFSF. It is primarily expressed on activated DC later in the immune response. CD70 binds the cognate TNF receptor CD27 expressed on naive CD4+ and CD8+ T

19 cells. CD27 signaling can lead to the activation of classical NFκB and c-Jun kinase pathway.

Previous studies have revealed that CD70-CD27 is particularly important for the primary generation of a CTL response against virus infection103,104. In contrast to CD28 signaling that drives cell cycle progression, CD27 is linked to IL-2 production which is essential for the maintenance of effector CTL105. Mice lacking CD27 and infected with influenza show reduced

CD4+ and CD8+ T cell accumulation in the lung106. In addition, a recent study also suggests that

CD27 is important for the induction of CXCL10 secretion by CD8+ T cells which contributes to their accumulation at the site of infection107.

1.3.3c 4-1BBL and 4-1BB

Like CD27, 4-1BB also known as CD137 or TNFRSF9 is a TNF receptor family member expressed on many cell types including mast cells, DC, granulocytes, NK cells, NKT cells and T cells. The ligand 4-1BBL is expressed predominantly on activated APCs such as macrophages,

DC and B cells, while the expression of the receptor is up-regulated on T cells upon activation.

4-1BB signaling can activate multiple pathways that include ERK, p38, JNK, classical NFκB and

PI3K that results in ubiquitin mediated degradation of pro-apoptotic proteins such as BIM, up- regulation of anti-apoptotic factors Bcl-XL, secretion of IL-2 and increased cell cycle progression108. Early studies showed that 4-1BB or 4-1BBL deficient mice have impaired CD8+

T cell responses to acute virus infections, suggesting 4-1BB may be important for T cell expansion. However, more recent evidence demonstrated the requirement for 4-1BBL is dependent on the severity of infection. During severe respiratory influenza infection, 4-1BBL is important for maintaining effector CD8 T cell survival, whereas during mild influenza infection,

4-1BBL is dispensable for primary CD8 T cell response, but it is essential for the maintenance of

CD8+ T cell memory109.

1.3.3d OX40L and OX40

OX40, also known as CD134 or TNFRSF4, is expressed primarily on activated T cells.

OX40 binds OX40L, or glycoprotein 34 (GP34) expressed on DC. In contrast to CD28 signaling,

OX40L-OX40 interaction occurs 2 – 3 days post TCR activation, whereas CD28 amplifies TCR signals simultaneously with T cell priming110. Hence, it is believed that OX40 signaling primarily prolongs T cell survival and continual proliferation after the initial burst of clonal expansion. OX40 signaling can lead to NFκB and PI3K-AKT activation, thus providing signals that support T cell survival. In addition, OX40 engagement can enhance CD4+ T cell differentiation by triggering the secretion of polarizing cytokines that induce TH1, TH2 and

TH17 responses111–115.

1.3.4 Antigen presentation

Like macrophages, DC are capable of taking up antigen through non-specific macropintocytosis and receptor-mediated phagocytosis. In addition to foreign antigen, DC express surface receptors CD205 and FcγR for uptake of immune complexes, and integrins αVβ3 and αVβ5 for uptake of dying cells. Captured antigens are processed and loaded onto MHC-II and MHC-I molecules for presentation to CD4+ T cells and CD8+ T cells respectively.

Classically, MHC-II loading presents both endogenous (self) and exogenous (foreign) antigens through the endocytic pathway, whereas MHC-I selectively presents endogenous antigens.

However, later studies revealed that exogenous antigen can be presented on MHC-I through cross-presentation specifically by DC, and this form of antigen presentation is particularly important for the generation of cytotoxic T lymphocytes (CTL) for clearance of virus or bacteria infected cells116,117.

MHC II molecules are synthesized in the endoplasmic reticulum where three α/β MHC II dimers are linked to a trimer of invariant chains that block peptide binding. Once assembled, the nonamer complex exits the ER for further transport into early endosomes. Lysosomes fuse with endosomes resulting in a lowering of the pH and the delivery of proteases into this fused compartment that mediate the degradation of the invariant chain. Immature DC accumulate prepared MHC II molecules in these endosomes loaded with proteolytic cathepsin and chaperone proteins. Once antigens are captured, they are transported to these vesicles for degradation.

Chaperone proteins such as HLA-DM (in human) help conjugate the antigen peptide binding to

MHC-II. Following peptide conjugation, the vesicles are then transported to the cell surface, exposing the peptide bound MHC-II118.

In contrast to MHC-II expression typically found in innate immune cells, all nucleated cells can express MHC-I. Ubiquitin mediated degradation of endogenous antigens are processed by the proteasome and loaded onto MHC-I molecules within the endoplasmic reticulum (ER) via specialized peptide transporters called TAP. Within the ER, resident chaperones calnexin and calreticulin facilitate the formation of the peptide loading complex and tapasin mediatepeptide loading119. When the complex is assembled, the peptide bound MHC-I are transferred through the Golgi to the plasma membrane exposing the peptide bound MHC-I for presentation. In order for exogenous antigens to be cross-presented through MHC-I, the antigen is usually taken up by

DC through macropinocytosis or phagocytosis. Captured exogenous antigens can either be loaded onto MHC-I through the endocytic pathway similar to MHC-II loading (TAP independent)120,121 or the ER (TAP dependent)122,123.

1.4 T cell responses

Naive T cells continually traffick through SLO searching for APC expressing their cognate peptide bound to MHC molecules. As mentioned, DC can secrete chemoattractants that allow T cells to better locate them within the T cell zone. During the early phase of T cell - DC interaction, the antigen specific T cells receive low levels of TCR stimulation, and the peptide affinity and quantity presented will determine the contact duration and whether threshold TCR signaling can be achieved124. Following initial contact, DC provide T cells with peptide bound

MHC, co-stimulatory signals B7-H1 and B7-H2, and cytokines required for T cell programming.

All of these signals are essential to promote optimum T cell responses.

1.4.1 CD4+ T cell Responses

Antigen-presenting DC can trigger CD4+ T cell differentiation into, broadly speaking, four subsets of T helper cells, TH-1, TH-2, TH-17 and TFH that mediate very different types of adaptive immunity. In response to different antigens, DC can secrete different cytokines that facilitate the differentiation of CD4+ T cells. PRR activation generally triggers inflammatory responses, resulting in the induction of IL-12 and Type I IFNs125,126 which activate T-bet, a master regulator of TH-1 differentiation. Activated T cells further produce IL-2 that triggers

STAT5 activation, positively regulating T-bet binding and transcription of effector cytokines such as TNF-α and IFN-γ127.

For TH-2 responses, in vitro studies have shown that naive CD4+ T cells require IL-2 and

IL-4 to initiate TH-2 differentiation. IL-4 mediates STAT6 activation that further promotes the expression of GATA-3, the key transcription factor that regulates TH-2 programming128. Similar to TH-1 differentiation, IL-2 is also required for STAT5 activation that synergizes with IL-4 for

23 transcription of TH-2 relevant genes. In addition, the presence of IL-4 suppresses IL-12 activity to allow for restricted TH-2 cell differentiation129. Whether DC can directly induce IL-4 production is unclear: a previous study showed that TH-2 differentiation in vivo can proceed independently of IL-4. Furthermore, activation of the IRF-4 in DC can trigger IL-10 and IL-33 production that indirectly promote TH-2 differentiation in atopic asthma130. Interestingly, several studies have also reported that DC can regulate TH-2 differentiation based on TCR strength, whereby strong TCR signals suppress the expression of GATA-3 and inhibit STAT5 activity to allow for TH-1 differentiation, while weak TCR responses favor TH-2 differentiation131–133.

TH-17 cells are a unique subset of CD4+ T cell that requires IL-6, IL-21, IL-23 and TGF-β for differentiation. TH-17 differentiation can be divided into three stages: 1) Upon TCR activation, TGF-β triggers the expression of the IL-23 receptor and ROR-γt, the key TH17 regulatory transcription factor to initiate differentiation as well as the expression of IL-17 cytokines, meanwhile IL-6 activates STAT3 that results in IL-21 production. 2) IL-21 acts in a paracrine or autocrine manner to promote TH-17 cell expansion. 3) The production of IL-23 further maintains the TH-17 pool134. TH-17 cells are pro-inflammatory cells capable of producing the IL-17 cytokines (IL-17A and IL-17F). They are particularly important for the clearance of bacterial and fungal infections, however they have also been associated with numerous autoimmune diseases including Multiple Sclerosis and Crohn's Disease.

TFH are different from other T helper cells in that they are specialized in providing help signals specifically to B cells to regulate humoral responses. Their unique expression of the chemokine receptor CXCR5 directs their migration into B cell follicles in a CXCL13 dependent manner. TFH interaction with cognate B cells triggers B cell differentiation and subsequent germinal center formation, resulting in somatic hypermutation and antibody class switching.

Early studies proposed that IL-6 and IL-21 are sufficient for TFH differentiation since naive

CD4+ T cells stimulated with these cytokines up-regulate CXCR5135. However mice deficient in

IL-6 and IL-21 showed normal TFH differentiation suggesting other factors can compensate136.

Bcl6 is the master transcription factor that regulates TFH differentiation, and as mentioned earlier, ICOS signaling is critical for the induction of Bcl6 expression in TFH95.

1.4.2 CD8+ T cell responses

CD8+ T cells mediate immune responses against numerous infectious pathogens, tumor immunity, autoimmune diseases and graft rejection in organ transplants. In order for DC to prime

CD8+ T cells for efficient generation of CTLs, DC must be optimally activated. For immune responses that produce robust inflammation and directly induce high expression levels of peptide-bound MHC molecules, co-stimulatory molecules and secretion of cytokines such as IL-

12 and Type I IFNs, DCs have been shown to have the capacity for direct CD8+ T cell priming in for the initial response137,138, albeit the secondary recall response is impaired in the absence of

CD4+ T cell help139–141. As for immune responses that produce lower levels of inflammation, DC require help from CD4+ T cell to boost their immunogenic potential to mediate optimal CD8+ T cell activation, a process known as cross-priming (Figure 1.3)142. Cross-presentation and cross- priming of CD8+ T cells are two different immunologic events whereby cross-presentation of exogenous antigen by DC may not be required to cross-prime CD8+ T cells for the generation of

CTLs. In cross-priming, activated CD4+ T cells up-regulate help signals that induce further DC maturation. This process, which is also known as DC licensing, is critical for the generation of

CTL responses. Infections with viruses such as influenza, lymphocytic choriomeningitis virus

(LCMV) and vesicular stomatitis virus (VSV) can produce a strong CD8+ T cell response in the absence of help143,144. These pathogens create a robust inflammatory response through PPR

Figure 1.3 Type I IFNs promote T cell priming during viral infection versus soluble antigens

Viruses can trigger PRR activation on immature DCs, leading to DC maturation and production of various pro-inflammatory cytokines and a large quantity of Type I IFNs. In this scenario, DCs are strongly activated, and they are capable of directly interacting with CD8+ T cells for the generation of virus-specific CTLs. In the case of soluble antigens, DCs are poorly activated due to the absence of PRR-stimulus. Semi-mature DCs must first interact with helper CD4+ helper T cells which rapidly up-regulate "help signals" CD40L and LTαβ upon activation. DC-intrinsic LTβR and CD40 activation promotes DC maturation, with LTβR signaling producing a modest amount of Type I IFNs for a sustained period that facilitates CD8+ T cell expansion.

26 activation in DC, leading to the secretion of pro-inflammatory cytokines such as IL-12 and Type

I IFNs which are critical for DC activation as well as CD8+ T cell programming. Specifically recent studies have shown that antigen or virus specific CD8+ T cells with high expression of microRNA miR-155 can down-regulate STAT1 to inhibit the anti-proliferative effect of Type I

IFNs145. With respect to pathogens that do not produce strong inflammation or trigger Type I

IFN responses such as Vaccinia virus140,146, CD4+ T cell help is more important for their clearance. In addition to TCR activation and CD28 co-stimulation, IL-12 and Type I IFNs have both been described as important third signals in T cell priming. Their functions which will be described in the later sections are pivotal for the generation of CTLs147–149.

1.6 DC licensing

DC licensing is an early event during initial T cell - DC interaction. Upon activation, antigen specific CD4+ T cells quickly up-regulate specific DC licensing factors, CD40L and lymphotoxin-α1β2 (LTαβ), within 24 hours of cross-linking of the T cell receptor. These signals trigger the cognate TNF receptor, CD40 and lymphotoxin-β receptor (LTβR) expressed on DC that in turn modulate DC functions by augmenting the expression of peptide bound MHC molecule, co-stimulatory signals and secretion of cytokines that are essential for T cell programming.

1.6.1 CD40 mediated licensing

CD40 signaling was the first ever identified T cell help signal that modulates DC function for priming CD8+ T cells150–152. As mentioned, CD40L belongs to the TNF superfamily and it is expressed on activated T cells and B cells. CD40L binds and activates its cognate receptor CD40

27 expressed on T cells, B cells, platelets, macrophages and DC. The expression of CD40 on DC is regulated, where it is highly expressed upon immune activation.

1.6.1a CD40 signaling

CD40 is a transmembrane protein and a member of the TNFSRF. Like other TNFRSF, its activation by its corresponding ligand leading to receptor trimerization. The assembled CD40 complex exposes the binding site for the recruitment of TRAFs. TRAF-1, -2, -3, -5, and -6 are recruited to the cytoplasmic tail of CD40, binding to CD40 either directly through the TRAF binding domain or indirectly through TRAF-TRAF interactions153. CD40 regulates many signaling pathways that include p38, AKT, JNK, STAT5 and both classical and alternative NFκB activation, leading to transcription of numerous genes.

The TRAF molecules are critical adapter proteins that mediate signaling downstream of all

TNFRSF, and they work together to facilitate numerous signaling events. Within the context of

CD40 signaling, TRAF1 can bind weakly to the TRAF binding domain of CD40, and a recent study has revealed that TRAF1 is important for TRAF2 recruitment to CD40 where TRAF1-/- mice showed reduced TRAF2 binding to CD40, resulting in loss of TRAF2 activity154. In addition TRAF1 has also been implicated in regulating the activity of TRAF2 as well as cellular

Inhibitor of Apoptosis (cIAP) 1 and 2. In the steady state, a pre-assembled complex containing

TRAF2, TRAF3, TRAF5 and cIAP 1/2 constitutively bind the NFκB-inducing kinase (NIK) through the NIK binding domain on TRAF3, and this complex mediates NIK ubiquitin-directed degradation. NIK is a critical kinase that processes the NFκB2 p100 molecule into the p52 active moiety which can further bind RelB for nuclear translocation and gene transcription. Upon CD40 activation, TRAF1 can recruit the complex to the receptor for TRAF2 activation. TRAF2 acts as

28 an E3 ubiquitin ligase to mediate the K63 ubiquitination of cIAP1/2 which further directs the

K48-ubiquitination of TRAF3 for its degradation. Once TRAF3 is degraded, NIK can resume its function of propagating alternative NFκB signaling155,156. The exact role of TRAF5 within the complex is unclear, but a TRAF5 deficient system showed that it is required for both classical and alternative NFκB activation157. Lastly, TRAF6 can signal classical NFκB activation through a similar mechanism downstream of the MyD88-mediated signaling pathway. In addition, recent studies have shown that TRAF6 can recruit Cbl-b and PI3K which results in Akt phosphorylation that is critical for DC survival158.

1.6.1b The role of CD40 signaling

CD40 signaling is a potent help signal, whereby activation of CD40 alone can compensate for CD4+ T cell help to generate efficient CTL responses150,159. CD40 activation in DC can lead to up-regulation of the adhesion molecule ICAM-1, co-stimulatory molecules B7-H1 and B7-H2, and can increase the stability of peptide-bound MHC class I and class II complexes160,161. In addition, CD40 mediated activation of AKT and the classical NFκB pathway can lead to the up- regulation of anti-apoptotic proteins such as Bcl-XL to improve DC survival, which results in increased duration of DC : T cell interactions162. Lastly, CD40 engagement in DC can trigger robust IL-12 expression163 which is critical for early CD8+ T cell programming.

IL-12 is a heterodimer composed of a p35 and p40 subunit and binds specifically the IL-12 receptor expressed on NK cells and T cells. Both in vitro and in vivo mouse studies have shown that IL-12 can directly promote antigen specific CD8+ T cell proliferation in response to peptide bound MHC-I presentation164–166. IL-12 deficient mice succumb to mCMV infection, and they exhibit significant loss of CTL antiviral responses167. However, for help-independent infections

29 such as Vesicular Stomatitis virus, IL-12 deficient mice can produce normal CD8+ T cell responses for viral clearance, further defining the importance of CD40 and IL-12 in cross- priming168. A recent study examining the transcriptional profile of antigen specific CD8+ T cells activated in the presence of IL-12 or Type I IFNs revealed that both cytokines are important for regulating the expression of key transcription factors such as T-bet and eomesodermin (Eomes).

These transcription factors are essential in T cell programming for the primary expansion and production of effector cytokines such as IFN-γ and granzyme B. Moreover, IL-12 and Type I

IFNs can regulate histone acetylation activity that promotes chromatin remodeling, leading to increased gene accessibility and transcriptional activity147,169.

Altogether, CD40 signaling in DC can promote further DC maturation and secretion of IL-

12 which is critical for CD8+ T cell programming. Both CD40 and the resulting induction of IL-

12 are particularly important for clearance of pathogens that require CD4+ T cell help. Aside from its function in DC, CD40L signals provided by CD4+ T cell help can also play an important role in B cell activation, leading to antibody isotype switching and affinity maturation. The role of CD40 in B cell activation will not be discussed here.

1.6.2 Lymphotoxin-α1β2 and Lymphotoxin-β receptor

Lymphotoxin was named after the cytotoxic activity observed in activated lymphocytes170.

LTαβ is composed of one LT-α unit and two of LT-β to form a heterotrimeric complex, with

LTα being tethered to the cell surface via LTβ. In the mid 1980's, with the advance of molecular cloning, LT-α was categorized as a TNF family member along with TNF-α and 20 other TNF family like molecules171. TNF-α and LT-α share approximately 50% homology and were thought to be functionally redundant, especially since LT-α can form a homotrimer that signals through

30 the same receptors that bind TNF-α: TNF-R1 and TNF-R2. Given the similarity, LT-α was later renamed as TNF-β. However, upon the discovery of LT-β as the second subunit of the membrane-bound LT complex (LTαβ) which binds a specific cognate receptor LTβR, it was appreciated that LTαβ could exert distinct physiological functions from TNF-α and LT-α172.

Furthermore, TNF-α is primarily produced by innate immune cells such as macrophages and neutrophils, while LTαβ is mainly expressed by lymphocytes, hence the name TNF-β has since been abandoned.

Other than LTαβ, LTβR can be activated by a second TNF family member LIGHT

(lymphotoxin-like, exhibits inducible expression, and competes with herpes simplex virus glycoprotein D for HVEM, a receptor expressed by T lymphocytes). LIGHT is expressed by immature DC, but down-regulated upon DC activation. Some evidence have suggested that

LIGHT expression is required to mount an efficient CTL response173, but an in vivo-ex vivo study showed that LIGHT is dispensable for CD8+ T cell cross-priming174. LIGHT is expressed on activated T cells and it can bind another TNFRSF member, the herpesvirus entry mediator

(HVEM). However, HVEM can interact with the B and T lymphocyte attenunator (BTLA), an immunoglobulin superfamily member that resembles the inhibitory co-stimulatory receptor

CTLA-4 and PD1175. BTLA binding to HVEM has been shown to negatively regulate T and B cell functions, and LIGHT expression can sequester HVEM to prevent its inhibitory effects

(Figure 1.4).

1.6.2a LTβR signaling

Like other members of the TNFRSF, LTβR homotrimerizes upon activation and recruits

TRAF-2, -3 and -5 to mediate downstream signaling pathways that lead to JNK, classical

Figure 1.4 Members of the lymphotoxin system and the interacting partners.

LTα is expressed as a soluble homotrimer that can bind TNFR1 and TNFR2. LTα can also heterotrimerize with two units of LTβ forming the LTα1β2 complex that specifically bind LTβR. The homotrimeric LIGHT also recognizes and bind LTβR, in addiction it also engages HVEM which can also be inhibited by BTLA through competitive binding.

NFκB176 and alternate NFκB activation177. TRAF2 activation has been shown to be required for the classical NFκB pathway and JNK activation downstream of LTβR signaling176. Similar to

CD40 signaling, TRAF-2, -3, -5 and cIAP1/2 work together to negatively regulate alternative

NFκB activation through constitutive degradation of NIK in a similar mechanism as described in an earlier section. The NFκB family is a group of transcription factors that comprises Rel-A, c-

Rel, Rel-B, p50 (p105) and p52 (p100). In the steady state, a family of cytoplasmic inhibitory proteins IκBs sequester NFκB members to prevent them from entering the nucleus for gene transcription. Following receptor activation, LTβR recruits the IKK complex which includes the

IKKα, IKKβ and IKKγ for their phosphorylation. The IKK complex further phosphorylates the

IκB kinases for their degradation, resulting in the release of p105 for proteasome mediated processing into the active p50 subunit. LTβR mediated classical NFκB activation can induce pro- inflammatory mediators such as MIP-1β and MIP-2, as well as DC survival and homeostatic proliferation177. LTβR activated classical NFκB activation is rapid while the alternative NFκB activation is gradual. Studies examining alternative NFκB signaling showed that it is critical for lymphoid organogenesis, leading to production and secretion of chemokines CCL19, CCL21 and

CXCL13 in stromal cells178. Although other TNFRSF members such as CD40, RANK and

BAFF receptor can also activate the alternate NFκB signaling, organogenesis is primarily governed by LTβR mainly due to restricted expression of the receptors and ligand availability.

1.6.2bThe role of LTβR signaling in immune system homeostasis

LTαβ is expressed on lymphoid tissue inducer cells (LTi), natural killer cells, resting follicular B cells and activated T and B cells, whereas LTβR is found on stromal cells such as

FDC and FRC in the SLO, high endothelial venules (HEV), macrophages and DC179–182. Since the discovery of the LT pathway, studies have characterized the importance of LT-α, LT-β and

LTβR in organogenesis and maintenance of secondary lymphoid organs. Mice lacking any one of the LT components fail to develop peripheral lymphoid organs that include lymph nodes, Peyer's

Patches and disrupted structure in the spleen. During embryonic development, LTi cells generated from fetal liver precursors interact with stromal cells through LTβR signaling to direct lymphoid organogenesis. LTβR activation leads to stromal cell differentiation and secretion of

CXCL13 that further activates and recruits LTi cells that express the cognate receptor CXCR5, resulting in a positive feedback signal. This interaction promotes iterative interactions between

LTi and stromal cells in order to initiate early lymphoid structures. Further recruitment of endothelial cells to these structures trigger their differentiation into HEV183. In the adult animal, a matured stromal cell network is supported by LTβR signaling: Constitutive LTβR signaling results in the secretion of chemokine gradients of CXCL13 and CCL19/CCL21 that defines the B and T cell zones184,185 hence, LT-deficient animals exhibit disrupted splenic structure.

In addition to the establishment and maintenance of secondary lymphoid organs, LTβR signaling is also important for the homeostatic proliferation of some DC subsets. In particular, the homeostatic proliferation of CD11b+ CD8- Esam+ DCs in the spleen has been shown to rely on LTαβ ligand expressed by B cells12,186. Similarly in the gut, the CD11b+ CD103+ DC subset within the lamina propria of the small intestine relies on Notch 2 signaling for development, while LTβR signaling is important for their homeostatic proliferation8. In order to maintain a balance within DC populations, the LTβR-mediated DC proliferative signals is also counter- regulated by BTLA binding to HVEM that negatively regulates the CD8- DC subset187.

1.6.2c The involvement of the LT pathway in immune responses.

Many studies have implicated LTαβ and LTβR in immunity against different infections.

LTα, LTβ or LTβR deficient mice have been shown to be susceptible to Mycobacterium, Listeria and murine cytomegalovirus (mCMV) infections188–191. Even though the lack of secondary lymphoid structures may contribute in part to observed susceptibility to these infections, other groups have generated chimeric mice or conditional knock out mice that restricted the absence of

LTβR signaling within the immune compartment, while leaving lymph node organogenesis intact. Even under these conditions, mice lacking LTβR signaling succumb to LCMV infection with impaired CTL immunity, suggesting that LTβR signaling can modulate T cell responses and affect disease status192,193. Moreover, LTβR signaling has also been shown to be involved in some autoimmune diseases, where inhibition of the LTβR pathway using a decoy LTβR inhibitor

(LTβR-Ig) can successfully relieve or interrupt pathology associated with a number of T cell mediated autoimmune disease models including experimental autoimmune encephalomyelitis

(EAE)194, non-obese diabetes195–197, collagen induced arthritis198, inflammatory bowel disease199,200 and graft-versus host disease201–203. The mechanism for this protection in the setting of autoimmune disease has not yet been well characterized. In some cases, impaired T cell responses have been invoked, whereas others have implicated the malformation of ectopic lymphoid follicles as an important aspect of therapeutic efficacy. In addition, previous studies have hinted that LTβR signaling may contribute to Type I IFN expression, where in vitro treatment of human fibroblasts with soluble LTαβ can induce LTβR-mediate IFN-β expression that protect the cells from human CMV infections189. In addition, transcriptional analysis of

LTβR deficient mice infected with Listeria showed a loss of IFNα expression as well as all of the

Type I IFN regulated genes204, further suggesting that LTβR signaling regulates Type I IFN

35 responses. Therefore, in this thesis I will explore the link between the LT pathway and Type I

IFNs (the LT/Type I IFNs axis) as a potential driver of T cell responses to foreign and self- antigen.

1.7 Type I IFNs

IFNs were first discovered to be important anti-viral factors that interfere with viral replication in mammalian cells. There are three classes of IFN (Type I, II and III) categorized based on their structural homology and the specific receptor with which they associate 205. The

Type I IFNs family is very diverse and includes a single IFN–β member, numerous IFN-α variants, and the lesser known IFN -ε, -κ, - ω, and –δ 206. Despite this diversity, all Type I IFNs bind exclusively to the interferon-α receptor (IFNAR). Upon receptor activation, IFNAR1 and

IFNAR2 dimerize and phosphorylate the Janus Kinase family members TYK2 and JAK1.

Activated TYK2 and JAK1 phosphorylate STAT1 and STAT2, and together they bind IRF-9 to form a trimeric transcription factor, ISGF3. ISGF3 translocates into the nucleus and interacts with the ISRE elements to activate IFN-related gene transcription 205. Type I IFNs are secreted by various cell types including fibroblasts, epithelial cells, innate immune cells and lymphocytes, and they represent a key initiating factor against viral infections and an important signal that modifies T cell responses.

1.7.1 Pro-inflammatory functions of Type I IFNs

Type I IFNs are generally regarded as a pro-inflammatory cytokine in numerous immune settings such as autoimmune diseases like psoriasis and systemic lupus erythematosus 34,207,208, allograft rejection in transplantation 209, and immunity against tumors 210,211. Type I IFNs have robust pro-inflammatory effects that can act in both an autocrine and paracrine manner on

36 immune cells to modulate their functions. During an immune response with minimal PRR activation, DCs cannot reach maximal immunogenic status, and Type I IFNs overcome this hurdle by promoting DC maturation 212. IFNAR activation in DC triggers NFκB and p38 MAPK activation, resulting in the up-regulation of MHC class I and class II as well as co-stimulatory molecules B7-H1 and B7-H2 213. In addition, blood circulating monocytes, when differentiated into DCs in the presence of Type I IFNs, up-regulate the chemokine receptor CCR7 thus allowing them to migrate more efficiently into SLOs214. Type I IFNs can complement IL-12 to drive TH-1 differentiation (human T cells), where IFNAR-mediated STAT2 phosphorylation recruits and activates STAT4, a transcription factor that potentiates IL-12R signaling215.

Moreover, Type I IFNs signaling through STAT4 has also been implicated in the induction of

IFN-γ in natural killer cells and T cells, particularly in the absence of STAT1216. Various studies have reported that Type I IFNs acts as a potent third signal that promotes antigen cross- priming148,212,217. IFNAR activation in CD8+ T cell can induce chromatin remodeling through histone acetelyation that promotes the transcription of many genes required for clonal expansion and production of effector molecules147. Lastly, Type I IFNs prolong the CD8+ T cell expansion phase in response to cross-presented antigen, and it enhances the responsiveness of antigen specific CD8+ T cells to IL-2 and IL-15 for increased survival217. Hence, Type I IFNs can independently act on DCs, CD4+ T cells and CD8+ T cells through very different mechanisms that facilitate inflammatory immune responses.

1.7.2 Anti-inflammatory roles of Type I Interferon

Many experimental and clinical settings have used Type I IFNs as a treatment to quiet inflammatory conditions, suggesting that Type I IFNs can also exert immunoregulatory functions. In particular, IFN-β has been shown to be an effective treatment for collagen-induced

37 arthritis 218,219, relapsing-remitting multiple sclerosis (MS)220 and autoimmune familial

Mediterranean fever221,222. The anti-inflammatory functions of Type I IFNs, particularly in MS, have been characterized by numerous studies, and yet the exact mechanism of action remains unclear. Blood-derived DCs become activated upon IFNAR activation, however astrocytes and microglial cells in the central nervous system (CNS) down-regulate MHC-II in response to Type

I IFNs 223,224. T cell recruitment into the CNS may require Type I IFNs to induce relevant

CXCR3 attracting chemokines, however other studies have also showed that prolonged IFN-β treatment in MS patients down-regulates the expression of cell adhesion molecules such as

VCAM-1 and ICAM-1 in brain endothelial cells, resulting in reduced immune cell infiltration into the CNS225,226. IFN-β or IFNAR1 deficient mice have been shown to produce an enhanced number of antigen specific CD8+ T cells when immunized with a DNA-based vaccine, suggesting that Type I IFNs are also required to control T cell proliferation227. Recent studies showed that Type I IFNs induced STAT1 activation negatively regulates the expression and function of the proto-oncogene c-myc in CD8+ T cells which is important for homeostatic proliferation228,229. Furthermore, other studies have also shown that IFNAR or IFN-β deficient mice exhibit lower numbers of IL-10 producing T cells, which may also explain the increased

CD8+ T cell expansion in the absence of IFNAR signaling227,230. It is important to point out, however, that scenarios where IFNAR signaling is completely absent may lead to different effects on shaping T cell responses than situations where the levels/kinetics of Type I IFN production have been altered.

1.8 Summary

Given the pleiotropic nature of Type I IFNs that produce potent effects on various levels of the immune system, and strong evidence that demonstrate that LTβR signaling can regulate Type

I IFN induction 191,204,231,232, this study was set out to identify and characterize how LTβR signaling in DC regulates Type I IFN expression and the impact of the LT/Type I IFNs axis in

CD8+ T cell cross-priming. This thesis contains two result chapters: The first chapter describes the role of LTβR signaling in DC, and examines how DC-intrinsic LTβR signaling can affect the clonal expansion of CD8+ T cells. In addition, I show that LTβR stimulation can directly elicit a

Type I IFN response in DC.

In the second data chapter, I further explore the importance of the LTβR/Type I IFNs axis in an inducible autoimmune disease mouse model. I found that LTβR signaling in DC not only affects antigen specific CD8+ T cell expansion, but the expression of adhesion molecules LFA-1 and VLA-4 by antigen specific CD8+ T cells and their ability to infiltrate a target tissue.

Furthermore, I identified a role for TRAF3 in the LTβR-mediated Type I IFN response.

CHAPTER 2

LTβR signaling in Dendritic Cells induces a Type I IFN response that is required for optimal clonal expansion of CD8+ T cells

______

This chapter is taken from a published manuscript: Summers deLuca L1, Ng D1, Gao Y1, Wortzman ME, Watts TH, Gommerman JL. Proc Natl Acad Sci U S A. 2011 Feb 1;108(5):2046-51. doi: 10.1073/pnas.1014188108. Epub 2011 Jan 18. 1These authors contributed equally Leslie-Summers deLuca and Yun Fei- Gao provided figure 2.1, and 2.2 Leslie-Summers deLuca provided figure 2.3-2.5, 2.7 and 2.8

2.1 Abstract

During an immune response, antigen-bearing dendritic cells (DC) migrate to the local draining lymph node and present antigen to CD4+ helper T cells. Antigen-activated CD4+ T cells then up-regulate Tumour Necrosis Factor (TNF) super-family members including CD40-ligand and Lymphotoxin (LT)-α β. Although it is well accepted that CD40 stimulation on DC is required for DC licensing and cross-priming of CD8+ T cell responses, it is likely that other signals are integrated into a comprehensive DC activation program. Here we show that a cognate interaction between LTαβ on CD4+ helper T cells and LTβ-Receptor (LTβR) on DC results in unique signals that are necessary for optimal CD8+ T cell expansion via a Type I IFN-dependent mechanism. In contrast, CD40 signaling appears to be more critical for CD8+ T cell IFNγ production. Therefore, different TNF family members provide integrative signals that shape the licensing potential of antigen-presenting DC.

2.2 Introduction

CD8+ T cell responses are crucial for host responses to viral infection and are also involved in allograft rejection and tumour immunity. In some settings, provision of T cell help is required to adequately license DC for cross-priming of CD8+ T cell responses233. However, the nature of these help signals remains incompletely characterized, and it is unclear how these signals integrate into a program of DC maturation. Manipulation of such signals represents a promising therapeutic approach for promoting tumour immunity or for quieting autoimmune disease.

Tumour necrosis factor (TNF) family members including CD40-ligand (CD40L), RANK- ligand, LIGHT and Lymphotoxin-αβ (LTαβ) are rapidly up-regulated on antigen (Ag)-activated

CD4+ T helper cells234,235. During the immune response, DC:T cell interactions result in the

41 ligation of CD40, CD70 and RANK on DC, and this has been shown to promote DC cross- priming capacity and survival150–152,236,237. We have recently shown that LTαβ expression on Ag- specific CD4+ T cells is also critical for DC function in the context of protein Ag235. These studies provoke the question of whether different TNF family pathways are redundant, or somehow act cooperatively in the context of DC maturation.

Previous studies have hinted at a role for the LT pathway in T cell function238. Focusing on cases where CD8+ T cell responses rely on T cell help, such as CD8+ T cell responses to allo-

Ag239,240, and tumour-Ag241, LTβR signaling has a significant effect on CD8+ T cell activation and clonal expansion. To resolve how this pathway may impact the maturation of a CD8+ T cell response in vivo, we used different approaches whereby we selectively inhibited LTβR signaling on the hematopoietic compartment, or specifically on DC, to evaluate effects of LTβR signaling on DC-mediated CD8+ T cell cross-priming. Our data revealed that the LT pathway was important for CD8+ T cell clonal expansion, but not effector function, whereas the CD40 pathway was necessary for CD8+ T cell function but was dispensable for T cell expansion in response to cell-associated Ag. LTβR stimulation on DC was found to provoke a Type I IFN response, even in the absence of added TLR agonist, and exogenous IFN-α could recover CD8+

T cell proliferation, suggesting a mechanism for the effects of this pathway on CD8+ T cell priming. Therefore, the LT pathway provides necessary and non-redundant DC-intrinsic signals that provoke optimal CD8+ T cell clonal expansion.

2.3 Results

2.3.1 LTβR signaling cooperates with CD40-derived signals for priming of CD8+ T cells in vivo

We previously demonstrated that the expression of LTαβ ligand on Ag-specific CD4+ T cells in vivo is required for DC function ex vivo235. However, a hallmark of DC licensing is the ability to cross-prime a CTL response. In order to ascertain the relevance of LTβR signaling on

DC licensing, we assessed whether LTβR signaling was required for priming of a CD8+ T cell response to cell-associated Ag in vivo. To achieve a relatively CD4+ T cell help-dependent system, spleen and LN cells from Bm1 mice were used as syngeneic vehicles for Ag ovalbumin

(OVA) delivery so that Bm1 cells could not directly present Ag to OTI T cells in Bm1xB6 F1 recipient hosts142. Bm1 cells were hypotonically loaded with OVA protein150 and were used to immunize mice that had received congenically labeled OTI T cells one day prior. We first assessed the consequences of global inhibition of LTβR signaling by treating recipient mice with a decoy fusion protein, LTβR-Ig, which was administered the day prior to immunization with

Bm1-OVA. We found that, compared to the control treatment group, OTI T cell expansion was significantly impaired in LTβR-Ig treated recipient mice (Figure 2.1A). Reduced frequency of

OTI T cells was observed throughout the immune response, from days 3-21, in the spleen (see

Figure 2.2 for representative FACS) and the blood (Figure 2.1A). However, in spite of this observed decrease in clonal expansion, OTI T cells exhibited no defect in IFNγ production at any time point, with equivalent proportions of IFNγ-producing OTI T cells in both control and

LTβR-Ig treated mice (Figure 2.1B, see Figure 2.2 for representative FACS). Since the homeostasis of DC in naïve mice is perturbed in LTβR-/- animals242–244, we confirmed that short term treatment with the LTβR-Ig agent did not result in a reduction in DC

Figure 2.1 DC-derived LTR and CD40 signals contribute distinctly to CD8+ T cell cross- priming in vivo.

(A-B) WT BM1xB6 F1 mice were given an A/T of responder CD45.1 or Thy1.1 OTI T cells, were treated with either control huIgG or LTR-Ig, and were immunized the next day with OVA-loaded Bm1 cells. At multiple days post-immunization, and OTI expansion (A) and IFN production (B) was measured. Data is representative of 6 (A) or 2 (B) independent experiments (n=3-6 per experiment). (C-D) WT -> WT or LTR-/- -> WT chimeric mice were given an A/T of responder CD45.1 or Thy1.1 OTI T cells, were treated with either control Ab or CD40L, and were immunized with OVA-loaded Bm1 cells, and OTI expansion (C) and IFN production (D) was measured at day 3. These experiments were performed 4 times with similar results, and results shown represent the average of 4 mice/group. (D-E) Mixed chimeric WT + CD11c-DTR - > WT or LTR-/- + CD11c-DTR-> WT mice were given an A/T of responder CD45.1 OTI T cells, were treated with diphtheria toxin, and were immunized with OVA-loaded Bm1 cells, and OTI expansion (E) and IFN production (F) was measured at day 3. Data is representative of 2 experiments (n=4 per experiment). Empty circles = control Ig treated mice, filled circles = LTR-Ig treated mice. White bars = control treated mice, Grey bars = -CD40L treated mice. *p<0.05, ***p<0.0001 (2-way ANOVA for (A)).

Figure 2.2 Representative FACS plots of OTI T cell expansion/IFN production following LTR-Ig treatment.

WT BM1xB6 F1 were given an A/T of responder CD45.1 or Thy1.1 OTI T cells, were treated with either control huIgG or LTR-Ig, and were immunized the next day with OVA-loaded Bm1 cells. At multiple days mice post-immunization OTI expansion (A) and IFN production (B) was measured in the spleen re-stimulated with SIINFEKL peptide (1uM) ex vivo. Data is representative of 3 independent experiments (n=3-6 per experiment). Black histogram = huIgG- treated; grey histogram = LTR-Ig-treated, grey filled histogram = unimmunized.

45 numbers, an alteration in DC phenotype, or an inability to acquire and process OVA protein for presentation on MHC class I (Figure 2.3). Therefore, LTβR-Ig treatment results in sub-optimal clonal expansion, but not effector function, of CD8+ T cells in response to cell-associated Ag.

CD40 signaling is a potent maturation cue during DC activation and can also induce CD86 expression on DC245. However these data indicate that in the context of abrogated LTβR signaling, physiological CD40 signaling cannot compensate to maintain DC stimulatory function, and it is unclear whether signals derived from both CD40 and LTβR act additively or synergistically to license DC for CD8+ T cell cross-priming, or alternatively whether these two pathways contribute something distinct to DC maturation. To resolve this, we generated WT ->

WT or LTβR-/- -> WT chimeric mice, and at 8 weeks post-reconstitution these mice were injected with OTI T cells and immunized with cell-associated OVA as in Figure 2.1A. At day 3, which in our system is the peak of the CD8 (OTI) response (see Figure 2.1A for kinetics), we observed robust expansion of the OTI T cells in WT -> WT chimeras, however in mice lacking

LTβR expression on Ag-presenting cells (APC), we observed a statistically significant reduction in OTI T cell accumulation (Figure 2.1C white bars). Not unlike LTβR-Ig treated mice, OTI T cells primed in LTβR-/- -> WT mice exhibited no defect in IFNγ production, with equivalent proportions of IFNγ-producing OTI T cells in both groups (Figure 2.1D white bars). We also confirmed that DC from LTβR-/- mice expressed normal levels of maturation markers and could acquire and process OVA protein for presentation on MHC class I in a manner that was comparable with DC from WT mice (Figure 2.4). Therefore, under circumstances where the organization of splenic stroma is normal (which depends on LTβR expression in radio-

Figure 2.3 Evaluation of DC phenotype/function in LTR-Ig treated mice

(A) Mice were treated on day 0 with control huIgG or LTR-Ig. In some cases, mice were immunized with BM1-OVA on day 1 and then splenic DC were harvested 18 hours later and co- incubated with the B3Z responder T cells over-night to evaluate peptide loading on MHCI. SIINFEKL:MHCI presentation was measured by colorimetric detection of β- galactosidaseactivity using CPRG substrate. These experiments were performed 2 times with similar results, and results shown represent the average of 3 mice/group. *p<0.05, ***p<0.001. (B) Mice were treated on day 0 with control huIgG or LTR-Ig and splenic DC were enumerated as a % of total splenocytes (B) or as total DC per spleen (C) following Collagenase D (Sigma) and DNase I (Sigma) digestion. DC were further analysed with respect to the distribution of DC subsets and expression of surface markers with a representative example shown in (D). No statistically significant differences were observed between huIgG or LTR-Ig treated groups. The experiment was performed using 3 mice per group.

Figure 2.4 Evaluation of DC phenotype/function in LTR-/- treated mice.

(A) WT -> WT or LTR-/- -> WT chimeric mice were immunized with BM1-OVA on day 1 and then splenic DC were harvested 18 hours later and co-incubated with the B3Z responder T cells over-night to evaluate peptide loading on MHCI. SIINFEKL:MHCI presentation was measured by colorimetric detection of β-galactosidase activity using CPRG substrate. This experiment was performed 2 times with similar results, and results shown represent the average of 3 mice/group. *p<0.05, ***p<0.001. (B) Splenic DC from WT -> WT or LTR-/- -> WT chimeric mice were enumerated as a % of total splenocytes (B) or as total DC per spleen (C). DC were further analysed with respect to the distribution of DC subsets and expression of surface markers with a representative example shown in (D). No statistically significant differences were observed between WT -> WT or LTR-/- -> WT chimeric mice with the exception of that the frequency of CD4+CD11b+ DC was reduced in LTR-/- -> WT chimeric mice. The experiment was performed twice using 3 mice per group for each experiment. **p<0.001.

48 resistant host cells), expression of LTβR in radio-sensitive hematopoietic cells is required for full clonal expansion of OTI T cells.

To determine if this CD8+ T cell defect would be exacerbated by the additional absence of

CD40 licensing cues, WT -> WT and LTβR-/- -> WT BM chimeric mice were treated with α-

CD40L blocking Ab. In contrast to the LTβR-/- -> WT BM chimeric mice, we observed no defect in OTI T cell expansion in WT -> WT mice in which CD40 signaling was abrogated, and there was a minimal compound defect beyond the impaired OTI expansion in the absence of LTβR signaling when both CD40 and LTβR signaling were simultaneously abrogated (Figure 2.1C grey bars). While secretion of IFNγ by OTI T cells was uncompromised in the absence of LTβR licensing, it was grossly impaired in α-CD40L treated mice (Figure 2.1D grey bars). Together these results identify unique roles for LTβR and CD40 signaling in promoting CD8+ T cell responses to cell-associated Ag, the former in regulating CD8+ T cell expansion and the latter in instructing CD8+ T cell effector function.

2.3.2 DC-intrinsic LTβR signaling is required for optimal CD8+ T cell clonal expansion

We next addressed whether DC, which express LTβR243, are the relevant LTβR+ hematopoietically-derived cell required for cross-priming CD8+ T cells. We therefore generated mixed BM chimeras using CD11c-DTR/GFP donor BM along with either WT or LTβR-/- donor

BM transferred into lethally irradiated WT hosts, and reconstitution of BM-derived cells was confirmed using GFP and CD45 congenic markers. CD11c-DTR mice express a diphtheria toxin

(DT) receptor under the control of the CD11c promoter, and treatment of these mice with DT results in thorough but temporary depletion of DC246. Treatment of LTβR-/- + CD11c-DTR ->

WT chimeric mice with DT would therefore specifically deplete WT CD11c+ cells while

49 preserving LTβR-/- DC. At 12 weeks post-reconstitution, chimeric mice were given an adoptive transfer (A/T) of responder OTI T cells, and the following day were immunized with OVA- loaded Bm1 splenocytes. Chimeric mice were treated with DT on day 0 and day 1 post- immunization, and depletion of CD11c-DTR+ (GFP+) DC in the blood, spleen and LN was confirmed. At the peak of the CD8+ response, expansion of OTI T cells was significantly impaired in LTβR-/- + CD11c-DTR chimeras in terms of frequency (p<0.001; Figure 2.1E), and also in the case of total numbers of OTI (p<0.05; Figure 2.5), while IFNγ production remained intact (Figure 2.1F), recapitulating the defect observed in LTβR-/- -> WT chimeras. Consistent with these data, mice that received LTβ-deficient helper T cells also exhibited a significant reduction in the peak expansion of splenic OTI CD8+ T cells (Figure 2.6A) but not IFNγ production (Figure 2.6B) following immunization with OVA-loaded Bm1 cells, suggesting that cross-talk between LTαβexpressing Ag-specific T cells and LTβR+ DC is required for optimal clonal expansion of CD8+ T cells in response to cell-associated Ag. Collectively, these data identify DC intrinsic LTβR signaling as a requirement for CD8+ T cell clonal expansion, but not for CD8+ T cell-derived IFNγ production.

