X Chromosome Gene Dosage in Autoimmune Disease Susceptibility and B Cell Development

X chromosome gene dosage in autoimmune disease susceptibility and B cell development

A dissertation submitted to the Graduate School of the University of Cincinnati In partial fulfillment of the Requirements for the degree of

Doctor of Philosophy (Ph.D.)

in the Immunology Graduate Program of the College of Medicine 2016 by Ke (Coco) Liu

B.S., Nanchang University, 2010

Committee Chair: John B. Harley, M.D., Ph.D.

Abstract

Autoimmune disorders are the result of abnormal immune responses that cause tissue damage.

Collectively, these diseases affect about 8% of U.S. population. Most autoimmune diseases affect women more than men, with up to 90% of patients being females. No conclusive explanations for this female predominance has been demonstrated with clear mechanisms.

Our lab has discovered that men with two X chromosomes (Klinefelter's syndrome, 47,XXY) have the same risk as women for developing the autoimmune diseases, systemic lupus erythematosus (SLE). This is consistent with the possibility that the number of X chromosomes contributes to sex bias in SLE. We explored whether a third X chromosome in women with triple

X syndrome (47,XXX) would further increase the risk of developing SLE and other female- biased autoimmune diseases, primary Sjögren's syndrome (pSS), primary biliary cirrhosis (PBC) and rheumatoid arthritis (RA). We found that 47,XXX is more frequent in patients with SLE and pSS. The risk of SLE and pSS were increased approximately 25- and 41- fold, respectively, in

47,XXX women compared to men. We did not find enrichment of 47,XXX in PBC or RA.

Therefore, the number of X chromosomes appears to be a key factor impacting the risk of SLE and pSS development.

In order to identify what genes on the X chromosome are responsible for SLE gender bias, we took a candidate gene approach by selecting genes that escape X-inactivation in both humans and mice. We hypothesize that overexpression of these genes in women who have twice as many X chromosomes as men will contribute to lupus susceptibility bias. Among the candidate genes,

Ddx3x encodes an RNA helicase gene implicated in IFN-β signaling.

We generated conditional knockout mice with Ddx3x deficient in hematopoietic cells (Ddx3x-

Vav1). Unexpectedly, Ddx3x expression level was not different among female WT, female heterozygous and male WT mice in relevant immune tissues. However, we discovered a B cell deficiency in the lymphoid tissues of male hemizygous knockout mice (Vav1ddx3x) relative to ddx3x sufficient control male mice. Reduction of Ddx3x expression in hematopoietic cells was associated with reduced numbers of early pro-B, late pre-B, and immature B cells in the bone marrow. In the spleen, the numbers of transitional, follicular, and germinal center B cells, as well as plasmablasts, were decreased. Mixed bone-marrow chimeric mice demonstrated that these B- cell deficit phenotypes were B cell intrinsic. Moreover, Vav1ddx3x B cells were found to proliferate less at the late pro-B cell and early pre-B cell stages, possibly contributing to the generalized phenotype of low B-cell numbers. Surprisingly, despite the paucity of mature B cells, the levels of immunoglobulins in the serum of Vav1ddx3x mice were increased compared to WT mice. Preliminary evidence suggests that marginal zone B cells in the Vav1ddx3x mice may undergo altered class-switching and have an enhanced ability to produce immunoglobulins. Our data reveal that Ddx3x deficiency affects B-cell development resulting in a widespread loss of B cells and a paradoxical increase in serum immunoglobulin levels. These are the first data to link

Ddx3x to B-cell development and function.

iii iv

Acknowledgement

I would like to thank my mentor, Dr. John Harley. Firstly, for providing such a diverse research environment he created in the Center for autoimmune genetics and etiology that I was exposed to not only immunology but also a variety of other fields. Secondly, I was lucky to get to pick a thesis project that I was, am and will still be interested in, the female predominance phenomenon in autoimmune diseases. Lastly, thanks for always encouraging and pushing me to be more independent, to make my own choices, to try to make mistakes and learn from it as I will eventually go on the road by myself. It is an interesting journey to work with you.

Next, I would extend my gratitude to Dr. Steve Waggoner, who co-mentored me. It is not a small task to take another graduate student. I learned so much, from how to design an experiment, interpret the results, to how to present and sell my work better. Indeed, you have trained me how to think critically, systemically and scientifically. You are easily accessible, share your own stories as a scientist, and provide us lots of advice as how to thrive as a graduate student. It has been delight learning from you and I am truly thankful for all your time and efforts.

I’m sincerely thankful for my committee members, Dr. Betty Tsao, Dr. Fred Finkelman, Dr. Hal

Scofield, Dr. Ken Kaufman, and Dr. Senad Divanovic. The help they have provided and time been devoted to my dissertation project has been invaluable. They helped me to keep on track, keep focused and think profoundly.

I would like to thank other really important people who have helped me many times. Dr. Diana

Taft, Dr. Erin Zoller, and Dr. Leah Kottyan, for helping me millions of times on presentation, writing skills, scientific thinking, experimental design and manuscript revisions. I could not survive without your kind support. I would like to thank our amazing past and current lab

v members, Albert, Carrie, Sara, Zubin, for the daily and countless support and help you have provided. Moreover, I would like to thank Dr. Maria Fields, Dr. Jana Reynor, Dr. Jill Fritz, Dr.

Jared Klarquist, Dr. Hesham Shehata and Erik Karmele, who are my fellow students and colleagues in the immunology program, from whom I learned so many experimental skills. They offered so much time and efforts to teach me that I am truly thankful for. I would also like thank other faculties and students in our division and immunology programs for their encouragements and supports throughout these years.

Finally, yet importantly, I would like to thank my all friends and family who provide their endless unconditional love and support to me. For Xiaoming, Shiyu, Yuan, it’s nice to have lunch with you guys every day, and the fun moments when we hang out with other Chinese friends. With you people around, I am less homesick. For the people in Steve’s lab, I really enjoyed just stop by whenever during the day and talk about everything from experiment struggles to life nuances. I really enjoyed the Seattle trip with you people. For my friends, remotely in China, and my parents, although it is to a point that you do not understand what I do for my research, you guys are always there cheering me up and support me when I had exams, writing to do, and experiments remotely. When it’s a Chinese holiday, I would receive tons of lovely greetings and wishes from home, which make me, feel warm. Most importantly, I would dedicate this work to my friend, family, and husband, Shuo. He is always so supportive and encouraging. When I had late night experiments or weekend duties, he would bring me food in case I have no time to buy and eat. Or just simply being there working on his own engineering graduate project aside. When I have ups and downs about experiments or writing, he would always go through those with me and get me re-energized and motivated. I loved every trip we

vi made together to explore outside world. I enjoy and appreciate you being on my side and I love you forever.

vii

Table of Contents

Abstract ...... ii Acknowledgement ...... v List of Abbreviations ...... xi Chapter 1: Introduction ...... 1 1. Autoimmunity ...... 1 1.1. Autoimmune diseases ...... 1 1.2. Systemic lupus erythematosus (SLE)/lupus ...... 2 1.3. SLE mouse models ...... 3 2. Genetics of SLE ...... 5 3. Immune regulation in SLE ...... 6 3.1. The role of type I IFN ...... 7 3.2. The role of B cells ...... 9 4. Gender bias in SLE and other autoimmune diseases ...... 11 4.1. Female gender bias in autoimmune diseases ...... 11 4.2. Current proposed hypotheses for gender bias ...... 12 4.3. X chromosome dosage hypothesis for female preponderance of lupus ...... 14 5. The X chromosome ...... 15 5.1. The nature of the X chromosome ...... 15 5.2. X chromosome inactivation ...... 15 5.3. Some genes escape X chromosome inactivation ...... 16 5.4. The function of XCI escapee: DDX3X/Ddx3x ...... 18 Chapter 2: X chromosome dose and sex bias in autoimmune diseases: increased 47,XXX in SLE and primary Sjögren’s Syndrome (pSS) ...... 20 1. Research questions and rationale...... 20 2. Materials and Methods ...... 21 2.1. Subjects and Genotyping Methods ...... 21 2.2. 47,XXX identification ...... 22 2.3. Fluorescence in situ hybridization (FISH)...... 23 2.4. Polymerase Chain Reaction (PCR) ...... 24

viii

2.5. Quality control and statistical analysis ...... 24 3. Results ...... 26 3.1. Subjects ...... 26 3.2. 47,XXX enrichment in systemic lupus erythematosus and primary Sjögren’s syndrome...... 27 3.3. 47,XXX risk is independent of ancestry ...... 29 3.4. 47, XXX in other autoimmune diseases: primary biliary cirrhosis (PBC) and rheumatoid arthritis (RA) and sarcoidosis...... 30 4. Summary ...... 31 Tables and figures ...... 32 Supplemental tables and figures ...... 38 Chapter 3: Gene dose effect of DDX3X/Ddx3x in propagating female preponderance in SLE .. 47 1. Research questions and rationale ...... 47 2. Materials and Methods ...... 50 2.1. Mice ...... 50 2.2. Tissue harvest and cell culture...... 51 2.3. RNA isolation and real-time polymerase chain reaction ...... 52 2.4. IFN-β in vivo and in vitro stimulation ...... 52 2.5. Flow cytometry ...... 53 2.6. Statistical analysis...... 53 3. Results ...... 53 3.1. Ddx3x deficient mouse and breeding strategy...... 53 3.2. DDX3X/Ddx3x expression...... 54 3.3. IFN-β expression in Ddx3x deficient mice...... 54 3.4. Ddx3x deficient mice phenotyping...... 55 4. Summary ...... 56 Tables and Figures ...... 59 Supplementary tables and figures ...... 65 Chapter 4: The role of Ddx3x in mouse B cell development and function ...... 66 1. Research questions and rationale ...... 66 2. Materials and Methods ...... 67

2.1. Mice ...... 67 2.2. Tissue harvest and cell isolation ...... 68 2.3. Flow cytometry ...... 68 2.4. Bone marrow transplantation ...... 69 2.5. B cell proliferation in vivo...... 69 2.6. ELISA ...... 69 2.7. Statistical analysis...... 69 3. Results ...... 70 3.1. Ddx3x deficiency alters B cell development and populations ...... 70 3.2. Ddx3x deficiency altering B cell development and differentiation is not B cell extrinsic ...... 71 3.3. Ddx3x deficiency affects B cell proliferation ...... 73 3.4. Ddx3x deficiency alters B cell function in the periphery ...... 74 4. Summary ...... 75 Tables and Figures ...... 77 Supplementary tables and figures ...... 86 Chapter 5: Summary and Discussion ...... 91 Bibliography ...... 104

List of Abbreviations

AA African ancestry

AITD Autoimmune thyroid disease

ANA Antinuclear antibodies

APC Antigen presenting cell

APRIL A proliferation-inducing ligand

BAF B allele frequency

BAFF B cell activating factor

BCR B cell receptor

BM Bone marrow

BMDCs Bone marrow derived dendritic cells

BrdU 5-Bromo-2′-deoxyuridine

CK1ε Casein kinase 1 ε

CLP Common lymphoid progenitor cells cGVHD Chronic graft versus host disease

D Diversity gene

DAI DNA-dependent activator of IFN-regulatory factors

DC Dendritic cell

EA European ancestry

FISH Fluorescence in situ hybridization

FOB Follicular B cells

GC-Bs Germinal center B cells

GN Glomerulonephritis

GWAS Genomewide association studies

IFN Interferon

IFNAR Interferon alpha receptor

Ig Immunoglobulin

IKKε Inhibitor of nuclear factor kappa-B kinase subunit epsilon iLN Inguinal lymph node

IPS-1 IFN-β promoter stimulator-1

IRF Interferon regulatory factor

ISGs Interferon stimulated genes

J Joining gene

LRR LogR ratio mLN Mesenteric lymph node

MDA5 Melanoma differentiation associated gene 5

MHC Major histocompatibility complex

MS Multiple sclerosis

MZB Marginal zone B cell

NK NK cell

OR Odds ratio

PAMPs Pathogen associated molecular patterns

PAR Pseudoautosomal region

xii

PBC Primary biliary cirrhosis

PBMCs Peripheral blood mononuclear cells

PCR Polymerase chain reaction

PPRs Pattern recognition receptors pSS Primary Sjögren's syndrome

QC Quality control

RA Rheumatoid arthritis

RIG-I Retinoic-acid-inducible gene I

RNP Ribonucleoprotein

SLE Systemic lupus erythematosus

SNP Single nucleotide polymorphism

T1D Type I diabete

T1/2/3 Transitional B cell 1/2/3

TBK1 TANK-binding kinase 1

TLRs Toll like receptors

V Variable gene

WAT White adipose tissue

XCI X chromosome inactivation

xiii

Chapter 1: Introduction

1. Autoimmunity

The immune system must be tightly regulated in order to maintain its ability to avidly defend the host against pathogens while avoiding inappropriate self-targeted responses that can lead to disease. The random rearrangement of antigen-specific receptors on T and B cells is intended to protect against a wide array of unknown foreign antigens, and various mechanisms of self- tolerance limit the ability of this randomly generated repertoire to strongly recognize self- antigens. Nevertheless, every individual harbors self-reactive T and B cells that have the potential to become aberrantly activated, and thereby contribute to a pathological state known as autoimmune disease.

1.1. Autoimmune diseases

Autoimmune diseases are a group of chronic disorders resulting from dysregulation of immune responses, that lead the body to attack its own cells and tissues. More than 80 conditions have been categorized as autoimmune diseases.

The prevalence of autoimmune disease in the general population rose from 3.2% in 1965 to 5.3% in 1995, then to 9.4% in 2007. According to the National Institutes of Health (NIH), an estimated

14.7–23.5 million Americans suffer from autoimmune disease—as much as 8% of the population[2]. The American Autoimmune Related Diseases Association (AARDA) provides a still higher estimate, suggesting that 50 million Americans suffer from diseases that have an autoimmune basis [3]. Patients who have autoimmune diseases usually suffer from chronic

1 illness. Worse, the major treatments for autoimmune disease are immunosuppressants that can cause devastating long-term side effects.

Some autoimmune diseases are systemic, such as systemic lupus erythematosus (SLE), characterized by various tissues damage and production of autoantibodies, while others are more organ-specific, such as rheumatoid arthritis (RA), affecting majorly joints. The causes of autoimmune disease are heterogeneous, wherein genetic, epigenetic, environmental, and immunoregulatory factors all contributing to the pathogenesis of autoimmune diseases. Although researchers have discovered much about autoimmune diseases, much of their true pathogenic mechanism remains unknown. The immune system’s propensity to attack its own tissues is particularly perplexing. Increased understanding of the pathogenesis of autoimmune disease would greatly aid the design of ever more effective therapeutic approaches for patients who have autoimmune disorders.

1.2. Systemic lupus erythematosus (SLE)/lupus

Systemic lupus erythematosus (SLE, or lupus) is a prototypic multisystem autoimmune disease that is characterized by the production of autoantibodies, inflammatory cytokines, and immune complexes. These immune products collectively damage and ultimately contribute to the failure of the organs, including the kidney, heart, and lung [4].

There are 20–150 cases of SLE per 100,000 people. The vast majority of SLE patients are females of child bearing age. Individuals having African, Native American, or Asian ancestry are at higher risk for SLE. The survival rate for SLE is 70% for 10 years[5]. SLE is a very heterogeneous disease having diverse clinical manifestations. Indeed, one patient’s phenotype

2 can be entirely different from that of the next. Accordingly, the American College of

Rheumatology developed 11 criteria to aid defining SLE; a patient meeting four of these criteria can be designated as an SLE patient [6]. Like most autoimmune diseases, SLE is also caused by multiple factors (Genetics, environmental, immune-regulatory, epigenetic, etc.) [4].

1.3. SLE mouse models Mouse models of lupus have been useful tools in the study of SLE pathogenesis. There are two types of lupus mouse models: (1) spontaneous models, in which disease is driven by genetic features of the mouse strain background (e.g., NZB/W F1 and its derivatives, MRL/lpr, and

BXSB/Yaa); and (2) induced models, including those triggered by chemicals or drugs such as

Pristane; and the chronic graft-versus-host-disease models (cGVHD)[7]. Each of these models represents one aspect or several, of the aggregate human SLE phenotype, such as autoantibody production, lymphoid activation, and lupus nephritis. Not surprisingly, then, each has its own merits and limitations when used to study such a complex human disease.

NZB/WF1 mice are the classic spontaneous SLE mouse model thought to resemble human disease the most. These mice develop high levels of antinuclear antibodies (ANA), including anti-dsDNA IgG, hemolytic anemia, proteinuria, progressive immune complex glomerulonephritis (GN), and B cell dysfunction. GN is observed at around five to six months, and kidney failure happens at around 10 to 12 months. As in SLE patients, this strain has a female sex bias phenotype: female mice spontaneously develop IgG antibodies to DNA and severe immune complex GN earlier than males. However, this strain lacks the autoantibodies against RNA-containing complexes that are seen in SLE patients[8].

Pristane (2, 6, 10, and 14 tetramethylpentadecane or TMPD) -induced lupus is a well-established inducible murine model of SLE. Pristane administration can induce SLE manifestations and production of autoantibodies such as anti-ribonucleoprotein (RNP) antibodies (anti-Su, anti-Sm, and anti-U1RNP), anti-DNA, and anti-histone. In some strains, the Pristane induced model also presents with a phenotype of immune-complex deposition in the kidney, leading to severe proteinuria and nephritis, but at a much lower frequency than compared to NZB/WF1. Pristane- induced SJL/J mice exhibit a female-favored phenotype, with female mice having higher mortality rates, more severe kidney disease, and higher serum levels of antinuclear and anti- dsDNA antibodies than their male siblings [9-12].

The chronic graft-versus-host disease (cGVHD) model can present a lupus-prone phenotype and is used as a lupus mouse model as well. These models are generated by only a single injection of donor cells designed to induce a very fast lupus-prone phenotype in the recipient mice. The disease is induced because of allografted cells. One example is the Bm12-B6 model, where the donor and recipient mice are identical except for three amino acid difference on their major histocompatibility complex (MHC) class II molecule. When injecting Bm12 splenocytes or

CD4+ T cells into a C57BL/6 recipient, autoantibodies can be induced and detected at around two weeks. Donor T cells and recipient B cells are also expanded and activated, with immune complex-mediated GN first observable around three months [13].

Spontaneous lupus models are quite useful for studying the genetic factors that contribute to SLE, whereas the inducible Pristane model confirms that environmental factors may also be important in lupus. However, there are obvious limitations of mouse models of lupus. No mouse model can resemble the complexity of SLE. The genetic background of SLE patients is far more complicated than mouse models can replicate. For one thing, mouse models usually do not

4 present the multiple tissue damage phenotype found in human SLE. In particular, some inducible models develop only a subset of lupus phenotypes and sometimes no renal disease.

Moreover, contradictory results can sometimes be obtained when studying lupus using different strains of mice. For example, a drug study testing the treatment of a soluble B cell activating factor (BAFF) receptor induced remission in NZM2410 mice but can not reverse nephritis in

NZB/WF1 mice[7].

Accordingly, variability and genetic background, among other factors, should be taken into account when using mouse models to study lupus. Moreover, using multiple appropriate models can help ensure discoveries’ applicability to human disease.

2. Genetics of SLE Despite the heterogeneity of SLE’s pathogenesis, genetics plays a large part in its etiology.

People of African, Native American, or Asian ancestry have a three- to sevenfold greater chance of getting lupus than do individuals of European descent [14]. Moreover, monozygotic twins’ concordance rate (approximately 30%) is 10 times greater than that of dizygotic twins (about

3%). What’s more, SLE heritability is estimated to be as high as 66%[15].

Genetic factors predispose certain people to onset of the disease. During the past 20 years, from linkage studies to hypothesis-driven targeted studies, then to genomewide association studies

(GWAS), about 85 genetic variants have been convincingly established as being associated with

SLE[16]. Among these genetic variant, altered copy number of certain genes (C4, FCGR3B,

TLR7) has been linked to SLE[17-19]. Most of the genetic associations are single nucleotide polymorphism (SNPs). Although in rare cases SLE is associated with a single variant (e.g., C1q and C4 deficiency)[20, 21], these independent associations contribute, in aggregate, to a greater

5 effect. C4 deficiency can lead to compromised negative selection of self-reactive B cells and lack of C1q can diminish the ability to remove and eliminate necrotic materials[22]. Some of these conditions associated with SLE have been shown to be associated with multiple autoimmune diseases (as is the case with STAT4 and PTPN22 for RA and diabetes, respectively). These genotyping SNPs can sometimes inform functional results, as seen in the arginine-to-histidine change in ITGAM position 77, which can lead to leukocyte adhesion deficiency. Most other

SNPs associated with SLE fall within the noncoding regions[23].

The genetic loci so far identified—although they are mostly in the noncoding region and thus cannot inform the direct etiology of SLE—hint at critical signaling pathways that might be relevant in the pathogenesis of SLE. For example, 30 of these 85 associated genes, are involved in NF-κB signaling pathways (e.g., IRF5 and IRF7)[16]. Another pathway indicated by these genetic associations is the B cell receptor (BCR) signaling pathway, and the genetic variants involved in this pathway are LYN, BLK, LYP, BANK1, and CD40-all SLE risk genes. These genes account for phosphorylation, transcription, cytokine production, proliferation, and antibody secretion in B cell activation[16].

Future fine mapping studies that will investigate more genetic variants at each genetically associated locus offer some hope for identifying causal variants and relating genetics to biological function.

3. Immune regulation in SLE

In SLE, immune system dysregulation is a major aspect of disease pathogenesis, for systemic abnormalities are mediated by global loss of tolerance. Immune cells change in both number and

6 function, and under- or overproduction of certain cytokines can exacerbate the disease[4]. We will focus on the roles of the cytokine type I IFN and the B cell in SLE.

