Title of Dissertation Goes Here in All Caps

A CHEMICALLY INDUCED COLITIS SCREEN REVEALS THE NECESSITY FOR MEMBRANE TRAFFIC IN INTESTINAL HOMEOSTASIS

APPROVED BY SUPERVISORY COMMITTEE

Bruce Beutler, M.D.

Ezra Burstein, M.D., Ph.D.

Philipp Scherer, Ph.D.

Sandra Schmid, Ph.D.

Sebastian Winter, Ph.D.

DEDICATION

To my wife and our families for their unyielding support and encouragement.

A CHEMICALLY INDUCED COLITIS SCREEN REVEALS THE NECESSITY FOR MEMBRANE TRAFFIC IN INTESTINAL HOMEOSTASIS

WILLIAM ELLIOTT MCALPINE

DISSERTATION

Presented to the Faculty of the Graduate School of Biomedical Sciences

The University of Texas Southwestern Medical Center

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

The University of Texas Southwestern Medical Center

Dallas, Texas

May, 2019

WILLIAM ELLIOTT MCALPINE

A CHEMICALLY INDUCED COLITIS SCREEN REVEALS THE NECESSITY FOR MEMBRANE TRAFFIC IN INTESTINAL HOMEOSTASIS

Publication No.

William Elliott McAlpine, Ph.D.

The University of Texas Southwestern Medical Center, 2019

Supervising Professor: Bruce Beutler, M.D.

Inflammatory bowel disease is most commonly a complex disorder caused by the interaction of environmental and genetic aberrations. Under normal conditions, a genetic program actively prevents inflammatory bowel disease, preventing invasion of microbes without permitting severe inflammation of the gut. To identify genes that maintain this balance, we performed a sensitized screen of 49,420 third generation (G3) germline mutant mice derived from N-ethyl-N- nitrosourea-mutagenized grandsires, bearing 104,658 coding/splicing mutations. We induced mild mucosal damage in these mice by orally administering dextran sodium sulfate (DSS) and found mutations that led to diarrhea and weight loss under these conditions. Causative mutations were

v mapped concurrently with screening using an automated mapping procedure. Among 114 DSS phenotypes identified and mapped, 36 have been validated by CRISPR/Cas9 targeting.

Three vesicle trafficking genes, Myo1d, Smcr8, and Tvp23b, were selected for mechanistic evaluation. MYO1D is a class I myosin that binds both actin and lipid. MYO1D localizes to the basolateral membrane of enterocytes and functions in the intestinal epithelium to protect against colitis. SMCR8, along with C9ORF72 and WDR41, is a member of a tripartite complex that functions as a guanine exchange factor. SMCR8 localizes to the lysosome, and its absence results in perturbations to endocytic and phagocytic pathways. Hyperactivation of endosomal Toll-like receptors in Smcr8-/- mice causes spontaneous inflammation, and hyperactivation of multiple pathways contributes to DSS susceptibility. TVP23B is a trans-Golgi protein that binds YIPF6.

Both TVP23B and YIPF6 are necessary for the formation of secretory granules in goblet and

Paneth cells of the intestinal epithelium. These studies reveal non-redundant molecules required for the return of normal physiologic balance within the intestine after DSS insult.

ACKNOWLEDGEMENTS

I am extremely grateful to those who have made my research training possible. First, I must extend my gratitude to Dr. Bruce Beutler who provided a training environment second to none. He is the epitome of everything a scientist should be, and I hope that his qualities have been imprinted on me.

My scientific training was also enhanced by Dr. Emre Turer who provided excellent mentorship on a daily basis. Interaction with him was always positive and productive, and we made some exciting discoveries together. I hope that we will have many more interactions over the course of our scientific careers.

Members of my thesis committee, Drs. Ezra Burstein, Philipp Scherer, Sandra Schmid, and

Sebastian Winter, have been very helpful in guiding my projects.

Thank you to Dr. Anne Murray and Dr. Eva Moresco who edited my manuscripts for publication. A special thanks to Dr. Murray who edited this thesis as well.

Members of the wet lab provided technical advice for many experiments. Thank you to

Drs. Takuma Misawa, Jin Huk Choi, Zhao Zhang, Duanwu Zhang, Lei Sun, Hexin Shi, Ying

Wang, Lijing Su, Xue Zhong, Evan Nair-Gill, and Subhajit Poddar. Aijie Liu, Miguel San Miguel, and Dr. Ruchi Jain provided critical assistance for FACS and molecular biology.

Forward genetics in mice is quite the enterprise that requires the involvement of many individuals. Jamie Russell managed the ENU mutagenesis and the phenotypic pipeline. Kuan-wen

Wang executed the DSS screen with me along with help from Braden Hayse and Jianhui Wang.

The bioinformatics team which consists of Stephen Lyon, Chun Hui Bu, Darui Xu, Tao Wang, and

Sara Hildebrand provided many tools including Linkage and Candidate Explorer to determine the

vii likelihood that a mutation was causative for an observed phenotype. Mihwa Choi, Xiaoming Zhan,

Xiaohong Li, and Miao Tang produced CRISPR/Cas9 mice for validation. Sara Ludwig managed the CRISPR mouse colony. Thank you all for the substantive support.

The expansive Beutler laboratory cannot operate without proper administration and management. Thank you to Betsy Layton, Lindsay Scott, Linda Watkins, Wanda Simpson, and

Jenny Geisbert in the administrative office for everything you do, but especially for the aid you provided in securing and utilizing grant funding. Thank you to Elena Mahrt for your management on the front lines of the wet lab.

Stephanie Arnett and Sheila Davis took tremendous care of the mouse strains used in this dissertation. Qihua Sun, Baifang Qin, and John Santoyo performed genotyping for these strains.

Rick Bearden managed our supply chain and was normally the first person I encountered each morning. I cannot think of a better way to start my day.

My family has provided enormous support in my pursuit to become a physician scientist.

My parents, Warren and Kelly, have always exhibited unparalleled love and sacrifice. My brothers,

Warren and Coan, have always provided encouragement and interest in my educational pursuits.

My sister, Sarah Grace, a budding biologist in her own right, helped me performed some of the experiments detailed in this work.

Finally, I would like to express a special thank you to my wife, Katie. I can confidently state, even in the absence of a control, that I would not have survived this progression without her unflinching support.

viii

TABLE OF CONTENTS

Abstract ...... v

Acknowledgements ...... vii

Table of Contents ...... x

Prior Publications ...... xii

List of Figures ...... xiv

List of Tables ...... xvii

List of Abbreviations ...... xviii

Chapter 1 - A DSS-induced colitis screen to identify non-redundant molecules required for intestinal homeostasis ...... 1 Preface...... 1 Introduction ...... 1 Results and Discussion ...... 7 Materials and Methods ...... 19

Chapter 2 - The class I myosin MYO1D binds lipid and protects against colitis ...... 21 Preface...... 21 Abstract ...... 21 Introduction ...... 21 Results ...... 22 Figures...... 27 Discussion ...... 37 Materials and Methods ...... 39

Chapter 3 - Loss of SMCR8 or WDR41 causes immune dysregulation and colitis susceptibility ...... 45 Preface...... 45 Abstract ...... 45 Introduction ...... 46 Results ...... 48 Figures...... 54 Discussion ...... 81 Materials and Methods ...... 83

Chapter 4 - TVP23B regulates secretory granule formation in Paneth and goblet cells ...... 87 Preface...... 87

x Introduction ...... 87 Results ...... 89 Figures...... 92 Discussion ...... 105 Materials and Methods ...... 108

Chapter 5 – Conclusions and Recommendations...... 111

Bibliography ...... 117

PRIOR PUBLICATIONS

McAlpine W, Sun L, Wang K-w, Liu A, Jain R, San Miguel M, Wang J, Zhang Z, Hayse B,

McAlpine SG, Choi JH, Zhong X, Ludwig S, Russell J, Zhan X, Choi M, Li X, Tang M, Moresco

EMY, Beutler B, Turer E. Excessive endosomal TLR signaling causes inflammatory disease in mice with defective SMCR8-WDR41-C9ORF72 complex function. Proceedings of the National

Academy of Sciences of the United States of America. 2018.

McAlpine W, Wang K-w, Choi JH, San Miguel M, McAlpine SG, Russell J, Ludwig S, Li X,

Tang M, Zhan Z, Choi M, Wang T, Bu C-H, Murray A, Moresco EMY, Turer EE, Beutler B. The class I myosin MYO1D binds to lipid and protects against colitis. Disease Models & Mechanisms.

2018. doi: 10.1242/dmm.035923.

Turer E, McAlpine W, Wang K-w, Lu T, Li X, Tang M, Zhan X, Wang T, Zhan X, Bu C-H,

Murray AR, Beutler B. Creatine maintains intestinal homeostasis and protects against colitis.

Proceedings of the National Academy of Sciences of the United States of America.

2017;114(7):E1273-E81. doi: 10.1073/pnas.1621400114. PubMed PMID: PMC5321020.

Wang T, Zhan X, Bu CH, Lyon S, Pratt D, Hildebrand S, Choi JH, Zhang Z, Zeng M, Wang KW,

Turer E, Chen Z, Zhang D, Yue T, Wang Y, Shi H, Wang J, Sun L, SoRelle J, McAlpine W,

Hutchins N, Zhan X, Fina M, Gobert R, Quan J, Kreutzer M, Arnett S, Hawkins K, Leach A, Tate

C, Daniel C, Reyna C, Prince L, Davis S, Purrington J, Bearden R, Weatherly J, White D, Russell

J, Sun Q, Tang M, Li X, Scott L, Moresco EM, McInerney GM, Karlsson Hedestam GB, Xie Y,

xii Beutler B. Real-time resolution of point mutations that cause phenovariance in mice. Proceedings of the National Academy of Sciences of the United States of America. 2015;112(5):E440-9. Epub

2015/01/22. doi: 10.1073/pnas.1423216112. PubMed PMID: 25605905; PMCID: PMC4321302.

xiii

LIST OF FIGURES

CHAPTER TWO

Figure 2.1: Mapping of the Myo1d mutation to the whisper phenotype ...... 27

Figure 2.2: Mapping of the Myo1d mutation to the horton phenotype ...... 28

Figure 2.3: Mapping of Myo1d in multiple pedigrees ...... 29

Figure 2.4: Protein domain organization of mouse MYO1D and production of MYO1D

mutant alleles ...... 30

Figure 2.5: Validation that Myo1d mutations cause a colitis phenotype...... 31

Figure 2.6: Colitis readouts for Myo1d96ins/96ins mice ...... 32

Figure 2.7: Non-hematopoietic defects contribute to colitis ...... 33

Figure 2.8: Myo1d mutant mice exhibit normal intestinal epithelial cell development . 34

Figure 2.9: Myosin 1D functions in lipid binding ...... 35

Figure 2.10: Myosin 1D functions in actin binding ...... 36

CHAPTER THREE

Figure 3.1: Mapping of the Smcr8 mutation to the patriot phenotype ...... 54

Figure 3.2: Mapping of the Smcr8 mutation to the patriot2 phenotype ...... 55

Figure 3.3: Validation of DSS weight loss phenotype caused by loss of SMCR8 ...... 56

Figure 3.4: Smcr8 CRISPR mice are susceptible to DSS-induced colitis ...... 57

Figure 3.5: Hematopoietic contributions to DSS phenotype...... 58

Figure 3.6: Splenomegaly and lymphadenopathy in Smcr8 mutant mice ...... 59

Figure 3.7: Equal expansion of major cell populations in spleens of Smcr8 CRISPR

mice ...... 60

xiv Figure 3.8: Increased activation of T cells in Smcr8 CRISPR mice ...... 61

Figure 3.9: Elevated IL-12p40 levels in serum of Smcr8 CRISPR mice...... 62

Figure 3.10: Screening for TNF responses to TLR9 stimulation...... 63

Figure 3.11: Analysis of TLR signaling in Smcr8-/- and Smcr8I2T/I2T peritoneal macrophages ...... 64

Figure 3.12: Analysis of TLR signaling in Smcr8-/- and Smcr8I2T/I2T BMDCs...... 65

Figure 3.13: Removal of TLR 3, 7, and 9 signaling prevents enlargement of peripheral immune organs ...... 66

Figure 3.14: Removal of TLR 3, 7, and 9 signaling normalizes IL-12(p40) serum levels...... 67

Figure 3.15: Removal of TLR 3, 7, and 9 signaling rescues spontaneous T cell activation ...... 68

Figure 3.16: Removal of TLR 3, 7, and 9 signaling fails to rescue DSS susceptibility . 69

Figure 3.17. Increased LysoTracker positive staining in Smcr8-/- BMDMs ...... 70

Figure 3.18: Impaired acidification of BioParticles by Smcr8-/- BMDMs ...... 71

Figure 3.19: Normal uptake of zymosan BioParticles ...... 72

Figure 3.20: Exaggerated TNF production by Smcr8-/- BMDMs in response to phagocytosed cargo ...... 73

Figure 3.21: Increased phagosomal ROS production by Smcr8-/- BMDMs ...... 74

Figure 3.22: Wdr41gogi/gogi mice exhibit increased CD44+ T cells in peripheral blood ... 75

Figure 3.23: Wdr41gogi/gogi mice exhibit increased DSS susceptibility ...... 76

Figure 3.24: Elevated IL-12p40 plasma levels in Wdr41gogi/gogi mice ...... 77

Figure 3.25: Wdr41gogi/gogi peritoneal macrophages produce elevated TNF in response to

endosomal TLR ligands ...... 78

Figure 3.26: Wdr41mca/mca peritoneal macrophages produce elevated TNF in response to

endosomal TLR ligands ...... 79

Figure 3.27: Increased percentages of CD44+ T cells in Wdr41mca/mca mice ...... 80

CHAPTER FOUR

Figure 4.1: Mapping of the Tvp23b mutation to the Chipotle phenotype ...... 92

Figure 4.2: CRISPR/Cas9 validation of Tvp23b as a DSS susceptibility gene ...... 93

Figure 4.3: H&E staining of Tvp23b-/- intestines ...... 94

Figure 4.4: PAS and AB staining of Tvp23b-/- intestines ...... 95

Figure 4.5: Transmission electron microscopy (TEM) of Tvp23b-/- intestines ...... 96

Figure 4.6: Reduced levels of LYZ in Paneth cells of Tvp23b-/- mice ...... 97

Figure 4.7: Reduced levels of REG3γ in Paneth cells of Tvp23b-/- mice ...... 98

Figure 4.8: Assessment of LYZ and REG3γ levels in small intestine crypts ...... 99

Figure 4.9: TVP23B binds YIPF6 ...... 100

Figure 4.10: TVP23B localizes to the TGN...... 101

Figure 4.11. Mapping of a Golga4 mutation to the deranged phenotype...... 102

Figure 4.12: Tvp23b-/- mice exhibit normal production of antibodies, TNF, and

insulin ...... 103

CHAPTER FIVE

Figure 5.1: Predicted genetic variance explained using logarithmic fit...... 111

xvi

LIST OF TABLES

CHAPTER ONE

Table 1.1: DSS Screen saturation statistics ...... 16

Table 1.2: Genes mapped during DSS screening...... 18

xvii

LIST OF ABBREVIATIONS

96ins – 96 base pair insertion

AB – alcian blue

ALPS3 – autoimmune lymphoproliferative syndrome, type III

ALS – amyotrophic lateral sclerosis

AMPs – antimicrobial proteins

ANOVA – analysis of variance

AP – alkaline phosphatase

APC – adenomatous polyposis coli

BSA – bovine serum albumin bp – base pair

CD – Crohn’s disease

Cpt – Chipotle

CVID8 – common variable immunodeficiency 8 with autoimmunity

DCG – dense core granule drg – deranged

Dsg – desmoglein

DSS – dextran sodium sulfate

EDTA – ethylenediaminetetraacetic acid

ENU – N-ethyl-N-nitrosourea

FTD – frontotemporal dementia

xviii

GATM – glycine amindotransferase

G3 – third generation

GEF – guanine exchange factor

GI – gastrointestinal

GWAS – genome-wide association studies

H2DCFDA – 2′,7′-dichlorodihydrofluorescein diacetate

H&E – hematoxylin and eosin

HCG – human chrorinic gonadotropin htn – horton

IBD – inflammatory bowel disease

IgA – immunoglobulin A

Insr – insulin receptor

ISG – immature secretory granule

Ksr1 – kinase suppressor of Ras-1

Lars2 – leucyl-tRNA synthetase

LPA – lysophosphatidic acid

LPC – lysophosphocholine

Lypd8 – Ly6/PLAUR domain containing 8

MFI – mean fluorescence intensity

Mitf3 – mitochondrial translational initiation factor-3

Myo1a – myosin IA

Myo1c – myosin IC

xix

Myo1d – myosin ID

Muc2 – mucin-2

NOX2 – NADPH oxidase p – patriot p2 – patriot2

PA – phosphatidic acid

PAS – period acid-Schiff

PBS – phosphate-buffered saline

PE – phosphatidylethanolamine

PI – phosphatidylinositol

PI(3)P – phosphatidylinositol 3-phosphate

PI(3,4,5)P3 – phosphatidylinositol 3,4,5-trisphosphate

PI(4)P – phosphatidylinositol 4-phosphate

PI(5)P – phosphatidylinositol 5-phosphate

PI(3,5)P2 – phosphatidylinositol 4,5-bisphosphate

PI(4,5)P2 – phosphatidylinositol 4,5-bisphosphate

PMSG – pregnant mare’s serum gonadotropin

PNAS – Proceedings of the National Academy of Sciences of the United States of America

Ppap2c – phosphatidic acid phosphatase type 2C

PolyPhen-2 - Polymorphism Phenotyping v2

PS – phosphatidylserinea

ROS – reactive oxygen species

S1P – sphingosine-1-phosphate sgRNA – small guide RNA

SWC – SMCR8-WDR41-C9ORF72

TCR – T-cell receptor complex

TEM – transmission electron microscopy

TGN – trans-Golgi network

TLR – Toll-like receptors

TNF – tumor necrosis factor

Tvp23b - trans-Golgi network vesicle protein homolog B

UC – ulcerative colitis

V-ATPase – vacuolar ATPase wpr – whisper

xxi

CHAPTER ONE

A DSS-induced colitis screen to identify non-redundant molecules required for intestinal homeostasis

PREFACE

The methods for real-time identification of mutations that cause phenovariance in mice and a subset of genes identified by the DSS-induced colitis screen have already been reported by the

Beutler laboratory in two separate journal articles published in Proceedings of the National

Academy of Sciences of the United States of America (PNAS) (1, 2). Authors published in PNAS retain the right to include the article, in whole or in part, in a thesis or dissertation.

INTRODUCTION

Inflammatory Bowel Disease Pathogenesis

Inflammatory bowel diseases (IBD) are characterized by chronic inflammation within the gastrointestinal (GI) tract. The exact pathophysiologic mechanisms of IBD are not fully understood, but it is established that the disease results from inappropriate or excessive immune responses to resident microbiota in a genetically susceptible host. Crohn’s disease (CD) and ulcerative colitis (UC) are the two most prevalent forms of IBD. Inflammation in CD is transmural and can be observed in all layers of the intestinal tract, while inflammation in UC is confined to the mucosa. The main pathological features of CD are fistulas, ulcers, and granulomas which primarily affect the terminal ileum but can be found in any part of the GI tract from mouth to the

1 2 anus. UC consists of superficial ulcerations limited to the colon, starting in the rectum and extending to the proximal colon.

IBD susceptibility has a strong genetic component. The greatest risk for acquiring CD or

UC is having a family member with the disease (3). First-degree relatives have a relative risk of 1 in 25 to 42 and 1 in 8 to 15 for CD and UC, respectively (4). When concordance rates for monozygotic twins are considered, the heritability of CD is about 40% and approximately 15% for

UC, hundreds of times higher than the disease rate of age-matched individuals in the general population (5). The two primary methods used to explore the genetics of IBD are genome-wide association studies (GWAS) and monogenic studies, but these approaches have only accounted for a small percentage of disease heritability.

GWAS and IBD

GWAS have identified over 200 risk loci associated with IBD (6-15); however, contrary to common misconception, the causative functional variant for many of these loci has not been identified. Because of linkage disequilibrium within the human genome, GWAS loci often span large genomic regions, extending from tens to hundreds of thousands of base pairs (16). These regions include five genes on average, making it difficult to distinguish between causal and neutral variants (16). Fine-mapping studies have attempted to address this issue with limited success (15,

16). When presented with a collection of genes within a locus, the inclination is to attribute causality to a particular gene because it is involved in a familiar aspect of disease biology; however, this is often performed with little to no evidence, which risks attribution of causality to

3 a gene that is not involved and omission of the causative gene that participates in an unknown aspect of disease biology.

GWAS focus on the identification of common genetic variants that occur at a minimum frequency of 5% (17). Highly deleterious alleles with large effects on disease susceptibility are typically found at low frequencies because of natural selection. GWAS fails to detect these rare variants (minimum allele frequency of less than 0.5%) that greatly outnumber common variants

(18, 19). So far, GWAS have only accounted for a small proportion of inherited risk, estimated to be less than 30%, suggesting that undetected rare variants with large effects may account for a substantial percentage of missing heritability (20). Indeed, it is estimated that 85% of disease- causing mutations with large effects are found in coding regions or canonical splice sites (21), but loci detected by GWAS are often found in non-coding regions which have modest effects on IBD susceptibility (22).

Despite GWAS limited ability to assign causality and detect rare variants, some causative genes have been unambiguously identified when multiple coding variants are detected, including

NOD2 (23, 24), ATG16L1 (25, 26), and IL-23R (27). The functional characterization of these genes has revealed significant insights into the underlying mechanisms of IBD establishing the importance of pattern recognition signaling, autophagy, and the IL-23/IL-17 axis (28).

Monogenic causes of IBD

Monogenic disorders that lead to IBD can also be utilized to explore genes with known causality. Of the 50 monogenic mutations that lead to IBD-like symptoms, only three (IL10,

IL10RB, and IL2RA) are located in loci determined by GWAS (29). The study of single gene

4 defects has provided functional understanding of IBD phenotypes and demonstrated the necessity for epithelial barrier function, neutrophil and macrophage killing, appropriate T and B cell activation, T-regulatory mechanisms, and intestinal innervation in healthy individuals (29).

