Horizons in Sjögren’s Syndrome Genetics

Abstract

Sjögren’s syndrome (SS) is a complex polygenic autoimmune disorder. A few major genetic effects have been identified. Historically, HLA and non-HLA genetic associations have been reported. Recently, the HLA region continued to reveal association findings. A new susceptibility region has been suggested by a study of a D6S349 microsatellite marker. Among non-HLA studies, recent association of immunoglobulin κ chain allotype KM1 with anti-La autoantibodies in primary Sjögren’s syndrome confirms findings in a study from two decades ago. Meanwhile, mouse models have been employed to study the genetic contribution to salivary lymphadenitis or dry eyes and mouth. Gene transfer exploration in mouse models shows promise. The authors review the HLA and non-HLA association studies and the mouse model work that has been reported. Newly developed genomic capacity will provide, in the future, a much closer approximation of the true picture of the genetic architecture of Sjögren’s syndrome.

Introduction

The technical capacity to explore the human genome in detail has arrived. We can evaluate almost any of the approximate 3 billion human DNA base pairs or estimated 10 million common Homo sapiens DNA polymorphisms. The genetic revolution has already generated billions of data points in attempts to identify genetic risk for those common disease phenotypes with high mortality, such as breast cancer, type 1 diabetes, and heart disease. Sjögren’s syndrome (SS) and a large number of more rare phenotypes have been slower to benefit from new technical capacities. The present progress trajectory hopefully will envelop these phenotypes of less compelling public interest, resulting in gene discoveries that can redefine their pathogenic mechanisms.

What are the prospects for genetic discovery in Sjögren’s syndrome? Certainly, no easily identified Mendelian pattern has been reported percolating through SS patient families. In fact, there is no population-based study of SS familial aggregation patterns or comparison of concordance between monozygotic and dizygotic twins, either of which would be very helpful to estimate the genetic contribution to disease causation.

Modern geneticists characterize the SS phenotype that consists of suspected, yet undiscovered, interplay of various genes, multiple genetic mechanisms, and environmental triggers, as “complex”. Sjögren’s syndrome features genetic complexities that are compounded by clinical setting presentations with diverse combinations of organ specific, systemic, exocrine and/or nonexocrine manifestations.

Diagnosis ambiguity and prognosis variability present another layer of complexity for Sjögren’s study. No specific diagnostic criteria, exclusively or conclusively, establish Sjögren’s syndrome. Likewise, no specific set of criteria is uniformly employed by all current clinical investigators, albeit a recently adopted European–American Consensus Criteria [1] has been added to the panoply of choices.

Genetic etiology in autoimmune disorders is traditionally surmised by recognition of familial aggregation. Genetic predisposition in SS development was first recognized in a 1937 study of three-generation kindred [2]. Since that time, genetic contributions to particular disease symptoms in SS are only beginning to be understood. To date, the best family study in primary SS for establishing a tendency for familial aggregation evaluated six multiplex kindreds. Segregation analysis suggested that a Mendelian dominant genetic effect for autoimmunity was present in these pedigrees. This genetic effect was not apparently linked to HLA or heavy chain immunoglobulin genes. These authors made the still valid conclusion that the genetic origin of SS results from the interaction of several genes [3].

Although only a relatively few genetic studies of Sjögren’s syndrome have been performed, a few of the major genetic effects have been identified. Two convincing associations dominate identified genetic effects:

1. Association with the locus called human leukocyte antigen (HLA) locus, located on human chromosome 6p21.3 (Fig. 1)

2. Confirmed association of immunoglobulin κ chain allotype KM1 with anti-La autoantibodies in primary Sjögren’s syndrome [4, 5] (Table 1).

Fig. 1.

The human leukocyte antigen (HLA) allele reported to have associations with primary Sjögren’s syndrome. Confirmed associations are in bold type

Table 1.

Reported Sjögren’s syndrome genetic associations outside the human leukocyte antigen (HLA) region