3.1. The role of type I IFN

Type I interferon (IFN-α, β, ω, ε, and κ) plays a major role in the pathogenesis of SLE[24]. Type

I IFN can be rapidly produced and amplified by all nucleated cells on recognition of conserved

PAMPs (pathogen associated molecular patterns) and signal through IFNAR1 and 2 (interferon α receptors) to mediate antiviral and antitumor activities[25-28]. Both humans and mice that have deficiencies in the genes involved in type I IFN signaling pathway are susceptible to various infections. Type I IFN possesses numerous immunoregulatory functions, having the capacity to control both innate and adaptive immune responses.

Several lines of evidence support a role for type I interferon in the pathogenesis of SLE. First, type I IFN levels are elevated in most SLE patients [29-33]. The elevation in serum IFN-α could be driven by immune complexes as well as by apoptotic cells [34, 35]. In some chronic infection patients, IFN immunotherapy has been known to trigger a lupus-like phenotype that resolves after the therapy is discontinued [36]. Type I IFN-induced gene expression from peripheral blood mononuclear cells (PBMCs) in SLE patients is higher than in healthy controls [29, 37-39].

Moreover, polymorphisms in IFN pathway genes –for example, IRF5 and STAT4 - are associated with an increased risk of systemic autoimmune disease [31, 40, 41].

In lupus, circulating immune complexes and apoptotic cells drive production of type I IFN via recognition of pattern recognition receptors (PRRs)[34, 35]. PRRs are ubiquitously expressed cytoplasmic nucleic acid sensing molecules—for example, retinoic-acid-inducible gene I (RIG-

I)–like RNA helicases,, melanoma differentiation associated gene 5 (MDA5), DNA-dependent activator of IFN-regulatory factors (DAI) and membrane-bound Toll-like receptors (TLRs) such as TLR3, TLR4, TLR7, TLR8, and TLR9[24]. With induction, type I IFN binds to IFNARs and initiates transcription of hundreds of interferon-stimulated genes (ISGs), resulting in direct and indirect antiviral effects as well as regulation of the global immune response, including pathways in target cells, antigen-presenting cells, and effector cells[42].

The induction of ISG by type I IFN not only includes some antiviral function genes (such as

APBEC, Mx GTPase) but also enhances the production of some autoantigens. One example is

Ro52/TRIM21, a member of the tripartite motif family, which can inhibit viral replication at multiple stages in the viral life cycle [43-45]. It is shown that this autoantigen Ro52 is highly expressed upon exposure to IFNs[46]. Moreover, increased production of Ro52 may enhance the likelihood of the presentation of immunostimulatory Ro52 epitopes via MHC class I molecules, which would facilitate their recognition by autoreactive T cells. In addition to their ability to influence the initiation of an adaptive immune response, type I IFNs have the capacity to enhance the survival and effector functions of cells that mediate immunopathology in systemic autoimmunity[24]. IFN-α increases the survival and activation/maturation of dendritic cells (DCs) and induces up-regulation of the MHC and co-stimulatory molecules such as CD40 and CD86

[47, 48]. IFN also induces DCs to secrete BAFF and a proliferation-inducing ligand (APRIL), which, in turn, lead to increases B-cell survival, differentiation, and isotype class-switching [49].

Type I IFNs directly and indirectly affect T cells by increasing survival, activation, and cross- priming and by promoting Th1 differentiation [50-53]. Accordingly, in the setting of chronic IFN production, increased levels of mature DCs could break tolerance by continuously presenting

8 self-antigen in a pro-immune context, potentially activating autoreactive cells that have survived negative selection in the thymus and periphery[24].

3.2. The role of B cells

One key component of SLE autoimmunity is dysregulated B cell function. The B cell is a type of lymphocyte that participates in humoral immune responses in adaptive immunity, and also serves as an antigen presenting cell and a cytokine-producing cell, as well as producing cytokines. B cells develop from common lymphoid progenitor cells (CLP) in bone marrow. Bone marrow stroma cells expressing IL-7 and Flt3 ligand support B cells throughout their development. B cells pass through distinctive stages (pre-pro-B cell, early pro-B, late pro-B, early/large pre-B, late/small pre-B, immature B cells), during which they express different surface antigens, sequentially proliferate, and rearrange their BCRs to acquire antigen specificity (Figure 1)[54-

56]. The B cell receptor (immunoglobulin) is composed of heavy chains and light chains.

Rearrangement of heavy and light chains involves the recombination of Diversity (D), Joining (J)

Figure 1. Schematic cartoon of B cell development and differentiation. and Variable (V) genes (VDJ for the heavy chain and just VJ for light chains, with the addition of a constant (C) region. This process starts at the pre-pro-B cell stage; after B cells finish

9 rearrangement of their heavy chain and express the pre-BCR, B cells finally commit to their lineage. After light chain rearrangement finishes at the immature B cell stage, B cells express

IgM and then exit from the bone marrow (Figure 1)[54-56]. Immature cells that have high a affinity for self-antigens will be deleted or change their specificity to maintain tolerance[57].

After immature B cells exit the bone marrow and migrate to secondary lymphoid tissues, they go through transitional B cell stages and became mature B cells, expressing both IgM and IgD on their surface. In the spleen, B cells can differentiate into follicular B cells (FOB) and marginal zone B cells (MZB). FOB cells reside inside the follicles in the white pulp of secondary and tertiary lymphoid tissue. FOB cells require T cell-dependent help to promote immunoglobulin class switching and produce high affinity antibodies. The MZBs, which are innate-like B cells, are the first line of defense against blood-borne pathogens. In the peripheral tissues, B cells go through somatic hypermutation and affinity maturation to acquire high-affinity antigen receptors through mutation of their Variable (V) genes. Upon activation, B cells can class-switch from

IgM to other isotypes (IgA, IgM, IgD, IgG or IgE). BAFF and APRIL are required for B cell differentiation, survival, and activation [57].

In SLE, the B cell is central to disease. Several B cell depletion-based treatments have been shown to be effective in some patients (e.g., rituximab, belimumab). In SLE patients, there is a B cell population shift toward immature and naïve B cells. CD27+ plasmablasts are also increased, correlating with autoantibody production and tissue damage [57]. The BAFF level also increases, allowing autoreactive B cells to survive longer. Most of the autoantibodies produced in SLE are

IgG and are somatically mutated[58]. B cells not only are the cells producing those autoantibodies but also serve as antigen presenting cells and reciprocally regulate other cell types.

B cells have the capacity to present autoantigens to T cells. In a mutant IgM MRL/lpr lupus

10 mouse model, in which B cells cannot secrete antibodies, but can express them on the cell surface, mice still develop spontaneous lupus and the same level of T cell activation, suggesting that B cell antigen-presenting cell activity is important in inducing abnormal T cell function [59].

Indeed, in humans, B cell expression of co-stimulatory molecules, such as CD40, is enhanced.

SLE B cells also produce increased levels of IL-6 and IL-10, further stimulating B cell function[58]. However, the way in which B cells break tolerance in SLE and become autoreactive is incompletely understood, despite attempts to discover its mechanism.

4. Gender bias in SLE and other autoimmune diseases

4.1. Female gender bias in autoimmune diseases

It was noticed a century ago that females

disproportionately exhibit autoimmune

diseases, a phenomenon that has become

a hallmark of the great majority of

autoimmune conditions. Figure 2[1]

depicts some notable disparities in sex

ratio for certain autoimmune diseases.

The disorders that have most extreme

sex differences are primary Sjögren's

syndrome (pSS), SLE, thyroid disease,

Figure 2. The sex distribution of some major autoimmune and scleroderma, in which more than 90% diseases. The numbers above the bars refer to the total number of disease cases (x1,000,000) in the USA[1]. of patients are female. The female:male

11 ratio is as high as approximately 14:1 in pSS and about 10:1 in SLE. In RA, multiple sclerosis

(MS), and Graves’ disease, around 60–75% of patients are females. For example, in RA, the female–male ratio is about 2–3:1. However, some putatively autoimmune diseases are not female-biased, such as type I diabetes (T1D), for which the female–male ratio is 1:1 in humans.

In recent years, increasing numbers of studies have focused on understanding female predominance in autoimmune diseases. Investigators have identified similarities between different autoimmune diseases and the potential mechanisms underlying female predominance in hopes of developing shared therapeutic approaches.

4.2. Current proposed hypotheses for gender bias

The basic immune response in female and male is sexually dimorphic. Women have more CD4+ lymphocytes than men do, as well as higher levels of production of Th1 cytokines[60, 61]. Many clinical vaccination trials have yielded different results for females and males [62, 63].

The mechanism underlying female preponderance in autoimmune disease is still unclear, but several hypotheses have been developed that attempt to explain this phenomenon [64]. These hypotheses include sex hormones[65], microchimerism[66], X chromosome monosomy[67], and

X chromosome inactivation (XCI)[68].

Much of the attention on female sex bias in autoimmune diseases has focused on sex hormones, such as estrogen/estradiol, progesterone, and testosterone, which serve as primary mediators of immune responses, initiating or accelerating autoimmune processes. The dominance of estrogen or androgen can determine whether an immune response favors Th1 or Th2; estrogen favors a

Th2 response, leading to B cell activation and production of antibodies[1]. One remarkable

12 indication of the role of hormones comes during pregnancy, where for the concentration of estrogen and progesterone greatly increases during the third trimester. In RA and MS patients, disease activity significantly decreases at this time, usually followed by a flare-up postpartum when hormone levels drop [69, 70]. In SLE, by contrast, the disease phenotype remains unchanged or worsens [71-73], but SLE also flares up after parturition.

Another hypothesis suggests microchimerism, which describes the bidirectional cell traffic between mother and child. These foreign cells persist long after parturition. Some studies suggest that microchimerism might play a role in the etiology of specific autoimmune diseases. Nelson and colleagues found higher levels of male DNA in female patients who had scleroderma than in controls[66], a result later confirmed by Artlett et al[74]. Yet, the mechanisms by which this small amount of foreign DNA or cells leads to female sex bias in autoimmunity remains unexplained.

The X chromosome has 1,973 genes; many are immune-associated. X chromosome inactivation

(XCI) in females equalizes the quantity of X chromosome gene expression between men and women. Because of XCI, women are mosaics for the X chromosome. In most cases, the distribution of parentally derived X chromosome is random, with around half of the cells posessing the maternal copy. Sometimes, however, nonrandom XCI occurs that favors inactivation of one of the parental X chromosomes. The resulting skewed XCI was found to be associated with patients who have scleroderma (systemic sclerosis): 64% of patients had skewed

XCI compared to only 8% of controls[64]. This could lead to homozygosity of X-linked alleles that might mask or unmask X-linked mutations or polymorphisms.

Another hypothesis associated with the X chromosome is X chromosome monosomy. X chromosome monosomy is an aneuploidy with the presence of only one X chromosome. When

13 studying primary biliary cirrhosis (PBC), autoimmune thyroid disease (AITD), and systemic sclerosis, Gershwin et al. have discovered an increased percentage of X monosomy in the

PBMCs than in normal individuals [67, 75]. X chromosome monosomy might cause a haploinsufficiency of X-linked genes that ordinarily escape XCI. However, the authors did not explain how this haploinsufficiency in females, which results in similar expression level among the genes escaping XCI between females and males for genes that escape XCI, would lead to female sex bias in certain autoimmune diseases.

4.3. X chromosome dosage hypothesis for female preponderance of lupus

Our lab has previously identified a particular risk factor for SLE—the addition of an extra X chromosome—by studying patients who have Klinefelter’s syndrome (47 chromosomes, sex chromosome: XXY) and Turner’s syndrome (45 chromosomes, sex chromosome: XO).

Strikingly, the risk of SLE in males who have Klinefelter’s syndrome is similar to the risk in normal females (46,XX) and is 14-fold higher than that of normal males (46,XY)[76, 77].

Correspondingly, there are fewer reports than expected of patients who have Turner’s syndrome who also have lupus. In fact, after scouring global and current literature for cases, our group has identified only two case reports in which Turner’s syndrome (45,XO) and SLE occurred together

[78, 79]. Certainly 45,XO is not overrepresented among females who have SLE, although the present data do not allow a precise determination of relative risk of SLE between 45,XO and

46,XX females. Because females have one more X chromosome than males, this increased number of X chromosomes might predispose females to develop SLE.

5. The X chromosome

5.1. The nature of the X chromosome

X and Y chromosomes appear to have been derived from autosomes 300 million years ago. After the X and Y chromosomes separated from autosomes, mutations on the Y chromosome made it male-determining. The develop of these two sex chromosomes thus diverged[80].

There are still homologous regions between the X and Y chromosomes—namely, the pseudoautosomal regions (PAR1 and PAR2) on each end of X and Y. These two regions guarantee proper line up and segregation of X and Y sex chromosome during meiosis. PAR1 comprises 2.6 Mbp of the short-arm tips of both X and Y chromosomes in humans. PAR2 is located at the tips of the long arms, spanning 320 kbp. Crossing over between the X and Y chromosomes is restricted to the pseudoautosomal regions in men; thus, pseudoautosomal genes exhibit an autosomal, rather than sex-linked, pattern of inheritance. Accordingly, females can inherit an allele originally present on the Y chromosome of their father and males can inherit an allele originally present on the X chromosome of their father [80, 81].

5.2. X chromosome inactivation

Since the separation of the X and Y chromosomes in evolution, the Y chromosome kept the male sex-determining genes, and has become shorter, because most of the genes from the autosome before Y became sex-determining are expressed on the X chromosome. In turn, females developed a compensation mechanism whereby one of the X chromosomes became inactivated, a process called X inactivation. During XCI, one of the X chromosomes becomes

15 heterochromatin and aggregates as the Barr body. Consequently, women with two X chromosome (46,XX) have one bar body, as do male patients with Kleinfelter’s syndrome

(47,XXY). Female patients with triple X syndrome (47,XXX) have two bar bodies. This process provides X chromosome gene dosage compensation between females and males [80, 81].

During the early stages of embryo development, cells randomly inactivate either paternal or maternal X chromosome. X inactivation is initiated by expression of a long non-coding RNA named Xist. Xist then acts in its entirety on the X chromosome that expresses itself, covering it and inactivating transcription of that X chromosome. This process is irreversable. A After the early progenitor cells that have inactivated one X chromosome pass certain developmental stages, descendents of these cells will inactivate the same X chromosome [82]. One famous example of

X inactivation is the coat color of the calico cat, inasmuch as it depends on which locus of the gene (orange/non-orange) is inactivated in which cells during early development[80].

Because of XCI in females, women are considered mosaics of the X chromosomes, and the severity of some X-linked diseases depends on the proportion of the expression of the mutant gene in key cell populations.

5.3. Some genes escape X chromosome inactivation

X chromosome inactivation balances gene expression between females and males by condensing one of the X chromosomes into a Barr body [82]. However, the gene silencing is not 100%, some genes escape inactivation, and both alleles of these genes are expressed from the X chromosomes.

The pattern of X inactivation escape is quite different between human and mice. In humans, around 15% of X genes escape X inactivation; these gene tend to be clustered in specific

16 chromosomal regions[83]. In mice, however, only 13 of 393 (3.3%) murine genes are significant expressed by the inactive X chromosome [84]. Interestingly, these murine escapees do not cluster together but are spread over the length of the murine X chromosome[84]. Among those genes that escape inactivation in mice, only eight are common between humans and mice:

XIST/Xist, CA5B/Car5b, DDX3X/Ddx3x, EIF2S3X/Elf2s3x, KDM5C/Kdm5c, KDM6A/Kdm6a,

MID1/Mid1, and 1810030O07RIK/1810030O07Rik, (Table 1).

XIST/Xist Inactive X specific transcripts CA5B/Car5b Carbonic anhydrase 5b, mitochondrial DDX3X/Ddx3x DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 3, X-linked EIF2S3X/Eif2s3x Eukaryotic translation initiation factor 2, subunit 3 gamma KDM5C/Kdm5c(JARID1C) Lysine (K)-specific demethylase 5C KDM6A/Kdm6a(UTX) lysine (K)-specific demethylase 6A MID1/Mid1 Midline 1 1810030O07RIK/1810030O07Rik RIKEN cDNA 1810030O07 gene Table 1. Genes escape XCI in both human and mice.

XIST/Xist confers X inactivation. CA5B/Car5b is carbonic anhydrase 5b, which belongs to a large zinc metalloenzyme family that that catalyze the reversible hydration of carbon dioxide.

DDX3X/Ddx3x belongs to DEAD/H (Asp-Glu-Ala-Asp/His) box RNA helicase family.

EIF2S3X/Eif2s3x is one of the genes that initiate translation. KDM5C/Kdm5c and

KDM6A/Kdm6a are lysine (K)-specific demethylases. KDM5C/Kdm5c thought to be involved in the regulation of transcription and chromatin remodeling. KDM6A/Kdm6a catalyzes the demethylation of tri/di-methylated histone H3. MID1/Mid is likely to be involved in forming multi-protein complexed and acting as anchor points to microtubules.

1810030O07RIK/1810030O07Rik lacks a known function. The differential expression of these genes between females and males could contribute to mechanisms of gender bias in

17 autoimmune diseases. Of these eight genes, DDX3X/Ddx3x is the only one whose function has been shown to be involved in the immune system.

5.4. The function of XCI escapee: DDX3X/Ddx3x

DDX3X belongs to the DEAD (Asp-Glu-Ala-Asp) box RNA helicase family and is ubiquitously expressed in various tissues[85]. It has a homologue on Y chromosome, DDX3Y

(91% identity), which is only expressed in the testis[86-88]. The homology between human

DDX3X protein and mouse Ddx3x protein is around 97%. DEAD-box helicase family members are involved in most cellular processes involving RNA[85]. DDX3Xhas also been found to be involved in type I IFN signaling, viral replication, protein translation, cell cycle control, and apoptosis.

As an RNA helicase, DDX3X can unwind RNA and also interact with RNA splicing factors and mRNPs[89-91]. DDX3X shuttles between cytoplasm and the nucleus, and has the ability to export RNA out of the nucleus together with CRM1 or TAP[92]. As a transcriptional regulator; DDX3X down regulates E-cadherin promoter activity but enhances IFN-β promoter activity as well as the activity of the p21waf promoter. Moreover, DDX3X interacts with many translation initiation factors such as: eIF4a, eIF2α, PABP and eIF3[92]. DDX3X also plays a role in tumourigenesis: lower expression of DDX3X would increase hepatocarcinoma prevalence by 3 fold[93, 94], while in breast cancer DDX3X overexpression in breast epithelial cells increases cell growth, mortality and invasiveness [95]. The role of DDX3X in apoptosis is not very clear. One study suggested that caspase 6 and 9 cannot be activated in the absence of DDX3X hence, DDX3X knock-down cells should be protected from apoptosis[94].

Another study concludedconclude that DDX3X, through interaction with GSK3 and c-IAPa, is involved in an anti-apoptotic protein complex, which binds to death receptors and prevents signaling[95].

Among its multiple functions, DDX3X participates in innate immune signaling, inasmuch as it is involved in a nucleic acid recognition pathway that culminates in the activation of IFN-β transcription. Viral invasion of the immune system is recognized via cell surface or endosomal

Toll-like receptors (PRRs)[96], and the cytoplamic RIG-I like receptor or MDA5[97].

Signaling through both pathways converges at a complex containing TBK1 and the IKKε complex, and futher activates interferon regulatory factors IRF3 and IRF7[97, 98]. DDX3X directly interacts with IKKε to promote activation and phosphorylaton and, in turn, is phosphorylated by TBK1[99]. DDX3X promotes the recruitment of IRF3[99]. Moreover,

DDX3X can also bind IFN-β promoter stimulator-1 (IPS-1)[100, 101]. In addition, knocking down endogenous DDX3X using siRNA reduces IFN-β promoter activity[102]. The N terminus of DDX3X contributes to IFN-β promoter induction by acting as a coactivator[103].

In fact, some viruses target DDX3X’s function in IFN-β promoter induction in order to evade the host immune response [102, 104].

Chapter 2: X chromosome dose and sex bias in autoimmune diseases: increased 47,XXX in SLE and primary Sjögren’s

Syndrome (pSS)

1. Research questions and rationale.

Autoimmune disease affects 5-10% of the global population. Female preponderance is highlighted by the fact that almost 80% of autoimmune patients are women[105]. Systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS), primary biliary cirrhosis (PBC), rheumatoid arthritis (RA) all have a female sex bias, though the ratios vary from fourteen women to one man in pSS to only two to six women to one man in RA[106-109].

Several hypotheses have been proposed to explain the female preponderance in autoimmunity, such as sex hormones[110] and X monosomy[67, 75], but origins remain unknown. We have previously shown that Klinefelter's syndrome (47,XXY) is found in excess among men with

SLE, and the calculated prevalence of SLE among Klinefelter’ syndrome is similar to the prevalence in women (46,XX)[76]. Data for SLE in four core genotype mice show that the disease is associated with the number of X chromosomes, and not phenotypic sex [111, 112]. In aggregate, these data are consistent with the gene dose of chromosome X being a major determining factor for the female predominance of SLE in man and mouse.

In order to further test this X chromosome dose hypothesis, we asked whether having a third X chromosome increases the risk of developing SLE or other autoimmune diseases. Trisomy X

(47,XXX) is present in approximately 1 out of every 1000 live-born girls[113, 114]. There are no consistently recognized abnormalities in sex hormone levels, sexual development, fertility, or

20 pregnancy[115], though premature ovarian failure has been associated with carriage of

47,XXX[116-118]. One suggested phenotype difference is that 47,XXX women may have more anxiety and shyness and lower self-esteem[119]. In fact, 47,XXX is usually unrecognized, with over 90% of the patients undiagnosed unless patients are screened for chromosomal abnormalities for other reasons[115]. Herein, we show that 47,XXX is enriched among patients with SLE and pSS, but not enriched in patients with PBC, RA or the non-autoimmune disease sarcoidosis.