However, examination of monogenic disorders is limited because of the limited number of mutations that come to clinical attention and are subsequently systematically studied in families

(29). When considering the limitations of GWAS and the study of monogenic disorders, a desirable method for the elucidation of genetic risk in IBD requires the ability to examine the entire genome to determine causal genes rather than a few regions that may only be associated.

ENU mutagenesis to generate genetic diversity

Genetically engineered animal models represent one avenue to explore genes involved in

IBD. Ninety-nine percent of human genes have mouse homologues, and the two species share many similarities related to epithelial biology, immunology, and microbial composition (22). The discovery of monogenic causes of IBD in humans were preceded by the implication of some of these genes in mouse models (e.g., Il10r and Adam17) (30-34).

Knockout mice have frequently been used to study the effects of gene disruption in the context of a whole organism. Gene targeting is often performed based on an a priori assumption of a gene’s function and its role in disease, which prevents survey of genes that may be involved in unknown processes. A large proportion of human disease is caused by alterations at the protein level that affect structure, binding, and/or function (35). Complete knockout of a gene may result in a more severe phenotype or allow for compensation by other genes (35). For example, mice with gene-trap mediated-disruptions of Atg16l1 exhibit more extreme disorganization and loss of

5 Paneth cell granules compared with mice homozygous for a knock-in for the Atg16l1 T300A risk allele, which have phenotypes that more closely resemble what is observed in humans with the susceptibility variant (22, 36). Lastly, an estimated 34% of genes are essential for life in mice and cannot be analyzed in the context of complete loss of function (37).

Chemically-induced mutagenesis, such as treatment with N-ethyl-N-nitrosourea (ENU), offers an alternative method to generate mouse models of human diseases. ENU is a potent alkylating agent that primarily induces point mutations, leading to a wide range of missense, nonsense, splicing, or make-sense mutations (38, 39). Approximately one base pair (bp) per Mbp of DNA is altered in the haploid genome of an ENU-treated mouse, equating to approximately

3,000 total mutations (39). The use of ENU-mutagenesis in combination with a phenotypic screen represents a viable approach to uncover genes not previously implicated in IBD. Random generation of mutations and assessment of resulting phenotypes in an unbiased manner allows for the determination of all non-redundant molecules involved in a biologic phenomenon and reveals new pathways that were previously undetected. In addition to null alleles, ENU mutagenesis can generate hypomorphs, hypermorphs, and neomorphs which more closely resemble disease causing alleles in humans (35). Hypomorphic mutations can overcome early lethality, an important advantage considering viable mutations in essential genes are more likely to cause phenotype compared with non-essential genes (37).

DSS-induced colitis

One of the most widely used experimental models of IBD in mice is dextran sodium sulfate

(DSS)-induced colitis. DSS is a sulfonated polysaccharide that is toxic to the intestinal epithelium

6 (40). The severity of DSS-induced colitis depends on a variety of factors, including dose (generally

1 to 5%), duration and frequency of the administration, strain of mice, and housing conditions

(germ-free vs. specific pathogen free) (40). Clinical manifestations of DSS-induced colitis are similar to what is observed in humans with IBD and include weight loss, rectal bleeding, diarrhea, occult blood in stools, and mortality (40). The distal colon is primarily affected by DSS treatment, and in this aspect the model more closely resembles UC. DSS-induced histological changes include mucin depletion, epithelial cell death, and necrosis that leads to loss of the intestinal epithelium. The epithelium loss is accompanied by infiltration of neutrophils, cryptitis, crypt abscesses, and in extreme cases vacuolar hydropic degeneration of cells (40).

DSS-induced colitis reflects UC in both clinical manifestations and organ pathogenesis, but the model does not completely recapitulate human IBD. In the DSS model, inflammation can develop in the absence of the adaptive immune system (e.g., SCID and Rag1-/- mice) (40); however, in human IBD, CD4+ T lymphocytes play a crucial role in IBD pathogenesis (41, 42). Heavy infiltration of T cells is observed at sites of tissue injury in IBD. Depletion of CD4+ T cells improves clinical remission in IBD patients infected with human immunodeficiency virus and in therapeutic interventions designed to target T cells (43). Additionally, primarily T-cell-mediated autoimmune diseases (e.g., psoriasis, rheumatoid arthritis, and multiple sclerosis) are observed with a higher frequency in patients with IBD (44).

Despite the differences between DSS-induced colitis and human IBD, it is the best model to use when screening for intestinal homeostatic aberrations in large numbers of mice due to its rapidity (less than 2 weeks), ease of administration (oral), reproducibility, and controllability

(dose-dependence) (45). Moreover, many of the genes involved in human IBD, identified through

7 GWAS or the study of monogenic disorders, also result in DSS colitis phenotypes when altered in mice, including Atg16l1 (46, 47), Il10 (48), Il23r (49), and Nod2 (50).

In this chapter, the results of a forward genetic screen of ENU-mutagenized mice exposed to DSS are detailed. Screening of approximately 50,000 mice harboring over 100,000 mutations resulted in an estimated 21% genome saturation. We mapped 114 phenotypes and 36 were validated by literature verification or CRISPR/Cas9 targeting. Many of these genes have not been implicated in intestinal homeostasis and some could account for the large percentage of missing heritability in IBD.

RESULTS AND DISCUSSION

Fractional destruction of the genome and real time identification of mutations that cause phenotype

The Beutler laboratory has developed a system for real-time identification of ENU- introduced mutations that cause phenotypes in mice (2). Real-time identification of causative mutations is based on the premise that the conventional steps taken in positional cloning (i.e., establishment of a homozygous stock, outcrossing, backcrossing, and meiotic mapping to establish a critical region) are unnecessary to assign cause and effect if all candidate mutations are known and if a large pedigree has been derived from the G1 founder. An average of 60 coding changes are transmitted from every ENU-mutagenized G0 progenitor to the G1 founder of each pedigree, and almost all ENU-induced phenotypes emanate from a coding change (39). The mutations are identified by whole-exome sequencing, which is performed on all G1 mice in advance of breeding.

8 Ten G2 daughters are derived from each G1 male, and all daughters are backcrossed to the G1, yielding 20 to 50 G3 animals. G2 and G3 mice are genotyped at every mutation site and zygosities are recorded prior to phenotypic screening. The upload of quantitative or qualitative phenotype data triggers a program called Linkage Analyzer to analyze all mutations for statistical association to each phenotype using recessive, dominant, and semi-dominant models of inheritance.

Mapping and validation of DSS hits

To identify genes necessary for intestinal homeostasis, we performed a forward genetic screen in which G3 mutant mice were treated with 1.3 to 1.5% DSS in their drinking water for seven days and then switched to regular drinking water for an additional three days. Body weights were recorded on day 0, 7, and 10 with respect to initiation of DSS treatment and percentage of initial body weight was used as a continuous mapping variable. We screened 49,420 G3 mice derived from 1,885 G1 grandsires carrying 104,658 variant alleles of 19,360 genes (Table 1.1);

72,886 mutations in 16,356 genes were transmitted to the homozygous state in three or more G3 mice. Among these mutations, 19,622 were null alleles (premature stop codons or critical splice junction errors) in 7,983 genes, while 36,724 mutations in 11,923 genes were predicted to be

“probably damaging” by Polymorphism Phenotyping v2 (PolyPhen-2) with a score of 0.95 or greater (51). Using our recently described method to calculate the damaging effects of mutations, we estimate that we have severely damaged or destroyed 21.3% of all protein-encoding mouse genes and tested the mutant alleles three or more times in the homozygous state in the DSS screen

(37).

9 Linkage mapping identified putative causative mutations in 101 genes in 114 pedigrees

(Table 1.2). Nine genes, Ern2, Myocd, Myo1d, Lrba, Lrmda, Pcnx2, Prckd, Smcr8, and Tg, were mapped multiple times in ancestrally unrelated pedigrees accounting for the difference between gene and pedigree number. Eight genes, Arhgap17, Ern2, Hgf, Klf5, Muc2, Slc9a8, Tnfrsf14, and

Yap1, have previously been reported in the literature to be required for protection from DSS- induced colitis (52-57). We targeted 36 of the remaining 94 genes with CRISPR/Cas9 to validate the mapping data. Nineteen genes were verified, and 16 genes were excluded. Lck was validated with commercially available knockout mice. Thus, the current validation rate for genes not previously been reported in the literature to be involved in intestinal homeostasis is 56% (20/36).

A machine learning model was developed by the Beutler laboratory to better predict if a high scoring mutation is causative for a phenotype. Candidates are scored on the type of screen, predicted damage probability, essentiality of the gene, number of mice used in the screening, the screening result qualities (e.g., outliers, big variations, overlaps between VAR and REF results, etc.), mapping in more than one pedigree (superpedigree), and other criteria were categorized as excellent. This prediction model exhibits 68% precision for candidate genes categorized as good or excellent. Using this model, it is predicted that 28 of the remaining 41 pending genes that are rated as good or better will validate.

Lrba, Prkcd, and Tnfrs14 are associated with IBD in humans (15, 58). Mutations in LRBA cause common variable immunodeficiency 8 with autoimmunity (CVID8, OMIM #614700).

Affected individuals develop variable autoimmune disorders, including idiopathic thrombocytopenic purpura, autoimmune hemolytic anemia, and IBD (58-61). Mutations in PRKCD are linked to autoimmune lymphoproliferative syndrome, type III (ALPS3; OMIM

10 #615559) (62-64). Presentation of ALPS3 varies, but most patients have lymphadenopathy and other autoimmune manifestations. GWAS have associated PRKCD with IBD (15). Tnfrsf14 encodes a member of the TNF super family and has also been linked to human IBD by GWAS

(15). The detection of these genes validates the screen as a means to identify human disease genes and components of IBD pathogenesis.

When treated with 1.3 to 1.5% DSS, wild-type mice in our colony lose on average 5% of their initial body weight, while mice harboring mutations in genes that are required for protection from colitis lose between 20 to 30% of their body weight. The vast majority of mutations mapped to an increased susceptibility phenotype, but in some pedigrees mutations seemed to impart resistance to DSS-induced colitis (i.e., no body weight loss or even body weight gain during the treatment). These “resistance” genes represent potential drug targets for IBD therapy. Three of these resistance genes, Lck, Tnfrsf14, and Plpp2 have been verified either through literature or by our laboratory. Lck is critical for T cell development, and its function is discussed in more detail below. TNFRSF14 activates the non-canonical NF-κβ signaling pathway and Tnfrsf14-deficient mice exhibit reduced susceptibility to DSS (65). Ppap2c encodes phosphatidic acid phosphatase

(PAP) type 2C (PPAP2C). PAP family members catalyze the conversion of phosphatidic acid to diacylglycerol and inorganic phosphate (66). Phosphatidic acid and diacylglycerol are important lipid messengers that regulate many processes including polarized cell growth, pathogen defense, oxidative burst, and secretion (67). The ENU allele, Trust, imparted resistance to colitis in homozygous mice (Ppap2cTrust/Trust). CRISPR-Cas9 targeting was used to generate Ppap2c knockout mice (Ppap2c-/-). Ppap2c+/- mice were resistant to DSS-induced colitis, while Ppap2c-/- mice were susceptible. This indicates that some loss of enzyme activity is protective, but complete

11 loss is harmful. Whether this protection is caused by increased phosphatic acid or reduced diacylglycerol is not known. We are currently in the process of generating CRISPR/Cas9 replacement mice for the original ENU allele (Ppap2cT51M) to verify that this mutation leads to resistance in the homozygous state.

The severity of DSS-induced colitis is dose dependent. Polydipsic mice are often susceptible to DSS treatment because they consume more DSS water, effectively exposing them to a higher dose. Mice homozygous for a missense mutation in insulin receptor (Insr), which presumably develop diabetes metillus with accompanying polydipsia, were detected in the screen.

Similarly, a nonsense mutation in Dnajc3 was identified. Dnajc3-/- mice exhibit glucosuria and hyperglycemia due to increased apoptosis of pancreatic islet cells (68). Mutations in two kidney transporters, CLCNKA, a voltage-gated chloride channel, and SLC5A2, a sodium glucose transporter, also mapped. Clcnka-/- and Slc5a2-/- mice both exhibit excessive thirst due to osmotic diuresis (69, 70). Development of colitis in these mice is an artifact of the model, and these genes do not represent candidates for IBD in humans.

DSS screen highlights important cell types and pathways required for intestinal homeostasis

GWAS and monogenic studies have established the requirement for epithelial barrier function in IBD. Molecules with key functions in the epithelium, including those involved in mucous layer formation and tight junction formation, were also identified in the DSS screen.

Mucin-2 (MUC2) is the major proteinaceous component of mucin secreted by goblet cells (57).

SLC9A8 functions as sodium-hydrogen exchanger at the apical membrane of intestinal epithelial cells, and loss of this protein leads to reduced goblet cell number and disorganization of the mucin

12 layer (71). ARHGAP17 is a RhoGTPase required for mucin layer and tight junction integrity (52).

DSG2, a member of the desmoglein (DSG) family of cadherins, promotes tight junction integrity in the epithelium (72) and prevents epithelial apoptosis during inflammation (73).

Molecules required for intestinal epithelium proliferation and differentiation were also mapped, especially proteins that mediate Wnt/β-catenin signaling. When inactivated, β-catenin is phosphorylated by the adenomatous polyposis coli (APC)/Axin/GSK3/CK1 destruction complex leading to its ubiquitination and proteasomal degradation (74). In the presence of Wnt ligand, the destruction complex is disrupted allowing for nuclear translocation of β-catenin and transcription of target genes. Genes encoding proteins that interact with β-catenin and act as co-activators for target gene transcription, including Yap1, Tcf4, and Klf5, were detected in the screen. (55, 56).

GWAS and monogenic studies have also established the importance for appropriate T cell polarization and regulation in IBD. LCK is a member of the Src family of nonreceptor tyrosine kinases and phosphorylates CD3, ζ-chains of the T-cell receptor complex (TCR), and Zap70 when a TCR engages a specific antigen presented by major histocompatibility complex (MHC) class I and II (75, 76). The tyrosine phosphorylation cascade initiated by LCK is critical for T-cell activation and LCK-deficient mice exhibit severe T cell developmental defects (77, 78). During the DSS screen and upon subsequent validation, Lck-/- mice exhibited resistance to developing colitis, establishing that although not required, T cells do mediate inflammation in this model.

While some mapped genes could be categorized in pathways previously associated with

IBD, many mapped genes operate in pathways that have never been reported to be linked with the disease including those required for mitochondrial function, lipid and steroid metabolism, and vesicle trafficking pathways beyond autophagy. Glycine amindotransferase (GATM) catalyzes the

13 rate limiting step for creatine biosynthesis, mediating the transfer of an amidino group from arginine to glycine to form ornithine and guanidinoacetate. CRISPR/Cas9 targeting validated mapping of Gatm to a DSS phenotype, and colitis susceptibility caused by loss of GATM function was rescued by supplementation of exogenous creatine (79). Mtif3 encodes mitochondrial translational initiation factor-3 (MTIF3), which binds and increases the availability of free 28S ribosomal subunits subsequently promoting mitochondrial protein synthesis (80). Mutations in

MTIF3 are associated with Parkinson’s disease which is predicted to be caused by oxidative stress- induced mitochondrial dysfunction (81-83). Mitochondrial leucyl-tRNA synthetase (LARS2) is a member of the class 1 aminoacyl-tRNA synthetase family (84).

Genes that function in lipid and steroid metabolism (i.e., Degs2, Hsd11b, and Ppap2c) were all validated by CRISPR/Cas9 targeting. Degs2 encodes a bi-functional enzyme that is involved in the synthesis of ceramides and phytoceramides and can function either as a sphingolipid delta(4)-desaturase or a sphingolipid-C4-hydroxylase. While DEGS1 can also function to produce ceramide, the ability of DEGS2 to synthesize phytoceramide is unique. It is not known if loss of phytoceramide synthesis is causative for the phenotype. The mapping of Hsd11b was validated by

CRISPR replacement of the original ENU allele which results in a valine to glutamic acid change at amino acid 270 (V270E). Hsd11b-/- mice exhibit early lethality with half of Hsd11b-/- mice dying within 48 hours of birth (85). Surviving Hsd11b-/- mice exhibit apparent mineralocorticoid excess accompanied by hypertension, hypokalemia, and hypochloremia. The intestinal loops and distal convoluted tubules of Hsd11b-/- mice are dilated. It is possible that the cause of DSS susceptibility is polydipsia secondary to kidney dysfunction; these mice will be given DSS by oral gavage.

Ppap2c was also validated by CRISPR, and, as discussed above, the enzyme functions to convert

14 LPA into diacylglycerol and inorganic phosphate (66). Lpar6, another gene mapped in the screen, is a G-protein receptor that binds LPA endogenously (86). Perhaps the resistance phenotype observed with loss of PPAP2C function, which presumably results in increased LPA, is mediated through activation of LPAR6.

Autophagy is the major cellular pathway for the recycling and degradation of cytoplasmic cellular contents and is an essential component of innate host defense. Several genes that function in the autophagy pathway have been linked with IBD, with the strongest association represented by ATG16L1 (25, 26). Autophagy genes Lrba and Trappc8 were also detected in the DSS screen.

LRBA-deficient B cells harvested from patients exhibited increased apoptosis and reduced autophagy (59). TRAPPC8, the mammalian orthologue of the yeast autophagy‐specific TRAPP subunit Trs85, functions in both secretory transport and autophagy through its regulation of ATG9

(87). In addition to autophagy, molecules that control other membrane trafficking processes were identified and are the focus for the remainder of this thesis. MYO1D (Chapter 2) is a class I myosin that binds both lipid and actin. MYO1D localizes to the basolateral membrane of the colon epithelium. SMCR8 and WDR41 (Chapter 3) are members of a guanine exchange factor (GEF) complex that regulates lysosomal traffic in myeloid cells. TVP23B (Chapter 4) functions in the transport of proteins from the Golgi apparatus and loss of this molecule leads to aberrations in antimicrobial protein and mucus production by Paneth and goblet cells.

Table 1.1: DSS Screen saturation statistics.

No. of hits Pedigrees G3 mice Genes Alleles Saturation Type of mutation

All mutations >=0 1885 49,420 19,360 104,658 77.51 >=1 1875 46,827 18,211 74,940 72.91 >=2 1875 46,817 17,307 74,036 69.29 >=3 1875 46,796 16,356 72,886 65.48 Probably null* or probably >=0 1885 49,420 16,248 56,445 65.05 damaging** >=1 1870 43,097 14,324 39,332 57.35 >=2 1868 43,038 13,130 38,138 52.57 >=3 1868 42,908 11,923 36,724 47.74 Probably null* >=0 1885 49,420 12,531 31,534 50.17 >=1 1850 36,044 10,280 22,076 41.16 >=2 1839 35,883 9,108 20,904 36.47 >=3 1827 35,521 7,983 19,622 31.96 * Probably null: nonsense, makesense, splicing errors, frameshift indels, start loss, or start gain mutations; designation not provided by PolyPhen- 2 ** Probably damaging: PolyPhen-2 score of 0.95–1.0.

Table 1.2: Genes mapped during DSS screening.