2.3.3 LTβR signaling is required for full activation and cell cycle progression of Ag-specific CD8+ T cells

Given the reduction in the clonal burst of Ag-specific CD8+ T cells in immunized LT- inhibited mice, we asked whether the activation, proliferation or persistence OVA-specific CD8+

T cells was impaired. At two days post-immunization, we measured cell division/activation, and noted a lag in CFSE dilution as well as a significant reduction in CD25 up-regulation on OTI T cells derived from LTβR-/- -> WT mice as compared to OTI T cells derived from WT -> WT mice (Figure 2.7A, B). The reduction in CFSE dilution suggested that in the absence of LTβR-

Figure 2.5 Expression of LTR on DC is required for optimal numbers of OTI at the peak of the immune response in vivo

Mixed chimeric WT + CD11c-DTR -> WT or LTR-/- + CD11c-DTR-> WT mice were given an A/T of responder CD45.1 OTI T cells, were treated with diphtheria toxin, and were immunized with OVA-loaded Bm1 cells, and OTI expansion. Number of OTI per spleen was enumerated for each mouse. *p<0.05.

Figure 2.6 Expression of LT on Ag-specific CD4+ T cells is required for optimal expansion of OTI in vivo and in vitro.

WT B6 mice were given an A/T of either WT or LT-/- helper OTII T cells, along with responder CD45.1 OTI T cells, and were immunized the next day with OVA-loaded Bm1 cells. At 3 days post-immunization, OTI expansion (A) and IFN production (B) was measured in the spleen. *p<0.05. Black bars = unimmunized mice, white bars = immunized mice that received WT OTII T cells, grey bars = immunized mice that received LT-/- OTII T cells. The experiment was performed 2 times with similar results. (C) BMDC were pre-incubated with LPS and OVA protein for 18 hours, washed, and then plated with OVA-specific CFSE-labeled OTI T cells and in some cases WT or LT-/- OTII T cells. 72 hours later, OTI T cells were assessed for CFSE dilution. This experiment was performed two times with similar results.

Figure 2.7 LTR signaling is required for normal CD8+ T cell activation and cell cycle completion.

Chimeric mice (WT -> WT or LTR-/- -> WT) were given an A/T of CFSE-labeled responder OTI T cells, then were either left unimmunized or were immunized one day later with OVA- loaded Bm1 cells. At day 2 post-immunization, CFSE dilution and CD25 up-regulation were evaluated (A), and the extent of CD25 up-regulation on divided versus undivided OTI was measured (B). Data is representative of 3 independent experiments (n=3-5 mice per experiment), **p<0.01. (C) To assess cell cycling in the described in vivo experiment from A-B, CFSE- stained OTI T cells were gated and analysed with DyeCycle Violet to measure cell cycle status, with a representative FACS plot shown in (i). Enumeration of OTI that had divided (CFSEmed (ii)), had “terminally divided” (CFSE- (iii)), or had remained undivided (CFSEhigh (iv)), was performed. Data is representative of 2 experiments (in one case performed at day 2, and another at day 3, n=4 per experiment). Note that in the presence of DyeCycle Violet that CFSE intensity appears different than without DyeCycle Violet (A versus C).

53 derived DC licensing, OTI T cells were either not dividing, or were dividing but were failing to survive. We therefore measured the cell cycle status of CFSE-labeled OTI from OVA-Bm1 immunized chimeric mice. Interestingly, a significant increase in the percent of cycling, CFSEint

OTI in LTβR-/- -> WT compared to WT -> WT chimeric mice was observed (Figure 2.7C, p <

neg 0.01), and this was accompanied by a significant reduction in the percent of resting/G1 CFSE

(fully divided) OTI in LTβR-/- -> WT versus WT -> WT chimeric mice (p < 0.005). Staining for cleaved caspase 3 leading up to, and at the peak of the OTI response, showed no differences in apoptotic OTI in WT -> WT and LTβR-/- -> WT mice, however analysis of dying cells in vivo is unreliable since they are quickly phagocytosed, and while apoptotic demise is common, there exist additional death pathways that may not be captured by techniques designed to measure apoptotic cell death247. It therefore seems likely that OTI T cells primed in LTβR-/- -> WT mice are cycling in equivalent proportion to those primed in WT -> WT mice, but are dying prior to reaching a terminally-divided resting state.

2.3.4 Stimulation of LTβR on DC results in the production of Type I IFNs

To determine the mechanism whereby DC-intrinsic LTβR signaling contributes to the clonal expansion of CD8+ T cells in vivo, we investigated the possibility that production of Type

I IFNs was downstream of LTβR activation in DC for several reasons: Firstly, LTβR signaling has been shown to induce Type I IFNs in radio-resistant stromal cells independent of TLR- derived signals248. Secondly, Type I IFNs have been shown to induce CD25 expression on T cells249, is required for the optimal clonal expansion of CD8+ T cells139,148, can exert these pro- proliferation effects independent of CD40/CD40L activity250. Finally, Type I IFNs induce the expression of CD86 on DC251, and we have observed a transient decrease in CD86 expression on

DC post OVA immunization in vivo that is recovered in the presence WT DC (Figure 2.8)

54 suggesting a factor that acts in trans can rescue CD86 expression. Therefore, we reasoned that the poor OTI expansion, the reduced expression of CD25 on OTI T cells, and the failure to up- regulate CD86 on LTβR-/- DC could all reflect a defect in Type I IFN production. To assess whether LTβR signaling induces Type I IFNs in conventional DC, we generated bone marrow derived DC (BMDC) and co-cultured WT versus LTβR-/- BMDC with OVA-specific CD4+ helper OTII T cells as a source of licensing signals (CD40L, LTαβ). These co-cultures were then supplemented with either LPS alone or LPS plus OVA323-339 to induce the expression of

LTαβ/CD40L on OTII 235. Exposure to LPS alone induced a baseline level of IFNβ and IFNα5 gene expression which did not differ between WT and LTβR-/- DC (open vs closed circles,

Figure 2.9A, B). However, the addition of the OTII cognate peptide, OVA323-339 to the DC/T cell co-cultures, which activates OTII CD4+ T cells and stimulates them to express LTαβ/CD40L235, resulted in a robust induction of both IFNβ and IFNα5 gene expression that was blunted in the absence of LTβR expression on DC (open vs closed circles, Figure 2.9C, D). Indeed, the induction of IFNβ and IFNα5 gene expression, both of which have been shown to be expressed at early time points following PRR activation252, from LTβR-/- DC in response to LPS plus

OVA323-339 was not any different than what was observed with LPS alone. The reduction in IFNβ and IFNα5 gene expression from LTβR-/- DC was not due to any developmentally-associated

DC-intrinsic defect since LTβR-/- BMMDC were similar to WT BMDC in terms of DC surface marker expression and their capacity to make IL-12 (Figure 2.10). Moreover, blockade of

LTβR/LTαβ interactions between T cells and WT BMDC with LTβR-Ig in vitro recapitulated the reduction in IFNβ and IFNα5 expression (Figure 2.11). These data indicate that DC-intrinsic

LTβR ligation induces a necessary and unique signal(s) that collaborates with TLR4 stimulation to provoke optimal IFNβ and IFNα5 gene expression.

Figure 2.8 LTR-/- DC display altered CD86 up-regulation

CD45.1-> WT CD45.2 (white fill) and LTR-/- CD45.1-> WT CD45.2 (grey fill) chimeric mice were given an A/T of helper WT OTII T cells and were immunized one day later with OVA protein and LPS. At 18h (A) and 36h (B), dLN were collected, digested and stained to measure CD86 expression on CD11c+ DC by flow cytometry. A reduction in CD86 expression on LTR-/- DC was observed at 18 hours post-immunization, but CD86 expression was recovered at 36 hours post-immunization. Experiment was performed 3 times with similar results. *p<0.05 (C) WT CD45.2 + LTR-/- CD45.1 -> LTR+/- CD45.1/2 mixed bone marrow chimeras were given an A/T of helper WT OTII T cells and were immunized one day later with DQ-OVA protein and LPS in one foot. CD86 expression was tracked on WT (filled circles) vs LTR-/- (empty circles) CD11c+ dLN DC (black lines) as well as on DQ-OVA+ CD11c+ dLN DC (grey lines). Note that in the case of mixed BM chimeras, no difference in CD86 was observed suggesting that WT can rescue CD86 expression on LTR-/- DC in trans.

Figure 2.9 LTR synergizes with TLR4 to maximize Type I IFN production in BMDC

WT (open circles) and LTR-/- (grey circles) BMDC were co-incubated with OVA-specific CD4+ T cells (OTII) and LPS (A-B). In some cases, these co-cultures were also supplemented with OVA323-339 (C-D). DC were isolated at indicated time-points and the expression of IFN (A,C) and IFN5 (B,D) was measured by real time RT-PCR. This experiment was performed 3 times with similar results. *p<0.05

Figure 2.10: LTR+/-BMDC express normal levels of co-stimulation markers before and after LPS stimulation, and are capable of producing IL-12 production in vitro.

WT (open circles/bars) and LTR-/- (grey circles/bars) BMDC were directly stimulated with anti- CD40 and anti-LTR Abs. At 30 hours, DC were collected and IL-12p40 expression was evaluated by flow cytometry (A, B), where fold over-background indicates the amount of IL12 expression, expressed as mean fluorescence intensity (MFI) compared to un-stimulated BMDC. Note that anti-LTR treatment does not induce IL-12p40 expression. Experiment was repeated 2 times with similar results. (C) Indicated surface markers on WT and LTR-/- BMDC were evaluated before and after 18 hour treatment with LPS.

Figure 2.11 Type I IFN production induced by LPS + OVA is attenuated in WT BMDC by LTR-Ig treatment

WT BMDC were co-incubated with OVA-specific CD4+ T cells (OTII) and LPS (A-B) and

OVA323-339. In some cases, these co-cultures were also supplemented with LTR-Ig to block LTR signaling in DC. The expression of IFN and IFN5 was measured by real time RT-PCR. This experiment was performed twice with similar results. * p < 0.05

We next assessed whether LTβR signals could provoke Type I IFN expression independent of TLR signaling by co-culturing WT versus LTβR-/- BMDC with recombinant LTαβ ligand or with agonist Abs directed at LTβR. Interestingly, in the absence of any added TLR signal, we detected a modest induction of IFNβ and IFNα5 mRNA in response to LTβR ligation in BMDC

(open versus closed bars, Figure 2.12A-B, p<0.001 and p<0.01 for IFNβ and IFNα5 respectively between WT and LTβR-/- BMDC). IFNβ and IFNα5 induction was specific to the LT pathway as

LTβR-/- BMDC failed to induce IFNα/β under these conditions. Furthermore, IFNα/β was not significantly induced following stimulation with α-CD40 (Figure 2.12A-B), even though α-CD40

(but not α-LTβR) Abs readily provoked IL-12 secretion from BMDC (Figure 2.10).

Since the amount of IFNα/β expression was relatively modest in response to LTα/β ligand or agonist Abs directed at LTβR we evaluated whether signals through TLR4 and LTβR could collaborate to increase Type I IFN expression. Indeed, BMDC pre-treated with LPS showed evidence of further up-regulation of IFNαβ expression when subsequently treated with anti-

LTβR agonist Ab, and this effect was not observed for LTβR-/- BMDC (Figure 2.12C,D). Thus, we postulate that Pathogen/Danger-associated molecular patterns (PAMPs/DAMPs) pre- condition DC to receive signals through the LTβR which further augment IFNα/β expression.

2.3.5 LTβR-/- DC fail to support OTI proliferation in vitro but proliferation can be rescued by exogenous IFNα

To determine if reduced Type I IFN production contributes to the failure of LTβR-/- DC to induce full CD8+ T cell activation and expansion in vivo, we established an in vitro for measuring OTI T cell proliferation. Given the flexible nature of such a system, we were able to carefully titrate the amount of LPS and the ratio of DC to T cells such that OTI T cell proliferation was rendered help-dependent, and indeed, in the absence of OTII CD4+ T cells, we

Figure 2.12 LTR signaling can mediate Type I IFN production independent of TLR activation in BMDC

BMDC were stimulated with anti-LTR, LTor anti-CD40 in the absence of any added LPS. Expression of IFN (A) and IFN5 (B) was measured by real time RT-PCR at 6 and 18 hours post-stimulation of WT (open bars) and LTR-/- (grey bars) BMDC. These experiments were performed at least 3 times with similar results. *p<0.05, **p<0.01, ***p<0.001. BMDC were also pre-incubated with LPS for 2 hours, washed and then stimulated with anti-LTR. Expression of IFN (C) and IFN5 (D) was measured by real time RT-PCR at 4, 5 and 6 hours post-stimulation of WT (open bars) and LTR-/- (grey bars) BMDC. This experiment was performed 3 times with similar results. P<0.05.

observed minimal OTI proliferation which was equivalent to what was observed in the absence of any OVA Ag (Figure 2.13A, left 2 panels). Although this in vitro system is an artificial system, we observed a similar defect in OTI proliferation when OVA-BM1 was introduced into the cultures (Figure 2.14). However, since the OTI proliferation observed with BM1-OVA stimulation in vitro was less robust and somewhat delayed compared to LPS plus OVA (Figure

2.14), we focused on the LPS-version of the in vitro assay. Interestingly, using the LPS/OVA system, in the presence of OTII CD4+ T cells we noted more than a two-fold reduction in OTI

CFSE dilution when OTI were cultured with LTβR-/- versus WT DC, confirming a requirement for DC-intrinsic LTβR signaling for optimal OTI expansion in vitro (Figure 2.13A versus

2.13B), and recapitulating the nature and magnitude of our OTI expansion defect that was observed in vivo (Figure 2.1). Similar defects in CFSE dilution were observed with LTβR-/-

BMDC co-cultured with BM1-OVA in vitro (Figure 2.14), and also when WT OTII helper T cells were compared with LTβ-/- OTII CD4+ T cells (Figure 2.6). These data demonstrate that

LTαβ-LTβR DC licensing signals are required for stimulating OTI expansion in vitro.

To determine whether Type I IFNs could rescue the proliferation of OTI T cells primed by

LTβR-/- DC, we added IFNα into our WT or LTβR-/- DC-OTII co-cultures and measured OTI proliferation by CFSE dilution. Exogenous IFNα restored proliferation of OTI T cells stimulated with LTβR-/- DC, but had a minimal effect on proliferation of OTI stimulated by WT DC (Figure

2.13A vs 2.13B), where maximal Type I IFN induction would have been achieved through the combination of LPS and LTβR ligation. Furthermore, although we noted that CD25 levels on

OTI T cells from LTβR-/- DC-OTII co-cultures were reduced to 65% of the normal CD25 levels observed in WT DC-OTII co-cultures, addition of IFNα restored CD25 levels on OTI T cells co-

Figure 2.13 LTR-/- BMDC do not support full OTI proliferation in vitro and this can be rescued with Type I IFNs

WT (A) and LTR-/- (B) BMDC were pre-incubated with LPS and OVA protein for 18 hours, washed, and then plated with CFSE-labeled OVA-specific CD8+ T cells (OTI) with or without OTII CD4+ T cells. 72 hours later, OTI T cells were gated based on CD45.1 expression and/or CD8 expression, then assessed for CD25 expression and CFSE dilution. In some cases 25unit/mL of IFN were supplemented to the cultures at 48 hours. The experiment is a representative example of 3 independent experiments.

Figure 2.14 BMDC/OTI co-cultures promotes OTI division which is compromised in the presence of LTR-/- BMDC

WT or LTR-/- BMDC were plated with BM1-OVA protein and CFSE-labeled OVA-specific CD8+ T cells (OTI) with or without OTII CD4+ T cells. 5 days later, OTI T cells were gated based on CD45.1 expression and/or CD8 expression, then assessed for CFSE dilution and entry into cell cycle using dye cycle violet. Note that the extent of proliferation is less than what is observed in Figure 2.13, and the proliferation is observed on day 5 rather than day 3. For these reasons, LPS-OVA was judged to be more amenable for in vitro BMDC evaluation. This experiment was performed two times with similar results.

64 cultured with LTβR-/- DC to levels equivalent to those observed in WT DC-OTII co-cultures

(94% of normal). Therefore, production of Type I IFNs is a critical mediator downstream of

LTβR signaling in DC, and recovery of Type I IFN levels can rescue the otherwise impaired proliferation of CD8+ T cells primed by LTβR-/- DC. These data provide a mechanism for the unique function of LTβR in DC-mediated T cell priming.

2.4 Discussion

T cell help is required in many cases for optimal cross-priming of CD8+ T cell responses to cell-associated Ag233. The nature of these help signals, however, remains incompletely characterized. Here we show that DC-intrinsic LTβR signaling is required for ex vivo DC function and for optimal cross-priming of a CD8+ T cell response in vivo. Our study identifies a non-redundant function for LTβR signaling alongside CD40 stimulation in promoting DC maturation and identifies LTβR-dependent Type I IFN production as a unique contribution from this TNF family member in shaping an effective CD8+ T cell response.

We have previously shown that LTαβ is rapidly up-regulated on OVA-specific CD4+ helper T cells in response to immunization with Ag, and the kinetics of LTαβ expression is very similar to CD69 expression235. Such kinetics are very similar to the induction of CD40L on CD4+ helper T cells, and represent a novel form of “help” for DC conditioning in vivo via interaction with LTβR on DC. Although our studies indicate that LTαβ on cells other than Ag-specific CD4+

T cells (B cells for example) is insufficient to trigger LTβR on DC to prime CD8+ T cells, it is possible that LIGHT expression on CD4+ helper T cells may also contribute to LTβR signaling in DC to facilitate CD8+ T cell responses to cell-associated Ag, and indeed the defect we observe with LTβR-/- DC is typically larger than what is observed when we transfer LTβ-/- OTII (Figure

2.6). Nevertheless, the expression of LTαβ on CD4+ helper T cells is likely strictly regulated since we observe very little expression on resting CD4+ helper T cells235, and indeed, the expression of LTαβ on Th1 and Th17 cells has been exploited for depletion strategies in the context of rodent models of autoimmunity253.

One possible explanation for our findings is that DC-intrinsic LTβR signaling is required for their homeostatic maintenance in the spleen242–244. However, we have confirmed that the reduction in OTI expansion is due to impaired DC function and not reduced DC numbers in three ways. First, purification and culture of equivalent numbers of WT and LTβR-/- DC results in a profound reduction in OTI proliferation ex vivo. Second, in WT mice whose splenic DC populations are intact, the absence of LTβ expression on a small population of adoptively transferred helper CD4+ T cells recapitulates the defect in OTI clonal expansion observed in

LTβR-/- settings. Finally, treatment of WT mice with a LTβR signaling inhibitor, LTβR-Ig, one day prior to immunization again results in a two-fold reduction in the peak OTI expansion without any impact on DC numbers. Therefore in scenarios in which DC numbers remain equivalent but the LTαβ-LTβR signal is absent, peak OTI expansion is significantly and comparably reduced in vivo, and this phenotype was confirmed using LTβR-/- BMDC in vitro.

Thus, the LTβR licensing requirement for DC stimulatory function is independent of its role in

DC homeostasis.

The discrepant capacity of CD40 and LTβR licensing signals to induce IFNα β may explain, at least in part, the inability of physiological CD40 signals to compensate for the absence of DC-intrinsic LTβR signaling. The exception to this scenario is in cases where anti-

CD40 agonist Abs are added in vivo, which we have previously shown can compensate for the absence of LTαβ on helper OTII T cells235. However, anti-CD40 Abs have been shown to

66 provoke robust and sustained up-regulation of LTαβ on B cells (up to 11 days) which could conceivably have overcome the absence of LTαβ on helper T cells under those circumstances254.

In any case, our finding that CD40 signaling was dispensable for CD8+ T cell clonal expansion contradicts other reports that suggest the sufficiency of CD40 ligation for full DC licensing150,255 and for early CD8+ T cell expansion256–259. Indeed, the evidence for DC-derived CD40 signals supporting CD8+ T cell expansion are mixed260,261. The variable requirement for CD40 in CD8+

T cell cross-priming could be explained by the recent identification of CD40L expression on DC, which may drive CD8+ T cell priming in more help-independent systems262. Furthermore, CD70 has been shown to play a more predominant role in CD40-independent CD8+ T cell responses263.

Given that the defects we observe in CD8+ T cell clonal expansion are significant, but not absolute (ie, residual CD8+ T cell proliferation is observed), a model of complex interplay of multiple TNF family receptors, rather than CD40 alone, in mediating DC licensing seems likely.

In addition, the relative importance of each of these molecules will likely depend on the stimulation conditions.

We found that IFNα/β mRNA was synthesized in response to LTβR stimulation even in the absence of TLR co-stimulation (Figure 2.12). Interestingly, IFNα/β is a potent inducer of co- stimulatory molecule expression on DC, including CD86264,265. Consistent with a role for IFNα/β in regulating CD86 expression, we observed a transient decrease in CD86 expression on DC from LTβR-/- mice. Moreover, this defect was rescued by the presence of WT DC, indicating that a soluble mediator such as IFNα/β may stimulate the expression of CD86 on LTβR-/- DC in trans

(Figure 2.8). IFNα/β is also induced by TLR ligation, and additionally the IFNα/β receptor

(IFNAR) has been shown to be critical for MyD88-independent DC maturation in response to

Salmonella infection265. Thus, it is likely that collaboration between PAMPs/DAMPs and LTβR

67 signaling is required for optimal IFNα/β production, and indeed we found that treatment of LPS stimulated DC with anti-LTβR augmented Type I IFN expression in vitro.

IFNAR expression on CD8+ T cells is critically required for CD8+ T cell priming and expansion250 and prolongs expression of genes involved in T cell programming by modulating chromatin accessibility149. As in the LT-deficient scenarios, Ag-specific IFNAR-/- CD8+ T cells fail to expand following priming, and overzealous cell cycling followed by defective persistence of IFNAR-/- CD8+ T cells has been reported139, suggesting that normal proliferation and subsequent loss of CD8+ T cells in the absence of LTβR signaling may be the result of a suboptimal Type I IFN response. Consistent with our finding that LT-derived signals are required for optimal CD25 expression, IFNα can also induce a dramatic up-regulation of CD25 on CD8+ T cells in vivo249. Since add-back of IFNα to LTβR-/- BMDC cultures rescued poor

CD8+ T cell proliferation in vitro, such collaboration between innate signals and LTβR signals may be required for optimal CD8+ T cell proliferation in vivo. Future experiments where Type I

IFN levels can be carefully titrated in vivo will shed light on how the LT-derived Type I IFNs might modify CD8+ T cell responses to cell-associated Ag such as auto-Ag or tumour-Ag.

Although the role for LTβR signaling in CD8+ responses during infectious disease has been mixed232,238,266, this could reflect varying levels of help-dependency in different systems where there may have been robust Type I IFN production elicited by PAMPs/DAMPs. Our study focuses on the role for DC-intrinsic LTβR signaling in provoking a CD8+ T cell response to cell- associated Ag, and this may have functional significance to help-dependent situations such as tumour eradication and graft rejections, scenarios where LTβR signaling has been implicated239–

241,267. The evaluation of LTβR signaling in DC during auto-immunity remains to be fully elucidated and would provide therapeutic insight into the potential value of LTβR inhibition for

68 treatment of such chronic diseases.

2.5 Materials and methods

In vivo help-dependent CD8+ T cell responses

OTI (1x106) T cells were purified (as above) and adoptively transferred into C57BL/6 mice treated with LTβR-Ig or control Ig or alternatively into WT -> WT and LTβR-/- -> WT BM chimeric mice. In some cases, 3x106 either WT OTII or LTβ-/-OTII were transferred into mice.

On the following day, these mice were primed with 25 x 106 of OVA-loaded BM1 splenocytes.

OVA-loaded splenocytes were prepared by osmotic shock. Briefly, 20 x 107 BM1 splenocytes were resuspended in 1ml of hypertonic solution (0.5M sucrose, 10% polyethyleneglycol 1000, and 10 mM Hepes in RPMI 1640) containing 10 mg/ml OVA protein for 10 min at 37˚C. 14 ml of pre-warmed hypotonic solution (40% H2O, 60% PRMI 1640) was added, and the cells incubated for an additional 2 min at 37˚C. The cells were spun immediately after the incubation, washed five times with PBS, and injected into mice. The OTI response was evaluated at day 3, 7,

14, 21 in the blood and/or spleen.

In vitro help-dependent CD8+ T cell responses

Day 10 BMDC cultures were treated with LPS (100ng/mL) or LPS with OVA (50µg/mL) overnight. OTI and OTII T cells were purified as described in the online supplemental methods, and OTI T cells were CFSE labeled (1µM). Activated BMDC were co-cultured with OTII and

OTI T cells at a ratio of 1:10:10. Cultures were treated with or without IFNα (25U/mL) in rescue experiments. Similar results were obtained when OVA-loaded BM1 splenocytes were substituted for LPS/OVA.

In vitro BMDC stimulation

Stimulations were performed in 96-well plates with cell density at 1 X 106 cells/mL in complete RPMI 1640 medium (Sigma Aldrich). Plates were coated for 1h at 21°C with anti-

Armenian Hamster antibody (BD Bioscience) (4g/mL) in 0.05M NaHCO3 pH 9.6, and agonistic anti-mLTβR antibody AFH6 (20g/mL), recombinant LTαβ (100ng/ml) or anti- mCD40 antibody FGK4/5 (10g/mL) was added overnight in PBS at 4°C. Supernatant was removed prior to adding DC. In some cases, DC were pre-treated with 100ng/ml LPS for 2 hours, washed and then transferred to plates coated with agonistic anti-mLTβR Ab AFH6. In co- culture experiments, OVA-specific OTII CD4+ T cells were purified using negative bead isolation as above and then rested overnight. T cells were then incubated with BMDCs at a ratio of 2:1 and the mixed cultures were stimulated with either LPS alone (100ng/ml) or LPS with

OVA peptide 323-339 (5µg/ml) for the indicated time points. In some cases, LPS/OVA was substituted with 50,000 BM1 cells hypotonically loaded with OVA. BMDC were then isolated from T cells by CD11c-based positive selection as described in the online supplemental methods.

Isolated DC fractions were homogenized in Trizol for cDNA preparation and qPCR analysis.

Mice:

WT C57BL/6 mice were purchased from Charles River. C57BL/6 OTII mice were purchased from Charles River and were bred in-house. LTβ-/- mice purchased from B&K

Universal were crossed with OTII mice. BM1 mice were purchased from The Jackson

Laboratory and were bred in-house. LTβR-/- mice were a kind gift from Dr. Rodney Newberry

(Washington University School of Medicine). All animals were housed in specific pathogen-free

70 conditions and all experiments were performed according to the University of Toronto approved animal use protocols.

Bone Marrow Chimeras:

Bone marrow was collected from femurs and tibia of WT C57BL/6 (CD45.1) mice, LTβR-

/- CD45.1 mice, or CD11c-DTR/GFP CD45.2 mice, filtered through a 70 m nylon mesh (BD

Falcon, Mississauga, ON, Canada), and red blood cells were lysed using RBC lysing buffer

(Sigma-Aldrich). For single chimeras, 2-4 x 106 bone marrow cells were then injected iv into

C57BL/6 CD45.2 mice that had been lethally irradiated (2 x 550 rads). Recipient mice were left for 8-10 weeks to reconstitute, and were given water supplemented with 2 mg/mL neomycin sulfate (Bio-Shop, Burlington, ON, Canada) for the first two weeks. In some cases, bone marrow cells from WT CD45.2 and LTβR-/- CD45.1 donors was mixed 1:1 prior to transfer into LTβR+/-

CD45.1/2 irradiated recipients, or bone marrow cells from CD11c-DTR and WT CD45.1 or

LTβR-/- CD45.1 was mixed 1:1 prior to transfer into C57BL/6 CD45.2 irradiated mice.

DC-depletion of CD11c-DTR chimeras:

Diphtheria toxin (DT) was obtained from Sigma and was reconstituted as per the manufacturer's instructions and stored in PBS at 1mg/mL at -20C until use. 6 hours prior to immunization, and again 24 hours later, chimeric mice were given DT at 4ng/g intraperitoneally, diluted to 0.5µg/mL in sterile PBS. Depletion of CD11c+ (EGFP+) cells was measured in the blood, spleen and LN at 24h-post-DT treatment by measuring specific depletion of EGFP+ DC by flow (data not shown).

Cell preparation for ex vivo assays and flow cytometry:

Unless otherwise noted, draining axillary, brachial and inguinal LN were collected at indicated time-points. Briefly, LN were suspended in Hank’s Buffered Salt Solution (Invitrogen),

10 mM Hepes, 150 mM NaCl, 5 mM KCl, 1 mM MgCl2, and 1.8 mM CaCl2 supplemented with

1 mM collagenase D (Roche, Mississauga, ON, Canada) and 60 g/ml DNase I (Sigma-Aldrich).

LN were mashed with glass slides, and the suspensions were incubated at 37°C with 5% CO2 for

30 min. At this time, tissues were disrupted by pipetting up and down, and suspensions were further incubated at 37°C with 5% CO2 for 10 min. EDTA was added to a final concentration of

1 mM, and the cell suspension was incubated at room temperature for 10 min. Cells were flushed through a 70m filter (BD Falcon), spun, and resuspended in PBS (Sigma)/1 mM EDTA/2%

FBS (Gibco, Carsbad, CA, USA). DCs were enriched using CD11c+ positive selection kit (Stem

Cell Technologies) and post-purification DC purity was measured by flow cytometry. Cells were resuspended in 10% media (RPMI, 10% FBS (Gibco),1% β-mercaptoethanol, 1% L-Glutamine,

1% penicillin-streptomycin), irradiated at 2000 rads and plated in triplicate at indicated number of DC per well in a 96 well flat bottom plate (VWR, Mississauga, ON, Canada), along with responder OTII or OTI cells. DC-depleted fraction was plated where indicated in two-fold serial dilutions starting at 2 x 105 cells per well. For responder T cells, CD8+ or CD4+ T cells were enriched from WT OTI or OTII mice, respectively, and were purified as above. Post-purification, cells were stained for purity with antibodies against Vβ5.1, Vα2, and CD4 or CD8 as appropriate, and OTI and OTII T cells were plated with DC at 5 x 105 cells per well in 10% serum (as above). DC–T-cell co-cultures were incubated at 37°C with 5% CO2 for 72 h. 100 l of supernatant was removed for IFNγ quantification, and was replaced by 100 l of fresh media with [3H]Thymidine ([3H]Td) at 0.1 Ci/ml (GE Healthcare, Mississauga, ON, Canada); cultures were incubated for an additional 18 h and [3H]Td incorporation was measured.

Flow cytometry Antibodies against CD4, CD8, CD11c, CD25, CD45.1, CD45.2, CD80,

CD86, CD40, MHCII, CD11b, IFNγ, Thy1.1 (FITC, PE-Cy5), Thy1.2 as well as streptavidin- fluorochrome conjugates were purchased from eBiosciences. Antibodies against CD69, Thy1.1

(PerCP-Cy5.5), Vα2 and Vβ5.1 were otained from BD Biosciences. Antibody against PDCA-1 was purchased from Miltenyi. All surface stains were performed in PBS/2% FBS (Gibco)/0.05% sodium azide (EM Science, Merck, Darmstadt, Germany). Intracellular staining was performed using BD Cytofix/Cytoperm kit. DyeCycle violet was obtained from Invitrogen, and was used according to manufacturer's instructions. All stained samples were acquired on the FACSCalibur,

Canto or LSRII (BD Bioscience) as appropriate.

For the generation of BMDC, femurs and tibia were removed from C57BL/6 (LTβR+/+ and

LTβR-/-) mice and bone marrow cells were harvested by flushing with PBS. The BMDC cultures were prepared at a density of 2 X 106 cells/mL in RPMI 1640 medium (Sigma-Aldrich) supplemented with 10% FBS (Gibco inactivated for 60min at 56oC), 0.05mM 2- mercaptoethanol, 100 U/mL penicillin, 100 g/mL streptomycin, 40 U/mL murine recombinant

GM-CSF (Peprotech). Fresh medium was added every 3 days until day 6, and every 2 days thereafter. Non-adherent cells were harvested on day 10 for analysis.

Primer sequences:

Oligonucleotide primers for qPCR analysis were as follow: for IFN-α4 (Forward) 5' -GCC

ATC CTT GTG CTA AGA G- 3'; (Reverse) 5' -TCA AGA GGA GGT TCC TGC ATC AC- 3'; for IFN-α5 (Forward) 5'-ACA GGT CGG GGT GCA GGA ATC T- 3'; (Reverse) 5' -CAC TCC

TCC TTG CTC AAT CTT- 3'; for IFN-β (Forward) 5' -TGC GTT CCT GCT GTG CTT CT- 3';

(Reverse) 5' -TTG GAT GGC AAA GGC AGT GT- 3'; for HMBS (Forward) 5’-TCC AAG

AGC CCA GCT A– 3’; (Reverse) 5’ –ATT AAG CTG CCG TGC AAC A- 3’. PCR reaction was activated by heating samples at 95oC for 10min, and cycled at 95oC for 15s, 61oC for 60s for

35 cycles.

CHAPTER 3

A Lymphotoxin/Type I IFN axis programs CD8+ T cells to infiltrate a self-tissue and propagate immunopathology.

Dennis Ng1, Derek Cumming4, Albert Lin3, Lesley A. Ward1, Ramtin Rahbar3, Derek Clouthier1, Tania Watts1, Karen Mossman4, Pamela S. Ohashi3, Jennifer L. Gommerman1,2

Derek Cumming – provided IRF3-/- bone marrow Albert Lin and Ramtin Rahbar – provided technical support on the RIP-GP model Derek Clouthier – provided transgenic P14 and SMART mice

3.1 Abstract

Type I IFNs are cytokines that can mediate both immune suppression and activation, and they have been shown to exert diverse functions in shaping the CD8+ T cell response. Here we report that DC-derived Type I IFNs are required to program the proliferative and pro-migratory potential of CD8+ T cells, ultimately dictating their ability to infiltrate target tissues. To achieve optimal levels of DC-derived Type I IFNs, signals through the Lymphotoxin-β Receptor (LTβR) via a TRAF3-dependent mechanism are required, and this LTβR/Type I IFN axis is essential for

CD8+ T cell programming and disease induction. Collectively, our data demonstrate that input from both the LTβR and pattern recognition receptors (PRR) regulate Type I IFN production in

DC in order to program CD8+ T cells to access and exert immuopathology in a self-tissue. A further understanding of the Type I IFN/LTβR axis will provide valuable therapeutic insights for treatment of CD8+ T cell mediated autoimmune diseases.

3.2 Introduction

Type I IFNs are pleiotropic cytokines comprised of a single IFN-β, numerous IFN-α's, and several IFN variants, including IFN -ε, -κ, - ω, and –δ, that facilitate various immune responses. Despite this diversity in IFN members, all Type I IFNs bind exclusively to the

Interferon-α receptor (IFNAR) expressed ubiquitously on all cell types205. One of the confounding issues concerning Type I IFNs is the multifaceted effects of this cytokine that have obvious discrepant biological outcomes such as cell proliferation versus cytotoxicity and immune activation versus immune suppression51,268. Various studies have demonstrated the importance of Type I IFNs in the generation of cytotoxic T lymphocytes147,212,269, while others have shown the suppressive effects of Type I IFNs on CD8+ T cell mediated immune responses227,230,270,271. How Type I IFNs balance both proliferative and cytotoxic effects remains

76 unclear, though recent evidence suggest that the expression of MicroRNA mir155 in antigen specific CD8+ T cells can down-regulate STAT1 that protects the cells from the anti-proliferative effect of Type I IFNs145, understanding the impact of this cytokine on T cell responses is important for rationalizing immunomodulatory therapies.

DC survey the body and rely on pattern recognition receptors (PRRs) such as toll-like receptors (TLRs) for the detection of invading microbes. As efficient antigen-presenting cells

(APCs), DCs activate the adaptive immune response by providing several signals for optimizing

T cell activation. These include: 1) Antigen-specific peptide-MHC complex, 2) Co-stimulatory molecules such as CD80 and CD86, and, 3) cytokine production such as IL-12 and Type I IFNs.

In the case of bacterial or viral infection, the presence of inflammatory cytokines triggered by

DC-intrinsic PRR activation often generates highly immunogenic DC that can directly prime

CD8+ T cell responses. However, during immune responses that elicit lower levels of inflammation, DCs often require help from CD4+ T cells in order to boost DC activation status and their capacity to secret cytokines233. Previous studies have shown that activated CD4+ T cells up-regulate several TNF superfamily members (TNFSF) including CD40-L (CD154),

RANK-ligand and LTαβ during an immune response. These TNFSF ligands signal the cognate

TNFRs expressed on DC as a form of "help" in order to potentiate the T cell priming capacity of

DC for the generation of CTLs in the context of help-dependent immune responses150,152,235,272.

DC are major producers of Type I IFNs, and PRR activation has been well characterized as the primary means of triggering Type I IFN production from DC. Signaling through TLR-3, -

4, -7, and -9 as well as the RIG-I-like receptors can induce rapid and robust Type I IFN production in response to infection51. Interestingly, recent studies have hinted that the TNF receptor super family (TNFRSF) members, TNFR-1, TNFR-2, RANK and Lymphotoxin-β

77 receptor (LTβR) are also capable of inducing Type I IFN production; however in contrast to PRR activation, TNFRSF members typically incite a very modest, albeit sustained level of Type I IFN production273–277. Previously we showed that expression of LTαβ on Ag-specific CD4+ T cells provide a "help" signal through the LTβR on DC, and this LTβR-dependent T cell:DC cross-talk was required to optimize CD8+ T cell expansion in vitro and in vivo274. While the induction of

Type I IFNs via PRR signaling has been well-characterized278, the relevance and mechanism of

TNFRSF facilitated Type I IFNs, specifically in the context of CD8+ T cell activation and chronic inflammation, is not well understood.

Here we have examined the involvement of DC-intrinsic LTβR signaling and Type I IFN production in the context of a CD8+ T cell-mediated disease model. In this study, we found that

LTβR-deficient DC were inefficient at promoting antigen specific CD8+ T cell expansion in response to a neo-self-antigen expressed in the pancreas, resulting in significantly lower expression of adhesion molecules on antigen-specific CD8+ T cells and restricted access to the target organ. We determined that LTβR signaling requires TRAF3 for the induction of Type I

IFNs. Add-back of IFNα in the presence of LTβR-deficient DC restored CD8+ T cell activation and infiltration into the target tissue, resulting in the destruction of islet β-cell mass and glucose intolerance. Taken together, our study sheds new light on the importance of a LTβR/Type I IFNs axis in shaping the early CD8+ T cell response to a non-replicating self-antigen, and provides a mechanism for how TNFRSF cooperate with TLR to promote Type I IFN production by conventional DC.

3.3 Results

3.3.1 DC-intrinsic LTβR signaling is required to induce pancreatic inflammation and glucose intolerance

To study the effect of DC-intrinsic LTβR signaling and the priming of CD8+ T cells in the context of an autoimmune disease, we compared the immunogenic potential of WT and LTβR-/-

Bone Marrow-Derived DC (BMDC) in provoking an autoimmune response to a neo-self-antigen.

Specifically, we used RIP-GP mice that express the lymphocytic choriomeningitis virus glycoprotein (LCMV-GP) under the control of the rat insulin inducible promoter such that

LCMV-GP is expressed by pancreatic islet β cells. Normally the peripheral LCMV-GP reactive

T cells are ignorant to the pancreatic islet β cell expression of LCMV-GP antigen279. This tolerance can be breached through the adoptive transfer of GP-peptide loaded BMDC that have been activated with a TLR agonist, thereby triggering the activation of GP-specific CD8+ T cells in the endogenous repertoire, leading to their subsequent infiltration into the pancreas and loss of islet cell mass280,281.

We determined whether this disease model requires CD4+ T cell help to elicit antigen specific CD8+ T cell responses. This would imply that CD4+ T cell:DC cross-talk is important for instigating autoimmunity in this setting. To determine whether diabetes induction by peptide- loaded BMDC is dependent on CD4+ T cell help in RIP-GP mice, we monitored diabetes incidence induced by BMDC that were loaded with or without the MHC-II restricted GP peptide.

Indeed, BMDC loaded with only MHC-I restricted peptides did not develop diabetes, suggesting that CD4+ T cell help is critical for the proper activation of GP-specific CD8+ T cells in this model (Figure 3.1).

Figure 3.1 Diabetes induction in RIP-GP mice by peptide loaded BMDC requires LCMV-GP specific CD4+ T cell activation.

RIP-GP mice were given an adoptive transfer of activated WT BMDC that were loaded with or without the Class II restricted LCMV-GP 61 peptide. The blood glucose level was monitored for up to 15 days to assess the diabetes status. The data depicted are representative from 2 independent experiments.

Since we had previously shown that the expression of LTαβ on CD4+ T was involved in T cell:DC cross-talk235, we next asked whether LTβR signaling in DC was required for diabetes induction in RIP-GP mice. To test this, we generated BMDC from WT or LTβR-/- donor mice and adoptively transferred the GP-peptide loaded BMDC into RIP-GP mice, comparing the immunogenic potential of WT or LTβR-/- DC in priming GP-specific CD8+ T cells for their capacity to mediate immune pathology in the pancreas. RIP-GP mice that received WT BMDC exhibited a marked increase in blood glucose levels by day 10 after DC transfer, and the mice remained diabetic for up to 20 days. In contrast, less than 10% of the RIP-GP mice that received

LTβR-/- BMDC developed diabetes (Figure 3.2A). Histological and immunofluorescence examination of pancreata from RIP-GP mice that had received WT BMDC exhibited significant immune cell infiltration (primarily CD8+ T cells) into the islets, whereas pancreata from RIP-GP mice that received LTβR-/- BMDC displayed low level of islet cell death and significantly less immune cell infiltration (Figure 3.2B and C). Therefore, the induction of pancreatic CD8+ T cell infiltration and diabetes by the transfer of GP-peptide pulsed BMDC requires CD4+ T cell help and BMDC-intrinsic LTβR signaling in the RIP-GP mouse model.

3.3.2 Robust CD8+ T cell proliferation and up-regulation of VLA4 and LFA-1 correlate with the induction of diabetes in RIP-GP mice

In order to assess whether CD8+ T cell expansion in response to an endogenous auto-antigen is likewise dependent on DC-intrinsic LTβR signaling, we examined the frequency of GP-specific

CD8+ T cells from the endogenous repertoire. Interestingly, similar to what was previously observed in the context of foreign protein antigen274, we noted approximately a 50% reduction in the frequency and numbers of CD8+ T cells specific for GP- peptide in all of the

Figure 3.2 DC-intrinsic LTβR signaling is required for the induction of diabetes

(A) Diabetes incidence in RIP-GP mice that were adoptively transferred with GP-peptide loaded WT or LTβR-/- BMDC. Data show blood glucose levels post BMDC transfer and disease incidence. Data shown is representative of 10 independent experiments totalling 62 mice per group. (B) Pancreata isolated from RIP-GP mice 6 days post transfer of WT or LTβR-/- GP- peptide loaded BMDC were stained with hematoxylin and eosin to assess infiltration of islets. (C) Immunofluorescent staining was performed on pancreata from the experiment in (B) using fluorescent conjugated antibodies specific for CD8, CD4 and CD3. Data shown for (B) and (C) is representative of 3 or more independent experiments.

83 compartments examined (Figure 3.3A and Figure 3.4A), and this was further confirmed upon transfer of GP33-specific P14 transgenic T cells (Figure 3.4B). However, despite impaired CD8+

T cell expansion in the mice that received LTβR-/- GP-peptide loaded BMDC, we found that the

Ag-specific CD8+ T cells could still produce equivalent levels of the cytokines IFN-γ and TNF-α

(Figure 3.3C), which is consistent with our previous findings274.

Even though DC-intrinsic LTβR signaling contributes to reduced antigen specific CD8+ T cell numbers in the periphery, the dramatic paucity of CD8+ T cells in the pancreata of mice that received GP-peptide loaded LTβR-/- BMDC (Figure 3.3C) is likely not fully explained by this reduced expansion alone, and in fact the normal production of IFNγ and TNFα by LTβR-/- primed CD8+ T cells implies that other CD8+ T cell defects may manifest as a consequence of improper priming. In order for effector T cells to successfully migrate into target organs, they must up-regulate the expression of adhesion molecules including the very-late antigen 4 (VLA-4) and lymphocyte function-associated antigen-1 (LFA-1). Indeed, all of the infiltrating CD8+ T cells found in the pancreas of RIP-GP mice express high levels of both VLA-4 and LFA-1, and they are actively proliferating as evidenced by the expression of Ki67 (Figure 3.5A). We therefore examined the expression of VLA-4 and LFA-1 on GP-specific CD8+ T cells in the periphery to determine if there was a defect in the up-regulation of these markers on CD8+ T cells in mice that did not exhibit pancreatic CD8+ T cell infiltration nor developed diabetes.

Indeed, we found that the expression of both VLA-4 and LFA-1 on GP-specific CD8+ T cells was significantly reduced in the mice that received LTβR-/- BMDC, particularly on Day 7 immediately preceding what would normally be the onset of diabetes for mice receiving WT

BMDC (Figure 3.5B). Hence, a reduction in VLA4 and LFA1 expression on peripheral CD8+ T

Figure 3.3 DC-intrinsic LTβR signaling is required for optimal expansion of antigen-specific CD8+ T cells in the periphery and their accumulation in the pancreas

RIP-GP mice were given an adoptive transfer of activated WT or LTβR-/- BMDC loaded with GP-peptides to induce diabetes. (A) At 7 and 10 days post transfer, the frequency of GP-specific CD8+ T cells was measured in the blood, pancreatic LN and pancreas *p < 0.05. (B) The frequency of total CD8+ T cells found in the pancreas was determined by FACS *p < 0.05. (C) CD8+ T cells in the blood harvested on Day 7 were re-stimulated with GP-peptides to assess cytokine production. Data shown is representative of 3 or more independent experiments.