Gene	Polymorphism site	Allele or genotype	Association found	Study (case/control)	Association not found (case/control)
ApoE		ApoE ε4	Early onset Primary SS	Pertovaara et al. [28] (63/64)
CCR5		CCR53Δ2	Lower in Primary SS	Petrek et al. [22] (39/76)
Fas	Nt–671	G/G	Primary SS	Bolstad et al. [26] (70/72)	Mullighan et al. [27] (108/101)
	IVS2nt176	SNP C/T
	IVS5nt82	SNP C/G
GSTM1		Homozygous null genotype	Primary SS	Wang et al. [29] (106/143)
HA-1	Nt+500/504*	168His	Lower in Primary SS	Harangi et al. [30] (88 /371)
IL-1RA	Intron-2	IL1rn*2	Primary SS	Perrier et al. [24] (36/100)	Petrek et al. [22] (39/76)
IL-4Rα		Q576R	Primary SS	Youn et al. et al. [31] (45/74)
				Ramos-Casals et al. [25] (48/98)
				Wang et al. [29],
IL-10	Promoter -1082	SNP G/A	Primary SS	Hulkkonen et al. [23] (62/400)	Rischmueller et al. [34], (108/165)
	Promoter -819	SNP C/T
	Promoter -592	SNP C/A
	Promoter (-1,082, -819, -592)	ATA/ATA		Origuchi et al. [32], (47/107)
	Promoter (-1,082, -819, -592)	GCC/ATA		Font et al. [33], (63/150)
		G9	SS w/ cutaneous vasculitis	Gomez et al. [17], (39/15)
Ig KM 26/1204 anti-La in 1 SS [4]		KM1	Anti-La in Primary SS	Whittingham et al. [4], (26 /1204) Pertovaara et al. [28], (6/26)
MBL	Codon 54	Wild-type allele	Primary SS	Wang et al. [29], (104/143)	Mullighan et al. [27], (97/106)
		Homozygous mutant allele	Lupus, RA, SS	Tsusumi et al. [35], (266/129)	Aittoniemi et al. [36], (62/400)
PTPN22	Nt. 1858	SNP C/T	Primary SS	Gomez et al. [17], (70/308)	Ittah et al. [21], (183/172)
					Criswell et al. [9], (16/2064)
Ro52	Intron 1 (nt. 7216)	SNP C/T	Anti-Ro in SS	Nakken et al. [37], (97/72)
	Intron-3 (137 from exon4)	SNP A/G	Anti-Ro52 in Primary SS	Imanishi et al. [38] (111/97)
TCRBV		Deletion/Deletion	Primary SS	Lawson et al. [39], (61/121)
TGFβ1	Codon 10	SNP C/T	Primary SS with anti-La	Gottenberg [10], (92/254)

^*Two nucleotide polymorphism

Susceptibility Genes

Susceptibility gene identification provides diagnostic and prognostic tools. Susceptibility genes define SS disease risk variation within the human species. Each gene discovered to contribute to disease risk establishes part of disease causation. Genes that carry risk are causation genes that define the mechanism of disease and are critical to basic clinical and scientific comprehension of pathogenesis. Identification and inventorying susceptibility genes is the first step towards clinical applications of such knowledge.

Susceptibility gene identification appears to be only now barely possible in complex diseases, such as Sjögren’s syndrome. Once identified, how quickly we achieve a subsequent understanding of how they operate is completely unpredictable. Some genes will be like HLA-B27 in ankylosing spondylitis where the mechanism is not yet agreed three decades after discovery [6], while others will be like sickle cell anemia, where the discovery of the structural changes and oxygen-binding behavior of the mutant form clearly implies mechanism [7].

With only ~25,000 genes in the human genome, the task looks less daunting than a few years ago when estimates were that humans consisted of >150,000 genes. Even so, no effort has been made in Sjögren’s syndrome to evaluate all of these genes and the intervening sequence simultaneously in what is commonly called a “genome scan”. No doubt, the ink will hardly be dry on this review before this more global approach will be accomplished for Sjögren’s syndrome, as well. This discussion is therefore historically restricted to candidate gene results from association and linkage studies.

Before being able to understand how candidate genes interplay to increase disease risk, we need to identify the players. The specter of false positive results is always a risk in association studies. Some have estimated that only about a third of published association results are replicated and, thereby, sufficiently confirmed to conclude that they are robust to a degree of true-positive associations [8]. Replication is a minimum, although powerful requirement of the scientific method, and is the major problem for most genetic associations now alleged.

Working List of Candidate Genes

HLA Region

Recent findings from larger samples are consistent with older studies supporting a genetic predisposition from HLA class II marker alleles [9]. Except for a strong association of anti-Ro(SS-A) and anti-La(SS-B) autoantibodies to the HLA region (Fig. 1), there is, as yet, only modest support for disease severity or other clinical subsets being associated with the HLA region, despite reports to the contrary that have not been replicated. Indeed, some show that without these autoantibodies, they find no HLA association with Sjögren’s syndrome, whatsoever [10].

