Sjögren’s syndrome is an autoimmune epithelitis characterized by production of antibodies and focal lymphocytic infiltrates that cause severe dysfunction of the lacrimal glands and salivary glands. The secretory dysfunction leads to ocular surface and oral pathology and symptoms. When manifestations of the autoimmune processes arise in extraglandular sites, such as the kidneys, vasa vasorum, and peripheral vessels, they further impair quality of life and significantly increase risks of life-threatening vasculitides and non-Hodgkin’s mucosa-associated lymphoid tissue (MALT) lymphomas.
An understanding of how the disease processes develop and evolve could improve diagnosis and treatment. However, the etiopathogenesis of Sjögren’s syndrome remains a formidable challenge. Sjögren’s syndrome comprises two major subtypes, i.e., primary Sjögren’s syndrome and secondary Sjögren’s syndrome [1
], and both subtypes present in multiple clinical phenotypes. The cellular and molecular processes that underlie the clinical subtypes are diverse. Primary Sjögren’s infiltrates may be either T-cell predominant or B-cell predominant [2
], and may have either positive- or negative type I interferon (IFN) signatures [3
]. B-cell predominant infiltrates may have or lack germinal centers [4
]. These features imply that Sjögren’s syndrome may have multiple etiologies and may develop along multiple pathogenic pathways.
Identified risk factors offer clues to Sjögren’s syndrome etiology. It is known for some time that certain human leukocyte antigen (HLA) class II molecule variants favor B-cell responses to the anti-Sjögren’s-syndrome-related A (Ro/SSA) and anti-Sjögren’s-syndrome-related B (La/SSB) autoantigens, formation of lymphocytic infiltrates with germinal centers, and risks for MALT lymphomas and vasculitides, evidently because the high-risk HLA class II variants have high affinities for Ro/SSA and La/SSB epitopes [5
]. Additional genetic polymorphisms reported to be associated with risk for Sjögren’s syndrome include variants of interleukin 10 (IL-10) [8
], tumor necrosis factor (TNF) [11
], antigen peptide transporter 2 (TAP2) [12
], and 2′-5′-oligoadenylate synthetase 1 (OAS1) [13
]. Female sex is a well-known risk factor [14
], and parity was also reported to be a risk factor [15
]. The concordance rate for monozygotic twins was estimated to be low [17
]. Therefore, interactions between genetic predisposition, sex, and parity are not sufficient to determine Sjögren’s etiology. Other phenomena, assumed to be environmental in nature, must be involved.
In 1983, after observing that thyrocytes in Graves’ disease patients express HLA class II molecules [18
], Bottazzo, Pujol-Borrell, and colleagues proposed that induction of aberrant HLA class II molecule expression allows non-immune system cells to present autoantigen epitopes directly to T cells and initiate formation of autoimmune lesions [19
]. In the time since, epithelial cells in labial salivary glands of patients with Sjögren’s syndrome were confirmed to express HLA class II molecules [20
], as well as HLA class I molecules. [23
]. They were also shown to express cluster of differentiation 80 (CD80) and CD86 [24
], which provide the costimulatory signals necessary for T cells to become activated after antigenic stimulation; the costimulatory molecule CD40 [25
]; several chemokines (C–C motif chemokine ligand 3 (CCL3), CCL4, CCL5, CCL17, CCL21, CCL22, CCL28, C–X–C motif chemokine ligand 8 (CXCL8), CXCL9, CXCL10, CXCL12, CXCL13, and CX3CL1) [26
]; several cell adhesion molecules (CD2, intercellular adhesion molecule 1 (ICAM-1), lymphocyte function-associated 1 (LFA-1), and LFA-3) [34
]; and several cytokines (B-cell activating factor (BAFF), granulocyte-macrophage colony-stimulating factor (GM-CSF), IFN-γ, IL-lα, IL-1β, IL-2, IL-6, IL-10, IL-18, IL-28/IL-29, transforming growth factor beta (TGF-β), and TNF-α) [34
Many of the immune response-related molecules expressed by epithelial cells of salivary glands with Sjögren’s lesions also are expressed by epithelial cells in salivary glands from healthy control subjects, but generally at much lower levels than in affected glands. Therefore, the data suggest that exposures to certain risk factors increase the probability that exocrine gland epithelial cells will upregulate their expression of genes that cause them to either (1) function as surrogate antigen presenting cells, or (2) create local microenvironments where interactions between professional antigen-presenting cells, autoreactive T cells, and autoreactive B cells are likely lead to immune cell activation and self-organization of ectopic lymphoid structures.