Figure 3.4 Expansion of P14 CD8+ T cells and the absolute number of GP-specific CD8+ T cell of RIP-GP mice that received WT or LTβR-/- GP-peptide loaded BMDC

(A) RIP-GP mice received an adoptive transfer of 5000 P14 transgenic CD8+ T cells, and 1 day later, WT or LTβR-/- GP-peptide loaded activated BMDC were administered. (B) Related to Figure 2, the absolute number of GP-specific CD8+ T cell in the blood and pancreatic LN was calculated for the experiments depicted in Figure 2. *p < 0.05

Figure 3.5 Pancreas infiltrating GP-specific CD8+ T cells express high level of adhesion molecules VLA-4 and LFA-1

(A) The pancreata of RIP-GP mice that had previously received GP-peptide loaded WT BMDC were analysed by FACS. Expression of VLA4, LFA1 and Ki67 was assessed on pancreas CD8+ T cells (empty histogram) relative to the respective FMO controls (shaded histogram). (B) The expression of VLA-4, LFA-1 and Ki67 on peripheral GP-specific CD8+ T cells was compared in mice that received either WT or LTβR GP-peptide loaded BMDC on Day 7 and Day 10 post transfer *p < 0.05. Data is representative of 3 or more independent experiments.

89 cells primed in the absence of LTβR signaling on BMDC may explain the dramatic paucity of

CD8+ T cells in the pancreas.

3.3.3 Production of Type I IFNs by BMDC is required for optimal CD8+ T cell activation/expansion and diabetes induction in RIP-GP mice.

Since LTβR signaling can facilitate Type I IFN expression in DC274, and Type I IFNs are particularly important for the programming of cytotoxic T lymphocytes148,212,217,269, we queried whether adoptive transfer of GP-peptide loaded BMDC that are incapable of producing Type I

IFNs would have the same effect as the absence of BMDC-intrinsic LTβR signaling in the context of the RIP-GP diabetes mouse model. Accordingly, we transferred BMDC generated from IRF3 deficient mice into the RIP-GP host to assess their capacity to induce disease. IRF3 is a critical transcription factor that mediates Type I IFN expression downstream of TLR-3, TLR-4 and the RIG-I-like receptor in response to infection. IRF3-/- DC cannot produce Type I IFNs in response to LPS282. When we transferred IRF3-/- BMDC into RIP-GP mice, they also fail to develop diabetes (Figure 3.6A), concomitant with a reduction in the frequency and number of

GP-specific CD8+ T cells in the periphery, and a significant reduction in total infiltrating CD8+ T cells into the pancreas (Figure 3.6B and C). Moreover, GP-specific CD8+ T cells primed by

IRF3-/- BMDC in vivo expressed significantly lower levels of VLA-4 and LFA-1 (Figure 3.6B), reminiscent of the LTβR deficient setting (Figure 3.5B). Thus, administration IRF3-/- BMDC or

LTβR-/- BMDC to RIP-GP mice results in similar outcomes with respect to CD8+ T cell phenotype and diabetes susceptibility.

Given the apparent phenocopy between BMDC derived from IRF3-/- and LTβR-/- mice, we next ascertained whether stimulation of the LTβR in BMDC can induce the phosphorylation and nuclear translocation of IRF3. To test this, we generated BMDC from WT or LTβR-/- mice

Figure 3.6 DC-intrinsic expression of IRF3 is required to prime CD8+ T cells and induce diabetes in RIP-GP mice.

(A) Blood glucose levels and diabetes incidence of RIP-GP mice that have received an adoptive transfer of activated WT or IRF3-/- BMDC were monitored. (B) The frequency of GP-specific CD8+ T cells and their expression of VLA-4, LFA-1 and Ki67 was quantified in the periphery of RIP-GP mice on Day 7 and Day 10 post-transfer of WT versus IRF3-/- GP-peptide loaded BMDC *p < 0.05. (C) Pancreata from RIP-GP mice receiving WT or IRF3-/- GP-peptide loaded BMDC were compared using H&E staining to assess islet damage and by FACS to determine the frequency of CD8+ T cells *p < 0.05. These data are pooled from 2 independent sets of experiments.

92 and stimulated the cells in vitro with an LTβR agonistic antibody. A hallmark of LTβR activation is degradation of TRAF3 and activation of the NFκB2 pathway, both of which have been demonstrated in a number of cell lines and murine embryonic fibroblast systems155,283,284. In the BMDC system we also found that treatment with the anti-LTβR antibody led to TRAF3 degradation and p100 processing into the p52 active NFκB2 subunit within 8hr of stimulation

(Figure 3.7), suggesting that LTβR signaling also activates the alternative NFκB pathway in DC.

To understand how LTβR signaling contributes to Type I IFN expression, we asked whether any of the signaling events leading to Type I IFN production downstream of PRRs could be observed in the context of LTβR activation in BMDC. In our previous study, we showed that

LTβR stimulation triggers a gradual induction of IFNβ and IFNα by quantitative PCR274. When we stimulated WT BMDC with LTβR agonistic antibody, we found that IRF3 was transiently phosphorylated 1hr post LTβR stimulation and also again at 18hr (Figure 3.8A and B). In addition, we could detect IRF3 in the nucleus 1hr post LTβR stimulation (Figure 3.8C). To further confirm that Type I IFNs are produced, we measured STAT1 phosphorylation in BMDC after overnight stimulation with anti-LTβR agonistic antibody, and we found that only WT

BMDC but not LTβR-/- BMDC exhibited STAT1 phosphorylation. However no impairment in

STAT1 phosphorylation upon LPS treatment of LTβR-/- BMDC was observed, suggesting that

LTβR-/- BMDC do not have a global defect in their ability to produce Type I IFNs (Figure

3.8D).

3.3.4 LTβR-dependent Type I IFN production requires TRAF3

All TNFR members signal through multiple TNF receptor associated factors (TRAFs) that further recruit and activate other signaling molecules for various responses. Interestingly, both

Figure 3.7 BMDC stimulation with agonistic αLTβR

(A) WT or LTβR-/- BMDC were stimulated with αLTβR for the specified time points. The protein levels of p100, p52 and TRAF3 were analyzed by western blotting and compared with the level of actin as a loading control. The data depicted are representative from 2 independent experiments.

Figure 3.8 TRAF3 is required for LTβR-dependent IFN-I production

(A) BMDC generated from WT or LTβR-/- mice were stimulated in culture with an agonistic antibody against LTβR. Cells were harvested at various time points post stimulation and analysed by western blot to assess IRF3 phosphorylation with actin as a comparative loading control. (B) BMDC generated from WT or IRF3-/- mice were stimulated with LPS or agonistic α-LTβR antibody for 20h to assess IRF3 phosphorylation. (C) WT BMDC were transfected with scrambled negative siRNA or TRAF3 specific siRNA, and cells were harvested 24hr post transfection and examined for TRAF3 levels by western blotting. siRNA transfected cells were also stimulated with LPS or αLTβR at the specified time points and harvested for quantitative PCR to assess IFNβ and IFNα5 expression *p < 0.05.

TRAF-3 and TRAF-6 have recently been discovered to play important roles in regulating the induction of Type I IFNs downstream of TLR signaling56,285. Since, LTβR signals through

TRAF-2, -3 and -5 upon receptor activation, we next queried whether TRAF3 is responsible for the expression of Type I IFNs upon LTβR activation. To test this, we performed a knock-down of TRAF3 with siRNA in BMDC and found that the siRNA partially eliminated TRAF3 expression. When we examined the TRAF3 deficient BMDC, the expression of both IFNβ and

IFNα5 was impaired in response to LPS and completely abolished in response to LTβR stimulation. Furthermore, TRAF3 knockdown in BMDC also suppressed LTβR-dependent

STAT1 phosphorylation (Figure 3.8E). Therefore, LTβR signaling depends on TRAF3 for Type

I IFN production and subsequent STAT1 phosphorylation in BMDC, and this appears to be a feature that is shared by both TLR4 and LTβR signaling pathways.

3.3.5 Exogenous administration of IFNα restores CD8+ T cell activation and immunopathology in RIP-GP mice that received LTβR-/- BMDC.

Given the strong phenocopy observed between IRF3-/- versus LTβR-/- BMDC in the context of the RIP-GP model, we hypothesized that the production of Type I IFNs facilitated by

DC-intrinsic LTβR signaling was important for optimizing CD8+ T cell proliferation and their potential to invade and destroy pancreas tissue. To test this hypothesis, we added recombinant

IFNα to our established in vitro and in vivo systems. For the in vitro experiments, we isolated

GP61-specific CD4+ T cells from transgenic SMARTA mice and GP33-specific CD8+ T cells from P14 mice, and co-cultured these T cells with LPS-activated WT or LTβR-/- BMDC (Figure

3.9A). When we titrated the amount of GP33 peptide in the culture, we were able to generate a help-dependent system whereby CD8+ P14 T cells fail to expand in the absence of CD4+

SMARTA T cells. Using this in vitro system, we found that GP-specific CD8+ T cells co-

Figure 3.9 DC-intrinsic LTβR-dependent IFN-I production is required for optimal CD8+ T cell expansion and expression adhesion molecules VLA4 and LFA1

(A) Graphical representation of the experiment: BMDC generated from WT or LTβR-/- mice were activated with LPS and loaded with LCMV peptides. GP61 specific CD4+ T cells derived from SMARTA mice were co-cultured with GP33-specific CD8+ T cells derived from P14 mice. P14 CD8+ T cells were further labelled with CFSE, and co-cultured for 3 days with activated BMDC, SMARTA CD4+ T cells and in some cases recombinant IFNα. P14 CD8+ T cells were subsequently analysed by FACS to assess proliferation and expression of adhesion molecules. Representative flow cytometry data are shown. (B) Quantification of VLA4 and LFA1 expression on CD8+ T cells in the various conditions depicted in (A) *p < 0.05. The data shown are representative of 4 independent experiments.

99 cultured with peptide-loaded LTβR-/- BMDC expand poorly when compared to GP-specific

CD8+ T cells co-cultured with peptide-loaded WT BMDC. Furthermore, CD8+ T cells co- cultured with peptide-loaded LTβR-/- BMDC also expressed significantly lower levels of VLA-4 and LFA-1 (Figure 3.9B), reminiscent of our in vivo observation (Figure 3.5B). We found that in the presence of added recombinant IFNα, the expression of VLA4 and LFA1 on P14 CD8+ T cells primed by LTβR-/- LPS-activated BMDC was restored (Figure 3.9A and B). Likewise, in a similar experimental set-up, add-back of recombinant IFNα restored the expression of VLA4 on

OVA-specific CD8+ T cells to WT levels, confirming that the CD8+ T cell defects we observed in the presence of LTβR-/- DC were not peculiar to the GP-system (Figure 3.10). Collectively, these data show that DC-intrinsic LTβR signaling shapes early events in CD8+ T cell priming in vitro, optimizing CD8+ T cell clonal expansion and up-regulation of adhesion molecules in an

Type I IFNs dependent manner.

Next, to test if Type I IFNs facilitated by DC-intrinsic LTβR signaling was relevant for inducing CD8+ T cell responses targeted at the pancreas, RIP-GP mice that were injected with

LTβR-/- BMDC were subsequently treated with exogenous IFNα. As shown in Figure 1, GP- peptide loaded LTβR-/- BMDCs could not induce diabetes in the RIP-GP mice. However, when

RIP-GP mice were also provided with recombinant IFNα, more than 80% of the RIP-GP mice that received GP-peptide loaded LTβR-/- BMDCs became diabetic by day 10 (Figure 3.11A). Of the RIP-GP mice that became diabetic, GP-specific CD8+ T cells were found to expand normally and up-regulated VLA-4 and LFA-1 to comparable levels as that found on GP-specific CD8+ T cells derived from RIP-GP mice administered GP-pulsed WT BMDC (Figure 3.11B). Moreover,

CD8+ T cell infiltration into the pancreas was restored in mice that received peptide pulsed

LTβR-/- BMDCs along with exogenously added IFNα (Figure 3.11C). Collectively, our data

100

Figure 3.10 DC-intrinsic LTβR signaling is required for optimal expression of VLA4 and LFA1 on OVA-specific CD8+ T cells through a Type-I IFN dependent mechanism

(A) BMDC generated from WT or LTβR-/- mice were activated with LPS and incubated with OVA overnight. CD4+ OVA-specific T cells (OTII) and CFSE labelled CD8+ OVA-specific (OTI) T cells and co-cultures withactivated BMDC and OVA protein for 3 days and analysed by FACS to assess proliferation and expression of adhesion molecules. In some cases IFNα was added. Representative FACS plots are shown. (B) Co-cultures were set up with OTI T cells and WT or LTβR-/-BMDC with or without OT-II CD4+ T cells and with or without the addition of IFNα. Differing concentrations of OVA was added *p < 0.05. The data were tabulated and are representative of 2 independent experiments.

101

Figure 3.11 Exogenous administration of IFN-I restores CD8+ T cell expansion and immunopathology in RIP-GP mice that received LTβR-/- BMDC

(A) GP-peptide loaded WT or LTβR-/- BMDC were administered to RIP-GP mice and on Day 3 post transfer, the mice that received LTβR-/-BMDC were further administered recombinant IFNα intravenously. Glucose levels and disease incidence up to 15 days post-BMDC administration are depicted. (B) The frequency of GP-specific CD8+ T cell in the blood, and their expression of VLA-4 and LFA-1 were compared by FACS at Day 10 post-BMDC administration *p < 0.05 Student's t-test. (C) Pancreata extracted at Day 10 post-immunization were subjected to immunofluorescence to examine CD8+ T cell infiltration. Data are representative of 2 independent sets of experiments.

102 indicate that Type I IFN production by DCs is necessary and sufficient for CD8+ T cell expansion, expression of VLA-4/LFA1 on CD8+ T cells and diabetes induction in the absence of

DC-intrinsic LTβR signaling. These results highlight an important role for the LTβR/Type I IFN axis in promoting auto-aggressive CD8+ T cell responses.

103

3.4 Discussion

Type I IFNs can act on different immune cell types with varying effects and functions.

They play an important role in antiviral responses during infection and can promote immune cell recruitment and activation. Type I IFNs are also widely used as a therapeutic treatment for autoimmune diseases, ostensibly dampening over-zealous immune responses286, and Type I IFNs have been shown to have both anti-proliferative and pro-apoptotic functions in various systems229,287. Whether Type I IFNs are pro- or anti-inflammatory, the relevant cells that produce

Type I IFNs and how the effects of Type I IFNs are regulated in vivo remain largely unclear. One possible molecular explanation for differential effects of Type I IFNs may be that activated antigen specific CD8+ T cells express microRNA mir155 that down-regulates STAT1 which protects the cells from the anti-proliferative effect of Type I IFNs145. Our study provides new insights into how TNFSF and PRR cooperate in provoking Type I IFN production, and in the context of the RIP-GP model, we showed that integration of DC-intrinsic LTβR-derived signals with PRR activation is critical for optimizing Type I IFN production to provoke a CD8+ T cell response to a self-antigen.

Immune responses that do not elicit strong inflammatory cytokines or viruses that do not trigger robust Type I IFN expression, such as vaccinia virus, require CD4+ T cell help for optimal generation of CTLs137. Our studies show that DC-intrinsic LTβR signaling represents an important help signal that regulates the induction of Type I IFNs. The lack of this help signal, or alternatively the absence of a DC-derived Type I IFN response (as in IRF3-/- DC) results in poor antigen specific CD8+ T cell expansion and impaired up-regulation of VLA-4 and LFA-1. LFA-1 is a pro-migratory receptor that is expressed on all leukocytes as well as activated T cells, and its expression allows immune cells to bind intercellular adhesion molecule (ICAM) family members

104 that are largely expressed on endothelial and target tissue cells288. As such, expression of LFA-1 allows immune cells to migrate from the vasculature into the tissue parenchyma. In addition to migration, LFA-1 expression is also important for optimizing T cell - DC interactions 289, where suboptimal LFA-1 - ICAM-1 binding can result in poor T cell clonal expansion290. Furthermore, the induced expression of LFA-1 can be an important consequence of CD8+ T cell asymmetrical division, where the daughter CD8+ T cell population that expresses higher levels of LFA-1 produces better cell contact with DCs, leading to its preferential up-regulation of CD25 and sustained proliferation291. Interestingly, this work also demonstrated the importance of complete

CD8+ T cell expansion in order to become fully competent to infiltrate target tissues.

Specifically, the adhesion molecule VLA-4 was found to be highly expressed on T cells that had undergone several divisions, as was also observed in our in vitro system (Figure 3.9A and

3.10A). Like LFA-1, VLA-4 is an adhesion molecule that mediates immune cell migration into inflamed tissues. Indeed, in the spontaneous non-obese diabetes (NOD) mouse model, treatment with anti-VLA4 neutralizing antibody effectively suppressed T cell infiltration into the pancreas and completely prevented diabetes development292. The same NOD mouse model is sensitive to

LTβR inhibition, suggesting a putative contributing mechanism of action for LT inhibitors in the context of NOD diabetes may be via prevention of the full maturation of the CD8+ T cell response by interfering with DC activity196,293. Thus, early events such as the up-regulation of

LFA-1 and VLA-4 on CD8+ T cells, which appears to be coupled with proliferation in response to antigen, can contribute to their immunogenic and tissue infiltrating potential.

Early studies have shown that patients suffering from chronic lymphocytic leukemia exhibit increased levels of LFA-1 on neoplastic B cells when treated with recombinant IFN-

α2294,295. In contrast, leukocytes from Multiple Sclerosis patients who received prolonged

105 treatment of IFN-β exhibit a significant reduction of VLA-4 and LFA-1 expression296. Our experiments clearly demonstrate that addition of recombinant Type I IFNs can induce the expression of LFA-1 on CD8+ T cells both in vitro and in vivo, and this may reflect an immediate and transient effect of Type I IFNs on CD8+ T cells that is not observed during prolonged therapeutic IFN-β treatment. In addition, recent studies have shown that Type I IFNs are important in help-dependent responses to direct immune cell recruitment through the induction of chemokines such as CXCL9 and CXCL10, leading to CTL infiltration into the site of infection

297. While we did not measure chemokine levels in the target organ (pancreas), it is possible that a combination of poor expansion of Ag-specific CD8+ T cells, impaired up-regulation of pro- migratory molecules (LFA-1, VLA-4) as well as a reduction of target tissue-intrinsic chemokines, would conspire to thwart the development of diabetes.

In order to ascertain if the LTβR-dependent BMDC phenotype is a result of Type I IFNs deficiency, we adoptively transferred BMDC lacking the transcription factor IRF3 into RIP-GP mice, and found that a similar phenotype was observed. A caveat to this experiment is that the defect we observed with IRF3-/- BMDC may be caused by insufficient activation of the BMDC with LPS prior to transfer. However in terms of LTβR-/- BMDC, we had previously shown that they can produce normal levels of Type I IFNs and IL-12 in response to LPS stimulation, and they express comparable level of co-stimulatory molecules such as CD80/CD86 and MHC-II upon activation274. Hence, the defect observed in the LTβR deficient scenario is likely at the level of CD8+ T cell priming rather than a poor response to TLR4 stimulation.

PRR activation represents the major signaling pathways that triggers Type I IFN expression. TNF receptor associated factors (TRAF)-3 and 6 play a critical role in the induction of Type I IFNs downstream of TLR55,56,285,298. TRAFs are signaling adaptor molecules that also

106 mediate signals downstream of the TNFR superfamily. LTβR activation recruits TRAF-2, -3 and

-5 to facilitate numerous physiological responses including chemokine production by stromal cells, maintenance of high endothelial venule competency, homeostatic proliferation of some splenic DC and death in fibroblasts and tumor cells235,243,299–302. How LTβR signals facilitate

Type I IFN expression is currently unclear, but TRAF3 signaling is particularly of interest since it is a key signaling component that is shared between LTβR and the Type I IFNs inducing TLR pathways. Upon LTβR activation, TRAF3 is ubiquitinated at lysine (K) 48 for degradation that results in the release of the NFB inducing kinase (NIK) and NFB2 processing283,303. While

TLR-4 activation also triggers K48-ubiquitination of TRAF3 leading to TRAF3 degradation that is required for the MAPK signaling pathways, TLR4 ligation can also promote TRAF3 K-63 auto-ubiquitination, resulting in the recruitment of TBK1, IKKε and IRF3 which are critical signaling events that lead to Type I IFN induction56,285.

Given TRAF3 is a common signaling molecule that is shared by the two pathways, we queried whether TRAF3 can regulate LTβR-mediated Type I IFN induction. TRAF3-deficient mice are embryonic lethal, hence we knocked down TRAF3 expression with siRNA in BMDC and examined the impact on Type I IFN production in the context of LTβR signaling. Our experiments showed that the loss of TRAF3 in DCs led to complete inhibition of IFN-α/β expression upon LTβR ligation. Whether TRAF3 facilitates the recruitment of IRF3 for its phosphorylation upon LTβR activation remains to be determined. Given that the dynamics of

Type I IFN expression induced by LTβR signaling are different from that of TLR activation274, we suspect that the signaling modality downstream of LTβR may involve both unique and shared features with TLR4 to induce IFN-α/β gene expression. Interestingly, a recent study showed that

NIK can negatively regulate Type I IFN induction in conventional DC304. NIK itself is

107 constitutively targeted by TRAF3 for ubiquitin mediated degradation downstream of LTβR, and

NIK degradation is prevented upon the K48-ubiquitination of TRAF3. We hypothesize that since

TRAF3 degradation is gradual and incomplete following LTβR stimulation (Figure 3.7), a portion of cellular TRAF3 in LTβR stimulated BMDC may undergo K-63 auto-ubiquitination resulting in a relative "détente" between these two ubiquitinated states, thus explaining the comparatively modest Type I IFN induction triggered by LTβR activation.

Previous studies have implicated the importance of both direct and indirect LTβR- mediated Type I IFN induction in stromal cells and macrophages against cytomegalovirus and vesicular stomatitis virus231,275,305. Furthermore, the absence of LTβR signaling in mice results in a significant loss of Type I IFN transcriptional signatures in the spleen in response to Listeria monocytogenes306. A recent study showed that both Type I IFNs and LTβR signaling are important for the control of hepatitis B virus in human hepatocytes, however in this case LTβR ligation does not trigger Type I IFN expression within hepatocytes307. It is therefore possible that the facilitation of Type I IFN production in antigen presenting cells (DC) by LTβR signaling is unique to antigen presenting cells, and is particularly manifested in the context of CD8+ T cell priming.

In summary, our current study characterizes LTβR signaling as a crucial help signal that regulates Type I IFN expression in DCs, and demonstrates the importance of this signal for CD8+

T cell programming and generation of tissue infiltrating CTLs for the induction of diabetes. This implies that inhibitors of the LT pathway, through their effect on Type I IFN production, may have therapeutic potential in CD8+ T cell driven disease states.

108

3.5 Materials and methods

Mice

C57BL/6 wild-type mice were purchased from Charles River Laboratories. LTβ-receptor

(LTβR)-/- mice were a generous gift from Rodney Newberry (Washington University School of

Medicine, St. Louis, MO). RIP-GP, P14 and SMARTA Transgenic mice were kind gifts from

Pamela Ohashi. All animals were housed under specific pathogen-free conditions and all experiments were performed according to animal-use protocols approved by the University of

Toronto.

BMDC Culture

Femurs and tibia were dissected and BM cells were harvested by flushing with PBS. The BMDC cultures were prepared at a density of 2 X 106 cells/mL in RPMI 1640 medium (Sigma-Aldrich) supplemented with 10% FBS (Gibco inactivated for 60min at 56oC), 0.05mM 2- mercaptoethanol, 100 U/mL penicillin, 100 ug/mL streptomycin, 40 U/mL murine recombinant

Gm-CSF (Peprotech). Fresh medium was added every 3 days until day 6, and every 2 days thereafter. Non-adherent cells were harvested on day 10 for analysis.

RIP-GP diabetes model

For induction of diabetes in RIP-gp mice, we activated BMDC with 10ng/mL of lipopolysaccharide (Sigma) for 18h and pulsed with 1µg/mL of gp33-41 (KAVYNFATM), gp276-286 (SGVENPGGYCL) and gp61-80 (GLNGPDIYKGVYQFKSVEFD) for 10h. Pulsed

DCs were washed three times with PBS and transferred intravenously at 2 x 106 cells per mouse.

109

For IFN-α add-back experiments, mice were injected 10,000U of IFN-α (purchased from PBL

Assay Science Cat# 12100-1) intravenously on day 3 post BMDC transfer.

Quantitative real-time PCR Analysis

Expression of mRNAs was quantified on the Applied BiosystemsTM 7300 Real-Time PCR machine using MaximaTM SYBR Green Master Mix (Fermentas). Total cellular RNA was isolated using TRIzol Reagent (Invitrogen) according to manufacturer's protocol, and cDNA libraries were generated with SuperScript® III First-Strand Synthesis System (Invitrogen) using

1µg of total RNA and oligo (dT) primers. Oligonucleotide primers for qPCR analysis were as follows: for IFN-α4 (Forward) 5' -GCC ATC CTT GTG CTA AGA G- 3'; (Reverse) 5' -TCA

AGA GGA GGT TCC TGC ATC AC- 3'; for IFN-α5 (Forward) 5'-ACA GGT CGG GGT GCA

GGA ATC T- 3'; (Reverse) 5' -CAC TCC TCC TTG CTC AAT CTT- 3'; for IFN-β (Forward) 5'

-TGC GTT CCT GCT GTG CTT CT- 3'; (Reverse) 5' -TTG GAT GGC AAA GGC AGT GT- 3'; for (HMBS) (forward) 5′-TCC AAG AGC CCA GCT A-3′; (reverse) 5′-ATT AAG CTG CCG

TGC AAC A-. The PCR reaction was activated by heating samples at 95C for 10min, and cycled at 95oC for 15s, 61oC for 60 s for 35 cycles.

Flow Cytometry

Antibodies against CD4, CD8, CD11c, CD11a, CD49d, CD45.1, CD45.2, MHCII, CD11b, IFNγ,

TNFα, as well as streptavidin-fluorochrome conjugates were purchased from eBioscience. For analysis of GP-specific CD8+ T cells, monomers of H-2Kb: KAVYNFATM were purchased from the Baylor College of Medicine, and tetramers were made by conjugating with PE or APC

– streptavidin purchased from Molecular Probes, Life Technologies Inc.

110 siRNA

TRAF3 siRNA was composed of a 21-nucleotide long duplexes were synthesized by integrated

DNA Technologies. The sequence for TRAF3 knock down: Sense – 5’ CGA GGA GAA CUU

AUG AAA UCA UAA C – 3’ and Anti-sense – 3’ AU GCU CCU CCU CUU GAA UAC UUU

AGU AUU G – 5’

Western Blot Analysis

For western blot analysis, cells were harvested and lysed in ice-cold RIPA buffer purchased from

Cell SignalingTM. Cell lysates were incubated on ice for 10 min and centrifuged at 12,500 rpm for 15 min at 4 °C. Cell extracts were run through SDS-polyacrylamide gels followed by transfer onto nitrocellulose membranes. The membranes were blocked at room temperature for 1 h with

5% BSA and then incubated overnight at 4oC with primary antibodies. The membranes were washed and developed using an enhanced chemiluminescence detection system (Pierce).

Statistical analysis

Data were analyzed based on two-tailed Student’s t-test using GraphPad Prism software.

Responses were considered significant when a probability value p < 0.05 was obtained.

111

CHAPTER 4

Discussion

112

4.1 Summary

DC are capable of fine tuning immune responses through their expression of co- stimulatory or co-inhibitory molecules, and production of inflammatory or regulatory cytokines to mediate clearance of pathogens while avoiding excessive immune pathology. In order to maintain and regulate the homeostasis of our immune system, factors that directly modulate DC activation status can serve as potential therapeutic targets against numerous diseases.

In Chapter 2, I showed that LTαβ, a TNF family member, expressed on activated antigen specific CD4+ T cells, can stimulate DC to facilitate further DC maturation. Mice lacking LTβR signaling failed to efficiently prime antigen specific CD8+ T cells during the primary immune response. Previous studies have extensively characterized the role of LTβR signaling in the development and maintenance of lymphoid structures as well as DC homeostasis. The work in

Chapter 2 clearly demonstrated that intrinsic LTβR signaling in DC directly modulates their immunogenic potential, and through multiple approaches, we showed that the defect in antigen specific CD8+ T cell expansion observed in the absence of LTβR signaling is independent of lymphoid architecture, DC migration, antigen processing, antigen presentation or DC homeostasis. By using an in vitro BMDC stimulation assay with agonistic antibody or the active ligand LTαβ, I showed that LTβR stimulation can trigger IFNα5 and IFNβ expression. Other studies have previously hinted that LTβR signaling can mediate a Type I IFN response in different infection models 189,204,305,306, however our studies provided the first direct evidence of

LTβR engagement leading to Type I IFN expression. Type I IFNs are generally regarded as a pro-inflammatory cytokine secreted during bacterial or viral infections, and the signals that trigger Type I IFN responses are primarily the result of PRR activation. In recent years, a few studies have emerged describing members of the TNFRSF such as TNFR1, TNFR2, RANK and

113

LTβR contributing to Type I IFN induction. However, the signaling pathways involved and the physiological relevance of this TNFRSF-derived response have not been fully characterized.

In Chapter 3, I further examined the LTβR-Type I IFN axis in T cell mediated immunity using the RIP-GP mouse inducible autoimmune disease model. The RIP-GP model involves the adoptive transfer of BMDC to elicit endogenous T cell responses against pancreatic islet cells in expressing the LCMV-GP protein. This allowed us to circumvent any complications involving the role of LTβR signaling in maintaining lymphoid architecture or DC homeostasis normally presented in an LTαβ/LTβR deficient animal. In addition, I was able to study the functional aspect of LTβR signaling specifically in DC. Similar to what we have observed in the LTβR deficient chimeric mice or during LTβR-Ig treatment, I showed that adoptive transfer of BMDC lacking LTβR into RIP-GP mice led to poor GP-specific CD8+ T cell expansion and failed to induce disease. Moreover the expanded CD8+ T cells exhibit an overall lower expression of adhesion molecules, LFA-1 and VLA-4. LFA-1 expression has been shown to affect the quality of T cell - DC interaction, and VLA-4 is essential for T cell infiltration into effector sites. The lack of GP-specific CD8+ T cell infiltration is likely a result of poor expression of these adhesion molecules found on the antigen specific T cells. When I further examined whether Type I IFNs can compensate for the loss of LTβR signaling, I found that the provision of exogenous recombinant IFN-α was indeed able to restore the CD8+ T cell expansion to WT levels, as well as the expression of both LFA-1 and VLA-4 both in vitro and in vivo.

Despite ample evidence showing the capacity of Type I IFNs can directly program CD8+ T cell expansion and effector function148,149,169, all nucleated cells can respond to Type I IFNs due to the ubiquitous expression of its cognate receptor IFNAR. Therefore, we need to consider that the LTβR-induced Type I IFN response can also trigger positive immunogenic effects in DC and

114

CD4+ T cells, thereby contributing to CD8+ T cell priming and disease induction. Previous studies have shown that Type I IFNs can promote DC activation, induce the up-regulation of co- stimulatory molecules CD80 and CD86 and enhance peptide-MHC expression. In Chapter 2, we showed that LTβR deficient DC do exhibit lower expression of CD86 compared to WT DC.

Furthermore, this defect is lost in mixed chimeric mice containing both WT and LTβR-/- DC, suggesting a LTβR-dependent factor (perhaps Type I IFNs) derived from WT DC can rescue

CD80 and CD86 expression in the LTβR-/- DC in a paracrine manner.

Although the LT/Type I IFNs axis directly affects CD4+ T cell function was not tested in our studies, others have shown that Type I IFNs can synergize with IL-12 to promote TH-1 differentiation, TH-1 cells are potent producer of IL-2 which is essential for T cell proliferation, and may indirectly modulate the antigen specific CD8 T cell expansion. In addition, CD4+ T cells have been shown to express Type I IFNs-induced chemokines such as CXCL-9 and CXCL-

10 that direct CTL infiltration into effector sites during herpes simplex virus (HSV) infection297.

Therefore LTβR-mediated Type I IFNs may also trigger similar events that promote LCMV-GP specific CD8+ T cell infiltration into the pancreas in RIP-GP mice. Lastly, a recent study showed that Type I IFNs can directly inhibit regulatory T cell activation and proliferation to promote the generation of an optimal CTL response against LCMV infection308. Hence, LTβR sufficient DC may require an overall lower level of immunogenicity to prime CD8+ T cell in RIP-GP mice due to reduced regulatory T cells number and activity. Therefore, the cumulative effects of Type I

IFNs exerted on a variety of immune cell types may conspire together to induce disease pathology.

The data in Chapter 3 showed that LTβR signaling triggers IRF-3 activation and nuclear translocation, and I demonstrated that LTβR-mediated Type I IFNs requires TRAF3 signaling,

115 whereby knock-down of TRAF3 by siRNA significantly reduced IFNβ and IFNα5 gene expression. Interestingly, TRAF3 and IRF3 are signaling molecules that facilitate Type I IFN induction downstream of TLR3 and TLR4. TLR3/4 activation recruits the adaptor protein Trif which further recruits TRAF3 for its K63 ubiquitination, eventually leading to IRF-3 phosphorylation and transcription of IFN-genes. Downstream of TLR-4, TRAF3 signaling can mediate both Type I IFN expression through IRF-3 phosphorylation and production of inflammatory cytokines such as TNF-α and IL-12 through activation of MAPKs. Interestingly, the regulation between IRF-3 versus MAPK activation highly depends on the status of TRAF3 ubiquitination. TRAF3 is an E3-ubiquitin ligase, and TRIF binding leads to TRAF3 activation and promotes its self-ubiquitionation at the K63 position for the recruitment of TBK-1, IKK-ε and IRF-3 for Type I IFN induction. In contrast, MyD88 signaling recruits TRAF6, IKK-γ, cIAP1/2, Ubc13, TAK1 and MAPKs to TLR4 upon activation. Once the complex is assembled,

TRAF6 and cIAP1/2 can mediate the ubiquitination of TRAF3 at the K48 position for its degradation. Following TRAF3 degradation, the MyD88 signaling complex can be internalized into the cytosol for subsequent MAPKs activation. Downstream of LTβR, TRAF3 has already been shown to undergo TRAF2 and cIAP1/2 mediated K48-ubiquitination and degradation to facilitate activation of the alternate NFκB pathway. Whether TRAF3 has a bimodal ubiquitination status, and self-ubiquitination capacity to facilitate the recruitment of IRF-3 for its subsequent phosphorylation, remains unknown.

Given that TRAF3 and IRF-3 are common signaling components downstream of TLR and

LTβR, we suspect that there is cross-talk between the two pathways. Indeed, in Chapter 2 we found that under conditions where DC were initially activated with LPS and further stimulated with LTβR agonists or co-cultured with antigen specific CD4+ T cells to provide a source of

116

LTαβ, DC display a very robust Type I IFNs signature compared to LPS stimulation alone. How these two pathways are interacting and perhaps synergizing with each other to boost Type I IFN expression remains unclear. A possible scenario may involve TLR-4 activation resulting in a skew of TRAF3 cytosolic protein abundance. Since TRAF3 has been shown to determine the activity status of the alternate NFκB pathway309, where high levels of cytosolic TRAF3 strongly inhibits its activation, and NIK, the key components in the processing of NFκB2, has recently been identified as a major inhibitor of the Type I IFNs signaling pathway304, I hypothesize that

TLR-4 activation may have a stabilizing effect on TRAF3 levels, programming the DC for Type

I IFN induction upon LTβR engagement.

The discovery of TNF and the TNF receptor superfamily has led to major advances in therapeutic treatments against various inflammatory diseases. The LTαβ/LIGHT-LTβR pathways have been extensively studied, and inhibitors such as the soluble LTβR-immunoglobulin (LTβR-

Ig) fusion protein attenuate a number of chronic inflammatory and autoimmune disease models such as rheumatoid arthrithis, multiple sclerosis, and Type I diabetes, albeit without a clear mechanism of action defining the role of LTβR signaling in these scenarios. This thesis describes the Lymphotoxin pathway in T cell mediated immune responses. LTαβ displayed by activated antigen specific CD4+ T cells acts as a licensing factor that activates the LTβR expressed on DC to trigger further DC maturation and production of Type I IFNs. LTβR mediated Type I IFN induction is mediated through TRAF3 signaling, and it is required to prime CD8+ T cells for their optimal expansion and expression of adhesion molecules, LFA-1 and VLA-4, for their infiltration into effector site to induce disease pathology (Figure 4.1). Our studies unravel a novel role for LTβR in Type I IFN induction, which may potentially explain the efficacy provided by

LTβR

117

Figure 4.1 Lymphotoxin-dependent Type I IFNs (IFN-I)are required for early programming of CD8+ T cells resulting in their clonal expansion, up-regulation of LFA-1 and VLA-4 and their ability to cause immunopathology in a self tissue

1) Cross-talk between CD4+ T cells and DC results in LTβR signaling in DC leading to TRAF3- dependent phosphorylation and nuclear localization of IRF3. 2) These DC-intrinsic LTβR signals cooperate with TLR4 signals for Type I IFN production. 3) Type I IFNs induced by LTβR/TLR4 signaling in DC are required for CD8+ T cell clonal expansion. 4) Up-regulation of VLA4 and LFA1. 5) CD8+ T cells subsequently invade the pancreas where they exert immunopathology.

118 treatment. Thus our studies further our understanding of the LTβR pathway to allow for better design of therapeutic treatments for the aforementioned diseases.

4.2 Future directions

Our study demonstrates that the lymphotoxin pathway plays an important role in the priming of T cell responses against host protein antigens, which implicate clinical relevance in T cell mediated autoimmune diseases in human. Human DC express LTβR, but whether this signal is also required for the priming of CD8+ T cell is unclear. Furthermore, whether LTβR signaling can trigger Type I IFN expression in human DC is also unknown. Lastly, the mechanism of Type

I IFN induction by TNFR family members have not yet been fully characterized. Whether LTβR signaling synergizes or work independently with PRRs such as TLR-4 in the induction of Type I

IFNs can further our understanding of this unique LTβR/Type I IFN axis that can provide insights into therapeutic treatments against autoimmunity or tumor response.

Early studies showed that mice lacking LTα, LTβ or LTβR exhibit substantial loss of splenic

DC, and this defect was initially attributed to DC migration as a result of chemokine deficiences310,311. Contrary to this finding, others have reported that mice lacking chemokines required for T cells and DC homing to the SLO, CCL19 and CCL21, showed normal numbers of

DC despite disrupted splenic structure312. Further studies eventually revealed that LTβR signaling is required for the homeostatic maintenance of the CD11b+ CD8- subset of DC in the spleen. In recent years, Notch signaling has also been attributed to regulate DC homeostasis313.

Notch signaling is an evolutionary conserved pathway that plays essential roles in dictating cell fates. Within the context of the immune system, Notch signaling is important for lymphocyte development. The expression of Delta-like 4 on the thymic epithelium promotes T cell

119 development through Notch1, while Delta-like 1 expressed on the splenic marginal zone stroma and regulates marginal zone B cells through Notch2. In collaboration with Boris Reizis' group who first identified a role for Notch signaling in DC homeostasis, we examined how LTβR and

Notch signaling governs splenic DC homeostasis. The Reizis study (Appendix 1) showed that

Notch2 signaling is required for the terminal differentiation of CD11b+ CD8- CX3cr1- Esamhi

DC in the spleen and CD11b+ CD103+ DC within the lamina propria of the intestine. I generated chimeric mice that lack LTβR in the hematopoietic compartment and found that LTβR signaling regulates the same subset of splenic DC (Figure 4.2). Whether the two signaling pathways work together or act independently from each other to maintain DC homeostasis is unknown.In order to further identify whether there is a link between Notch2 and LTβR signaling, I also examined the intestinal CD11b+ CD103+ DC subset by generating mixed bone marrow chimeras using WT and LTβR-/- donor bone marrow in a competitive assay. Indeed, there was a consistently lower ratio of LTβR deficient CD11b+ CD103+ DC compared to the WT controls (Figure 4.3). In collaboration with Ken Murphy's group, we examined the CD11b+ CD103+ subset of DC in the intestine. The Murphy study revealed that this subset of Notch2/LTβR dependent DC are an essential source of IL-23, and this source of IL-23 is necessary to trigger the activation of

NKp46+ innate lymphoid cells to mount an efficient immune responses against attaching and effacing bacterial pathogens8. Whether Notch2 signaling or LTβR signaling directly or indirectly contributes to IL-23 production in DC, and how the two signaling pathways converge to regulate a very specific subset of DC in the spleen and in the intestine are unanswered questions.

120

Figure 4.2 Splenic DC subset of LTβR-/- chimeric mice

Splenic CD11b+Esamhi DC among gated wild-type or LTβR-/- bone marrow chimeras (mean ± SD of three animals). *P<0.05 Student’s t-test. Figure adapted from Lewis et al314.

121

Figure 4.3 CD103+ CD11b+ DC isolated from the intestinal laminar propria of mixed WT and LTβR-/-bone marrow chimera

Chimeric mice were generated with WT or LTβR-/- or WT+LTβR-/- bone marrow. Intestinal laminar propria preparation of cells were analyzed by flow cytometry pre-gated on CD11c+, MHCIIhi and CD11b+. Quantification of chimerism was calculated as follows: (percent contribution of mutant cells to intestinal cDCs / percent contribution of wild-type cells to intestinal cDCs) / (percent contribution mutant cells to splenic T cells / percent contribution of wild-type cells to splenic T cells). ***P<0.001 (Student's t-test) Figure adapted from Satpathy et al8.

122

REFERENCES

1. Ng, D. & Gommerman, J. L. The Regulation of Immune Responses by DC Derived Type I

IFN. Front. Immunol. 4, 94 (2013).

2. Steinman, R. M., Lustig, D. S. & Cohn, Z. A. Identification of a novel cell type in

peripheral lymphoid organs of mice. 3. Functional properties in vivo. J. Exp. Med. 139,

1431–45 (1974).

3. Steinman, R. M. & Cohn, Z. A. Identification of a novel cell type in peripheral lymphoid

organs of mice. II. Functional properties in vitro. J. Exp. Med. 139, 380–97 (1974).

4. Karsunky, H., Merad, M., Cozzio, A., Weissman, I. L. & Manz, M. G. Flt3 ligand

regulates dendritic cell development from Flt3+ lymphoid and myeloid-committed

progenitors to Flt3+ dendritic cells in vivo. J. Exp. Med. 198, 305–13 (2003).

5. Sathaliyawala, T. et al. Mammalian target of rapamycin controls dendritic cell

development downstream of Flt3 ligand signaling. Immunity 33, 597–606 (2010).

6. Miller, J. C. et al. Deciphering the transcriptional network of the dendritic cell lineage.

Nat. Immunol. 13, 888–99 (2012).

7. Satpathy, A. T. et al. Zbtb46 expression distinguishes classical dendritic cells and their

committed progenitors from other immune lineages. J. Exp. Med. 209, 1135–52 (2012).

8. Satpathy, A. T. et al. Notch2-dependent classical dendritic cells orchestrate intestinal

immunity to attaching-and-effacing bacterial pathogens. Nat. Immunol. 14, 937–48

(2013).

123

9. Dudziak, D. et al. Differential antigen processing by dendritic cell subsets in vivo. Science

315, 107–11 (2007).

10. Kushwah, R. & Hu, J. Complexity of dendritic cell subsets and their function in the host

immune system. Immunology 133, 409–19 (2011).

11. McKenna, H. J. et al. Mice lacking flt3 ligand have deficient hematopoiesis affecting

hematopoietic progenitor cells, dendritic cells, and natural killer cells. Blood 95, 3489–97

(2000).

12. Lewis, K. L. et al. Notch2 Receptor Signaling Controls Functional Differentiation of

Dendritic Cells in the Spleen and Intestine. Immunity 3, 1–12 (2011).

13. Allan, R. S. et al. Epidermal viral immunity induced by CD8alpha+ dendritic cells but not

by Langerhans cells. Science 301, 1925–8 (2003).

14. Belz, G. T. et al. Cutting edge: conventional CD8 alpha+ dendritic cells are generally

involved in priming CTL immunity to viruses. J. Immunol. 172, 1996–2000 (2004).

15. Maldonado-López, R., Maliszewski, C., Urbain, J. & Moser, M. Cytokines regulate the

capacity of CD8alpha(+) and CD8alpha(-) dendritic cells to prime Th1/Th2 cells in vivo.

J. Immunol. 167, 4345–50 (2001).

16. Schlecht, G., Leclerc, C. & Dadaglio, G. Induction of CTL and nonpolarized Th cell

responses by CD8alpha(+) and CD8alpha(-) dendritic cells. J. Immunol. 167, 4215–21

(2001).

124

17. McLellan, A. D. et al. Anatomic location and T-cell stimulatory functions of mouse

dendritic cell subsets defined by CD4 and CD8 expression. Blood 99, 2084–93 (2002).

18. Helft, J., Ginhoux, F., Bogunovic, M. & Merad, M. Origin and functional heterogeneity of

non-lymphoid tissue dendritic cells in mice. Immunol. Rev. 234, 55–75 (2010).

19. Ginhoux, F. & Merad, M. Ontogeny and homeostasis of Langerhans cells. Immunol. Cell

Biol. 88, 387–92 (2010).

20. Merad, M., Ginhoux, F. & Collin, M. Origin, homeostasis and function of Langerhans

cells and other langerin-expressing dendritic cells. Nat. Rev. Immunol. 8, 935–47 (2008).

21. Brewig, N. et al. Priming of CD8+ and CD4+ T cells in experimental leishmaniasis is

initiated by different dendritic cell subtypes. J. Immunol. 182, 774–83 (2009).

22. Bedoui, S. et al. Cross-presentation of viral and self antigens by skin-derived CD103+

dendritic cells. Nat. Immunol. 10, 488–95 (2009).