2. Materials and Methods

2.1. Subjects and Genotyping Methods

We used genotyping records of subjects enrolled in multiple different genetic studies (some published) for this work. We excluded all male subjects from our study. To be included as a case in this study, subjects need to have a confirmed diagnosis of a disease of interest: SLE, pSS, PBC,

RA or sarcoidosis. For controls, we used non-affected SLE family controls, non-auto- inflammatory subjects, population controls and non-sarcoidosis controls. A summary of recruitment and inclusion information for each cohort used in this study is available in the supplementary information (Supplementary Table 1a, 1b). Institutional review boards (IRB) at each site provided approval for this study, and individual informed consent was obtained from all participants.

All patients with SLE met the American College of Rheumatology (ACR) classification criteria for systemic lupus erythematosus[120]. Subjects with pSS met the 2002 American-European

Consensus Group (AECG) classification criteria for primary Sjögren's syndrome [121]. PBC

21 patients met the American Association for the Study of Liver Diseases Diagnostic Criteria for

PBC[122]. Participants with RA met 1987 American College of Rheumatology criteria for the classification of RA [123]. Sarcoidosis patients had a) tissue confirmation of granuloma, or b) chest radiographic evidence of bilateral symmetrical hilar adenopathy with either a history of erythema nodosum or at least two years observation during which time no other disease was found to explain radiographic abnormalities [124-127] (Supplementary Table 1a).

Controls include first-degree family members (sisters and mothers) of SLE patients, all of whom were shown not to have SLE [128]. Other controls are healthy individual with no auto- inflammatory diseases from three cohorts [129-131]. Population controls of European decedents recruited globally were also added [132]. Another population control group consisted of self- identified African Americans receiving care at Mt. Sinai Hospital in New York. The last control cohorts were recruited as part of the sarcoidosis studies and were without sarcoidosis [124-127]

(Supplementary Table 1b).

Genotyping records are from various platforms: ImmunoChip [196,524 single nucleotide polymorphisms (SNPs), Illumina], HumanHap370 Beadchip (300,000 SNPs, Illumina). Omni1 chip (1,134,514 SNPs, Illumina), OmniExpress chip (741,000 SNPs, Illumina) or Omni5-Quad chip (4,301,331 SNPs, Illumina). A summary of genotyping platforms for each cohort used in this study is available in the supplementary information (Supplementary Table 1a, 1b).

2.2. 47,XXX identification

X chromosome copy number variations were found by visually inspecting B allele frequency and

LogR ratio (LRR) plot of each subject's X chromosome for abnormalities using Genome Studio

(Illumina), then mean logR ratio was calculated to confirm gain of copy number. In the B allele frequency plot, the fluorescence intensity of the B allele of any SNP is plotted over the total fluorescence for that SNP in a given individual. Therefore, at an SNP with BB homozygosity the result is 100%, with AA homozygosity the result is 0%, and with AB heterozygosity, the result is

50%. In the LRR plot, the fluorescence intensity of total allele A and B of the given sample is normalized by the reference genome to differentiate between copy number gains and losses.

Normal females (46,XX) were identified by the heterozygous "three band" pattern in the B allele frequency plot, which corresponds to the 0%, 50%, and 100% SNP frequencies (i.e. AA, AB,

BB). The mean logR ratio, shown by a red line at 0.0 in the LRR plot, indicates there are two copies of the X chromosome (Figure 1).

Trisomy X (47,XXX) subjects were identified by the "four band" pattern in the B allele frequency plot, which corresponds to 0%, 33%, 66%, and 100% SNP frequencies (i.e. AAA,

AAB, ABB, BBB) with the accompanying mean LRR near 0.2 (Figure 1).

2.3. Fluorescence in situ hybridization (FISH)

When cells were available, samples with chromosome abnormalities identified by B allele frequency and LRR plots were validated by FISH, as previously reported[133]. For subjects that had frozen peripheral blood mononuclear cells (PBMCs), we used commercial FISH probes that recognize the centromeres of chromosomes X and Y to confirm aneuploidies of chromosome X

(alpha-satellite repeats DXZ1 & DYZ3, PID# KI-20030, Veridex). Fifty nuclei were scored for each subject. For subjects that only had Epstein-Barr virus transformed B lymphocyte cell lines

(LCLs or lymphoblastoid cell lines) available, we used commercial FISH probes that recognize

23 the X centromeres, Yp and chromosome 18 (DXZ1, DYZ3, and D18Z1, Vysis) to confirm aneuploidies of chromosome X. Chromosome 18 probe signals served as controls to establish that the cell lines tested had a normal diploid genome. Two hundred nuclei were scored for each subject. Possible mosaicism was assessed according to the percentage of 45,X, 46,XX, or

47,XXX cells enumerated (Figure 2).

2.4. Polymerase Chain Reaction (PCR)

47,XXX samples having only DNA available were confirmed by PCR using TaqMan® copy variation assays (Life technology). The copy number of a gene on the X chromosome (YIPF6) was normalized to TaqMan® copy number reference assay, RNase P, on chromosome 14. Real- time PCR was performed on an Applied Biosystems 7500 thermocycler. Data were analyzed using CopyCaller v2.0 software (Figure 3).

2.5. Quality control and statistical analysis

Subjects that had a call frequency lower than 0.90, genetically XY individuals self-reported as females, subjects whose samples indicated mixing of samples from multiple individuals or duplicated samples were removed.

The enrichment of 47,XXX in SLE, pSS, PBC, RA subjects versus control subjects were tested by two-tailed Fisher’s exact test using R (Version 3.2.2) (Table 1).

The prevalence of SLE and pSS among women with 47,XXX was estimated using Bayes’ theorem. Below are the Bayes’ theorem (P [B/A] = (P [A/B]*P [B])/P [A]) calculations to estimate the prevalence of SLE and primary Sjögren’s syndrome in females with 47,XXX.

A = the frequency of trisomy X, B = the frequency of SLE or pSS in females, P [A/B] represents the frequency of individuals with 47,XXX in SLE or pSS, and P [B/A] would indicate the probability of SLE or pSS within the group with trisomy X.

Condition Prevalences used in the Bayes’ theorem calculations

1) SLE = 1 in 1,000 women in the USA[106]

2) pSS = 1 in 200 women in the USA[134]

3) 47,XXX = 1 in 1,000 women[135]

4) 47,XXX with SLE = 1 in 404 SLE women (data from this study)

5) 47,XXX with pSS = 1 in 344 pSS women (data from this study)

The Bayes’ theorem calculations:

Prevalence of S퐿퐸 푖푛 females with 47, 푋푋푋

= [퐹푟푒q (47, 푋푋푋 푖푛 푆퐿퐸) ∗퐹푟푒푞 (푆퐿퐸 푖푛 푓푒푚푎푙푒s)] / 퐹푟푒푞 (47, 푋푋푋 푖푛 푓푒푚푎푙푒s)

Prevalence of pSS 푖푛 females with 47, 푋푋푋

= [퐹푟푒q (47, 푋푋푋 푖푛 푆S) ∗퐹푟푒푞 (푆S 푖푛 푓푒푚푎푙푒)] / 퐹푟푒푞 (47, 푋푋푋 푖푛 푓푒푚푎푙푒s)

Population stratification was performed using SNP & Variation Suite software (Golden Helix) for SLE cases and controls. European ancestry and African ancestry were segregated within three s.d. of the mean of the first two principal components. Non-European or African ancestry subjects were dropped as well as the ones that their self-identified races did not match the

25 principal component analysis results. Logistic regression modeling in R (Version 3.0.0) using glm function was performed to adjust odds ratio values for ancestry.

To confirm our results were not due to biases in the control groups, we also performed tests comparing 47,XXX enrichment in our case subjects to different control groups. The enrichment of 47,XXX in SLE versus non-affected SLE family controls was tested by two-tailed Fisher’s exact test using R (Version 3.2.2). The enrichment of 47,XXX in pSS versus healthy non-auto- inflammatory controls was also tested by two-tailed Fisher’s exact test using R (Version 3.2.2).

The enrichment of SLE and pSS subjects versus a) controls without non-sarcoidosis subjects, b) controls without population controls and c) controls without non-sarcoidosis and population controls were also tested by two-tailed Fisher’s exact test using R (Version 3.2.2). Finally, the same analysis was used to test the enrichment of 47,XXX in SLE, pSS, PBC, RA subjects versus unselected newborn infants controls as reported by Jacob et.al[135].

3. Results

3.1. Subjects

After quality control, we had 2826 SLE, 1033 pSS, 1118 PBC, 1710 RA, 939 sarcoidosis and

7074 control subjects for our study. For controls, we had 2090 non-affected first-degree family member of SLE patients, 1684 non-auto-inflammatory healthy control subjects, 2680 population control subjects and 620 non-sarcoidosis control subjects (Supplementary Table 1a, 1b). There is no difference in prevalence of 47,XXX between SLE family control versus the rest controls by

Fisher exact test (p=1, data not shown). A risk with the population controls is that some subjects might have the diseases of interest because this group represents a general population and was

26 not specifically screened for the diseases of interest to this study. The non-sarcoidosis controls

(n=620) may also contain subjects with diseases of interest (other than sarcoidosis), but the smaller number of subjects makes it less likely to contain SLE/pSS/PBC/RA patients. Since the non-sarcoidosis control and population control each contains one of the two 47,XXX in the controls, we decided to retain them for modeling purpose. This decision potentially biases our results towards the null hypothesis. We therefore considered all three control cohorts in aggregate for our primary analysis.

3.2. 47,XXX enrichment in systemic lupus erythematosus and primary Sjögren’s syndrome.

Using single nucleotide polymorphism typing of the X chromosome, we found 47,XXX in seven of 2,826 women with SLE and in three of 1,033 women with pSS, but only two in the 7,074 controls (p=0.003, OR=8.78, 95% CI: 1.67-86.79 and p=0.02, OR=10.29, 95% CI: 1.18-123.47; respectively; Table 1, Fig. 1). Five of the seven 47,XXX subjects with SLE were validated by

FISH (Fig. 2). We validated one using frozen PBMCs and another four using lymphoblastoid cell lines (LCLs). Four subjects had 100% of 47,XXX cells; one had 98.5% of 47,XXX cells and

1.5% of 48,XXXX cells, which led us to consider this subject with the other 47,XXX subjects.

We validated the two other 47,XXX subjects with SLE and the three with primary pSS using q-

PCR, which demonstrated a three X chromosome complement in all five DNA samples (Fig. 3).

No DNA or cells were available from the two 47,XXX controls, but the complete fidelity between the genotyping screening and more traditional method supports the likelihood of accuracy in these samples.

Thus, we find that 1 in 404 (95% CI: 196-1004) women with SLE and 1 in 344 (95% CI: 115-

1620) with pSS have 47,XXX, ~2.5 and ~2.9 times higher, respectively, than the published population prevalence of 47,XXX which is ~1/1000[113, 114].

Controls in this study consisted of non-affected SLE family members, non-auto-inflammatory subjects, population controls, and non-sarcoidosis controls. The best controls for the SLE patients are the 2090 non-affected SLE family members. However, there are no 47,XXX participants found in this control cohort. There is a significant difference in 47,XXX enrichment when comparing SLE patients only to the family control cohort (Supplementary Table 3). The non-affected SLE family members could possibly contain subjects with pSS, so we tested

47,XXX enrichment in pSS with only the 1684 healthy non-auto-inflammatory controls and the results remained significant (Supplementary Table 4). The 2680 population controls and 620 non-sarcoidosis controls could potentially contain subjects with SLE or pSS. We tested 47,XXX enrichment in SLE and pSS with the removal of either or both of these control datasets and the results remained significant (Supplementary Table 5a, 5b, 5c). We also compared 47,XXX in

SLE and pSS with a published unselected newborn infants 47,XXX screening datasets (n=20,790)

[135]. Since a reported 47,XXX phenotype is increased anxiety and shyness combined with lower self-esteem[119], we feared 47,XXX participants potential may be less willing to participate in research studies, although this should equally affect case and control recruitment.

Using the unselected newborn infants’ dataset should have no recruitment bias but a potential bias in cases remains. We still have a significant enrichment of 47,XXX in SLE (p=0.036,

OR=2.58, 95% CI: 0.92-6.36) and the same trend for 47,XXX enrichment in pSS (p=0.093,

OR=3.02, 95% CI: 0.57-10.22) (Supplementary Table 6). We are therefore inclined to conclude

28 that the comparison with the combination of all control cohorts is a fair representation of the enrichment in SLE and pSS. The loss of significance in pSS may be due to small sample size.

Using the published population prevalence of 47,XXX in the population (~1/1000) and Bayes’ theorem[76, 113, 114], we predict that SLE exists in 1 of 404 women with 47,XXX and pSS in

1of 69 women with 47,XXX. This calculation assumes that SLE is present in 1/1,000, and pSS in

1/200 women in the USA. We estimate the prevalence of SLE and pSS with 47,XXX is ~2.5 times higher than that with 46,XX and ~25 and ~41 times higher than 46,XY, respectively

(Table 3).

3.3. 47,XXX risk is independent of ancestry

Individuals of African ancestry have a higher risk for SLE compared to those of European ancestry[136], and we observed a majority of the 47,XXX subjects self-identified as of African ancestry. While there are no studies on 47,XXX prevalence by ancestry, there is a study demonstrating 47,XXY differences by ancestry[135]. Therefore, we asked whether the SLE risk associated with 47,XXX is dependent on ancestry. After population stratification, 375 samples were categorized as non-European/African ancestries and therefore excluded. An additional 144 samples were dropped because of a self-identified race mismatch with the principal component analysis, giving us a final of 1592 subjects of African ancestry (AA) and 1083 subjects of

European ancestry (EA) with SLE, along with 3755 AA and 2951 EA control subjects. Among the seven 47,XXX women with SLE, five self-identified as being of African ancestry, and two of

European ancestry. The two control 47,XXX women self-identified as of African ancestry. After principal component analysis for population stratification, one African ancestry SLE 47,XXX

29 subject was removed as an admixed outlier (Table 2). We performed a logistic regression analysis of SLE status with 47,XXX and ancestry as independent variables. The ancestry- adjusted OR was 7.36 with p=0.01, similar to the unadjusted model (Supplementary Table 2).

We also modeled using the interaction term between ancestry and 47,XXX to test for effect modification, but the interaction term was not significant. So we conclude that the odds of

47,XXX in SLE are independent of African ancestry in our data.

In addition, there was an increased incidence of 47,XXX among the pSS patients. The pSS patient cohort was almost entirely comprised of individuals of European ancestry (>95%); all three 47,XXX pSS cases self-identified as being of European ancestry. This meant that adjusting for ancestry effects in pSS is unnecessary.

3.4. 47, XXX in other autoimmune diseases: primary biliary cirrhosis (PBC) and rheumatoid arthritis (RA) and sarcoidosis.

We found 47,XXX in 1 of 1,118 women with PBC, 1 of 1,710 with RA, and none of 939 with sarcoidosis (p>0.05, Table 1). The 47,XXX subject with PBC was validated to have three X chromosomes with q-PCR (Figure 3). No DNA or cells were available for the 47,XXX subject with RA. There was no statistically significant increase in the prevalence of 47,XXX seen in

PBC, RA or sarcoidosis compared to our controls. The comparison of 47,XXX in PBC, RA and sarcoidosis to the unselected newborn infants’ dataset also found no significant difference

(Supplementary Table 6).

4. Summary

We showed that 47,XXX is more frequent in patients with SLE and pSS. Indeed, this and previous work show that the risk of SLE and pSS increases with each additional X chromosome

(Table 3). The ~25-fold increase in predicted prevalence for SLE and ~41-fold increase for pSS observed in 47,XXX women compared to 46,XY men reveals an effect on risk exceeding by many fold that of all other known genetic risk factors for SLE or pSS. The importance of these findings is not for the few individuals with 47,XXX, but rather lies in the fact that rare events and phenotypes reveal insights into the mechanism for the general disease circumstances.

Everyone has an X chromosome, and increased 47,XXX among women with either SLE or pSS informs the potential mechanism underpinning the disparate risk of these diseases found for men and women with a normal sex chromosome complement. Because sexual development and sex hormones are normal in 47,XXX women, these data suggest that the number of X chromosomes is a key factor.

We did not find excess 47,XXX among women with the other autoimmune diseases studied here

(PBC and RA) compared with our controls or the unselected newborn infant dataset (Table 1,

Supplementary Table 6) suggesting the existence of multiple mechanisms for sex bias in autoimmunity. In particular, PBC has been previously associated with increased acquired X- monosomy [67, 75, 137]. Our next step will be exploring the element or elements on X chromosome that responsible for female sex bias in SLE and pSS.

Tables and figures

Table 1. 47,XXX incidence. 47,XXX in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS), primary biliary cirrhosis (PBC), rheumatoid arthritis (RA), sarcoidosis and controls. Each disease state is compared to the control group. P-value, OR with 95% CI were calculated by Fisher exact test.

F/M Sample Odds Ratio Disease 47,XXX p value Incidence ratio Size (95%CI) Systemic lupus 8.78 10:1 2826 7 0.003 1/404 erythematosus (1.67-86.79) Primary Sjögren’s 10.29 14:1 1033 3 0.02 1/344 syndrome (1.18-123.47) Primary Biliary 3.16 10:1 1118 1 NS 1/1118 Cirrhosis (0.05-60.89) 2.07 Rheumatoid Arthritis 2~3:1 1710 1 NS 1/1710 (0.03-39.75) Sarcoidosis - 939 0 - - -

Controls - 7074 2 - - 1/3537

Table 2. Numbers of 47,XXX by ancestry in SLE cases and controls.

Disease Race 47,XXX Sample Size Total AA 4 1592 Systemic lupus erythematosus 2675 EA 2 1083 AA 2 3755 Controls 6706 EA 0 2951 AA 6 5347 Total 9381 EA 2 4034

Table 3. Systemic lupus erythematosus (SLE) and primary Sjögren's syndrome p(SS) relative risk with X chromosome numbers.

X number Karyotype SLE Relative Risk pSS Relative Risk 1 45X; 46XY 1 1 2 46XX; 47XXY* ~10 ~14 3 47XXX ~25 ~41 *47,XXY data with primary Sjögren's syndrome is not available here.

Fig. 1. 47,XXX identification.

B allele frequency and logR ratio plot of 46,XY from a representative normal man (first panel, for comparison purposes only as men were not included in the present study), 46,XX from a normal woman (second panel) and 47,XXX from a trisomy X syndrome patient (third panel).

Fig. 2. 47,XXX validation by fluorescence in situ hybridization assay (FISH).

Validation of 46,XX from a normal female (upper) and 47,XXX from a trisomy X syndrome patient (bottom). The images are showing a single representative nucleus. 200 nuclei were counted.

Fig. 3. 47,XXX validation by PCR.

DNA amplification validation of 47,XXX for two systemic lupus erythematosus patients (SLE1,

SLE2), three primary Sjögren's syndrome patients (pSS1, pSS2, pSS3) and one primary biliary cirrhosis patient (PBC1). As calibrators, we used known 47,XXX (shown in blue), 46,XX

(shown in red) and 45X (shown in purple) samples. Calibrators were all validated by FISH and used to determine copy number in the experimental samples.

Supplemental tables and figures

Supplementary Table 1a. Recruitment information for patient datasets.

Pre- and post-quality control (QC) numbers, genotyping platforms, recruitment strategies, criteria, locations, time and references are listed for SLE, pSS, PBC, RA and sarcoidosis patient datasets. Age information is unavailable.

Pre- Post Genotyping Recruitment Datasets Recruitment Criteria Recruitment locations Recruitment time Reference QC QC platform strategy

Cedars-Sinai Medical Center, California; Hospital for special surgery, New York; Oklahoma 976 subjects: before American College of Rheumatology Omni 1, and Medical Research Foundation, Oklahoma; the University of Alabama at Birmingham, Alabama; 2010; 1821 subject: Not SLE 3449 2826 Consent in clinics (ACR) classification criteria for systemic immunochip Univeristy of Chicago, Illinois; University of Michigan, Michigan; University of Minnesota, 1992-2010; 29 published lupus erythematosus Minnesota; United States; 27 subjects unknown subjects unknown

Newcastle University, United Kingdom; National Institue of Health, USA; Uppsala University American-European Consensus Group Hospital, Sweden; Karolinska Institute, Sweden; The Queen Elizabeth Hospital and University of pSS 1238 1033 Immunochip Consent in clinics (AECG) classification criteria for primary N/A [132] Adelaide, Australia; Stavanger University Hospital, Norway; University of Bergen, Norway; Sjögren's syndrome Université Paris-Sud, France HumanHap American Association for the Study of PBC 1564 1118 370 Beadchip; Consent in clinics Liver Diseases Diagnostic Criteria for Canada, United States, and Europe 2005-2012 [138, 139] Immunochip PBC 1987 American College HumanHap RA_1 453 449 Consent in clinics of Rheumatology criteria for the Toronto, Canada 2004-2008 [131] 370 Beadchip classification of RA 1987 American College RA_2 1290 1261 Immunochip Consent in clinics of Rheumatology criteria for the Argentina, Mexico, Peru, and Chile 2008-2010 [140] classification of RA Tissue confirmation of granuloma; or had Beth Israel Deaconess Medical Center, Massachusetts; Georgetown University Medical Center, Consent in clinics, chest radiographic evidence of bilateral Washington D.C.; Case Western Reserve University and University of Cincinnati Medical Center, 167 subject: 1996- recruit from health symmetrical hilar adenopathy and either a Ohio; Henry Ford Health system, Michigan; Johns Hopkins University School of Medicine, 1999; 429 subjects: Sarcoidosis 956 939 Omni 1 fairs, advertise history of erythema nodosum or at least Maryland; Medical University of South Carolina, South Carolina; Mount Sinai Medical Center, late 90’s to early [124-127] through media, two years observation during which time Florida; National Jewish Medical and Research Center, Colorado; University of Iowa College of 2000’s; 343 subjects: hospital and church no other disease was found to explain Medicine, Iowa; University of Pennsylvania and Allegheny University of the Health Sciences, unknown radiographic abnormalities Pennsylvania; United States. 429 subjects unknown.