Gene Ref Het Var Inheritance Mutation Type Predicted Effect p-value Candidate status

Arhgap17 (LV) 7 8 4 Recessive L254Q probably damaging 4.7x10-6 excellent candidate Col1a1 19 17 8 Recessive R404L unknown 1.2x10-5 not good candidate Degs2 8 9 6 Recessive N189D probably damaging 2.2x10-6 excellent candidate Entpd5 14 17 4 Recessive R321* probably null 4.6x10-6 good candidate Ern2 (LV) 12 20 10 Recessive splice donor probably null 4.2x10-13 excellent candidate Ern2 (LV) 7 11 2 Semidominant splice donor probably benign 8.6x10-4 excellent candidate Ern2 (LV) 10 11 2 Recessive splice acceptor probably null 9.1x10-3 failed initial filter Gatm 8 7 6 Additive D254G probably damaging 2.0x10-7 excellent candidate Gphn 9 14 4 Recessive S608T probably benign 1.1x10-6 excellent candidate Hgf (LV) 6 10 2 Recessive H649L probably benign 8.1x10-9 excellent candidate Hr 5 15 5 Recessive L1062P probably damaging 1.9x10-7 excellent candidate Hsd11b2 2 6 4 Recessive V270E possibly damaging 2.4x10-9 excellent candidate Insr 12 8 1 Recessive S1084P probably damaging 1.7x10-5 good candidate Klf5 (LV) 24 29 5 Semidominant R360C probably damaging 1.8x10-7 good candidate Lck ® 2 9 6 Recessive E288G probably damaging 2.0x10-5 good candidate Lrba 16 11 5 Recessive Q1292* probably null 9.1x10-8 excellent candidate Lrba 10 17 9 Recessive Y2356* probably null 1.4x10-8 excellent candidate Lrmda 20 24 5 Semidominant Y13* probably null 5.9x10-8 excellent candidate Lrmda 15 20 3 Semidominant D37A probably damaging 1.8x10-4 excellent candidate Muc2 (LV) 19 20 4 Recessive C365R probably damaging 1.9x10-8 potential candidate Myo1d 4 11 3 Recessive N401I probably damaging 5.6x10-5 excellent candidate Myo1d 12 19 10 Recessive L972P probably damaging 1.5x10-9 excellent candidate Myo1d 4 7 3 Recessive V512E probably damaging 1.0x10-5 excellent candidate Myo1d 11 16 7 Recessive D926V probably damaging 8.0x10-5 excellent candidate Nlrp4d 5 19 4 Recessive K762N probably benign 1.9x10-5 potential candidate Ppap2c ® 17 20 5 Semidominant T51M probably damaging 1.3x10-7 potential candidate Rnps1 27 23 4 Recessive F181I probably damaging 1.4x10-6 excellent candidate Slc9a8 (LV) 10 23 1 Recessive M215K probably damaging 4.7x10-5 excellent candidate Smcr8 10 13 3 Recessive I2T probably damaging 9.1x10-7 excellent candidate Smcr8 6 13 4 Recessive M1V probably null 7.7x10-6 excellent candidate Smcr8 18 41 6 Recessive Q615* probably null 3.0x10-6 excellent candidate Tnfrsf14 ® (LV) 7 8 1 Recessive K115* probably null 3.8x10-4 potential candidate Tvp23b 8 14 2 Semidominant E60* probably null 6.3x10-5 excellent candidate Wdr41 15 23 8 Recessive splice donor probably null 6.0x10-6 potential candidate Yap1 (LV) 4 7 8 Semidominant G36D probably damaging 6.5x10-5 excellent candidate Zdhhc1 9 8 4 Recessive frame shift probably null 8.5x10-6 excellent candidate

17 Arhgef11 24 20 3 Recessive splice site probably null 2.3x10-6 excellent candidate Cox7a2 2 4 2 Recessive splice acceptor probably null 2.7x10-4 good candidate Fgfr2 8 8 1 Dominant R400L probably damaging 5.6x10-5 not good candidate Glyat 10 7 3 Recessive V247F probably benign 1.8x10-6 not good candidate Gpat2 ® 11 16 10 Semidominant V75M probably damaging 1.84x10-5 not good candidate Il1rap 17 24 6 Recessive splice donor probably null 2.1x10-5 excellent candidate Liph 15 25 5 Recessive I204N probably damaging 6.1x10-7 excellent candidate Mnd1 18 14 3 Recessive C62F probably benign 4.5x10-5 not good candidate Nfe2 16 27 5 Recessive L253S probably damaging 2.3x10-6 potential candidate Pcnx 11 15 3 Recessive Y1674F probably damaging 7.4x10-4 excellent candidate Pcnx 17 33 10 Recessive S1536* probably null 1.85x10-7 excellent candidate Pde8a ® 20 16 3 Recessive intron probably null 1.6x10-5 excellent candidate Rhobtb2 20 21 5 Recessive D88G probably damaging 7.8x10-4 not good candidate Sash1 12 15 4 Recessive P1000T probably damaging 1.6x10-6 not good candidate Sipa1 9 10 4 Recessive R199Q probably damaging 1.0x10-6 good candidate Syt9 13 29 3 Semidominant T408I 6.3x10-5 not good candidate Thyn1 9 9 5 Semidominant Q98P probably damaging 4.0x10-4 not good candidate Aaas 9 23 3 Recessive splice donor probably null 2.7x10-5 excellent candidate Arhgef7 11 15 Dominant L141P probably damaging 1.5x10-4 potential candidate Atp8b1 19 21 1 Recessive Q119* probably null 4.2x10-9 excellent candidate Camk2g 13 10 3 Semidominant E430G probably null 6.1x10-9 potential candidate Cars 9 14 2 Recessive Y883F probably damaging 1.1x10-6 excellent candidate Cd101 15 10 2 Recessive Y879F possibly damaging 4.1x10-7 not good candidate Cdh17 37 25 10 Semidominant Y270C probably damaging 1.8x10-6 excellent candidate Clcnka 25 37 10 Recessive Y179* probably null 3.5x10-11 excellent candidate Col3a1 19 17 2 Recessive splice donor probably null 1.6x10-5 excellent candidate Col4a1 7 9 0 Dominant G1341V probably damaging 1.3x10-5 potential candidate Creld1 7 20 6 Recessive C169F probably damaging 5.2x10-5 excellent candidate Cyb5r4 16 25 2 Recessive Y88* probably null 1.9x10-5 good candidate Dennd5a 23 22 5 Recessive E504* probably null 1.4x10-5 excellent candidate Dnajc3 7 23 3 Recessive Y291* probably null 6.2x10-5 excellent candidate Dsg2 18 23 5 Recessive Y226* probably null 1.3x10-8 excellent candidate Eif5b 11 8 2 Dominant S209A unknown 5.0x10-5 potential candidate Fam120a 20 26 7 Semidominant intron probably null 6.1x10-7 excellent candidate Gm5346 3 11 3 Recessive Y697* probably null 2.8x10-6 good candidate Golga4 9 13 6 Recessive nonsense probably null 1.0x10-9 excellent candidate Gulp1 5 15 2 Recessive Y27* probably null 1.0x10-4 potential candidate Hscb 12 21 5 Recessive L143P probably damaging 3.1x10-7 excellent candidate Lars2 3 10 4 Recessive I305N probably damaging 2.4x10-6 excellent candidate Ldhd ® 7 11 2 Dominant Y86H probably benign 1.9x10-5 potential candidate Limk1 27 24 5 Recessive T162I probably damaging 3.7x10-6 not good candidate Lpar6 25 28 3 Semidominant C36Y probably damaging 6.6x10-5 good candidate

18 Miga2 16 13 10 Recessive T168A probably benign 7.3x10-5 not good candidate Mtif3 8 7 2 Recessive R249S possibly damaging 1.1x10-5 potential candidate Myocd 11 12 1 Recessive splice site probably null 9.8x10-5 excellent candidate Myocd 10 8 4 Recessive S108P probably damaging 7.9x10-4 excellent candidate Nbas 18 20 6 Recessive N419D probably damaging 2.2x10-5 good candidate Nek8 36 42 14 Recessive M40T probably damaging 4.4x10-11 good candidate Nfkb1 10 11 2 Semidominant splice donor probably null 8.3x10-4 excellent candidate Nme5 37 45 7 Recessive M1R probably null 1.7x10-4 not good candidate Nos3 5 2 5 Recessive T572I probably damaging 8.2x10-4 excellent candidate Nr1i3 21 20 2 Recessive I91K probably damaging 1.5x10-5 excellent candidate Nr2e1 29 39 7 Recessive Y176* probably null 3.5x10-10 excellent candidate Ovol2 34 49 7 Recessive C120Y probably damaging 1.4x10-8 excellent candidate Parp14 10 7 4 Recessive splice acceptor probably benign 9.3x10-5 good candidate Pbld2 15 10 0 Dominant splice donor probably null 3.0x10-5 potential candidate Pcid2 5 6 1 Recessive I195F probably damaging 2.8x10-5 good candidate Pigc 14 26 1 Recessive Y166C probably damaging 3.5x10-7 excellent candidate Ppp3cb 17 36 12 Semidominant splice donor probably null 1.1x10-5 excellent candidate Prkcd 14 16 3 Recessive M1T probably null 4.2x10-5 excellent candidate Prkcd 19 23 2 Recessive L498P probably damaging 3.5x10-7 excellent candidate Ptprc 5 8 2 Recessive S405P probably damaging 3.3.x10-5 excellent candidate Ptprd 14 17 4 Recessive V288A probably damaging 2.2x10-6 good candidate Rb1 8 15 6 Recessive F838S probably damaging 5.3x10-7 good candidate Slc26a4 13 15 1 Recessive C706* probably null 8.7x10-7 not good candidate Slc5a2 15 26 5 Recessive splice site probably null 1.8x10-4 not good candidate Tcf7l2 5 9 2 Recessive splice acceptor probably null 8.9x10-6 good candidate Tg 21 16 3 Recessive C53Y probably damaging 2.3x10-4 not good candidate Tg 14 17 2 Recessive I1352K possibly damaging 5.6x10-12 not good candidate Tmem79 12 22 8 Recessive Y280C probably damaging 3.0x10-9 excellent candidate Tnip1 12 15 1 Recessive splice donor probably null 3.4x10-5 excellent candidate Trappc8 13 19 5 Recessive Y126C probably damaging 8.8x10-8 excellent candidate Tubgcp3 7 7 3 Recessive S338T probably damaging 1.0x10-6 excellent candidate Ube3c 14 12 10 Recessive splice acceptor probably benign 1.2x10-7 not good candidate Ube4a 10 11 2 Recessive splice donor probably null 3.4x10-6 excellent candidate Ugt2b38 32 29 9 Recessive S144P probably benign 1.1x10-5 potential candidate Wnt9b 11 13 2 Recessive splice acceptor probably null 1.4x10-6 excellent candidate Zbtb20 24 16 0 Dominant T627K probably damaging 4.5x10-8 good candidate Ref, wild type; Het, heterozygous; Var, homozygous variant. Genes colored in green have been verified. Genes colored in red mapped during screening but were excluded. ® Mutations conferred resistance. LV: literature verified * Stop codon

MATERIALS AND METHODS

Sequencing and Determination of Candidate Genes after DSS treatment

Whole-exome sequencing and mapping were performed as described (2). Briefly, exome-enriched

DNA from all G1 mice was sequenced using the Illumina HiSeq 2500 platform. All G3 mice were genotyped across coding mutations according to their pedigree using Ion Torrent AmpliSeq custom panels as previously described (2). For the DSS-induced colitis screen, 1.3-1.5% DSS (MP

Biomedicals) was administered to G3 mice for 7 d followed by 3 d of regular drinking water (1).

To correlate mice with the genotyping results, we used the percentage of original weight loss as a continuous variable.

Verification by CRISPR Knockout and Knockin mutations

Cas9 mRNA and sgRNA were generated as described (2). Briefly, CRISPR target sites for genes were chosen using web resource CRISPR Design (crispr.mit.edu). Oligo DNA pairs were cloned into plasmid pX330 (Addgene) which was transfected into Neuro-2A cells to assess CRISPR activity by surveyor assay. To synthesize CRISPR small guide RNA (sgRNA) and cas9 mRNA, template was first amplified by PCR followed by in vitro transcription.

Microinjection of Zygotes for CRISPR Targeting of Genes.

Microinjection was performed as described (2). Female C57BL/6J mice were superovulated by injecting them with 6.5 U of pregnant mare's serum gonadotropin (PMSG; Millipore, 367222) and

20 then 6.5 U of human chorionic gonadotropin (hCG; Sigma-Aldrich, C1063) 48 h later. The superovulated females were subsequently mated with C57BL/6JJcl male mice (The Jackson

Laboratory) overnight. The following day, fertilized eggs were collected from the oviducts of the female mice, and in vitro transcribed Cas9 mRNA (50 ng/μL) and sgRNA (20–50 ng/μL) were injected into the pronucleus or cytoplasm of the fertilized eggs. The injected embryos were cultured in M16 medium (Sigma-Aldrich, M7292) at 37 °C and 95% air/5% CO2. For the production of mutant mice, two-cell stage embryos were transferred into the ampulla of the oviduct

(10–20 embryos per oviduct) of pseudopregnant Hsd:ICR (CD-1) females (Harlan Laboratories).

CHAPTER TWO

The class I myosin MYO1D binds lipid and protects against colitis

PREFACE

This chapter has been accepted for publication by Disease, Models, and Mechanisms.

Authors publishing in journals managed by The Company of Biologists retain the right to reproduce the article, in whole or in part, in a thesis or dissertation.

ABSTRACT

Myosin ID (MYO1D) is a member of the class I myosin family. We screened 49,420 third generation germline mutant mice derived from N-ethyl-N-nitrosourea-mutagenized grandsires for intestinal homeostasis abnormalities after oral administration of dextran sodium sulfate (DSS). We found and validated mutations in Myo1d as a cause of increased susceptibility to DSS-induced colitis. MYO1D is produced in the intestinal epithelium, and the colitis phenotype is dependent on the non-hematopoietic compartment of the mouse. Moreover, MYO1D appears to couple cytoskeletal elements to lipid in an ATP dependent manner. These findings demonstrate that

MYO1D is needed to maintain epithelial integrity and protect against DSS-induced colitis.

INTRODUCTION

Myosins are a large family of motor proteins that are best known for their roles in muscle contraction and cellular motility. Myosins contain a motor domain which binds filamentous actin

21 22 and hydrolyzes ATP in order to provide force for movement towards the barbed end (+) of the filament (88). These molecular motors also contain a tail domain that is capable of binding cargo

(88). Among the myosin superfamily are class I myosins which have the ability to bind cellular membranes (89). Given their ability to generate force and bind membrane, class I myosins are well-suited to perform endocytosis and endocytosis, and to regulate membrane tension (89).

Orally administered dextran sodium sulfate (DSS) damages the intestinal mucosa, creating conditions where mechanisms mediating a return to homeostasis are tested. In normal C57BL/6J mice, all the requisite mechanisms are intact. We used random germline mutagenesis to damage or destroy genes and determined that mutations in Myo1d confer hypersensitivity to low dose DSS

(79, 90-92). In this study, we report that MYO1D is necessary for the maintenance of intestinal homeostasis and provide an initial phenotypic characterization of the colitis phenotype caused by loss of MYO1D function.

RESULTS

Recessive mutations in Myo1d lead to DSS-induced colitis susceptibility

ENU and the previously described inbreeding scheme in chapter 1 were used to generate mice with random mutations in the heterozygous and homozygous state (2). We subjected 49,420 third generation (G3) mice from 1,885 pedigrees to 1.3-1.5% DSS in their drinking water, and body weights of the mice were recorded daily. Susceptibility to DSS-induced colitis, manifested as body weight loss on day 7 or 10 of DSS treatment relative to initial weight, was detected in two pedigrees, R0096 and R0244. The colitis phenotype in both pedigrees mapped to damaging Myo1d alleles using a recessive model of inheritance. The phenotypes were designated whisper (wpr;

23 pedigree R0096) and horton (htn; pedigree R0244). The whisper phenotype was mapped to mutations in three genes on chromosome 11: Ksr1 and Myo1d (p = 7.2 x 10-9) and Lypd8 (p = 1.4 x 10-8) (Fig. 2.1). The horton phenotype was mapped to mutations in two genes on chromosome

11: Cfap52 and Myo1d (p = 9.5 x 10-5, Fig. 2.2). Both Myo1d mutations were missense errors predicted to be “probably damaging” by PolyPhen-2 (51) with scores of 0.986 and 1.000 for the whisper and horton alleles, respectively. In all, eleven ENU-induced alleles of Myo1d have been tested, and five of them appear to be damaging based on their phenotypic effects (Fig. 2.3).

MYO1D is a member of the class I myosin family and contains an N-terminal motor domain and a C-terminal tail homology (TH1) domain. The whisper allele encoded a Leu972Pro substitution in the TH1 domain, and the horton allele encoded an Asn401Ile substitution in the motor domain

(Fig. 2.4A). Immunoblotting of epithelial extracts from Myo1dwpr/wpr and Myo1dhtn/htn mice showed reduced levels of MYO1D protein (Fig. 2.4B), suggesting that both mutations affect protein stability.

To verify that Myo1d mutations were causative of phenotype in these pedigrees, we crossed horton heterozygotes (Myo1d+/htn) with whisper heterozygotes (Myo1d+/wpr) to generate Myo1d compound heterozygotes with simple heterozygosity for all other ENU-induced mutations.

Myo1dhtn/wpr mice remained susceptible to DSS challenge with 20% weight loss by day 9 of treatment, validating causation (Fig. 2.5A). To further confirm that mutations in Myo1d result in a DSS-induced colitis susceptibility phenotype, CRISPR/Cas9-mediated targeting was used to generate a 96 bp insertion (96ins) in Myo1d, which resulted in the in-frame addition of 32 amino acids in the TH1 domain of the MYO1D protein (Fig. 2.5B). During low-dose DSS treatment,

Myo1d96ins/96ins mice lost approximately 10% of body weight by day six, and more than 25% of

24 body weight by day eight (Fig. 2.5B). Weight loss was accompanied by higher disease activity index (DAI), colonic shortening, and increased expression of proinflammatory genes (Cxcl2, Ifng,

Il6, Il1b, Nos2, and Tnf) in distal colons of Myo1d96ins/96ins mice (Fig. 2.6A-C). Colons from DSS- treated Myo1d96ins/96ins mice showed marked histopathological changes characterized by infiltration of lymphocytes and loss of crypt architecture (Fig. 2.6D). DSS treatment was lethal for

Myo1d96ins/96ins mice by day 10.

Myo1d mutations sensitize to DSS in a hematopoietic extrinsic manner

Mutations can impair homeostasis through effects on the GI epithelium, the hematopoietic compartment, or both. To determine the relative contributions of hematopoietic and extrahematopoietic compartments, we generated bone marrow chimeric mice. Donors and/or recipients were either CD45.1 or horton (CD45.2) strains; chimeras were completely reconstituted with donor bone marrow as assessed by flow cytometry. After 1.4% DSS administration, chimeric

CD45.1 mice with horton hematopoietic cells did not exhibit significant weight loss, similar to control CD45.1 mice receiving transplants of WT bone marrow (Fig. 2.7A). Chimeric horton recipient mice, irrespective of donor bone marrow, were not protected from DSS challenge, and lost an average of 25% of their initial body weight by day eight. The weight loss coincided with reduced colon length (Fig. 2.7B) and increased rectal bleeding and diarrhea (Fig. 2.7C). Myo1d expression was previously reported in the kidney and brain (93). To determine whether increased

DSS-water consumption contributed to the colitis phenotype, we challenged Myo1dhtn/wpr mice with DSS by oral gavage. Myo1d+/+ mice exhibited no weight loss, while Myo1dhtn/wpr mice were sensitive to DSS administration and lost approximately 25% of their initial body weight by day

25 eight (Fig. 2.7D). These data demonstrate that functional MYO1D in non-hematopoietic cells is necessary for restoration of intestinal homeostasis following DSS challenge.

Myo1d mutant mice have normal intestinal differentiation

In wild-type enterocytes of the small intestine, MYO1D localizes to the lateral membrane, terminal web, and microvillar tips (94). We found that in the colons of wild-type mice, MYO1D was chiefly produced by colonocytes, in which it was localized primarily to the basolateral membrane (Fig. 2.8A). In Myo1d96ins/96ins mice, it was either absent or mislocalized (Fig. 2.8A), suggesting that the mutant phenotype might result from an epithelial specific defect. Myosin IA, the Drosophila homolog of MYO1D, interacts with β-catenin and mutations in Myosin IA lead to defects in left-right asymmetry (95, 96). Wnt/β-catenin signaling maintains homeostasis of the intestinal epithelium by regulating the balance between cell proliferation, differentiation, and death

(97). Therefore, we examined if epithelial homeostasis was altered due to loss of MYO1D function without environmental insult. Similar numbers of proliferative cells (Ki-67 positive), and goblet,

Paneth, and enteroendocrine cells were detected in Myo1d96ins/96ins and wild-type colon or ileum epithelium (Fig. 2.8A,B). These findings suggest that intestinal epithelial differentiation is normal in Myo1d mutant mice.(97). Therefore, we examined if epithelial homeostasis was altered due to loss of MYO1D function without environmental insult. Similar numbers of proliferative cells (Ki-

67 positive), and goblet, Paneth, and enteroendocrine cells were detected in Myo1d96ins/96ins and wild-type colon or ileum epithelium (Fig. 2.8A,B). These findings suggest that intestinal epithelial differentiation is normal in Myo1d mutant mice.

26 MYO1D couples membrane lipids to actin filaments

To better understand the role of MYO1D in the epithelium, we investigated its biochemical properties. The TH1 domain of class I myosins is capable of binding lipid moieties (89). To determine the lipids that MYO1D binds, we purified recombinant human MYO1D, which shares

98% sequence identity with mouse MYO1D, and assessed binding to lipid strips (Fig. 2.9A).

MYO1D exhibited strongest affinity for phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) and some binding to phosphatidylinositol (3,4,5)-trisphosphate (PI(3,4,5)P3). To further assess the lipid binding of MYO1D, we performed a complementary liposome co-flotation assay (Fig. 2.9B).

MYO1D showed highest co-flotation with liposomes containing both PIP2 and PIP3. The N- terminal motor domain of class I myosins mediates actin binding (89). To confirm the association of MYO1D with the actin cytoskeleton in the colon epithelium, we examined the association of

MYO1D with the particulate fraction of NP-40-treated mouse colonic epithelial cell lysates (Fig.

2.10). Under these conditions, the majority of MYO1D remained insoluble. Treatment of the insoluble pellet with 2 mM ATP resulted in near complete solubilization of MYO1D, consistent with ATP disruption of the myosin-actin rigor conformation. These data suggest that MYO1D binds to PIP2 or PIP3 and to filamentous actin in intestinal epithelial cells.

FIGURES

Figure 2.2: Mapping of the Myo1d mutation to the whisper phenotype. (A) Percentage of initial body weight on day 10 of DSS treatment plotted vs. genotype (REF, Myo1d+/+ (n=12); HET, Myo1d+/wpr (n=19); VAR, Myo1dwpr/wpr (n=10)). (B) Manhattan plot showing P values of association between the whisper phenotype and mutations identified in the whisper pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. P values for linkage of mutations in Myo1d, Ksr1, and Lypd8 with the whisper phenotype are indicated.

Figure 2.2: Mapping of the Myo1d mutation to the horton phenotype. (A) Percentage of initial body weight on day 7 of DSS treatment plotted vs. genotype (REF, Myo1d+/+ (n=4), HET, Myo1d+/htn (n=10), VAR, Myo1dhtn/htn(n=3)). (B) Manhattan plot showing P values of association between the horton phenotype and mutations identified in the horton pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. P values for linkage of mutations in Myo1d and Cfap52 with the horton phenotype are indicated.

Figure 2.3: Mapping of Myo1d in multiple pedigrees. (A) Percent of initial body weight plotted vs. Myo1d genotype for mice from 11 unrelated pedigrees (R0069, R0081, R0096, R0244, R0711, R1084, R3429, R3917, R3928, R4723, R5042) with distinct Myo1d mutations. (F) Manhattan plot showing P values of association between Myo1d and DSS phenotype calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively.

Figure 2.4: Protein domain organization of mouse MYO1D and production of MYO1D mutant alleles. (A) The locations of the horton, whisper, whisper2, and whisper3 mutations are shown. The horton mutation is an asparagine (N) to isoleucine (I) substitution at amino acid 401 in the head domain (N401I). The whisper mutation is a leucine (L) to proline (P) substitution at position 972 (L972P) in the TH1 domain. The whisper2 mutation is a valine (V) to glutamic acid (E) substitution at position 512 (V512E) in the head domain. The whisper3 mutation is a aspartic acid (D) to a valine (V) substitution at 926 (D926V) Abbreviations: TH1, tail homology-1; IQ, IQ motifs (IQXXXRGXXXR/K, where X is any amino acid) (B) Representative immunoblot showing MYO1D levels in colon epithelium isolated from wild-type, horton, whisper, and Myo1d96ins/96ins mice. GAPDH was used as a loading control (n = 3 samples from each genotype).