The first reported HLA association with Sjögren’s syndrome was in 1976 with B8, the HLA Class I allele I of HLA-B that is on the HLA-HR3 haplotype [11]. In 1975, HLA-DR3 had been shown to be associated with serologically defined Dw3 by Tom Chused and colleagues. This was the first evidence of the now well-known and many times replicated HLA-DR3 association with Sjögren’s syndrome [12]. DR3 and B8 are in disequilibrium, being found together more frequently than predicted by chance. As the B8 association vanishes when DR3 is considered but controlling for B8 does not entirely remove the DR3 effect [13], the associations with B8 and DR3 are probably measuring the same association, and furthermore, the primary association is with DR3. This means that the HLA-DR locus is closer to the causative gene than is HLA-B. The existing data are not sufficiently discriminating to determine whether the causative allele is DR3 itself, or not.

There are a number of important immune system genes in the HLA region in addition to the members of HLA Class I and II, including TNF, LTA, LTB, TAP, and LMP genes, some of which have also been associated. Of these, neither of the associations between anti-Ro positive primary Sjögren’s and TNFα [14] or TAP2 [15] were replicated in the subsequent study by Jean et al. [16].

An interesting example of association in the HLA region, but 2 Mb away from the HLA-DR, is the D6S349 microsatellite marker. Located in a region of otherwise undistinguished DNA, it is associated and may identify a new susceptibility region for primary Sjögren’s syndrome [17]. As of this writing, no known genes reside in the haplotype block containing D6S439.

The extent of disequilibrium has been estimated for some haplotypes in HLA [e.g. DR3) to be as large as 3.5 Mb, complicating the interpretation of any association result, as the example demonstrates of DR3 and B8, which are both on this haplotype and presented above.

Dissecting the interplay between various effects of multiple alleles of a gene and multiple genes remains a daunting and largely unsolved problem. Both systemic lupus erythematosus and Sjögren’s syndrome are associated with DR3 and DR2 and illustrate this issue. A large association study done by Graham and colleagues [18] shows that disequilibrium around the DR2 association disintegrates relatively rapidly, while that around DR3 does not, and hence, the 3.5 Mb haplotype. This is the same as appreciating that the DR2 containing haplotypes responsible for the observed association tend to be much smaller than the associated DR3 containing haplotype.

There is strong evidence that some haplotypes operate together. Indeed, gene complementation at HLA-DQ was described two decades ago with the combination of the DQw1 and DQw2 allelic serotypes producing an odds ratio >17 for high levels of anti-Ro and anti-La in primary Sjögren’s syndrome patients [19], suggesting a synergistic interaction between alleles.

PA Correa and colleagues present the interesting observation that a TNFα allele and haplotype increases risk for primary Sjögren’s syndrome, while the same allele appears to protect from tuberculosis [20]. Perhaps, what we currently view as “HLA susceptibility genes” should be thought of as an incredibly complicated equation that influences the survival past so many problems potentially faced in life and certainly not just susceptibility genes unique for Sjögren’s syndrome. Meanwhile, the TNFα -308A allele is thought to be associated secondarily with Sjögren’s syndrome through DRB1*03 [21], raising a question on whether tuberculosis risk and primary Sjogren’s syndrome are operating at different loci and whether their opposing risks are related only through linkage disequilibrium.

Genes Outside the HLA Region

Genes outside the HLA Region have also been associated with primary SS, a form of secondary SS or its autoanti-bodies (Table 1). These include chemokine and cytokine polymorphisms for CCR5 [22], IL-10 [23], IL-1 Ra [24], IL-6 [23], TGFβ1 [21] and TNF-α [14]. The IL-4 receptor polymorphism is a promising candidate [25]. Fas and Fas ligand gene polymorphisms have also been investigated as possible candidate susceptibility genes [26, 27], perpetuating themes of relatively dysfunctional apoptotic genes being important autoimmunity risk factors.

Lack of replication results is a serious problem with non-HLA genetic associations. Only primary Sjögren’s syndrome with the association of anti-La autoantibodies in immunoglobulin light chain allotype KM1 has been confirmed. The confirmation effort required 20 years, a reflection of minimal interest levels in Sjögren’s syndrome genetics within academic research communities. No other alleged non-HLA region genetic associations are considered confirmed; possible associations of many are obviously controversial. Perhaps closest to being repudiated is mannose binding ligand associations (Table 1).

There are many other possible genes based both on knowledge of inflammatory response physiology and upon other literature related to SS. CXCL13 and CXCR5 attract B cells to target organs and are expressed in Sjögren’s [40]. The known and alleged Sjögren’s syndrome autoantigens are obvious candidate genes, including Ro, La, immunoglobulin, α-fodrin [41], β-fodrin [42], SS-56 [43], ABCA1 [44], P230 trans-Golgi Network protein [45], 97 kD protein associated with Golgi complex [46], α-amylase [47], and muscarinic-3 receptor [48]. Polymorphisms in the RO52 (SSA1) gene have been associated with anti-Ro52 kD autoantibodies in primary Sjögren’s (Table 1). Perhaps, findings that different RO52 (SSA1) gene polymorphisms have been related to anti-Ro52 autoantibodies and to anti-Ro in Sjögren’s syndrome are interrelated.