Supplementary Table 1b. Recruitment information for control datasets.

Pre- and post-QC numbers, genotyping platforms, recruitment strategies, criteria, locations, time, and references are listed for control datasets. Age information is unavailable.

Pre- Post Genotyping Recruitment Recruitment Datasets Recruitment criteria Recruitment locations Reference QC QC platform strategy time

First-degree lupus family members Non-affected SLE family 2408 2090 Immunochip Consent in clinics (mothers and sisters) who have no United States 1990-2010 [128] controls SLE Community-based recruitment Healthy individuals without known strategies: including major health conditions recruited Cincinnati Sept 2006-Oct 501 497 Omni5 visits to from the geographic area served by Cincinnati, OH, United States [129, 130] cohorts 2010 neighborhood Cincinnati Children's Hospital schools and day-care Medical Center (CCHMC) centers Healthy Consent in individuals, with community clinics, Healthy individuals recruited who no auto- RDRCC recruits in health fair were screened negative for Not 792 747 Immunochip Oklahoma, United States 2009-2010 inflammatory cohorts and employer- autoimmune rheumatic disease by a published diseases sponsored recruiting validated questionnaire[141] events Individuals of European origin from [131], data HumanHap Canada Advertise through the Toronto area who had no history for 112 452 440 370 Beadchip, Toronto, Canada 2005-2011 cohort flyer of rheumatoid arthritis or other subjects not Immunochip inflammatory disease published

Uppsala University Hospital, Sweden; Karolinska Institute, Sweden; The Queen pSS Elizabeth Hospital and University of Adelaide, Australia; Oklahoma Medical 465 352 Immunochip N/A Population controls N/A [132] controls Research Foundation, USA; Stavanger University Hospital, Norway; University of Population Bergen, Norway; University of Minnesota, United States Controls African American patients receiving AA care at Mount Sinai without Not 2379 2328 OmniExpress Consent in clinics Mt Sinai Medical School hospitals, United States 2007-2012 controls selection for any specific condition published requirements Beth Israel Deaconess Medical Center, Massachusetts; Georgetown University Medical Center, Washington D.C.; Case Western Reserve University and University 186 subject: Consent in clinics, of Cincinnati Medical Center, Ohio; Henry Ford Health system, Michigan; Johns 1996-1999; 105 recruit in health Hopkins University School of Medicine, Maryland; Medical University of South subjects: late 90’s Sarcoidosis study controls 639 620 Omni1 fairs, advertise Non-sarcoidosis [124-127] Carolina, South Carolina; Mount Sinai Medical Center, Florida; National Jewish to early 2000’s; through media, Medical and Research Center, Colorado; University of Iowa College of Medicine, 329 subjects: hospital and church Iowa; University of Pennsylvania and Allegheny University of the Health Sciences, unknown Pennsylvania; United States. 106 subjects are unknown.

Supplementary Table2. Adjusted and unadjusted OR of 47,XXX in SLE.

Logistic regression modeling OR p-value

Model 1 (47,XXX only) 7.53 0.01

Model 2 (47,XXX and ancestry) 7.36 0.01

Supplementary Table 3. 47,XXX enrichment in SLE (compared to non-affected SLE family member control cohort).

47,XXX in systemic lupus erythematosus (SLE) and non-affected SLE family controls.

*Modified p-value, OR with 95% CI were calculated by Fisher’s exact test through adding 0.5 into each cell to solve the issue of zero 47,XXX in the controls.

Disease phenotype Sample Size 47, XXX p-value OR(95% CI)

Systemic lupus erythematosus 2826 7 0.02 11.12 (1.26-inf)

Non-affected SLE family controls 2090 0 - -

Supplementary Table 4. 47,XXX enrichment in pSS (compared to healthy controls with no auto-inflammatory diseases).

47,XXX in primary Sjögren's syndrome (pSS) and healthy controls with no auto-inflammatory diseases. *Modified p-value, OR with 95% CI were calculated by Fisher’s exact test through adding 0.5 into each cell to solve the issue of zero 47,XXX in the controls.

Sample 47, OR(95% Disease phenotype p-value Size XXX CI) 11.44(1.08- Primary Sjögren's syndrome 1033 3 0.02 inf) Healthy controls with no auto-inflammatory diseases 1684 0 - -

Supplementary Table 5a. 47,XXX enrichment in SLE and pSS (compared to controls excluding sarcoidosis study control cohort).

47,XXX in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS) and controls excluding sarcoidosis study controls. Each disease state is compared to controls excluding sarcoidosis study controls. P-value, OR with 95% CI were calculated by Fisher’s exact test.

Sampl 47, p- Disease phenotype OR(95% CI) e Size XXX value 16.02 (2.06- Systemic lupus erythematosus 2826 7 0.001 719.67) 18.78 (1.51- Primary Sjögren's syndrome 1033 3 0.009 980.69) Controls (Excluding sarcoidosis study 6454 1 - - controls)

Supplementary Table 5b. 47,XXX enrichment in SLE and pSS (compared to controls excluding population control cohort).

47,XXX in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS) and controls excluding population controls. Each disease state is compared to controls excluding population controls. P-value, OR with 95% CI were calculated by Fisher’s exact test.

Sample 47, Disease phenotype p-value OR(95% CI) Size XXX 10.90 (1.40- Systemic lupus erythematosus 2826 7 0.007 490.58) 12.79 (1.03- Primary Sjögren's syndrome 1033 3 0.02 669.11) Controls (Excluding population controls) 4394 1 - -

Supplementary Table 5c. 47,XXX enrichment in SLE and pSS (compared to controls excluding sarcoidosis study control cohort and population control cohort).

47,XXX in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS) and controls excluding both sarcoidosis study controls and population controls. Each disease state is compared to controls excluding sarcoidosis study controls and population controls. *Modified p- value, OR with 95% CI were calculated by Fisher’s exact test through adding 0.5 into each cell to solve the issue of zero 47,XXX in the controls.

Sample 47, Modified OR(95% Disease phenotype p-value* Size XXX CI)* Systemic lupus erythematosus 2826 7 0.001 20.08 (2.28-inf)

Primary Sjögren's syndrome 1033 3 0.002 25.64 (2.41-inf) Controls (Excluding sarcoidosis study controls and population 3774 0 - - controls)

Supplementary Table 6. 47,XXX incidence (compared to unselected newborn infants cohort).

47,XXX in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (pSS), primary biliary cirrhosis (PBC), rheumatoid arthritis (RA) and unselected newborn infants. Each disease state is compared to the unselected newborn infants group. P-value, OR with 95% CI were calculated by Fisher’s exact test.

Disease phenotype Sample Size 47, XXX OR(95% CI) p-value

Systemic lupus erythematosus 2826 7 2.58 (0.92-6.36) 0.036

Primary Sjögren's syndrome 1033 3 3.02 (0.57-10.22) 0.093

Primary biliary cirrhosis 1118 1 0.93 (0.02-5.82) 1

Rheumatoid arthritis 1710 1 0.61 (0.01-3.80) 1

Unselected newborn infants[135] 20790 20 - -

Chapter 3: Gene dose effect of DDX3X/Ddx3x in propagating female preponderance in SLE

1. Research questions and rationale

Female dominance in autoimmune diseases is a common observation of an unsolved scientific problem that is responsible for a substantial portion of disease risk. For lupus, sex chromosome related disease risk spans a 25-fold range (Table 3, Chapter 2). Understanding female dominance in lupus is likely to provide insight into female dominance in the at least 20 other female dominated autoimmune diseases, for the consistencies and differences between disorders providing much insight into mechanisms.

Systemic lupus erythematosus (SLE, or lupus) is preponderant in females compared to males by five- to 12-fold, depending upon age[106]. The mechanism of female preponderance in SLE is currently an unknown aspect of lupus pathogenesis, the understanding of which may offer specific effective therapies. Our data show that males who have Klinefelter’s syndrome (47,XXY) and females who have triple X syndrome (47,XXX) have a higher risk of SLE than normal males

(46,XY) and females (46,XX). Meanwhile, SLE is virtually unknown in Turner’s syndrome

(45,X), with only two cases reported in the medical literature[78, 142]. These observations are consistent with females’ (46,XX) preponderance of SLE in comparison to normal males’ (46,XY) being the result of a gene dose effect emanating from the X chromosome (Table 3, Chapter 2).

This interpretation was also supported in murine models of lupus using mice from the four core genotypes. In the four core genotype mice, the sry male mouse sex-determining gene has been removed from the Y chromosome and transferred to the autosome. Via breeding, four types of mice: XY male mice, XYsry- female mice, XX female mice, XXsry+ male mice were generated. In

47 these mice, gender and X chromosome number effect have been separated. In two experimental systems (pristane-induced lupus and spontaneous NZM2328 lupus mouse models), the XX chromosome complement, whether male or female, showed greater susceptibility to SLE than male and female mice having XY chromosome[111, 112]. In the first study using the pristane- induced lupus mouse model, the authors castrated the mice to neutralize hormonal effects, showing that the XX sex chromosome effect is independent of hormonal effect[111]. In the second study using the spontaneous NZM2328 lupus mouse model, the authors did not perform castration. They showed that the XX sex chromosome effect was not confounded by the gonadal type, including the hormonal effect [112].

None of the 12 GWASs representing all the major human ancestries shows any association that explains female dominance. Although studies have found TLR7[143, 144] and IRAK1-

MECP2[145] loci on the X chromosome to be associated with SLE, these variants do not explain the SLE sex bias. In fact, TLR7 variants increase the risk for male (46,XY) lupus more than for female (46,XX) lupus in Asians [143-146]. In our opinion, the studies to date have been sufficiently well designed to detect a genetic variant driving the sex bias - were there such a variant to explain the female dominance of SLE. The large number of samples tested in aggregate in these studies, the multiple genotyping platforms used, the multiple ancestries studied, and the reduced crossover in X (the larger regions of disequilibrium reduce the likelihood of missing genetic association), mean that an explanation based on a variant of the X chromosome is unlikely. These observations, in aggregate, lead us to one attractive hypothesis: Female dominance in lupus is explained by ordinarily invariant gene or genes

(including non-coding RNAs) on the X chromosome that escapes XCI and that thus is expressed at a higher level in females than in males, creating the gene dosage effect.

If animal models in mice and natural disease in humans both operate by a gene dose effect, with female propensity dominating in both species, then by parsimony we argue that the basic mechanisms may be similar in both species. The best candidates, given our mechanism predictions (invariant gene or genes and X

(13 genes escape, or 3.3%) than in humans (about ~15% escape) [83, 84]. The overlapping set consists of only eight genes: Xist, Kdm6a, Elf2s3x, Ddx3x, Kdm5c, Car5b, 1810030O07rik, and

Mid1. Of these, DDX3X has been shown to be involved in IFN-ß signaling [147]. Because type I

IFN is involved in the pathogenesis of SLE [24, 29, 31, 36, 40, 41], DDX3X is a particularly interesting candidate for explaining the X chromosome gene-dose effect in SLE.

DDX3X belongs to the DEAD (Asp-Glu-Ala-Asp) box RNA helicase family and is ubiquitously expressed in various tissues [148]. The amino acid sequence homology between human and mouse DDX3X/Ddx3x protein is 97%, implying a powerful evolutionary conservation. DDX3X functions downstream of Toll-like receptors, RIG-I receptor and MDA5 in IFN- β signaling.

DDX3X can activate IRF3, and IRF7, promote phosphrylation of IKKε, and itself be phosphorylated by TBK1[97, 98]. Moreover, DDX3X can also bind IFN-β promoter stimulator-1

(IPS-1)[100, 101]. DDX3X acts as a co-activator via direct or indirect binding to the IFN-β promoter [102, 103]. Knocking down endogenous DDX3X expression using siRNA reduces IFN-

β promoter activity [102].

In light of these information, our specific hypothesis we had planned to test is as follows:

DDX3X/Ddx3x escapes X chromosome inactivation resulting in higher expression levels of

DDX3X/Ddx3x in females than in males. This X-linked DDX3X/Ddx3x gene dose difference contributes to female preponderance in SLE by increasing IFN-β signaling.

Consistent with our hypothesis, studies showed that DDX3X is transcribed approximately 1.5- fold higher in women than in men using transformed B cell lines derived from four HapMap populations [83, 149]. Our preliminary RNA-Seq data from six lymphoblastoid cell lines and six

PBMCs show that DDX3X is increased by 1.40- and 1.43-fold relative to men (data not shown).

Other studies tested the mRNA level of Ddx3x show female compared to male tissue differences ranging from 1.2 to 2.6, with a mean between 1.6 and 2.0 [84, 150-153]. At present, no human biological system will allow us to fully ascertain the functional effects of DDX3X in lupus female dominance to SLE risk. If the human and murine systems have a similar origin for the female dominance of SLE, then murine experiments will provide important insight into the basis of human lupus. We have the first C57BL/6 (B6) Ddx3x floxed mice ever made to help us explore the role of Ddx3x in SLE sex bias. We will take a comprehensive approach with the combination of novel genetic tools and immunological methods to study the role of Ddx3x in

SLE sex bias.

2. Materials and Methods

2.1. Mice

Ddx3x floxed mice (Ddx3x-fl) were obtained from Dr. Josef Penninger in Institute of the

Molecular Biotechnology, Austria. There the mice have been backcrossed to B6 background for eight generations. We bred Ddx3x-fl with Vav1-Cre mice to generate Ddx3x female heterozygous mice and male hemizygous knockout (KO) mice. Mice were housed in a specific

50 pathogen-free barrier facility, and experiments were conducted with approval from the

Institutional Animal Care and Use Committee (IACUC) of the Cincinnati Children's Hospital

Medical Center. The mice used in these experiments were eight to 10 weeks old unless otherwise specified.

2.2. Tissue harvest and cell culture

We collected 50–100 mg of mouse muscle, liver, white adipose tissue (WAT), and heart into

Biopur RNase-free tube (Eppendorf) and immediately froze it with liquid nitrogen for isolation of RNA. Mouse femurs and tibias were harvested, and bone marrow cells were flushed with 1×

PBS (Corning) using a 10 mL syringe and 26 gauge 3/8-inch needle (BD Worldwide), then were filtered through a 70μM filter mesh. Mouse spleens were harvested and mashed on top of the

70μM filter mesh using the end of a plunger from a 1 mL syringe (Thermo Fisher Scientific,

Inc.). Mouse blood was harvested into EDTA tube (BD Biosciences). Red blood cells in bone marrow, spleen and blood were lysed using ACK lysing buffer (Thermo Fisher Scientific).

Leukocytes from bone marrow and spleen were obtained using a CD45 positive selection kit

(Miltenyi Biotec). Bone marrow-derived dendritic cells (BMDCs) were obtained by culturing bone marrow cells with 20 ng/mLGM-CSF (PeproTech) in RPMI1640 (Thermo Fisher Scientific) with 10% Fetal Bovine Serum (FBS) (Thermo Fisher Scientific) and 1% Anti-anti (Thermo

Fisher Scientific) for six days and harvested using a cell scraper.

2.3. RNA isolation and real-time polymerase chain reaction

Tissues collected in Biopur tubes were homogenized in 1 mL Trizol (Thermo Fisher Scientific) using a TissueLyser (Qiagen) set at 30 Hz for 5 minutes and 1/8 inch-diameter stainless steel beads (McMaster-Carr). Total RNA from tissues was extracted using Thermo Fisher Scientific

Trizol protocol. Total RNA from leukocytes was extracted using the mirVana™ miRNA

Isolation Kit (Thermo Fisher Scientific). cDNA was synthesized using the PrimeScript™ RT reagent Kit (Takara). Real-time polymerase chain reaction (PCR) was performed using SYBR

Green-Rox dye-based SYBR Premix Ex Taq (Takara), with specific mouse Ddx3x primer (F:

GGATCACGGGGTGATTCAAGAGG, R: CTATCTCCACGGCCACCAATGC) as published previously[150]. Ddx3x expression was normalized to actin; relative quantification was determined.

2.4. IFN-β in vivo and in vitro stimulation

For in vitro stimulation, BMDCs were stimulated with 25μg/mL high molecular weight (HWM) poly I:C (InvivoGen) for 6 hours or 24 hours. The IFN-β level in the supernatant was measured by VeriKine Mouse Interferon Beta ELISA Kit (PBL Assay Science). In vivo stimulation was performed by intraperitoneal injection of 150μg HWM poly I:C per mouse, and sera were collected 4 hours post stimulation. Serum IFN-β level was measured by an ISRE-L929 reporter assay, as previously described[154].

2.5. Flow cytometry

Cells from bone marrow, spleen, and lymph nodes were stained with fluorescence-labeled antibodies (Supplementary table 1) in 100μl flow buffer (2% FBS in PBS) containing

CD16/CD32 Fc blocking antibody (ebioscience) for 30 minutes at 4 °C, then fixed with 100μL fixation buffer for 30 minutes at 4 °C. Data were acquired on LSR II or LSR Fortessa flows cytometer (BD biosciences), and analyzed by FACsDiva (BD biosciences) and FlowJo software

(TreeStar, Inc.).

2.6. Statistical analysis

Experiments were analyzed using GraphPad Prism software (Version 6, GraphPad Software, Inc.

GraphPad Software, Inc.). Data are presented as mean ± SD. If not specified, data were calculated by two-tailed t-test for two groups and one-way ANOVA for three groups, with *p ≤

0.05, **p ≤ 0.01, ***p ≤ 0.001, or ****p ≤ 0.0001 indicating significance.

3. Results

3.1. Ddx3x deficient mouse and breeding strategy.

Ddx3x conditional KO mice (Figure 1a) were kindly provided by our collaborator Dr. Josef

Penniger in Austria. We first bred the Ddx3x floxed mice with EIIa-Cre mice, a full--body Cre- expressing mouse. However, no female or male Ddx3x KO mice were obtained after several rounds of breeding. The absence of KO survivors suggested that this gene is critical for embryogenesis. Next, we bred Ddx3x conditional KO mice with the Vav1-Cre mice, which

53 would delete the Ddx3x gene in all the hematopoietic cells. After breeding, we obtained heterozygous female mice, and hemizygous male mice. However, despite several rounds of breeding we were unable to obtain the homozygous female mice with homozygosity dysfunctional Ddx3x in their hematopoietic lineage (Figure 1b).

3.2. DDX3X/Ddx3x expression.

As the literature suggested, Ddx3x should have a higher expression level in females than in males in both human and mice [83, 84]. We tested the Ddx3x mRNA level in C57BL/6 mice from the

Jackson Laboratory and then in Ddx3x-Vav1 deficient mice we generated. Astonishingly, we found no significant expression differences between female and male C57BL/6 mice in spleen tissue (Figure 2a). In the mice we generated, female Ddx3x heterozygous mice have similar expression levels of Ddx3x to those of female WT and male WT mice in leukocytes from bone marrow cells (Figure 2b). The Ddx3x male hemizygous mice have 80% reduction of Ddx3x expression in leukocytes from BM and 60% reduction in leukocytes from spleen (Figure 2c, 2d).

In addition, we also tested Ddx3x expression in white adipose tissue (WAT), heart, blood and muscle. We found that female C57BL/6 mice have higher expression of Ddx3x in WAT and heart but not in blood, muscle, or heart (Figure 2e, 2f, 2g, 2h, 2i). These data suggest that Ddx3x escaping X inactivation can be tissue-specific and possibly related to genetic background.

3.3. IFN-β expression in Ddx3x deficient mice.

As Ddx3x has been previously shown to participate in the IFN-β signaling pathway, and because type I IFN plays a significant role in pathogenesis in SLE, we next quantified IFN-β production

54 in these Ddx3x deficient mice. Because there are no expression differences for Ddx3x in female

WT, female heterozygous, and male WT mice, we compared male WT and male Ddx3x-Vav1 hemizygous KO mice for the following experiment (designated as Vav1ddx3x). We generated bone marrow derived dendritic cells (BMDCs) from Vav1ddx3x mice, and their WT littermate control We found, surprisingly, that BMDCs from Vav1ddx3x mice produced more IFN-β, after in vitro stimulation using 25μg/mL of poly I:C for 6 hours or 24 hours than did the BMDCs from the WT mice (Figure 3). However, in vivo injection of 150μg poly I:C induced a lower level of serum IFN-β in the Vav1ddx3x mice than in the WT male (Figure 3). These results may appear contradictory and remain without a satisfactory explanation.

3.4. Ddx3x deficient mice phenotyping.

We then examined the cellular phenotypes of the male Vav1ddx3x mice. In the bone marrow, we found a fewer B220+ cells, and more CD11b+ and Ter119+ cells (Figure 4a). Vav1ddx3x mice had similar numbers of double positive, double negative and single positive T cells in thymus than did WT mice (data not shown). In the spleen, we found decreased numbers of B cells, T cells, and NK cells, while macrophage and DC numbers did not change (Figure 4b, 4c, 4d: macrophage and DC data not shown). For lymph nodes, we examined the inguinal (iLN) and mesenteric lymph nodes (mLN). In iLN, we found decreased number of B cells in the Vav1ddx3x mice than in the WT mice, whereas in the mLN, B cell number was similar (Figure 4e, 4f).

4. Summary

We picked the XCI escaping gene DDX3X/Ddx3x as our first candidate gene to study of SLE sex bias. We obtained Ddx3x floxed mice and bred with EIIa-Cre mice to generate different dose of

Ddx3x in female and male mice. But we never got any female or male KO mice. Even breeding

Ddx3x floxed mice with the Vav1-Cre did not allow for generation of female homozygous KO mice; however, the male hemizygous mice were viable (Vav1ddx3x). These findings suggest that

Ddx3x likely plays a critical role in embryogenesis. Indeed, Li et al. showed that in vitro knockdown of Ddx3x in zygote cytoplasm resulted in less viable blastocyst number and the cell cycle arrest between 2-cell to 4-cell stage[155]. Chen et al. further demonstrated that ablation of

Ddx3x in the epiblast caused widespread apoptosis and abnormal cell growth, suggesting Ddx3x deficiency leads to more genome damage and resulting in cell cycle arrest and apoptosis[156].