23. Henri, S. et al. CD207+ CD103+ dermal dendritic cells cross-present keratinocyte-derived

antigens irrespective of the presence of Langerhans cells. J. Exp. Med. 207, 189–206

(2010).

24. Stoitzner, P. et al. Langerhans cells cross-present antigen derived from skin. Proc. Natl.

Acad. Sci. U. S. A. 103, 7783–8 (2006).

25. Kaplan, D. H., Jenison, M. C., Saeland, S., Shlomchik, W. D. & Shlomchik, M. J.

Epidermal langerhans cell-deficient mice develop enhanced contact hypersensitivity.

Immunity 23, 611–20 (2005).

125

26. Imai, Y. et al. Freshly isolated Langerhans cells negatively regulate naïve T cell activation

in response to peptide antigen through cell-to-cell contact. J. Dermatol. Sci. 51, 19–29

(2008).

27. Beauchamp, N. M., Busick, R. Y. & Alexander-Miller, M. A. Functional divergence

among CD103+ dendritic cell subpopulations following pulmonary poxvirus infection. J.

Virol. 84, 10191–9 (2010).

28. Helft, J. et al. Cross-presenting CD103+ dendritic cells are protected from influenza virus

infection. J. Clin. Invest. 122, 4037–47 (2012).

29. Kim, T. S., Gorski, S. A., Hahn, S., Murphy, K. M. & Braciale, T. J. Distinct dendritic cell

subsets dictate the fate decision between effector and memory CD8(+) T cell

differentiation by a CD24-dependent mechanism. Immunity 40, 400–13 (2014).

30. Jang, M. H. et al. CCR7 is critically important for migration of dendritic cells in intestinal

lamina propria to mesenteric lymph nodes. J. Immunol. 176, 803–10 (2006).

31. Johansson-Lindbom, B. et al. Functional specialization of gut CD103+ dendritic cells in

the regulation of tissue-selective T cell homing. J. Exp. Med. 202, 1063–73 (2005).

32. Coombes, J. L. et al. A functionally specialized population of mucosal CD103+ DCs

induces Foxp3+ regulatory T cells via a TGF-beta and retinoic acid-dependent

mechanism. J. Exp. Med. 204, 1757–64 (2007).

33. Cerovic, V. et al. Intestinal CD103(-) dendritic cells migrate in lymph and prime effector

T cells. Mucosal Immunol. 6, 104–13 (2013).

126

34. Elkon, K. B. & Stone, V. V. Type I interferon and systemic lupus erythematosus. J.

Interferon Cytokine Res. 31, 803–12 (2011).

35. Bendriss-Vermare, N. et al. Human thymus contains IFN-alpha-producing CD11c(-),

myeloid CD11c(+), and mature interdigitating dendritic cells. J. Clin. Invest. 107, 835–44

(2001).

36. Res, P. C., Couwenberg, F., Vyth-Dreese, F. A. & Spits, H. Expression of pTalpha mRNA

in a committed dendritic cell precursor in the human thymus. Blood 94, 2647–57 (1999).

37. Naik, S. H., Corcoran, L. M. & Wu, L. Development of murine plasmacytoid dendritic

cell subsets. Immunol. Cell Biol. 83, 563–70 (2005).

38. Brawand, P. et al. Murine plasmacytoid pre-dendritic cells generated from Flt3 ligand-

supplemented bone marrow cultures are immature APCs. J. Immunol. 169, 6711–9 (2002).

39. Medina, K. L. et al. Separation of plasmacytoid dendritic cells from B-cell-biased

lymphoid progenitor (BLP) and Pre-pro B cells using PDCA-1. PLoS One 8, e78408

(2013).

40. Loschko, J. et al. Antigen targeting to plasmacytoid dendritic cells via Siglec-H inhibits

Th cell-dependent autoimmunity. J. Immunol. 187, 6346–56 (2011).

41. Galicia-Rosas, G. et al. A sphingosine-1-phosphate receptor 1-directed agonist reduces

central nervous system inflammation in a plasmacytoid dendritic cell-dependent manner.

J. Immunol. 189, 3700–6 (2012).

127

42. Joyee, A. G. & Yang, X. Plasmacytoid dendritic cells mediate the regulation of

inflammatory type T cell response for optimal immunity against respiratory Chlamydia

pneumoniae infection. PLoS One 8, e83463 (2013).

43. Satpathy, A. T. et al. Zbtb46 expression distinguishes classical dendritic cells and their

committed progenitors from other immune lineages. J. Exp. Med. 209, 1135–52 (2012).

44. Anderson, K. L. et al. Transcription factor PU.1 is necessary for development of thymic

and myeloid progenitor-derived dendritic cells. J. Immunol. 164, 1855–61 (2000).

45. Auffray, C. et al. Monitoring of blood vessels and tissues by a population of monocytes

with patrolling behavior. Science 317, 666–70 (2007).

46. Serbina, N. V, Salazar-Mather, T. P., Biron, C. A., Kuziel, W. A. & Pamer, E. G.

TNF/iNOS-producing dendritic cells mediate innate immune defense against bacterial

infection. Immunity 19, 59–70 (2003).

47. Casanova, J.-L., Abel, L. & Quintana-Murci, L. Human TLRs and IL-1Rs in host defense:

natural insights from evolutionary, epidemiological, and clinical genetics. Annu. Rev.

Immunol. 29, 447–91 (2011).

48. Reynolds, J. M. et al. Toll-like receptor 2 signaling in CD4(+) T lymphocytes promotes T

helper 17 responses and regulates the pathogenesis of autoimmune disease. Immunity 32,

692–702 (2010).

49. Akira, S. Innate immunity and adjuvants. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 366,

2748–55 (2011).

128

50. Meylan, E. & Tschopp, J. Toll-like receptors and RNA helicases: two parallel ways to

trigger antiviral responses. Mol. Cell 22, 561–9 (2006).

51. Stetson, D. B. & Medzhitov, R. Type I interferons in host defense. Immunity 25, 373–81

(2006).

52. Barton, G. M., Kagan, J. C. & Medzhitov, R. Intracellular localization of Toll-like

receptor 9 prevents recognition of self DNA but facilitates access to viral DNA. Nat.

Immunol. 7, 49–56 (2006).

53. O’Neill, L. A. J., Golenbock, D. & Bowie, A. G. The history of Toll-like receptors -

redefining innate immunity. Nat. Rev. Immunol. 13, 453–60 (2013).

54. O’Neill, L. A. J. & Bowie, A. G. The family of five: TIR-domain-containing adaptors in

Toll-like receptor signalling. Nat. Rev. Immunol. 7, 353–64 (2007).

55. Oganesyan, G., Saha, S., Guo, B. & He, J. Critical role of TRAF3 in the Toll-like

receptor-dependent and-independent antiviral response. Nature 439, 208–11 (2005).

56. Tseng, P.-H. et al. Different modes of ubiquitination of the adaptor TRAF3 selectively

activate the expression of type I interferons and proinflammatory cytokines. Nat.

Immunol. 11, 70–5 (2010).

57. Häcker, H., Tseng, P. & Karin, M. Expanding TRAF function: TRAF3 as a tri-faced

immune regulator. Nat. Rev. Immunol. 11, 457–468 (2011).

58. Takeda, K. & Akira, S. Toll-like receptors in innate immunity. Int. Immunol. 17, 1–14

(2005).

129

59. Asselin-Paturel, C. & Trinchieri, G. Production of type I interferons: plasmacytoid

dendritic cells and beyond. J. Exp. Med. 202, 461–5 (2005).

60. Honda, K. et al. IRF-7 is the master regulator of type-I interferon-dependent immune

responses. Nature 434, 772–7 (2005).

61. Ito, T. Interferon-alpha and Interleukin-12 Are Induced Differentially by Toll-like

Receptor 7 Ligands in Human Blood Dendritic Cell Subsets. J. Exp. Med. 195, 1507–1512

(2002).

62. Larangé, A., Antonios, D., Pallardy, M. & Kerdine-Römer, S. TLR7 and TLR8 agonists

trigger different signaling pathways for human dendritic cell maturation. J. Leukoc. Biol.

85, 673–83 (2009).

63. Yoneyama, M. & Fujita, T. RNA recognition and signal transduction by RIG-I-like

receptors. Immunol. Rev. 227, 54–65 (2009).

64. Seth, R. B., Sun, L., Ea, C.-K. & Chen, Z. J. Identification and characterization of MAVS,

a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3. Cell 122,

669–82 (2005).

65. Kawai, T. et al. IPS-1, an adaptor triggering RIG-I- and Mda5-mediated type I interferon

induction. Nat. Immunol. 6, 981–8 (2005).

66. Kato, H. et al. Cell type-specific involvement of RIG-I in antiviral response. Immunity 23,

19–28 (2005).

130

67. Rothfuchs, A. G. et al. Dectin-1 interaction with Mycobacterium tuberculosis leads to

enhanced IL-12p40 production by splenic dendritic cells. J. Immunol. 179, 3463–71

(2007).

68. Yamasaki, S. et al. Mincle is an ITAM-coupled activating receptor that senses damaged

cells. Nat. Immunol. 9, 1179–88 (2008).

69. Richard, M., Thibault, N., Veilleux, P., Gareau-Pagé, G. & Beaulieu, A. D. Granulocyte

macrophage-colony stimulating factor reduces the affinity of SHP-2 for the ITIM of

CLECSF6 in neutrophils: a new mechanism of action for SHP-2. Mol. Immunol. 43,

1716–21 (2006).

70. Rogers, N. C. et al. Syk-dependent cytokine induction by Dectin-1 reveals a novel pattern

recognition pathway for C type lectins. Immunity 22, 507–17 (2005).

71. 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).

72. Philpott, D. J., Sorbara, M. T., Robertson, S. J., Croitoru, K. & Girardin, S. E. NOD

proteins: regulators of inflammation in health and disease. Nat. Rev. Immunol. 14, 9–23

(2014).

73. Paradis, M. et al. A TNF-α-CCL20-CCR6 axis regulates Nod1-induced B cell responses.

J. Immunol. 192, 2787–99 (2014).

131

74. Martinon, F., Burns, K. & Tschopp, J. The inflammasome: a molecular platform

triggering activation of inflammatory caspases and processing of proIL-beta. Mol. Cell 10,

417–26 (2002).

75. Masumoto, J. et al. ASC is an activating adaptor for NF-kappa B and caspase-8-dependent

apoptosis. Biochem. Biophys. Res. Commun. 303, 69–73 (2003).

76. Fink, S. L. & Cookson, B. T. Pyroptosis and host cell death responses during Salmonella

infection. Cell. Microbiol. 9, 2562–70 (2007).

77. Sansonetti, P. J. et al. Caspase-1 activation of IL-1beta and IL-18 are essential for Shigella

flexneri-induced inflammation. Immunity 12, 581–90 (2000).

78. Tsuji, N. M. et al. Roles of caspase-1 in Listeria infection in mice. Int. Immunol. 16, 335–

43 (2004).

79. Fink, S. L., Bergsbaken, T. & Cookson, B. T. Anthrax lethal toxin and Salmonella elicit

the common cell death pathway of caspase-1-dependent pyroptosis via distinct

mechanisms. Proc. Natl. Acad. Sci. U. S. A. 105, 4312–7 (2008).

80. Yamagami, S. et al. CCR5 chemokine receptor mediates recruitment of MHC class II-

positive Langerhans cells in the mouse corneal epithelium. Invest. Ophthalmol. Vis. Sci.

46, 1201–7 (2005).

81. Van den Bossche, J., Malissen, B., Mantovani, A., De Baetselier, P. & Van Ginderachter,

J. A. Regulation and function of the E-cadherin/catenin complex in cells of the monocyte-

macrophage lineage and DCs. Blood 119, 1623–33 (2012).

132

82. Granucci, F. et al. Early events in dendritic cell maturation induced by LPS. Microbes

Infect. 1, 1079–84 (1999).

83. Förster, R. et al. CCR7 coordinates the primary immune response by establishing

functional microenvironments in secondary lymphoid organs. Cell 99, 23–33 (1999).

84. Kabashima, K. et al. CXCL12-CXCR4 engagement is required for migration of cutaneous

dendritic cells. Am. J. Pathol. 171, 1249–57 (2007).

85. Colvin, B. L., Morelli, A. E., Logar, A. J., Lau, A. H. & Thomson, A. W. Comparative

evaluation of CC chemokine-induced migration of murine CD8alpha+ and CD8alpha-

dendritic cells and their in vivo trafficking. J. Leukoc. Biol. 75, 275–85 (2004).

86. Castellino, F. et al. Chemokines enhance immunity by guiding naive CD8+ T cells to sites

of CD4+ T cell-dendritic cell interaction. Nature 440, 890–5 (2006).

87. Miller, J. et al. Two pathways of costimulation through CD28. Immunol. Res. 45, 159–72

(2009).

88. Boise, L. H. et al. CD28 costimulation can promote T cell survival by enhancing the

expression of Bcl-XL. Immunity 3, 87–98 (1995).

89. Appleman, L. J., van Puijenbroek, A. A. F. L., Shu, K. M., Nadler, L. M. & Boussiotis, V.

A. CD28 costimulation mediates down-regulation of p27kip1 and cell cycle progression

by activation of the PI3K/PKB signaling pathway in primary human T cells. J. Immunol.

168, 2729–36 (2002).

133

90. Grohmann, U. et al. CTLA-4-Ig regulates tryptophan catabolism in vivo. Nat. Immunol. 3,

1097–101 (2002).

91. Qureshi, O. S. et al. Trans-endocytosis of CD80 and CD86: a molecular basis for the cell-

extrinsic function of CTLA-4. Science 332, 600–603 (2011).

92. Botturi, K., Lacoeuille, Y., Cavaillès, A., Vervloet, D. & Magnan, A. Differences in

allergen-induced T cell activation between allergic asthma and rhinitis: Role of CD28,

ICOS and CTLA-4. Respir. Res. 12, 25 (2011).

93. Tesciuba, A. G. et al. ICOS costimulation expands Th2 immunity by augmenting

migration of lymphocytes to draining lymph nodes. J. Immunol. 181, 1019–24 (2008).

94. Schaefer, J. S., Montufar-Solis, D., Vigneswaran, N. & Klein, J. R. ICOS promotes IL-17

synthesis in colonic intraepithelial lymphocytes in IL-10-/- mice. J. Leukoc. Biol. 87, 301–

8 (2010).

95. Choi, Y. S. et al. ICOS receptor instructs T follicular helper cell versus effector cell

differentiation via induction of the transcriptional repressor Bcl6. Immunity 34, 932–46

(2011).

96. Zhang, X. et al. Structural and functional analysis of the costimulatory receptor

programmed death-1. Immunity 20, 337–47 (2004).

97. Yamazaki, T. et al. Expression of programmed death 1 ligands by murine T cells and

APC. J. Immunol. 169, 5538–45 (2002).

134

98. Rodig, N. et al. Endothelial expression of PD-L1 and PD-L2 down-regulates CD8+ T cell

activation and cytolysis. Eur. J. Immunol. 33, 3117–26 (2003).

99. Chemnitz, J. M., Parry, R. V, Nichols, K. E., June, C. H. & Riley, J. L. SHP-1 and SHP-2

associate with immunoreceptor tyrosine-based switch motif of programmed death 1 upon

primary human T cell stimulation, but only receptor ligation prevents T cell activation. J.

Immunol. 173, 945–54 (2004).

100. Carter, L. et al. PD-1:PD-L inhibitory pathway affects both CD4(+) and CD8(+) T cells

and is overcome by IL-2. Eur. J. Immunol. 32, 634–43 (2002).

101. Wang, S. et al. Molecular modeling and functional mapping of B7-H1 and B7-DC

uncouple costimulatory function from PD-1 interaction. J. Exp. Med. 197, 1083–91

(2003).

102. Butte, M. J., Keir, M. E., Phamduy, T. B., Sharpe, A. H. & Freeman, G. J. Programmed

death-1 ligand 1 interacts specifically with the B7-1 costimulatory molecule to inhibit T

cell responses. Immunity 27, 111–22 (2007).

103. Munitic, I., Kuka, M., Allam, A., Scoville, J. P. & Ashwell, J. D. CD70 deficiency impairs

effector CD8 T cell generation and viral clearance but is dispensable for the recall

response to lymphocytic choriomeningitis virus. J. Immunol. 190, 1169–79 (2013).

104. Penaloza-MacMaster, P. et al. Opposing effects of CD70 costimulation during acute and

chronic lymphocytic choriomeningitis virus infection of mice. J. Virol. 85, 6168–74

(2011).

135

105. Peperzak, V., Xiao, Y., Veraar, E. A. M. & Borst, J. CD27 sustains survival of CTLs in

virus-infected nonlymphoid tissue in mice by inducing autocrine IL-2 production. J. Clin.

Invest. 120, 168–78 (2010).

106. Hendriks, J., Xiao, Y. & Borst, J. CD27 promotes survival of activated T cells and

complements CD28 in generation and establishment of the effector T cell pool. J. Exp.

Med. 198, 1369–80 (2003).

107. Peperzak, V. et al. CD8+ T cells produce the chemokine CXCL10 in response to

CD27/CD70 costimulation to promote generation of the CD8+ effector T cell pool. J.

Immunol. 191, 3025–36 (2013).

108. Wang, C., Lin, G. H. Y., McPherson, A. J. & Watts, T. H. Immune regulation by 4-1BB

and 4-1BBL: complexities and challenges. Immunol. Rev. 229, 192–215 (2009).

109. Lin, G. H. Y. et al. Endogenous 4-1BB ligand plays a critical role in protection from

influenza-induced disease. J. Immunol. 182, 934–47 (2009).

110. Gramaglia, I., Weinberg, A. D., Lemon, M. & Croft, M. Ox-40 ligand: a potent

costimulatory molecule for sustaining primary CD4 T cell responses. J. Immunol. 161,

6510–7 (1998).

111. So, T., Song, J., Sugie, K., Altman, A. & Croft, M. Signals from OX40 regulate nuclear

factor of activated T cells c1 and T cell helper 2 lineage commitment. Proc. Natl. Acad.

Sci. U. S. A. 103, 3740–5 (2006).

136

112. Ito, T. et al. TSLP-activated dendritic cells induce an inflammatory T helper type 2 cell

response through OX40 ligand. J. Exp. Med. 202, 1213–23 (2005).

113. Nakae, S. et al. IL-17 production from activated T cells is required for the spontaneous

development of destructive arthritis in mice deficient in IL-1 receptor antagonist. Proc.

Natl. Acad. Sci. U. S. A. 100, 5986–90 (2003).

114. Rogers, P. R. & Croft, M. CD28, Ox-40, LFA-1, and CD4 modulation of Th1/Th2

differentiation is directly dependent on the dose of antigen. J. Immunol. 164, 2955–63

(2000).

115. Maxwell, J. R. et al. IL-18 bridges innate and adaptive immunity through IFN-gamma and

the CD134 pathway. J. Immunol. 177, 234–45 (2006).

116. Jung, S. et al. In vivo depletion of CD11c+ dendritic cells abrogates priming of CD8+ T

cells by exogenous cell-associated antigens. Immunity 17, 211–20 (2002).

117. Shedlock, D. J. et al. Role of CD4 T cell help and costimulation in CD8 T cell responses

during Listeria monocytogenes infection. J. Immunol. 170, 2053–63 (2003).

118. Van Lith, M. et al. Regulation of MHC class II antigen presentation by sorting of

recycling HLA-DM/DO and class II within the multivesicular body. J. Immunol. 167,

884–92 (2001).

119. Cresswell, P., Bangia, N., Dick, T. & Diedrich, G. The nature of the MHC class I peptide

loading complex. Immunol. Rev. 172, 21–8 (1999).

137

120. Bertholet, S. et al. Leishmania antigens are presented to CD8+ T cells by a transporter

associated with antigen processing-independent pathway in vitro and in vivo. J. Immunol.

177, 3525–33 (2006).

121. Shen, L., Sigal, L. J., Boes, M. & Rock, K. L. Important role of cathepsin S in generating

peptides for TAP-independent MHC class I crosspresentation in vivo. Immunity 21, 155–

65 (2004).

122. Van Kaer, L., Ashton-Rickardt, P. G., Ploegh, H. L. & Tonegawa, S. TAP1 mutant mice

are deficient in antigen presentation, surface class I molecules, and CD4-8+ T cells. Cell

71, 1205–14 (1992).

123. Chefalo, P. J., Grandea, A. G., Van Kaer, L. & Harding, C. V. Tapasin-/- and TAP1-/-

macrophages are deficient in vacuolar alternate class I MHC (MHC-I) processing due to

decreased MHC-I stability at phagolysosomal pH. J. Immunol. 170, 5825–33 (2003).

124. Henrickson, S. E. et al. T cell sensing of antigen dose governs interactive behavior with

dendritic cells and sets a threshold for T cell activation. Nat. Immunol. 9, 282–91 (2008).

125. Chang, J. T., Segal, B. M. & Shevach, E. M. Role of costimulation in the induction of the

IL-12/IL-12 receptor pathway and the development of autoimmunity. J. Immunol. 164,

100–6 (2000).

126. Imanguli, M. M. et al. Increased T-bet+ cytotoxic effectors and type I interferon-mediated

processes in chronic graft-versus-host disease of the oral mucosa. Blood 113, 3620–30

(2009).

138

127. Shi, M., Lin, T. H., Appell, K. C. & Berg, L. J. Janus-kinase-3-dependent signals induce

chromatin remodeling at the Ifng locus during T helper 1 cell differentiation. Immunity 28,

763–73 (2008).

128. Le Gros, G., Ben-Sasson, S. Z., Seder, R., Finkelman, F. D. & Paul, W. E. Generation of

interleukin 4 (IL-4)-producing cells in vivo and in vitro: IL-2 and IL-4 are required for in

vitro generation of IL-4-producing cells. J. Exp. Med. 172, 921–9 (1990).

129. Zhu, J., Cote-Sierra, J., Guo, L. & Paul, W. E. Stat5 activation plays a critical role in Th2

differentiation. Immunity 19, 739–48 (2003).

130. Rengarajan, J. et al. Interferon regulatory factor 4 (IRF4) interacts with NFATc2 to

modulate interleukin 4 gene expression. J. Exp. Med. 195, 1003–12 (2002).

131. Steinfelder, S. et al. The major component in schistosome eggs responsible for

conditioning dendritic cells for Th2 polarization is a T2 ribonuclease (omega-1). J. Exp.

Med. 206, 1681–90 (2009).

132. Everts, B. et al. Omega-1, a glycoprotein secreted by Schistosoma mansoni eggs, drives

Th2 responses. J. Exp. Med. 206, 1673–80 (2009).

133. Yamane, H., Zhu, J. & Paul, W. E. Independent roles for IL-2 and GATA-3 in stimulating

naive CD4+ T cells to generate a Th2-inducing cytokine environment. J. Exp. Med. 202,

793–804 (2005).

134. Bettelli, E., Korn, T., Oukka, M. & Kuchroo, V. K. Induction and effector functions of

T(H)17 cells. Nature 453, 1051–7 (2008).

139

135. Nurieva, R. I. et al. Generation of T follicular helper cells is mediated by interleukin-21

but independent of T helper 1, 2, or 17 cell lineages. Immunity 29, 138–49 (2008).

136. Zotos, D. et al. IL-21 regulates germinal center B cell differentiation and proliferation

through a B cell-intrinsic mechanism. J. Exp. Med. 207, 365–78 (2010).

137. Wiesel, M., Kratky, W. & Oxenius, A. Type I IFN substitutes for T cell help during viral

infections. J. Immunol. 186, 754–63 (2011).

138. Le Bon, A. et al. Cross-priming of CD8+ T cells stimulated by virus-induced type I

interferon. Nat. Immunol. 4, 1009–15 (2003).

139. Kolumam, G. A., Thomas, S., Thompson, L. J., Sprent, J. & Murali-Krishna, K. Type I

interferons act directly on CD8 T cells to allow clonal expansion and memory formation

in response to viral infection. J. Exp. Med. 202, 637–50 (2005).

140. Novy, P., Quigley, M., Huang, X. & Yang, Y. CD4 T cells are required for CD8 T cell

survival during both primary and memory recall responses. J. Immunol. 179, 8243–51

(2007).

141. Smith, C. M. et al. Cognate CD4(+) T cell licensing of dendritic cells in CD8(+) T cell

immunity. Nat. Immunol. 5, 1143–8 (2004).

142. Bennett, S. R., Carbone, F. R., Karamalis, F., Miller, J. F. & Heath, W. R. Induction of a

CD8+ cytotoxic T lymphocyte response by cross-priming requires cognate CD4+ T cell

help. J. Exp. Med. 186, 65–70 (1997).

140

143. Borrow, P. et al. CD40L-deficient mice show deficits in antiviral immunity and have an

impaired memory CD8+ CTL response. J. Exp. Med. 183, 2129–42 (1996).

144. Belz, G. T., Wodarz, D., Diaz, G., Nowak, M. A. & Doherty, P. C. Compromised

influenza virus-specific CD8(+)-T-cell memory in CD4(+)-T-cell-deficient mice. J. Virol.

76, 12388–93 (2002).

145. Gracias, D. T. et al. The microRNA miR-155 controls CD8(+) T cell responses by

regulating interferon signaling. Nat. Immunol. 14, 593–602 (2013).

146. Wiesel, M. et al. Th cells act via two synergistic pathways to promote antiviral CD8+ T

cell responses. J. Immunol. 185, 5188–97 (2010).

147. Agarwal, P. et al. Gene regulation and chromatin remodeling by IL-12 and type I IFN in

programming for CD8 T cell effector function and memory. J. Immunol. 183, 1695–704

(2009).

148. Curtsinger, J. J. M., Valenzuela, J. J. O., Agarwal, P., Lins, D. & Mescher, M. F. Type I

IFNs provide a third signal to CD8 T cells to stimulate clonal expansion and

differentiation. J. Immunol. 174, 4465–4469 (2005).

149. Agarwal, P. et al. Gene regulation and chromatin remodeling by IL-12 and type I IFN in

programming for CD8 T cell effector function and memory. J. Immunol. 183, 1695–704

(2009).

150. Bennett, S. R. et al. Help for cytotoxic-T-cell responses is mediated by CD40 signalling.

Nature 393, 478–80 (1998).

141

151. Ridge, J. P., Di Rosa, F. & Matzinger, P. A conditioned dendritic cell can be a temporal

bridge between a CD4+ T-helper and a T-killer cell. Nature 393, 474–8 (1998).

152. Schoenberger, S. P., Toes, R. E., van der Voort, E. I., Offringa, R. & Melief, C. J. T-cell

help for cytotoxic T lymphocytes is mediated by CD40-CD40L interactions. Nature 393,

480–3 (1998).

153. Pullen, S. S., Dang, T. T., Crute, J. J. & Kehry, M. R. CD40 signaling through tumor

necrosis factor receptor-associated factors (TRAFs). Binding site specificity and activation

of downstream pathways by distinct TRAFs. J. Biol. Chem. 274, 14246–54 (1999).

154. Xie, P., Hostager, B. S., Munroe, M. E., Moore, C. R. & Bishop, G. A. Cooperation

between TNF receptor-associated factors 1 and 2 in CD40 signaling. J. Immunol. 176,

5388–400 (2006).

155. Vallabhapurapu, S. et al. Nonredundant and complementary functions of TRAF2 and

TRAF3 in a ubiquitination cascade that activates NIK-dependent alternative NF-kappaB

signaling. Nat. Immunol. 9, 1364–70 (2008).

156. Zarnegar, B. J. et al. Noncanonical NF-kappaB activation requires coordinated assembly

of a regulatory complex of the adaptors cIAP1, cIAP2, TRAF2 and TRAF3 and the kinase

NIK. Nat. Immunol. 9, 1371–8 (2008).

157. Nakano, H. et al. Targeted disruption of Traf5 gene causes defects in CD40- and CD27-

mediated lymphocyte activation. Proc. Natl. Acad. Sci. U. S. A. 96, 9803–8 (1999).

142

158. Rowland, S. L. et al. A novel mechanism for TNFR-associated factor 6-dependent CD40

signaling. J. Immunol. 179, 4645–53 (2007).

159. Diehl, L. et al. CD40 activation in vivo overcomes peptide-induced peripheral cytotoxic

T-lymphocyte tolerance and augments anti-tumor vaccine efficacy. Nat. Med. 5, 774–9

(1999).

160. Cella, M. et al. Ligation of CD40 on dendritic cells triggers production of high levels of

interleukin-12 and enhances T cell stimulatory capacity: T-T help via APC activation. J.

Exp. Med. 184, 747–52 (1996).

161. Fujii, S.-I., Liu, K., Smith, C., Bonito, A. J. & Steinman, R. M. The linkage of innate to

adaptive immunity via maturing dendritic cells in vivo requires CD40 ligation in addition

to antigen presentation and CD80/86 costimulation. J. Exp. Med. 199, 1607–18 (2004).

162. Miga, A. J. et al. Dendritic cell longevity and T cell persistence is controlled by CD154-

CD40 interactions. Eur. J. Immunol. 31, 959–65 (2001).

163. Koch, F. et al. High level IL-12 production by murine dendritic cells: upregulation via

MHC class II and CD40 molecules and downregulation by IL-4 and IL-10. J. Exp. Med.

184, 741–6 (1996).

164. Keppler, S. J., Theil, K., Vucikuja, S. & Aichele, P. Effector T-cell differentiation during

viral and bacterial infections: Role of direct IL-12 signals for cell fate decision of CD8(+)

T cells. Eur. J. Immunol. 39, 1774–83 (2009).

143

165. Lee, J.-B., Lee, K.-A. & Chang, J. Phenotypic changes induced by IL-12 priming regulate

effector and memory CD8 T cell differentiation. Int. Immunol. 19, 1039–48 (2007).

166. Chang, J. et al. IL-12 priming during in vitro antigenic stimulation changes properties of

CD8 T cells and increases generation of effector and memory cells. J. Immunol. 172,

2818–26 (2004).

167. Carr, J. A., Rogerson, J. A., Mulqueen, M. J., Roberts, N. A. & Nash, A. A. The role of

endogenous interleukin-12 in resistance to murine cytomegalovirus (MCMV) infection

and a novel action for endogenous IL-12 p40. J. Interferon Cytokine Res. 19, 1145–52

(1999).

168. Oxenius, A., Karrer, U., Zinkernagel, R. M. & Hengartner, H. IL-12 is not required for

induction of type 1 cytokine responses in viral infections. J. Immunol. 162, 965–73

(1999).

169. Curtsinger, J. M., Valenzuela, J. O., Agarwal, P., Lins, D. & Mescher, M. F. Type I IFNs

provide a third signal to CD8 T cells to stimulate clonal expansion and differentiation. J.

Immunol. 174, 4465–9 (2005).

170. Ruddle, N. H. & Waksman, B. H. Cytotoxic effect of lymphocyte-antigen interaction in

delayed hypersensitivity. Science 157, 1060–2 (1967).

171. Gray, P. W. et al. Cloning and expression of cDNA for human lymphotoxin, a

lymphokine with tumour necrosis activity. Nature 312, 721–4

144

172. Browning, J. L. et al. Lymphotoxin beta, a novel member of the TNF family that forms a

heteromeric complex with lymphotoxin on the cell surface. Cell 72, 847–56 (1993).

173. Scheu, S. et al. Targeted disruption of LIGHT causes defects in costimulatory T cell

activation and reveals cooperation with lymphotoxin beta in mesenteric lymph node

genesis. J. Exp. Med. 195, 1613–24 (2002).

174. Summers-DeLuca, L. E. et al. Expression of lymphotoxin-alphabeta on antigen-specific T

cells is required for DC function. J. Exp. Med. 204, 1071–81 (2007).

175. Gonzalez, L. C. et al. A coreceptor interaction between the CD28 and TNF receptor

family members B and T lymphocyte attenuator and herpesvirus entry mediator. Proc.

Natl. Acad. Sci. U. S. A. 102, 1116–21 (2005).

176. Mackay, F., Majeau, G. R., Hochman, P. S. & Browning, J. L. Lymphotoxin beta receptor

triggering induces activation of the nuclear factor kappaB transcription factor in some cell

types. J. Biol. Chem. 271, 24934–8 (1996).

177. Dejardin, E. et al. The lymphotoxin-beta receptor induces different patterns of gene

expression via two NF-kappaB pathways. Immunity 17, 525–35 (2002).

178. Ngo, V. N. et al. Lymphotoxin alpha/beta and tumor necrosis factor are required for

stromal cell expression of homing chemokines in B and T cell areas of the spleen. J. Exp.

Med. 189, 403–12 (1999).

179. Murphy, M. et al. Expression of the lymphotoxin beta receptor on follicular stromal cells

in human lymphoid tissues. Cell Death Differ. 5, 497–505 (1998).

145

180. Gramaglia, I., Mauri, D. N., Miner, K. T., Ware, C. F. & Croft, M. Lymphotoxin

alphabeta is expressed on recently activated naive and Th1-like CD4 cells but is down-

regulated by IL-4 during Th2 differentiation. J. Immunol. 162, 1333–8 (1999).

181. Browning, J. L. et al. Characterization of lymphotoxin-alpha beta complexes on the

surface of mouse lymphocytes. J. Immunol. 159, 3288–98 (1997).

182. Ware, C. F., VanArsdale, T. L., Crowe, P. D. & Browning, J. L. The ligands and receptors

of the lymphotoxin system. Curr. Top. Microbiol. Immunol. 198, 175–218 (1995).

183. Onder, L. et al. Endothelial cell-specific lymphotoxin-β receptor signaling is critical for

lymph node and high endothelial venule formation. J. Exp. Med. 210, 465–73 (2013).

184. Nishikawa, S.-I., Honda, K., Vieira, P. & Yoshida, H. Organogenesis of peripheral

lymphoid organs. Immunol. Rev. 195, 72–80 (2003).

185. Yoshida, H. et al. Different cytokines induce surface lymphotoxin-alphabeta on IL-7

receptor-alpha cells that differentially engender lymph nodes and Peyer’s patches.

Immunity 17, 823–33 (2002).

186. Kabashima, K. et al. Intrinsic lymphotoxin-beta receptor requirement for homeostasis of

lymphoid tissue dendritic cells. Immunity 22, 439–50 (2005).

187. De Trez, C. et al. The inhibitory HVEM-BTLA pathway counter regulates lymphotoxin

receptor signaling to achieve homeostasis of dendritic cells. J. Immunol. 180, 238–48

(2008).

146

188. Suresh, M. et al. Role of lymphotoxin alpha in T-cell responses during an acute viral

infection. J. Virol. 76, 3943–51 (2002).

189. Benedict, C. A. et al. Lymphotoxins and cytomegalovirus cooperatively induce interferon-

beta, establishing host-virus détente. Immunity 15, 617–26 (2001).

190. Lucas, R. et al. A role for lymphotoxin beta receptor in host defense against

Mycobacterium bovis BCG infection. Eur. J. Immunol. 29, 4002–10 (1999).

191. Ehlers, S. et al. The lymphotoxin beta receptor is critically involved in controlling

infections with the intracellular pathogens Mycobacterium tuberculosis and Listeria

monocytogenes. J. Immunol. 170, 5210–8 (2003).

192. Junt, T. et al. Expression of lymphotoxin beta governs immunity at two distinct levels.

Eur. J. Immunol. 36, 2061–75 (2006).

193. Berger, D. P. et al. Lymphotoxin-beta-deficient mice show defective antiviral immunity.

Virology 260, 136–47 (1999).

194. Gommerman, J. L. et al. A role for surface lymphotoxin in experimental autoimmune

encephalomyelitis independent of LIGHT. J. Clin. Invest. 112, 755–67 (2003).

195. Ettinger, R. et al. A critical role for lymphotoxin-beta receptor in the development of

diabetes in nonobese diabetic mice. J. Exp. Med. 193, 1333–40 (2001).

196. Wu, Q. et al. Reversal of spontaneous autoimmune insulitis in nonobese diabetic mice by

soluble lymphotoxin receptor. J. Exp. Med. 193, 1327–32 (2001).

147

197. Lee, Y. et al. Recruitment and activation of naive T cells in the islets by lymphotoxin beta

receptor-dependent tertiary lymphoid structure. Immunity 25, 499–509 (2006).

198. Fava, R. A. et al. A role for the lymphotoxin/LIGHT axis in the pathogenesis of murine

collagen-induced arthritis. J. Immunol. 171, 115–26 (2003).

199. Stopfer, P. et al. Blocking lymphotoxin-beta receptor activation diminishes inflammation

via reduced mucosal addressin cell adhesion molecule-1 (MAdCAM-1) expression and

leucocyte margination in chronic DSS-induced colitis. Clin. Exp. Immunol. 136, 21–9

(2004).

200. Mackay, F. et al. Both the lymphotoxin and tumor necrosis factor pathways are involved

in experimental murine models of colitis. Gastroenterology 115, 1464–75 (1998).

201. Tamada, K. et al. Modulation of T-cell-mediated immunity in tumor and graft-versus-host

disease models through the LIGHT co-stimulatory pathway. Nat. Med. 6, 283–9 (2000).

202. Tamada, K. et al. Blockade of LIGHT/LTbeta and CD40 signaling induces allospecific T

cell anergy, preventing graft-versus-host disease. J. Clin. Invest. 109, 549–57 (2002).

203. Wu, Q., Fu, Y.-X. & Sontheimer, R. D. Blockade of lymphotoxin signaling inhibits the

clinical expression of murine graft-versus-host skin disease. J. Immunol. 172, 1630–6

(2004).

204. Banks, T. A. et al. A lymphotoxin-IFN-beta axis essential for lymphocyte survival

revealed during cytomegalovirus infection. J. Immunol. 174, 7217–25 (2005).

148

205. Platanias, L. C. Mechanisms of type-I- and type-II-interferon-mediated signalling. Nat.

Rev. Immunol. 5, 375–86 (2005).

206. Sen, G. C. Viruses and interferons. Annu. Rev. Microbiol. 55, 255–81 (2001).

207. Nestle, F. O. et al. Plasmacytoid predendritic cells initiate psoriasis through interferon-

alpha production. J. Exp. Med. 202, 135–43 (2005).

208. Lande, R. et al. Plasmacytoid dendritic cells sense self-DNA coupled with antimicrobial

peptide. Nature 449, 564–9 (2007).

209. Tovey, M. G. et al. Role of the type I interferons in allograft rejection. J. Leukoc. Biol. 59,

512–7 (1996).

210. Diamond, M. S. et al. Type I interferon is selectively required by dendritic cells for

immune rejection of tumors. J. Exp. Med. 208, 1989–2003 (2011).

211. Fuertes, M. B. et al. Host type I IFN signals are required for antitumor CD8+ T cell

responses through CD8 + dendritic cells. J. Exp. Med. 208, 2005–16 (2011).

212. LeBon, A. & Tough, D. F. Type I interferon as a stimulus for cross-priming. Cytokine

Growth Factor Rev. 19, 33–40 (2008).

213. Pollara, G., Handley, M. E., Kwan, a, Chain, B. M. & Katz, D. R. Autocrine type I

interferon amplifies dendritic cell responses to lipopolysaccharide via the nuclear factor-

kappaB/p38 pathways. Scand. J. Immunol. 63, 151–4 (2006).

149

214. Parlato, S. Expression of CCR-7, MIP-3beta , and Th-1 chemokines in type I IFN-induced

monocyte-derived dendritic cells: importance for the rapid acquisition of potent migratory

and functional activities. Blood 98, 3022–3029 (2001).

215. Farrar, J. D., Smith, J. D., Murphy, T. L. & Murphy, K. M. Recruitment of Stat4 to the

human interferon-alpha/beta receptor requires activated Stat2. J. Biol. Chem. 275, 2693–7

(2000).

216. Nguyen, K. B. et al. Interferon alpha/beta-mediated inhibition and promotion of interferon

gamma: STAT1 resolves a paradox. Nat. Immunol. 1, 70–6 (2000).

217. LeBon, A. et al. Direct stimulation of T cells by type I IFN enhances the CD8+ T cell

response during cross-priming. J. Immunol. 176, 4682–9 (2006).

218. Van Holten, J. et al. Treatment with recombinant interferon-beta reduces inflammation

and slows cartilage destruction in the collagen-induced arthritis model of rheumatoid

arthritis. Arthritis Res. Ther. 6, R239–49 (2004).

219. Adriaansen, J. et al. Intraarticular interferon-beta gene therapy ameliorates adjuvant

arthritis in rats. Hum. Gene Ther. 17, 985–96 (2006).

220. Weinstock-Guttman, B., Ramanathan, M. & Zivadinov, R. Interferon-beta treatment for

relapsing multiple sclerosis. Expert Opin. Biol. Ther. 8, 1435–47 (2008).

221. Tweezer-Zaks, N., Rabinovich, E., Lidar, M. & Livneh, A. Interferon-alpha as a treatment

modality for colchicine- resistant familial Mediterranean fever. J. Rheumatol. 35, 1362–5

(2008).

150

222. Guarda, G. et al. Type I interferon inhibits interleukin-1 production and inflammasome

activation. Immunity 34, 213–23 (2011).

223. Satoh, J., Paty, D. W. & Kim, S. U. Differential effects of beta and gamma interferons on

expression of major histocompatibility complex antigens and intercellular adhesion

molecule-1 in cultured fetal human astrocytes. Neurology 45, 367–73 (1995).

224. Hall, G. L., Wing, M. G., Compston, D. a. S. & Scolding, N. J. Β-Interferon Regulates the

Immunomodulatory Activity of Neonatal Rodent Microglia. J. Neuroimmunol. 72, 11–19

(1997).

225. Defazio, G. et al. Interferon beta-1a downregulates TNFalpha-induced intercellular

adhesion molecule 1 expression on brain microvascular endothelial cells through a

tyrosine kinase-dependent pathway. Brain Res. 881, 227–30 (2000).

226. Corsini, E. et al. Effects of beta-IFN-1b treatment in MS patients on adhesion between

PBMNCs, HUVECs and MS-HBECs: an in vivo and in vitro study. J. Neuroimmunol. 79,

76–83 (1997).

227. Dikopoulos, N. et al. Type I IFN negatively regulates CD8+ T cell responses through IL-

10-producing CD4+ T regulatory 1 cells. J. Immunol. 174, 99–109 (2005).

228. Gil, M., Salomon, R., Louten, J. & Biron, C. Modulation of STAT1 protein levels: a

mechanism shaping CD8 T-cell responses in vivo. Blood 107, 987–993 (2006).

229. Gil, M. P. et al. Regulating type 1 IFN effects in CD8 T cells during viral infections:

changing STAT4 and STAT1 expression for function. Blood 120, 3718–28 (2012).

151

230. Bochtler, P., Kröger, A., Schirmbeck, R. & Reimann, J. Type I IFN-induced, NKT cell-

mediated negative control of CD8 T cell priming by dendritic cells. J. Immunol. 181,

1633–43 (2008).

231. Iversen, A.-C., Norris, P. S., Ware, C. F. & Benedict, C. A. Human NK cells inhibit

cytomegalovirus replication through a noncytolytic mechanism involving lymphotoxin-

dependent induction of IFN-beta. J. Immunol. 175, 7568–74 (2005).

232. Lund, F. E. et al. Lymphotoxin-alpha-deficient mice make delayed, but effective, T and B

cell responses to influenza. J. Immunol. 169, 5236–43 (2002).

233. Bevan, M. J. Helping the CD8(+) T-cell response. Nat. Rev. Immunol. 4, 595–602 (2004).

234. Hochweller, K. & Anderton, S. M. Kinetics of costimulatory molecule expression by T

cells and dendritic cells during the induction of tolerance versus immunity in vivo. Eur. J.

Immunol. 35, 1086–96 (2005).

235. Summers-DeLuca, L. E. et al. Expression of lymphotoxin-alphabeta on antigen-specific T

cells is required for DC function. J. Exp. Med. 204, 1071–81 (2007).

236. Wong, B. R. et al. TRANCE (tumor necrosis factor [TNF]-related activation-induced

cytokine), a new TNF family member predominantly expressed in T cells, is a dendritic

cell-specific survival factor. J. Exp. Med. 186, 2075–80 (1997).

237. Wong, B. R., Josien, R. & Choi, Y. TRANCE is a TNF family member that regulates

dendritic cell and osteoclast function. J. Leukoc. Biol. 65, 715–24 (1999).

152

238. Puglielli, M. T. et al. Reversal of virus-induced systemic shock and respiratory failure by

blockade of the lymphotoxin pathway. Nat. Med. 5, 1370–4 (1999).

239. Guo, Z. et al. Cutting edge: membrane lymphotoxin regulates CD8(+) T cell-mediated

intestinal allograft rejection. J. Immunol. 167, 4796–800 (2001).

240. Tamada, K. et al. Blockade of LIGHT/LTbeta and CD40 signaling induces allospecific T

cell anergy, preventing graft-versus-host disease. J. Clin. Invest. 109, 549–57 (2002).

241. Kanodia, S. et al. Expression of LIGHT/TNFSF14 combined with vaccination against

human papillomavirus Type 16 E7 induces significant tumor regression. Cancer Res. 70,

3955–64 (2010).

242. Cells, D. et al. The Inhibitory HVEM-BTLA Pathway Counter Regulates Lymphotoxin

Receptor Signaling to Achieve Homeostasis of Dendritic Cells. J. Immunol. (2008).

243. Kabashima, K. et al. Intrinsic lymphotoxin-beta receptor requirement for homeostasis of

lymphoid tissue dendritic cells. Immunity 22, 439–50 (2005).

244. Wang, Y.-G., Kim, K. D., Wang, J., Yu, P. & Fu, Y.-X. Stimulating lymphotoxin beta

receptor on the dendritic cells is critical for their homeostasis and expansion. J. Immunol.

175, 6997–7002 (2005).

245. Caux, C. et al. Activation of human dendritic cells through CD40 cross-linking. J. Exp.

Med. 180, 1263–72 (1994).

246. Probst, H. C. et al. Histological analysis of CD11c-DTR/GFP mice after in vivo depletion

of dendritic cells. Clin. Exp. Immunol. 141, 398–404 (2005).