Figure 2.5: Validation that Myo1d mutations cause a colitis phenotype. (A) Weight loss analysis of Myo1d+/+, Myo1d+/htn, Myo1d+/wpr, Myo1dhtn/wpr after 1.4% DSS treatment (n=4 for all groups). (B) Weight loss analysis of Myo1d+/+ (n=5), Myo1d+/96ins mice (n=12), and Myo1d96ins/96ins mice (n=6) after 1.4% DSS treatment. Each experiment was performed a minimum of three times. Data are expressed as means ± s.d. and significance was determined by two-way analysis of variance (ANOVA) with Dunnett’s multiple comparisons (A,B) or unpaired Students t-test (C-E) (*P<0.05, **P<0.01, ***P<0.001. ****P<0.0001).

Figure 2.6: Colitis readouts for Myo1d96ins/96ins mice. (A) Disease activity index (DAI) and (B) colon length (cm) of Myo1d+/96ins (n=8) and Myo1d96ins/96ins (n=8) mice eight days after DSS challenge. (C) Relative mRNA expression levels of inflammatory cytokines Cxcl2, Ifng, Il1b, Il6, Nos2, and Tnf in the distal colon of Myo1d+/96ins (n=6) and Myo1d96ins/96ins (n=6) mice after four days of 1.4% DSS. (D) Representative H&E staining of Myo1d96ins/96ins colons after seven days of DSS treatment. For all experiments, 1.4% DSS was administered to mice. Each experiment was performed a minimum of three times. Data are expressed as means ± s.d. and significance was determined by two-way analysis of variance (ANOVA) with Dunnett’s multiple comparisons (A,B) or unpaired Students t-test (C-E) (*P<0.05, **P<0.01, ***P<0.001. ****P<0.0001).

Figure 2.7: Non-hematopoietic defects contribute to colitis. (A-C) Bone marrow chimeras were generated, and percent initial weight (A), colon length (B), and DAI (C) were determined after 1.4% DSS administration (n=5 for all groups). (D) Percent initial weight loss of Myo1d+/+ (n=4) or Myo1dhtn/wpr (n=5) mice after daily oral gavage with 2.5g/kg DSS. Data are expressed as means ± s.d. and significance was determined by two-way analysis of variance (ANOVA) with Dunnett’s multiple comparisons (A), one-way ANOVA with Dunnett’s multiple comparisons (B,C) or unpaired Students t-test (D) (*P<0.05, **P<0.01, ***P<0.001. ****P<0.0001).

Figure 2.8: Myo1d mutant mice exhibit normal intestinal epithelial cell development. (A) Representative images of immunofluorescence staining of ileum or colonic tissues for MYO1D, Ki-67, mucin 2 (Muc2), Dcamkl1, lysozyme (Lyz), and nuclei in Myo1d+/+ and Myo1d96ins/96ins mice. (B) Quantification of proliferating (Ki-67-positive) goblet (Muc2-positive), tuft (Dcamkl1- positive), and Paneth (LYZ-positive) cells in Myo1d+/+ (n=3) and Myo1d96ins/96ins (n=3) animals.

Figure 2.9: Myosin 1D functions in lipid binding. (A) Representative images of FLAG-MYO1D bound to a lipid-coated strip membrane. FLAG alone bound to lipid-coated strip membrane was used as a negative control. Abbreviations: LPA, lysophosphatidic acid; S1P, sphingosine-1- phosphate; LPC, lysophosphocholine; PI, phosphatidylinositol; PI(3)P, PI-(3)-phosphate; PI(4)P, PI-(4)-phosphate; PI(5)P, PI-(5)-phosphate; PI(3,4)P2, PI-(3,4-) bisphosphate; PI(3,5)P2, PI-(3,5)- bisphosphate; PI(4,5)P2, PI-(4,5)-bisphosphate; PI(3,4,5)P3, PI-(3,4,5)-trisphosphate; PA, phosphatidic acid; PE, phosphatidylethanolamine; PS, phosphatidylserine; PC, phosphatidylcholine. (B) Representative immunoblot of the products of a liposome/protein cofloatation assay. Basal liposomes were composed of phosphatidylcholine, phosphatidylethanolomine, and diacylglycerol.

Figure 2.10: Myosin 1D functions in actin binding. Representative immunoblot of supernatant (s) and pellet (p) fractions of colonic epithelial cells lysed with 1% NP-40. The NP-40 pellet fraction was subsequently treated with 2 mM ATP in the absence of detergent. GAPDH was used to indicate solubilization of cytoplasm by 1% NP-40.

37 DISCUSSION

Using forward genetics, we demonstrated that damaging mutations in the class I myosin

MYO1D led to a defect that rendered mice susceptible to DSS-induced colitis. Using multiple

ENU alleles and CRISPR/Cas9 targeting, we validated causation. MYO1D appears to function as a molecular link between membrane lipids and the actin cytoskeleton.

The whisper phenotype originally mapped to three putative ENU-introduced mutations in

Myo1d, Ksr1, and Lypd8. Both kinase suppressor of Ras-1 (Ksr1) and Ly6/PLAUR domain containing 8 (Lypd8) knockout (KO) mice are known to be susceptible to DSS-induced colitis (98,

99). However, it is not clear that the alleles generated here confer susceptibility, and MYO1D was shown to be essential for resistance to DSS challenge by complementation testing, as well as gene targeting.

Functional studies of class I myosins in the intestinal epithelium have so far been limited to myosin-1a (MYO1A). MYO1A associates with brush border membrane rafts and is required for the localization or retention of sucrase-isomaltase and cystic fibrosis transmembrane conductance regulator channels (100, 101). Additionally, MYO1A powers microvillar membranes to propel alkaline phosphatase-laden vesicles into the lumen (102). Myo1a KO mice have many perturbations at the cellular level, but fail to show any phenotype at the level of the whole organism without environmental stress presumably because of compensation from other class I myosins including MYO1C and MYO1D (103). Myo1a KO mice do exhibit increased mortality with 3%

DSS treatment, but the death rate of the Myo1d mutant mice observed in response to 1.4% DSS in our study suggests that MYO1D has evolved a non-redundant function that is not compensated by other class I myosins (103).

38 The physiologic function of MYO1D in the epithelium required to maintain intestinal homeostasis remains unknown, although several possibilities exist. One attractive hypothesis is that MYO1D regulates vesicle trafficking in the epithelium. In the epithelial Madin–Darby Canine

Kidney cell line, MYO1D is required for trafficking of transferrin from apical and basolateral early endosomes to recycling endosomes (104). It is possible that MYO1D plays a critical role in trafficking adhesion molecules to the basolateral membrane, where they are necessary for the integrity of adherens and tight junctions that are disrupted by DSS.

Class I myosins are well-suited to regulate membrane dynamics due to their ability to bind both actin and membrane lipids (89). Assessment of MYO1D binding properties suggest the protein likely binds membrane lipids and actin in vivo. DSS treatment wounds the epithelium and requires cell migration into the damaged area for epithelial restitution. Remodeling of the actin cytoskeleton underlies the membrane dynamics necessary for cell migration and disruption of actin remodeling is sufficient to cause chemically-induced colitis in Villin-/- mice (105, 106). Collective cell migration also requires planar cell polarity which is disrupted in rats lacking MYO1D (107).

Thus, impaired cell migration due to disruption of actin cytoskeletal remodeling may also contribute to DSS-induced colitis susceptibility in Myo1d mutant mice.

The DSS-induced colitis model does not recapitulate all aspects of human IBD, eliciting a different profile of T cell proinflammatory cytokines for example (108, 109). However, DSS- induced colitis shares many similarities with human IBD, particularly with ulcerative colitis, including clinical manifestations and organ pathogenesis (108, 109). Moreover, orthologs or paralogs of several genes linked to monogenic forms of human IBD have been implicated in DSS- induced colitis in mice by our laboratory (79, 90, 110) and others (48-50, 111, 112), indicating that

39 molecular mechanisms relevant to human IBD can be uncovered through study of the DSS model.

The present study identified an essential and non-redundant role for MYO1D in restoring intestinal homeostasis following disruption of the epithelial barrier in mice and should encourage consideration of MYO1D as a candidate locus in work seeking to identify genetic causes of IBD in humans.

MATERIALS AND METHODS

Mice

ENU mutagenesis and DSS-colitis induction were performed as previously described in Chapter

1. Mice gavaged with DSS received 2.5g DSS/kg of bodyweight daily for seven days via a gavage feeding needle. Body weight was recorded daily and reported as the amount of weight loss from the pre-treatment weight. Disease activity index score is a composite score of weight loss, stool bleeding and stool consistency determined as previously described (113). Briefly: weight loss: 0

(no loss), 1 (1–10% loss of body weight), 2 (10–15% loss of body weight), 3 (15–20% loss of body weight), and 4 (> 20% loss of body weight); stool consistency: 0 (normal), 2 (loose stool), and 4

(diarrhea); and bleeding: 0 (no blood), 1 (hemoccult positive), 2 (hemoccult positive and visual pellet bleeding), and 4 (gross bleeding and/or blood around anus). All mice were housed in the

University of Texas Southwestern vivarium.

Generation of the Myo1d96ins/96ins mouse strain using the CRISPR/Cas9 system

To generate the Myo1d96ins/96ins mouse strain, embryos were collected as described in Chapter 1.

The following day, fertilized eggs were collected from the oviducts and in vitro–transcribed Cas9

40 mRNA (50 ng/μl) and Myo1d small base-pairing guide RNA (50 ng/μl; 5’-

CCGTGCAGGCTGCACTGCACCGG-3’) were injected into the cytoplasm or pronucleus of the embryos. The injected embryos were cultured and transferred as previously described in Chapter

1. The resulting Myo1d96ins/96ins mice contain a 96 base pair insertion (5’-

ACAGAGAAGAAATGCACCGTCTCTGTGGAGACCCGGCTCAATCAGCCACAGCCTGA

CTTCACCAAGACCCGGCTCAATCAGCCACAGCCTGACTTC-3’) between nucleotides

80484343-80484342 on chromosome 11. Inserted sequence is composed of multiple duplication events from exon 22 of Myo1d. This insertion results in the in-frame addition of 32 amino acids

(TEKKCTVSVETRLNQPQPDFTKTRLNQPQPDF) between residues 968 and 969 of the protein.

Crypt Isolation

Colonic crypts were isolated as previously described (91). Briefly, colons were isolated from mice and stool removed from the lumen. Colons were then cut to 5-10 mm pieces and incubated at room temperature for 30 minutes in phosphate-buffered saline (PBS) containing 5 mM ethylenediaminetetraacetic acid (EDTA).

Antibodies

The following antibodies were used in this study: GAPDH (1:1000, Cell Signaling Technologies

D16H11), Ki-67 (1:200, Cell Signaling Technologies D3B5), Muc2 (1:200 Santa Cruz

Biotechnology H-300), myosin ID (1:200 (immunofluorescence) or 1:1000 (western blotting)

41 Santa Cruz Biotechnology H-60), Lysozyme (1: 200 DAKO 3.2.1.17), and Dcamkl1 (1:300

Abcam ab37994).

Quantitative RT-PCR

Total RNA from colonic epithelium was isolated using TRIzol reagent (Thermofisher) according to the manufacturer’s instructions. DNase treatment and clean-up was performed with the DNA- free DNase Treatment and Removal Reagents kit (Thermofisher). The isolated RNA was subsequently purified on a silica column (Invitrogen) to remove any excess DSS, which can interfere with the reverse transcriptase reaction. One microgram of RNA was reverse-transcribed to cDNA with SuperScript III First-Strand Synthesis System for RT-PCR (Life Technologies).

Transcript levels of Cxcl2, Ifng, Il1b, Il6, Nos2, and Tnf were analyzed using iTaq Universal SYBR

Green Supermix (Bio-Rad) on a Step One Plus Real-Time PCR System (Life Technologies) with the following primers: Cxcl2; 5’-GCTTCCTCCTTCCTTCTGGT-3’, 5’-

GGGCAGAAAGCTTGTCTCAA-3’; Ifng: 5’-TGAGCTCATTGAATGCTTGG-3’, 5’-

ACAGCAAGGCGAAAAAGGAT-3’; Il1b: 5’-GGTCAAAGGTTTGGAAGCAG-3’, 5’-

TGTGAAATGCCACCTTTTGA-3’; Il6: 5-GTCAGGGGTGGTTATTGCAT-3’, 5’-

AGTGAGGAACAAGCCAGAGC-3’; Nos2 5’- TTCTGTGCTGTCCCAGTGAG-3’, 5’-

TGAAGAAAACCCCTTGTGCT-3’; and Tnf: 5’-AGATGATCTGACTGCCTGGG-3’, 5’-

CTGCTGCACTTTGGAGTGAT-3’. Relative expression was calculated using the ΔΔCt standardization method using Gapdh (5’-TTGATGGCAACAATCTCCAC-3’ and 5’-

CGTCCCGTAGACAAAATGGT-3’).

42 Lipid strip binding assay

HEK293T cells (ATCC CRL-11268, authenticated and used between passages 5-10) overexpressing Human FLAG-MYO1D were lysed in Nonidet P-40 buffer [50 mM Tris⋅Cl, pH

8.0, 0.1 M NaCl, 1% (vol/vol) Nonidet P-40, 10% (vol/vol) glycerol, 1.5 mM EDTA, and Protease

Inhibitor Mixture]. Immunoprecipitation was performed using anti-FLAG M2 agarose (Sigma-

Aldrich). Binding of purified FLAG-MYO1D to lipid was analyzed using commercially available lipid strips (Echelon Biosciences) according to the manufacturer’s instructions.

Liposome co-floation assay

Phosphatidylcholine, PC; Phosphatidylethanolamine, PE; Phosphatidylserine, PS; Diacylglycerol,

DAG; Cholesterol, Chol; PI-(4,5)-bisphosphate, PI(4,5)P2; and PI-(3,4,5)-trisphosphate,

PI(3,4,5)P3 were purchased from Avanti Polar Lipid. Liposomes were composed of

PC:PE:Chol:DAG (58:20:20:2), PC:PE:Chol:DAG:PS (40:20:20:2:18),

PC:PE:Chol:DAG:PS:PIP2 (38:20:20:2:18:2), PC:PE:Chol:DAG:PS:PIP3 (38:20:20:2:18:2), or

PC:PE:Chol:DAG:PS:PIP2:PIP3 (36:20:20:2:18:2:2). Lipid mixtures were dried in glass tubes with nitrogen gas and under vacuum overnight followed by hydration in Tris-buffered saline (TBS, pH

7.6). Liposomes were prepared by extrusion 30 times through a 100-nm polycarbonate membrane.

MYO1D protein was purified as described above. Liposome protein mixtures were incubated at room temperature for 1 hour. Liposomes were isolated by floatation on a Histodenz density gradient (40%:35%:30%) as described and analyzed for binding to MYO1D by western blotting.

Immunofluorescence

43 Intestine was rapidly dissected and flushed with cold PBS, cut into small 3 mM concentric circles, and fixed in freshly prepared 4% PFA for 1 hr at room temperature. Tissue was cryoprotected in

15% sucrose for 1 hr and 30% sucrose overnight. Tissues were embedded in liquid nitrogen-cooled isopentane in OCT in the same mold to ensure identical processing for each sample. 7 uM sections were produced using a Leica cryostat. Sections were washed with PBS 3X for 5 minutes, and then blocked in 10% bovine serum albumin (BSA) in PBS. Primary antibodies were diluted in 5% BSA in PBS. Sections were incubated overnight at 4°C and then washed 3X for 10 minutes. Slides were then incubated for 1 hour in Alexa fluor antibodies (1:1000) in 5% BSA in PBS and washed 3X for 10 minutes. Slides were mounted in Prolong Antifade Gold and visualized using a Zeiss

LSM880 confocal microscope.

Statistical analysis

Age- and sex-matched mice were randomly allocated to experimental groups based on their genotypes. No pre-specified effect size was assumed, and 3-12 mice per genotype were used in experiments; this sample size was sufficient to demonstrate statistically significant differences in comparisons between two or more unpaired experimental groups by unpaired t-test or ANOVA, respectively. All mice were included during data analysis. All statistical analyses were performed using GraphPad Prism. Two-tailed student’s t-test was utilized for comparisons of a single parameter between two groups, one-way ANOVA with Dunnett’s test was utilized for comparison of one parameter between multiple groups, and two-way ANOVA with Dunnett’s test was utilized for comparison of two parameters between multiple groups. Because mice utilized in this study were inbred and age- and sex-matched, variance was assumed to be similar between treatment

44 groups. Phenotypic data were assumed to follow a normal distribution, as has been observed in large datasets from numerous phenotypic screens conducted by our group.

CHAPTER THREE

Loss of SMCR8 or WDR41 causes immune dysregulation and colitis susceptibility

PREFACE

This chapter is in review for publication by PNAS. Authors publishing in PNAS retain the right to reproduce the article, in whole or in part, in a thesis or dissertation.

ABSTRACT

The SMCR8-WDR41-C9ORF72 (SWC) complex is a regulator of autophagy and lysosomal function. Autoimmunity and inflammatory disease have been ascribed to loss-of-function mutations of Smcr8 or C9orf72 in mice. In humans, autoimmunity has been reported to precede amyotrophic lateral sclerosis caused by mutations of C9ORF72. However, the cellular and molecular mechanisms underlying autoimmunity and inflammation caused by C9ORF72 or

SMCR8 deficiencies remain unknown. Here, we show that splenomegaly, lymphadenopathy, and activated circulating T cells observed in Smcr8-/- mice were rescued by triple knockout of the endosomal Toll-like receptors (TLR), TLR3, TLR7, and TLR9. Myeloid cells from Smcr8-/- mice produced excessive inflammatory cytokines in response to endocytosed TLR3, TLR7, or TLR9 ligands administered in the growth medium, and in response to TLR2 or TLR4 ligands internalized by phagocytosis. These defects likely stem from prolonged TLR signaling caused by accumulation of LysoTracker positive vesicles, and by delayed phagosome maturation, both of which were observed in Smcr8-/- macrophages. Smcr8-/- mice also showed elevated susceptibility to dextran

45 46 sodium sulfate-induced colitis, which was not associated with increased TLR3, TLR7, or TLR9 signaling. Deficiency of WDR41 phenocopied loss of SMCR8. Our findings provide evidence that excessive endosomal TLR signaling resulting from prolonged ligand-receptor contact causes inflammatory disease in SMCR8 deficient mice.

INTRODUCTION

The lysosome is a dynamic organelle that serves as the degradative endpoint for endocytosis, phagocytosis, and autophagy. Recently, the SMCR8-WDR41-C9ORF72 (SWC) tripartite complex has been implicated as an important regulator of membrane trafficking pathways that converge on the lysosome (114-121). Two members of this complex, SMCR8 and C9ORF72, contain DENN domains (122, 123) which are commonly found in proteins that act as guanine nucleotide exchange factors (GEF) for specific Rab GTPases, molecular switches that govern membrane traffic within eukaryotes. Indeed, the SWC complex has been shown to provide GEF activity for RAB8A and RAB39B (116, 121) and interact with a multitude of other Rabs (116,

119-121). A role for the SWC complex in autophagy (115-119) and endocytic transport (120) has been established by studies documenting defects in these processes in the absence of a single complex component. However, the precise function of the SWC complex or its individual members remains poorly defined.

Hexanucleotide repeat expansions in C9ORF72 are the most common genetic cause of amyotrophic lateral sclerosis and frontotemporal dementia (ALS/FTD) in humans (124).

Numerous epidemiologic studies have also documented an increased prevalence of autoimmune diseases in patients with ALS/FTD with or without C9ORF72 mutations, leading to the proposal

47 that these diseases are linked to the same immunological mechanism(s) (125-128). Studies in mice support a causal role for SWC complex dysfunction in autoimmunity. Several groups have reported that loss of C9ORF72 function leads to autoimmunity characterized by enlarged peripheral immune organs, elevated proinflammatory cytokines, and autoantibody production

(114, 115, 129, 130). More recently, splenomegaly, lymphadenopathy, and increased autoantibody production in Smcr8-/- mice were reported (131), recapitulating the central phenotypes observed in

C9orf72-/- mice. Macrophages derived from C9orf72-/- mice exhibit increased lysosomal number and exaggerated responses to the proinflammatory stimuli peptidoglycan, CpG, and silica (114).

C9orf72-/- mice exhibited elevated splenic levels of LC3, P62, and LAMP1 supporting an in vivo role for the protein in regulation of autophagy (114, 115). The primary cause of spontaneous inflammation in C9orf72-/- and Smcr8-/- mice has not been identified, but altered autophagy (114,

115, 130) and elevated lysosomal exocytosis (131) have been proposed as potential mechanisms.

Here we show that mutations in Smcr8 or Wdr41 cause increased disease activity in response to DSS. Loss of complex function causes impaired acidification of phagocytosed cargo by the lysosome that corresponds with increased cytokine and ROS production. Additionally, we establish that loss of SMCR8 or WDR41 function leads to exaggerated signaling from TLR3,

TLR7, and TLR9, innate immune receptors sensitive to nucleic acids and whose function is restricted to endolysosomal compartments. Defective lysosome and phagosome maturation in

SMCR8-deficient mice causes protracted endosomal TLR stimulation which leads to spontaneous inflammation. Collectively, these data establish a role for SMCR8 and WDR41 in trafficking of cargo to the lysosome and identify these molecules as important regulators of myeloid cell responses to immunogenic cargo.

RESULTS

Hematopoietic deficiency of SMCR8 increases sensitivity to DSS-induced colitis

To identify genes necessary for intestinal homeostasis, we performed a forward genetic screen in which G3 mutant mice were treated with 1.3-1.5% DSS in their drinking water for 7 days and then switched to regular drinking water for an additional 3 days (1). Body weights were recorded on day 0, 7, and 10 with respect to initiation of DSS treatment and percent initial body weight was used as a continuous mapping variable.