Although our data and published reports support the role for Ddx3x in embryogenesis, the underlying causes as to why breeding Ddx3x floxed mice with Vav1-Cre mice generated only male but not female KO mice remains unknown. In the male mice, one may wonder if the Y chromosome analogue of Ddx3x: Ddx3y would compensate for Ddx3x’s function and support the survival of male KO mice. However, three groups have reported independently that the protein expression of ddx3y is limited to testis and ddx3y is important for sperm-genesis. This phenomenon is perplexing and remains without a satisfactory explanation.

Quantification of Ddx3x mRNA level in mice also provide unexpected results. Notably, we did not detect differential expression level of Ddx3x between female and male mice in the spleen.

Moreover, in the Ddx3x-Vav1 conditional KO mouse model we generated, the heterozygous female mice express the same level of Ddx3x mRNA as the female WT mice in leukocytes from bone marrow. In other tissues evaluated, female B6 mice have a higher Ddx3x expression in

WAT and heart, but not in liver, muscle, or blood, than did male B6 mice. Our results partially agree with those of Chen et al. [150]. Like them, we found that female B6 mice express more

Ddx3x in mouse adipose tissue and heart than do male B6 mice, but we did not replicate their results showing differential expression in the liver, perhaps because of the different mouse strains were used. Based on our data, Ddx3x escaping from X inactivation might be tissue specific, which varies from one tissue to another. Chen et al. did not measure Ddx3x expression in spleen, blood, or bone marrow. Ddx3x expression in these immune tissues is of particular importance when preparing to investigate the role of Ddx3x in promoting IFN-β production in the SLE context. Unfortunately, because we did not see differential expression between female

WT and female heterozygotes for Ddx3x and male WT mice, these mice are not appropriate model to use in answering our original questions. However, we found that Ddx3x male hemizygous KO mice (Vav1ddx3x) had an 80% reduction of mRNA of Ddx3x in leukocytes in bone marrow. Notably, deletion of Ddx3x mRNA is not 100% in the Vav1ddx3x mice could possibly due to Vav1-Cre efficiency not being 100% in every hematopoietic cells.

Since the Vav1ddx3x mice have a decreased dosage of Ddx3x, instead of testing whether increased expression of Ddx3x predisposes to elevated lupus susceptibility, we thought we could test whether decreased Ddx3x leads to attenuated lupus disease. Accordingly, we stimulated the

Vav1ddx3x mice in vitro (in BMDCs) and in vivo to induce an IFN-β response. Astonishingly, we found a totally opposite phenotype for IFN-β levels between in vitro and in vivo results—when we stimulated in vitro using poly I:C, BMDCs generated from Vav1ddx3x mice produced significantly more IFN-β than those from WT mice. Whereas the in vivo stimulation using poly

I:C induced reduced serum IFN-β levels in the Vav1ddx3x mice compared to the WT mice.

To define whether any other factors in the Vav1ddx3x mice might contribute to this unexpected phenotype, we next thoroughly explored the phenotype of immune cells populations in these mice. We learned that B cell number in the spleen, iLN, and bone marrow significantly decreased in the Vav1ddx3x mice compared to levels in the WT mice. T cell development appeared to be normal in the thymus, but T cell and NK cell numbers also dropped in the spleen.

After discovery of these immune cell population deficits phenotype, we concluded that Vav1ddx3x mice do not represent a proper model for testing the type I IFN hypothesis. Since every nucleated cell can produce type I IFN, it would be hard to distinguish if the decreased serum level of IFN-β following in vivo stimulation is due to the effect of reduced numbers of T cell, B cell and NK cell or due to Ddx3x function itself. Also, the BMDCs data suggests that Ddx3x might inhibit production of IFN-β in dendritic cells, that it could have different effects on different cell types.

Notably however, the observed B cell deficit phenotype in the Vav1ddx3x mice might provide a hint about a novel function for Ddx3x regulation in B cell development. Considering that B cell function is perturbed in SLE, Ddx3x’s effect on B cell numbers poses interesting questions:

Whether Ddx3x controls B cell development and function?

Tables and Figures

Figure 1. Ddx3x floxed mice targeting construct and breeding strategy.

(a) Ddx3x floxed mice (Ddx3x-fl) targeting construct. LoxP site was inserted into the flanking exon 2 in Ddx3x gene. With the presence of Cre, exon 2 will be deleted, resulting in an early stop codon in exon 3 and thus stopping translation. 64, 27, and 65 are Ddx3x genotyping primers. (b)

Ddx3x floxed mice breeding strategy: Ddx3x floxed mice were bred with Vav1-Cre mice to obtain female heterozygous mice and male hemizygous KO mice deficient in hematopoietic cell

Ddx3x expression.

Figure 2. Ddx3x expression is reduced in Vav1ddx3x mice. a) Ddx3x mRNA expression in female and male C57BL/6 mice’s spleen tissue. (b) Ddx3x mRNA expression in leukocytes from bone marrow in female WT, female Ddx3x-Vav1 heterozygous, male WT, and male Ddx3x-Vav1 hemizygous KO mice as quantified by qRT-PCR.

Ddx3x mRNA expression in leukocytes from (c) bone marrow and (d) spleen of male WT and male Ddx3x-Vav1 hemizygous KO mice. Ddx3x mRNA expression in (e) blood, (f) muscle, (g) liver, (h) white adipose tissue (WAT), and (i) heart of female and male C57BL/6 mice. Ddx3x expression was normalized to actin. ****p≤ 0.0001.

a) b)

c) d)

60 e) f) g)

h) i)

Figure 3. IFN-β expression with Ddx3x deficiency.

(a) BMDCs generated from Vav1ddx3x mice and WT bone marrow cells were stimulated with 25

μg/mL of poly I:C for 6 hours or 24 hours. IFN-β in the supernatant was measured by ELISA. (b)

Vav1ddx3x mice and WT were intraperitoneally injected with 150 μg poly I:C; 4 hours after stimulation, sera were collected and IFN-β production was measured by bioassay.

Figure 4. Vav1ddx3x mice have a decreased number of B cells in spleen and BM.

(a) B220+, CD11b+, and Ter119+ single positive cell numbers in bone marrow of WT versus

Ddx3x-Vav1 male hemizygous knockout mice (Vav1ddx3x). (b) B cell number, (c) T cell number, and (d) NK cell number in the spleen of WT versus Vav1ddx3x mice. B cells were gated as CD3-

CD19+, T cells were gated as CD3+ CD19-, and NK cells were gated as CD3-CD19-NKp46+. B cell numbe in (e) iLN and (f) mLN. B cells were gated as CD3-CD19+. *p ≤ 0.05, **p ≤ 0.01,

***p ≤ 0.001, ****p ≤ 0.0001.

b) Spleen c) Spleen

Spleen

Spleen d)

e) f)

B cell# B cell#

Supplementary tables and figures

Supplementary Table 1.

List of flow cytometry antibodies used.

Antigen Fluorochrome Company Catalog# Anti-MHC Class II (I-A/I-E) Alexa Fluor® 700 eBioscience 56-5321-82 CD11b Pacific Blue BioLegend 101224 CD11c APC eBioscience 17-0114-82 CD19 FITC BioLegend 115505 CD335 (NKp46) PE eBioscience 12-3351-82 CD3e V500 BD Bioscience 560771 CD4 BV421 Biolegend 100443 CD8a PerCP Biolegend 100731 F4/80 PerCP BioLegend 123126 Live/Dead Fixable Stain eF506 eBioscience 65-0866-14 LIVE/DEAD Fixable Stain Indo-1 Life technology L-34962 TCR β APC-Cy7 Biolegend 109220

Chapter 4: The role of Ddx3x in mouse B cell development and function

1. Research questions and rationale

In the previous chapter, we talked about to identify any genes on the X chromosome accountable for female sex bias in SLE, we took a candidate gene approach and chose DDX3X/Ddx3x, for the data then available showed that it would escape X inactivation in both humans and mice and promote the IFN-β signaling pathway. We hypothesized that DDX3X/Ddx3x’s escape of XCI results in higher expression levels of DDX3X/Ddx3x in females than in males and that this X- linked DDX3X/Ddx3x gene dose effect contributes to the female preponderance in SLE by increasing IFN-β signaling.

We obtained Ddx3x floxed mice and bred with Vav1-Cre mice to generate our working model.

Unexpectedly, female mice didn’t express a higher level of Ddx3x than male mice in spleen or bone marrow, which are the relevant immune tissues. We were left with the choice of comparing male Vav1ddx3x mice (80% reduction of Ddx3x in BM) with their littermate controls. We planned to test whether less Ddx3x in the Vav1ddx3x mice would lead to less IFN-β and confer less disease susceptibility.

However, we obtained contradictory results, leaving us uncertain whether Vav1ddx3x mice experience more IFN-β production or less. We also discovered that these mice have a B cell deficit phenotype in both spleen and bone marrow.

In SLE, the B cell is central to the disease phenotype. In SLE patients, immature, naïve B, and

CD27+ plasmablasts are expanded. The increased number of CD27+ plasmablasts correlates

66 with autoantibodies production and tissue damage. The level of BAFF is also increased, allowing autoreactive B cells to survive longer. Moreover, B cells generate mutations at a higher rate during the somatic hypermutation process—most of the autoantibodies produced in SLE are IgG and somatically mutated. Also, expression of the co-stimulatory molecules, such as CD40, is enhanced in T cell-B cell interaction. B cells can serve as an antigen-presenting cell and present self-antigens to T cells. B cells also produce increased levels of IL-6, and IL-10 in SLE which further stimulating its function [57-59].

B cells produce autoantibodies and play a major role in the pathogenesis in SLE, we were thus very interested in learning how Ddx3x regulates B cell biology in the Vav1ddx3x mice. In this chapter, we focused on and investigated how Ddx3x affects B cell population and function.

2. Materials and Methods

2.1. Mice

Ddx3x floxed mice (Ddx3x-fl) were obtained from Dr. Josef Penninger in Institute of the

Molecular Biotechnology, Austria. There the mice have been backcrossed to B6 background for eight generations. We bred Ddx3x-fl with Vav1-Cre mice and obtained male hemizygous KO mice (Vav1ddx3x). The control group used in this study was littermate control (WT) of Vav1ddx3x mice. B6 CD45.1 (BoyJ) mice used in bone marrow transplantation experiments were purchased from the Jackson Laboratory. Mice were housed in a specific pathogen-free barrier facility, and experiments were conducted with approval from the Institutional Animal Care and Use

Committee (IACUC) of the Cincinnati Children's Hospital Medical Center. The mice used in these experiments were eight to 10 weeks old unless otherwise specified.

2.2. Tissue harvest and cell isolation

Mouse femurs and tibias were harvested, and bone marrow cells were flushed with 1× PBS

(Corning) using a 10 mL syringe and a 26-gauge 3/8-inch needle (BD Worldwide), then filtered through a 70 μM filter mesh. Mouse spleens and lymph nodes were harvested and mashed on top of the 70 μM filter mesh using the end of the plunger for the 1 mL syringe. Red blood cells were lysed using an ACK lysing buffer (Thermo Fisher Scientific, Inc.) for bone marrow cells and splenocytes.

2.3. Flow cytometry

Cells from bone marrow and spleen were stained with fluorescence-labeled antibodies

(Supplementary table 1) in 100 μL flow buffer (2% FBS in PBS) containing CD16/CD32 Fc blocking antibody (ebioscience) for 30 minutes at 4 °C, then fixed with 100 μL fixation buffer for 30 minutes at 4 °C for extracellular staining. For intracellular antigens, cells were first stained with extracellular antibodies as previously described, then incubated with 100 μL

CytoFix/CytoPerm Solution (BD biosciences) for 30 minutes at 4 °C, washed by 100 μL 1×

Perm buffer (BD biosciences) twice, stained with fluorescent antibodies for 30 minutes at 4 °C, and then washed by flow buffer. Data were acquired on LSR II, the LSR Fortessa flow cytometer

(BD Biosciences), and analyzed by FACsDiva (BD Biosciences) and FlowJo software (TreeStar,

Inc.).

2.4. Bone marrow transplantation

Recipient mice were lethally irradiated with two doses of irradiation at 700 rad and then 475 rad.

Two and half million bone marrow cells from CD45.1 B6 WT (BoyJ) mice (Jackson Laboratory) and CD45.2 Vav1ddx3x mice were mixed together at a 1:4 ratio, then intravenously injected into recipient B6 CD45.1 mice (BoyJ) (n=6). Transplantation was performed by Cincinnati

Children’s Comprehensive Mouse and Cancer Core. Mice were examined 14 weeks and 17 weeks after transplantation.

2.5. B cell proliferation in vivo

Mice were intraperitoneally injected with 2 mg of 5-Bromo-2′-deoxyuridine (BrdU) (Sigma-

Alrich) 14 hours before sacrifice. The cells were stained using BrdU Staining Kit for Flow

Cytometry eFluor® 450 (ebioscience).

2.6. ELISA Serum immunoglobulin level was measured using Mouse IgA, IgG, IgM, IgE, IgG1, IgG2a,

IgG2b and IgG3 ELISA Ready-SET-Go kits (ebioscience). Serum BAFF level was measured using Mouse BAFF/BLyS/TNFSF13B Quantikine ELISA Kit (R&D system).

2.7. Statistical analysis

Experiments were analyzed using GraphPad Prism software (Version 6, GraphPad Software,

Inc.). Data are presented as mean ± SD. Unless otherwise specified, data were calculated by two-

69 tailed t-test for two groups and one-way ANOVA for three groups, with *p ≤ 0.05, **p ≤ 0.01,

***p ≤ 0.001, or ****p ≤ 0.0001 indicating significance.

3. Results

3.1. Ddx3x deficiency alters B cell development and populations

We previously observed that Vav1ddx3x mice (male Ddx3x-Vav1 hemizygous KO mice) have a B cell number deficit phenotype in both spleen and bone marrow. This led us to wonder whether

Ddx3x deficiency affects B cells during early lymphopoiesis. However, it was unclear at which stage of B cell development was effected by Ddx3x deficiency. B cell development in bone marrow is a sequential process of proliferation and BCR rearrangement through discrete stages.

To identify which stage or stages during B cell development are affected by Ddx3x deficiency, we used the Hardy classification system to discriminate different stages of B cells, as previously published[157]. Bone marrow cells were first gated on lineage negative markers (CD3–, CD11b–,

Ly6c–, Gr-1–, and Ter119–, data not shown). From the B220+CD43+ population, the four early stages (FrA: pre-pro-B cells, FrB: early pro-B cell, FrC: late pro-B cells, FrC’: early/large pre-B cells) during B cell development were defined by BP-1 and CD24 gating (Figure 1a). Vav1ddx3x mice had fewer cells at the early pro-B cell stage, but had a similar number of cells at the pre- pro-B, late pro-B and early pre-B stages (Figure 1b). The next two stages (FrD: small/late pre-B cells, FrE: immature B cells) were defined by IgM and IgD gated from the B220+CD43- population (Figure 1a). Immature cells then would exit out from the bone marrow and become mature in the periphery, then recirculate back into the bone marrow. Mature B cells (FrF) is

IgM+IgD+ (Figure 1a). B cell numbers were significantly reduced during at the late pre-B stage,

70 followed by immature B and mature B cell stages, in Vav1ddx3x mice in bone marrow (Figure

1b).

When immature B cells exit to the periphery, they go through transitional stages (T1, T2, T3), then acquire maturity. The numbers of all three transitional B cell stages decreased in the

Vav1ddx3x mice (Figure 1c). Mature B cells can differentiate into two main subpopulations: FOB and MZBs in the spleen. FOB proportion and number significantly decreased in the Vav1ddx3x mice in the spleen, whereas MZB proportion and number were increased. Besides FOB deficit phenotype in the periphery, the germinal center B cell (GC-B) and plasmablasts in spleen were also decreased in the Vav1ddx3x mice (Supplementary figure 1). These data suggest that ddx3x deficiency affects B cell development during early stages of lymphopoiesis and that this effect lasts to the periphery (Figure 1d).

3.2. Ddx3x deficiency altering B cell development and differentiation is not B cell extrinsic

In the Vav1ddx3x mice, Ddx3x is deleted not only in B cells but in all the hematopoietic cells. In fact, we have showed (Figure 4, chapter 3) that T cell and NK cell numbers also decreased in the spleen of the Vav1ddx3x mice. We then raised the question of whether the B cell number deficit phenotype we observed in the Vav1ddx3x mice is B cell intrinsic or extrinsic. To answer this question, we generated mixed bone marrow chimera mice. Bone marrow cells from CD45.2

Vav1ddx3x mice and CD45.1 B6 WT mice(BoyJ) were mixed together then transplanted into lethally irradiated CD45.1 B6 WT recipient mice. Results were analyzed at least 14 weeks after transplantation. When we mixed donor bone marrow cells from CD45.1 B6 WT mice and

CD45.2 Vav1ddx3x mice at a 1:1 ratio, we saw far fewer CD45.2 cells reconstituted in the mixed

71 bone marrow chimera (data not shown). Accordingly, we generated mixed chimera mice receiving four times more donor cells from the Vav1ddx3x mice than from CD45.1 B6 WT mice

(Figure 2a).

After reconstitution, the mean average percentage of each donor cells in the bone marrow was similar in the mixed bone marrow chimera mice, though there was variability among mice

(Figure 2b). Within the late pre-B, immature B, and mature B cell populations in bone marrow, the percentage of cells derived from the Vav1ddx3x mice was significantly decreased (Figure 2c).

Moreover, the reduction was exacerbated in FOB and MZB in spleen, suggesting that donor B cells from Vav1ddx3x mice had a competitive disadvantage over the WT donor cells (Figure 2d).

With the presence of many more WT cells in the mixed bone marrow chimera mice, the percentage of late pre-B, immature B and mature B in cells derived from Vav1ddx3x donor did not recapitulate the percentage of that in the cells derived from WT donor (Figure 2e). This is similar to the phenotypic differences we observed in the Vav1ddx3x mice for these three B cell stages in bone marrow compared to the WT mice. It suggests that B cell deficit phenotype for late pre-B, immature B and mature B in bone marrow is intrinsic, and that this phenotype does not require contributors from other cells. In the spleen, the percentage of FOB in cells derived from Vav1ddx3x donor was also lower than that in the cells derived from WT donor, suggesting that FOB deficit phenotype is also intrinsic (Figure 2f). However, the percentage of MZB is similar in cells derived from both donors, indicating that the increased number of MZBs in the

Vav1ddx3x was an extrinsic effect (Figure 2f).

In the mixed bone marrow chimera mice, we had far fewer B cells derived from the Vav1ddx3x donor and thus were unable to determine whether the Vav1ddx3x cells had any dominant negative effect on the WT cells.

3.3. Ddx3x deficiency affects B cell proliferation

To investigate whether Ddx3x deficiency could affect B cell proliferation as an underlying mechanism to explain the B cell deficit phenotype, we injected 2 mg of BrdU intraperitoneally per mouse to Vav1ddx3x and WT mice. After 14 hours, the proportion of proliferating, BrdU+ cells within different B-cell subpopulations from Vav1ddx3x mice compared to WT mice was analyzed by flow cytometry. A lower proliferation rate was observed among Vav1ddx3x mice than among the WT mice during the late pro-B and early pre-B cell stages (Figure 3a, 3b). The lowered proliferation rate at these two stages might explain the B cell number decrease for the following late pre-B, immature B, and mature B cell stages. However, we did not find proliferation rate differing at the early pro-B stage or the pre-pro-B stage between the Vav1ddx3x mice and WT mice, suggesting the early pro-B cell number deficit phenotype can be caused by some alternative mechanism(s). And these other mechanism(s) can synergistically or additively function together with lowered proliferation rate to create the B cell number loss phenotype in the Vav1ddx3x mice we observed.

We also measured serum BAFF level in the Vav1ddx3x versus WT mice, for BAFF is essential for

B cell survival and maturation. Serum BAFF level was increased in Vav1dd3x mice

(Supplemental figure 2). Increased BAFF level could promote B cell survival in the Vav1dd3x mice. With less B cell in the Vav1ddx3x mice, the increased serum BAFF level could be due to excessive BAFF production binding to every B cells. Alternatively, because BAFF is produced by various cell types—including monocytes, dendritic cells, and bone marrow stromal cells—as we observed an increased number of CD11b+ cells in the bone marrow in Vav1dd3x mice, the

BAFF-producing cells in the Vav1dd3x mice might be increased, contributing to increased serum

BAFF level.

3.4. Ddx3x deficiency alters B cell function in the periphery

While Ddx3x deficiency clearly affects B cell development and differentiation resulting in less B cell numbers, we wondered if Ddx3x deficiency would alter B cell functions regarding immunoglobulin production? Astonishingly, we found significantly increased levels of serum

IgA, IgG and IgM in the Vav1ddx3x mice while IgE level is decreased (Figure 4a, 4b, 4c, 4d).

IgG levels increased fourfold, IgM threefold, and IgA 1.75-fold in the Vav1ddx3x mice. Moreover, within the IgG family, serum IgG1, IgG2b, and IgG3 levels significantly increased in the

Vav1ddx3x mice (Figure 4e, 4f, 4g, 4h). Within IgG, IgG2b was the most abundant and increased four-fold in the Vav1ddx3x mice.

To investigate which cell type might be responsible for producing the increased level of immunoglobulins, we first examined B-1 cell, the innate-like B cell, which can produce natural

IgM, and IgA antibodies. We found that B-1 cells were not increased in the peritoneal cavity or spleen in the Vav1ddx3x mice (Supplementary figure 3). Next we checked which cells in the spleen express IgG by flow cytometry to initiate the investigation for higher serum Ig levels. The percentage of cells expressing IgG1 and IgG2b increased in the Vav1ddx3x mice whereas the number of cells expressing IgG1 and IgG2b did not change (Figure 5a, 5b, 5c, 5d). This suggests that in the Vav1ddx3x mice, besides the cells that express IgG1 and IgG2b as in the WT mice, some other cell types might also be contributing to the expression of IgG1 and IgG2b.