153

247. Driessens, G. et al. Micronuclei to detect in vivo chemotherapy damage in a p53 mutated

solid tumour. Br. J. Cancer 89, 727–9 (2003).

248. Schneider, K. et al. Lymphotoxin-mediated crosstalk between B cells and splenic stroma

promotes the initial type I interferon response to cytomegalovirus. Cell Host Microbe 3,

67–76 (2008).

249. Le Bon, A. et al. Direct stimulation of T cells by type I IFN enhances the CD8+ T cell

response during cross-priming. J. Immunol. 176, 4682–9 (2006).

250. Le Bon, A. et al. Cross-priming of CD8+ T cells stimulated by virus-induced type I

interferon. Nat. Immunol. 4, 1009–15 (2003).

251. Luft, T. et al. Type I IFNs enhance the terminal differentiation of dendritic cells. J.

Immunol. 161, 1947–53 (1998).

252. Huarte, E. et al. Recombinant adenoviral vectors turn on the type I interferon system

without inhibition of transgene expression and viral replication. Mol. Ther. 14, 129–38

(2006).

253. Chiang, E. Y. et al. Targeted depletion of lymphotoxin-alpha-expressing TH1 and TH17

cells inhibits autoimmune disease. Nat. Med. 15, 766–73 (2009).

254. Vu, F., Dianzani, U., Ware, C. F., Mak, T. & Gommerman, J. L. ICOS, CD40, and

lymphotoxin beta receptors signal sequentially and interdependently to initiate a germinal

center reaction. J. Immunol. 180, 2284–93 (2008).

154

255. Schoenberger, S. P., Toes, R. E., van der Voort, E. I., Offringa, R. & Melief, C. J. T-cell

help for cytotoxic T lymphocytes is mediated by CD40-CD40L interactions. Nature 393,

480–3 (1998).

256. Lind, E. F. et al. Cross-Presentation of Soluble Antigens 1. J. Immunol. 0–9 (2008).

257. Lefrançois, L., Altman, J. D., Williams, K. & Olson, S. Soluble antigen and CD40

triggering are sufficient to induce primary and memory cytotoxic T cells. J. Immunol. 164,

725–32 (2000).

258. Matthews, K. E. et al. Increasing the survival of dendritic cells in vivo does not replace

the requirement for CD4+ T cell help during primary CD8+ T cell responses. J. Immunol.

179, 5738–47 (2007).

259. Maxwell, J. R., Ruby, C., Kerkvliet, N. I. & Vella, A. T. Contrasting the roles of

costimulation and the natural adjuvant lipopolysaccharide during the induction of T cell

immunity. J. Immunol. 168, 4372–81 (2002).

260. Bachmann, M. F. et al. Cutting edge: distinct roles for T help and CD40/CD40 ligand in

regulating differentiation of proliferation-competent memory CD8+ T cells. J. Immunol.

173, 2217–21 (2004).

261. Hernandez, M. G. H., Shen, L. & Rock, K. L. CD40 on APCs is needed for optimal

programming, maintenance, and recall of CD8+ T cell memory even in the absence of

CD4+ T cell help. J. Immunol. 180, 4382–90 (2008).

155

262. Johnson, S. et al. Selected Toll-like receptor ligands and viruses promote helper-

independent cytotoxic T cell priming by upregulating CD40L on dendritic cells. Immunity

30, 218–27 (2009).

263. Van Deusen, K. E., Rajapakse, R. & Bullock, T. N. J. CD70 expression by dendritic cells

plays a critical role in the immunogenicity of CD40-independent, CD4+ T cell-dependent,

licensed CD8+ T cell responses. J. Leukoc. Biol. 87, 477–85 (2010).

264. Hoshino, K., Kaisho, T., Iwabe, T., Takeuchi, O. & Akira, S. Differential involvement of

IFN-beta in Toll-like receptor-stimulated dendritic cell activation. Int. Immunol. 14, 1225–

31 (2002).

265. Tam, M. a, Sundquist, M. & Wick, M. J. MyD88 and IFN-alphabeta differentially control

maturation of bystander but not Salmonella-associated dendritic cells or

CD11cintCD11b+ cells during infection. Cell. Microbiol. 10, 1517–29 (2008).

266. Kumaraguru, U., Davis, I. A., Deshpande, S., Tevethia, S. S. & Rouse, B. T. Lymphotoxin

alpha-/- mice develop functionally impaired CD8+ T cell responses and fail to contain

virus infection of the central nervous system. J. Immunol. 166, 1066–74 (2001).

267. Lukashev, M. et al. Targeting the lymphotoxin-beta receptor with agonist antibodies as a

potential cancer therapy. Cancer Res. 66, 9617–24 (2006).

268. Trinchieri, G. Type I interferon: friend or foe? J. Exp. Med. 207, 2053–63 (2010).

156

269. Xiao, Z., Casey, K. a, Jameson, S. C., Curtsinger, J. M. & Mescher, M. F. Programming

for CD8 T cell memory development requires IL-12 or type I IFN. J. Immunol. 182,

2786–94 (2009).

270. Teijaro, J. R. et al. Persistent LCMV infection is controlled by blockade of type I

interferon signaling. Science 340, 207–11 (2013).

271. Wilson, E. B. et al. Blockade of chronic type I interferon signaling to control persistent

LCMV infection. Science 340, 202–7 (2013).

272. Theill, L. E., Boyle, W. J. & Penninger, J. M. RANK-L and RANK: T cells, bone loss,

and mammalian evolution. Annu. Rev. Immunol. 20, 795–823 (2002).

273. Yarilina, A., Park-Min, K.-H., Antoniv, T., Hu, X. & Ivashkiv, L. B. TNF activates an

IRF1-dependent autocrine loop leading to sustained expression of chemokines and

STAT1-dependent type I interferon-response genes. Nat. Immunol. 9, 378–87 (2008).

274. Summers deLuca, L. et al. LTβR signaling in dendritic cells induces a type I IFN response

that is required for optimal clonal expansion of CD8+ T cells. Proc. Natl. Acad. Sci. U. S.

A. 108, 2046–51 (2011).

275. Moseman, E. A. et al. B cell maintenance of subcapsular sinus macrophages protects

against a fatal viral infection independent of adaptive immunity. Immunity 36, 415–26

(2012).

276. Ohshima, D. et al. RANK signaling induces interferon-stimulated genes in the fetal

thymic stroma. Biochem. Biophys. Res. Commun. 408, 530–6 (2011).

157

277. Otero, D. C., Baker, D. P. & David, M. IRF7-dependent IFN-β production in response to

RANKL promotes medullary thymic epithelial cell development. J. Immunol. 190, 3289–

98 (2013).

278. Ivashkiv, L. B. & Donlin, L. T. Regulation of type I interferon responses. Nat. Rev.

Immunol. 14, 36–49 (2014).

279. Ohashi, P. S. et al. Ablation of “tolerance” and induction of diabetes by virus infection in

viral antigen transgenic mice. Cell 65, 305–17 (1991).

280. Lin, A. C. C., Dissanayake, D., Dhanji, S., Elford, A. R. & Ohashi, P. S. Different toll-like

receptor stimuli have a profound impact on cytokines required to break tolerance and

induce autoimmunity. PLoS One 6, e23940 (2011).

281. Dissanayake, D. et al. Nuclear factor-κB1 controls the functional maturation of dendritic

cells and prevents the activation of autoreactive T cells. Nat. Med. 17, 1663–1668 (2011).

282. Sakaguchi, S. et al. Essential role of IRF-3 in lipopolysaccharide-induced interferon-beta

gene expression and endotoxin shock. Biochem. Biophys. Res. Commun. 306, 860–6

(2003).

283. Sanjo, H., Zajonc, D. M., Braden, R., Norris, P. S. & Ware, C. F. Allosteric regulation of

the ubiquitin:NIK and ubiquitin:TRAF3 E3 ligases by the lymphotoxin-beta receptor. J.

Biol. Chem. 285, 17148–55 (2010).

284. Razani, B. et al. Negative feedback in noncanonical NF-kappaB signaling modulates NIK

stability through IKKalpha-mediated phosphorylation. Sci. Signal. 3, ra41 (2010).

158

285. Oganesyan, G. et al. Critical role of TRAF3 in the Toll-like receptor-dependent and -

independent antiviral response. Nature 439, 208–11 (2006).

286. González-Navajas, J. M., Lee, J., David, M. & Raz, E. Immunomodulatory functions of

type I interferons. Nat. Rev. Immunol. 12, 125–35 (2012).

287. Chawla-Sarkar, M. et al. Apoptosis and interferons: role of interferon-stimulated genes as

mediators of apoptosis. Apoptosis 8, 237–49 (2003).

288. Smith, A. et al. The role of the integrin LFA-1 in T-lymphocyte migration. Immunol. Rev.

218, 135–46 (2007).

289. Scholer, A., Hugues, S., Boissonnas, A., Fetler, L. & Amigorena, S. Intercellular adhesion

molecule-1-dependent stable interactions between T cells and dendritic cells determine

CD8+ T cell memory. Immunity 28, 258–70 (2008).

290. Henrickson, S. E. et al. T cell sensing of antigen dose governs interactive behavior with

dendritic cells and sets a threshold for T cell activation. Nat. Immunol. 9, 282–91 (2008).

291. King, C. G. et al. T cell affinity regulates asymmetric division, effector cell

differentiation, and tissue pathology. Immunity 37, 709–20 (2012).

292. Yang, X. D., Karin, N., Tisch, R., Steinman, L. & McDevitt, H. O. Inhibition of insulitis

and prevention of diabetes in nonobese diabetic mice by blocking L-selectin and very late

antigen 4 adhesion receptors. Proc. Natl. Acad. Sci. U. S. A. 90, 10494–8 (1993).

293. Ettinger, R. et al. A critical role for lymphotoxin-beta receptor in the development of

diabetes in nonobese diabetic mice. J. Exp. Med. 193, 1333–40 (2001).

159

294. Jansen, J. H. et al. Induction of CD11a/leukocyte function antigen-1 and

CD54/intercellular adhesion molecule-1 on hairy cell leukemia cells is accompanied by

enhanced susceptibility to T-cell but not lymphokine-activated killer-cell cytotoxicity.

Blood 80, 478–83 (1992).

295. Marotta, G., Zagonel, V. & Pinto, A. Induction of LFA-1/CD11a and ICAM-1/CD54

adhesion molecules on neoplastic B cells during in vivo treatment of chronic lymphocytic

leukemia with interferon-alpha 2. Blood 81, 267–9 (1993).

296. Muraro, P. A. et al. Decreased integrin gene expression in patients with MS responding to

interferon-beta treatment. J. Neuroimmunol. 150, 123–31 (2004).

297. Nakanishi, Y., Lu, B., Gerard, C. & Iwasaki, A. CD8(+) T lymphocyte mobilization to

virus-infected tissue requires CD4(+) T-cell help. Nature 462, 510–3 (2009).

298. Häcker, H. et al. Specificity in Toll-like receptor signalling through distinct effector

functions of TRAF3 and TRAF6. Nature 439, 204–7 (2006).

299. Murphy, M. et al. Expression of the lymphotoxin beta receptor on follicular stromal cells

in human lymphoid tissues. Cell Death Differ. 5, 497–505 (1998).

300. Browning, J. L. et al. Signaling through the lymphotoxin beta receptor induces the death

of some adenocarcinoma tumor lines. J. Exp. Med. 183, 867–78 (1996).

301. Winter, H. et al. Tumor-specific T cells signal tumor destruction via the lymphotoxin beta

receptor. J. Transl. Med. 5, 14 (2007).

160

302. Upadhyay, V. & Fu, Y.-X. Lymphotoxin signalling in immune homeostasis and the

control of microorganisms. Nat. Rev. Immunol. 13, 270–9 (2013).

303. Razani, B., Reichardt, A. D. & Cheng, G. Non-canonical NF-κB signaling activation and

regulation: principles and perspectives. Immunol. Rev. 244, 44–54 (2011).

304. Jin, J. et al. Noncanonical NF-κB Pathway Controls the Production of Type I Interferons

in Antiviral Innate Immunity. Immunity 40, 342–54 (2014).

305. Schneider, K. et al. Lymphotoxin-mediated crosstalk between B cells and splenic stroma

promotes the initial type I interferon response to cytomegalovirus. Cell Host Microbe 3,

67–76 (2008).

306. Kutsch, S., Degrandi, D. & Pfeffer, K. Immediate lymphotoxin beta receptor-mediated

transcriptional response in host defense against L. monocytogenes. Immunobiology 213,

353–66 (2008).

307. Lucifora, J. et al. Specific and Nonhepatotoxic Degradation of Nuclear Hepatitis B Virus

cccDNA. Science (2014). doi:10.1126/science.1243462

308. Srivastava, S., Koch, M. A., Pepper, M. & Campbell, D. J. Type I interferons directly

inhibit regulatory T cells to allow optimal antiviral T cell responses during acute LCMV

infection. J. Exp. Med. (2014). doi:10.1084/jem.20131556

309. Sanjo, H., Zajonc, D. M., Braden, R., Norris, P. S. & Ware, C. F. Allosteric regulation of

the ubiquitin:NIK and ubiquitin:TRAF3 E3 ligases by the lymphotoxin-beta receptor. J.

Biol. Chem. 285, 17148–55 (2010).

161

310. Abe, K. et al. Distinct contributions of TNF and LT cytokines to the development of

dendritic cells in vitro and their recruitment in vivo. Blood 101, 1477–83 (2003).

311. Wu, Q. et al. The requirement of membrane lymphotoxin for the presence of dendritic

cells in lymphoid tissues. J. Exp. Med. 190, 629–38 (1999).

312. Gunn, M. D. et al. Mice lacking expression of secondary lymphoid organ chemokine have

defects in lymphocyte homing and dendritic cell localization. J. Exp. Med. 189, 451–60

(1999).

313. Caton, M. L., Smith-Raska, M. R. & Reizis, B. Notch-RBP-J signaling controls the

homeostasis of CD8- dendritic cells in the spleen. J. Exp. Med. 204, 1653–64 (2007).

314. Lewis, K. L. et al. Notch2 receptor signaling controls functional differentiation of

dendritic cells in the spleen and intestine. Immunity 35, 780–91 (2011).

162

Appendices

Immunity Article

Notch2 Receptor Signaling Controls Functional Differentiation of Dendritic Cells in the Spleen and Intestine

Kanako L. Lewis,1 Michele L. Caton,1 Milena Bogunovic,2 Melanie Greter,2 Lucja T. Grajkowska,1 Dennis Ng,3 Apostolos Klinakis,4 Israel F. Charo,5 Steffen Jung,6 Jennifer L. Gommerman,3 Ivaylo I. Ivanov,1 Kang Liu,1 Miriam Merad,2 and Boris Reizis1,* 1Department of Microbiology and Immunology, Columbia University Medical Center, New York, NY 10032, USA 2Department of Oncological Sciences, Mount Sinai School of Medicine, New York, NY 10029, USA 3Department of Immunology, University of Toronto, Toronto, Ontario M5S 1A8, Canada 4Biomedical Research Foundation, Academy of Athens, Athens 11527, Greece 5Gladstone Institute of Cardiovascular Disease, University of California, San Francisco, San Francisco, CA 94158, USA 6Department of Immunology, The Weizmann Institute of Science, Rehovot, 76100, Israel *Correspondence: [email protected] DOI 10.1016/j.immuni.2011.08.013

SUMMARY lymphoid organs and their CD103+CD11b counterparts in tissues mediate efficient cross-presentation to cytotoxic T cells Dendritic cells (DCs) in tissues and lymphoid organs (Shortman and Heath, 2010). The CD8a-negative CD8CD11b+ comprise distinct functional subsets that differen- subset is preferentially involved in MHC class II (MHC II)- tiate in situ from circulating progenitors. Tissue- restricted Ag presentation to CD4+ helper T cells (Dudziak + specific signals that regulate DC subset differ- et al., 2007). In the spleen, CD11b DCs are preferentially local- entiation are poorly understood. We report that ized to the marginal zone (MZ), a unique structure that filters the DC-specific deletion of the Notch2 receptor caused incoming blood (Mebius and Kraal, 2005). In the intestinal lamina propria (LP), the CD11b+ DC population is comprised of two a reduction of DC populations in the spleen. Within + + + distinct subsets. The CD11b CD103 subset is thought to the splenic CD11b DC subset, Notch signaling mediate Ag capture and transport to mesenteric lymph node blockade ablated a distinct population marked by (LN). Recently, it was shown to be particularly efficient for the high expression of the adhesion molecule Esam. induction of interleukin 17 (IL-17)-secreting helper T (Th17) cells hi The Notch-dependent Esam DC subset required in vitro (Denning et al., 2011), although its role in T cell differenti- lymphotoxin beta receptor signaling, proliferated ation in vivo remains unclear. Conversely, the CD11b+CD103 in situ, and facilitated CD4+ T cell priming. The population does not migrate to LN, is capable of high-level cyto- Notch-independent Esamlo DCs expressed mono- kine secretion, and appears closely related to macrophages cyte-related genes and showed superior cytokine (Bogunovic et al., 2009; Schulz et al., 2009; Varol et al., 2009). responses. In addition, Notch2 deletion led to the Classical DCs along with pDCs, monocytes, and macro- loss of CD11b+CD103+ DCs in the intestinal lamina phages originate from the common macrophage and DC progenitor (MDP) in the bone marrow (BM) (Fogg et al., 2006). propria and to a corresponding decrease of IL-17- + Commitment to the DC lineage occurs in the BM at the level of producing CD4 T cells in the intestine. Thus, Notch2 common DC progenitors (CDP) (Naik et al., 2007; Onai et al., is a common differentiation signal for T cell-priming 2007), whereas the terminal differentiation of classical DC + CD11b DC subsets in the spleen and intestine. subsets occurs in the periphery. All DCs in the lymphoid organs and CD103+ DCs in tissues are thought to develop from pre-DC (Ginhoux et al., 2009; Liu et al., 2009), a blood-derived INTRODUCTION progenitor originally defined in the spleen (Naik et al., 2006). Similarly, the unique CD11b+CD103+ subset in the intestinal LP Dendritic cells (DCs) represent the primary antigen (Ag)-present- is derived from pre-DCs. All pre-DC-derived subsets are low or ing cell population in the immune system. They can detect negative for fractalkine receptor Cx3cr1 and preferentially pathogens through pattern recognition receptors such as Toll- depend on signaling by Flt3 ligand through its receptor Flt3. like receptors (TLRs), migrate into the T cell areas of lymphoid On the other hand, CD11b+ DCs in tissues arise from MDP- organs, secrete immunostimulatory cytokines such as inter- derived monocytes, express Cx3cr1, and depend on macro- leukin-12 (IL-12), and present pathogen-derived peptides to phage colony-stimulating factor receptor Csf1r rather than on naive T cells (Steinman and Idoyaga, 2010). To initiate appro- Flt3 (Bogunovic et al., 2009; Ginhoux et al., 2009; Varol et al., priate immune responses to different pathogen types, DCs 2009). Thus, the homogeneity and single pre-DC origin of comprise distinct functional subsets including interferon-pro- DC subsets in lymphoid organs appears to contrast with the ducing plasmacytoid DCs (pDCs) and two main subsets of clas- functional heterogeneity and dual origin of DC subsets in sical DCs (cDCs). The CD8a-expressing CD8+CD11b DCs in tissues such as the intestine. Furthermore, little is known about

780 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

molecular signals that promote DC fate and impart subset and/or way to block canonical Notch signaling in DCs, we used Itgax- tissue specificity on DC progenitors. cre-induced expression of dominant-negative human MAML1 Notch is an evolutionarily conserved signaling pathway that (DNMAML1) protein fused to green fluorescent protein (GFP) allows cells to adopt cell fates dictated by their microenviron- (Tu et al., 2005). The analysis of GFP fluorescence confirmed ment (Bray, 2006). The interaction of Notch receptor with its DC-specific activation of DNMAML1 expression in the resulting ligand on a neighboring cell causes receptor cleavage that DC-DNMAML1 mice (see Figure S1 available online). Indeed, releases the intracellular domain of Notch (NICD), which translo- Notch2-dependent MZ B cells showed minimal recombination cates into the nucleus and binds the transcription factor CSL and were present in normal numbers in mice with Notch2 dele- (called RBPJ in the mouse). The resulting NICD-RBPJ complex tion (DC-Notch2D) or with DNMAML1 activation (Figure S1). As recruits coactivators of the Mastermind (MAML) family and shown in Figures 1A and 1B, mice with DC-specific deletion of activates Notch-dependent gene expression programs, includ- Notch1 (DC-Notch1D) had normal splenic DC populations. This ing canonical targets such as Hes1 and Deltex (Dtx1). Notch is consistent with normal DC development after Notch1 deletion signaling is regulated at multiple levels including ligand endocy- in hematopoietic chimeras (Radtke et al., 2000). In contrast, DC- tosis, receptor glycosylation, and composition of the transacti- Notch2D and DC-DNMAML1 mice had reduced fraction and vation complex. Mammals have five Notch ligands of the absolute numbers of splenic CD11chiMHC II+ classical DCs Delta-like and Jagged families and four Notch receptors (Figures 1A and 1B). The reduction was specific to cDCs, as (Notch1–4), of which only Notch1 and Notch2 contain transacti- shown by the fact that pDC populations were normal despite effi- vation domains. These two receptors appear to mediate the cient cre recombination (Figure S1). majority of Notch-dependent developmental processes and As with DC-specific Rbpj deletion (DC-RbpjD)(Caton et al., their deletion causes embryonic lethality. 2007), the CD8CD11b+ subset was significantly reduced in Notch signaling plays essential roles in lymphocyte develop- the spleens of DC-Notch2D and DC-DNMAML1 mice (Figures ment by inducing progenitor differentiation in unique anatomic 1C and 1D). In addition, the CD8+ DC subset was decreased locations (Radtke et al., 2010; Yashiro-Ohtani et al., 2010; in DC-Notch2D spleens, whereas a distinct population of Yuan et al., 2010). Thus, Delta-like 4 (Dll4) on thymic epithelium CD8CD11b double-negative cells became apparent within drives T cell lineage development through Notch1 on thymic the DC population. A similar albeit less pronounced phenotype progenitors, whereas Delta-like 1 (Dll1) on the splenic MZ was observed in DNMAML1-overexpressing DCs, suggesting stroma specifies MZ B cell subset through Notch2. However, the involvement of canonical Notch signaling. The CD8CD11b the role of Notch in the development of innate immune double-negative DCs arising after Notch2 deletion included two system remains poorly understood. Notch1 has been impli- distinct populations expressing either signal regulatory protein cated in DC differentiation in vitro (Cheng et al., 2003; Zhou alpha (SIRPa) or CD24, the markers associated with CD11b+ et al., 2009), although earlier studies on Notch1 deletion in vivo and CD8+ DC lineages, respectively (Figure 1E; Naik et al., argue against this notion (Radtke et al., 2000). We have shown 2006). Similar populations could be detected among the rare previously that DC-specific deletion of RBPJ caused partial CD8CD11b DCs in wild-type mice, suggesting that they reduction of the splenic CD11b+ DCs, which express Notch represent natural immature stages of CD11b+ and CD8+ DC target Dtx1 in an RBPJ-dependent manner (Caton et al., development. Earlier developmental stages such as pre-DCs 2007). However, major questions remain concerning the Notch were unaffected by Notch2 deletion (Figure S1), consistent receptor involved, the partial nature and functional conse- with a late differentiation defect. quences of the phenotype, and the role of Notch in DC differen- In addition to the decrease in number, splenic CD11b+ DCs in tiation in tissues. DC-Notch2D mice exhibited an abnormal surface phenotype We now report that the Notch2 receptor controls DC differen- including higher CD11b and lower (albeit heterogeneous) SIRPa tiation in the spleen. Among the CD11b+ DCs, Notch2-RBPJ expression (Figure 1F). Splenic CD11b+ DCs express the signaling specifies a unique Cx3cr1loEsamhi DC subset, which specific marker Dcir2 (33D1) (Dudziak et al., 2007) and a fraction was required for efficient T cell priming in the spleen. Moreover, of them express CD4 (Kamath et al., 2000). As shown in Fig- Notch2 was necessary for the development of lamina propria ure 1F, the fraction of Dcir2+ and CD4+ cells was profoundly CD11b+CD103+ DCs, which in turn maintain optimal numbers reduced in Notch2-deficient CD11b+ DCs. Altogether, these of Th17 cells. These results establish Notch2 signaling as an data demonstrate that Notch2 deletion impairs the development essential tissue-specific determinant of DC differentiation in of CD11b+ and CD8+ splenic DCs and causes phenotypic abnor- the spleen and intestine. Furthermore, they demonstrate that mality of the residual CD11b+ DC population. CD11b+ DCs both in tissues and in lymphoid organs are heterogeneous and include a distinct Notch-dependent subset Notch2-RBPJ Signaling Specifies a Distinct Population specialized in T cell priming. of Splenic CD8– DCs To explain the partial reduction and abnormal phenotype of RESULTS CD11b+ DCs in DC-Notch2D, DC-DNMAML1, and DC-RbpjD mice, we hypothesized that CD11b+ DCs are heterogeneous Notch2 Controls Splenic DC Differentiation and include a distinct population eliminated by Notch2-RBPJ To determine the role of individual Notch receptors in DC differ- blockade. We therefore crossed DC-RbpjD mice to a Cx3cr1- entiation, we used Itgax-cre (CD11c-cre) deleter strain for DC- GFP reporter strain, which has been used to resolve DC subsets specific deletion of conditional Notch1 (Radtke et al., 1999) in the spleen and tissues (Bar-On et al., 2010; Varol et al., 2009). and Notch2 (McCright et al., 2006) alleles. As a complementary We found that CD11b+ DCs in wild-type spleens included GFPhi

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 781 Immunity Notch2 Controls DC Differentiation

Figure 1. Notch2 Signaling Regulates A DC-Notch1 DC-Notch2 DC-DNMAML1 B cre- cre+ Splenic DC Development 1.4 1.1 1.7 2 *** Mice with Itgax-cre-mediated deletion of Notch1 1.5 *** or Notch2 or DC-speciﬁc overexpression of cre- 1 DNMAML1 were analyzed along with the respec- 0.5

Total (%) tive cre-negative littermate controls. Statistically 0 signiﬁcant differences are indicated as follows:

) 25 5 ** 20 * ***p < 0.001; **p < 0.01; *p < 0.05. 15 (A) Representative staining proﬁles of total sple- 1.4 0.4 0.8 10 hi + CD11c nocytes with the fraction of CD11c MHC II DCs Cells (10 5 0 indicated. cre+ (B) The fraction (top) and absolute number (bottom) of splenic DCs (mean ± SD of three to six -Notch1 -Notch2 MAML1 animals per group). DC DC (C) Staining proﬁles of gated CD11chiMHC II+ MHC II DC-DN splenic DCs, with CD11b+ (blue), CD8+ (green), and double-negative (purple) subsets highlighted. C DC-Notch1 DC-Notch2 DC-DNMAML1 E The percentages represent mean ± SD, n = 3–6 for D 54 DC-Notch1 and DC-DNMAML1 and 11–12 for 72 ±6 71 ±5 79 ±4 DC-Notch2D. 16 12 ±2 11 ±4 (D) The percentage among total splenocytes (top) - cre 7 ± 1 Ctrl and absolute number (bottom) of CD11b+ and CD8+ DC subsets (mean ± SD of 8–10 animals per 11 ±4 13 ± 3 11 ±2 group). (E) The expression of SIRPa and CD24 in gated 57 69 ±3 49 ±7*** 62 ±7*** CD11chiMHC II+CD11b CD8 double-negative CD11b SIRP 23 D 8±3* DCs from conditional Notch2 (DC-Notch2 ) and 14 ±2 + cre+ 10 ±2 DC- littermate control (Ctrl) spleens. The SIRPa and Notch2 CD24+ populations indicative of the differentiation toward CD11b+ and CD8+ DC subsets, 12 ±2 39 ±6*** 24 ±4*** respectively, are indicated (representative of two CD24 CD8 animals). (F) Expression of surface markers in CD11b+ DCs D D cre- cre+ F from DC-Notch2 and control spleens. See also Figure S1. CD11b+ CD8+ 1.5 *** 0.3 *** 1 0.2 *** *

0.5 0.1 D CD11b SIRP DC-Rbpj splenocytes (Figure 2B). The 0 0 RBPJ-dependent Esamhi population was

)Total (%) )Total 20 2.0 5 ** ** 15 1.6 cell no. Relative also deﬁned by low expression of 1.2 10 *** 0.8 Clec12a, a C-type lectin expressed by 5 + Cells (10 Cells 0.4 CD8 DCs, pDCs, and myeloid cells (La- 0 0 houd et al., 2009). Staining with Esam, particularly in combination with Clec12a, CD4 Dcir2 + -Notch1 -Notch2 MAML1 -Notch1 -Notch2 MAML1 resolved the two CD11b DC populations DC DC DC DC Ctrl and allowed their analysis in the absence DC-DN DC-DN DC-Notch2 of Cx3cr1-GFP reporter (Figure 2C). The Esamhi but not EsamloCD11b+ DCs were nearly absent in all examined Notch and GFPlo populations, whereas in DC-RbpjD spleens the GFPlo mutants including DC-RbpjD, DC-Notch2D, and DC-DNMAML1 population was missing (Figure 2A). To identify a positive marker (Figure 2D). In contrast, the low Esam expression on CD8+ of this Notch-dependent DC population, we performed microar- DCs was minimally affected by Notch2 loss (Figure S2). ray analysis of total CD11b+ DCs from DC-RbpjD spleens and To test the effect of Notch overactivation in DCs, we used the focused on genes that encode cell surface markers (data not Itgax-cre strain in conjunction with a Cre-inducible overexpres- shown). One prominently reduced gene encoded Esam, an sion cassette encoding Notch1 intracellular domain (NICD) (Buo- immunoglobulin superfamily adhesion molecule that is ex- namici et al., 2009). The NICDs of Notch1 and Notch2 are highly pressed on endothelium and regulates neutrophil extravasation homologous (Radtke et al., 2010) and should be interchange- (Wegmann et al., 2006). We found that among the splenocytes, able in gain-of-function experiments. In the resulting DC-NICD Esam was abundant in a subset of CD11b+ DCs, low in CD8+ animals, splenic CD11b+ DCs consisted almost exclusively of DCs, and absent from non-DCs (Figure S2). The combination EsamhiClec12a cells (Figure 2E). Furthermore, the expression of Esam with Cx3cr1-GFP deﬁned two distinct populations, of Esam was increased in all DCs including the CD8+ subset EsamloGFPhi and EsamhiGFPlo, the latter completely absent in (Figure 2F), showing Esam as a faithful readout of Notch activity

782 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

B Figure 2. Notch2-RBPJ Signaling Speciﬁes a A C + 22 16 Distinct Population of CD11b DCs (A) The expression of Cx3cr1-GFP reporter in CD11b+ Ctrl Ctrl splenic DCs from conditional RBPJ-deﬁcient (DC-RbpjD) mice. Shown is the histogram of GFP expression in gated 70 hi + + + 77 CD11c MHC II CD11b DCs from Cx3cr1-GFP control (Ctrl) or DC-RbpjD mice (representative of four animals per

GFP group). Relative cell no. Relative 81 68 + Clec12a (B) The expression of Esam versus GFP in gated CD11b GFP + D splenic DCs from Cx3cr1-GFP control or DC-Rbpj mice. DC- DC- Ctrl Rbpj Rbpj (C) The expression of Esam versus Clec12a in gated + D DC-Rbpj 9 CD11b splenic DCs from DC-Rbpj or control mice. 14 (D) The percentage among total splenocytes (top) hi Esam Esam and absolute number (bottom) of the Esam and EsamloCD11b+ DC subsets in mice with DC-specific tar- D geting of the indicated genes (mean ± SD of three to five Esamhi Esamlo animals). Significance is indicated as in Figure 1. 1.2 *** 0.7 0.6 (E) Splenic DC populations in mice with DC-specific 0.8 0.5 expression of Notch1 intracellular domain (DC-NICD). *** *** 0.4 0.3 Shown are staining profiles of gated splenic CD11chiMHC 0.4 0.2 + + 0.1 II DCs and of their CD11b subset from DC-NICD mice or Total (%) 0 0 cre- Cre-negative littermate controls (Ctrl). Representative of eight animals per genotype. ** 10 + ** cre + + ) 12 8 (F) The expression of Esam in splenic CD11b and CD8 5 8 ** 6 DCs from DC-NICD and control mice. 4 4 See also Figure S2. 2

Cells (10 0 0

-Rbpj -Rbpj MAML1 -Notch2 -Notch2 MAML1 DC DC hi lo DC DC Esam but not of Esam DCs was severely / DC-DN DC-DN reduced in Ltbr mice (Figures 3A and 3B). E F To circumvent the pleiotropic defects of 81 LTbR deficiency, we analyzed competitive 30 chimeras reconstituted with a mixture of wild- Ctrl b 7 CD11b+ type and LT R-deficient BM. The fraction of LTbR-deficient donor cells was reduced in 58 Esamhi DC population, suggesting that the phenotype was at least partially cell intrinsic 82 Clec12a

CD11b (Figures 3C and 3D). In contrast, mice with 1 Relative cell no. Relative DC-specific targeting of Irf4 (Klein et al., 2006), DC- + 8 CD8 + NICD a transcription factor regulating CD11b DC development, showed marginal decrease of 98 both Esamhi and Esamlo populations (Figure 3B CD8 Esam Esam and data not shown). Thus, EsamhiCD11b+ DCs lo Ctrl are genetically distinct from Esam DCs and DC-NICD specifically depend on Notch2-RBPJ and LTbR signaling and to a lesser extent on Flt3 signaling. in DCs. Altogether, these data suggest that splenic CD11b+ DCs Cell surface staining of CD11b+ DCs showed that the Esamhi contain a distinct Esamhi population that is directly regulated by Cx3cr1-GFPlo population was uniformly positive for CD4 and Notch2-RBPJ signaling. Dcir2 (Figure 4A), in agreement with the preferential loss of CD4+ and Dcir2+ DCs after Notch2 deletion (Figure 1F). Genetic and Phenotypic Analysis of Splenic Esamhi DCs Conversely, EsamloCx3cr1-GFPhi cells were heterogeneous for We have tested genetic requirements for the development of CD4 and Dcir2 and high for CD11b and Clec12a. Esamlo cells ex- Notch-dependent EsamhiCD11b+ DC population. Splenocytes pressed high amounts of Flt3 and LTbR, despite their indepen- from Flt3-deficient mice showed a modest but significant reduc- dence of these receptors (Figures 4A and 4B; Figure S3). Cyto- tion of Esamhi population (Figures 3A and 3B), suggesting that spin preparations of sorted Esamhi DCs revealed irregular Flt3 signaling contributes to its development or homeostasis. nuclei and thin cytoplasmic veils, whereas Esamlo DCs showed The blockade of lymphotoxin b receptor (LTbR) signaling typical mononuclear morphology (Figure 4C). Notably, Esamlo partially reduces the number of splenic CD11b+ DCs and impairs DCs appeared distinct from splenic macrophages (Figure 4C) their in situ proliferation in a cell-intrinsic manner (De Trez et al., or intestinal CD11b+CD103 DCs (Bogunovic et al., 2009), 2008; Kabashima et al., 2005). We found that the fraction of because they did not contain prominent cytoplasmic vesicles.

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 783 Immunity Notch2 Controls DC Differentiation

hi A Flt3-/- Ltbr-/- DC-Rbpj BrdU pulse revealed more proliferating cells among Esam DCs (Figure 4E). Altogether, the Esamhi subset of CD11b+ DCs appears distinct in its higher turnover rate and dependence on Notch2 and LTbR signaling. Ctrl 44 38 28 Gene Expression Profiles of Splenic CD11b+ DC Populations To confirm genetic differences between the two subsets of CD11b+ DCs, we compared genome-wide expression profiles CD11c of wild-type Esamhi and EsamloCD11b+ DC populations, along D Expt 22 7 2 with CD11b+ DCs from DC-Rbpj mice. No major differences in the expression of Notch2 or other pathway components were found by microarray or quantitative polymerase chain reac- Esam tion (PCR) (qPCR) (Figure S3), suggesting control by other factors such as ligand availability. Clustering analysis confirmed that RBPJ-deficient DCs were very similar to the Esamlo subset B Esamhi Esamlo (Figure 5A), as suggested by surface phenotype (Figure 2). Pair- hi lo *** wise comparison of Esam and Esam populations (Figure 5B; 0.7 0.4 0.6 Table S1) showed the expected differential expression of Esam 0.5 ** 0.3 Ctrl 0.4 0.2 and Clec12a. One of the most differentially expressed genes 0.3 Expt 0.2 0.1 was Gpr4, which encodes a G protein-coupled receptor and is Total (%) 0.1 hi 0 0 preferentially expressed in CD11c cDCs as revealed by a Flt3 Ltbr Irf4 Flt3Ltbr Irf4 Gpr4 locus-driven transgenic GFP reporter (Figure S3). This reporter confirmed the expression of Gpr4 in Esamhi but not Esamlo DCs (Figure S3). Finally, Esamhi DCs showed lower C D expression of cell cycle inhibitor Cdkn1a (p21), in accordance with their higher proliferation rate. Next, the Immgen database (Heng and Painter, 2008) was WT * used to derive expression signatures of splenic CD11b+CD4+ 0.15 23 ±6 and CD11bCD8+ DCs and of blood Ly6C+MHC II+ monocytes. 0.10 We then examined the distribution of these signature gene sets 0.05 among Esamhi and EsamloCD11b+ DC populations. As shown Total (%) CD11c 0 in Figure 5C, genes characteristic of all DCs were equally Esamhi Esamlo expressed by both populations, confirming that they represent Ltbr-/- bona fide DCs. In contrast, CD11b+ DC-specific genes including Ltbr-/- 9±2* WT the Notch target Dtx1 were preferentially expressed in the Esamhi subset. Conversely, monocyte-specific genes were pref- lo Esam erentially expressed in Esam cells, including such canonical myeloid genes as Ly6c, cytokine (Csf1r, Csf3r) and chemokine (Ccr2) receptors, and lysozyme (Lyz1, Lyz2). Notably, the expres- Figure 3. The Development of EsamhiCD11b+ DCs Requires LTbR Signaling sion of myeloid genes such as colony stimulating factor (CSF) lo (A) Splenic CD11b+Esamhi DC subset in experimental (Expt) Flt3/ or Ltbr/ receptors in Esam DCs appear significantly lower than in D hi animals or respective wild-type controls (Ctrl). The DC-Rbpj splenocytes that macrophages (Figure S3). Thus, the Esam population features lack Esamhi DCs (Figure 2D) are included for comparison. Shown are staining the unique expression profile of CD11b+ DCs, whereas Esamlo + hi hi profiles of gated B220 CD11b CD8 splenocytes, with the CD11c Esam population appears closer to but distinct from monocytes or DC population highlighted. macrophages. (B) The fraction of Esamhi and Esamlo DC subsets among total splenocytes from Flt3/, Ltbr/, or DC-Irf4D experimental (Expt) animals or controls (Ctrl). By using red fluorescent protein (RFP) reporter driven by Ccr2 hi Significance is indicated as in Figure 1 (mean ± SD of four to five animals). locus (Saederup et al., 2010), we found that Esam DCs were the (C) Splenic CD11b+Esamhi DC population among gated wild-type competitor only splenic DC population with low RFP expression, while (WT) or Ltbr/ donor populations in competitively reconstituted BM chimeras, Esamlo DCs and the majority of CD8+ DCs were Ccr2-RFPhi shown as in (A) (mean ± SD of three animals). (Figure 5D). To confirm the differential relationship of Esamhi / + (D) The fraction of WT competitor- or Ltbr donor-derived CD11b DC and Esamlo DCs to myeloid cells, we used lineage tracing with subsets among total splenocytes in the BM chimeras (mean ± SD of three a myeloid-specific Lyz2-cre (also known as LysM-cre) deleter animals). strain (Jakubzick et al., 2008; Liu et al., 2009). In Lyz2-cre mice crossed to a cre-inducible yellow fluorescent protein (YFP) After in vivo pulse with nucleoside analog 5-bromo-20-deoxyuri- reporter, splenic Esamhi DCs as well as CD8+ DCs harbored dine (BrdU), EsamhiCD11b+ DC incorporated more BrdU than only a small (5%–7%) fraction of recombined YFP+ cells (Fig- Esamlo DCs (Figure 4D), revealing a faster turnover. Although ure 5E). In contrast, the Esamlo DCs showed substantial recom- the contribution of progenitors cannot be ruled out, the short bination (30%) that was nevertheless lower than in monocytes

784 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

A B Figure 4. Phenotypic Analysis of the Splenic CD11b+ DC Subsets Unless indicated otherwise, Cx3cr1-GFP reporter mice were used to separate the EsamhiGFPlo and EsamloGFPhi subsets of CD11chiCD11b+CD8 splenic DCs. (A) Expression of the indicated surface markers in the hi lo lo hi Relative cell no. Relative gated Esam GFP and Esam GFP subsets. Dashed CD11c CD4 Dcir2 LT R lines indicate positive staining threshold. (B) Expression of LTbR on the indicated DC subsets from lo wild-type mice.

Relative cell no. Relative Esam Esamhi (C) Microphotographs of the sorted DC subsets stained by Giemsa stain on cytospin preparations. The CD11c + CD8 CD11b+F4/80+ Cx3cr1-GFP+ macrophages (MF) are shown for comparison (magniﬁcation, 4003). (D) Histograms of BrdU staining in the gated DC subsets Flt3 Clec12a CD11b after 2 days of in vivo BrdU pulse. The percentages of + Esamlo Esamhi BrdU cells are indicated; representative of four animals. (E) Cell cycle proﬁles of DC subsets after 2 hr of in vivo BrdU pulse. Shown are gated DC subsets stained for DNA C Esamhi Esamlo M content (DAPI) and BrdU incorporation; the fractions of cells in S and G2/M phases are indicated (representative of two animals). See also Figure S3.

ure 6A; Table S1). The expression of these genes in Esamlo DCs was nevertheless lower than in macrophages (Figure 6B). We therefore tested the ability of CD11b+ DC subsets to produce cytokines in response to TLR ligands. Esamlo DCs efﬁciently produced tumor a Esamlo Esamhi CD8+ necrosis factor alpha (TNF- ) and IL-12 in D E response to the TLR9 agonist CpG DNA, in 40 contrast to the weak production by Esamhi 10 + 1.8 3.7 2.4 DCs (Figure 6C). CD11c CD11b monocytes and macrophages produced more TNF-a but lo

BrdU 0.9 1.7 0.8 less IL-12 than Esam DCs, underscoring the Relative cell no. Relative distinct functionality of the latter population. A lo BrdU DAPI (linear) similarly superior cytokine production by Esam hi Esamlo Esamhi over Esam DCs was observed in response to TLR2 agonist heat-killed L. monocytogenes (Figure 6D) and TLR7 agonist gardiquimod and macrophages (45%). These data show that the Esamhi (data not shown). No expression of interferon-b or interleukin- DCs are uniquely low for Ccr2, have no developmental history 10 by either DC population could be detected under these condi- of Lyz2 expression, and therefore are not derived from the tions (data not shown). Furthermore, Esamlo DCs showed Esamlo population or from monocytes. stronger TLR-mediated induction of costimulatory molecules The differences in signaling requirements, gene expression than Esamhi DCs (Figure 6E), suggesting a higher overall sensi- profile, and Lyz2 expression history suggest that the Esamhi tivity to TLR ligands. and Esamlo population may arise from different precursors. Both subsets of CD11b+ DCs showed efficient phagocytosis Indeed, adoptively transferred Cx3cr1-GFP+ MDPs gave rise to of in vivo injected polystyrene beads (Figure S4). In an in vitro both GFPhi and GFPlo splenic CD11b+ DC populations, whereas Ag presentation assay, both Esamlo and Esamhi DCs sorted CDPs gave rise almost exclusively to GFPloCD11b+ DCs corre- from ovalbumin (OVA)-pulsed mice efficiently primed naive sponding to the Esamhi subset (Figure S3). These data suggest OVA-specific CD4+ OT-II T cells (Figure S4). Furthermore, that the GFPloEsamhi subset represents the canonical CDP- splenic DCs from DC-RbpjD mice elicited strong OVA-specific derived splenic DC population; conversely, GFPhi DCs probably OT-II proliferation in vitro, despite the lack of Esamhi DC subset. arise from different myeloid progenitors. These data suggest that Esamhi and Esamlo DCs are equally capable of Ag-specific T cell priming in vitro on a per cell basis. Functional Analysis of Splenic CD11b+ DC Populations To test the role of Esamhi DCs in T cell priming in vivo, we adop- The Esamlo DC subset showed preferential expression of tively transferred carboxyfluorescein diacetate succinimidyl multiple genes associated with TLR-mediated pathogen ester (CFSE)-labeled OT-II T cells into DC-RbpjD mice and immu- sensing, including several TLRs, MyD88, CD14, and CD36 (Fig- nized with OVA. Three days after immunization, OT-II T cells

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 785 Immunity Notch2 Controls DC Differentiation

A Esamhi Figure 5. Gene Expression Profiles of Splenic CD11b+ DC Subsets lo Esam (A) The comparison of control and RBPJ-deficient CD11b+ RBPJ DCs. Splenic CD11b+ DCs from Cx3cr1-GFP+ DC-RbpjD 10.80.60.40.20 mice were sorted along with the EsamhiGFPlo and Esamlo Relative distance GFPhi subsets of CD11b+ splenic DCs from control Cx3cr1-GFP+ mice. Shown is unsupervised clustering B analysis of the resulting microarray expression profiles. D (B) Pairwise comparison of the Esamhi and Esamlo 2.6% 4 subsets. The probes overexpressed >2-fold in the Cdkn1a respective populations are shown in green and red, and 3.5 + their percentage among the total probes is indicated. Clec12a CD8 3 T cells Probes for select relevant genes are highlighted in blue. (C) The expression of signature probe sets in the Esamhi 2.5 lo

log intensity log and Esam subsets. The signature sets of total DCs, lo Gpr4 + 2 CD11b DCs and monocytes were derived from Immgen Esam + database and their distribution among the Esamhi and

Esam CD8 DC 1.5 lo 1.4% Esam subsets is depicted as in (B). (D) The expression of Ccr2-RFP reporter in DC subsets. 1.523 2.5 3.5 4 Shown are histograms of RFP expression in the indicated hi - Esam log intensity Clec12a splenic populations including CD11c CD11b+ monocytes CD11b+ DC F + C and macrophages (Mo-M ), and Clec12a and Clec12a CD11b+ DCs corresponding to Esamhi and Esamlo 5.3% cell no. Relative subsets, respectively. 3.5 Clec12a+ (E) Efﬁciency of Lyz2-cre recombination in DC subsets. CD11b+ DC Shown are histograms of YFP expression in the indicated 3 splenic populations of Lyz2-cre+ mice with cre-inducible DC- YFP reporter allele. specific 2.5 See also Table S1. Mo-M

2 6.1% RFP Notch2 Is Required for the Development 232.5 3.5 of Intestinal CD11b+CD103+ DCs 0% E Given the tissue-speciﬁc nature of Notch 3.5 signaling, we examined the role of Notch2 in 2 DC differentiation outside of the spleen. We 3 CD8+ D CD11b+ found that DC-Notch2 mice had normal Dtx1 T cells + + log intensity log DC-specific numbers of CD11b and CD103 DCs and Lang-

lo 2.5 erhans cells in the skin and/or skin-draining LN + +

Esam 2 (Figure S5). Similarly, CD11b and CD103 DCs 7 CD8+ DC in the lung and liver were unaffected by Notch2 1.5 19.8% deletion (Figure S5). However, analysis of DCs 1.5 232.5 3.5 in the intestinal LP revealed selective depletion 5 + + Lyz2 Esamhi of the CD11b CD103 subset (Figures 7A and 39.7% CD11b+ DC 7B). In contrast, the monocyte-derived CD11b+- 3.5 CD103 population was unaffected, whereas the + 3 Relative cell no. CD103 population was moderately increased. Ccr2 29 Esamlo + + monocyte- Similarly, the fraction of CD11b CD103 DCs CD11b+ DC 2.5 specific migrating from the LP was selectively reduced in the mesenteric LN (Figure 7C). The LP 2 CD11b+CD103+ DCs did not express Esam or 43 Mo-M D 1.5 0.6% Dtx1 and were not reduced in DC-RBPJ or DC-DNMAML1 mice (data not shown), suggest- 1.5 232.5 3.5 ing that the Notch2 signal is transient or qualita- Esamhi log intensity YFP tively different in this subset. However, the expression of the Gpr4 reporter was detected in showed extensive proliferation (as measured by CFSE dye CD11b+CD103+ but not in CD11b+CD103 DCs (Figure S3). This dilution) in the spleens of control mice, but only minor prolifera- selective expression was similar to Gpr4 expression in Esamhi tion in the spleens of DC-RbpjD mice (Figure 6F). Because DC- but not Esamlo splenic DCs, illustrating the similarity between RbpjD spleens speciﬁcally lack Esamhi DCs, we conclude that Notch2-dependent DC subsets in the spleen and intestine. this subset is required for optimal T cell priming in the spleen To test the functional consequences of CD11b+CD103+ in vivo. DC depletion, we examined intestinal T cell populations in

786 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

A 6 DE Esamhi 5 Esamlo 0.8 269 1 168 4 Water Water 3

Relative expression 1 16 850 4 364 0 Dtx1 TLR9 MyD88 CD14 CpG CpG Relative cell no. Relative Relative cell no. Relative B 1000 Esamhi 9 1351 lo Esam 3 541 100 M HKLM HKLM

TNF- CD40 1 Relative expression Esamlo Esamlo 0.1 Esamhi Esamhi Tlr2 CD36 CD14

C 7.7 0.1 F 0.6 1.3 0.3 1.7 Water 27 36 37 3 23 74

58 28 27 46 Relative cell no. 10 18 CpG Relative cell no.