Among several phenovariants discovered (1), linkage mapping identified Smcr8 as a putative causative gene in two ancestrally unrelated pedigrees. The first allele, patriot, resulted in a likely damaging missense mutation (PolyPhen-2 score = 1.0) causing an isoleucine-to-threonine substitution in the second amino acid (I2T) of the protein. Homozygosity for the patriot mutation caused 20% weight loss by day 7 (P = 9.05x10-7) (Fig. 3.1). A second phenotype, patriot2, in which homozygotes lost 20% of their initial weight when measured on day 10, was mapped to co- segregating mutations in Lgl1 and Smcr8 (P = 7.7x10-6), of which the latter was considered a better candidate based on association of multiple alleles with the phenotype (Fig. 3.2). The patriot2 mutation (PolyPhen-2 = 0.999) resulted in a methionine-to-valine substitution of the first amino acid (M1V) presumably destroying the translation start codon. To validate the mapping data,

CRISPR/Cas9 targeting was utilized to generate mice with the I2T mutation (Smcr8I2T/I2T) originally detected in the patriot pedigree, as well as a frameshift mutant with a putative null allele

(Smcr8-/-). Mice homozygous for these CRISPR alleles displayed increased susceptibility to DSS

49 treatment compared to wild-type controls, exhibiting 20% weight loss by day 8 (Fig. 3.3). Both

CRISPR lines exhibited an elevated disease activity index (DAI) including weight loss, severe diarrhea, and intestinal bleeding, and colonic shortening (Fig. 3.4A,B). Histopathological alterations characterized by infiltrating leukocytes and loss of crypt architecture were also apparent in Smcr8-/- and Smcr8I2T/I2T mice (Fig. 3.4C).

To delineate the contributions of the hematopoietic compartment to the colitis phenotype, we generated bone marrow chimeras. Wild-type mice reconstituted with Smcr8-/- bone marrow exhibited increased body weight loss, DAI, and colonic shortening compared with those reconstituted with wild-type bone marrow, suggesting that SMCR8 functions in hematopoietic cells to protect against DSS-induced colitis (Fig. 3.5A,B).

Splenomegaly, lymphadenopathy, T cell activation, and elevated IL-12p40 in Smcr8-/- mice

Smcr8 CRISPR mice spontaneously exhibited signs of immune dysregulation.

Splenomegaly and lymphadenopathy were observed in 9-12 month old Smcr8-/- and Smcr8I2T/I2T mice (Fig. 3.6). In the spleens of these mice, percentages of major immune cell populations including macrophages, monocytes, neutrophils, B and T cells remained unchanged compared to wild-type mice (Fig. 3.7) Within the T cell compartment, we observed reduced percentages of naïve (CD3+CD62L+CD44-) and increased percentages of activated (CD3+CD44+CD62L-) cells in both CD4+ and CD8+ lineages, a phenotype that was present in spleen as well as peripheral blood

(Fig. 3.8). 23plex cytokine array analysis revealed elevated plasma levels of IL-12p40 in Smcr8-/- and Smcr8I2T/I2T mice compared to control animals (Fig. 3.9). These data indicate that an overabundance of myeloid and activated lymphoid cells develop in mice lacking SMCR8 function.

SMCR8 negatively regulates endosomal TLR signaling

To identify genes involved in the regulation of TLR signaling, we carried out a forward genetic screen in which peritoneal macrophages harvested from G3 mice carrying ENU-induced mutations were assayed for tumor necrosis factor (TNF) secretion after stimulation with TLR ligands in the culture medium. Using superpedigree analysis, a method that evaluates genotype- phenotype associations pooled from multiple pedigrees, we found that damaging mutations in

Smcr8 led to increased production of (TNF) in response to CpG oligodeoxynucleotides, a TLR9 ligand, in two ancestrally unrelated pedigrees (R1418 and R3960) (Fig. 3.10). Stimulation of peritoneal macrophages from Smcr8-/- and Smcr8I2T/I2T mice with CpG resulted in increased production of TNF compared to wild-type macrophages validating the original linkage mapping

(Fig. 3.11A).

Further investigation revealed elevated TNF responses were also induced by activation of

TLR3 with poly(I:C), or activation of TLR7 with R848 (Fig. 3.11A). Contrastingly, stimulation of TLR2 with Pam3CSK4 or TLR4 with LPS resulted in normal levels of TNF production (Fig.

3.11B). Similar defects of IL-6 production were observed in Smcr8-/- bone marrow derived dendritic cells (BMDCs), which displayed excessive IL-6 responses to poly(I:C), R848, and CpG, but normal responses to Pam3CSK4 and LPS. (Fig. 3.12). These findings show that SMCR8 is necessary to limit signaling from endosomal TLRs in macrophages and myeloid dendritic cells.

Knockout of endosomal TLR signaling rescues spontaneous inflammation

51 To determine whether endosomal TLR signaling contributed to the inflammatory phenotypes described above, we bred Smcr8-/- mice onto a Tlr3-/-;Tlr7-/-;Tlr9-/- (Tlr3/7/9-/-) background. Strikingly, spleen and lymph node size were restored to normal levels in Smcr8-/-

;Tlr3/7/9-/-mice (Fig. 3.13). Hyperactivation of T cells and high circulating plasma levels of IL-

12p40 observed in Smcr8-/- mice were also rescued to control levels on a Tlr3/7/9-/- background

(Fig. 3.14,15). To determine if exaggerated endosomal TLR activation in Smcr8-/- hematopoietic cells causes colitis susceptibility, we reconstituted lethally irradiated wild-type mice with bone marrow from Smcr8+/-, Smcr8-/-, Smcr8+/-;Tlr3/7/9-/-, or Smcr8-/-;Tlr3/7/9-/- mice and treated them with DSS to induce colitis (Fig 3.15). Recipients of Smcr8-/- bone marrow showed weight losses similar to recipients of Smcr8-/-;Tlr3/7/9-/- bone marrow on day 10 of the DSS treatment protocol

(Fig. 3.15). These data demonstrate that exaggerated signaling from endosomal TLRs results in immune cell hyperplasia and activation in SMCR8-deficient mice. However, distinct pathway(s) may be causative for the colitis phenotype.

SWC complex regulates acidification of cargo

TLR3, TLR7, and TLR9 engage ligands and signal from the endolysosomal compartment, a vesicular compartment formed by late endosome-lysosome fusion (132). Degradation of endocytic cargo by lysosomal hydrolases begins in endolysosomes and continues in lysosomes following maturation from endolysosomes, which involves gradually increasing acidification of the lumen. C9ORF72 reportedly localizes to lysosomes in 293T and neuronal cells (133).

Moreover, we found that Smcr8-/- macrophages exhibited an accumulation of putative lysosomes

(LysoTracker-positive vesicles) (Fig. 3.17), a phenotype also reported for C9ORF72-deficient

52 cells (114). Considering the hyperactive TLR3/7/9 signaling in Smcr8-/- cells, this finding suggests that a block in the cycle of lysosome function might result in failure to degrade endolysosomal cargo including TLRs and their ligands, leading to prolonged ligand-receptor interaction and consequently increased signaling.

Because TLR2 and TLR4 can become incorporated and signal from phagosomes following engulfment of a particle ligand (134-136), we assessed maturation and acidification of cargo within the phagocytic pathway. We utilized pHrodo bioparticle conjugates which exhibit a pH-dependent fluorescence that is maximal in the acidic environment of the phagolysosome. Smcr8-/- macrophages exhibited reduced MFI upon incubation with zymosan and E. coli pHrodo

BioParticles, but normal MFI after uptake of pH-insensitive BioParticles (Fig. 3.18,19), indicating normal uptake of cargo but impaired acidification of the phagolysosome. Similar to the exaggerated responses to endosomal TLR ligands, Smcr8-/- BMDMs produced increased levels of

TNF in response to zymosan (TLR2 ligand) and E. coli BioParticles (TLR4 ligand) (Fig. 3.20).

These findings support the idea that impaired vesicle acidification resulting in failure to degrade cargo in phagolysosomes leads to extended ligand-receptor contact and increased TLR signaling.

Phagosomal pH is dictated primarily by the activities of the vacuolar ATPase (V-ATPase) and

NADPH oxidase (NOX2). Generation of superoxide anions by NOX2 and their subsequent dismutation into hydrogen peroxide consumes protons; thus, NOX2 activity increases the pH of the phagosome. To assess levels of reactive oxygen species (ROS) within the phagosome, zymosan particles were covalently coupled with the ROS-sensitive dye OxyBURST Green H2DCFDA

53 (2′,7′-dichlorodihydrofluorescein diacetate) which fluoresces brightly upon conversion to dichlorofluorescein by oxidation. Higher MFI levels were detected in Smcr8-/- macrophages after incubation with OxyBURST-zymosan, indicating that impaired acidification is at least in part caused by elevated NOX2 activity (Fig. 3.21).

Wdr41 mutant mice phenocopy Smcr8 mutant mice

While screening ENU-mutagenized mice for hematopoietic alterations in peripheral blood, we detected an increase in CD44+CD62L- T cells in mice homozygous mice for a splice donor site mutation in Wdr41 (Fig. 3.22). The phenotype was named gogi. The gogi allele was also associated with increased DSS-induced colitis susceptibility with homozygotes losing 20% of their body weight by day 8 (Fig. 3.23). Wdr41gogi/gogi mice displayed elevated plasma levels of IL-12p40 under basal conditions without DSS treatment (Fig. 3.24). Stimulation of peritoneal macrophages harvested from Wdr41gogi/gogi mice with poly(I:C), R848, and CpG resulted in increased TNF production (Fig. 3.25). A second allele resulting in a premature stop codon in Wdr41 was also generated during ENU mutagenesis. Peritoneal macrophages homozygous for this allele, metallica

(mca), were hyperresponsive to endosomal TLR ligands (Fig. 3.26), and Wdr41mca/mca mice exhibited increased frequencies of activated T cells in the peripheral blood (Fig. 3.27).

54 FIGURES

Figure 3.1: Mapping of the Smcr8 mutation to the patriot phenotype. (A) Percentage of initial body weight on day 7 of DSS treatment plotted vs. genotype (REF, Smcr8+/+ (n=10), HET, Smcr8+/p (n=13), VAR, Smcr8p/p (n=3)). (B) Manhattan plot showing P values of association between the patriot phenotype and mutations identified in the patriot pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. Data are representative of one experiment. In A, data points represent individual mice and data are expressed as means ± s.d. P value for linkage of mutation in Smcr8 with the patriot phenotype is indicated.

Figure 3.2: Mapping of the Smcr8 mutation to the patriot2 phenotype. (A) Percentage of initial body weight on day 10 of DSS treatment plotted vs. genotype (REF, Smcr8+/+ (n=6), HET, Smcr8+/p2 (n=13), VAR, Smcr8p2/p2 (n=4)). (B) Manhattan plot showing P values of association between the patriot2 phenotype and mutations identified in the patriot2 pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. Data are representative of one experiment. In A, data points represent individual mice, and data are expressed as means ± s.d. P value for linkage of mutation in Smcr8 with the patriot phenotype is indicated.

Figure 3.3: Validation of DSS weight loss phenotype caused by loss of SMCR8. Weight loss analysis of Smcr8+/+ (n=6), Smcr8+/- (n=7), Smcr8+/I2T (n=7), Smcr8-/- (n=7), and Smcr8I2T/I2T (n=7) mice treated with 1.4% DSS for 7 days followed by 3 days of regular drinking water. Data are representative of three experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by two-way ANOVA with Dunnett’s multiple comparisons. (***P<0.001)

Figure 3.4: Smcr8 CRISPR mice are susceptible to DSS-induced colitis. (A) Disease activity index (DAI), (B) colon length (cm), and (C) representative H&E staining of Smcr8+/+ (n=5), Smcr8-/- (n=6), and Smcr8I2T/I2T (n=5) mice after 7 days of 1.4% DSS treatment. Data are representative of three experiments. In A and B, data points represent individual mice. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way ANOVA with Dunnett’s multiple comparisons. (***P<0.001)

Figure 3.5: Hematopoietic contributions to DSS phenotype. WT mice were lethally irradiated and +/+ -/- reconstituted with Smcr8 (n=4) or Smcr8 (n=5) bone marrow. Weight loss analysis (A) on the indicated days of treatment and DAI (B) after 7 days of treatment with 1.4% DSS. Data are representative of three experiments. In B, data points represent individual mice. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (**P<0.01, ***P<0.001).

Figure 3.6: Splenomegaly and lymphadenopathy in Smcr8 mutant mice. (A-F) Spleens and lymph nodes were harvested from Smcr8+/+ (n=5), Smcr8-/- (n=5), and Smcr8I2T/I2T (n=5) mice at 12 months of age. (A) Representative image of spleen harvested from Smcr8-/- mice. (B) Spleen weights normalized to body weight (bw). (C) Representative image and (D) total pooled cervical lymph node (LN) cell number harvested from two largest lymph nodes for each mouse. Data are representative of three experiments. In B and D, data points represent individual mice. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05, **P<0.01).

Figure 3.7: Equal expansion of major cell populations in spleens of Smcr8 CRISPR mice. Spleens were harvested from Smcr8+/+ (n=5), Smcr8-/- (n=5), and Smcr8I2T/I2T (n=5) mice at 12 months of age. Percent of total spleen cells for indicated cell populations. Data are expressed as means ± s.d. and significance was determined by one-way ANOVA with Dunnett’s multiple comparisons.

60 61

Figure 3.8: Increased activation of T cells in Smcr8 CRISPR mice. Peripheral blood (A,B) and spleens (C,D) were harvested from Smcr8+/+ (n=5), Smcr8-/- (n=5), and Smcr8I2T/I2T (n=5) mice at 12 months of age. Frequency of CD44+CD62L- CD4+ (A,B) and CD8+ (C,D) T cells is displayed. Data points represent individual mice. Data are representative of three independent experiments. Data are expressed as means ± s.d. and significance was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05, **P<0.01, ****P<0.0001).

Figure 3.9: Elevated IL-12p40 levels in serum of Smcr8 CRISPR mice. (A) 23plex cytokine analysis of plasma harvested from Smcr8+/+ (n=10), Smcr8-/- (n=10), and Smcr8I2T/I2T (n=10) mice at 6 months of age. (B) Plasma levels of IL-12p40 detected in Smcr8+/+ (n=6), Smcr8-/- (n=6), and Smcr8I2T/I2T (n=6) mice at 9 months of age. In B, data points represent individual mice. Data are representative of one or three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05, **P<0.01, ***P<0.001).

Figure 3.10: Screening for TNF responses to TLR9 stimulation. (A) Peritoneal macrophages harvested from pedigrees R1418 (REF, Smcr8+/+ (n=8), HET, Smcr8+/p (n=15), VAR, Smcr8p/p (n=2)) and R3960 (REF, Smcr8+/+ (n=11), HET, Smcr8+/p2 (n=18), VAR, Smcr8p2/p2 (n=5)) were stimulated with 25 ug/ml of CpG. TNF in the culture medium as measured by ELISA 4 h later. (B) Manhattan plot showing P values of association between the phenotype of elevated TNF production in response to CpG and mutations identified in the two pedigrees in (A) calculated using a recessive model of inheritance. The −log10 P values are plotted versus the chromosomal positions of mutations. Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. The P value for linkage of Smcr8 mutations with the elevated TNF production is indicated.

Figure 3.11: Analysis of TLR signaling in Smcr8-/- and Smcr8I2T/I2T peritoneal macrophages. ELISA analysis of TNF secretion by peritoneal macrophages (n=5 or 6 mice per genotype) stimulated for 6 h with indicated concentrations of endosomal TLR ligands poly(I:C), R848, and CpG (A) or surface TLR ligands Pam3CSK4 and LPS (B). Data points represent individual mice. Data are three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way or two-way ANOVA with Dunnett’s multiple comparisons. (*P<0.05, **P<0.01, ****P<0.0001).

Figure 3.12: Analysis of TLR signaling in Smcr8-/- and Smcr8I2T/I2T BMDCs. ELISA analysis of IL-6 secretion by BMDCs (n=4 mice per genotype) stimulated for 6 h with indicated concentrations of poly(I:C), R848, and CpG (A) or Pam3CSK4 and LPS (B). Data points represent individual mice. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (*P<0.05).

Figure 3.13: Removal of TLR 3, 7, and 9 signaling prevents enlargement of peripheral immune organs. Spleen weight (A) and cervical lymph node cell counts (B) from Smcr8+/- (n=5), Smcr8-/- (n=4), Smcr8+/-;Tlr3/7/9-/- (n=5) and Smcr8-/-Tlr3/7/9-/- (n=5) mice at 9 months of age. Data points represent individual mice. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences was determined by one-way ANOVA with Dunnett’s multiple comparisons (***P<0.001. ****P<0.0001).

Figure 3.14: Removal of TLR 3, 7, and 9 signaling normalizes IL-12(p40) serum levels. Data points represent individual mice. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05).

Figure 3.15: Removal of TLR 3, 7, and 9 signaling rescues spontaneous T cell activation. Data points represent individual mice. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05).

Figure 3.16: Removal of TLR 3, 7, and 9 signaling fails to rescue DSS susceptibility. (J) WT mice were lethally irradiated and reconstituted with bone marrow of the indicated donor genotype. Weight loss analysis of recipient mice treated with 1.4% DSS (n=5 for all groups). Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences was determined by one-way ANOVA with Dunnett’s multiple comparisons (*P<0.05).

Figure 3.17: Increased LysoTracker positive staining in Smcr8-/- BMDMs. (A) Representative images of increased LysoTracker positive vesicles in Smcr8-/- macrophages. Scale bars, 5 μm. (B) Quantification of LysoTracker positive area in Smcr8-/- cells. Each data point represents the average area of LysoTracker positive staining for independently derived cell lines (n=4). A minimum of 150 cells for each cell line were utilized for quantification. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (**P<0.01).

Figure 3.18: Impaired acidification of BioParticles by Smcr8-/- BMDMs. (A,B) Mean fluorescence intensity (MFI) of BMDMs (n=3 mice per genotype) incubated for indicated times with pHrodo zymosan (A) or pHrodo E. coli BioParticles (B). Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (*P<0.05, **P<0.01).

Figure 3.19: Normal uptake of zymosan BioParticles. Mean fluorescence intensity (MFI) of BMDMs (n=3 mice per genotype) incubated for the indicated times with Alexa Fluor 488 zymosan. Data are representative of three independent experiments. Data are expressed as means ± s.d.

Figure 3.20: Exaggerated TNF production by Smcr8-/- BMDMs in response to phagocytosed cargo. ELISA analysis of TNF secretion by BMDMs (n=3 mice per genotype) stimulated for 6 hours with 10 μg/ml zymosan or E. coli BioParticles. Data points represent individual mice. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (**P<0.01).

Figure 3.21: Increased phagosomal ROS production by Smcr8-/- BMDMs. MFI of BMDMs (n=3 mice per genotype) incubated for indicated times with OxyBURST zymosan. Data are representative of three independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (*P<0.05, **P<0.01).

Figure 3.22: Wdr41gogi/gogi mice exhibit increased CD44+ T cells in peripheral blood. Frequency of CD44+CD62L- CD4+ (A) and CD8+ (B) T cells in peripheral blood from C57BL/6J mice (n=7) and G3 mice from a single pedigree carrying the Wdr41gogi allele (Wdr41+/+ (n=17), Wdr41+/gogi (n=25), and Wdr41gogi/gogi (n=9)). Data points represent individual mice. Data are representative of two independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way ANOVA with Dunnett’s multiple comparisons (****P<0.0001).

Figure 3.23: Wdr41gogi/gogi mice exhibit increased DSS susceptibility. Weight loss analysis of C57BL/6J (n=10), Wdr41+/+ (n=7), Wdr41+/gogi (n=14), and Wdr41gogi/gogi (n=6) treated with 1.4% DSS. Data are representative of two independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by two-way ANOVA with Dunnett’s multiple comparisons (****P<0.0001).

Figure 3.24: Elevated IL-12p40 plasma levels in Wdr41gogi/gogi mice. ELISA analysis of plasma levels of IL-12p40 detected in untreated 6 month old gogi mice (Wdr41+/+ (n=5), Wdr41+/gogi (n=5), and Wdr41gogi/gogi (n=5)). Data points represent individual mice. Data are representative of two independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by one-way ANOVA with Dunnett’s multiple comparisons (****P<0.0001).

Figure 3.25: Wdr41gogi/gogi peritoneal macrophages produce elevated TNF in response to endosomal TLR ligands. ELISA analysis of TNF secretion by peritoneal macrophages (n=4 mice per genotype) stimulated for 6 hours with indicated concentrations of poly(I:C), R848, and CpG. Data are representative of two independent experiments. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by unpaired Student’s t-test (*P<0.05, **P<0.01, ****P<0.0001).

Figure 3.26: Wdr41mca/mca peritoneal macrophages produce elevated TNF in response to endosomal TLR ligands. ELISA analysis of TNF secretion by peritoneal macrophages (Wdr41+/+ (n=6), Wdr41+/mca (n=3), and Wdr41mca/mca (n=4)) stimulated for 6 hours with indicated concentrations of poly(I:C), R848, and CpG. Experiments were performed one time. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by two-way ANOVA with Dunnett’s multiple comparisons (*P<0.05, ***P<0.001, ****P<0.0001).

Figure 3.27: Increased percentages of CD44+ T cells in Wdr41mca/mca mice. Frequency of CD44+CD62L- CD4+ (A) and CD8+ (B) T cells in peripheral blood from Wdr41+/+ (n=4), Wdr41+/mca (n=3), and Wdr41mca/mca (n=5) mice. Data points represent individual mice. Experiments were performed one time. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by two-way ANOVA (A) or one-way ANOVA with Dunnett’s multiple comparisons (B,C) (*P<0.05, **P<0.01).

81 DISCUSSION

Here, we have established that exaggerated TLR signaling is elicited in myeloid cells from

Smcr8-/- and Smcr8I2T/I2T mice by ligands normally encountered in endolysosomes, and that such signaling in cell type(s) yet to be identified drives the spontaneous inflammation observed in

Smcr8-/- mice. Secondary lymphoid organ enlargement, activation of T cells, and elevated IL-

12p40 serum levels were rescued by ablation of endosomal TLR signaling. We note that the source of ligand (i.e. microbial vs. sterile) responsible for the spontaneous inflammation in Smcr8-\/- mice remains unknown, as does the identity of the cell type that primarily drives the immune dysregulation. Cell specific targeting of Smcr8 in myeloid and lymphoid lineages will aid in determining the roles each cell population plays in the observed phenotypes. In addition, the contribution of individual TLRs to the inflammatory phenotypes warrants further investigation.