Indeed, when we back-gated IgG1 and IgG2b into B cell subsets: FOB and MZB, we found

74 surprisingly, that MZB in the Vav1ddx3x mice expressed high levels of IgG2b and IgG1 (Figure

5e, 5f, 5g, 5h, 5i). In WT mice, it was FOB that expresses more IgG1 and IgG2b. MZB are innate-like B cells, and their BCR usually doesn’t class-switch to IgG, IgA, or IgE. With MZB expressing IgG1/2b, these cells might be dysregulated, and they gained the ability to class-switch in the Vav1ddx3x mice. Accordingly, the number of IgG1/2b-expressing B cells did not change in the Vav1ddx3x mice. However, the serum IgG1/2b level was higher than in the WT mice.

Moreover, the MZB acquired this unusual function of producing IgG1/2b in the Vav1ddx3x mice.

These results suggest that the MZB might have a higher capability to produce IgG1/2b resulting in the elevated serum IgG level.

4. Summary

In the Vav1ddx3x mice, we previously observed a B cell deficit phenotype in spleen and bone marrow. Using the Hardy classification system to distinguish different B cell development stages in bone marrow, we found that Ddx3x deficiency caused a reduction in B cell numbers at early pro-B, late pre-B and immature B stages in bone marrow. Continuing to the periphery, we found that transitional B cell, follicular B cell, germinal center B cell and plasmablast numbers decreased in the Vav1ddx3x mice, whereas the marginal zone B cells number increased in the spleen. In the mixed bone marrow chimera we generated, with the B cell population heavily skewed to WT B cells in both bone marrow and spleen, we still observed the B cell deficit phenotype in late pre-B, immature B and mature B cells in bone marrow as well as in FOB in the spleen. We conclude that the B cell deficit phenotype caused by Ddx3x deficiency is an intrinsic effect.

We also measured B cell proliferation rate by BrdU incorporation 14 hours after in vivo injection.

We found the proliferation rate significantly decreased at the late pro-B and early pre-B stages during B cell development in the Vav1ddx3x mice. This lowered proliferation rate at these two stages can explain the B cell number reduction in the next stages—namely late pre-B, immature

B and mature B cells. However, the proliferation rate at or before the early pro-B cell stage did not change, suggesting the early pro-B cell loss might due to alternative mechanism (s). Early pro-B cell stage is an important checkpoint when B cells rearrange the heavy chain D-J genes of their BCRs. Perhaps Ddx3x affects heavy chain rearrangment leading to less early pro-B cells in the Vav1ddx3x mice.

Beside cellular phenotypes in the Vav1ddx3x mice, we found that IgA, IgM, and IgG levels in serum at baseline were elevated, even with fewer plasmablasts and GC-B cell numbers. Within

IgG, IgG1, IgG2b, and IgG3, levels were increased in the Vav1ddx3x mice. We were surprised to identify MZBs in the spleen of Vav1ddx3x mice, expressing IgG2b and IgG1 on their cell surfaces.

This result suggests that marginal zone B cell might have an increased capability to secrete IgG.

Our data is the first piece of evidence that linked this X chromosome gene: Ddx3x, to B-cell development and function.

Tables and Figures

Figure 1. The deficiency of Ddx3x affects B cell development and differentiation.

(a) Representative contour plots from flow cytometric analyses of B cell development using the

Hardy classification in bone marrow (first two columns) and B cell differentiation in spleen

(third column) in WT versus Vav1ddx3x mice. (b) Absolute number of B cell stages during development in bone marrow. (c) Absolute number of transitional 1, 2, and 3 B cells in spleen in

WT versus Vav1ddx3x mice. Transitional B cells in the spleen were gated on CD19+ CD93+, T1:

IgM+CD23–, T2: IgM+CD23+, T3: IgM–CD23+. (d) Absolute number of FOB and MZB in WT versus Vav1ddx3x mice. ***p ≤ 0.001, ****p ≤ 0.0001.

b) 1  1 0 7 *** W T **** V a v 1 d d x 3 x

r **** e 6

b 1  1 0

m ****

u n 5

e 1  1 0

t u l T 3 % _ s p le e n o T 1 % _ s p le e n T 2 % _ s p le e n s 1 1 0 4 b 

4 5 % 3 5 % 3 0 % **** A

** **

3 3

9 9

140 %1 0 9

+ +

3 ) ) )

) 3 ) ) C C 3 0 % ) C 2 0 % B 3 2 B B B B B B

+ + - - + - - - 2 o o e e e

D o e

9 9 r 9 r r D r r

r r u D

1 1 P P P u 1 C P P t 3 5 % - C t

y y a a C +

e - D

e D e D

l + l r r t r t m

P a a (M

C C

a C M a M

( M

L L m

E ( E ( I F

f g ( ( ( g

A g 2 5 % r 1 0 %

I I

r 'I o C o D

B r r E F o

F r C r

t t 3 0 % F r F t

F F F

o o o

2 5 % 2 0 % 0 % d d x 3 x d d x 3 x d d x 3 x W T V a v 1 W T V a v 1 W T V a v 1 c)

T 1 # _ s p le e n T 2 # _ s p le e n T 3 # _ s p le e n

3  1 0 6 3  1 0 6 2 .0  1 0 6

**** **** ****

r r

e e

2  1 0 6 2  1 0 6 e

b b

b 1 .5  1 0 6

m m

u u

2  1 0 6 2  1 0 6 u

n n

e e

e 1 .0  1 0

t t t

6 6

u u

1  1 0 1  1 0 u

l l

o o o

s s s 5 .0 1 0

5 5 

b b

5  1 0 5  1 0 b

A A A

0 0 0 d d x 3 x d d x 3 x d d x 3 x W T V a v 1 W T V a v 1 W T V a v 1 d)

4 .0  1 0 7 ****

r W T e d d x 3 x b V a v 1

2 .0  1 0 7 m 6 1 0 6

u  ***

e 6

t 4  1 0

l o 6

s 2  1 0

b A 0 F O B M Z B

Figure 2. Evidence that Ddx3x plays an intrinsic role in B cells.

(a) Schematics of bone marrow transplants. Lethally irradiated recipient mice received mixed 2.5 million bone marrow cells from WT CD45.1 mice and Vav1ddx3x CD45.2 mice at 1:4 ratio. (b)

Percentage of bone marrow CD45+ cells that are CD45.1+ or CD45.2+ 14 weeks after reconstitution. (c) Percentage of BM late pre-B, immature B, and mature B cells that are

CD45.1+ or CD45.2+. (d) Percentage of splenic follicular B cells (FOB) and marginal zone B cells (MZB) that are CD45.1+ or CD45.2+. (e) Percentage of BM CD45.1+ and CD45.2+ cells that are late pre-B, immature B or mature B cells. (f) Percentage of splenic CD45.1+ and CD45.2+ cells that are FOB or MZB. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001.

a) b) 8 0 %

n 4

i 6 0 %

4 0 %

n e

D 2 0 %

o P 0 % d d x3 x W T V a v 1

*** **** **** c) 1 0 0 %

+ W T d o n o r c e lls

. a

B d d x 3 x

h V a v 1 d o n o r c e lls t

4 8 0 %

f 6 0 %

. g

o 4 0 %

C c

c 2 0 %

P a 0 % L a rg e p re -B Im m a tu re B M a tu re B

****

c ****

d) i 1 0 0 %

t 2

e W T d o n o r c e lls

5 h

p d d x 3 x t

4 8 0 % s

V a v 1 d o n o r c e lls

t r

a 6 0 %

. o

e 4 0 %

n C

c 2 0 %

a e

P 0 % F O B M Z B

e 2 5 % ****

1 r

e) . B

a W T d o n o r c e lls

t 4

e 2 0 % s

a d d x 3 x

l D

l V a v 1 d o n o r c e lls

1 5 %

u 1 0 %

B e

. ****

a 4

r ***

r n

5 %

e a P 0 % L a rg e p re -B Im m a tu re B M a tu re B

f) *

s 6 0 % l

l W T d o n o r c e lls

n 5 0 %

Z d d x 3 x +

e V a v 1 d o n o r c e lls

p r

5 4 0 %

o B

3 0 %

g 2 0 %

1 c

. 1 0 %

P D

C 0 % F O B M Z B

Figure 3. Ddx3x deficiency affects B cell proliferation during development.

Mice were injected with 2mg of BrdU and after 14 hours were stained with BrdU antibody for flow. (a) Representative histograms from flow cytometry analysis of B cell progenitors that positive for BrdU. (b) Average frequencies of BrdU+ B cell progenitors. **p ≤

0.01, ****p ≤ 0.0001.

Figure 4. Serum immunoglobulin levels in Vav1ddx3x mice.

Baseline serum levels of (a) IgA, (b) IgG, (c) IgM, (d) IgE, (e) IgG1, (f) IgG2a, (g) IgG2b and (h)

IgG3 in WT and Vav1ddx3x mice. *p ≤ 0.05, ****p ≤ 0.0001.

a) b)

c) d)

e) f)

82 g) h)

Figure 5. Ddx3x deficiency alters MZB function. a) Percentage and (b) absolute number of IgG1-expressing cells within B cell in the spleen. (c)

Percentage and (d) absolute number of IgG2b-expressing cells within B cells in the spleen. (e)

Representative dot plot showing cells that express IgG2b (yellow dot) backgate to FOB and

MZB (blue dot showing B cell). (f) Percentage and (g) absolute number of FOB and MZB in

IgG1-expressing B cells. (h) Percentage and (i) absolute number of FOB and MZB in IgG2b- expressing B cells. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

a) b)

c) d)

84 e)

f) g)

h) i)

Supplementary tables and figures

Supplementary table 1.

List of flow cytometry antibodies used.

Antigen Fluorochrome Company Catalog# Anti-MHC Class II (I-A/I-E) Alexa Fluor® 700 eBioscience 56-5321-82 B220 AF700 Biolegend 13101 B220 APC TONBO Biosciences 20-0452-U025 B220 BV650 BioLegend 103241 BP-1 PE BD Biosciences 553735 BrdU eF450 eBioscience 8848-6600-42 CD11b BV650 Biolegend 101239 CD11b Pacific Blue BioLegend 101224 CD11c APC eBioscience 17-0114-82 CD138 BV421 Biolegend 142507 CD19 APC-Cy7 Biolegend 115529 CD19 FITC BioLegend 115505 CD19 PerCP/Cy5.5 eBioscience 45-0193-82 CD21 APC/Cy7 Biolegend 123417 CD21 eF450 ebioscience 48-0212-80 CD21/35 PE Biolegend 123409 CD23 FITC Biolegend 101605 CD23 PE/Cy7 BioLegend 101613 CD24 (HSA) APC Biolegend 138505 CD3 FITC BioLegend 100306 CD335 (NKp46) PE eBioscience 12-3351-82 CD3e V500 BD Bioscience 560771 CD43 BV605 BD Biosciences 563205 CD45.1 PerCP/Cy5.5 Biolegend 110727 CD45.2 AF700 Biolegend 109821 CD5 APC BD Bioscience 561895 CD93 (AA4.1) APC Biolegend 136509 F4/80 PerCP BioLegend 123126 FasR APC-R700 BD 565130 FasR PE BD Bioscience 554258 GL-7 FITC BD Biosciences 553666 Gr-1 BV650 Biolegend 108441 Gr-1 Pacific Blue Biolegend 108429 IgD AF700 BioLegend 405730

IgD PE/Cy7 BioLegend 405719 IgD PerCP/Cy5.5 Biolegend 405709 IgG1 APC Biolegend 406609 IgG2b PE Biolegend 406707 IgG3 BV711 BD 565809 IgM APC-Cy7 Biolegend 406515 IgM PE Biolegend 406507 IgM PE-Cy7 eBioscience 25-5790-82 LIVE/DEAD® Fixable Stain Indo-1 Life technology L-34962 Ly6c FITC Biolegend 128005 Ter119 FITC Biolegend 116205

Supplementary figure 1. Plasmablasts and germinal center (GC) B cells in Vav1ddx3x mice.

Absolute number of (a) Plasmablasts and (b) GC-B cells in the spleen of WT versus Vav1ddx3x mice. Plasmablasts were gated as CD19mediumCD138+. GC-B cells were gated as CD19+GL-

7+Fas+. **p ≤ 0.01, ***p ≤ 0.001.

a) b)

Supplementary figure 2. Serum BAFF levels in Vav1ddx3x mice.

Serum BAFF levels in WT and Vav1ddx3x mice. Two-tailed t-test was performed. **p ≤ 0.01.

B A F F in s e ru m

3  1 0 4 **

2  1 0 4

l 2  1 0 4

/ g 4

p 1  1 0

5  1 0 3

0 d d x3 x W T V a v 1

Supplementary figure 3. B-1 cells in Vav1ddx3x mice.

Absolute number of B-1a and B-1b cells in (a) spleen and (b) peritoneal cavity in WT versus

Vav1ddx3x mice. B-1a cells were gated as CD19+CD43+CD23-IgMhiCD5+. B-1b cells were gated as CD19+CD43+CD23-IgMhiCD5+. * p ≤ 0.05.

Chapter 5: Summary and Discussion

Autoimmune diseases are characterized by female predominance, which despite several hypotheses for different diseases having been suggested, remain unexplained with no clearly identified mechanism.

In Chapter II, we tested a prediction of the X chromosome dosage effect hypothesis. Men with

Klinefelter's syndrome in SLE have been shown to have increased risk for SLE, predicted to be at the same level as the female risk[76, 77]. Accordingly, we predicted that with more X chromosomes, there will be greater risk for autoimmune diseases. We tested systemic lupus erythematosus (SLE) and other autoimmune diseases with data available to us: primary Sjögren's syndrome (pSS), primary biliary cirrhosis (PBC) and rheumatoid arthritis (RA) as well as an inflammatory disease, sarcoidosis, which has no female sex bias that we used as a diseases control.

We studied SLE and pSS women and found 47,XXX was enriched. We identified seven 47,XXX out of 2826 SLE patients, and three 47,XXX out of 1033 pSS patients, whereas in controls, there were only two 47,XXX out of 7074 controls. By Bayes’ theorem, the calculated risk for 47,XXX of getting SLE and pSS, increased by ~25 fold and ~41 fold compared to 46,XY men. This risk effect greatly exceeded of all other known genetic risk factors for SLE or pSS. The importance of these findings is not for the few individuals with 47,XXX, but rather lies in the fact that rare events and phenotypes reveal insights into the mechanism for the general disease circumstances.

Everyone has an X chromosome. And increased 47,XXX among women with either SLE or pSS informs the potential mechanism underpinning the disparate risk of these diseases found for men

91 and women with a normal sex chromosome complement. Because sexual development and sex hormones are normal in 47,XXX women, these data suggest that the number of X chromosomes is a key factor impacting the 10-fold risk difference between men and women.

However, we did not find excess 47,XXX among women with the other autoimmune diseases studied here (PBC and RA). While it is possible that we had inadequate sample size for PBC and RA, the relationship of Turner’s syndrome (45,X) to autoimmunity also suggests heterogeneity in the mechanisms of the sex bias found in autoimmune diseases. Autoimmune thyroiditis, type 1 diabetes mellitus, and celiac disease are found in excess among Turner’s syndrome (45,X) patients[158]. On the other hand, there is no evidence that prevalence of SLE or pSS is increased in Turner’s syndrome. While rarer than the Turner’s-associated diseases, the paucity of Turner’s syndrome among SLE patients [78, 142] suggests that SLE risk in these patients is more similar to 46,XY males rather than to 46,XX females. Moreover, PBC, autoimmune thyroid disease (AITD) and systemic sclerosis have increased acquired X- monosomy, but SLE does not[67, 75, 137], also consistent with the possibility of multiple mechanisms for sex bias in autoimmunity.

Because 47,XXX is a rare event, in order to get a sufficiently large sample size, we included non-affected SLE family controls, non-auto-inflammatory subjects, population controls and non- sarcoidosis controls in our study. The 2090 non-affected SLE family controls may contain subjects with pSS; we showed that when comparing 47,XXX enrichment in pSS to only those with the healthy non-auto-inflammatory controls, the result is still significant (p=0.02). The 2680 population controls and 620 non-sarcoidosis controls may contain subjects with SLE or pSS. We showed that even when removing these control groups, our 47,XXX enrichment in SLE remained significant, if not more so (p=0.001, p=0.002, respectively).

The literature reported 47,XXX prevalence is 1/1,000 female live births, whereas, in our controls, the rate is 1/3,573. One possible phenotype associated with 47,XXX is anxiety, shyness and low self-esteem (18), a potential source of bias as 47,XXX subjects may be less willing to participate in research studies. This raises the question of whether 47,XXX would be less willing to participate in our study. This personality trait of 47,XXX women would lead to a non-differential error that would equally affect both case and control group recruitment. The great majority (90%) of the 47,XXX are unrecognized (14). Since they only rarely exhibit a clear phenotype that would bring them to an evaluation.

We also compared our rate of 47,XXX in SLE, pSS, RA, PBC, and sarcoidosis with a published unselected newborn infant 47,XXX screening dataset (34). The newborn infant dataset has no bias in recruitment since consecutive births were studied. Even compared to this dataset, there is a significant enrichment of 47,XXX in SLE and a trend for 47,XXX enrichment in pSS, which supports our hypothesis of X chromosome does effect in SLE and pSS (p=0.04, p=0.09, respectively). Moreover, more than half of both our cases and controls are people with African

American ancestry, where no report of 47,XXX prevalence has ever been assembled. The

47,XXX 1/1,000 prevalence were generated majorly by birth health screening in Scandinavia where the majority of the participants are European ancestry. Perhaps 47,XXX has a different prevalence with different ancestries.

If SLE and 47,XXX both occur at 1 in ~1000 and were independent, then only 1 in a million women would have both SLE and 47,XXX. Our data suggest that both are coincident at 1 in

~40,000 women. The lupus families provide an interesting insight into how these risks may be modified. There are seven women with SLE and 47,XXX in our sample of 2826 SLE cases, but no primary female relatives of lupus patients with 47,XXX of the 2090 available to us. That this

93 difference is significant (p=0.02 by Fisher’s exact test with OR≥9.6) suggests that 47,XXX is a strong contributor to SLE when present within the family milieu of other genetics and environmental factors that establish risk for SLE.

Sex hormones have long been considered a candidate mechanism of gender bias. However, several observations argue against sex hormones being a major contributing factor to the gender bias observed in SLE. First, we previously reported Klinefelter’s men (47,XXY) have a similar prevalence of SLE as women[76], while in this study, 47,XXX women, who do not produce abnormal sex hormones[115, 159], have excess risk for SLE and pSS. In addition, four core genotype mice (XXfemale, XYfemale, XX male and XX female) display an X chromosome dose effect for susceptibility in lupus in mouse models[111, 112]. The mechanism by which the yaa mouse develops lupus susceptibility[146] is another example of an X chromosome gene dose effect that is unrelated to sex hormones, but instead to an X-Y chromosome translocation.

Furthermore, there is no difference in the levels of sex hormones between SLE patients at disease onset and healthy controls[160]. Other data suggest that sex hormone levels in men with SLE are abnormal, but similar to those found in men with other non-female-biased chronic illnesses[161].

The female bias in SLE also exists before puberty with girls being five times more likely to develop SLE than boys[162]. Similarly, the sex bias in SLE is unaffected by the lower estrogen levels after menopause, and is not statistically different from those with disease onset in the reproductive years[163]. These observations tend to bolster the importance of the X chromosome dose contributing to sex bias in SLE.

The ~100-fold increase in estrogens during pregnancy differentially affects various autoimmune diseases. There is no evidence, for example, for meaningful effects of pregnancy in SLE[164]. In some autoimmune diseases, particularly RA, a suppressive effect is observed in about two thirds

94 of RA patients during pregnancy[164]. Oral estrogens for female contraceptive or post- menopausal replacement also have different effects on different autoimmune diseases; conflicting data exist for SLE and RA. Estrogen therapy is generally safe in women with SLE and RA[165, 166], but, on the other hand, there are data supporting an increased risk of SLE and

RA among women who have taken oral contraception[167, 168]. That we did not find increased

47,XXX among patients with PBC and RA (with the current sample size available to us), both of which have a female bias, again consistent with gender bias in autoimmunity operating through multiple mechanisms. However, the sex ratio in the latter disease is not as extremely biased toward women as SLE and pSS (10/1 and 14/1 in SLE and pSS, respectively versus 2~6/1 in

RA). This distinction may suggest different mechanisms of gender bias or potentially additive mechanisms in SLE and pSS.

To conclude, the available data strongly support a genetic cause of the powerful female bias in

SLE and pSS being located on the X chromosome, and this mechanism appears to be at least potentially independent of circulating sex hormones. Our new data demonstrating increased risk of SLE and pSS among 47,XXX women indicate that the female bias may be explained, at least in part, by an X chromosome gene dose effect. Identifying the specific properties of the X chromosome responsible for the increase in SLE and pSS risk will be critical to advancing the understanding of these disorders.

In Chapter III, we set out to explore what elements (s) on X chromosome are responsible for SLE female sex bias. Since the X chromosome dosage effect was observed in both human and lupus mouse models [12, 112], we applied the assumption that female bias in humans and mice had the same basic mechanism and narrowed our candidate list by choosing genes escape X chromosome

95 inactivation (XCI) in both human and mice. Those genes express at a higher expression level in females than males could possibly conferfemale preponderance in SLE. [81, 83, 84]. Within the final list of 8 genes escaping XCI in human and mice, we choose an RNA helicase gene:

DDX3X/Ddx3x, which promotes type I IFN signaling, as our first candidate to study, as type I

IFN is critical to SLE pathogenesis.