CFSE IL-12 TNF- Ctrl Mo-M Esamlo Esamhi DC-Rbpj

Figure 6. Functional Properties of Splenic CD11b+ DC Subsets (A) The expression of indicated genes in GFPloEsamhi and GFPhiEsamlo subsets of splenic CD11b+ DCs sorted from Cx3cr1-GFP mice. Data represent normalized expression values relative to the Esamhi sample as determined by qPCR (mean ± SD of triplicate reactions). (B) The expression of indicated genes in the same DC subsets as well as in CD11cCD11b+F4/80+GFP+ macrophages (MF), determined by qPCR as in (A) and shown on a log scale. (C) Cytokine production by CD11b+ DCs in response to TLR9 ligand CpG. Lymphocyte-depleted splenocytes from Cx3cr1-GFP mice were incubated for 6 hr with CpG and stained for surface markers and intracellular TNF-a or IL-12. Shown are cytokine expression profiles in GFPloEsamhi or GFPhiEsamlo subsets of CD11chiCD11b+ DCs or in CD11cCD11b+GFPhi monocytes and macrophages (Mo-MF). Data are representative of three independent experiments. (D) TNF-a production by CD11b+ DCs in response to TLR ligands. Splenocytes from Cx3cr1-GFP mice were activated with CpG or TLR2 ligand heat-killed L. monocytogenes (HKLM) and stained as above. (E) Phenotypic maturation of CD11b+ DCs in vitro. Splenocytes from Cx3cr1-GFP mice were activated with CpG or HKLM and stained for DC markers as above. Shown are staining profiles of CD11b+ DC subsets for CD40, with mean fluorescence intensities indicated. (F) In vivo CD4+ T cell priming in the absence of Esamhi DCs. DC-RbpjD or littermate control (Ctrl) mice were administered CFSE-labeled OVA-specific OT-II T cells and immunized with OVA. Shown are CFSE staining profiles of gated OT-II T cells from the recipient spleens 3 days after immunization. The fractions of T cells that did not divide or divided 1–3 and >3 times are indicated (mean of two recipients). See also Figure S4.

DC-Notch2D mice. Whereas the fraction of regulatory CD4+ T (Figure 7D). Similar results were observed in the small intestine (Treg) cells expressing transcription factor FoxP3 was un- (Figure 7D) and colon (not shown). Thus, Notch2 provides a changed, the IL-17-producing Th17 cells were reduced 50% tissue-speciﬁc differentiation signal for the CD11b+CD103+

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 787 Immunity Notch2 Controls DC Differentiation

A B 30 4 C *** ) 25 18 ±12 3 3 18 ±3 20 15 2 11 Ctrl 10 Ctrl 15 1 ±4 Cells (10 ±12 5 0 0 81 ±14 CD11b+ CD11b+ 50 ±14 CD103- CD103+

CD103 3±1* 11 ±4** CD103

CD103 12 ) * 3 10 23 DC- B 8 Cre- DC- 38 ±4*** Notch2 6 Notch2 ±12* + 4 Cre Cells (10 57 ±13* 2 39 ±20 0 CD11b CD11b F4/80 CD11b- CD103+

D DCs Ex vivo CD4+ T cells Activated CD4+ T cells

18 Ctrl 45 29 10 8 26

32 IL-17 CD103 TCR

DC- 1.4 50 17 Notch2 11 13 12

CD11b FoxP3 IL-17

Figure 7. Notch2 Controls the Development of Intestinal CD11b+CD103+ DCs (A) DC populations in the intestinal lamina propria from DC-Notch2D mice and littermate controls. Shown are staining profiles of CD45+CD11chiMHC II+ DCs and of the gated CD11b+ DCs, highlighting the CD11b+CD103+ DC subset. Fractions represent average ± SD of four animals per group. Statistical significance is indicated as in Figure 1. (B) Absolute numbers of the indicated DC subsets (mean ± SD of four animals). (C) Staining profiles of CD11chiMHC II+ DCs from the mesenteric lymph nodes. Fractions represent average ± SD of seven animals per group. (D) Effector T cell populations in the intestine. Lymphocytes were isolated from the LP of small intestine and stained ex vivo for intracellular FoxP3 or activated in vitro with PMA plus ionomycin and stained for intracellular IL-17. Shown are profiles of CD11chiMHC II+ LP DCs and of gated TCRb+CD4+ T cells isolated ex vivo or activated in vitro (two pairs of littermates shown; representative of four DC-Notch2D animals). See also Figure S5.

DC population, which in turn support optimal effector T cell Canonical Notch signaling occurs through the NICD-RBPJ- differentiation in the intestine. MAML transcriptional activation complex, and Rbpj deletion (Han et al., 2002) or DNMAML1-mediated complex disruption DISCUSSION (Maillard et al., 2004) recapitulates Notch receptor deletion in lymphocytes. Indeed, CD11b+ splenic DCs were equally Although Notch signaling has been indirectly implicated in affected by the loss of Notch2 or RBPJ or by DNMAML expres- various aspects of DC development or function (Caton et al., sion. However, Notch2 deletion impaired the development of 2007; Cheng et al., 2003; Feng et al., 2010; Zhou et al., 2009), splenic CD8+ DCs and intestinal CD11b+CD103+ DCs, whereas the specific Notch receptor involved and DC populations Rbpj deletion spared both subsets (Caton et al., 2007; data not affected were unclear. Here we have demonstrated that Notch2 shown). Although Notch2 signaling in these DC subsets may receptor controls the development of specific DC subsets in the include a noncanonical RBPJ-independent pathway, this dis- spleen and intestine. In contrast, Notch1 was dispensable for DC crepancy probably reflects technical aspects of gene targeting development, consistent with previous results of global inducible in DCs. Because DCs have a rapid turnover and Itgax-cre Notch1 deletion (Radtke et al., 2000). Similarly, Notch4 defi- recombination commences after DC commitment, delayed ciency did not affect DC populations (data not shown), establish- recombination kinetics and/or increased protein perdurance of ing Notch2 as a nonredundant Notch receptor regulating the DC RBPJ versus Notch2 would mitigate the phenotype. Indeed, lineage development. continuous global deletion of RBPJ caused apparent loss of

788 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

CD8+ DCs (Feng et al., 2010), suggesting the predominant role of over of RBPJ-deficient CD11b+ DCs (Caton et al., 2007) was canonical Notch2 signaling in DC development. based on prolonged BrdU pulse that aggregated proliferation Notch signaling can provide the initial commitment signal as with cell death and migration. Most importantly, the higher prolif- well as sustain further differentiation steps and/or homeostasis eration rate of Esamhi DCs is consistent with their dependence of a given cell type (Radtke et al., 2010; Yashiro-Ohtani et al., on Flt3 and LTbR signaling, both of which promote DC prolifera- 2010). In splenic DCs, Notch2 deletion affected both DC subsets tion in situ (Kabashima et al., 2005; Waskow et al., 2008). and caused accumulation of immature CD11bCD8 cells. In All blood-derived DCs in lymphoid organs are thought to addition, mature CD11b+ DCs demonstrate ongoing Notch originate from CDPs in the BM via the committed pre-DCs signaling activity, as evidenced by RBPJ-dependent expression entering from the blood and differentiating in situ. Indeed, the of Dtx1 and sensitivity to Notch2-RBPJ loss or DNMAML1 induc- EsamhiCx3cr1loCD11b+ splenic DCs apparently arise in this tion. In contrast, CD8+ DCs do not express Dtx1 and are less pathway because they can be derived from CDPs and lack affected by Rbpj deletion or DNMAML1 induction. Thus, Notch2 developmental history of myeloid-specific Lyz2 gene expres- signaling appears to initiate and maintain the differentiation of sion. In contrast, the EsamloCx3cr1hi subset could not be derived splenic CD11b+ DC subset, starting at the common pre-DC from transferred CDPs and preferentially expressed myeloid stage and thereby promoting CD8+ DC development as well. In genes, although at much lower level than monocytes or macro- tissues such as the intestine, Notch signaling in CD11b+CD103+ phages. Because splenic DCs cannot be derived from mono- DCs may commence after the pre-DC stage and would not affect cytes (Varol et al., 2007), the Esamlo DCs probably arise from CD103+ DCs. earlier progenitors such as MDPs. Splenic DC development is Our data illustrate a tissue-specific role of Notch signaling in not dependent on myeloid cytokine receptors Csf1r (Ginhoux DC differentiation in the spleen and intestinal LP, in contrast to et al., 2009) or Csf2r (our unpublished data), consistent with its proposed general role in DC development (Cheng et al., the lower expression of these genes in Esamlo DCs relative to 2003; Zhou et al., 2009). Indeed, both tissues are classical sites macrophages (Figure S3). In any case, our results suggest that of Notch activity, which drives the respective differentiation of the spleen is similar to nonlymphoid organs in the dual origin of splenic MZ B cells and of the intestinal epithelium (Wilson and its resident DCs from the CDPs and from a parallel myelomono- Radtke, 2006). MZ B cell development is mediated solely by cytic pathway. Notch ligand Dll1 (Hozumi et al., 2004), whereas the latter In the spleen, the absence of Esamhi DCs after DC-specific involves both Dll1 and Dll4 (Pellegrinet et al., 2011). Because Rbpj deletion impaired CD4+ T cell priming. This was not both Dll1 and Dll4 are also expressed in the splenic red pulp because of higher Ag uptake or presentation capacity of Esamhi and MZ (Tan et al., 2009), these two ligands may redundantly DCs on a per cell basis; therefore, Esamhi DCs may be uniquely regulate Notch2-mediated DC development both in the spleen suited for T cell priming because of their specific localization in and intestine. A notable molecular feature of the splenic MZ is the MZ and/or migration properties. In the intestinal LP, the active LTbR signaling, which is required both for MZ formation CD11b+CD103+ cells represent a unique DC population that and MZ B cell maintenance, probably in a cell-extrinsic manner migrates to the mesenteric LN in the steady state and upon (Mebius and Kraal, 2005). We found that both Notch2 and inflammation, suggesting its likely role in T cell priming (Bogu- LTbR signals control the development of CD11b+ splenic DCs. novic et al., 2009). We found that the loss of this population after The contribution of LTbR was at least in part intrinsic to the DC-specific Notch2 targeting was associated with a reduced hematopoietic compartment, consistent with previous studies fraction of Th17 cells, a major effector CD4+ T cell population on LTbR in DC development (Kabashima et al., 2005) and func- in the intestine. Although this finding remains to be confirmed tion (Summers deLuca et al., 2011). Our results establish these in mice with pure genetic background and more defined micro- two pathways as common regulators of the splenic MZ and flora, it is consistent with the recently reported capacity of of its two major immune constituents, the MZ B cells and CD11b+CD103+ DCs to induce Th17 cell differentiation in vitro CD11b+ DCs. (Denning et al., 2011). Thus, DC-specific Notch2 targeting The CD11b+ and CD8+ DC populations in the lymphoid organs provided the first insight into the specific function of CD11b+ of naive animals were thought to be genetically and functionally CD103+ DC subset in vivo, opening the way for future studies homogeneous. However, it has been demonstrated recently of immune responses and tolerance in the intestine. Collectively, that CD8+ DCs are comprised of two distinct populations, the our results establish Notch2 as a common regulator of the two Cx3cr1-GFP classical CD8+ DCs and Cx3cr1-GFPhi cells key T cell-priming DC populations in the spleen and the intestine. affiliated with the pDC lineage (Bar-On et al., 2010). Here we combined genetic analysis with the Cx3cr1-GFP reporter and EXPERIMENTAL PROCEDURES surface marker Esam to similarly subdivide the splenic CD11b+ b DCs. In particular, we found that Notch or LT R signaling Animals hi blockade causes a complete ablation of the Esam DC subset, The Itgax-cre hemizygous transgene was used to generate the following rather than a partial decrease of all splenic CD11b+ DCs. The experimental animals on C57BL/6 (B6) background unless indicated other- D D D Notch2-dependent Esamhi population exhibited properties of wise: DC-Rbpj (Rbpjflox/flox), DC-Notch1 (Notch1flox/flox), DC-Notch2 flox/flox D flox/flox classical DCs, including cytoplasmic ‘‘veils’’ and rapid turnover. (Notch2 , mixed B6/129), DC-Irf4 (Irf4 ), DC-DNMAML1 (Rosa26- + + StopFlox-DNMAML1, B6 or B6129F1), and DC-NICD (Eef1a1-StopFlox- Although the CD4 subset of CD11b DCs was reported to incor- D NICD1, mixed B6/129). Where indicated, DC-Rbpj mice carried heterozygous porate BrdU at a slower rate (Kamath et al., 2000), Esam is Cx3cr1-GFP reporter allele. Because the Itgax-cre transgene does not affect + expressed only in a fraction of CD4 DCs and thus provides DC numbers or phenotype (Caton et al., 2007), Cre-negative littermates a more precise subset definition. Furthermore, the higher turn- were used as controls for the respective Itgax-cre+ animals. For the analysis

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 789 Immunity Notch2 Controls DC Differentiation

of intestinal T cells, cohoused sex-matched littermates were used. Germline ACCESSION NUMBERS null mutants of Flt3 and LTbR on B6 background, as well as competitive chimeras established from LTbR-null BM have been described (Summers The microarray data are available in the Gene Expression Omnibus (GEO) deLuca et al., 2011; Waskow et al., 2008). Chimeras were analyzed 11 weeks database (http://www.ncbi.nlm.nih.gov/gds) under the accession number after reconstitution and the donor- or competitor-derived cells were distin- GSE31551. guished by differential CD45 isoform expression. Homo- or heterozygous Cx3cr1-GFP and heterozygous Ccr2-RFP reporter mice were on the B6 back- SUPPLEMENTAL INFORMATION ground. The Lyz2-cre lineage tracing was performed as described (Jakubzick et al., 2008) and utilized mice hemizygous for Lyz2-cre and heterozygous for Supplemental Information includes Supplemental Experimental Procedures, Rosa26-StopFlox-EYFP. All animal studies were performed according to the five figures, and one table and can be found with this article online at doi:10. investigator’s protocol approved by the Institutional Animal Care and Use 1016/j.immuni.2011.08.013. Committee of Columbia University. ACKNOWLEDGMENTS Cell Analysis Cells were isolated ex vivo from lymphoid organs and stained for cell surface We thank U. Klein, A. Efstratiadis, F. Radtke, T. Gridley, and W. Pear for mouse markers as described (Caton et al., 2007; Liu et al., 2009). The antibodies used strains; A. Capobianco, D. Lin, and A. Ferrante for animal donations; M. La- for staining are listed in Supplemental Experimental Procedures. The isolation houd for antibodies; and J. Ericson for Immgen data. Supported by the Amer- and analysis of DCs from the intestinal LP and other tissues was done as in ican Asthma Foundation (B.R.), NIH grants AI072571 (B.R.) and DK085329 Bogunovic et al. (2009) and Ginhoux et al. (2009). The isolation and analysis (I.I.I.), NIH training grants AI007161 (K.L.L.) and HD055165 (M.L.C.), and of intestinal T cells was done as described (Ivanov et al., 2006). Fluorochrome- CIHR/IRSC award MOP #67157 (J.L.G.). or biotin-conjugated antibodies were from eBioscience or BD Biosciences. Biotinylated anti-Clec12a was kindly provided by M. Lahoud at the Walter Received: May 7, 2011 and Eliza Hall Institute. The samples were acquired on a LSR II flow cytometer Revised: July 28, 2011 or sorted on FACSAria flow sorter (BD Immunocytometry Systems) and Accepted: August 30, 2011 analyzed with FlowJo software (Treestar Inc.). Cytospin preparations were Published online: October 20, 2011 done on Cytospin 3 centrifuge (Thermo Shandon) and stained by Giemsa stain. For BrdU incorporation, Cx3cr1-GFP reporter mice were either injected i.p. REFERENCES with 1.5 mg BrdU and analyzed 2 hr later or injected with 1.5 mg BrdU for 2 consecutive days and analyzed 48 hr after the first injection. BrdU staining Bar-On, L., Birnberg, T., Lewis, K.L., Edelson, B.T., Bruder, D., Hildner, K., was performed with APC BrdU Flow Kit (BD Biosciences) and combined Buer, J., Murphy, K.M., Reizis, B., and Jung, S. (2010). CX3CR1+ with DAPI staining for 2 hr pulse. To measure cytokine production, spleno- CD8alpha+ dendritic cells are a steady-state population related to plasmacy- cytes from Cx3cr1-GFP mice were depleted of lymphoid and erythroid toid dendritic cells. Proc. Natl. Acad. Sci. USA 107, 14745–14750. cells by negative selection on MACS columns (Miltenyi). The DC-enriched Barnden, M.J., Allison, J., Heath, W.R., and Carbone, F.R. (1998). Defective flow-through fraction was incubated in serum-containing medium at TCR expression in transgenic mice constructed using cDNA-based alpha- 8 37 C with 100 nM CpG-B oligonucleotides (ODN1826) or 10 heat-killed and beta-chain genes under the control of heterologous regulatory elements. m L. monocytogenes (Invivogen). After 2 hr, Brefeldin A (10 g/ml) was added Immunol. Cell Biol. 76, 34–40. and the cells were incubated for an additional 4 hr. Cells were stained for Bogunovic, M., Ginhoux, F., Helft, J., Shang, L., Hashimoto, D., Greter, M., Liu, surface markers, fixed, permeabilized, and stained with APC-conjugated anti- K., Jakubzick, C., Ingersoll, M.A., Leboeuf, M., et al. (2009). Origin of the lamina bodies to IL-12 or TNF-a (BD Biosciences) according to the manufacturer’s propria dendritic cell network. Immunity 31, 513–525. instructions. Bray, S.J. (2006). Notch signalling: a simple pathway becomes complex. Nat. Gene Expression Analysis Rev. Mol. Cell Biol. 7, 678–689. For microarray analysis, spleens from Cx3cr1-GFP+ DC-RbpjD mice or Cre- Buonamici, S., Trimarchi, T., Ruocco, M.G., Reavie, L., Cathelin, S., Mar, B.G., negative littermate controls were isolated and depleted of lymphoid and Klinakis, A., Lukyanov, Y., Tseng, J.C., Sen, F., et al. (2009). CCR7 signalling as erythroid cells by negative selection on MACS columns. Total CD11chiCD11b+ an essential regulator of CNS infiltration in T-cell leukaemia. Nature 459, 1000– DCs from DC-RbpjD spleens (uniformly EsamloGFPhi, see Figure 2) or Esamlo 1004. GFPhi and EsamhiGFPlo subsets thereof from control mice were sorted and Caton, M.L., Smith-Raska, M.R., and Reizis, B. (2007). Notch-RBP-J signaling total RNA was isolated with RNeasy columns (QIAGEN). Precipitated RNA controls the homeostasis of CD8- dendritic cells in the spleen. J. Exp. Med. was reverse-transcribed, labeled, and hybridized to GeneChip Mouse Gene 204, 1653–1664. 1.0 ST Array (Affymetrix) according to the manufacturer’s instructions. Cheng, P., Nefedova, Y., Miele, L., Osborne, B.A., and Gabrilovich, D. (2003). RMA-normalized gene expression values were analyzed with NIA Array soft- Notch signaling is necessary but not sufficient for differentiation of dendritic ware (Sharov et al., 2005). The signatures of normal DC populations were cells. Blood 102, 3980–3988. derived by clustering analysis of Immgen subsets DC.4+.Sp, DC.8+.Sp, and De Trez, C., Schneider, K., Potter, K., Droin, N., Fulton, J., Norris, P.S., Ha, Mo.6c+II+.Bl (Heng and Painter, 2008) and are available from the authors S.W., Fu, Y.X., Murphy, T., Murphy, K.M., et al. (2008). The inhibitory HVEM- upon request. Real-time qPCR was performed as described (Caton et al., BTLA pathway counter regulates lymphotoxin receptor signaling to achieve 2007). homeostasis of dendritic cells. J. Immunol. 180, 238–248.

T Cell Proliferation Assay Denning, T.L., Norris, B.A., Medina-Contreras, O., Manicassamy, S., Geem, Splenic and LN CD4+ T cells from OVA-specific OT-II transgenic mice (Barn- D., Madan, R., Karp, C.L., and Pulendran, B. (2011). Functional specializations den et al., 1998) were enriched by negative selection with MACS columns, of intestinal dendritic cell and macrophage subsets that control Th17 and regu- labeled with CFSE, and injected intravenously (i.v.) (2 3 106 per animal) into latory T cell responses are dependent on the T cell/APC ratio, source of mouse DC-RbpjD or control mice. After 24 hr, recipient mice received 10 mg OVA strain, and regional localization. J. Immunol. 187, 733–747. (Sigma) i.v. and were analyzed 72 hr later. Transferred T cells were identified Dudziak, D., Kamphorst, A.O., Heidkamp, G.F., Buchholz, V.R., Trumpfheller, as CD4+ TCR Va2+ TCR Vb5+. C., Yamazaki, S., Cheong, C., Liu, K., Lee, H.W., Park, C.G., et al. (2007). Differential antigen processing by dendritic cell subsets in vivo. Science 315, Statistical Analysis 107–111. Statistical significance was estimated with unpaired, two-tailed Student’s Feng, F., Wang, Y.C., Hu, X.B., Liu, X.W., Ji, G., Chen, Y.R., Wang, L., He, F., t test. Dou, G.R., Liang, L., et al. (2010). The transcription factor RBP-J-mediated

790 Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. Immunity Notch2 Controls DC Differentiation

signaling is essential for dendritic cells to evoke efficient anti-tumor immune Pellegrinet, L., Rodilla, V., Liu, Z., Chen, S., Koch, U., Espinosa, L., Kaestner, responses in mice. Mol. Cancer 9, 90. K.H., Kopan, R., Lewis, J., and Radtke, F. (2011). Dll1- and dll4-mediated Fogg, D.K., Sibon, C., Miled, C., Jung, S., Aucouturier, P., Littman, D.R., notch signaling are required for homeostasis of intestinal stem cells. Cumano, A., and Geissmann, F. (2006). A clonogenic bone marrow progenitor Gastroenterology 140, 1230–1240, e1–e7. specific for macrophages and dendritic cells. Science 311, 83–87. Radtke, F., Wilson, A., Stark, G., Bauer, M., van Meerwijk, J., MacDonald, H.R., Ginhoux, F., Liu, K., Helft, J., Bogunovic, M., Greter, M., Hashimoto, D., Price, and Aguet, M. (1999). Deficient T cell fate specification in mice with an induced J., Yin, N., Bromberg, J., Lira, S.A., et al. (2009). The origin and development of inactivation of Notch1. Immunity 10, 547–558. nonlymphoid tissue CD103+ DCs. J. Exp. Med. 206, 3115–3130. Radtke, F., Ferrero, I., Wilson, A., Lees, R., Aguet, M., and MacDonald, H.R. Han, H., Tanigaki, K., Yamamoto, N., Kuroda, K., Yoshimoto, M., Nakahata, T., (2000). Notch1 deficiency dissociates the intrathymic development of dendritic Ikuta, K., and Honjo, T. (2002). Inducible gene knockout of transcription factor cells and T cells. J. Exp. Med. 191, 1085–1094. recombination signal binding protein-J reveals its essential role in T versus B Radtke, F., Fasnacht, N., and Macdonald, H.R. (2010). Notch signaling in the lineage decision. Int. Immunol. 14, 637–645. immune system. Immunity 32, 14–27. Heng, T.S., and Painter, M.W.; Immunological Genome Project Consortium. Saederup, N., Cardona, A.E., Croft, K., Mizutani, M., Cotleur, A.C., Tsou, C.L., (2008). The Immunological Genome Project: networks of gene expression in Ransohoff, R.M., and Charo, I.F. (2010). Selective chemokine receptor usage immune cells. Nat. Immunol. 9, 1091–1094. by central nervous system myeloid cells in CCR2-red fluorescent protein Hozumi, K., Negishi, N., Suzuki, D., Abe, N., Sotomaru, Y., Tamaoki, N., knock-in mice. PLoS ONE 5, e13693. Mailhos, C., Ish-Horowicz, D., Habu, S., and Owen, M.J. (2004). Delta-like 1 Schulz, O., Jaensson, E., Persson, E.K., Liu, X., Worbs, T., Agace, W.W., and is necessary for the generation of marginal zone B cells but not T cells in vivo. Pabst, O. (2009). Intestinal CD103+, but not CX3CR1+, antigen sampling cells Nat. Immunol. 5, 638–644. migrate in lymph and serve classical dendritic cell functions. J. Exp. Med. 206, Ivanov, I.I., McKenzie, B.S., Zhou, L., Tadokoro, C.E., Lepelley, A., Lafaille, 3101–3114. J.J., Cua, D.J., and Littman, D.R. (2006). The orphan nuclear receptor Sharov, A.A., Dudekula, D.B., and Ko, M.S. (2005). A web-based tool for prin- RORgammat directs the differentiation program of proinflammatory IL-17+ T cipal component and significance analysis of microarray data. Bioinformatics helper cells. Cell 126, 1121–1133. 21, 2548–2549. Jakubzick, C., Bogunovic, M., Bonito, A.J., Kuan, E.L., Merad, M., and Shortman, K., and Heath, W.R. (2010). The CD8+ dendritic cell subset. Randolph, G.J. (2008). Lymph-migrating, tissue-derived dendritic cells are minor Immunol. Rev. 234, 18–31. constituents within steady-state lymph nodes. J. Exp. Med. 205, 2839–2850. Steinman, R.M., and Idoyaga, J. (2010). Features of the dendritic cell lineage. Kabashima, K., Banks, T.A., Ansel, K.M., Lu, T.T., Ware, C.F., and Cyster, J.G. Immunol. Rev. 234, 5–17. (2005). Intrinsic lymphotoxin-beta receptor requirement for homeostasis of Summers deLuca, L., Ng, D., Gao, Y., Wortzman, M.E., Watts, T.H., and lymphoid tissue dendritic cells. Immunity 22, 439–450. Gommerman, J.L. (2011). LTbR signaling in dendritic cells induces a type I Kamath, A.T., Pooley, J., O’Keeffe, M.A., Vremec, D., Zhan, Y., Lew, A.M., IFN response that is required for optimal clonal expansion of CD8+ T cells. D’Amico, A., Wu, L., Tough, D.F., and Shortman, K. (2000). The development, Proc. Natl. Acad. Sci. USA 108, 2046–2051. maturation, and turnover rate of mouse spleen dendritic cell populations. Tan, J.B., Xu, K., Cretegny, K., Visan, I., Yuan, J.S., Egan, S.E., and Guidos, J. Immunol. 165, 6762–6770. C.J. (2009). Lunatic and manic fringe cooperatively enhance marginal zone Klein, U., Casola, S., Cattoretti, G., Shen, Q., Lia, M., Mo, T., Ludwig, T., B cell precursor competition for delta-like 1 in splenic endothelial niches. Rajewsky, K., and Dalla-Favera, R. (2006). Transcription factor IRF4 controls Immunity 30, 254–263. plasma cell differentiation and class-switch recombination. Nat. Immunol. 7, Tu, L., Fang, T.C., Artis, D., Shestova, O., Pross, S.E., Maillard, I., and Pear, 773–782. W.S. (2005). Notch signaling is an important regulator of type 2 immunity. Lahoud, M.H., Proietto, A.I., Ahmet, F., Kitsoulis, S., Eidsmo, L., Wu, L., Sathe, J. Exp. Med. 202, 1037–1042. P., Pietersz, S., Chang, H.W., Walker, I.D., et al. (2009). The C-type lectin Varol, C., Landsman, L., Fogg, D.K., Greenshtein, L., Gildor, B., Margalit, R., Clec12A present on mouse and human dendritic cells can serve as a target Kalchenko, V., Geissmann, F., and Jung, S. (2007). Monocytes give rise to for antigen delivery and enhancement of antibody responses. J. Immunol. mucosal, but not splenic, conventional dendritic cells. J. Exp. Med. 204, 182, 7587–7594. 171–180. Liu, K., Victora, G.D., Schwickert, T.A., Guermonprez, P., Meredith, M.M., Yao, Varol, C., Vallon-Eberhard, A., Elinav, E., Aychek, T., Shapira, Y., Luche, H., K., Chu, F.F., Randolph, G.J., Rudensky, A.Y., and Nussenzweig, M. (2009). Fehling, H.J., Hardt, W.D., Shakhar, G., and Jung, S. (2009). Intestinal lamina In vivo analysis of dendritic cell development and homeostasis. Science propria dendritic cell subsets have different origin and functions. Immunity 31, 324, 392–397. 502–512. Maillard, I., Weng, A.P., Carpenter, A.C., Rodriguez, C.G., Sai, H., Xu, L., Waskow, C., Liu, K., Darrasse-Je` ze, G., Guermonprez, P., Ginhoux, F., Merad, Allman, D., Aster, J.C., and Pear, W.S. (2004). Mastermind critically regulates M., Shengelia, T., Yao, K., and Nussenzweig, M. (2008). The receptor tyrosine Notch-mediated lymphoid cell fate decisions. Blood 104, 1696–1702. kinase Flt3 is required for dendritic cell development in peripheral lymphoid McCright, B., Lozier, J., and Gridley, T. (2006). Generation of new Notch2 tissues. Nat. Immunol. 9, 676–683. mutant alleles. Genesis 44, 29–33. Wegmann, F., Petri, B., Khandoga, A.G., Moser, C., Khandoga, A., Volkery, S., Mebius, R.E., and Kraal, G. (2005). Structure and function of the spleen. Nat. Li, H., Nasdala, I., Brandau, O., Fa¨ ssler, R., et al. (2006). ESAM supports Rev. Immunol. 5, 606–616. neutrophil extravasation, activation of Rho, and VEGF-induced vascular permeability. J. Exp. Med. 203, 1671–1677. Naik, S.H., Metcalf, D., van Nieuwenhuijze, A., Wicks, I., Wu, L., O’Keeffe, M., and Shortman, K. (2006). Intrasplenic steady-state dendritic cell precursors Wilson, A., and Radtke, F. (2006). Multiple functions of Notch signaling in self- that are distinct from monocytes. Nat. Immunol. 7, 663–671. renewing organs and cancer. FEBS Lett. 580, 2860–2868. Naik, S.H., Sathe, P., Park, H.Y., Metcalf, D., Proietto, A.I., Dakic, A., Carotta, Yashiro-Ohtani, Y., Ohtani, T., and Pear, W.S. (2010). Notch regulation of early S., O’Keeffe, M., Bahlo, M., Papenfuss, A., et al. (2007). Development of plas- thymocyte development. Semin. Immunol. 22, 261–269. macytoid and conventional dendritic cell subtypes from single precursor cells Yuan, J.S., Kousis, P.C., Suliman, S., Visan, I., and Guidos, C.J. (2010). derived in vitro and in vivo. Nat. Immunol. 8, 1217–1226. Functions of notch signaling in the immune system: consensus and controver- Onai, N., Obata-Onai, A., Schmid, M.A., Ohteki, T., Jarrossay, D., and Manz, sies. Annu. Rev. Immunol. 28, 343–365. M.G. (2007). Identification of clonogenic common Flt3+M-CSFR+ plasmacy- Zhou, J., Cheng, P., Youn, J.I., Cotter, M.J., and Gabrilovich, D.I. (2009). Notch toid and conventional dendritic cell progenitors in mouse bone marrow. Nat. and wingless signaling cooperate in regulation of dendritic cell differentiation. Immunol. 8, 1207–1216. Immunity 30, 845–859.

Immunity 35, 780–791, November 23, 2011 ª2011 Elsevier Inc. 791 A rt i c l e s

Notch2-dependent classical dendritic cells orchestrate intestinal immunity to attaching- and-effacing bacterial pathogens

Ansuman T Satpathy1, Carlos G Briseño1, Jacob S Lee1, Dennis Ng2, Nicholas A Manieri1, Wumesh KC1, Xiaodi Wu1, Stephanie R Thomas1, Wan-Ling Lee1, Mustafa Turkoz3, Keely G McDonald4, Matthew M Meredith5, Christina Song1, Cynthia J Guidos2,6, Rodney D Newberry4, Wenjun Ouyang7, Theresa L Murphy1, Thaddeus S Stappenbeck1, Jennifer L Gommerman2, Michel C Nussenzweig5,8, Marco Colonna1, Raphael Kopan3 & Kenneth M Murphy1,9

Defense against attaching-and-effacing bacteria requires the sequential generation of interleukin 23 (IL-23) and IL-22 to induce protective mucosal responses. Although CD4+ and NKp46+ innate lymphoid cells (ILCs) are the critical source of IL-22 during infection, the precise source of IL-23 is unclear. We used genetic techniques to deplete mice of specific subsets of classical dendritic cells (cDCs) and analyzed immunity to the attaching-and-effacing pathogen Citrobacter rodentium. We found that the signaling receptor Notch2 controlled the terminal stage of cDC differentiation. Notch2-dependent intestinal CD11b+ cDCs were an obligate source of IL-23 required for survival after infection with C. rodentium, but CD103+ cDCs dependent on the transcription factor Batf3 were not. Our results demonstrate a nonredundant function for CD11b+ cDCs in the response to pathogens in vivo.

The cytokines interleukin 23 (IL-23) and IL-22 are critical for immune DCs may be the main source of IL-23 (refs. 11,12). A role for macro- responses that maintain mucosal integrity against infections by phages was proposed on the basis of the greater pathogen burden of attaching-and-effacing bacterial pathogens1,2. Isolated lymphoid mice deficient in the chemokine receptor CX3CR1 (Cx3cr1−/− mice), follicles (ILFs) in the small and large intestine contain dendritic cells as well as the greater susceptibility of CD11c-DTR mice (which, after © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature (DCs), B cells and innate lymphoid cells (ILCs) that orchestrate protec- treatment with diphtheria toxin, are selectively depleted of cells that tion against those pathogens2–4. During such infections, ILCs produce express diphtheria toxin receptor (DTR) under the control of the pro- IL-22, which promotes barrier integrity by inducing the production of moter of the gene encoding the integrin CD11c) to such infection12. npg antimicrobial peptides, including RegIIIβ and RegIIIγ, by epithelial Alternatively, a role for classical DCs (cDCs) was proposed on the basis cells2,5,6. The importance of IL-22 is indicated by the susceptibility of the greater pathogen burden and lower IL-23 production in mice of Il22−/− mice to the attaching-and-effacing bacterium Citrobacter that lack the lymphotoxin-β receptor (LTβR) in CD11c-expressing rodentium, a model for the infection of humans with enteropathogenic cells11 and by the separate observation that cDCs are the main source and enterohemorrhagic Escherichia coli2,6,7. of IL-23 after stimulation of Toll-like receptor 5 (ref. 13). IL-22-producing ILCs are heterogeneous and include a CD4+ The opposing conclusions of those studies emphasize the difficulty subset5 and a CD4−NKp46+ subset8. Both subsets express the tran- in distinguishing the roles of macrophages and cDCs in vivo, particu- scription factor RORγt, which is required for their development9,10. larly for studies that rely on depletion methods based on CD11c14. An important unresolved question is the identity of the cells that Zbtb46 has been identified as a transcription factor expressed in stimulate ILCs to produce IL-22. ILCs do not directly detect infection cDCs but not macrophages, monocytes or plasmacytoid DCs (pDCs), by attaching-and-effacing pathogens but instead seem to depend on and expression of DTR under the control of the Zbtb46 promoter IL-23 produced by other innate cells for their activation2,8. It has been (Zbtb46DTR) allows selective depletion of cDCs15,16. Other systems suggested that in response to C. rodentium, either macrophages or have been developed that also allow selective depletion of individual

1Department of Pathology and Immunology, Washington University in St. Louis, School of Medicine, St. Louis, Missouri, USA. 2Department of Immunology, University of Toronto, Toronto, Ontario, Canada. 3Department of Developmental Biology, Washington University in St. Louis, School of Medicine, St. Louis, Missouri, USA. 4Department of Internal Medicine, Washington University in St. Louis, School of Medicine, St. Louis, Missouri, USA. 5Laboratory of Molecular Immunology, The Rockefeller University, New York, New York, USA. 6Program in Developmental and Stem Cell Biology, Hospital for Sick Children Research Institute, Toronto, Ontario, Canada. 7Department of Immunology, Genentech, South San Francisco, California, USA. 8Howard Hughes Medical Institute, The Rockefeller University, New York, New York, USA. 9Howard Hughes Medical Institute, Washington University in St. Louis, School of Medicine, St. Louis, Missouri, USA. Correspondence should be addressed to K.M.M. ([email protected]).