While the requirement of endolysosomal acidification for TLR3, TLR7, and TLR9 activation is well known (137-140), our data instead show increased TLR signaling despite impaired acidification, at least in phagolysosomes. We conclude that the level of acidification of phagolysosomes in Smcr8-/- cells is sufficient for TLR activation, but perhaps inadequate for optimal function of the hydrolases responsible for degradation of phagocytosed TLR ligands. This idea is supported by the observation of lysosome accumulation in Smcr8-/- macrophages, because lysosome turnover requires degradation and removal of degraded cargo via diffusion or specific transport (132, 141). Our data thus suggest a critical role for SMCR8, and likely the SWC complex itself, that occurs after late endosome or phagosome fusion with the lysosomal compartment in the membrane trafficking cycle. One possibility suggested by the elevated phagosomal ROS levels observed in Smcr8-/- macrophages is that the SWC complex negatively regulates recruitment of

82 NOX2 to phagolysosomes and endolysosomes; this may be achieved through modulation of Rab activity, which has been shown to regulate NOX2 recruitment to phagosomes in dendritic cells

(142) As a consequence of impaired degradation of lysosomal cargo including TLR3, TLR7,

TLR9, and/or their ligands, prolonged ligand-receptor interactions result in correspondingly prolonged signaling.

The proposed model for negative regulation of endosomal TLR signaling by the SWC complex provides a plausible mechanistic explanation for the autoimmune phenotypes observed in C9orf72-/- and Smcr8-/- mice. Effective and timely degradation of phagocytosed dead cells by macrophages precludes activation by nucleic acids released from dying cells and presentation of self peptides to T cells. However, in Smcr8-/- macrophages degradation of such cargo may be both delayed and impaired, resulting in preservation of nucleic acids and peptides presented to T cells, and subsequent development of cellular and humoral immunity against self DNA and autoantigens.

In humans with ALS/FTD caused by C9ORF72 mutations, it may be that such autoantigens include nervous system proteins important for motor neuron function.

Loss of SMCR8 function in hematopoietic cells also rendered mice susceptible to chemically-induced colitis. Removal of signaling through TLRs 3, 7, and 9 in hematopoietic cells did not rescue colitis susceptibility observed in Smcr8-/- mice and perhaps worsened disease.

Moreover, mice reconstituted with Smcr8+/-;Tlr3/7/9-/- bone marrow were more susceptible compared with mice reconstituted with Smcr8+/- bone marrow. While complete ablation of endosomal TLR responses appears to increase sensitivity to DSS, the possibility remains that hyperactivation of these receptors also increases susceptibility and is contributory to the DSS phenotype observed in Smcr8 -/- mice. With that said, Smcr8-/-Tlr3/7/9-/- chimeric mice lost more

83 body weight than Tlr3/7/9-/- chimeric mice after DSS challenge indicating the presence of defects independent of endosomal TLR signaling. DSS insult physically compromises the intestinal epithelial barrier allowing for the entry of luminal microbes that must be cleared for restoration of intestinal homeostasis. We hypothesize that challenge of Smcr8 CRISPR mice with DSS may result in the hyperactivation of pathogenic immune signaling responses associated with endosomes and phagosomes after internalization of these microbes. Moreover, Smcr8-/- macrophages exhibited increased production of phagosomal ROS which has been shown to contribute to disease severity in DSS colitis (48, 143).

Spontaneous inflammation has been reported in mice with C9ORF72 or SMCR8 deficiency, but loss of function of the third SWC complex member, WDR41, has not been examined. We showed that mice homozygous for damaging mutations in Wdr41 phenocopied

Smcr8-/- mice in hyperactivation of T cells and elevated serum IL-12p40 levels under basal conditions, as well as colitis susceptibility after DSS challenge. Taken together, all members of the tripartite complex appear to be necessary for immune regulation.

MATERIALS AND METHODS

Mice

DSS-colitis induction was performed as described in Chapter 1. Disease activity index was scored as described in Chapter 2.

Smcr8-/-; Tlr3-/-; Tlr7-/-;Tlr9-/- mice were generated by intercrossing single knockout mouse strains. The Tlr3-/- (144), Tlr7-/- (145) and Tlr9-/- (146) mutant mice have been previously

84 described. All mice were housed in the University of Texas Southwestern vivarium and all procedures were performed in accordance with institutionally approved protocols.

Generation of the Smcr8I2T/I2T and Smcr8-/- mouse strains using the CRISPR/Cas9 system

To generate Smcr8I2T/I2T and Smcr8-/- mouse strains, embryos were harvested as describe in

Chapter 1 and in vitro–transcribed Cas9 mRNA (50 ng/μl) and Smcr8 small base-pairing guide

RNA (50 ng/μl; 5’- GGGATCTTCGTCTTCTGACG -3’ for Smcr8-/- and 5’-

GATCAGCGCCCCTGATGTGG -3’ for Smcr8I2T/I2T mice) were injected into the cytoplasm or pronucleus of the embryos. The injected embryos were cultured and transferred as described in

Chapter 1.

Smcr8-/- mice contain a 26-bp deletion of chromosome 11 (60,778,048-60,778,073) in exon

1 resulting in a predicted frameshifted protein product beginning after amino acid 7 of the protein and terminating after the inclusion of 6 aberrant amino acids. Smcr8I2T/I2T mice contain 3 separate point mutations on chromosome 11: 60,778,032 (T->C), 60,778,048 (G->A), and 60,778,078 (G-

>A). The first mutation results in an isoleucine to threonine change in amino acid 2 of the protein while the latter two are silent and result in no coding changes.

Hematopoietic chimeras

The indicated recipient mice were irradiated with 2 doses of 6.5 Gy spaced 4 h apart. Bone marrow cells from the tibiae and femurs of donors were intravenously injected into recipients.

Mice were maintained for two weeks on water containing trimethroprim-sulfamethoxazole antibiotics and experiments were performed 8-10 weeks after reconstitution.

Peritoneal macrophage, bone marrow-derived dendritic cell and bone marrow-derived macrophage cultures

Peritoneal macrophages were isolated as previously described (147). Mice were injected intraperitoneally with Brewer’s modified thioglycolate (3% wt/vol; BD Biosciences). Cells were collected by peritoneal lavage with 5 ml of PBS on day 5 after injection, plated onto 96-well plates at a density of 1 x105 cells/well overnight, and stimulated with the indicated ligands. Cells were subjected to MTT assay (Sigma-Aldrich) for normalization.

BMDCs and BMDMs were generated by standard protocols. Briefly, bone marrow was isolated from femurs and tibias of mice. Bone marrow cells were cultured in recombinant GM-

CSF (33 ng/ml) or M-CSF (20 ng/ml) for 6 days. BMDCs were plated onto 96-well plates at a density of 1x105 cells/well and stimulated with the indicated ligands. BMDMs were plated onto

24-well plates at a density of 5x105 cells/well for microbial BioParticle uptake and response experiments.

Antibodies and reagents

The following antibodies were used: B220, CD3, CD4, CD11c, CD19, CD44, Ly-6G (BD

Biosciences), CD8, CD11b, CD86, Ly-6C, (BioLegend), and CD62L, F4/80 (Tonbo Biosciences).

The following reagents were used: Pam3CSK4 and R848 (InvivoGen), poly(I:C) (GE

Healthcare), LPS and MALP-2 (Enzo Life Sciences), CpG-ODN 1668 (Sigma-Aldrich), mouse

IL-6, IL-12p40, and TNF-α Ready-SET-Go kits (eBioscience).

86 Flow cytometry

Spleen and blood cells were isolated and incubated as previously described (2). Data were acquired on a LSR II Fortessa cell analyzer (BD Biosciences) and analyzed with FlowJo software

(FlowJo, LLC). Cell sorting was performed on a FACSAria II cell sorter (BD Biosciences).

Microscopy

BMDMs were plated onto 8-well glass bottom μ-slides (ibidi) at a density of 2x105 cells/well. Cells were incubated with 300 nM LysoTracker for 15 minutes for visualization of lysosomes. Images were acquired using 63X objectives of a Zeiss LSM880 inverted confocal microscope.

Histology

Freshly isolated distal colons were fixed in formalin and embedded in paraffin.

Hematoxylin-eosin staining was conducted using a standard protocol by the UT Southwestern

Histology core.

Statistical Analysis

Statistical analysis was performed as detailed in Chapter 2.

CHAPTER FOUR

TVP23B regulates secretory granule formation in Paneth and goblet cells

PREFACE

The colitis phenotype caused by TVP23B deficiency has only begun to be analyzed. Thus, this chapter is less developed compared with others. Nevertheless, the phenotype is quite strong, consistent with the subject of this thesis, and the amount of data generated to date is sufficient for a standalone chapter.

INTRODUCTION

The gastrointestinal tract is lined by an intestinal epithelium that separates luminal microbes from the underlying immune cells residing in the lamina propria. This physical separation is required for the maintenance of immunologic quiescence within the intestine. The intestinal epithelial monolayer is primarily composed of enterocytes interlocked through tight junctions that present a significant physical barrier to microbial invasion (148). Enterocytes also play an active role in microbial defense through their secretion of antimicrobial proteins (AMPs), transportation of secretory immunoglobulin A (IgA) from the basolateral to apical surface, and participation in innate and adaptive immunity (148).

The intestinal epithelium is composed of other specialized cell types, such as Paneth and goblet cells, that are also critical for immune defense. Positioned at the bottom of small intestine crypts, Paneth cells are pyramidal-shaped cells with basally oriented nuclei and apically oriented

88 secretory granules. These large dense core granules (DCGs) occupy most of the cytoplasm and are composed of a number of antimicrobial proteins (AMPs) including α-defensins, C-type lectins, and lysozyme (148). AMPs protect against enteric pathogens, shape microbiota composition, and limit bacterial-epithelial cell contact (148). The importance of Paneth cells in intestinal homeostasis has been highlighted by association of NOD2 and ATG16L1 with IBD (36, 149).

Loss of function of NOD2 or ATG16L1 leads to Paneth cell aberrations including reduced defensin production and abnormalities in granule formation (36, 149).

Goblet cells support the epithelial barrier by producing mucus that forms a thick gel-like layer covering the luminal surface of the epithelium (148). This mucus limits physical contact of microbiota with the epithelium and protects against invasion (148). The critical role for the mucus layer in maintenance of intestinal homeostasis has been demonstrated by studies with Muc2-/- mice and in patients with IBD. MUC2 deficiency in mice results in increased bacterial penetration, increased susceptibility to DSS-induced colitis, and development of spontaneous colitis (57). IBD patients show reduced goblet cell number and mucus production (150).

Proteins destined for secretion are transported from the ER through the Golgi and then to the cell surface. Unlike constitutively secreted proteins, antimicrobial proteins and mucins are stored in secretory granules positioned at the apical side of the cell where they are poised for exocytosis into the lumen for regulation of resident microbiota. While screening, a mutation in trans-Golgi network vesicle protein homolog B (Tvp23b) mapped to a DSS susceptibility phenotype and causation was validated by CRISPR/Cas9 targeting. Examination of the intestinal epithelium of Tvp23b-/- mice revealed aberrations in both Paneth and goblet cell secretory granules.

Interaction with YIPF6, a previously published DSS mutant (92), was established. Together, these

89 Golgi-localizing proteins appear to be essential for regulation of the secretory pathway in the intestinal epithelium, but dispensable for secretion in other cell types.

RESULTS

Mapping and validation of Tvp23b

While screening pedigree R4840 for susceptibility to DSS-induced colitis, nine mice were strongly affected, losing over 20% of their initial body weight. This phenotype, named Chipotle

(Cpt), was mapped to a nonsense mutation in Tvp23b using an additive model of inheritance (Fig.

4.1A). Tvp23bCpt/Cpt mice lost on average 25% of their initial body weight while Tvp23b+/+ and

Tvp23b+/Cpt mice lost 13% and 19%, respectively (Fig. 4.1B). CRISPR/Cas9 targeting of Tvp23b

(Tvp23b-/-) generated a 1 bp deletion in exon 2 which resulted in a frame-shifted protein product beginning at amino acid 19. Treatment of Tvp23b-/- mice with DSS resulted in significant body weight loss compared with Tvp23b+/+ mice, confirming that Tvp23b is required for protection from

DSS-induced colitis (Fig. 4.2).

Loss of Tvp23b causes intestinal epithelium aberrations

Tvp23b is broadly expressed in the intestinal epithelium (151). To determine if loss of

TVP23B leads to epithelial dysfunction, untreated Tvp23b-/- small intestine and colon tissues were stained with hematoxylin and eosin (Fig. 4.3). Within the small intestine, mucin-containing vacuoles were smaller and the distinct granules normally present in Paneth cells were noticeably absent. Mucin vacuoles were also diminutive in goblet cells of the colon. These phenotypes were

90 even more obvious in tissues stained with period-acid Schiff (PAS) and alcian blue (AB) (Fig.

4.4).

Transmission electron microscopy (TEM) of Tvp23b-/- colons and small intestines revealed ultrastructural aberrations in both Paneth and goblet cells granules (Fig. 4.5). Apical DCGs were less frequent, smaller, and more disorganized in Tvp23b-/ - Paneth cells. Granules were also smaller in Tvp23b-/ - goblet cells and had higher electron density variation.

The DCGs of Paneth cells are composed of many different AMPs including lysozyme and

REG3γ (148). Immunostaining revealed that the DCG abnormalities detected by TEM corresponded with near absence of these proteins in Tvp23b-/- Paneth cells (Fig. 4.6,7).

Immunoblotting of small intestine crypt lysates confirmed the loss of lysozyme (Fig. 4.8). Bands of equal intensity were detected when blotting for REG3γ (Fig. 4.9). Although this band is the correct molecular weight, a knockout control will be used in the future to confirm specificity.

Unlike lysozyme, REG3γ is also secreted by enterocytes of the epithelium (148). Secretion of this protein appears to be unaffected in Tvp23b-/- enterocytes based on immunostaining (Fig. 4.7), making it difficult to detect reduced levels in whole crypt lysates that include both Paneth cells and enterocytes.

TVP23B interacts with YIPF6 and localizes to the Golgi

The intestinal abnormalities observed in Tvp23b-/- mice were strikingly similar to the Klein-

Zschocher mutant reported in 2012 by the Beutler laboratory (92), indicating that these two molecules may operate within the same pathway. Moreover, interaction between TVP23B and

YIPF6, detected using yeast two hybrid, has been reported by the CCSB preliminary interactome

91 database (152). To determine if TVP23B and YIPF6 interact, 293T cells were transiently cotransfected with FLAG-YIPF6 and HA-TVP23B. Lysates were incubated with anti-FLAG- coupled beads and immunoblot analysis revealed co-immunoprecipitation of YIPF6 and TVP23B, confirming interaction between the two molecules (Fig. 4.9).

TVP23B strongly colocalizes with GOLGA4, a peripheral membrane protein associated with the cytoplasmic face of the trans-Golgi in HT-29-MTX-E12 colonic epithelial cells (Fig.

4.10). Golga4 was another gene mapped in the DSS screen (Fig. 4.11), and it is possible that

GOLGA4 functions in the same pathway as TVP23B and YIPF6.

TVP23B is dispensable for many secretory processes

TVP23B likely operates in both goblet and Paneth cells, but whether this protein carries out important functions in other cell types has not been determined. Immunization of Tvp23bcpt/cpt mice resulted in normal production of IgG antibodies (Fig. 4.12A). TNF secretion by Tvp23bcpt/cpt peritoneal macrophages in response to innate immune ligands was unchanged (Fig. 4.12B).

Tvp23bcpt/cpt mice exhibited normal insulin production after glucose challenge (Fig. 4.12C). No neurologic defects have been observed in Tvp23b-/- mice suggesting that secretion of neuropeptides is intact. Both insulin and neuropeptides are packaged in DCGs indicating that loss of TVP23B does not cause DCG aberrations in all cell types. Thus, TVP23B is dispensable for many secretory processes, and the defects detected thus far are specific to the intestinal epithelium.

92 FIGURES

Figure 4.1: Mapping of the Tvp23b mutation to the Chipotle phenotype. (A) Percentage of initial body weight on day 10 of DSS treatment plotted vs. genotype (REF, Tvp23b+/+ (n=8), HET, Tvp23b+/Cpt (n=14), VAR, Tvp23bCpt/Cpt (n=2)). (B) Manhattan plot showing P values of association between the Chipotle phenotype and mutations identified in the Chipotle pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. Data are representative of one experiment. In A, data points represent individual mice, and data are expressed as means ± s.d. P value for linkage of mutation in Tvp23b with the Chipotle phenotype is indicated.

Figure 4.2: CRISPR/Cas9 validation of Tvp23b as a DSS susceptibility gene. Weight loss analysis of Tvp23b+/+ (n=5), Tvp23+/- (n=18), and Tvp23b-/- (n=6) mice treated with 1.3% DSS for 7 days followed by 3 days of regular drinking water. Data are representative of one experiment. Data are expressed as means ± s.d. and the significance of differences between genotypes was determined by two-way ANOVA with Dunnett’s multiple comparisons. (****P<0.0001)

Figure 4.3: H&E staining of Tvp23b-/- intestines. Representative H&E staining of small intestines and colons harvested from Tvp23b+/+ (n=3) and Tvp23b-/- (n=3) mice without treatment with DSS. Scale bar, 50 μm.

Figure 4.4: PAS and AB staining of Tvp23b-/- intestines. Representative period acid-Schiff (PAS) staining of small intestines and alcian blue (AB) staining of colons harvested from Tvp23b+/+ (n=3) and Tvp23b-/- (n=3) mice without treatment with DSS. Scale bar, 50 μm.

Figure 4.5. Transmission electron microscopy (TEM) of Tvp23b-/- intestines. Representative electron micrographs of small intestine Paneth and colonic goblet cells of Tvp23b-/- (n=2) mice. Scale bar, 4 uM.

Figure 4.6. Reduced levels of LYZ in Paneth cells of Tvp23b-/- mice. Confocal images of Tvp23b-/- sections immunostained with anti-LYZ. Scale bar, 50 μm.

Figure 4.7: Reduced levels of REG3γ in Paneth cells of Tvp23b-/- mice. Confocal images of Tvp23b-/- sections immunostained with anti-REG3γ. Scale bar, 50 μm.

Figure 4.8: Assessment of LYZ and REG3γ levels in small intestine crypts. Small intestine crypt extracts were immunoblotted with anti-LYZ and anti-REG3γ antibodies. β-tubulin was used as a loading control.

100

Figure 4.9: TVP23B binds YIPF6. HEK293T cells were transfected with FLAG-tagged YIPF6 and/or HA-tagged TVP23B. Cell lysates were immunoprecipitated using anti-FLAG M2 agarose and immunoblotted with antibodies against HA and FLAG. GAPDH was used as a loading control. Reproduced with permission by Miguel San Miguel.

101

Figure 4.10: TVP23B localizes to the TGN. Immunostaining of HT-29-MTX-E12 cells using anti- GOLGA4 and anti-TVP23B antibodies. Scale bar, 10 μm.

102

Figure 4.11. Mapping of a Golga4 mutation to the deranged phenotype. (A) Percentage of initial body weight on day 10 of DSS treatment plotted vs. genotype (REF, Golga4+/+ (n=9), HET, Golga4+/drg (n=13), VAR, Golga4drg/drg (n=6)). (B) Manhattan plot showing P values of association between the Chipotle phenotype and mutations identified in the deranged pedigree calculated using a recessive model of inheritance. The −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis). Horizontal red and purple lines represent thresholds of P = 0.05 with or without Bonferroni correction, respectively. Data are representative of one experiment. In A, data points represent individual mice, and data are expressed as means ± s.d. P value for linkage of mutation in Golga4 with the deranged phenotype is indicated.

103

104 Figure 4.12: Tvp23b-/- mice exhibit normal production of antibodies, TNF, and insulin. (A) T cell- dependent β-gal-specific antibodies 14 days after immunization of REF (n=8), HET (n=16), VAR (n=2) mice with a recombinant SFV vector encoding the model antigen, β-gal (rSFV-βGal). (B) ELISA analysis of TNF production after stimulation of REF (n=10), HET (n=16), VAR (n=2) with 25 ng/ml R848. (C) Serum insulin levels for REF (n=8), HET (n=16), VAR (n=2) 30 min after intraperitoneal injection of glucose (1g/kg body weight). REF, Tvp23b+/+; HET, Tvp23b+/-; VAR, Tvp23b-/-. Data points represent individual mice.

105 DISCUSSION

In polarized epithelium, proteins destined for the apical/basolateral membrane, constitutive and regulatory secretion (i.e., DCG), and lysosomes are sorted in the trans-Golgi network (TGN)

(153). The formation of DCGs begins with the budding of immature secretory granules (ISGs) from the TGN followed by a series of steps to form mature granules, including removal of missorted non-granule proteins as well as continued luminal acidification and condensation of granule proteins (154).

The lipid composition of the membrane is critical for ISG budding. Cholesterol along with other lipids such as glycosphingolipid and sphingomyelin form lipid rafts from which ISGs bud

(154). Lipid rafts serve as docking stations for granule proteins that will be packaged in the DCG and the machinery that provides the driving force for ISG formation (154). In addition to cholesterol, the TGN is also composed of high levels of diacylglycerol and phosphatic acid.

Cholesterol, diacylglycerol, and phosphatic acid exhibit a conical structure that induces negative curvature in the membrane, facilitating budding (154).

The mechanisms by which specific proteins are sorted into DCGs are unclear, but it is likely a combination of “sorting by entry” and “sorting by retention” (155). In the sorting by entry model, granule proteins are selectively included in budding ISGs because they possess positive sorting signals that interact with a receptor in the budding membrane (155, 156). Entry of a secretory protein into the developing ISG is influenced by its aggregative properties and ability to associate with the budding membrane (156). The lumen of the TGN is slightly acidic and has a high calcium concentration relative to the earlier secretory compartments (156). Under these conditions, DCG proteins tend to self-associate because they contain a preponderance of acidic

106 amino acids, while constitutively secreted proteins remain soluble (156). The association of aggregated proteins with lipid rafts promotes inclusion in the ISG and may also contribute the driving force for membrane budding (154). In the sorting by retention model, a large fraction of secretory proteins is included in the budding ISG, and proteins are selectively withdrawn for targeting to constitutive or lysosomal vesicles (156).