Our collaborator provided us the Ddx3x floxed mice; we used these conditional knockout mice as our working model. We first bred them with EIIa-cre mice, but could not obtain female homozygous or male hemizygous knockout mice. Consequently, we bred Ddx3x floxed mice with Vav1-Cre mice to delete Ddx3x in all the hematopoietic cells, with the assumption that the role of Ddx3x in the immune cell should be critical to its function in type I IFN and female lupus bias. Still, we cannot obtain female homozygous knockout mice. However, we obtained male hemizygous knockout mice. When we measured Ddx3x expression level, to our surprise, we found no differential expression level of Ddx3x from leukocytes in bone marrow among female

WT, female heterozygous and male WT mice. In addition, in spleen from B6 female and male

WT mice, the Ddx3x expression level was similar. In tissues other than spleen and BM, we found

Ddx3x expression higher in white adipose tissue and heart of females than of males, but not in white blood cells, muscle or liver. However, the male Ddx3x Vav1 hemizygous mice (Vav1ddx3x) have a reduction of Ddx3x mRNA expression for 80% in leukocytes from the bone marrow and

60% in leukocytes from the spleen.

As the Ddx3x floxed mouse is the first in vivo Ddx3x conditional knockout mice ever made, lots of the phenotypes were novel to us. These results firstly suggest that ddx3x is critical to embryogenesis, as we did not obtain any female or male Ddx3x-EIIa knockout mice. This suggestion is consistent with the work of two independent groups who show Ddx3x plays a

96 major role in cell survival and cell cycle control of early embryogenesis, and extraembryonic development [155, 156]. Li et al. showed that decreaed level of Ddx3x led to less viable blastocyst number and the cell cycle arrested between 2-cell to 4-cell stage [155]. Chen et al. suggested that Ddx3x deficiency causes genome damage with widespread apoptosis and abnormal cell growth in the epiblast [156]. Our data and the two published reports all support the role of Ddx3x in embryogenesis, although the underlying mechanism remains unclear.

However, when we bred Ddx3x floxed mice with Vav1-cre mice, we still did not obtain female homozygous knockout mice but male hemizygous knockout mice were viable. It seems like the role of Ddx3x in hematopoietic system is vital to early embryo survival as well, but how do male hemizygous male mice survive? Ddx3x has a homologous gene on the Y chromosome, namely ddx3y, which was reported to be important in sperm genesis, and is only translated into protein in the testis[86-88]. Perhaps, ddx3y compensates for the role of Ddx3x in males so that hemizygous male mice (Vav1ddx3x) were viable. There would be some amount of ddx3y protein transferred to zygote cytosol along with the sperm, and this small amount of protein might be critical for early survivial of the male.

In theory, although we did not obtain any female homozygous mice, we should have three tiers of Ddx3x expression levels with the Ddx3x-Vav1 conditional knockout mice system, with female

WT expressing the highest amount, then in the middle female heterozygous mice and male WT mice as they only have one copy of Ddx3x, and male Ddx3x hemizygous mice express the lowest amount. With this system, we could still explore whether more expression of Ddx3x would confer greater disease susceptibility. The fact that we did not observed differential levels of

Ddx3x in these three tiers of mice in relevant immune tissues (spleen and bone marrow), and that we did measure a higher level of Ddx3x in white adipose tissue and heart, but not in blood, liver

97 or muscle in female WT mice compared to male WT mice suggests that Ddx3x escapes from

XCI is tissue specific. In public literature, Yang et al and Chen et al showed Ddx3x expression is higher in females in liver, adipose, muscle, heart and brain tissues using different mouse strains

[84, 150]. As our Ddx3x expression results between female and male mice in liver, and muscle are not in agreement with theirs. Perhaps, Ddx3x escape from XCI is dependent on the genetic background. It would require researchers to test more tissues and mouse strains to establish whether this idea is true or not.

Although ideally we would prefer to use a “gain of function” model to test if increased Ddx3x leads to more lupus susceptibility, we were left with the choice of using the male hemizygous knockout (Vav1ddx3x) mouse as our working model. In the Vav1ddx3x mice, we observed an 80% reduction of Ddx3x expression level in bone marrow and 60% reduction in spleen compared to male WT mice, providing us a model of decreased dose of Ddx3x. As the efficiency of lots of loxP and cre system is not 100%, this possibly cause the deletion of Ddx3x not being 100% in bone marrow and spleen. In addition, the efficiency can vary from tissue to tissue. In the spleen,

Ddx3x reduction was 60%. Using Vav1ddx3x mice, we thought we can still ask, the reverse of our original hypothesis, if the reduction of Ddx3x expression would lead to the decreased type I IFN production and less lupus phenotype. However, when we measured IFN-β production in vitro and in vivo by poly I:C stimulation, we found inconsistent results. By in vitro stimulation of

BMDCs generated from Vav1ddx3x mice, we measured more IFN-β production. Whereas in vivo stimulation of poly I:C resulted in less production of serum IFN-β in the Vav1ddx3x mice. As these results are contradictory, we examined the Vav1ddx3x mice carefully for if any other confounding factors were creating the phenotypes we saw. We found B cell, NK cell, and T cell numbers decreased in the spleen of these mice. In bone marrow, Vav1ddx3x mice had an increased

98 number of Ter119+ cells, CD11b+ cells and decreased B220+ cells. The B linage deficit phenotype was also observed in inguinal lymph nodes.

With these cells deficit phenotypes, Vav1ddx3x mice cannot serve as an appropriate model to study type I IFN or SLE female sex bias, as it cannot be distinguished the possibilities of whether the cell deficit phenotype or the ddx3x function lead to the phenotype. However, the B cell deficit phenotype was of particular interest to us since B cell is central to the lupus phenotype, by being the cell type that produces autoantibodies and presents autoantigens. In SLE patients, there is a B cell population shift with more immature and naïve B cell expansion. CD27+ plasmablasts are also increased which correlates with autoantibodies production and tissue damage[57]. Most of the autoantibodies produced in SLE are IgG and somatically mutated[58]. Therefore, this novel finding that deficiency of Ddx3x leads to fewer B cells suggests insight into B cell development may emerge from this study. We then explored the underlying mechanism of why

Ddx3x deficiency affects B cell number and investigate if B cell function is affected.

In Chapter IV, we revealed that the deficiency of Ddx3x in the Vav1ddx3x mice results in a B cell deficit phenotype early during B cell lymphopoiesis in the bone marrow compartment, that early pro-B, late pre-B and immature B cell number is decreased. When immature B cells migrate to the periphery, transitional B cell numbers continued to decline. When B cells differentiate, there were a significantly decreased number of transitional B cells, follicular B cells, plasmablasts, and germinal center B cells, but an increased number of marginal zone B cells in the Vav1ddx3x mice.

Mixed bone marrow chimera experiment demonstrated that the late pre-B, immature and FOB deficit phenotype was B cell intrinsic whereas the marginal zone B cell increase was B cell extrinsic that it requires the participation of other cells to help. Preliminary data of BrdU

99 incorporation assay showed B cells proliferate slower at late pro-B and early pre-B stage during development, which could be one of the reasons to explain B cell number decreased for the stages subsequently. Serum B cell activating factor (BAFF) level was increased in the Vav1ddx3x mice which might promote B cell survival. Although B cell number decreased in general in the

Vav1ddx3x mice, including plasmablasts and germinal center B cells, we measured an increased level of IgA, IgM and IgG (1/2b/3) at baseline without simulation. We were surprised to find that

MZB cells in the Vav1ddx3x mice can abnormally class switch to from IgM to IgG and might have an enhanced ability to produce immunoglobulins, especially IgG2b.

We demonstrated that Ddx3x deficiency leads to a general B cell loss during development in the bone marrow and differentiation in the periphery. Moreover, this effect of Ddx3x deficiency seems to affect B2 cells the most, since we did not observe changes for the B1 cells in the spleen or peritoneal cavity. Also the mixed bone marrow chimera experiment further demonstrates that the B cell deficit phenotype is B cell intrinsic and does not require help from other cell types.

The fact that when we first mixed the bone marrow together at 1:1 ratio to generate mixed bone marrow chimera mice, we saw 20 fold or even more cells from the WT donor in the mixed bone marrow chimera further suggesting a profound intrinsic defects in the Vav1ddx3x cells donor cells which cannot compete with WT cell in the same environment (data not shown).

B cell development in bone marrow is a step-by-step process characterized by proliferation and heavy and light chain rearrangement for B cell receptor. The formation of pre-B cell receptor

(pre-BCR) followed by BCR formation with normal signaling being required for proper B cell development and function. There are several possibilities for how Ddx3x deficiency confers the

B cell deficit phenotype. Our preliminary data showed that late pre-B cells and early pre-B cells proliferate slower in the Vav1ddx3x mice, which can explain B cell deficit phenotype for the

100 subsequent stages: the late pre-B, immature B cells in the bone marrow, transitional B cells and

FOB cells in the periphery. We can not, however, conclude this is the only mechanism. For the decreased cell numbers observed at the early pro-B stage, since theproliferation rate at this stage or before it did not change, a decreased proliferation rate is not an explanation for the pro-B deficit., This suggests multiple mechanisms might be synergistically or additively playing a role for B cell number loss during development in bone marrow. As early pro-B cell stage is an important development checkpoint when B cells have finished their D-J gene rearrangement for heavy chain of pre-BCR, it is possible that Ddx3x affects heavy chain rearrangement or function causing less B cells at this stage. Other possibilities are, in the Vav1ddx3x mice, there might be fewer numbers of lineage progenitors before B linages are initiated, such as with the common lymphoid progenitor cell (CLP), or even before that, at the multi-pluripotent progenitor cell

(MPP), or hematopoietic stem cell levels. Alternatively, B cell in the Vav1ddx3x might be pro- apoptotic and have shorter survival time than WT B cells.

As a DEAD box RNA helicase, how does Ddx3x regulate B cell development? Perhaps, shortage of Ddx3x in the Vav1ddx3x mice affects RNA level of key genes during B cell development, regulating transportation, unwinding, and splicing role or change the transcriptional or translational activity of those critical genes. One possible explanation would be, DDX3 has been shown to participate in Wnt–β-catenin signaling pathway, that ddx3 is a regulatory subunit of

CK1e (casein kinase 1 e)[169]. Since Wnt signaling has been suggested to have a mitogenic role in early B cell development[170]. Perhaps, Ddx3x regulates B cell development through Wnt signaling pathway.

Although Vav1ddx3x had a decreased B cell numbers, the serum IgA, IgG, and IgM level at baseline were increased in the Vav1ddx3x mice. For IgG, the IgG1, IgG2b, and IgG3 levels were

101 increased. We did not find increased number of B-1 cells in the Vav1ddx3x mice unless their Ig production functions were promoted; B-1 cells were less likely to be responsible for producing more IgA and IgM in the Vav1ddx3x mice. When we stained for IgG expressing cells in the spleen, we were surprised to find that MZB expressed these IgGs, especially IgG2b, which is most abundant. Somehow, these innate-like B cells, which are known not to class switch, were functionally changed in the Vav1ddx3x mice; they might have greater capacity to make these IgGs leading to an increased level in serum. Since the phenotype of increased MZB were shown to be extrinsic in the mixed bone marrow chimera experiment, the increased IgG levels were possibly an B cell extrinsic effect as well. That some other cells might also be deficient for Ddx3x, helped

MZB alter its function, though how Ddx3x regulates MZB remains unknown.

The serum BAFF level was also increased in the Vav1ddx3x mice. The Vav1ddx3x mice, like the

μMT mice, which lack mature B cell and have an increased amount of BAFF in serum. With fewer B cells, these BAFF proteins might be excessive and probably are more than enough to bind to every B cell, resulting in higher levels in serum. BAFF is produced by macrophages, monocytes, DCs. As we observed an increased number of CD11b+ cells in the bone marrow, the increased level of BAFF may also be due to the number of its generators are increased. The increased level of BAFF, could also in turn, promote B cell survival and induce immunoglobulin production.

In conclusion, the Ddx3x-Vav1 conditional knockout mouse system was not appropriate for us to assess the question about whether Ddx3x is accountable for SLE female preponderance.

DDX3X/Ddx3x could still a right candidate gene accounting for SLE female sex bias. Although it did not escape XCI in spleen and bone marrow on B6 background mice, in human it is

102 expressed more in female PBMCs than in male PBMCs (data not shown). Based on our experience learned in this study, X chromosome inactivation is tissue specific and could be different in human and mice. Perhaps, when narrowing the candidate genes to investigate SLE female sex bias, one should not overlap the genes escaping XCI in both human and mice, as inbred mouse genetic background is far simpler than outbred human and the profiles of XCI escapees are quite different. The many RNA sequencing databases now available with profiling results in female and male samples may provide the capacity to evaluate gene expression differences that follow an X chromosome gene dose pattern across many individual human and mouse strain. Also choosing the optimal mouse model to work with is critical as well, generally, engineering a knock in model which creates more expression of the candidate gene would be the best suitable model to use to study increased X gene dosage hypothesis for SLE sex bias.

Although our approach and data does not approve or disapprove, whether DDX3X/ddx3x confers

SLE female preponderance, however, with the male hematopoietic deficient (Vav1ddx3x) mice, we, for the first time, showed that Ddx3x plays a role in B cell development ad function. Ddx3x deficiency in the hematopoietic system led to decreased proliferation rate of B cells and a reduced number of B cell in the bone marrow and the spleen, and resulted in somewhat paradoxical elevated basal immunoglobulin levels. These findings highlight the importance of

Ddx3x in B cell development and function.

103

Bibliography 1. Whitacre, C.C., Sex differences in autoimmune disease. Nat Immunol, 2001. 2(9): p. 777- 80. 2. National Institutes of Health Autoimmune Disease Coordinating Committee Report. 2002. 3. Association, A.A.R.D. Autoimmune Statistics. 2016 6/8/16; Available from: http://www.aarda.org/autoimmune-information/autoimmune-statistics/. 4. Tsokos, G.C., Systemic lupus erythematosus. N Engl J Med, 2011. 365(22): p. 2110-21. 5. Pons-Estel, G.J., et al., Understanding the epidemiology and progression of systemic lupus erythematosus. Semin Arthritis Rheum, 2010. 39(4): p. 257-68. 6. Aletaha, D., et al., 2010 Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum, 2010. 62(9): p. 2569-81. 7. Perry, D., et al., Murine models of systemic lupus erythematosus. J Biomed Biotechnol, 2011. 2011: p. 271694. 8. Theofilopoulos, A.N. and F.J. Dixon, Murine models of systemic lupus erythematosus. Adv Immunol, 1985. 37: p. 269-390. 9. Satoh, M. and W.H. Reeves, Induction of lupus-associated autoantibodies in BALB/c mice by intraperitoneal injection of pristane. J Exp Med, 1994. 180(6): p. 2341-6. 10. Satoh, M., et al., Anti-nuclear antibody production and immune-complex glomerulonephritis in BALB/c mice treated with pristane. Proc Natl Acad Sci U S A, 1995. 92(24): p. 10934-8. 11. Satoh, M., et al., Widespread susceptibility among inbred mouse strains to the induction of lupus autoantibodies by pristane. Clin Exp Immunol, 2000. 121(2): p. 399-405. 12. Smith, D.L., et al., A female preponderance for chemically induced lupus in SJL/J mice. Clin Immunol, 2007. 122(1): p. 101-7. 13. Morris, S.C., P.L. Cohen, and R.A. Eisenberg, Experimental induction of systemic lupus erythematosus by recognition of foreign Ia. Clin Immunol Immunopathol, 1990. 57(2): p. 263-73. 14. Alarcon-Segovia, D., et al., Familial aggregation of systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases in 1,177 lupus patients from the GLADEL cohort. Arthritis Rheum, 2005. 52(4): p. 1138-47. 15. Deapen, D., et al., A revised estimate of twin concordance in systemic lupus erythematosus. Arthritis Rheum, 1992. 35(3): p. 311-8. 16. Vaughn, S.E., et al., Genetic susceptibility to lupus: the biological basis of genetic risk found in B cell signaling pathways. J Leukoc Biol, 2012. 92(3): p. 577-91.

104

17. Blanchong, C.A., et al., Genetic, structural and functional diversities of human complement components C4A and C4B and their mouse homologues, Slp and C4. Int Immunopharmacol, 2001. 1(3): p. 365-92. 18. Niederer, H.A., et al., FcgammaRIIB, FcgammaRIIIB, and systemic lupus erythematosus. Ann N Y Acad Sci, 2010. 1183: p. 69-88. 19. Kelley, J., et al., Variation in the relative copy number of the TLR7 gene in patients with systemic lupus erythematosus and healthy control subjects. Arthritis Rheum, 2007. 56(10): p. 3375-8. 20. Moser, K.L., et al., Recent insights into the genetic basis of systemic lupus erythematosus. Genes Immun, 2009. 10(5): p. 373-9. 21. Tsokos, G.C. and G.M. Kammer, Molecular aberrations in human systemic lupus erythematosus. Mol Med Today, 2000. 6(11): p. 418-24. 22. Manderson, A.P., M. Botto, and M.J. Walport, The role of complement in the development of systemic lupus erythematosus. Annu Rev Immunol, 2004. 22: p. 431-56. 23. Harley, J.B., et al., The genetics of human systemic lupus erythematosus. Curr Opin Immunol, 1998. 10(6): p. 690-6. 24. Hall, J.C. and A. Rosen, Type I interferons: crucial participants in disease amplification in autoimmunity. Nat Rev Rheumatol, 2010. 6(1): p. 40-9. 25. Kotenko, S.V., et al., IFN-lambdas mediate antiviral protection through a distinct class II cytokine receptor complex. Nat Immunol, 2003. 4(1): p. 69-77. 26. Sheppard, P., et al., IL-28, IL-29 and their class II cytokine receptor IL-28R. Nat Immunol, 2003. 4(1): p. 63-8. 27. Ank, N., et al., Lambda interferon (IFN-lambda), a type III IFN, is induced by viruses and IFNs and displays potent antiviral activity against select virus infections in vivo. J Virol, 2006. 80(9): p. 4501-9. 28. Doyle, S.E., et al., Interleukin-29 uses a type 1 interferon-like program to promote antiviral responses in human hepatocytes. Hepatology, 2006. 44(4): p. 896-906. 29. Baechler, E.C., et al., Interferon-inducible gene expression signature in peripheral blood cells of patients with severe lupus. Proc Natl Acad Sci U S A, 2003. 100(5): p. 2610-5. 30. Gottenberg, J.E., et al., Activation of IFN pathways and plasmacytoid dendritic cell recruitment in target organs of primary Sjogren's syndrome. Proc Natl Acad Sci U S A, 2006. 103(8): p. 2770-5. 31. Niewold, T.B., et al., Association of the IRF5 risk haplotype with high serum interferon- alpha activity in systemic lupus erythematosus patients. Arthritis Rheum, 2008. 58(8): p. 2481-7. 32. Emamian, E.S., et al., Peripheral blood gene expression profiling in Sjogren's syndrome. Genes Immun, 2009. 10(4): p. 285-96. 33. Kirou, K.A., et al., Activation of the interferon-alpha pathway identifies a subgroup of systemic lupus erythematosus patients with distinct serologic features and active disease. Arthritis Rheum, 2005. 52(5): p. 1491-503.

105

34. Vallin, H., et al., Anti-double-stranded DNA antibodies and immunostimulatory plasmid DNA in combination mimic the endogenous IFN-alpha inducer in systemic lupus erythematosus. J Immunol, 1999. 163(11): p. 6306-13. 35. Bave, U., G.V. Alm, and L. Ronnblom, The combination of apoptotic U937 cells and lupus IgG is a potent IFN-alpha inducer. J Immunol, 2000. 165(6): p. 3519-26. 36. Ioannou, Y. and D.A. Isenberg, Current evidence for the induction of autoimmune rheumatic manifestations by cytokine therapy. Arthritis Rheum, 2000. 43(7): p. 1431-42. 37. Bennett, L., et al., Interferon and granulopoiesis signatures in systemic lupus erythematosus blood. J Exp Med, 2003. 197(6): p. 711-23. 38. Han, G.M., et al., Analysis of gene expression profiles in human systemic lupus erythematosus using oligonucleotide microarray. Genes Immun, 2003. 4(3): p. 177-86. 39. Kirou, K.A., et al., Coordinate overexpression of interferon-alpha-induced genes in systemic lupus erythematosus. Arthritis Rheum, 2004. 50(12): p. 3958-67. 40. Sigurdsson, S., et al., Polymorphisms in the tyrosine kinase 2 and interferon regulatory factor 5 genes are associated with systemic lupus erythematosus. Am J Hum Genet, 2005. 76(3): p. 528-37. 41. Remmers, E.F., et al., STAT4 and the risk of rheumatoid arthritis and systemic lupus erythematosus. N Engl J Med, 2007. 357(10): p. 977-86. 42. Turan, K., et al., Nuclear MxA proteins form a complex with influenza virus NP and inhibit the transcription of the engineered influenza virus genome. Nucleic Acids Res, 2004. 32(2): p. 643-52. 43. Harris, R.S. and M.T. Liddament, Retroviral restriction by APOBEC proteins. Nat Rev Immunol, 2004. 4(11): p. 868-77. 44. Ozato, K., et al., TRIM family proteins and their emerging roles in innate immunity. Nat Rev Immunol, 2008. 8(11): p. 849-60. 45. Strandberg, L., et al., Interferon-alpha induces up-regulation and nuclear translocation of the Ro52 autoantigen as detected by a panel of novel Ro52-specific monoclonal antibodies. J Clin Immunol, 2008. 28(3): p. 220-31. 46. Gallucci, S., M. Lolkema, and P. Matzinger, Natural adjuvants: endogenous activators of dendritic cells. Nat Med, 1999. 5(11): p. 1249-55. 47. Luft, T., et al., Type I IFNs enhance the terminal differentiation of dendritic cells. J Immunol, 1998. 161(4): p. 1947-53. 48. Le Bon, A., et al., Cross-priming of CD8+ T cells stimulated by virus-induced type I interferon. Nat Immunol, 2003. 4(10): p. 1009-15. 49. Biron, C.A., et al., Natural killer cells in antiviral defense: function and regulation by innate cytokines. Annu Rev Immunol, 1999. 17: p. 189-220. 50. Curtsinger, J.M., et al., Type I IFNs provide a third signal to CD8 T cells to stimulate clonal expansion and differentiation. J Immunol, 2005. 174(8): p. 4465-9.