Protein expression changes have been found in Sjögren’s syndrome including the cysteine-rich secretory protein 3 (CRISP3) [49], aquaporin 5 [50], and aquaporin 1 [51], COX-1 [52], that also suggest candidate genes. There are unpublished studies that will greatly expand the gene expression data available for Sjögren’s syndrome.

The genes responsible for effects in the NOD mouse model of Sjögren’s on mouse chromosomes 1 [53], 3 [54], and 4 [55] will become candidates when they are identified. Other findings in the NOD, such as the LGP10 autoantigen [56], high levels of matrix metalloproteases [57] and cysteine proteases [58], also suggest candidate genes.

The technical capacity to evaluate large numbers of polymorphisms relatively quickly should lead to an evaluation of all of these candidates and the entire complement of human genes that is now accessible for evaluation.

Mouse Models and Gene Transfer

Meanwhile, genetic contributions toward understanding the salivary lymphadenitis or dry eyes and mouth of Sjögren’s syndrome are available from work in mouse models.

The non-obese diabetic (NOD) mouse is an inbred strain that has been used for many years as a model of Type 1 diabetes. More recently, lymphocytic infiltrates of the salivary glands and reduced salivary and lacrimal gland function have been noted in these animals [59]. Strikingly, the NOD mouse is the only previously reported model of SS, which shows exocrine glandular dysfunction. There are circulating antibodies that bind the surface and the nucleus of salivary gland cells. NOD mice make antibodies against alpha-fodrin [60], but there is little evidence that anti-Ro or anti-La are present in these mice. There is, however, evidence that antibodies are critical to disease in this model. NOD mice without functional B cells (NOD IgMnull) have salivary infiltrates consisting of T cells but have normal salivary gland function. Infusion of IgG from NOD mice or from humans with SS induces a decrement of salivary gland function in the NOD IgMnull mouse, though [61]. Antibodies against muscarinic receptors may be important in this regard.

Like some other reported animals with SS, the NOD mouse may be considered a model of secondary SS as another autoimmune disease is present—namely, type 1 diabetes. The diabetes in these animals is dependent on the MHC haplotype. Studies with NOD congenic for H-2^b demonstrate that while diabetes does not develop, salivary gland infiltrates do develop [62]. Thus, these congenic animals appear to be a model of primary SS. Because the cellular infiltrate, cell surface binding antibodies, and glandular dysfunction are present, the NOD mouse and its variants will no doubt prove useful in deciphering the pathogenesis of SS. Nonetheless, while replicating the disease in several ways, this model does not have appreciable anti-Ro and anti-La, a near universal finding in human disease.

The MRL-lpr/lpr mouse has a genetic defect in the fas gene, and therefore, a defect in the apoptotic death of lymphocytes. There is massive accumulation of lymphocytes and development of a disease similar to human systemic lupus erythematosus. This animal has been used extensively to study lupus autoimmunity [63]. The salivary glands of MRL-lpr/lpr mice are infiltrated with activated T lymphocytes. After transfer of these cells into SCID mice (no functional T or B cells), salivary and lacrimal lesions develop. Anti-CD4 or anti-Vβ8 monoclonal antibodies prevent development of lesions in the SCID recipients [64].

Unfortunately, there are inherent problems with the MRL-lpr/lpr mouse as a model of SS. Clearly, this is a model of secondary, not primary disease, as these mice also have a lupus-like illness. Second, there is no decreased function of the salivary glands [65]. Third, these animals do not develop anti-Ro or anti-La in a manner similar to human patients. In one study, 6 of 17 mice had antibodies to 52 kD Ro, but the titer was high in only 3. Meanwhile 1 of 17 was reported to have anti-60 kD Ro, and no titer or inhibition data were given [66].