The sorting of lysozyme by retention is the only process that has been studied mechanistically in Paneth cells (157, 158). After activation by commensal microbes, NOD2 recruits LRRK2, RIP2, and RAB2A to the surface of the DCG to direct retention of lysozyme in mature DCGs (157, 158). Loss of any of these molecules causes mistargeting of lysozyme to the lysosome where it is degraded. This retention mechanism is specific to lysozyme as procryptidin and Reg3γ trafficking to DCGs is unaltered. In Tvp23b-/- mice, there is a near absence of DCGs within Paneth cells, and both lysozyme and Reg3γ levels appear to be reduced; thus, loss of

TVP23B causes a more general defect in DCG maturation. It is possible that TVP23B prevents mistargeting of all granule proteins to the lysosome; however, due to its localization at the TGN, it is more likely that TVP23B coordinates events early on during DCG maturation related to initial packaging of granule proteins or budding of ISGs.

Similar to DCG proteins of the Paneth cell, gel-forming mucins of the goblet cells aggregate at low pH and in the presence of increasing calcium concentrations (153). Hydrogen and calcium ions prevent the hydration of these polyanionic molecules, facilitate protein-protein interactions between mucin molecules, and stabilize mucin entanglements (153). Moreover, acidic pH is required for the formation of disulfide bonds between mucin molecules. Tvp23b-/- mice contain mucin vacuoles, but they are smaller in size (Fig. 4) Ultrastructural evaluation by TEM

107 revealed that the granules that make up these vacuoles are also smaller and have varying electron density (Fig. 5). Thus, the mucus polymerization properties in these Tvp23b-/- mice are likely altered. When formulating a unifying mechanism to explain the Paneth and goblet cell phenotypes, it is reasonable to postulate that TVP23B is required for the maintenance of low pH and high calcium in the TGN. Alternatively, given its membrane-bound nature, TVP23B might also regulate the organization of lipid rafts or budding of ISGs from the TGN.

TVP23B regulates the packaging of proteins into secretory granules of Paneth and goblet cells, but it remains to be studied if constitutive secretion pathways are affected in these cell types or if TVP23B is required for regulatory secretion in other epithelial cells. Alkaline phosphatase

(AP) is a constitutively secreted enzyme in enterocytes that is important in host defense. AP detoxifies endotoxin by removing its phosphate esters, and loss of AP results in DSS susceptibility

(159, 160). Assessment of AP levels in Tvp23b-/- mice will help to delineate if constitutive secretion of AMPs is altered. In addition to Paneth cells, enteroendocrine cells are also identifiable by the presence of DCGs and the formation of secretory granules in this cell type should also be assessed.

If DCG formation is the only secretion process altered in Tvp23b-/- mice, then this model will represent a fantastic opportunity to study the molecules that regulate DCG maturation. This area has been relatively understudied because of a lack of cell lines that undergo regulatory secretion. The ability to culture differentiated intestinal epithelium in particular has only recently become possible (161). Moreover, it is well established that the proper formation of mucin and

AMP-containing granules is necessary for protection from IBD, and TVP23B should be considered as a susceptibility gene when exploring causes of IBD in humans.

108

MATERIALS AND METHODS

Mice

ENU mutagenesis and DSS-induced colitis induction were performed as previously described in Chapter 1. All mice were housed in the University of Texas Southwestern vivarium. the University of Texas Southwestern Medical Center and were performed in accordance with institutionally approved protocols.

Generation of the Tvp23b-/- mouse strain using the CRISPR/Cas9 system

To the Tvp23b-/- mouse strain, embryos were harvested as describe in Chapter 1 and in vitro–transcribed Cas9 mRNA (50 ng/μl) and Smcr8 small base-pairing guide RNA (50 ng/μl; 5’-

GGGATCTTCGTCTTCTGACG-3’) were injected into the cytoplasm or pronucleus of the embryos. The injected embryos were cultured and transferred as described in Chapter 1.

CRISPR/Cas9 targeting resulted in a 1 bp deletion at location 62,881,980 of chromosome

11. This is predicted to cause a frame-shifted protein product beginning after amino acid 19 of the protein, which is normally 205 amino acids in length, and terminating after the inclusion of 46 aberrant amino acids.

Small Intestine Crypt Isolation

Small intestines were isolated from mice and stool removed from the lumen. Intestines were incubated at 4°C with rotation for 30 minutes in PBS containing 2 mM EDTA. Intestines were gently agitated to remove villi and then moved to a new tube containing PBS with 2mM

109 EDTA. Intestines were rotated for an additional 15 min at 4°C and then vigorously agitated to release crypts. Crypts were separated from contaminating villi by filter through a 70 μm strainer.

Antibodies

The following antibodies were used in this study: β-tubulin (1:1000, Cell Signaling

Technologies 2146S), Lysozyme (1: 200 DAKO 3.2.1.17), and REG3γ (a generous gift from the laboratory of Lora Hooper (162))

Histology and Immunostaining

Freshly isolated colons and small intestines were swiss-rolled, fixed in formalin, and embedded in paraffin. H&E, PAS, and AB staining were conducted using a standard protocol by the UT Southwestern Histology core. For immunostaining, sections were deparaffinized and rehydrated through an ethanol gradient. Antigen retrieval was peformed by boiling of slides for 15 minutes in a citrate-buffered solution (DAKO). Immunostaining was then performed as described in Chapter 2.

Electron microscopy

Mice were exsanguinated with 0.9% saline followed by perfusion fixation with 4% paraformaldehyde, 1.5% glutaraldehyde, and 0.02% picric acid in 0.1 M cacodylate buffer and colons and small intestines were dissected and cut into concentric circles. Tissues were incubated overnight at 4°C and processed using a standard protocol by the UT Southwestern EM core.

Tissues mounted on grids were imaged using a JOEL 1400+ microscope.

110

Statistical Analysis

Statistical analysis was performed as detailed in Chapter 2.

111

CHAPTER FIVE

Conclusions and recommendations

Over the past few decades, GWAS and monogenic family studies have accomplished noteworthy advancements in the understanding of the genetics of IBD (6-15, 29). However, these studies have only accounted for a portion of disease heritability (20). The contributions from

GWAS in particular have slowed. Large meta-analyses which are surveyed thousands of patient samples continue to implicate new loci, but the effect size is constantly decreasing (Fig. 5.1) (163).

Even in the extreme situation that tens of thousands of loci are eventually identified, less than half of genetic variance is predicted to be accounted for (163). It is clear that continued expansion of

GWAS cohorts will not have the ability to explain unresolved IBD heritability.

Figure 5.1. Predicted genetic variance explained using logarithmic fit. Adapted from Franke et al. (163).

To determine the genetic elements required for intestinal homeostasis, we performed a forward genetic screen in mice to identify genes that are required for return of normal physiologic

112 balance in the intestine after DSS insult. The DSS screen was extremely productive, revealing many genes required for intestinal homeostasis that have not previously been implicated. After saturation of approximately 20% of the genome, 28 genes were mapped and verified, including 20 that have not previously been reported. Of the 58 genes pending validation, 41 score good or better using a machine learning model. Given the 68% precision for genes of this classification, 28 genes pending CRISPR/Cas9 validation are predicted to validate. When considering this number and the current level of genome saturation, as many as 250 genes may result in a DSS susceptibility phenotype when damaged.

With less than a quarter of the genome saturated so far, screening large numbers of G3 mice should be a priority moving forward in order to elucidate all elements of intestinal homeostasis. The current level of genome saturation required 5 years of screening and 50,000 G3 mice (approximately 200 mice per week). Over the last few years, production has at times reached

600 mice per week, and ideally this number should be sustained in the long term. This is certainly feasible for the DSS screen as it is not as laborious as other phenotypic screens performed by the

Beutler laboratory.

GWAS fails to detect rare variants, which are more abundant in the genome and can have larger effect size (17). As the cost of next generation sequencing continues to decline, large-scale exome and whole genome sequencing will identify rare variants missed by GWAS. Many of the genes that have been implicated in IBD have corresponding phenotypes in animal models (46-50).

Considering this overlap, genes identified by the DSS screen should be considered as candidates for human IBD. If a putative variant is located in a gene identified by the DSS screen, then this may increase the likelihood that it is causative for the disease.

113 After identification of genes required for intestinal homeostasis, a major goal will be to determine the functional mechanisms and biological pathways associated with these genes. Not only will this aid in developing a better understanding of the pathogenesis of IBD, but it could potentially provide avenues for personalized therapy in the future. For example, patients with

IL10R mutations with predominant immune defects can be cured with bone marrow transplantation

(164), while this would not be an appropriate therapy with patients with primary epithelial defects such as what is observed in patients with EPCAM mutations (165, 166).

This last point is particularly important when considering the number of genes and pathways identified by the DSS screen. Previous genetic studies have established the necessity for epithelial barrier function, canonical immune signaling, and autophagy in intestinal homeostasis maintenance (28, 29). The DSS screen has ascertained more non-redundant molecules that fall within these categories but has revealed the necessity for other pathways as well, including lipid and steroid metabolism, mitochondrial function, and vesicle trafficking pathways beyond autophagy. Patients with a primary defect in phytoceramide biosynthesis, which is observed in

Degs2-/- mice, may need a different therapy compared with those that have an aberrant interferon signature, which is observed in Lrba-/- mice. Our ability to identify genes when screening has so far outpaced our mechanistic evaluation of these genes. Understanding where and how these molecules function will be a key focal point of future studies. In this thesis, three molecules that regular vesicle trafficking were characterized in more detail.

In chapter two, MYO1D was demonstrated to be a lipid-binding motor protein required for protection from DSS-induced colitis. Mutations in either the motor or tail domain of the protein resulted in a phenotype. This class I myosin localized to the basolateral membrane in intestinal

114 enterocytes and loss of MYO1D in the intestinal epithelium accounted for DSS susceptibility.

Although the homologs of the protein have been implicated in the regulation of Wnt/β-catenin signaling (95), intestinal epithelial differentiation was normal in MYO1D-deficient mice. In vitro,

MYO1D has been shown to regulate basolateral/apical traffic and membrane tension (89, 104).

Loss of MYO1D in rats alters planar cell polarity in tracheal epithelium (107). Future work will explore whether these processes are altered in the colonic epithelium of MYO1D-deficient mice.

In chapter three, SMCR8 and WDR41 were established as important regulators of lysosomal function. These molecules form the SWC complex with C9ORF72 to regulate phagosome and endosomal trafficking. Loss of function in SMCR8, WDR41, and C9ORF72 caused aberrant TLR signaling and increased ROS production within these compartments in myeloid cells. In addition to colitis susceptibility, Smcr8-/- and Wdr41-/- mice exhibited spontaneous inflammation characterized by lymphoproliferation and elevated serum cytokines.

Hyperactivation of endosomal TLRs was causative for spontaneous inflammation; however, colitis susceptibility is likely caused by aberrant myeloid function. Mice with floxed sites flanking exon

1 of Smcr8 have been generated, and conditional targeting of different cell lineages will be performed to determine the relevant cell types that contribute to the whole-body phenotypes.

Moreover, Smcr8-/- mice will be rederived in a germ-free setting to determine if the driver of spontaneous inflammation is sterile or microbial.

Several phenotypes caused by loss of SWC function have been described at the cellular level. We established that loss of SMCR8 results in the accumulation of LysoTracker positive vesicles and impaired acidification of phagocytosed cargo, and others have established that increased lysotracker exocytosis occurs in this context. The molecular mechanism responsible for

115 these phenotypes has not been determined, and studies to identify interactors of the SWC complex in myeloid cells are underway to better understand how it functions at the molecular level. Another avenue for further molecular exploration relates to the I2T mutation that causes phenotype. This point mutant is found at normal levels in cells and interaction with C9ORF72 and WDR41 is not disrupted. Future work will determine if this point mutation causes mislocalization, reduces catalytic activity, or disrupts other interactors of the complex.

In chapter four, TVP23B was established to be important for Paneth and goblet cell function. Histological evaluation of Tvp23b-/- small intestine and colons revealed diminished numbers of Paneth cell and goblet cell granules, and ultrastructural analysis revealed disorganization and reduced size of granules in these cells. Aberrations in granule formation corresponded with diminished levels of LYZ and REG3γ in Paneth cells. Interaction with YIPF6, another protein identified by the DSS screen, was established. Future studies will be conducted to elucidate which secretory pathways are affected in Tvp23b-/- mice. Moreover, whether TVP23B functions to regulate “sorting by entry” or “sorting by retention” during secretory granule formation will be determined.

All three of these vesicle trafficking strains exhibit strong colitis phenotypes when challenged with DSS. Smcr8-/- and Tvp23b-/- mice exhibit immunologic and epithelial dysfunction under basal conditions. These phenotypes should encourage further exploration of MYO1D,

SMCR8, and TVP23B as candidate genes for human IBD.

While host genetics is likely the major contributor, other factors such as microbiota composition, epigenetics, and environmental factors also influence the heritability of IBD.

Microbial dysbiosis is a key element of IBD pathogenesis (167), and the composition of gut

116 microbiota is heavily influenced by maternal-offspring exchanges that occur in infancy (168). A functional methylome map has been produced for UC which indicates a role for epigenetic regulation of disease pathogenesis (169). Many environmental factors that confer risk are shared within families including diet and exposure to environmental pollutants (170). Overall, “curing”

IBD will require an understanding of the complex interplay between host genetics, epigenetics, microbiota, and environmental factors.

BIBLIOGRAPHY

1. Turer E, et al. (2017) Creatine maintains intestinal homeostasis and protects against colitis. Proceedings of the National Academy of Sciences of the United States of America 114(7):E1273-E1281. 2. Wang T, et al. (2015) Real-time resolution of point mutations that cause phenovariance in mice. Proceedings of the National Academy of Sciences of the United States of America 112(5):E440-449. 3. Russell RK & Satsangi J (2008) Does IBD run in families? Inflamm Bowel Dis 14 Suppl 2:S20-21. 4. Halfvarson J, et al. (2005) Anti-Saccharomyces cerevisiae antibodies in twins with inflammatory bowel disease. Gut 54(9):1237-1243. 5. Brant SR (2011) Update on the heritability of inflammatory bowel disease: the importance of twin studies. Inflamm Bowel Dis 17(1):1-5. 6. de Lange KM, et al. (2017) Genome-wide association study implicates immune activation of multiple integrin genes in inflammatory bowel disease. Nature genetics 49(2):256-261. 7. Liu JZ, et al. (2015) Association analyses identify 38 susceptibility loci for inflammatory bowel disease and highlight shared genetic risk across populations. Nature genetics 47(9):979-986. 8. Parkes M, et al. (2007) Sequence variants in the autophagy gene IRGM and multiple other replicating loci contribute to Crohn's disease susceptibility. Nature genetics 39(7):830-832. 9. Yamazaki K, et al. (2013) A genome-wide association study identifies 2 susceptibility Loci for Crohn's disease in a Japanese population. Gastroenterology 144(4):781-788. 10. Anderson CA, et al. (2011) Meta-analysis identifies 29 additional ulcerative colitis risk loci, increasing the number of confirmed associations to 47. Nature genetics 43(3):246- 252. 11. Kenny EE, et al. (2012) A Genome-Wide Scan of Ashkenazi Jewish Crohn's Disease Suggests Novel Susceptibility Loci. PLoS genetics 8(3):e1002559. 12. Julia A, et al. (2014) A genome-wide association study identifies a novel locus at 6q22.1 associated with ulcerative colitis. Hum Mol Genet 23(25):6927-6934. 13. Yang SK, et al. (2014) Genome-wide association study of Crohn's disease in Koreans revealed three new susceptibility loci and common attributes of genetic susceptibility across ethnic populations. Gut 63(1):80-87. 14. Ellinghaus D, et al. (2016) Analysis of five chronic inflammatory diseases identifies 27 new associations and highlights disease-specific patterns at shared loci. Nature genetics 48(5):510-518. 15. Jostins L, et al. (2012) Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease. Nature 491(7422):119-124. 16. Momozawa Y, et al. (2018) IBD risk loci are enriched in multigenic regulatory modules encompassing putative causative genes. Nature Communications 9(1):2427. 17. Yu C, et al. (2018) Low-frequency and rare variants may contribute to elucidate the genetics of major depressive disorder. Translational Psychiatry 8(1):70.

117 118 18. Tennessen JA, et al. (2012) Evolution and Functional Impact of Rare Coding Variation from Deep Sequencing of Human Exomes. Science (New York, N.Y.) 337(6090):64-69. 19. The International HapMap C (2010) Integrating common and rare genetic variation in diverse human populations. Nature 467(7311):52-58. 20. McClellan J & King MC (2010) Genetic heterogeneity in human disease. Cell 141(2):210-217. 21. Choi M, et al. (2009) Genetic diagnosis by whole exome capture and massively parallel DNA sequencing. Proceedings of the National Academy of Sciences of the United States of America 106(45):19096-19101. 22. Liu T-C & Stappenbeck TS (2016) GENETICS AND PATHOGENESIS OF INFLAMMATORY BOWEL DISEASE. Annual review of pathology 11:127-148. 23. Hugot JP, et al. (2001) Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn's disease. Nature 411(6837):599-603. 24. Ogura Y, et al. (2001) A frameshift mutation in NOD2 associated with susceptibility to Crohn's disease. Nature 411(6837):603-606. 25. Hampe J, et al. (2007) A genome-wide association scan of nonsynonymous SNPs identifies a susceptibility variant for Crohn disease in ATG16L1. Nature genetics 39(2):207-211. 26. Rioux JD, et al. (2007) Genome-wide association study identifies five novel susceptibility loci for Crohn's disease and implicates a role for autophagy in disease pathogenesis. Nature genetics 39(5):596-604. 27. Duerr RH, et al. (2006) A Genome-Wide Association Study Identifies IL23R as an Inflammatory Bowel Disease Gene. Science (New York, N.Y.) 314(5804):1461-1463. 28. Cho JH & Brant SR (2011) Recent Insights Into the Genetics of Inflammatory Bowel Disease. Gastroenterology 140(6):1704-1712. 29. Uhlig HH (2013) Monogenic diseases associated with intestinal inflammation: implications for the understanding of inflammatory bowel disease. Gut 62(12):1795- 1805. 30. Brandl K, Tomisato W, & Beutler B (2012) Inflammatory bowel disease and ADAM17 deletion. The New England journal of medicine 366(2):190; author reply 190. 31. Glocker EO, et al. (2009) Inflammatory bowel disease and mutations affecting the interleukin-10 receptor. The New England journal of medicine 361(21):2033-2045. 32. Blaydon DC, et al. (2011) Inflammatory skin and bowel disease linked to ADAM17 deletion. The New England journal of medicine 365(16):1502-1508. 33. Kuhn R, Lohler J, Rennick D, Rajewsky K, & Muller W (1993) Interleukin-10-deficient mice develop chronic enterocolitis. Cell 75(2):263-274. 34. Chalaris A, et al. (2010) Critical role of the disintegrin metalloprotease ADAM17 for intestinal inflammation and regeneration in mice. The Journal of Experimental Medicine 207(8):1617-1624. 35. Oliver PL & Davies KE (2012) New insights into behaviour using mouse ENU mutagenesis. Human Molecular Genetics 21(R1):R72-R81. 36. Cadwell K, et al. (2008) A unique role for autophagy and Atg16L1 in Paneth cells in murine and human intestine. Nature 456(7219):259-263.

119 37. Wang T, et al. (2018) Probability of phenotypically detectable protein damage by ENU- induced mutations in the Mutagenetix database. Nat Commun 9(1):441. 38. Moresco EMY, Li X, & Beutler B (2013) Going Forward with Genetics: Recent Technological Advances and Forward Genetics in Mice. The American Journal of Pathology 182(5):1462-1473. 39. Arnold CN, et al. (2012) ENU-induced phenovariance in mice: inferences from 587 mutations. BMC Research Notes 5:577-577. 40. Perse M & Cerar A (2012) Dextran sodium sulphate colitis mouse model: traps and tricks. Journal of biomedicine & biotechnology 2012:718617. 41. Boden EK & Lord JD (2017) CD4 T Cells in IBD: Crossing the Line? Digestive diseases and sciences 62(9):2208-2210. 42. Caprioli F, Marafini I, Facciotti F, Pallone F, & Monteleone G (2013) Targeting T-cells in chronic inflammatory bowel diseases. J Clin Cell Immunol 4(155):2. 43. Viazis N, et al. (2010) Course of inflammatory bowel disease in patients infected with human immunodeficiency virus. Inflammatory bowel diseases 16(3):507-511. 44. Halling ML, Kjeldsen J, Knudsen T, Nielsen J, & Hansen LK (2017) Patients with inflammatory bowel disease have increased risk of autoimmune and inflammatory diseases. World Journal of Gastroenterology 23(33):6137-6146. 45. Chassaing B, Aitken JD, Malleshappa M, & Vijay-Kumar M (2014) Dextran Sulfate Sodium (DSS)-Induced Colitis in Mice. Current protocols in immunology / edited by John E. Coligan ... [et al.] 104:Unit-15.25. 46. Zhang H, et al. (2017) The protection role of Atg16l1 in CD11c(+) dendritic cells in murine colitis. Immunobiology 222(7):831-841. 47. Cadwell K, et al. (2010) Virus-Plus-Susceptibility Gene Interaction Determines Crohn’s Disease Gene Atg16L1 Phenotypes in Intestine. Cell 141(7):1135-1145. 48. Li B, Alli R, Vogel P, & Geiger TL (2014) IL-10 modulates DSS-induced colitis through a macrophage-ROS-NO axis. Mucosal Immunol 7(4):869-878. 49. Cox JH, et al. (2012) Opposing consequences of IL-23 signaling mediated by innate and adaptive cells in chemically induced colitis in mice. Mucosal Immunol 5(1):99-109. 50. Watanabe T, et al. (2008) Muramyl dipeptide activation of nucleotide-binding oligomerization domain 2 protects mice from experimental colitis. J Clin Invest 118(2):545-559. 51. Adzhubei IA, et al. (2010) A method and server for predicting damaging missense mutations. Nature methods 7(4):248-249. 52. Lee S-y, et al. (2016) Arhgap17, a RhoGTPase activating protein, regulates mucosal and epithelial barrier function in the mouse colon. Scientific Reports 6:26923. 53. Bertolotti A, et al. (2001) Increased sensitivity to dextran sodium sulfate colitis in IRE1β- deficient mice. Journal of Clinical Investigation 107(5):585-593. 54. Tahara Y, et al. (2003) Hepatocyte growth factor facilitates colonic mucosal repair in experimental ulcerative colitis in rats. The Journal of pharmacology and experimental therapeutics 307(1):146-151. 55. McConnell BB, et al. (2011) Krüppel-Like Factor 5 Protects Against Dextran Sulfate Sodium-Induced Colonic Injury by Promoting Epithelial Repair in Mice. Gastroenterology 140(2):540-549.e542.