106

51. Kolumam, G.A., et al., Type I interferons act directly on CD8 T cells to allow clonal expansion and memory formation in response to viral infection. J Exp Med, 2005. 202(5): p. 637-50. 52. Marrack, P., J. Kappler, and T. Mitchell, Type I interferons keep activated T cells alive. J Exp Med, 1999. 189(3): p. 521-30. 53. Andrews, B.S., et al., Spontaneous murine lupus-like syndromes. Clinical and immunopathological manifestations in several strains. J Exp Med, 1978. 148(5): p. 1198- 215. 54. LeBien, T.W. and T.F. Tedder, B lymphocytes: how they develop and function. Blood, 2008. 112(5): p. 1570-80. 55. Bonilla, F.A. and H.C. Oettgen, Adaptive immunity. J Allergy Clin Immunol, 2010. 125(2 Suppl 2): p. S33-40. 56. Fuxa, M. and J.A. Skok, Transcriptional regulation in early B cell development. Curr Opin Immunol, 2007. 19(2): p. 129-36. 57. Tobon, G.J., J.H. Izquierdo, and C.A. Canas, B lymphocytes: development, tolerance, and their role in autoimmunity-focus on systemic lupus erythematosus. Autoimmune Dis, 2013. 2013: p. 827254. 58. Lipsky, P.E., Systemic lupus erythematosus: an autoimmune disease of B cell hyperactivity. Nat Immunol, 2001. 2(9): p. 764-6. 59. Shlomchik, M.J., et al., The role of B cells in lpr/lpr-induced autoimmunity. J Exp Med, 1994. 180(4): p. 1295-306. 60. Amadori, A., et al., Genetic control of the CD4/CD8 T-cell ratio in humans. Nat Med, 1995. 1(12): p. 1279-83. 61. Huygen, K. and K. Palfliet, Strain variation in interferon gamma production of BCG- sensitized mice challenged with PPD II. Importance of one major autosomal locus and additional sexual influences. Cell Immunol, 1984. 85(1): p. 75-81. 62. Michaels, R.M. and K.D. Rogers, A sex difference in immunologic responsiveness. Pediatrics, 1971. 47(1): p. 120-3. 63. Vranckx, R., et al., HBV vaccinations in medical and paramedical staff: the impact of age on immunization results. Vox Sang, 1986. 50(4): p. 220-2. 64. Invernizzi, P., et al., Female predominance and X chromosome defects in autoimmune diseases. J Autoimmun, 2009. 33(1): p. 12-6. 65. Cutolo, M., et al., Estrogens and autoimmune diseases. Ann N Y Acad Sci, 2006. 1089: p. 538-47. 66. Nelson, J.L., et al., Microchimerism and HLA-compatible relationships of pregnancy in scleroderma. Lancet, 1998. 351(9102): p. 559-62. 67. Invernizzi, P., et al., X chromosome monosomy: a common mechanism for autoimmune diseases. J Immunol, 2005. 175(1): p. 575-8.

107

68. Ozcelik, T., X chromosome inactivation and female predisposition to autoimmunity. Clin Rev Allergy Immunol, 2008. 34(3): p. 348-51. 69. Nelson, J.L. and M. Ostensen, Pregnancy and rheumatoid arthritis. Rheum Dis Clin North Am, 1997. 23(1): p. 195-212. 70. Confavreux, C., et al., Rate of pregnancy-related relapse in multiple sclerosis. Pregnancy in Multiple Sclerosis Group. N Engl J Med, 1998. 339(5): p. 285-91. 71. Silver, R.M. and D.W. Branch, Autoimmune disease in pregnancy. Baillieres Clin Obstet Gynaecol, 1992. 6(3): p. 565-600. 72. Lockshin, M.D., Pregnancy does not cause systemic lupus erythematosus to worsen. Arthritis Rheum, 1989. 32(6): p. 665-70. 73. Petri, M., Systemic lupus erythematosus and pregnancy. Rheum Dis Clin North Am, 1994. 20(1): p. 87-118. 74. Artlett, C.M., J.B. Smith, and S.A. Jimenez, Identification of fetal DNA and cells in skin lesions from women with systemic sclerosis. N Engl J Med, 1998. 338(17): p. 1186-91. 75. Invernizzi, P., et al., Frequency of monosomy X in women with primary biliary cirrhosis. Lancet, 2004. 363(9408): p. 533-5. 76. Scofield, R.H., et al., Klinefelter's syndrome (47,XXY) in male systemic lupus erythematosus patients: support for the notion of a gene-dose effect from the X chromosome. Arthritis Rheum, 2008. 58(8): p. 2511-7. 77. Dillon, S., et al., Klinefelter's syndrome (47,XXY) among men with systemic lupus erythematosus. Acta Paediatr, 2011. 100(6): p. 819-23. 78. Takegami, T., et al., [A case of SLE associated with Turner's syndrome of 45, XO/46, XXq+ mosaicism (author's transl)]. Nihon Naika Gakkai Zasshi, 1980. 69(7): p. 861-6. 79. Rakfal, S.M. and M. Deutsch, Radiotherapy for malignancies associated with lupus: case reports of acute and late reactions. Am J Clin Oncol, 1998. 21(1): p. 54-7. 80. Gunter, C., Genome biology: she moves in mysterious ways. Nature, 2005. 434(7031): p. 279-80. 81. Ross, M.T., et al., The DNA sequence of the human X chromosome. Nature, 2005. 434(7031): p. 325-37. 82. Augui, S., E.P. Nora, and E. Heard, Regulation of X-chromosome inactivation by the X- inactivation centre. Nat Rev Genet, 2011. 12(6): p. 429-42. 83. Carrel, L. and H.F. Willard, X-inactivation profile reveals extensive variability in X- linked gene expression in females. Nature, 2005. 434(7031): p. 400-4. 84. Yang, F., et al., Global survey of escape from X inactivation by RNA-sequencing in mouse. Genome Res, 2010. 20(5): p. 614-22. 85. Sharma, D. and E. Jankowsky, The Ded1/DDX3 subfamily of DEAD-box RNA helicases. Crit Rev Biochem Mol Biol, 2014. 49(4): p. 343-60.

108

86. Ditton, H.J., et al., The AZFa gene DBY (DDX3Y) is widely transcribed but the protein is limited to the male germ cells by translation control. Hum Mol Genet, 2004. 13(19): p. 2333-41. 87. Foresta, C., A. Ferlin, and E. Moro, Deletion and expression analysis of AZFa genes on the human Y chromosome revealed a major role for DBY in male infertility. Hum Mol Genet, 2000. 9(8): p. 1161-9. 88. Rauschendorf, M.A., et al., Complex transcriptional control of the AZFa gene DDX3Y in human testis. Int J Androl, 2011. 34(1): p. 84-96. 89. Yang, Q., et al., DEAD-box proteins unwind duplexes by local strand separation. Mol Cell, 2007. 28(2): p. 253-63. 90. Merz, C., et al., Protein composition of human mRNPs spliced in vitro and differential requirements for mRNP protein recruitment. Rna, 2007. 13(1): p. 116-28. 91. Stevens, S.W., et al., Composition and functional characterization of the yeast spliceosomal penta-snRNP. Mol Cell, 2002. 9(1): p. 31-44. 92. Lai, M.C., Y.H. Lee, and W.Y. Tarn, The DEAD-box RNA helicase DDX3 associates with export messenger ribonucleoproteins as well as tip-associated protein and participates in translational control. Mol Biol Cell, 2008. 19(9): p. 3847-58. 93. Chao, C.H., et al., DDX3, a DEAD box RNA helicase with tumor growth-suppressive property and transcriptional regulation activity of the p21waf1/cip1 promoter, is a candidate tumor suppressor. Cancer Res, 2006. 66(13): p. 6579-88. 94. Chang, P.C., et al., DDX3, a DEAD box RNA helicase, is deregulated in hepatitis virus- associated hepatocellular carcinoma and is involved in cell growth control. Oncogene, 2006. 25(14): p. 1991-2003. 95. Botlagunta, M., et al., Oncogenic role of DDX3 in breast cancer biogenesis. Oncogene, 2008. 27(28): p. 3912-22. 96. Paludan, S.R. and A.G. Bowie, Immune sensing of DNA. Immunity, 2013. 38(5): p. 870- 80. 97. Dixit, E. and J.C. Kagan, Intracellular pathogen detection by RIG-I-like receptors. Adv Immunol, 2013. 117: p. 99-125. 98. Fitzgerald, K.A., et al., IKKepsilon and TBK1 are essential components of the IRF3 signaling pathway. Nat Immunol, 2003. 4(5): p. 491-6. 99. Gu, L., et al., Human DEAD box helicase 3 couples IkappaB kinase epsilon to interferon regulatory factor 3 activation. Mol Cell Biol, 2013. 33(10): p. 2004-15. 100. Mulhern, O. and A.G. Bowie, Unexpected roles for DEAD-box protein 3 in viral RNA sensing pathways. Eur J Immunol, 2010. 40(4): p. 933-5. 101. Oshiumi, H., et al., DEAD/H BOX 3 (DDX3) helicase binds the RIG-I adaptor IPS-1 to up-regulate IFN-beta-inducing potential. Eur J Immunol, 2010. 40(4): p. 940-8. 102. Schroder, M., M. Baran, and A.G. Bowie, Viral targeting of DEAD box protein 3 reveals its role in TBK1/IKKepsilon-mediated IRF activation. EMBO J, 2008. 27(15): p. 2147- 57.

109

103. Soulat, D., et al., The DEAD-box helicase DDX3X is a critical component of the TANK- binding kinase 1-dependent innate immune response. EMBO J, 2008. 27(15): p. 2135-46. 104. Oshiumi, H., et al., Hepatitis C virus core protein abrogates the DDX3 function that enhances IPS-1-mediated IFN-beta induction. PLoS One, 2010. 5(12): p. e14258. 105. Eaton, W.W., et al., Epidemiology of autoimmune diseases in Denmark. J Autoimmun, 2007. 29(1): p. 1-9. 106. Helmick, C.G., et al., Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part I. Arthritis Rheum, 2008. 58(1): p. 15-25. 107. Selmi, C., et al., Primary biliary cirrhosis. Lancet, 2011. 377(9777): p. 1600-9. 108. Gabriel, S.E., The epidemiology of rheumatoid arthritis. Rheum Dis Clin North Am, 2001. 27(2): p. 269-81. 109. Massardo, L., et al., Early rheumatoid arthritis in Latin America: low socioeconomic status related to high disease activity at baseline. Arthritis Care Res (Hoboken), 2012. 64(8): p. 1135-43. 110. Quintero, O.L., et al., Autoimmune disease and gender: plausible mechanisms for the female predominance of autoimmunity. J Autoimmun, 2012. 38(2-3): p. J109-19. 111. Smith-Bouvier, D.L., et al., A role for sex chromosome complement in the female bias in autoimmune disease. J Exp Med, 2008. 205(5): p. 1099-108. 112. Sasidhar, M.V., et al., The XX sex chromosome complement in mice is associated with increased spontaneous lupus compared with XY. Ann Rheum Dis, 2012. 71(8): p. 1418- 22. 113. Nielsen, J. and M. Wohlert, Chromosome abnormalities found among 34,910 newborn children: results from a 13-year incidence study in Arhus, Denmark. Hum Genet, 1991. 87(1): p. 81-3. 114. Jacobs, P.A., et al., Estimates of the frequency of chromosome abnormalities detectable in unselected newborns using moderate levels of banding. J Med Genet, 1992. 29(2): p. 103-8. 115. Tartaglia, N.R., et al., A review of trisomy X (47,XXX). Orphanet J Rare Dis, 2010. 5: p. 8. 116. Devi, A. and P.A. Benn, X-chromosome abnormalities in women with premature ovarian failure. J Reprod Med, 1999. 44(4): p. 321-4. 117. Baronchelli, S., et al., Cytogenetics of premature ovarian failure: an investigation on 269 affected women. J Biomed Biotechnol, 2011. 2011: p. 370195. 118. Jiao, X., et al., Cytogenetic analysis of 531 Chinese women with premature ovarian failure. Hum Reprod, 2012. 27(7): p. 2201-7. 119. Otter, M., C.T. Schrander-Stumpel, and L.M. Curfs, Triple X syndrome: a review of the literature. Eur J Hum Genet, 2010. 18(3): p. 265-71. 120. Tan, E.M., et al., The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum, 1982. 25(11): p. 1271-7.

110

121. Shiboski, S.C., et al., American College of Rheumatology classification criteria for Sjogren's syndrome: a data-driven, expert consensus approach in the Sjogren's International Collaborative Clinical Alliance cohort. Arthritis Care Res (Hoboken), 2012. 64(4): p. 475-87. 122. Lindor, K.D., et al., Primary biliary cirrhosis. Hepatology, 2009. 50(1): p. 291-308. 123. Arnett, F.C., et al., The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum, 1988. 31(3): p. 315-24. 124. Design of a case control etiologic study of sarcoidosis (ACCESS). ACCESS Research Group. J Clin Epidemiol, 1999. 52(12): p. 1173-86. 125. Rybicki, B.A., et al., A sarcoidosis genetic linkage consortium: the sarcoidosis genetic analysis (SAGA) study. Sarcoidosis Vasc Diffuse Lung Dis, 2005. 22(2): p. 115-22. 126. Iannuzzi, M.C., et al., Sarcoidosis susceptibility and resistance HLA-DQB1 alleles in African Americans. Am J Respir Crit Care Med, 2003. 167(9): p. 1225-31. 127. Adrianto, I., et al., Genome-wide association study of African and European Americans implicates multiple shared and ethnic specific loci in sarcoidosis susceptibility. PLoS One, 2012. 7(8): p. e43907. 128. Rasmussen, A., et al., The lupus family registry and repository. Rheumatology (Oxford), 2011. 50(1): p. 47-59. 129. Hollenbach, J.A., et al., Juvenile idiopathic arthritis and HLA class I and class II interactions and age-at-onset effects. Arthritis Rheum, 2010. 62(6): p. 1781-91. 130. Hinks, A., et al., Dense genotyping of immune-related disease regions identifies 14 new susceptibility loci for juvenile idiopathic arthritis. Nat Genet, 2013. 45(6): p. 664-9. 131. Gregersen, P.K., et al., REL, encoding a member of the NF-kappaB family of transcription factors, is a newly defined risk locus for rheumatoid arthritis. Nat Genet, 2009. 41(7): p. 820-3. 132. Lessard, C.J., et al., Variants at multiple loci implicated in both innate and adaptive immune responses are associated with Sjogren's syndrome. Nat Genet, 2013. 45(11): p. 1284-92. 133. Dillon, S.P., et al., Sex chromosome aneuploidies among men with systemic lupus erythematosus. J Autoimmun, 2012. 38(2-3): p. J129-34. 134. Vitali, C., et al., Classification criteria for Sjogren's syndrome: a revised version of the European criteria proposed by the American-European Consensus Group. Ann Rheum Dis, 2002. 61(6): p. 554-8. 135. Jacobs, P.A., The incidence and etiology of sex chromosome abnormalities in man. Birth Defects Orig Artic Ser, 1979. 15(1): p. 3-14. 136. Danchenko, N., J.A. Satia, and M.S. Anthony, Epidemiology of systemic lupus erythematosus: a comparison of worldwide disease burden. Lupus, 2006. 15(5): p. 308- 18. 137. Invernizzi, P., et al., X monosomy in female systemic lupus erythematosus. Ann N Y Acad Sci, 2007. 1110: p. 84-91.

111

138. Hirschfield, G.M., et al., Primary biliary cirrhosis associated with HLA, IL12A, and IL12RB2 variants. N Engl J Med, 2009. 360(24): p. 2544-55. 139. Juran, B.D., et al., Immunochip analyses identify a novel risk locus for primary biliary cirrhosis at 13q14, multiple independent associations at four established risk loci and epistasis between 1p31 and 7q32 risk variants. Hum Mol Genet, 2012. 21(23): p. 5209- 21. 140. Lopez Herraez, D., et al., Rheumatoid arthritis in Latin Americans enriched for Amerindian ancestry is associated with loci in chromosomes 1, 12, and 13, and the HLA class II region. Arthritis Rheum, 2013. 65(6): p. 1457-67. 141. Karlson, E.W., et al., A connective tissue disease screening questionnaire for population studies. Ann Epidemiol, 1995. 5(4): p. 297-302. 142. Cooney, C.M., et al., 46,X,del(X)(q13) Turner's syndrome women with systemic lupus erythematosus in a pedigree multiplex for SLE. Genes Immun, 2009. 10(5): p. 478-81. 143. Shen, N., et al., Sex-specific association of X-linked Toll-like receptor 7 (TLR7) with male systemic lupus erythematosus. Proc Natl Acad Sci U S A, 2010. 107(36): p. 15838-43. 144. Deng, Y., et al., MicroRNA-3148 modulates allelic expression of toll-like receptor 7 variant associated with systemic lupus erythematosus. PLoS Genet, 2013. 9(2): p. e1003336. 145. Kaufman, K.M., et al., Fine mapping of Xq28: both MECP2 and IRAK1 contribute to risk for systemic lupus erythematosus in multiple ancestral groups. Ann Rheum Dis, 2013. 72(3): p. 437-44. 146. Pisitkun, P., et al., Autoreactive B cell responses to RNA-related antigens due to TLR7 gene duplication. Science, 2006. 312(5780): p. 1669-72. 147. Schroder, M., Human DEAD-box protein 3 has multiple functions in gene regulation and cell cycle control and is a prime target for viral manipulation. Biochem Pharmacol, 2010. 79(3): p. 297-306. 148. Kim, Y.S., et al., Gene structure of the human DDX3 and chromosome mapping of its related sequences. Mol Cells, 2001. 12(2): p. 209-14. 149. Johnston, C.M., et al., Large-scale population study of human cell lines indicates that dosage compensation is virtually complete. PLoS Genet, 2008. 4(1): p. e9. 150. Chen, X., et al., The number of x chromosomes causes sex differences in adiposity in mice. PLoS Genet, 2012. 8(5): p. e1002709. 151. Isensee, J., et al., Sexually dimorphic gene expression in the heart of mice and men. J Mol Med (Berl), 2008. 86(1): p. 61-74. 152. Yang, X., et al., Tissue-specific expression and regulation of sexually dimorphic genes in mice. Genome Res, 2006. 16(8): p. 995-1004. 153. Raefski, A.S. and M.J. O'Neill, Identification of a cluster of X-linked imprinted genes in mice. Nat Genet, 2005. 37(6): p. 620-4. 154. Jiang, Z., et al., CD14 is required for MyD88-independent LPS signaling. Nat Immunol, 2005. 6(6): p. 565-70.

112

155. Li, Q., et al., DDX3X regulates cell survival and cell cycle during mouse early embryonic development. J Biomed Res, 2014. 28(4): p. 282-91. 156. Chen, C.Y., et al., Targeted inactivation of murine Ddx3x: essential roles of Ddx3x in placentation and embryogenesis. Hum Mol Genet, 2016. 157. Li, Y.S., K. Hayakawa, and R.R. Hardy, The regulated expression of B lineage associated genes during B cell differentiation in bone marrow and fetal liver. J Exp Med, 1993. 178(3): p. 951-60. 158. Bonamico, M., et al., Prevalence and clinical picture of celiac disease in Turner syndrome. J Clin Endocrinol Metab, 2002. 87(12): p. 5495-8. 159. Barr, M.L., et al., The triplo-X female: an appraisal based on a study of 12 cases and a review of the literature. Can Med Assoc J, 1969. 101(5): p. 247-58. 160. Chang, D.M., et al., Hormonal profiles and immunological studies of male lupus in Taiwan. Clin Rheumatol, 1999. 18(2): p. 158-62. 161. Mackworth-Young, C.G., et al., Sex hormones in male patients with systemic lupus erythematosus: a comparison with other disease groups. Eur J Rheumatol Inflamm, 1983. 6(3): p. 228-32. 162. Kamphuis, S. and E.D. Silverman, Prevalence and burden of pediatric-onset systemic lupus erythematosus. Nat Rev Rheumatol, 2010. 6(9): p. 538-46. 163. Lalani, S., et al., Clinical features and prognosis of late-onset systemic lupus erythematosus: results from the 1000 faces of lupus study. J Rheumatol, 2010. 37(1): p. 38-44. 164. Buyon, J.P., The effects of pregnancy on autoimmune diseases. J Leukoc Biol, 1998. 63(3): p. 281-7. 165. Petri, M., et al., Combined oral contraceptives in women with systemic lupus erythematosus. N Engl J Med, 2005. 353(24): p. 2550-8. 166. Vandenbroucke, J.P., et al., Oral contraceptives and rheumatoid arthritis: further evidence for a preventive effect. Lancet, 1982. 2(8303): p. 839-42. 167. Bernier, M.O., et al., Combined oral contraceptive use and the risk of systemic lupus erythematosus. Arthritis Rheum, 2009. 61(4): p. 476-81. 168. Demers, R., J.A. Blais, and H. Pretty, [Rheumatoid arthritis treated by norethynodrel associated with mestranol: clinical aspects and laboratory tests]. Can Med Assoc J, 1966. 95(8): p. 350-4. 169. Cruciat, C.M., et al., RNA helicase DDX3 is a regulatory subunit of casein kinase 1 in Wnt-beta-catenin signaling. Science, 2013. 339(6126): p. 1436-41. 170. Busslinger, M., Transcriptional control of early B cell development. Annu Rev Immunol, 2004. 22: p. 55-79.

113