Other animal models of SS have been reported (and are discussed elsewhere in this journal), each of them with significant problems. An acute sialoadenitis occurs in a graft-vs-host transplant model [67]. Other than a pathological description, no other data have been reported. Mice transgenic for the envelope protein of the hepatitis C virus develop salivary lesions resembling SS [68], and humans with hepatitis C can develop a sicca illness [69]. The NFS/ sld mouse bears a mutation resulting in sublingual gland differentiation arrest. After neonatal thymectomy, these mice develop pathology in the other salivary and lacrimal glands that resembles that found in SS, with females affected more often than males [64]. The infiltrate was predominantly composed of CD8-positive lymphocytes, and while there were anti-salivary antibodies in the sera, no anti-Ro or anti-La was detected [64]. Mice homozygous for the alymphoplasia (aly) mutation lack lymph nodes and Peyer’s patches. These animals develop inflammatory infiltrates (CD4 lymphocyte rich) of the salivary glands, lungs, pancreas, and lacrimal glands. The mechanism by which this defect gives the absent lymph node and absent Peyer’s patches phenotype is not known [70].

Of particular interest, Fleck and colleagues reported infection of C57BL/6-lpr/lpr mice with murine cytomegalovirus with resultant acute and chronic sialoadenitis [71]. This inflammation persisted after clearance of the virus. High levels of anti-Ro, anti-La, rheumatoid factor, and anti-dsDNA were produced. Thus, this is the only previously reported mouse model of SS with convincingly high levels of anti-Ro and anti-La, although whether the anti-Ro is anti-60 kD Ro or anti-52 kD Ro, or both, is not clear. Pathology of other organs is not given, but the presence of anti-dsDNA, which is usually considered specific to lupus, implies that this is also a model of secondary SS [71].

Therapeutic genetic transfer exploration in mouse salivary glands shows great promise as models for gene expression. These experiments are being done to explore the potential advantages of salivary glands as vehicles for gene expression. This approach toward improving salivary gland functioning or toward inhibiting immune mediated gland destruction in Sjögren’s syndrome has been pursued, particularly in the laboratory led by Bruce Baum at the National Institute of Dental and Craniofacial Research (NIDCR) at the National Institutes of Health [72].

Salivary tissue can be accessed noninvasively, as a site for gene transfers. For example, a recombinant adenovirus-associated virus vector (AAV) transfer of interleukin-10 cDNA into the non-obese diabetic (NOD) murine model has been reported to suppress inflammatory salivary disease [72].

Efficacy of a transgenic gastrointestinal hormone vaso-active intestinal peptide (VIP) in prevention of NOD sialadenitis was also recently reported [72]. The immunomodulatory properties of VIP increase IL-10 expression [73], while inhibiting lymphocyte migration, some macrophage functions, T cell proliferation, and expression of certain inflammatory mediators [74—76]. VIP may have the potential to improve the management of several autoimmune disorders. Interestingly, there is sufficient homology between mouse and human VIP that transgenic human VIP (hVIP) acts through mouse VIP receptors in transgenic mice [72].

Transfer by retrograde ductal infusion into the NOD mouse salivary gland of recombinant serotype 5 adeno-associate virus, encoded as a vector for human VIP cDNA (rAd5CMVhVIP or as rhVIP below) had astonishing consequences [72]. The mice were treated before onset of disease. Relative to either empty vector controls or uninfected controls, these animals had twice the salivary flow of control animals. Perplexingly, there was no difference in the inflammatory changes in the salivary gland between the rhVIP NOD mice and their controls. Levels of IL-2, IL-10, IL-12, and TNFα were much lower in the mice expressing rhVIP, while levels of IL-4, IL-6, INFγ, and RANTES did not differ from controls.

VIP gene transfer showed local disease modifying and immunosuppressive effects in the NOD model. These results suggest that this approach may be useful in the prevention of Sjögren’s syndrome salivary manifestations [72] and leaving us to wonder whether there might be susceptibility genes in this pathway influencing the human disease.

Future Horizons and Perspectives

Clearly, we know very little about the genetics of Sjögren’s syndrome. Other than the strong association with alleles of the HLA Class II genes, HLA-DR and HLA-DQ, there is not much else that is at present sufficiently convincing to be considered conclusive. As in virtually all of the HLA-Class II associations, the molecular mechanism explaining why these alleles are susceptibility genes is not known. To know how susceptibility creates these associations would be an important advance in our understanding, indeed. The prospects are that the genes now being found in other related autoimmune diseases, especially rheumatoid arthritis and systemic lupus erythematosus will be evaluated in Sjögren’s syndrome patients and controls. This work is likely to greatly expand the list of genes associated with Sjögren’s syndrome.

Previously, the impediment to progress was the inability to evaluate genes directly and rapidly. Recent technical progress has removed this problem. The evaluation of hundreds of thousands of allelic differences between human chromosomes is within our grasp and would accompany suitable patient and appropriate control materials and databases. In addition, the newly developed capacity to evaluate the entire genome will provide, in the future, a much closer approximation of the genetic architecture of Sjögren’s syndrome.