Received 22 November 2012; accepted 28 June 2013; published online 4 August 2013; doi:10.1038/ni.2679

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 937 A rt i c l e s

cDC subsets in vivo17–20. Mice deficient in the transcription factor were uniformly GFP+ and were mainly in the CD103+CD11b− and Batf3 lack the CD8α+ cDC subset that is identified by its expression CD103+CD11b+ gates (Fig. 1b), consistent with the ability of cDCs, of CD103 in the periphery; as a result, these mice have defective CD8+ but not macrophages, to migrate to the MLNs28. Resident cDCs were T cell responses to several viral pathogens17,21 and are susceptible also uniformly GFP+, but were predominantly in the CD103−CD11b− to infection with Toxoplasma gondii22. Similarly, mice with selective or CD103−CD11b+ gate. Thus, expression of Zbtb46 seemed to distin- depletion of pDCs have defects in the production of type I interferon guish cDCs from macrophages in the intestine independently of their and are susceptible to chronic infection by viruses such as lymphocytic expression of the integrins CD103 (αEβ7) or CD11b (αM). Histological choriomeningitis virus19,23. Thus, studies using systems of selective assessment of the small intestine of wild-type mice reconstituted with depletion have identified nonredundant roles for CD8α+ cDCs and Zbtb46gfp bone marrow (Zbtb46gfp chimeras) showed that GFP+ cDCs pDCs but not for the third major subset of DCs, the CD11b+ cDCs. were present in organized lymphoid structures, including Peyer’s A subset of CD11b+ cDCs in vivo expressing the adhesion mol- patches and ILFs, as well as at points of antigen encounter in intesti- ecule ESAM undergoes depletion after conditional deletion of the nal villi (Fig. 1c). In the large intestine, GFP+ cDCs were also present gene encoding the signaling receptor Notch2 (ref. 18). That study in colonic patches, ILFs and surrounding villi (Fig. 1c). proposed that Notch2 signaling is selectively required for the develop- The results reported above suggested that the administration of ment of splenic CD11b+ESAM+ cDCs and intestinal CD103+CD11b+ diphtheria toxin to Zbtb46DTR mice should selectively eliminate cDCs cDCs derived from the precursor of the cDC (pre-cDC), analogous to but spare macrophages. Indeed, treatment of the Zbtb46DTR chime- the unique dependence of CD8α+ cDCs on Batf3 (ref. 17). However, ras with diphtheria toxin resulted in a significantly lower frequency CD11b+ESAM− cDCs persist in Notch2-deficient mice, which prompts of both CD103+CD11b− cDCs and CD103+CD11b+ cDCs in the the question of how those cells are related to CD11b+ESAM+ cDCs lamina propria and the F4/80− populations of CD103−CD11b− cDCs and whether they provide compensatory functions. Furthermore, and CD103−CD11b+ cDCs, but did not affect F4/80+ macrophages although that study suggested that Notch2 specifically regulates the (Fig. 2a,b). In addition, lymphocytes and ILCs in the lamina propria CD11b+ branch of cDCs, some evidence indicated a developmental did not undergo depletion after treatment of Zbtb46DTR chimeras with defect in the CD8α+ branch as well, although that was not further diphtheria toxin and did not express Zbtb46-GFP (Supplementary analyzed18. Unexpectedly, mice that lack CD103+CD11b+ cDCs have Fig. 1a–c). Thus, we used this system to determine whether cDCs only 50% less production of IL-17 by T cells stimulated ex vivo, and were required for protection against C. rodentium. We found that responses during infection have not been examined to show a specific Zbtb46DTR chimeras treated with diphtheria toxin were unable to function for such cells in immunological defense. recover after challenge with C. rodentium and died within 8–12 d Here we demonstrate a selective function for CD11b+ cDCs in of infection (Fig. 2c). In contrast, wild-type mice reconstituted with immunological defense against pathogens. Through the use of several wild-type bone marrow (wild-type chimeras) and Zbtb46DTR chi- genetic models with selective depletion of various subsets of cDCs meras not treated with diphtheria toxin survived beyond 15 d; they in vivo, we found that Notch2-dependent intestinal CD103+CD11b+ underwent some weight loss but recovered normal weight by 30 d cDCs provided nonredundant protection against infection with after infection (Fig. 2c,d). attaching-and-effacing pathogens, but macrophages or Batf3- To confirm a specific role for cDCs and to investigate a possible dependent CD103+CD11b− cDCs did not. Notch2-dependent cDCs role for monocyte-derived cells in protection from C. rodentium, we were required for host survival mediated by the local generation next analyzed mice deficient in the cytokine Flt3L (Flt3l−/− mice)29 of IL-23-dependent antimicrobial responses early after infection. and mice deficient in the chemokine receptor CCR2 (Ccr2−/− mice)30. © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature Developmentally, Notch2 regulated the terminal differentiation of Flt3l−/− mice have fewer lamina propria cDCs but maintain normal both the CD8α+ and CD11b+ branches of cDCs, which allowed their numbers of macrophages and monocytes24,25. Similar to diphthe- subsequent homeostatic population expansion via signaling through ria toxin–treated Zbtb46DTR chimeras, Flt3l−/− mice succumbed to + + −/− npg LTβR. Our results demonstrate that intestinal CD103 CD11b cDCs C. rodentium infection after 10–16 d (Fig. 2e,f). In contrast, Ccr2 detect infection with attaching-and-effacing pathogens and secrete mice have a specific defect in the recruitment of monocytes to the IL-23, which is critical for IL-22 production by ILCs and the mainte- intestinal lamina propria but maintain normal populations of cDCs31. nance of mucosal integrity. Although Ccr2−/− mice underwent more weight loss than did their wild-type (Ccr2+/+) counterparts, only 25% succumbed to infection RESULTS (Fig. 2e,f). These results demonstrated that cDCs, rather than mac- cDCs mediate defense against attaching-and-effacing pathogens rophages or monocyte-derived cells, were required for early innate The generation of Zbtb46gfp mice, which express green fluorescent defense against C. rodentium, but did not indicate whether a particu- protein (GFP) driven by the Zbtb46 promoter (Zbtb46-GFP), as well lar cDC subset was required. as Zbtb46DTR mice, has allowed the selective visualization and depletion of cDCs, which can help in distinguishing the requirements for Notch2-deficient mice lack CD11b+ cDCs in vivo cDCs and macrophages in immune responses in vivo15,16. We sought Notch2 is required for the development of the CD11b+ESAM+ cDC to determine whether expression of Zbtb46-GFP could be used to dis- fraction18. However, the specific function of those cells has not been tinguish cDCs from macrophages in the intestinal lamina propria and determined in vivo. To characterize how Notch2 signaling influences mesenteric lymph nodes (MLNs) (Fig. 1a,b). Among MHCII+CD11c+ cDC development, we examined the requirement for Notch2 and its lamina propria cells, CD103+CD11b− cells and CD103+CD11b+ cells transcriptional partner RBPJ, as well as the γ-secretase complex com- were uniformly GFP+, consistent with their identity as cDCs. Both posed of presenilin 1 (PSEN1) and PSEN2, in the development of CD103−CD11b+ cells and CD103−CD11b− cells also included a sub- splenic CD8α+ or CD11b+ cDCs (Fig. 3a). Hematopoietic loss of stantial GFP+ fraction (Fig. 1a), in agreement with published studies either Rbpj or Notch2 mediated by Cre recombinase expressed from suggesting that cDCs may also reside in those gates16,24,25. In contrast, the hematopoietic compartment–specific promoter Vav1 (Vav1-Cre) F4/80+ macrophages were present mainly in the MHCII−CD11c+ gate resulted in a lower frequency of CD11c+MHCII+ cDCs and CD8α+ and did not express GFP (Fig. 1a). In the MLNs, migratory cDCs26,27 cDCs (identified by expression of the surface marker CD24) and

938 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology A rt i c l e s

Figure 1 Zbtb46-GFP identifies intestinal cDC a + + – – populations. (a,b) Flow cytometry of live cells Lamina propria CD103 CD103 CD103 CD103 + – from the lamina propria (a) and MLNs (b) of CD45 B220 CD11b– CD11b+ CD11b– CD11b+ gfp/+ Zbtb46 mice (n = 9); pregating, above plots. 17 38 Numbers adjacent to outlined areas indicate CD11c+ 97 97 61 56 percent cells in each throughout. MHCII, MHCII+ 4.6 24 0.7 0.5 12 21 major histocompatibility complex class II. 10

(c) Fluorescence microscopy of Zbtb46-GFP, + 5 5 5 CD11c 10 10 4.0 7.0 10 CD4, B220 and β-catenin, and of staining with MHCII+ 4 4 4 10 24 10 10 78 94 28 3.1 the DNA-intercalating dye DAPI, in sections of & 3 3 3 + 10 10 80 10 CD11c 2.2 2.6 5.9 88 small intestine (top) and colon (bottom) from – 0 0 5.1 0 MHCII MHCII CD103 Zbtb46-GFP chimeras ( = 5) generated by the reconstitution 3 4 5 3 4 5 3 4 5 n 0 10 10 10 0 10 10 10 0 10 10 10 gfp of wild-type mice with Zbtb46 bone marrow. CD11c CD11b F4/80 Scale bars, 200 µm (Peyer’s patch and colonic b + + – – patch) or 100 µm (isolated lymphoid follicles CD103 CD103 CD103 CD103 MLN + – – + – + (ILF) and villi). Data are representative of CD45 B220 CD11b CD11b CD11b CD11b three independent experiments (a,b) or two 1.5 independent experiments (c). 23 44 Migratory 97 93 85 77 cDC 13 0.4 0.5 0.0 8.6 5.2 elimination of CD11b+ESAM+CD4+ cDCs (also identifiable by expression of the signal- 105 105 105 4 4 11 3.3 4 Resident 10 10 10 99 96 99 80 transduction molecule CD172a rather than 3 1.0 3 3 cDC 10 10 49 10 0.0 1.0 0.2 0.8 CD11b; Fig. 3a). Vav1-Cre–mediated dele- 0 0 0 CD103 Zbtb46-GFP MHCII 19 tion of Psen1 and Psen2 also resulted in a 0 103 104 105 0 103 104 105 0 103 104 105 + + similarly lower frequency of CD11c MHCII CD11c CD11b F4/80 cDCs and CD8α+ cDCs and elimination of Peyer’s patch ILF Villi ESAM+CD4+ cDCs (Fig. 3a). These results c indicated that deletion of RBPJ affected cDC development through loss of canonical Notch signaling rather than through derepres- Small sion of genes that are not targets of Notch. intestine Histologically, deletion of Notch2 eliminated the characteristic clusters of CD11b+ cDCs in

the marginal zone and bridging channels of the Zbtb46-GFP CD4 B220 Zbtb46-GFP CD4 β-catenin DAPI spleen32, which left a distribution of cDCs Colonic patch ILF Villi loosely scattered throughout the T cell zones and red pulp (Fig. 3b). As loss of Notch2 also affects the develop-

© 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature 33 ment of additional hematopoietic lineages , Colon we used Cre expressed from the promoter of the gene encoding CD11c (CD11c-Cre)

npg instead of the Vav1-Cre system to restrict the

effect of Notch2 deletion to myeloid lineages Zbtb46-GFP CD4 β-catenin DAPI for analysis of cDC function during infection in vivo. Deletion of Notch2 by CD11c-Cre (Notch2cKO) gener- In contrast, Batf3−/− mice selectively lacked the complementary ated defects in cDC development identical to those that resulted lamina propria CD103+CD11b− cDC subset (Fig. 3e) but retained from Vav1-Cre–induced deletion of Notch2 (Fig. 3c,d) but prevented normal numbers of CD103+CD11b+ cDCs34. Similarly, Notch2cKO effects on other hematopoietic lineages (Supplementary Fig. 2). mice had considerably fewer migratory CD103+CD11b+ cDCs Notch2cKO mice had fewer splenic cDCs than did their Notch2-sufficient in the MLNs, and Batf3−/− mice had no CD103+CD11b− cDCs counterparts and had no ESAM+CD4+ cDCs (Fig. 3c,d). A fraction (Fig. 3e). The development of cryptopatches and ILFs containing of CD24+ cDCs also expressed ESAM and was dependent on CD11c+ cDCs and ILCs was unaffected in Notch2cKO mice (Fig. 3f and Notch2 signaling (Fig. 3c,d), which suggested that Notch2 acted in a Supplementary Fig. 2d), consistent with a published report show- similar way, although to a different extent, in the development of ing that CD11b+ DCs are present mainly in the lamina propria and the CD11b+ and CD8α+ subsets of cDCs. Notch2-dependent not in ILFs35. Thus, Notch2cKO mice provided an in vivo system ESAM+ cDCs were abundant not only in the spleen but also in the for the analysis of CD11b+ cDC function in the presence of intact resident cDC fraction in the MLNs (Supplementary Fig. 2a). lymphoid structures. Similarly, a small fraction of resident ESAM+, Notch2-dependent cDCs was present in skin-draining lymph nodes (Supplementary Notch2 controls the terminal differentiation of cDC subsets Fig. 2a). In contrast, migratory dermis-derived cDCs in skin-draining Although the role of Notch2 in cDC development is thought to be lymph nodes and peripheral tissue–resident cDCs in the lungs limited to the CD11b+ subset of cDCs18, we observed that Notch2cKO and kidneys were not affected by Notch2 deletion18 (Supplementary mice also had defects in the CD8α+ cDC subset (Fig. 3c,d). As CD8α Fig. 2a,b). In the intestinal lamina propria, deletion of Notch2 resulted expression can be altered by manipulation of the Notch pathway36, in a much lower frequency of CD103+CD11b+ cDCs (Fig. 3e). we used expression of CD24 and the DC marker DEC-205 (CD205)

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 939 A rt i c l e s

Zbtb46DTR WT 100 a – DT + DT – DT + DT b → c DTR 5 Zbtb46 → WT + DT 6.2 0.3 10 80 4 8.1 19 0.3 4.1 10 15 NS 60 Spleen 103 WT → WT (– DT) 40 WT → WT + DT ) DTR 3 0 10 Survival (%) 20 Zbtb46 → WT (– DT) ** DTR 3 4 5 Zbtb46 → WT + DT 0 10 10 10 0 CD11b ** 0 5 10 15 20 25 30 105 24 16 5 105 Cells ( × 10 Time after infection (d) 104 Lamina 4 ** 10 propria 103 WT → WT (– DT) 3 38 84 0 30 10 d WT → WT + DT 0 + 20 DTR – + MΦ MHCII CD103 0 cDC cDC Zbtb46 → WT (– DT) 10 DTR 3 4 5 Zbtb46 → WT + DT 0 10 10 10 3 4 5 F4/80 0 10 10 10 + CD11b + CD11b 0

CD11c Body F4/80 –10 CD103 CD103 –20 weight change (%) –30 Figure 2 Zbtb46+ cDCs are essential for survival after infection with C. rodentium. (a) Expression of 0 5 10 15 20 25 30 Time after infection (d) CD103 and F4/80 (right) in lamina propria cells from chimeras generated by the reconstitution of wild-type (WT) mice with Zbtb46 DTR bone marrow, left untreated (− DT) or treated with diphtheria e 100 toxin (40 ng per gram body weight (ng/g)) on day −3 and day −1 (+ DT) and assessed on day 0 80 + + (CD45 B220 cells pregated at left). (b) Quantification of cDCs and macrophages (MΦ) in the 60 DTR lamina propria of the chimeras in a ( WT (donor recipient); = 4), presented per +/+ Zbtb46 → → n 40 Ccr2 1 × 106 lamina propria cells. (c,d) Survival (c) and weight loss (d) of chimeras (key: donor→recipient; Ccr2–/– Survival (%) +/+ 9 20 Flt3l n = 8 per group, except WT → WT + DT, n = 5) given oral inoculation of C. rodentium (2 × 10 colony- ** Flt3l–/– forming units) and otherwise left untreated (− DT) or treated with diphtheria toxin (20 ng/g) 1 d before 0 0 5 10 15 20 25 30 infection and every third day (5 ng/g) for the remainder of the experiment (+ DT). (e,f) Survival (e) Time after infection (d) and weight loss (f) of C. rodentium–infected mice (n = 5 (Ccr2 +/+), 4 (Ccr2 −/−) or 10 (Flt3l +/+ and Ccr2+/+ −/− 30 –/– Flt3l )). NS, not significant; *P < 0.05 and **P < 0.001 (Student’s t-test (b,d,f) or log-rank f Ccr2 Mantel-Cox test (c,e)). Data are from two independent experiments (error bars (b,d,f), s.e.m.). 20 Flt3l+/+ Flt3l–/– 10 0 Body –10 to identify this subset of cDCs (Supplementary Fig. 3a). To assess –20 * the transcriptional effects of Notch2, we analyzed gene expression weight change (%) –30 in both CD11b+ cDCs and DEC-205+ cDCs from mice with loxP- 0 5 10 15 20 25 30 Time after infection (d) flanked Notch2 alleles (Notch2f/f mice) and Notch2cKO mice (Fig. 4a and Supplementary Fig. 3a). As expected, we noted substantial differ- Cx3cr1, we found that CD11b+ESAM+ cDCs, which underwent deple- ences between Notch2f/f CD11b+ cDCs and Notch2cKO CD11b+ cDCs. tion in Notch2cKO mice, were distinguished from CD11b+ESAM− In particular, the expression of known Notch targets such as Hes1 and cDCs on the basis of GFP expression (Fig. 4e, top). Similarly, two Dtx1 was lower in Notch2cKO CD11b+ cDCs, as was the expression populations were also distinguished on the basis of their expression of cDC-specific genes such as Lphn3, Spint1 and Dnase1l3 (Fig. 4a). of GFP and CD4 (Fig. 4e, bottom). These subsets probably reflected © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature Most of the genes regulated by Notch2 in CD11b+ cDCs were also true differences in the maturity of CD11b+ cDCs, as Zbtb46, which regulated by Notch2 in DEC-205+ cDCs (Fig. 4a). In CD11b+ cDCs, has intermediate expression in cDC progenitors and is upregulated Notch2 influenced gene expression in both the ESAM+ and ESAM− in mature cDCs16, had higher expression in the ESAM+ fraction of + − npg fractions (Supplementary Fig. 3b), which suggested that Notch2 CD11b cDCs than in the ESAM fraction (Fig. 4d,e). Accordingly, acted early after differentiation of the CD11b+ cDC subset from pre- both ESAM+ cDCs and ESAM− cDCs developed from wild-type com- cDCs, before the induction of ESAM expression, and that its actions mon DC progenitors and pre-DCs in vivo (Supplementary Fig. 4a,b), were not simply restricted to the development of an ESAM+ subset. which supported a model in which Notch2 influenced the terminal Likewise, in DEC-205+ cDCs, Notch2 regulated the same set of genes cDC differentiation of pre-DC–derived cells. in both the ESAM+ and ESAM− fractions, which largely overlapped To evaluate the stage at which Notch2 first affected cDC develop- the genes regulated in CD11b+ cDCs (Supplementary Fig. 3b). ment, we analyzed competitive mixed chimeras generated by recon- Principal-component analysis showed that similar genes were influ- sititution of wild-type mice with bone marrow from wild-type donors enced by Notch2 deficiency in CD11b+ cDCs and DEC-205+ cDCs. and donors with Vav1-Cre–mediated deletion of Notch2 (Notch2vav), Principal component 1 (PC1) segregated cDCs by lineage subset, in which Notch2 is deleted at an early stage of hematopoietic devel- which distinguished CD11b+ cDCs from DEC-205+ cDCs (Fig. 4b). opment (Fig. 4f). We observed equal competition between wild-type In contrast, PC2 segregated both cDC lineages by the presence or and Notch2vav progenitors for cells in the lineage-negative (Lin−) absence of Notch2 signaling, which distinguished Notch2-sufficient Sca-1+c-Kit+ fraction and up to the pre-DC stage of DC development cDCs from Notch2-deficient cDCs of each lineage (Fig. 4b). Next (Fig. 4f and Supplementary Fig. 5a). However, in the mature splenic we assessed the expression of genes most heavily weighted in PC2 cDC compartment, wild-type cells outcompeted Notch2vav cells along the developmental pathway from common myeloid progeni- (Fig. 4f and Supplementary Fig. 5b), which indicated that Notch2 tor to splenic cDC. Genes with the highest ‘loadings’ in PC2, which first affected development after the pre-DC. We noted this effect in were thus induced by Notch2, had high expression in terminally dif- both the CD11b+ and CD8α+ subset of mature cDCs and, further- ferentiated cDCs but low expression in DC progenitors (Fig. 4b–d). more, in both the ESAM+ and ESAM− fraction of each cDC subset. Conversely, genes with the most negative ‘loadings’ in PC2 had high In summary, our analyses of gene expression and competitive bone expression in progenitor cells. Through the use of reporter mice marrow chimeras indicated that Notch2 acted in the terminal dif- with sequence encoding GFP knocked into two such genes, Ccr2 and ferentiation of CD11b+ cDCs and CD8α+ cDCs.

940 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology A rt i c l e s

a b Notch2f/f Notch2vav d 3.1 76 68 f/f WT Notch2 35 ** cKO 15 Notch2 30 )

3 25 ** 1.3 57 20 8.2 ** Rbpjvav 24 15 CD11c IgD B220 Cells ( × 10 10 * 5 ** 1.6 0 66 5.6 + + + + Psen1vav vav 24 cDC Psen2 CD24 CD172 + ESAM+ ESAM

CD24 CD172 105 1.5 105 105 4 4 69 4 8.6 CD11c MAdCAM-1 vav 10 10 10 Notch2 3 3 18 3 10 10 10 e Notch2f/f Notch2cKO Batf3–/– 0 0 0 + + CD4 CD172 MHCII CD45 CD11c 3 4 5 3 4 5 3 4 5 0 10 10 10 0 10 10 10 0 10 10 10 2.5 7.2 2.9 1.2 0.2 8.7 CD11c CD24 ESAM CD11c+ + + Lamina CD11c+MHCII+CD172+ MHCII CD24 c propria 2.8 17 81 – + hi f/f B220 CD11c MHCII Notch2 57 65 5 17 22 10 25 5.6 26 21 3.0 31 104 MLN 103 2.1 53 32 52 36 Notch2f/+ 81 0 CD103 CD11c-Cre 22 23 3 4 5 12 0 10 10 10 16 CD11b

5 5 5 5 5 f/f cKO 10 1.3 10 10 10 72 10 f Notch2 Notch2 104 104 64 104 104 104 cKO Notch2 103 103 103 4.5 103 3.5 103 5.0 0 0 33 0 0 0 CD11b CD11b CD4 MHCII CD172 CP 0 103 104 105 0 103 104 105 0 103 104 105 0 103 104 105 0 103 104 105 CD11c CD24 ESAM ESAM ESAM

Figure 3 Canonical Notch2 signaling is required for the development of splenic and intestinal CD11b+ cDCs. (a) Flow cytometry of live splenocytes from wild-type mice and mice with Vav1-Cre–mediated deletion of Rbpj ILF (Rbpjvav), Psen1 and Psen2 (Psen1vavPsen2 vav) or Notch2 (Notch2 vav), pregated as MHCII+CD11c+ (left) and CD172a+CD24− (middle) cells. (b) Fluorescence microscopy of CD11c, IgD and B220 (top) or CD11c and

© 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature MAdCAM-1 (bottom) in sections of spleen from Notch2f/f and Notch2 vav mice. Scale bars, 200 µm. (c) Flow cytometry of splenocytes from Notch2 f/f mice (top), mice with CD11c-Cre–mediated deletion of a single CD11c CD90 DAPI loxP-flanked Notch2 allele (Notch2 f/+; middle) and Notch2cKO mice (bottom), stained for various markers (plot margins). (d) Quantification of cDC development in the Notch2 f/f and Notch2 cKO mice in c, presented per 1 × 106 splenocytes. (e) Flow cytometry npg of cells from the lamina propria of Notch2 f/f, Notch2 cKO and Batf3 −/− mice, stained for various markers (plot margins). (f) Immunofluorescence microscopy of cryptopatches (CP) and ILFs from sections of small intestine from Notch2 f/f and Notch2 cKO mice, stained for CD11c and CD90 and with DAPI. Scale bars, 50 µm. *P < 0.01 and **P < 0.001 (Student’s t-test). Data are representative of three independent experiments (a,c,e; n = 6–8 (a), 7 (c) or 3–5 (e) mice per group) or two independent experiments (b,f; n = 4 (b) or 5 (f) mice) or are from three independent experiments (d; n = 7 mice per group; error bars, s.e.m.).

LTbR mediates the population expansion of Notch2-dependent cDCs bridging channels38, we investigated whether signaling via both As Notch2 signaling controls the terminal differentiation of cDCs in Notch2 and LTβR selectively influenced the same cDC subset. the spleen and intestine, its effects should be subsequent to those of We compared the development of splenic cDC subsets in wild- Flt3L signaling, which can be observed in the population expansion type, Notch2cKO and LTβR-deficient (Ltbr−/−) mice. Ltbr−/− mice, of DC progenitors in the bone marrow37. To test that hypothesis, similar to Notch2cKO mice, had fewer CD11c+MHCII+ cDCs and we compared the development of Notch2-dependent cDC subsets in CD8α+ cDCs and considerably fewer ESAM+CD4+ cDCs (Fig. 5a). wild-type and Flt3l−/− mice (Supplementary Fig. 6a,b). As expected, Similarly, Nik−/− mice, which are deficient in activation of the non- Flt3l−/− mice had a much lower abundance of all subsets of cDCs, canonical pathway of the transcription factor NF-κB downstream including CD11b+ESAM+ cDCs. However, treatment of wild-type of LTβR signaling, and Nfkb1−/− mice, which lack the p105 sub mice with Flt3L resulted in 25-fold more CD11b+ESAM− cDCs but unit of NF-κB, showed selective loss of ESAM+ cDCs but retained only 2.5-fold more CD11b+ESAM+ cDCs (Supplementary Fig. 6c,d). ESAM− cDCs (Supplementary Fig. 7a,b). We did not observe Thus, Flt3L was necessary but not sufficient for development of those defects in mice lacking the costimulatory receptor CD40 CD11b+ESAM+ cDCs, which suggested that Notch2 regulated a step (Supplementary Fig. 7a), a member of the tumor-necrosis factor in cDC development subsequent to the actions of Flt3L. receptor family that can also activate NF-κB39. As Ltbr−/− mice 40 Because the lymphotoxin LTα1β2 is required for the prolif- lack lymph nodes and Peyer’s patches , we sought to determine eration of splenic CD8− cDCs located in the marginal zone and whether the loss of ESAM+ cDCs in these mice was cell intrinsic

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 941 A rt i c l e s

a CD11b+ cDC DEC-205+ cDC b Dtx1 0 Hes1 0 Clec10a + cDC+ cDC + cDC+ cDC

) Clec10a ) Hes1

10 –2 Spint1 10 –2 Dnase1l3 Positive CMPGMPCDPCD8CD4Negative CMPGMPCDPCD8CD4 Dtx1 Ccr2 Spint1 Lphn3 Ccr2 –4 –4 150 1300002K09Rik Cd226 Dnase1l3 Lphn3 f/f1 CD11b Gp2 Fcrls –6 Lphn3 –6 100 Ccr2 f/f2 CD11b f/f1 DEC-205 Spint1 Clec10a P value (log P value (log f/f3 CD11b f/f3 DEC-205 Hes1 Fn1 –8 –8 50 f/f2 DEC-205 Dnase1l3 Mgl2 0.01 0.1 1 10 100 0.01 0.1 1 10 100 Slc12a2 Lpl Expression (fold) Expression (fold) 0 Abcb1a Cfn PC2 f/f vs KO f/f vs KO KO1 DEC-205 Arhgap42 Cysltr1 –50 Arap2 Ccdc109b DEC-205+ cDC CD11b+ cDC KO3 CD11b KO3 DEC-205 KO1 CD11b KO2 DEC-205 Hic1 Gda 0 0 –100 KO2 CD11b Apol7c Klrb1b ) ) Rundc3b Capg 10 –2 10 –2 –150 Gpm6b Clec4b1 Clec4a4 Dpep2 –4 –4 –150 –100 –50 0 50 100 150 PC1 Serpinb2 Cd84 –6 –6 Esam Clec4e Ndst1 Fgfr1 P value (log P value (log –8 –8 Cacna1e Tnip3 0.01 0.1 1 10 100 0.01 0.1 1 10 100 Arhgap42 Olfm1 Expression (fold) Expression (fold) f/f vs KO f/f vs KO 2.75 14.75 Expression (log2) Positive ** Negative * Figure 4 Notch2 controls the c d Ccr2 Cd4 6 6 ** Cx3cr1 Esam terminal differentiation of Zbtb46 ) 4 ) 6 2 2 4

+ + )

CD11b and DEC-205 cDCs. 2 2 2 4 (a) Microarray analysis of cDC 2 f/f 0 0 subsets from Notch2 (f/f) 0 cKO –2 –2 and Notch2 (KO) mice, –2 + + − sorted as MHCII CD11c CD24 Expression (log –4 Expression (log –4 –4 + + Expression (log CD11b cells (CD11b cDCs; –6 –6 –6 left) or MHCII+CD11c+CD24+ + ) + ) + ) − + CMP GMP CDP + cDC + cDC CMP GMP CDP + cDC + cDC CMP GMP CDP + cDC+ cDC CD11b DEC-205 cells (DEC- Pre-DC Pre-DC Pre-DC 205+ cDCs; right); P values CD8 CD4 CD8 CD4 CD8 CD4 (vertical axes), Welch’s t-test. Pre-cDC (Zbtb46 Pre-cDC (Zbtb46 Pre-cDC (Zbtb46 Colors indicate expression more CD11c+MHCII+CD11b+ e f Notch2f/f + WT than twofold higher (red) or lower vav 5 5 5 Notch2 + WT ** * ** ** ** (blue) in Notch2 f/f cells than in 10 10 10 2 4 4 4 cKO cells. (b) Principal- 10 10 10 1 Notch2 3 3 3 component analysis (left) of 10 10 10 0.5 0.25 Notch2 f/f and Notch2 cKO CD11b+ 0 0 0 0.125 + 3 4 5 3 4 5 3 4 5 cDCs (CD11b) and DEC-205 cDCs 0 10 10 10 0 10 10 10 0 10 10 10 0.063 ESAM ESAM ESAM (DEC-205), analyzed by individual 0.031

replicate (f/f1–f/f3 and KO1–KO3; 105 105 105 Chimerism ratio 0.016 variance: PC1, 54.0%; PC2, 104 104 104 0.008 © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature 21.0%), and gene expression 103 103 103 0.004 (right) in progenitors (common 0 0 CDP cDC – + – + 0 CX3CR1-GFP GMP cDC cDC cDC cDC Zbtb46-GFP CCR2-GFP Pre-DC myeloid progenitors (CMP), 3 4 5 3 4 5 0 103 104105 0 10 10 10 0 10 10 10 MonocyteNeutrophil + ESAM+ ESAM+ ESAM+ ESAM granulocyte-macrophage progenitors CD4 CD4 CD4 npg (GMP) and common DC progenitors CD24 CD24 CD11b CD11b (CDP)) and in CD8+ and CD4+ cDC subsets, derived from the Immunological Genome Project database for probe sets corresponding to the 20 most positive and negative ‘loadings’ in PC2 (log-transformed values). (c) Gene expression in DC progenitors and subsets of probe sets corresponding to the 50 most positive and negative ‘loadings’ in PC2 (presented as mean-centered, log-transformed values). Each symbol represents an individual gene; small horizontal lines indicate the mean. (d) Gene expression of selected probe sets, analyzed as in c. (e) Flow cytometry of splenic CD11b+ cDCs from Ccr2 gfp mice (left), Cx3cr1gfp mice (middle) and Zbtb46 gfp mice (right). (f) Contribution of donor cells to bone marrow progenitors and splenocytes in mixed chimeras generated with CD45.2+ Notch2 f/ f and CD45.1+ wild-type bone marrow (Notch2 f/f + WT) or CD45.2+ Notch2 vav bone marrow and CD45.1+ wild-type bone marrow (Notch2 vav + WT), analyzed 8–10 weeks after lethal irradiation and transplantation; results are presented as the ratio of Lin−Sca-1+c-Kit+ chimerism in the same mouse (percent contribution of CD45.2+ cells / percent CD45.2+ cells among Lin−Sca-1+c-Kit+ cells), with chimerism of monocytes, neutrophils and cDCs analyzed among splenocytes. Each symbol represents an individual mouse; small horizontal lines indicate the mean. *P < 0.01 and **P < 0.001 (Kruskal-Wallis test with Dunn’s multiple comparison test (c) or Student’s t-test (f)). Data are from two independent experiments (a,b, left; n = 3 biological replicates per cell type), three independent experiments (b, right, c,d; n = 3–5 biological replicates per cell type) or two independent experiments (f; n = 3–5 mice per group) or are representative of three independent experiments (e; n = 5–6 mice per group).

by analyzing mixed chimeras generated by reconsitution of wild- similar competitive disadvantage relative to that of wild-type bone type mice with wild-type and Ltbr−/− bone marrow. We observed marrow in the generation of both branches of cDCs. substantial outcompetition of Ltbr−/− bone marrow by wild-type To evaluate the epistatic relationship between Notch2 and LTβR bone marrow in the generation of splenic and MLN-resident signaling, we examined cDC development in mixed chimeras gener- CD11b+ cDCs (Fig. 5b,c), as well as outcompetition of Ltbr−/− ated by the reconsitution of wild-type mice with Notch2cKO and Ltbr−/− CD8α+ cDCs by wild-type CD8α+ cDCs in this setting (Fig. 5b,c). bone marrow. In these mixed chimeras, CD11b+ cDCs in the spleen Thus, Ltbr−/− bone marrow and Notch2cKO bone marrow had a and MLNs developed equally from Notch2cKO bone marrow and Ltbr−/−

942 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology A rt i c l e s

a B220–CD11c– CD172+CD24– – + – b B220 CD11b DEC-205 1.7 45 11 6.4 68 24 Ltbr–/– + WT 57 WT 6.9 70 34 17 5 5 5 5 10 10 1.1 10 45 10 104 104 104 104 13 41 –/– cKO 11 72 Ltbr + Notch2 103 103 103 103 2.2 cKO 2.3 9.2 38 Notch2 0 0 0 0 MHCII CD172 CD45.1 CD11b 44 0 103 104 105 0 103 104 105 0 103 104 105 0 103 104 105 CD24 CD11c CD24 ESAM Neutrophil cDCs CD172+CD24– CD172–CD24+ 105 105 44 105 64 4 4 4 0.3 0.5 10 10 10 0.4 –/– 28 3 2.8 3 3 72 80 65 Ltbr 10 10 10 49 Ltbr–/– + WT 36 0 0 0 51 22 17 29 MHCII CD11b CD11b 0 103 104 105 0 103 104 105 0 103 104 105 CD11c DEC-205 ESAM 105 1.0 1.0 1.1 104 68 61 66 60 –/– cKO Ltbr + Notch2 103 Ltbr–/– + WT 32 33 32 34 c 0 Ltbr–/– + Notch2cKO CD45.2 3 4 5 2 0 10 10 10 *** *** ** ** CD45.1

1 d e WT WT Ltbr–/– WT Ltbr–/– +WT WT CD11b+CD103+ cDCs CD11c+CD11b+ *** 5 10 0.5 10 11 8.9 14 104 3

Chimerism ratio 10 0.25 1 0 CD103 0 103 104 105 0.125 F4/80 0.1

105 0.1 99 7.1 Chimerism ratio cDC cDC + + cDC cDC + cDC 104 Monocyte 103 94 0.3 91 0.01 CD11b CD24 CD11b Spleen Resident MLN 0

CD45.1 + WT + WT –/– –/– 101 103104105 Ltbr CD45.2 Ifnar1

Figure 5 LTβR signaling mediates the homeostatic population expansion of Notch2-dependent cDCs. (a) Flow cytometry of live splenocytes from © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature wild-type, Notch2 cKO and Ltbr−/− mice, stained for various markers (plot margins; pregating, left and middle). (b) Flow cytometry of live splenocytes from mixed chimeras generated with CD45.1+ Ltbr −/− and CD45.2+ wild-type bone marrow (Ltbr−/− + WT) or with CD45.1+ Ltbr −/− and CD45.2+ Notch2 cKO bone marrow (Ltbr −/− + Notch2 cKO), analyzed 8–10 weeks after lethal irradiation and transplantation (pregating, above plots; neutrophils, CD11c−CD11b+CD24+ cells). Bottom row, red numbers in top right corners indicate chimerism ratio (percent contribution of CD45.1+ Ltbr −/− npg cells/percent CD45.1+ Ltbr −/− cells among splenic neutrophils). (c) Contribution of Ltbr −/− bone marrow to each cell type in the mixed chimeras in b, presented as a ratio of neutrophil chimerism in the same mouse. Each symbol represents an individual mouse; small horizontal lines indicate the mean. (d) Contribution of wild-type or Ltbr −/− bone marrow to CD103+CD11b+ cDCs in the lamina propria from chimeras generated by the reconstitution of CD45.1+CD45.2+ wild-type mice with CD45.1+CD45.2+ wild-type bone marrow (WT → WT), CD45.1+ Ltbr −/− bone marrow (Ltbr −/− → WT) or a mixture of Ltbr −/− and wild-type bone marrow (Ltbr −/−+WT → WT). (e) Quantification of chimerism in mixed chimeras generated with a mixture of Ltbr −/− and wild-type bone marrow (as in d) or Ifnar1 −/− and wild-type bone marrow, calculated as follows: (percent contribution of mutant cells to intestinal cDCs / percent contribution of wild-type cells to intestinal cDCs) / (percent contribution mutant cells to splenic T cells / percent contribution of wild-type cells to splenic T cells). Each symbol represents an individual mouse; small horizontal lines indicate the mean (± s.e.m.). *P < 0.05, **P < 0.01 and ***P < 0.001 (Student’s t-test). Data are representative of three independent experiments (a; n = 6 mice per group) or five independent experiments (d; n = 10–11 mice per group) or are from two independent experiments (b,c; n = 2–3 mice per group) or five independent experiments (e; n = 10 mice (Ltbr −/−) or 5 mice (Ifnar1 −/−)).

+ bone marrow (Fig. 5b,c). Similarly, CD8α cDCs from either donor or undergo LTα1β2-mediated population expansion (Supplementary also developed equally. These results suggested that Notch2 and LTβR Fig. 7c). Given those results, we sought to determine if LTβR also had acted along a similar pathway in cDC development. Furthermore, in a similar role in development of Notch2-dependent CD103+CD11b+ these mixed chimeras, Notch2cKO bone marrow did not generate any cDCs in the lamina propria. Indeed, in mixed chimeras generated by ESAM+ cDCs, whereas Ltbr−/− bone marrow generated a small popu- the reconstitution of wild-type mice with wild-type and Ltbr−/− bone lation of ESAM+ cDCs (Fig. 5b), which suggested that the require- marrow, we noted outcompetition of Ltbr−/− bone marrow by wild-type ment for Notch2 preceded the requirement for LTβR. In this model, bone marrow in the production of CD103+CD11b+ cDCs, unlike mice Ltbr−/− cDCs were able to activate Notch2 signaling and progress to deficient in the receptor for interferon-αβ (Ifnar1−/− mice), which did an ESAM+ subset but were unable to undergo homeostatic expansion, not show defects in cDC development (Fig. 5d,e). Thus, LTβR signal- which resulted in diminished fitness relative to that of wild-type cDCs; ing was required for the population expansion of Notch2-dependent in contrast, Notch2cKO cDCs were unable to progress to ESAM+ cells cDC subsets.

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 943 A rt i c l e s

a b c d 100 30 9 6 9 *** 20 * *** 80 ** 8

) ) *** 10 8 * f/f f/f 10 10 Notch2 Notch2 60 5 cKO 7 cKO 0 Notch2 Notch2 –10 7 Batf3–/– Batf3–/– 40 Batf3–/– ** Batf3–/– 6

Survival (%) 4 Notch2f/f –20 Notch2f/f

(CFU/g log LOD (CFU/g log 20 *** cKO cKO 6 5 Notch2 –30 *** Notch2 Colon length (cm)

*** C.rodentium in colon Body weight change (%) 0 –40 C. rodentium in spleen 3 5 4 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Time after infection (d) Time after infection (d) f **** g f/f cKO 6 100 e Notch2 Notch2 Notch2f/f 5 cKO Notch2 80 4 Batf3–/– 60 WT 3 –/– 40 *** Irf4 2 NS Ccr7–/–

Survival (%) +/+

Histology score 20 Batf2 1 NS Batf2–/– 0 0 0 5 10 15 20 25 30 35 40 45 Time after infection (d)

h +/+ WT Batf2 + 30 Figure 6 Notch2-dependent CD11b cDCs are essential for host defense against infection with C. rodentium. Irf4–/– Batf2–/– (a,b) Survival (a) and weight loss (b) of Batf3 −/−, Notch2 f/f and Notch2cKO mice (n = 9–10 per group) orally 20 Ccr7–/– inoculated with C. rodentium (2 × 109 colony-forming units). (c) C. rodentium titers in the spleen (left) and 10 colon (right) of mice (n = 6–7 per group) 9 d after inoculation as in a, presented as colony-forming units (CFU). 0 (d) Colon lengths in mice (n = 3 per group) 9 d after infection as in a. (e) Hematoxylin-and-eosin staining of –10 * * f/f cKO colon sections from Notch2 and Notch2 mice (n = 6–7 per group) 9 d after infection as in a. Scale bars, –20 200 µm. (f) Histological scores of the sections in e (and similar sections from Batf3 −/− mice). (g,h) Survival (g) Body weight change (%) –30 and weight loss (h) of mice (n = 8 (wild-type), 5 (Ccr7−/−, Batf2 −/− and Batf2 +/+) or 6 (Irf4−/−)) infected with 0 5 10 15 20 25 30 C. rodentium as in a. *P < 0.05, **P < 0.01, ***P < 0.001 (log-rank Mantel-Cox test (a,g), Student’s t-test (b,h), Time after infection (d) one-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test (c,d) or Kruskal-Wallis test with Dunn’s multiple-comparison test (f)). Data are from three independent experiments (a,b) or two independent experiments (c,d,g,h) or are representative of two independent experiments (e,f; error bars (b–d,f,h), s.e.m.).

A nonredundant role for CD11b+ cDCs in C. rodentium infection mice, whereas lamina propria CD103+CD11b+ cDCs were present As Notch2cKO and Batf3−/− mice lacked CD103+CD11b+ cDCs and but less abundant (Supplementary Fig. 8c). Irf4−/− mice survived CD103+CD11b− cDCs, respectively, we sought to determine whether infection with C. rodentium until at least day 28 (Fig. 6g,h), which either subset was specifically required for host defense against an excluded the possibility that early resistance required cDC migration. attaching-and-effacing bacterial pathogen. We compared the survival However, those mice did eventually succumb to C. rodentium by day of Notch2f/f, Batf3−/− and Notch2cKO mice after oral infection with 42 after infection (Fig. 6g,h), perhaps because of defects in antibody © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature C. rodentium. Although Notch2f/f and Batf3−/− mice were resistant, responses2,43. Next we analyzed Ccr7−/− mice, which have a general Notch2cKO mice were highly susceptible to infection, which caused defect in cDC migration26 but intact antibody responses. These mice death within 7–10 d (Fig. 6a). Unlike Notch2f/f or Batf3−/− mice, were completely resistant to infection with C. rodentium (Fig. 6g,h), cKO npg Notch2 mice underwent rapid weight loss after infection, and which suggested that protection against such infection was provided when killed at day 9, they had significantly greater pathogen bur- by the local action of CD103+CD11b+ cDCs. den and shorter colons (Fig. 6b–d and Supplementary Fig. 8a,b). To determine whether the defects in mucosal immunity in Colons of Notch2cKO mice had considerable infiltration of inflamma- Notch2cKO mice were specific to infection with C. rodentium, we tory cells and crypt elongation, scattered loss of mucosal architecture, infected these mice with T. gondii. In contrast to Batf3−/− mice, which and ulceration and coagulation necrosis, unlike those of Notch2f/f are highly susceptible and uniformly succumb by 9 d after infec- mice (Fig. 6e,f). The development of Batf3-independent CD8α+ tion22, Notch2cKO mice survived infection and were indistinguish- cDCs during infection with intracellular pathogens such as Listeria able from their wild-type counterparts (Supplementary Fig. 8d). monocytogenes has been shown to occur as a result of compensation These results demonstrated that CD103+CD11b+ cDCs were by the AP-1 factor Batf2 (ref. 41). However, CD103+CD11b− cDCs not required for resistance to T. gondii and that the function of were not restored in Batf3−/− mice during infection with C. rodentium Batf3-dependent CD103+CD11b− cDCs seemed to be unaffected in (data not shown), and Batf2−/− mice survived infection without sub- Notch2cKO mice. stantial weight loss (Fig. 6g,h), which suggested that resistance to this attaching-and-effacing pathogen did not require compensatory CD11b+ cDCs are not essential for healing mucosal wounds development of CD8α+ cDCs. We sought to determine whether the enhanced susceptibility of To determine whether CD103+CD11b+ cDCs had to migrate to Notch2cKO mice to C. rodentium reflected local defects in colonic draining lymph nodes to provide protection against C. rodentium, we wound repair44 rather than defects in pathogen-specific immunologi- studied mice deficient in the transcription factor IRF4 (Irf4−/− mice) cal defense. Colonic wound repair involves local prostaglandin produc- or the chemokine receptor CCR7 (Ccr7−/− mice)26,42. Irf4−/− mice tion by the cyclooxygenase COX-2 (encoded by Ptgs2), which supports have a selective defect in the migration of CD11b+ cDCs from epithelial proliferation required for the resolution of inflammation45. tissues to draining lymph nodes42. We confirmed that migra- We noted that CD103+CD11b+ cDCs and macrophages in the intestine tory CD103+CD11b+ cDCs were absent from the MLNs of Irf4−/− expressed Ptgs2 (Fig. 7a), which suggested that either cell type may

944 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology A rt i c l e s

a b NS c d f/f 75 Notch2 100 100 f/f f/f

) Day 2 Notch2 Day 6 Notch2 + NS-398 2 50 75 75

25 50 50 Notch2cKO index ( × 10

Ptgs2 expression 0

+ – 25 SMA loss (%) 25

+ Mφ Wound healed (%) – cDC+ cDC 0 γδT Vγγδ5T Vγ5 0 Serosal M φ f/f f/f f/f CD11b –/– + CD11b+ CD11b cKO cKO β-catenin αSMA DAPI Batf3 CD103CD103 Notch2 Notch2 Notch2 Notch2 Notch2 + NS-398 e Notch2f/f Notch2cKO Batf3–/– Figure 7 Notch2-dependent cDCs are dispensable for colonic wound repair. (a) Ptgs2 expression in various intestinal cell types, derived from the Immunological Genome Project database55, assessed by microarray. (b) Whole-mount images of Notch2 f/f and Notch2 cKO colonic wounds 6 d after excision (left), and quantification of epithelial coverage in individual wound beds at day 6 after excision, assessed in whole-mount images (right). Outlined areas (left) indicate the initial injury site. Scale bars (left), 1 mm. Each symbol (right) represents an individual β-catenin αSMA DAPI mouse (n = 6 per group); small horizontal lines indicate the mean (± s.e.m.). (c) Loss of α-smooth muscle actin (SMA) underlying wound beds at day 6 after excision from Notch2 f/f, Notch2 cKO and Batf3 −/− mice, and from Notch2 f/f mice treated with NS-398 (far right), measured in histological sections (gap length / wound bed length). Each symbol represents an individual mouse (n = 2–6 per group); small horizontal lines indicate the mean (± s.e.m.). (d) Colonic sections from an f/f f/f untreated Notch2 mouse (left) or Notch2 mice treated with β-catenin F4/80 DAPI NS-398 (5 µg/g; right), stained for β-catenin and α-smooth muscle actin (αSMA) and with DAPI on day 2 (left) or day 6 (right) after excision (n = 2–3 mice per group). Scale bars, 200 µm. (e) Colonic sections from Notch2 f/f, Notch2 cKO and Batf3 −/− mice (n = 3–6 per group) stained for β-catenin, α-smooth muscle actin and F4/80 and with DAPI at 6 d after excision. Scale bars, 200 µm. Statistical results (b, right), Student’s t-test. Data are from one experiment with two to four replicate arrays (a; mean and s.e.m.) or two independent experiments (b,c) or are representative of two independent experiments (d,e).

be involved in supporting wound repair. We used endoscopy-guided and Mup1 had lower expression in the colons of infected Notch2cKO mucosal excision to induce colonic injury and found that Notch2cKO mice than in their Notch2f/f counterparts (Fig. 8c), which indicated mice had normal wound repair, as did Notch2f/f and Batf3−/− mice, the importance of the molecules encoded by those genes in the defense in contrast to Notch2f/f mice treated with the COX-2 inhibitor against attaching-and-effacing pathogens2. Genes encoding the anti- NS-398 (Fig. 7b,c). Inhibition of COX-2 after mucosal excision microbial peptides S100A8 and S100A9 had higher expression in © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature resulted in failed regeneration of the β-catenin-positive epithelial Notch2cKO mice than in Notch2f/f mice (Fig. 8c), consistent with a pub- layer and impaired maintenance of the underlying α-smooth muscle lished report indicating the insufficiency of S100A8 and S100A9 in the actin–positive muscularis propria layer (Fig. 7d), as reported before44. defense against C. rodentium2. Consistent with the findings reported f/f cKO −/− cKO npg In contrast, Notch2 , Notch2 and Batf3 mice had normal epi- above, IL-22 production was much lower in Notch2 mice, but not in thelial regeneration and maintenance of the muscularis propria at Batf3−/− mice, 9 d after infection with C. rodentium (Fig. 8d). To elimi- day 6 after mucosal excision (Fig. 7e). Thus, cDCs were not required nate the effect of inflammation on gene expression, we also assessed for the prostaglandin production involved in mucosal healing, which changes in gene expression at day 4 after infection, a time at which no suggested that the susceptibility of Notch2cKO mice to infection with change in histology, pathogen dissemination, colonic length or death C. rodentium was not due to any defects in wound repair. had occurred (Fig. 8e and Supplementary Fig. 8e,f). Again, Notch2cKO mice had lower expression of Il22 and of the IL-22-responsive genes CD11b+ cDCs regulate IL-23-dependent mucosal immunity Reg3b and Reg3g (which, as noted above, encode antimicrobial pep- To determine the basis of the susceptibility of Notch2cKO mice to infec- tides) (Fig. 8e and Supplementary Table 1). Accordingly, ILCs iso- tion with C. rodentium, we analyzed differences in gene expression in lated from Notch2cKO mice had lower intracellular expression of IL-22 colonic cells isolated from Notch2f/f, Batf3−/− and Notch2cKO mice 9 d protein at day 4; however, that effect was ‘rescued’ by the addition of after infection (Fig. 8a). Overall, for colonic gene expression, the dif- IL-23 ex vivo (Fig. 8f), which indicated that the lower IL-22 production ference between Notch2cKO mice and Notch2f/f mice was greater than was extrinsic of ILCs in Notch2cKO mice and instead probably resulted the difference between Batf3−/− mice and Notch2f/f mice (Fig. 8a and from a defect in the stimulation of ILCs. Supplementary Table 1). Ptgs2 expression was not lower in Notch2cKO Mice deficient in the p19 subunit of IL-23 (Il23a−/− mice) are suscep- mice 9 d after C. rodentium infection (Fig. 8a,b), consistent with the tible to infection with C. rodentium1,2, and intestinal CD103+CD11b+ intact wound healing noted above (Fig. 7e). However, Notch2cKO mice cDCs have been reported to produce IL-23 in vivo after stimula- had upregulation of various genes encoding inflammatory molecules, tion of Toll-like receptor 5 (ref. 13). However, the requirement including Il1a, Il1b, Il33 and Ccl3, and substantial downregulation for those cells as an obligate source of IL-23 during infection with of the genes Reg3b and Reg3g, which encode antimicrobial peptides attaching-and-effacing bacteria remains unclear. Functional IL-23 (Fig. 8a–c). Notably, among the set of genes shown to be induced by is a heterodimeric protein composed of p19 (encoded by Il23a) and the stimulation of colonic cells with IL-22 ex vivo, only Reg3b, Reg3g p40 (encoded by Il12b). In the lamina propria, we found that Il23a