120 56. Deng F, et al. (2018) YAP triggers the Wnt/β-catenin signalling pathway and promotes enterocyte self-renewal, regeneration and tumorigenesis after DSS-induced injury. Cell Death & Disease 9(2):153. 57. Van der Sluis M, et al. (2006) Muc2-deficient mice spontaneously develop colitis, indicating that MUC2 is critical for colonic protection. Gastroenterology 131(1):117-129. 58. Alangari A, et al. (2012) LPS-responsive beige-like anchor (LRBA) gene mutation in a family with inflammatory bowel disease and combined immunodeficiency. The Journal of allergy and clinical immunology 130(2):481-488.e482. 59. Lopez-Herrera G, et al. (2012) Deleterious Mutations in LRBA Are Associated with a Syndrome of Immune Deficiency and Autoimmunity. American Journal of Human Genetics 90(6):986-1001. 60. Charbonnier L-M, et al. (2015) Regulatory T Cell Deficiency and Immune dysregulation, Polyendocrinopathy, Enteropathy, X-Linked-Like Disorder Due to Loss of Function Mutations in LRBA. The Journal of allergy and clinical immunology 135(1):217- 227.e219. 61. Lo B, et al. (2015) AUTOIMMUNE DISEASE. Patients with LRBA deficiency show CTLA4 loss and immune dysregulation responsive to abatacept therapy. Science 349(6246):436-440. 62. Kuehn HS, et al. (2013) Loss-of-function of the protein kinase C delta (PKCdelta) causes a B-cell lymphoproliferative syndrome in humans. Blood 121(16):3117-3125. 63. Salzer E, et al. (2013) B-cell deficiency and severe autoimmunity caused by deficiency of protein kinase C delta. Blood 121(16):3112-3116. 64. Belot A, et al. (2013) Protein kinase cdelta deficiency causes mendelian systemic lupus erythematosus with B cell-defective apoptosis and hyperproliferation. Arthritis and rheumatism 65(8):2161-2171. 65. Kim W-K, et al. (2011) Role of TNFR-Related 2 Mediated Immune Responses in Dextran Sulfate Sodium-Induced Inflammatory Bowel Disease. Molecules and Cells 31(2):99-104. 66. Hooks SB, Ragan SP, & Lynch KR (1998) Identification of a novel human phosphatidic acid phosphatase type 2 isoform. FEBS letters 427(2):188-192. 67. Wang X, Devaiah SP, Zhang W, & Welti R (2006) Signaling functions of phosphatidic acid. Progress in lipid research 45(3):250-278. 68. Ladiges WC, et al. (2005) Pancreatic beta-cell failure and diabetes in mice with a deletion mutation of the endoplasmic reticulum molecular chaperone gene P58IPK. Diabetes 54(4):1074-1081. 69. Matsumura Y, et al. (1999) Overt nephrogenic diabetes insipidus in mice lacking the CLC-K1 chloride channel. Nature genetics 21(1):95-98. 70. Jurczak MJ, et al. (2011) SGLT2 deletion improves glucose homeostasis and preserves pancreatic beta-cell function. Diabetes 60(3):890-898. 71. Xu H, et al. (2012) Impaired mucin synthesis and bicarbonate secretion in the colon of NHE8 knockout mice. American journal of physiology. Gastrointestinal and liver physiology 303(3):G335-343.

121 72. Schlegel N, et al. (2010) Desmoglein 2-mediated adhesion is required for intestinal epithelial barrier integrity. American journal of physiology. Gastrointestinal and liver physiology 298(5):G774-783. 73. Nava P, et al. (2007) Desmoglein-2: a novel regulator of apoptosis in the intestinal epithelium. Molecular biology of the cell 18(11):4565-4578. 74. Stamos JL & Weis WI (2013) The beta-catenin destruction complex. Cold Spring Harbor perspectives in biology 5(1):a007898. 75. Zamoyska R, et al. (2003) The influence of the src-family kinases, Lck and Fyn, on T cell differentiation, survival and activation. Immunological reviews 191:107-118. 76. Palacios EH & Weiss A (2004) Function of the Src-family kinases, Lck and Fyn, in T- cell development and activation. Oncogene 23(48):7990-8000. 77. Molina TJ, et al. (1992) Profound block in thymocyte development in mice lacking p56lck. Nature 357(6374):161-164. 78. Levin SD, Anderson SJ, Forbush KA, & Perlmutter RM (1993) A dominant-negative transgene defines a role for p56lck in thymopoiesis. The EMBO journal 12(4):1671-1680. 79. Turer E, et al. (2017) Creatine maintains intestinal homeostasis and protects against colitis. Proceedings of the National Academy of Sciences of the United States of America 114(7):E1273-e1281. 80. Christian BE & Spremulli LL (2009) Evidence for an active role of IF3mt in the initiation of translation in mammalian mitochondria. Biochemistry 48(15):3269-3278. 81. Anvret A, et al. (2010) Possible involvement of a mitochondrial translation initiation factor 3 variant causing decreased mRNA levels in Parkinson's disease. Parkinson's disease 2010:491751. 82. Behrouz B, et al. (2010) Mitochondrial translation initiation factor 3 polymorphism and Parkinson's disease. Neuroscience letters 486(3):228-230. 83. Abahuni N, et al. (2007) Mitochondrial translation initiation factor 3 gene polymorphism associated with Parkinson's disease. Neuroscience letters 414(2):126-129. 84. Bonnefond L, et al. (2005) Toward the full set of human mitochondrial aminoacyl-tRNA synthetases: characterization of AspRS and TyrRS. Biochemistry 44(12):4805-4816. 85. Kotelevtsev Y, et al. (1999) Hypertension in mice lacking 11beta-hydroxysteroid dehydrogenase type 2. J Clin Invest 103(5):683-689. 86. Pasternack SM, et al. (2008) G protein-coupled receptor P2Y5 and its ligand LPA are involved in maintenance of human hair growth. Nature genetics 40(3):329-334. 87. Lamb CA, et al. (2016) TBC1D14 regulates autophagy via the TRAPP complex and ATG9 traffic. The EMBO journal 35(3):281-301. 88. Hartman MA & Spudich JA (2012) The myosin superfamily at a glance. Journal of Cell Science 125(7):1627-1632. 89. McConnell RE & Tyska MJ (2010) Leveraging the membrane - cytoskeleton interface with myosin-1. Trends in cell biology 20(7):418-426. 90. Brandl K & Beutler B (2012) Creating diseases to understand what prevents them: genetic analysis of inflammation in the gastrointestinal tract. Current opinion in immunology 24(6):678-685.

122 91. Brandl K, et al. (2009) Enhanced sensitivity to DSS colitis caused by a hypomorphic Mbtps1 mutation disrupting the ATF6-driven unfolded protein response. Proceedings of the National Academy of Sciences of the United States of America 106(9):3300-3305. 92. Brandl K, et al. (2012) Yip1 domain family, member 6 (Yipf6) mutation induces spontaneous intestinal inflammation in mice. Proceedings of the National Academy of Sciences of the United States of America 109(31):12650-12655. 93. Bahler M, Kroschewski R, Stoffler HE, & Behrmann T (1994) Rat myr 4 defines a novel subclass of myosin I: identification, distribution, localization, and mapping of calmodulin-binding sites with differential calcium sensitivity. The Journal of cell biology 126(2):375-389. 94. Benesh AE, et al. (2010) Differential localization and dynamics of class I myosins in the enterocyte microvillus. Molecular biology of the cell 21(6):970-978. 95. Speder P, Adam G, & Noselli S (2006) Type ID unconventional myosin controls left- right asymmetry in Drosophila. Nature 440(7085):803-807. 96. Hozumi S, et al. (2006) An unconventional myosin in Drosophila reverses the default handedness in visceral organs. Nature 440(7085):798-802. 97. Fevr T, Robine S, Louvard D, & Huelsken J (2007) Wnt/β-Catenin Is Essential for Intestinal Homeostasis and Maintenance of Intestinal Stem Cells. Molecular and Cellular Biology 27(21):7551-7559. 98. Okumura R, et al. (2016) Lypd8 promotes the segregation of flagellated microbiota and colonic epithelia. Nature 532(7597):117-121. 99. Goettel JA, et al. (2011) KSR1 protects from interleukin-10 deficiency-induced colitis in mice by suppressing T lymphocyte interferon-γ production. Gastroenterology 140(1):265-274. 100. Kravtsov DV, et al. (2012) Myosin Ia is required for CFTR brush border membrane trafficking and ion transport in the mouse small intestine. Traffic (Copenhagen, Denmark) 13(8):1072-1082. 101. Tyska MJ & Mooseker MS (2004) A role for myosin-1A in the localization of a brush border disaccharidase. The Journal of cell biology 165(3):395-405. 102. McConnell RE, et al. (2009) The enterocyte microvillus is a vesicle-generating organelle. The Journal of cell biology 185(7):1285-1298. 103. Tyska MJ, et al. (2005) Myosin-1a Is Critical for Normal Brush Border Structure and Composition. Mol Biol Cell 16(5):2443-2457. 104. Huber LA, et al. (2000) Both calmodulin and the unconventional myosin Myr4 regulate membrane trafficking along the recycling pathway of MDCK cells. Traffic (Copenhagen, Denmark) 1(6):494-503. 105. Ubelmann F, et al. (2013) Enterocyte loss of polarity and gut wound healing rely upon the F-actin–severing function of villin. Proceedings of the National Academy of Sciences of the United States of America 110(15):E1380-E1389. 106. Ferrary E, et al. (1999) In Vivo, Villin Is Required for Ca(2+)-Dependent F-Actin Disruption in Intestinal Brush Borders. The Journal of cell biology 146(4):819-830. 107. Hegan PS, Ostertag E, Geurts AM, & Mooseker MS (2015) Myosin Id is required for planar cell polarity in ciliated tracheal and ependymal epithelial cells. Cytoskeleton (Hoboken, N.J.) 72(10):503-516.

123 108. Eichele DD & Kharbanda KK (2017) Dextran sodium sulfate colitis murine model: An indispensable tool for advancing our understanding of inflammatory bowel diseases pathogenesis. World Journal of Gastroenterology 23(33):6016-6029. 109. Kiesler P, Fuss IJ, & Strober W (2015) Experimental Models of Inflammatory Bowel Diseases. Cellular and molecular gastroenterology and hepatology 1(2):154-170. 110. Brandl K, et al. (2010) MyD88 signaling in nonhematopoietic cells protects mice against induced colitis by regulating specific EGF receptor ligands. Proceedings of the National Academy of Sciences of the United States of America 107(46):19967-19972. 111. Bertolotti A, et al. (2001) Increased sensitivity to dextran sodium sulfate colitis in IRE1beta-deficient mice. J Clin Invest 107(5):585-593. 112. Staley EM, Schoeb TR, & Lorenz RG (2009) Differential susceptibility of P-glycoprotein deficient mice to colitis induction by environmental insults. Inflammatory bowel diseases 15(5):684-696. 113. Kim JJ, Shajib MS, Manocha MM, & Khan WI (2012) Investigating Intestinal Inflammation in DSS-induced Model of IBD. Journal of Visualized Experiments : JoVE (60):3678. 114. O'Rourke JG, et al. (2016) C9orf72 is required for proper macrophage and microglial function in mice. Science 351(6279):1324-1329. 115. Sullivan PM, et al. (2016) The ALS/FTLD associated protein C9orf72 associates with SMCR8 and WDR41 to regulate the autophagy-lysosome pathway. Acta neuropathologica communications 4(1):51. 116. Yang M, et al. (2016) A C9ORF72/SMCR8-containing complex regulates ULK1 and plays a dual role in autophagy. Science Advances 2(9):e1601167. 117. Ugolino J, et al. (2016) Loss of C9orf72 Enhances Autophagic Activity via Deregulated mTOR and TFEB Signaling. PLoS genetics 12(11):e1006443. 118. Jung J, et al. (2017) Multiplex image-based autophagy RNAi screening identifies SMCR8 as ULK1 kinase activity and gene expression regulator. eLife 6. 119. Webster CP, et al. (2016) The C9orf72 protein interacts with Rab1a and the ULK1 complex to regulate initiation of autophagy. The EMBO journal 35(15):1656-1676. 120. Farg MA, et al. (2014) C9ORF72, implicated in amytrophic lateral sclerosis and frontotemporal dementia, regulates endosomal trafficking. Human Molecular Genetics 23(13):3579-3595. 121. Sellier C, et al. (2016) Loss of C9ORF72 impairs autophagy and synergizes with polyQ Ataxin‐2 to induce motor neuron dysfunction and cell death. The EMBO journal 35(12):1276-1297. 122. Zhang D, Iyer LM, He F, & Aravind L (2012) Discovery of Novel DENN Proteins: Implications for the Evolution of Eukaryotic Intracellular Membrane Structures and Human Disease. Frontiers in genetics 3:283. 123. Levine TP, Daniels RD, Gatta AT, Wong LH, & Hayes MJ (2013) The product of C9orf72, a gene strongly implicated in neurodegeneration, is structurally related to DENN Rab-GEFs. Bioinformatics 29(4):499-503. 124. DeJesus-Hernandez M, et al. (2011) Expanded GGGGCC hexanucleotide repeat in noncoding region of C9ORF72 causes chromosome 9p-linked FTD and ALS. Neuron 72(2):245-256.

124 125. Turner MR, Goldacre R, Ramagopalan S, Talbot K, & Goldacre MJ (2013) Autoimmune disease preceding amyotrophic lateral sclerosis: an epidemiologic study. Neurology 81(14):1222-1225. 126. Miller ZA, et al. (2016) Increased prevalence of autoimmune disease within C9 and FTD/MND cohorts: Completing the picture. Neurology(R) neuroimmunology & neuroinflammation 3(6):e301. 127. de Pasqua S, et al. (2017) Amyotrophic lateral sclerosis and myasthenia gravis: association or chance occurrence? Neurological sciences : official journal of the Italian Neurological Society and of the Italian Society of Clinical Neurophysiology 38(3):441- 444. 128. Gotaas HT, Skeie GO, & Gilhus NE (2016) Myasthenia gravis and amyotrophic lateral sclerosis: A pathogenic overlap. Neuromuscular disorders : NMD 26(6):337-341. 129. Burberry A, et al. (2016) Loss-of-function mutations in the C9ORF72 mouse ortholog cause fatal autoimmune disease(). Science translational medicine 8(347):347ra393- 347ra393. 130. Atanasio A, et al. (2016) C9orf72 ablation causes immune dysregulation characterized by leukocyte expansion, autoantibody production, and glomerulonephropathy in mice. Scientific Reports 6:23204. 131. Zhang Y, et al. (2018) The C9orf72-interacting protein Smcr8 is a negative regulator of autoimmunity and lysosomal exocytosis. Genes & development 32(13-14):929-943. 132. Huotari J & Helenius A (2011) Endosome maturation. The EMBO journal 30(17):3481- 3500. 133. Amick J, Roczniak-Ferguson A, & Ferguson SM (2016) C9orf72 binds SMCR8, localizes to lysosomes, and regulates mTORC1 signaling. Molecular biology of the cell 27(20):3040-3051. 134. Underhill DM, et al. (1999) The Toll-like receptor 2 is recruited to macrophage phagosomes and discriminates between pathogens. Nature 401(6755):811-815. 135. Pauwels AM, Trost M, Beyaert R, & Hoffmann E (2017) Patterns, Receptors, and Signals: Regulation of Phagosome Maturation. Trends in immunology 38(6):407-422. 136. Wang Y, et al. (2007) Lysosome-associated small Rab GTPase Rab7b negatively regulates TLR4 signaling in macrophages by promoting lysosomal degradation of TLR4. Blood 110(3):962-971. 137. Hacker H, et al. (1998) CpG-DNA-specific activation of antigen-presenting cells requires stress kinase activity and is preceded by non-specific endocytosis and endosomal maturation. The EMBO journal 17(21):6230-6240. 138. Park B, et al. (2008) Proteolytic cleavage in an endolysosomal compartment is required for activation of Toll-like receptor 9. Nature immunology 9(12):1407-1414. 139. Ewald SE, et al. (2008) The ectodomain of Toll-like receptor 9 is cleaved to generate a functional receptor. Nature 456(7222):658-662. 140. Ewald SE, et al. (2011) Nucleic acid recognition by Toll-like receptors is coupled to stepwise processing by cathepsins and asparagine endopeptidase. J Exp Med 208(4):643- 651.

125 141. Schulze H, Kolter T, & Sandhoff K (2009) Principles of lysosomal membrane degradation: Cellular topology and biochemistry of lysosomal lipid degradation. Biochimica et biophysica acta 1793(4):674-683. 142. Jancic C, et al. (2007) Rab27a regulates phagosomal pH and NADPH oxidase recruitment to dendritic cell phagosomes. Nature cell biology 9(4):367-378. 143. You Y, Fu JJ, Meng J, Huang GD, & Liu YH (2009) Effect of N-acetylcysteine on the murine model of colitis induced by dextran sodium sulfate through up-regulating PON1 activity. Digestive diseases and sciences 54(8):1643-1650. 144. Alexopoulou L, Holt AC, Medzhitov R, & Flavell RA (2001) Recognition of double- stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 413(6857):732-738. 145. Hemmi H, et al. (2002) Small anti-viral compounds activate immune cells via the TLR7 MyD88-dependent signaling pathway. Nature immunology 3(2):196-200. 146. Hemmi H, et al. (2000) A Toll-like receptor recognizes bacterial DNA. Nature 408(6813):740-745. 147. Sun L, et al. (2017) HCFC2 is needed for IRF1- and IRF2-dependent Tlr3 transcription and for survival during viral infections. The Journal of Experimental Medicine 214(11):3263-3277. 148. Hooper LV (2015) Epithelial cell contributions to intestinal immunity. Advances in immunology 126:129-172. 149. Kobayashi KS, et al. (2005) Nod2-dependent regulation of innate and adaptive immunity in the intestinal tract. Science 307(5710):731-734. 150. Boltin D, Perets TT, Vilkin A, & Niv Y (2013) Mucin function in inflammatory bowel disease: an update. Journal of clinical gastroenterology 47(2):106-111. 151. Haber AL, et al. (2017) A single-cell survey of the small intestinal epithelium. Nature 551(7680):333-339. 152. Rolland T, et al. (2014) A proteome-scale map of the human interactome network. Cell 159(5):1212-1226. 153. Perez-Vilar J (2008) Formation of mucin granules. The Golgi Apparatus, (Springer), pp 535-562. 154. Kim T, Gondre-Lewis MC, Arnaoutova I, & Loh YP (2006) Dense-core secretory granule biogenesis. Physiology (Bethesda, Md.) 21:124-133. 155. Dikeakos JD & Reudelhuber TL (2007) Sending proteins to dense core secretory granules: still a lot to sort out. The Journal of cell biology 177(2):191-196. 156. Bowman GR, Cowan AT, & Turkewitz AP (2009) Biogenesis of dense-core secretory granules. Trafficking Inside Cells, (Springer), pp 183-209. 157. Wang H, et al. (2017) Rip2 Is Required for Nod2-Mediated Lysozyme Sorting in Paneth Cells. Journal of immunology (Baltimore, Md. : 1950) 198(9):3729-3736. 158. Zhang Q, et al. (2015) Commensal bacteria direct selective cargo sorting to promote symbiosis. Nature immunology 16(9):918-926. 159. Tuin A, Huizinga-Van der Vlag A, van Loenen-Weemaes AM, Meijer DK, & Poelstra K (2006) On the role and fate of LPS-dephosphorylating activity in the rat liver. American journal of physiology. Gastrointestinal and liver physiology 290(2):G377-385.

126 160. Ramasamy S, et al. (2011) Intestinal Alkaline Phosphatase Has Beneficial Effects in Mouse Models of Chronic Colitis. Inflammatory bowel diseases 17(2):532-542. 161. Sato T, et al. (2009) Single Lgr5 stem cells build crypt-villus structures in vitro without a mesenchymal niche. Nature 459(7244):262-265. 162. Cash HL, Whitham CV, Behrendt CL, & Hooper LV (2006) Symbiotic bacteria direct expression of an intestinal bactericidal lectin. Science 313(5790):1126-1130. 163. Franke A, et al. (2010) Genome-wide meta-analysis increases to 71 the number of confirmed Crohn's disease susceptibility loci. Nature genetics 42:1118. 164. Kotlarz D, et al. (2012) Loss of interleukin-10 signaling and infantile inflammatory bowel disease: implications for diagnosis and therapy. Gastroenterology 143(2):347-355. 165. Kammermeier J, et al. (2014) Targeted gene panel sequencing in children with very early onset inflammatory bowel disease—evaluation and prospective analysis. Journal of Medical Genetics 51(11):748-755. 166. Uhlig HH & Muise AM (2017) Clinical Genomics in Inflammatory Bowel Disease. Trends in genetics : TIG 33(9):629-641. 167. Ni J, Wu GD, Albenberg L, & Tomov VT (2017) Gut microbiota and IBD: causation or correlation? Nature reviews. Gastroenterology & hepatology 14(10):573-584. 168. Mueller NT, Bakacs E, Combellick J, Grigoryan Z, & Dominguez-Bello MG (2015) The infant microbiome development: mom matters. Trends in molecular medicine 21(2):109- 117. 169. Häsler R, et al. (2012) A functional methylome map of ulcerative colitis. Genome Research 22(11):2130-2137. 170. Turpin W, Goethel A, Bedrani L, & Croitoru MK (2018) Determinants of IBD Heritability: Genes, Bugs, and More. Inflammatory bowel diseases 24(6):1133-1148.