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 945 A rt i c l e s

a 14.5 14.5 e 14.5

ll33 12.5 ll1b 12.5 12.5 Sectm1b Ccl3 Ptgs2 Ptgs2 Slc37a2 f/f ) ) ) 2 10.5 Cxcl2 2 10.5 2 10.5 Spp1

colon ll1b colon

ll1a colon

cKO 8.5 8.5 cKO 8.5 Lrg1 –/– Ccl3 Reg3g Cxcl2 Ccl3 Reg3b S100a9 6.5 Reg3b Batf3 6.5 Spp1 ll33 6.5 Reg3b Notch2 Notch2

Expression (log Expression (log Expression (log S100a8 Reg3g ll1a Reg3g ll1a KO 4.5 4.5 4.5 Cxcl5 ll22 2.5 2.5 2.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5 2.5 4.5 6.5 8.5 10.5 12.5 14.5

Expression (log2) Expression (log2) Expression (log2) Notch2f/f colon Notch2f/f colon Notch2f/f colon

Notch2f/f Notch2f/f Notch2cKO b 20 c 9 f (UI) (day 4) (day 4) g 1.5 – – – + 8 B220 CD3 NK1.1 CD90 7 15 1.0 6 2.1 ± 0.2 15 ± 6.2 1.4 ± 0.1* 5 – IL-23 10

4 ll23a (AU) 0.5 KO vs f/f f/f vs KO 3 105 Expression (fold) 5 Expression (fold) 2 4 5.2 ± 0.3 15 ± 6.4 5.4 ± 0.3 10 ND ND 1 + IL-23 3 0 10 CD11b+ MΦ CD11b+ MΦ 0 0 0 cDC cDC Ltf CD4 ll33 lrg1 ll1a ll1b Cfd Hp Ccl3 Ccl8Saa3Expi ligp1 Arg1 Cxcl3Ptgs2 Mmp7 Cxcr2 Mmp8 Retnlg Spp1Cxcl2 Mup1 Cldn1 3 4 5 Ly6g6c Clec4e Reg3gReg3b S100a8 S100a9 0 10 10 10 UI Day 2 IL-22

f/f + d 0.75 ** 2.0 * Notch2 h 100 Figure 8 Notch2-dependent CD11b cDCs cKO regulate IL-23-dependent antimicrobial responses * ** Notch2 Batf3–/– 80 to . (a) Microarray analysis of gene 1.5 C. rodentium 0.50 expression in colonic cells from mice 9 d after 60 1.0 –/– infection with C. rodentium, presented as M-plots. II23a → WT NS II23a–/– + Notch2cKO → WT Colors indicate higher (red) or lower (blue) Il22 (AU) 40 –/– f/f Reg3g (AU)

0.25 Survival (%) II23a + Notch2 → WT *** expression in Notch2 cKO cells (left) or Batf3 −/− 0.5 WT → WT 20 cells (right) than in Notch2 f/f cells. (b,c) Expression of selected inflammatory molecule–encoding genes 0 0 0 from a with higher expression in Notch2 cKO (KO) 0 5 10 15 20 25 30 mice than in Notch2 f/f (f/f) mice (b), and expression Time after infection (d) of IL-22-stimulated genes2 (c). (d) Quantitative RT-PCR analysis of Il22 and Reg3g mRNA in colons from Notch2 f/f, Notch2 cKO and Batf3 −/− mice (n = 6–7 per group) 9 d after infection with C. rodentium, presented in arbitrary units (AU) relative to expression of the housekeeping gene Hprt. (e) Microarray analysis of gene expression in colonic cells from Notch2 f/f and Notch2 cKO mice 4 d after infection with C. rodentium, presented as an M-plot (left), and hematoxylin-and-eosin staining of colons from those mice (n = 4 per group; right). Scale bars (right), 100 µm. (f) Intracellular IL-22 © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature expression in ILCs obtained from the MLNs of uninfected Notch2 f/f mice (UI) or Notch2 f/f and Notch2 cKO mice (n = 4 per group) 4 d after infection with C. rodentium (day 4) and then stimulated ex vivo with the phorbol ester PMA and ionomycin in the presence (+ IL-23) or absence (− IL-23) of IL-23 (pregating, above plots). (g) Quantitative RT-PCR analysis of Il23a mRNA in CD11b+ cDCs (CD11c+MHCII+CD103+CD11b+) and macrophages (CD11c+MHCII−CD11b+F4/80+) sorted from the lamina propria of wild-type uninfected mice (UI) or wild-type mice 2 d after infection with C. rodentium npg (day 2), presented as in d (n = 3 mice per group). (h) Survival of mixed chimeras (n = 5–6 per group) generated by the reconstitution of wild-type mice with Il23a −/− bone marrow (Il23a −/− → WT), a mixture of Il23a−/− and Notch2 cKO bone marrow (Il23a −/−+Notch2 cKO → WT) or Il23a −/− and Notch2 f/f bone marrow (Il23a −/− + Notch2 f/f → WT), or wild-type bone marrow (WT → WT), orally inoculated with 2 × 109 colony-forming units C. rodentium 8–10 weeks after lethal irradiation and transplantation. ND, not detected. *P < 0.05, **P < 0.01 and ***P < 0.001 (one-way ANOVA with Tukey’s multiple-comparison test (d), Student’s t-test (f) or log-rank Mantel-Cox test (h)). Data are from one experiment (a–c; average of two to three biological replicates per sample), three independent experiments (d; bars, s.e.m.), one experiment with two biological replicates per sample (e, left) or two independent experiments (g,h; error bars (g), s.e.m.) or are representative of two independent experiments (e, right, f).

was expressed specifically in CD11b+CD103+ cDCs in the steady and Il23a−/− bone marrow and analyzed resistance to C. rodentium. state and 2 d after infection by C. rodentium, but this transcript was In Notch2cKO–Il23a−/− mixed chimeras, intestinal CD103+CD11b+ undetectable in macrophages in either setting (Fig. 8g). In addition, cDCs developed only from the Il23a−/− bone marrow, but all other we found that intracellular expression of p40 was undetectable in hematopoietic populations were unaffected in their ability to gen- CD11b+ cDCs at steady state but increased substantially after acti- erate IL-23. The survival of Notch2f/f–Il23a−/− chimeras after infec- vation (Supplementary Fig. 8g). The inducible expression of Il12b tion by C. rodentium was similar to that of chimeras generated by (encoding p40) was not dependent on the presence of CD8α+ cDCs transplantation of wild-type bone marrow alone; however, the sur- or on the transcription factors Batf, Batf2 or Batf3 (Supplementary vival of Notch2cKO–Il23a−/− chimeras was similar to that of chimeras Fig. 8g,h). Thus, expression of IL-23 was both inducible after activa- generated with Il23a−/− bone marrow alone, which succumbed to tion and restricted to the CD11b+ cDC subset. C. rodentium at 7–11 d after infection (Fig. 8h). The observation To determine whether Notch2-dependent cDCs were the obligate that Notch2cKO–Il23a−/− mixed chimeras were as susceptible to source of IL-23 during infection with C. rodentium, we generated C. rodentium infection as were chimeras generated with Il23a−/− bone mixed chimeras by reconstituting wild-type mice with Notch2f/f bone marrow alone suggested that Notch2-dependent CD11b+CD103+ marrow and Il23a−/− bone marrow or with Notch2cKO bone marrow cDCs were the critical source of IL-23 required for protection.

946 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology A rt i c l e s

DISCUSSION monocyte-derived cells31 can produce IL-6 (ref. 52), which could sup- Efforts to distinguish the functions of macrophages and DCs have port CD4+ T cell–derived IL-22 production later during infection53. been limited by the availability of systems for the selective depletion of Our observation that Ccr2−/− mice had diminished mortality relative each lineage in vivo46. Genetic models have been generated that have to that of Notch2cKO mice, yet still lost weight, is also in agreement led to the characterization of nonredundant roles for CD8α+ cDCs with a model in which monocytes are important in immune responses and for pDCs17–19, but a unique role for CD11b+ cDCs has not been to C. rodentium. Thus, future studies should characterize the non studied by a selective depletion model in vivo so far, to our knowl- redundant functions of pre-cDC–derived and monocyte-derived cells edge. Here we used the Notch2 dependence of CD11b+ cDCs to ana- in resistance to infection with attaching-and-effacing bacteria. lyze their role in host defense. We observed that Notch2-dependent The role of cDCs in immunological defense has been thought CD103+CD11b+ cDCs were required for IL-23 production to protect to reside mainly in their ability to prime the responses of CD4+ the host from early susceptibility to infection with C. rodentium, but T cells and CD8+ T cells54. Our results presented here, along with the Batf3-dependent CD103+CD11b− cDCs were not. That role in innate similar requirement for CD8α+ cDCs in IL-12 production for early defense was pathogen specific, as CD103+CD11b+ cDCs were not innate protection against T. gondii, demonstrate that both branches of necessary for resistance to T. gondii or for the healing of intestinal cDCs are also critical for the initiation of innate immunity. These mucosa after injury. These findings settle an unresolved question findings may indicate that the DC lineage separated from monocytes about the critical source of IL-23 required for driving IL-22-dependent and macrophages before the emergence of adaptive immunity and that antimicrobial responses to infection with attaching-and-effacing it has a dedicated role in orchestrating the reactions of the expanding pathogens2. As IL-22 also acts in other host-defense processes, such as family of ILCs3. in chronic inflammatory skin disease6, future studies should examine the requirement for CD11b+ cDCs in those settings as well. Methods Given our observation that Notch2 also regulated gene expression Methods and any associated references are available in the online in CD8α+ cDCs, we investigated whether defects in these cDCs could version of the paper. account for early susceptibility to C. rodentium. However, Batf3−/− and Batf2−/− mice showed normal resistance to C. rodentium. We Accession code. GEO: microarray data, GSE45698. also excluded the possibility of a role for other Notch2-independent + Note: Any Supplementary Information and Source Data files are available in the cDCs. In the absence of Notch2, some CD11b cDCs developed online version of the paper. in the spleen; likewise, Zbtb46-expressing CD103−CD11b+ cDCs may have also developed in the intestine. However, any ‘Notch2- Acknowledgments −/− inexperienced’ CD11b+ cDCs remaining in Notch2cKO mice were We thank B. Sleckman (Washington University in St. Louis) for Nik mice; T. Watts (University of Toronto) for Ifnar1−/− mice; J. Boothroyd (Stanford insufficient for IL-23 production and protection against infection University) for the plasmid PRU-FLuc-GFP; the Immunological Genome Project with C. rodentium. Notch signaling also regulates development of consortium for use of their database54; and the Alvin J. Siteman Cancer Center at IL-22-producing ILCs47,48, but we found that ILC function was not Washington University School of Medicine for use of the Center for Biomedical affected by CD11c-Cre–mediated deletion of Notch2. Nonetheless, Informatics and Multiplex Gene Analysis Genechip Core Facility. Supported by this dual role of Notch signaling in defense against infection with the Howard Hughes Medical Institute, the US National Institutes of Health (AI076427-02 to K.M.M., R01 GM55479 to R.K., R01 DE021255-01 and U01 attaching-and-effacing pathogens might represent a coordinated AI095542-01 to M.C., R01 DK071619 to T.S.S. and R01 DK064798 to R.D.N.), immunological strategy involving the intentional expression of Notch the US Department of Defense (W81XWH-09-1-0185 to K.M.M.), the American © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature ligands in pathogen-responsive cellular niches. Heart Association (12PRE8610005 to A.T.S. and 12PRE12050419 to W.K.), the Antibody blockade of LTβR in vivo has also been found to dimin- Canadian Institutes of Health Research (MOP 67157 to J.L.G. and FRN 11530 to 49 C.J.G.) and the National Cancer Institute (P30 CA91842 for the Alvin J. Siteman ish IL-23 production during infection with C. rodentium , and Cancer Center).

npg conditional deletion of Ltbr by CD11c-Cre increases susceptibility to this pathogen11,50. In addition, a published study has reported AUTHOR CONTRIBUTIONS a requirement for LTβR in the homeostasis of CD4+ cDCs in the A.T.S. and K.M.M. designed the study; A.T.S., C.G.B., J.S.L. and C.S. did 38 experiments related to infection with C. rodentium, with guidance from W.O., spleen . Such results would suggest a possible relationship between M.C. and K.M.M.; A.T.S., C.G.B. and D.N. did experiments related to Ltbr−/− mice, signaling via Notch2 and signaling via LTβR in cDC differentiation. with guidance from C.J.G. and J.L.G.; N.A.M. did experiments related to wound Indeed, we found that LTβR selectively influenced the development healing, with guidance from T.S.S.; A.T.S. and X.W. did microarray analysis; A.T.S., of splenic ESAM+ cDCs and intestinal CD103+CD11b+ cDCs, but S.R.T., W.K., W.-L.L., M.T., T.L.M. and K.G.M. did experiments related to cDC development in mice deficient in Notch2, Irf4 or Batf3, with guidance from R.K., Flt3L did not. It has been shown that loss of LTβR signaling directly gfp 11 R.D.N. and K.M.M.; A.T.S., C.G.B. and M.M.M. did experiments with Zbtb46 impairs Il23a expression by cDCs . However, our results suggest that and Zbtb46DTR mice, with guidance from M.C.N. and K.M.M.; and A.T.S. and LTβR signaling may also control the development of CD103+CD11b+ K.M.M. wrote the manuscript with contributions from all authors. cDCs, which are a source of IL-23, independently of any effects on Il23a expression. COMPETING FINANCIAL INTERESTS + The authors declare competing financial interests: details are available in the online The tissue-resident CD11b cDCs critical for early defense against version of the paper. C. rodentium were affected in Zbtb46DTR, Notch2cKO and Flt3l−/− mice and thus seemed to derive from pre-cDCs, rather than from Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. monocytes. However, our demonstration that Notch2-dependent CD11b+ cDCs were required for IL-23-dependent protection 1. Mangan, P.R. et al. Transforming growth factor-β induces development of the TH17 against C. rodentium does not exclude the possibility of contribu- lineage. Nature 441, 231–234 (2006). tions from macrophages and monocyte-derived cells. Indeed, 2. Zheng, Y. et al. Interleukin-22 mediates early host defense against attaching and monocytes serve a protective role in the eradication of C. rodentium effacing bacterial pathogens. Nat. Med. 14, 282–289 (2008). 3. Spits, H. & Di Santo, J.P. The expanding family of innate lymphoid cells: regulators after their recruitment to the lamina propria during later stages of and effectors of immunity and tissue remodeling. Nat. Immunol. 12, 21–27 infection51. So, whereas cDC-derived IL-23 may be critical early, (2011).

nature immunology VOLUME 14 NUMBER 9 SEPTEMBER 2013 947 A rt i c l e s

4. Colonna, M. Interleukin-22-producing natural killer cells and lymphoid tissue 30. Boring, L. et al. Impaired monocyte migration and reduced type 1 (Th1) cytokine inducer-like cells in mucosal immunity. Immunity 31, 15–23 (2009). responses in C–C chemokine receptor 2 knockout mice. J. Clin. Invest. 100, 5. Sonnenberg, G.F. et al. CD4+ lymphoid tissue-inducer cells promote innate immunity 2552–2561 (1997). in the gut. Immunity 34, 122–134 (2011). 31. Zigmond, E. et al. Ly6C hi monocytes in the inflamed colon give rise to 6. Zheng, Y. et al. Interleukin-22, a TH17 cytokine, mediates IL-23-induced dermal proinflammatory effector cells and migratory antigen-presenting cells.Immunity 37, inflammation and acanthosis. Nature 445, 648–651 (2007). 1076–1090 (2012). 7. Mundy, R. et al. Citrobacter rodentium of mice and man. Cell. Microbiol. 7, 32. Dudziak, D. et al. Differential antigen processing by dendritic cell subsets in vivo. 1697–1706 (2005). Science 315, 107–111 (2007). 8. Cella, M. et al. A human natural killer cell subset provides an innate source of 33. Radtke, F., Fasnacht, N. & MacDonald, H.R. Notch signaling in the immune system. IL-22 for mucosal immunity. Nature 457, 722–725 (2009). Immunity 32, 14–27 (2010). 9. Eberl, G. et al. An essential function for the nuclear receptor RORγ(t) in the 34. Edelson, B.T. et al. Peripheral CD103+ dendritic cells form a unified subset generation of fetal lymphoid tissue inducer cells. Nat. Immunol. 5, 64–73 developmentally related to CD8α+ conventional dendritic cells. J. Exp. Med. 207, (2004). 823–836 (2010). 10. Sanos, S.L. et al. RORgammat and commensal microflora are required for 35. McDonald, K.G. et al. Dendritic cells produce CXCL13 and participate in the the differentiation of mucosal interleukin 22-producing NKp46+ cells. Nat. development of murine small intestine lymphoid tissues. Am. J. Pathol. 176, Immunol. 10, 83–91 (2009). 2367–2377 (2010). 11. Tumanov, A.V. et al. Lymphotoxin controls the IL-22 protection pathway in gut 36. Caton, M.L., Smith-Raska, M.R. & Reizis, B. Notch-RBP-J signaling controls the innate lymphoid cells during mucosal pathogen challenge. Cell Host Microbe 10, homeostasis of CD8− dendritic cells in the spleen. J. Exp. Med. 204, 1653–1664 44–53 (2011). (2007). + 12. Manta, C. et al. CX3CR1 macrophages support IL-22 production by innate lymphoid 37. Waskow, C. et al. The receptor tyrosine kinase Flt3 is required for dendritic cells during infection with Citrobacter rodentium. Mucosal Immunol. 6, 177–188 cell development in peripheral lymphoid tissues. Nat. Immunol. 9, 676–683 (2012). (2008). 13. Kinnebrew, M.A. et al. Interleukin 23 production by intestinal CD103+CD11b+ 38. Kabashima, K. et al. Intrinsic lymphotoxin-β receptor requirement for homeostasis dendritic cells in response to bacterial flagellin enhances mucosal innate immune of lymphoid tissue dendritic cells. Immunity 22, 439–450 (2005). defense. Immunity 36, 276–287 (2012). 39. Summers deLuca, L. & Gommerman, J.L. Fine-tuning of dendritic cell biology by 14. Bennett, C.L. & Clausen, B.E. DC ablation in mice: promises, pitfalls, and the TNF superfamily. Nat. Rev. Immunol. 12, 339–351 (2012). challenges. Trends Immunol. 28, 525–531 (2007). 40. Fütterer, A. et al. The lymphotoxin β receptor controls organogenesis and affinity 15. Meredith, M.M. et al. Expression of the zinc finger transcription factor zDC (Zbtb46, maturation in peripheral lymphoid tissues. Immunity 9, 59–70 (1998). Btbd4) defines the classical dendritic cell lineage. J. Exp. Med. 209, 1153–1165 41. Tussiwand, R. et al. Compensatory dendritic cell development mediated by BATF-IRF (2012). interactions. Nature 490, 502–507 (2012). 16. Satpathy, A.T. et al. Zbtb46 expression distinguishes classical dendritic cells 42. Bajaña, S. et al. IRF4 promotes cutaneous dendritic cell migration to lymph nodes and their committed progenitors from other immune lineages. J. Exp. Med. 209, during homeostasis and inflammation. J. Immunol. 189, 3368–3377 (2012). 1135–1152 (2012). 43. Klein, U. et al. Transcription factor IRF4 controls plasma cell differentiation and 17. Hildner, K. et al. Batf3 deficiency reveals a critical role for CD8α+ dendritic cells class-switch recombination. Nat. Immunol. 7, 773–782 (2006). in cytotoxic T cell immunity. Science 322, 1097–1100 (2008). 44. Manieri, N.A. et al. Igf2bp1 is required for full induction of Ptgs2 mRNA in colonic 18. Lewis, K.L. et al. Notch2 receptor signaling controls functional differentiation of mesenchymal stem cells in mice. Gastroenterology 143, 110–121 (2012). dendritic cells in the spleen and intestine. Immunity 35, 780–791 (2011). 45. Brown, S.L. et al. Myd88-dependent positioning of Ptgs2-expressing stromal 19. Swiecki, M. et al. Plasmacytoid dendritic cell ablation impacts early interferon cells maintains colonic epithelial proliferation during injury. J. Clin. Invest. 117, responses and antiviral NK and CD8+ T cell accrual. Immunity 33, 955–966 258–269 (2007). (2010). 46. Satpathy, A.T. et al. Re(de)fining the dendritic cell lineage. Nat. Immunol. 13, 20. Hashimoto, D., Miller, J. & Merad, M. Dendritic cell and macrophage heterogeneity 1145–1154 (2012). in vivo. Immunity 35, 323–335 (2011). 47. Possot, C. et al. Notch signaling is necessary for adult, but not fetal, development 21. Torti, N. et al. Batf3 transcription factor-dependent DC subsets in murine CMV of RORγt+ innate lymphoid cells. Nat. Immunol. 12, 949–958 (2011). infection: differential impact on T-cell priming and memory inflation. Eur. J. 48. Lee, J.S. et al. AHR drives the development of gut ILC22 cells and postnatal Immunol. 41, 2612–2618 (2011). lymphoid tissues via pathways dependent on and independent of Notch. Nat. 22. Mashayekhi, M. et al. CD8a+ dendritic cells are the critical source of interleukin-12 Immunol. 13, 144–151 (2012). that controls acute infection by Toxoplasma gondii tachyzoites. Immunity 35, 49. Ota, N. et al. IL-22 bridges the lymphotoxin pathway with the maintenance of 249–259 (2011). colonic lymphoid structures during infection with Citrobacter rodentium. Nat. 23. Cervantes-Barragan, L. et al. Plasmacytoid dendritic cells control T-cell response Immunol. 12, 941–948 (2011). to chronic viral infection. Proc. Natl. Acad. Sci. USA 109, 3012–3017 (2012). 50. Wang, Y. et al. Lymphotoxin beta receptor signaling in intestinal epithelial © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature 24. Bogunovic, M. et al. Origin of the lamina propria dendritic cell network. Immunity cells orchestrates innate immune responses against mucosal bacterial infection. 31, 513–525 (2009). Immunity 32, 403–413 (2010). 25. Varol, C. et al. Intestinal lamina propria dendritic cell subsets have different origin 51. Kim, Y.G. et al. The Nod2 sensor promotes intestinal pathogen eradication via the and functions. Immunity 31, 502–512 (2009). chemokine CCL2-dependent recruitment of inflammatory monocytes. Immunity 34, npg 26. Ohl, L. et al. CCR7 governs skin dendritic cell migration under inflammatory and 769–780 (2011). steady-state conditions. Immunity 21, 279–288 (2004). 52. Rivollier, A. et al. Inflammation switches the differentiation program ofhi Ly6C 27. Jakubzick, C. et al. Lymph-migrating, tissue-derived dendritic cells are minor monocytes from antiinflammatory macrophages to inflammatory dendritic cells in constituents within steady-state lymph nodes. J. Exp. Med. 205, 2839–2850 the colon. J. Exp. Med. 209, 139–155 (2012). (2008). 53. Basu, R. et al. Th22 cells are an important source of IL-22 for host protection 28. Randolph, G.J., Ochando, J. & Partida-Sanchez, S. Migration of dendritic cell against enteropathogenic bacteria. Immunity 37, 1061–1075 (2012). subsets and their precursors. Annu. Rev. Immunol. 26, 293–316 (2008). 54. Steinman, R.M. Decisions about dendritic cells: past, present, and future. Annu. 29. McKenna, H.J. et al. Mice lacking flt3 ligand have deficient hematopoiesis affecting Rev. Immunol. 30, 1–22 (2011). hematopoietic progenitor cells, dendritic cells, and natural killer cells. Blood 95, 55. Heng, T.S. & Painter, M.W. The Immunological Genome Project: networks of gene 3489–3497 (2000). expression in immune cells. Nat. Immunol. 9, 1091–1094 (2008).

948 VOLUME 14 NUMBER 9 SEPTEMBER 2013 nature immunology ONLINE METHODS Antibodies and flow cytometry. Samples were stained at 4 °C in the Mice. All animals were bred and maintained in a specific pathogen-free animal presence of Fc Block (2.4G2; BD Biosciences) in flow cytometry buffer facility according to institutional guidelines and with protocols approved by (DPBS plus 0.5% BSA plus 2 mm EDTA). The following antibodies the Animal Studies Committee at Washington University in St. Louis. The were from BD Biosciences: allophycocyanin-conjugated antibody to generation of Zbtb46gfp and Zbtb46DTR mice has been described15,16. For CD4 (anti-CD4 (RM4-5)), V450–anti-GR1 (RB6-8C5), fluorescein depletion of cDCs from Zbtb46DTR mice, diphtheria toxin (40 ng/g; Sigma) isothiocyanate–anti-CD3ε (145-2C11), fluorescein isothiocyanate–anti- was administered on day −3 and day −1 before analysis on day 0. For deple- CD45.1 (A20), fluorescein isothiocyanate–anti-CD21/CD35 (7G6), tion of cDCs during infection with C. rodentium , mice were injected with phycoerythrin-indotricarbocyanine–anti-CD24 (M1/69), phycoerythrin- 20 ng/g diphtheria toxin 1 d before infection, and doses of 5 ng/g were given indotricarbocyanine–anti-CD8α (53-6.7) and allophycocyanin–anti-CD172a every 2–3 d thereafter to maintain depletion. Mice of the following genotypes (anti-SIRPα; P84). The following antibodies were from eBioscience: peridinin were from Jackson Laboratories: CD11c-cre (B6.Cg-Tg(Itgax-cre)1-1Reiz/J), chlorophyll protein–cyanine 5.5–anti-CD11b (M1/70), allophycocyanin– Vav1-cre (B6.Cg-Tg(Vav1-cre)A2Kio/J), Notch2f/f (B6.129S-Notch2tm3Grid/J), anti-CD90.2 (53-2.1), allophycocyanin–eFluor 780–anti-CD11c (N418), Ccr2−/− (B6.129S4-Ccr2tm1Ifc/J), Cd40−/− (B6.129P2-Cd40tm1Kik/J), Nfkb1−/− phycoerythrin–anti-CD23 (B3B4), phycoerythrin–anti-RORγt (AFKJS-9), (B6;129P-Nfkb1tm1Bal/J), Ccr7−/− (B6.129P2(C)-Ccr7tm1Rfor/J) and Cx3cr1gfp phycoerythrin–anti-IL22 (1H8PWSR), phycoerythrin–anti-IL12/23 p40 (B6.129P-Cx3cr1tm1Litt/J). The generation of Rbpjf/f, Psen1f/f and Psen2−/− mice (C17.8), phycoerythrin–anti-ESAM (1G8), eFluor 450–anti-MHCII (I-A–I-E; has been described56,57. Flt3l−/− (C57BL/6-flt3Ltm1Imx) mice were obtained M5/114.15.2), phycoerythrin–anti-CD103 (2E7), peridinin chlorophyll from Taconic. The generation of Il23−/− mice2, Ltbr−/− mice40 and Batf−/−, protein–cyanine 5.5–anti-CD16/32 (93), allophycocyanin–anti-CD45.2 (104), Batf2−/− and Batf3−/− mice17,41 has been described. The generation of Nik−/− allophycocyanin–eFluor 780–anti-CD117 (2B8), phycoerythrin–anti-CD135 mice has been described58; these mice were a gift from B. Sleckman. Ifnar1−/− (A2F10), V500–anti-B220 (RA3-6B2), allophycocyanin–anti-F4/80 (B-M8), mice were a gift from T. Watts. For the generation of Ccr2-eGFP ‘knock-in’ mice, Alexa Fluor 700–anti-Sca1 (D7) and peridinin chlorophyll protein–eFluor710– a targeting construct was designed to insert a DNA fragment encoding anti-Siglec-H (eBio440C). Phycoerythrin– and allophycocyanin–conjugated enhanced GFP (eGFP), followed by a polyadenylation signal and a loxP- anti-CD205 (anti-DEC-205; NLDC-145) were from Miltenyi. Biotin anti- flanked neomycin-resistance cassette at the translation start site of Ccr2. The CD127 (A7R34) was from BioXCell. In general, all antibodies were used at construct was transfected by electroporation into LK-1 (C57BL/6J) embry- a dilution of 1:200; anti-DEC-205 was used at a dilution of 1:20. Cells were onic stem cells59, and one correctly targeted clone identified by Southern blot analyzed on a FACSCanto II or FACSAria II (BD) and data were analyzed analysis was used for the generation of chimeras. Chimeras were bred to mice with FlowJo software (TreeStar). For immunofluorescence histology experi- with transgenic expression of Cre recombinase from the cytomegalovirus ments, the following reagents were used: rabbit polyclonal antibody to GFP (CMV) promoter (B6.C-Tg(CMV-Cre)1Cgn/J, Jackson Laboratories)60 for (A11122), Alexa Fluor 488–conjugated antibody to rabbit IgG (A11034) and removal of the neomycin-resistance cassette. For the generation of Irf4−/− mice, streptavidin–Alexa Fluor 555 (S32355; all from Invitrogen); Alexa Fluor 647– Irf4f/f mice (B6.129S1-Irf4tm1Rdf/J; Jackson Laboratories) were crossed with anti-CD11c (N418; BioLegend); biotin–anti-F4/80 (BM8; Caltag); Alexa Fluor CMV-Cre mice. For bone marrow–chimera experiments, the CD45.1+ B6.SJL 488– and biotin-conjugated anti-B220 (RA3-6B2), biotin–anti-IgD (11-26) (B6.SJL-PtprcaPepcb/BoyJ) mice were from Jackson Laboratories. Unless oth- and biotin–anti-MADCAM-1 (MECA-367; all from eBioscience); and rabbit erwise indicated, experiments used sex-matched littermates at 8–12 weeks of polyclonal antibody to β-catenin (C2206) and indocarbocyanine–conjugated age. Unirradiated mice used for C. rodentium experiments weighed less than antibody to α-smooth muscle actin (1A4; both from Sigma). 20 g when infected, unless specified otherwise. C. rodentium. Mice were infected by intraoral inoculation of 2 × 109 DC preparation. DCs were collected from lymphoid organs and nonlym- colony-forming units of C. rodentium, strain DBS100 (American Type phoid organs and were prepared as described16. Spleens, MLNs, skin-draining Culture Collection) as described2. Survival and weight loss were monitored (inguinal) lymph nodes and kidneys were minced and digested for 30 min at for 30–45 d. Survival studies were done in accordance with institutional 37 °C, with stirring, in 5 ml Iscove’s modified Dulbecco’s medium plus 10% FCS guidelines and with protocols approved by the Animal Studies Committee at © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature (cIMDM) with 250 µg/ml collagenase B (Roche) and 30 U/ml DNase I (Sigma- Washington University in St. Louis. Colon lengths were measured for mice Aldrich). Cells were passed through a 70-µm strainer before red blood cells infected for 4 or 9 d. For histology, distal colons were collected and fixed over- were lysed with ammonium chloride–potassium bicarbonate lysis buffer. Cells night at 25 °C in 10% buffered formalin phosphate (4% (wt/wt) formaldehyde, 6 6 npg were counted on a Vi-CELL analyzer (Beckman Coulter), and 5 × 10 to 10 × 10 0.4% (wt/vol) sodium phosphate (monobasic monohydrate), 0.65% (wt/vol) cells were used for each antibody-staining reaction. Lung-cell suspensions were sodium phosphate (dibasic anhydrous) and 1.5% (wt/vol) stabilizer metha- prepared after perfusion by injection of 10 ml Dulbecco’s PBS (DPBS) into the nol; SF100-4; Fisher Scientific), embedded in paraffin, sectioned and stained right ventricle after transection of the lower aorta. Dissected and minced lungs with hematoxylin and eosin. For analysis of colony-forming units, spleens and were digested for 1 h at 37 °C, with stirring, in 5 ml cIMDM with 4 mg/ml col- colons from mice infected for 4 or 9 d were weighed and homogenized and lagenase D (Roche). Suspensions of cells from the small intestine were prepared serial dilutions were plated for 24 h at 37 °C in duplicate onto MacConkey agar after removal of Peyer’s patches and fat. Intestines were opened longitudinally, plates (M7408; Sigma). The severity of colitis (by histology of samples assessed washed of fecal contents and cut into pieces 1 cm in length and were incub by a researcher ‘single-blinded’ to sample identity) was assigned a score on ated for 40 min at 37 °C, with rotation at 1g, in Hank’s balanced-salt solution a scale from 1 to 7, which corresponded to the following: 1, no evidence of (Life Technologies) plus 5 mM EDTA. Tissue pieces were washed in DPBS and inflammation; 2, little cellular infiltration in <10% of the field; 3, moder- minced and were incubated for 90 min at 37 °C, with stirring, in 25 ml RPMI ate cellular infiltration in 10–25% of the field, crypt elongation, bowel-wall medium plus 2% FBS with collagenase VIII (100 U/ml; C2139; Sigma). Cell thickening and no evidence of ulceration; 4, substantial cellular infiltration suspensions were pelleted, suspended in 40% Percoll (P4937; Sigma), overlaid in 25–50% of field, thickening of the bowel wall beyond the muscular layer on 70% Percoll and centrifuged for 20 min at 850g. Cells in the interphase and high vascular density; 5, substantial infiltration in >50% of field, crypt were recovered, washed in DPBS and stained. For DC population expansion elongation and distortion and transmural bowel thickening with ulceration; in vivo, 10 µg Flt3L (250-31L; PeproTech) was injected intraperitoneally for 6, complete loss of mucosal architecture with ulceration and loss of mucosal two consecutive days and organs were collected after 7 d. For induction of p40, vasculature; and 7, coagulation necrosis. For analysis of IL-22 expression in splenocyte samples were enriched for CD11c+ cells through the use of MACS ILCs, MLN cells isolated from mice infected for 4 d were stimulated for 4 h beads (Miltenyi). After enrichment, cells were stimulated for 24 h ex vivo with with 50 ng/ml PMA (phorbol 12-myristate 13-acetate; P1585; Sigma) and 1 µg/ml LPS (L2630; Sigma) and 50 ng/ml interferon-γ (315-05; PeproTech). 1 µM ionomycin (I0634; Sigma), with or without 10 ng/ml recombinant IL-23 Brefeldin A (1 µg/ml) was added for the final 4 h. Cells were then washed, (1887-ML-010; R&D Systems), in the presence of 1 µg/ml brefeldin A (B6542; stained for the expression of surface markers (antibodies identified below), Sigma). Cells were washed with DPBS and stained for surface expression of permeabilized with 0.25% saponin and stained for intracellular p40. specified markers (antibodies identified above). Cells were then fixed in 2%

doi:10.1038/ni.2679 nature immunology methanol-free paraformaldehyde (PFA), permeabilized with 0.25% saponin singular value decomposition without additional centering or scaling. Scores and stained for intracellular IL-22. were plotted in R software.

T. gondii. The type II Prugniaud strain of T. gondii expressing a transgene Quantitative RT-PCR. Gene expression was analyzed in colonic cells isolated encoding firefly luciferase and GFP (PRU-FLuc-GFP) used was provided by from mice infected for 2 or 9 d with C. rodentium. Bone marrow–derived J. Boothroyd. The parasites were grown in human foreskin fibroblasts cultures DCs cultured for 9 d with Flt3L (200 ng/ml) were stimulated with LPS and as described61. For infection, freshly egressed parasites were filtered, counted, interferon-γ as described above. RNA and cDNA were prepared with an and 100 tachyzoites were injected intraperitoneally into mice. Survival was RNeasy Mini kit (Qiagen), Superscript III reverse transcriptase (Invitrogen) monitored for 30 d. and Oligo(dT)20 Primer (Invitrogen). For analysis of Il23a expression in cDCs and macrophages, CD11b+ cDCs (Aqua−CD45+B220−CD11c+MH Endoscopy-guided mucosal excision. Mice were anesthetized, and colon CII+CD103+CD11b+) and macrophages (Aqua−CD45+B220−CD11c+MH lumens were visualized with a high-resolution miniaturized colonoscope system. CII−CD11b+F4/80+) from mice infected for 2 d were sorted to a purity of After inflation of the colon with DPBS, 3F flexible biopsy forceps were inserted >95%. Total RNA was isolated from sorted cells with an RNAqueous-Micro into the sheath adjacent to the camera. Three to five full-thickness areas of kit (Ambion). The StepOnePlus Real-Time PCR system was used accord- the entire mucosa and submucosa were removed from along the dorsal side ing to manufacturer’s instructions (Applied Biosystems) with the relative of the colon. For this study, wounds that averaged approximately 1 mm2 quantitation standard-curve method with HotStart-IT SYBR Green qPCR (equivalent to 250–300 crypts) were evaluated. Wounded mice were killed Master Mix (Affymetrix/USB). PCR conditions were as follows: 10 min at 2 or 6 d after injury, and each wound was frozen individually in optimal cutting 95 °C, followed by 40 two-step cycles consisting of 15 s at 95 °C and 1 min at temperature compound. Sections were prepared and fixed in 4% PFA, boiled 60 °C. Primers used for measurement of Il22, Reg3g, Il23a, Il12a, and Il12b in 10 mmol/l citrate buffer and rinsed in DPBS, then nonspecific binding was expression were as follows: IL22-F, 5′-TGACGACCAGAACATCCAGA-3′; blocked by incubation for 20 min with 3% bovine serum albumin and 0.1% IL22-R, 5′-CGCCTTGATCTCTCCACTCT-3′; HPRT-F, 5′-TCAGTCA Triton X-100, followed by incubation for 1 h with primary antibodies (identi- ACGGGGGACATAAA-3′; HPRT-R, 5′-GGGGCTGTACTGCTTAACCAG-3′; fied above). Slides were then rinsed with DPBS, incubated with secondary IL23a-F, 5′-AATAATGTGCCCCGTATCCA-3′; IL23a-R, 5′-GGATCCTTT antibodies (identified above), stained with bis-benzimide and then mounted GCAAGCAGAAC-3′; IL12a-F: 5′-GTGAAGACGGCCAGAGAAA-3′; IL12a-R, with Mowiol 4-88 (EMD Chemicals). Sections were viewed with a Zeiss 5′-GGTCCCGTGTGATGTCTTC-3′; IL12b-F, 5′-AGCAGTAGCAGTTC (Oberkochen, Germany) Axiovert 200 microscope equipped with an Axiocam CCCTGA-3′; IL12b-R, 5′-AGTCCCTTTGGTCCAGTGTG-3′; Reg3g-F, MRM camera. The selective COX-2 inhibitor NS-398 (Cayman Chem) was 5′-ATCATGTCCTGGATGCTGCT-3′; and Reg3g-R, 5′-AGATGGGGCA dissolved in dumethyl sulfoxide for preparation of the stock solution. Stocks TCTTTCTTGG-3′. were further diluted in 5% NaHCO3, and doses of 5 mg per kg body weight were administered intraperitoneally daily after wounding. Immunofluorescence microscopy. Spleens and intestines were fixed for 8 h in 2% PFA, incubated overnight in 30% sucrose in H2O and embedded in Isolation and transfer of bone marrow progenitor cells. Bone marrow optimal cutting temperature compound, and then cryosections 8 µm in thick- was collected from the femur, tibia, humerus and pelvis. Bones were frag- ness produced. After two washes in DPBS, nonspecific binding in sections was mented by mortar and pestle, and debris were removed by gradient centrifu- blocked by incubation for 10 min at 21 °C in CAS Block (00-8120; Invitrogen) gation with Histopaque 1119 (Sigma). Cells were passed through a 70-µm containing 0.2% Triton X-100. Sections were stained with primary and second- strainer and red blood cells were lysed with ammonium chloride–potassium ary reagents (identified above) diluted in CAS Block containing 0.2% Triton bicarbonate lysis buffer. Cells were counted on a Vi-CELL analyzer, and X-100 and were mounted with ProLong Gold Antifade reagent containing 5 × 106 to 10 × 106 cells were stained for analysis or the entire bone mar- DAPI (4,6-diamidino-2-phenylindole; P36935; Invitrogen). For staining of row was stained for sorting. The gates used to define common myeloid pro- CD11c, spleens were frozen without fixation and then were fixed in acetone genitors, common DC progenitors and pre-DCs were based on published for 15 min at 4 °C before being washed in DPBS and stained. A Zeiss AxioCam © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature studies62,63: common myeloid progenitors were identified as Lin−c-KithiSca- MRn microscope equipped with an AX10 camera was used for four-color epi 1−CD11c−CD135+CD16/32−CD127− cells; common DC progenitors were fluorescence microscopy. Monochrome images were acquired with AxioVision identified as Lin−c-KitintSca-1−CD135+CD16/32−CD127− cells; pre-DCs were software with either a 10× or a 20× objective and then were exported into − lo − + + − − npg identified as Lin c-Kit Sca-1 CD11c CD135 CD16/32 CD127 cells; and ImageJ software for subsequent color balancing and overlaying. cDC-restricted pre-cDCs were identified as Zbtb46-GFP+ cells in the pre- DC gate. For cell purification by sorting, a FACSAria II (BD Biosciences) was Bone marrow chimeras. Bone marrow cells from donor mice were collected used. Cells were sorted into DPBS supplemented with 0.5% BSA and 2 mM as described above, and 5 × 106 to 10 × 106 total bone marrow cells were EDTA. Cell purities of at least 95% were confirmed by analysis after sorting. transplanted by retroorbital injection into wild-type B6.SJL recipients. For For transfer experiments, purified cell populations were transferred retroorb- developmental competition experiments, recipients received a single dose of itally into congenically marked mice (CD45.1+) after sublethal irradiation with 1,200 rads of whole-body irradiation on the day of transplantation, and mice 600 rads of whole-body irradiation. were analyzed 8–10 weeks later. For infection experiments, recipients received a total of two doses of 525 rads of whole-body irradiation, with an interval Expression microarray analysis. Total RNA was isolated from cDCs with 4 h between the doses, on the day of transplantation, and mice were infected an RNAqueous-Micro kit (Ambion) or from colonic cells with an RNeasy 8–10 weeks later. For mixed–bone marrow competition, bone marrow was Mini kit (Qiagen). For Mouse Gene 1.0 ST Arrays, RNA was amplified with obtained from mice of two genotypes, then cells were counted and mixed at a WT Expression kit (Ambion) and labeled, fragmented and hybridized a ratio of 1:1, and Lin−Sca-1+c-Kit+ cells were analyzed by flow cytometry with a WT Terminal Labeling and Hybridization kit (Affymetrix). Data were to confirm 1:1 chimerism before transplantation. Il23a−/− bone marrow was processed by robust multiarray average summarization and quantile nor- distinguished from Notch2f/f or Notch2cKO bone marrow by GFP expression. malization, and expression values were modeled with ArrayStar software Ltbr−/− bone marrow was distinguished by expression of CD45.1. (DNASTAR version 4). For principal-component analysis, microarray data sets were pre-processed by ArrayStar with robust multiarray average summa- Statistical analysis. In general, differences between groups were analyzed rization and quantile normalization, then replicates were grouped by sample. by unpaired, two-tailed Student’s t-tests. Results with a P value of 0.05 or less Either log-transformed expression values from each replicate or mean log- were considered significant (Prism; GraphPad Software). Survival studies were transformed expression values from each replicate group were exported in analyzed by the log-rank Mantel-Cox test. For multiple comparisons, data were table format, imported into software of the R project for statistical computing analyzed by the one-way ANOVA followed by Tukey’s multiple-comparisons (version 2.13.2), mean-centered by gene, root-mean-square–scaled by sample, test. The appropriate nonparametric test was used when data failed to meet transposed and subjected to principal-component analysis computed by assumptions for parametric statistics.

nature immunology doi:10.1038/ni.2679 56. Han, H. et al. Inducible gene knockout of transcription factor recombination signal 60. Schwenk, F., Baron, U. & Rajewsky, K. A cre-transgenic mouse strain for the binding protein-J reveals its essential role in T versus B lineage decision. Int. ubiquitous deletion of loxP-flanked gene segments including deletion in germ cells. Immunol. 14, 637–645 (2002). Nucleic Acids Res. 23, 5080–5081 (1995). 57. Yu, H. et al. APP processing and synaptic plasticity in presenilin-1 conditional 61. Robben, P.M. et al. Production of IL-12 by macrophages infected with Toxoplasma knockout mice. Neuron 31, 713–726 (2001). gondii depends on the parasite genotype. J. Immunol. 172, 3686–3694 (2004). 58. Yin, L. et al. Defective lymphotoxin-beta receptor-induced NF-kappaB transcriptional 62. Akashi, K. et al. A clonogenic common myeloid progenitor that gives rise to all activity in NIK-deficient mice. Science 291, 2162–2165 (2001). myeloid lineages. Nature 404, 193–197 (2000). 59. Keskintepe, L. et al. Derivation and comparison of C57BL/6 embryonic stem cells 63. Onai, N. et al. Identification of clonogenic common + Flt3 M-CSFR+ plasmacytoid to a widely used 129 embryonic stem cell line. Transgenic Res. 16, 751–758 and conventional dendritic cell progenitors in mouse bone marrow. Nat. Immunol. 8, (2007). 1207–1216 (2007). © 2013 Nature America, Inc. All rights reserved. America, Inc. © 2013 Nature npg

doi:10.1038/ni.2679 nature immunology