Working with histologically normal lacrimal glands from healthy rabbits, we confirmed Frey and coworkers’ finding that lacrimal gland epithelial cells express exhibit immunohistochemical positivity for prolactin (PRL) [40
], which has pleiotropic actions as a cytokine [41
]. We found that corneal injury, corneal adenovirus infection, and pregnancy altered levels of PRL immunoreactivity [45
]. We also found that lacrimal glands express messenger RNA (mRNA) for PRL and also mRNAs for class II major histocompatibility complex (MHC II), CD80, and CD86, numerous cytokines, and numerous chemokines. The abundances of many transcripts exhibited considerable gland-to-gland variability. Some of the abundance variations appeared to be related to the dryness, while others were related to the temperatures and to the environment the animals experienced prior to study [47
]; additional variability was induced by the hormonal environment of pregnancy [49
]. However, much of the variability appeared related to stochastic phenomena localized within glands. Pearson’s analysis of the variability showed that variations of clusters of transcripts were significantly correlated. Principal component analyses confirmed the major correlation cluster and indicated that certain transcripts might be expressed by two or more correlation clusters in the same gland. We interpreted the transcript correlation clusters as signatures of clusters of cells that were interacting with each other coordinately.
The major correlation cluster was of interest, as it included mRNA for BAFF, a B-cell mitogen and activating factor, and mRNAs for the B-cell chemokine, CXCL13, and the T-cell cytokine, CCL21, all associated with Sjögren’s lymphocytic foci. A preliminary laser capture microdissection of glands from a nulliparous animal confirmed that epithelial cells expressed a number of immune response-related gene transcripts, that immune cells in small clusters expressed certain other transcripts, and that both epithelial cells and immune cells expressed certain transcripts. Notably, the immune cell clusters, or “accumulations”, were too small to be identified as Sjögren’s foci. These characteristics suggest that epithelial cells and immune cells can engage in multiple signaling interactions with each other, forming localized networks that might expand over time to eventually manifest as histopathologically identifiable Sjögren’s lesions.
We collated and analyzed data from our previously published studies to discern how age, pregnancy, and exposures varying ambient conditions influence the formation of correlation clusters. We also analyzed the abundances of selected transcripts in samples of acinar cells, intralobular duct cells, interlobular duct cells, intralobar duct cells, and clustered immune cells obtained by laser capture microdissection of three glands from term-pregnant animals. Two glands were representative of the subgroup of glands in which the Sjögren’s-resembling transcript correlation cluster was present at low levels, and one gland was representative of the subgroup in which the Sjögren’s-resembling transcript correlation cluster was present at high levels.
We found that epithelial cells rarely express mRNA for MHC II, but frequently express mRNA for CD1d, which presents both bacterial and autologous glycolipids invariant α-chain natural killer (NK) T cells. They even more frequently express mRNA for MHC I, which presents epitopes of intracellular proteins, both viral and autologous, to CD8+ T cells. We also found that samples of the epithelial segments heterogeneously express costimulatory molecules, chemokines, and cytokines. These findings support the hypothesis that epithelial cells in histologically normal glands can interact with clusters of immune cells to comprise networks with stochastically varying cellular compositions and transcript expression profiles. Environmental and hormonal exposures increase the likelihoods that small networks will expand and, perhaps, evolve as ectopic lymphoid structures of different phenotypes. The ability to detect molecular signatures of the small networks in histologically normal glands may have important implications for the future diagnosis and proactive treatment of incipient Sjögren’s syndrome.
Our findings indicate that clusters of immune cells expressing high levels of many of the same transcripts that are highly expressed in Sjögren’s infiltrates are present in histologically normal lacrimal glands. As the clusters can be detected before classical Sjögren’s infiltrates are identified, we suggest that they may be early precursors of Sjögren’s infiltrates. This conjecture is consistent with the finding that levels of the typical Sjögren’s autoantibodies can be elevated before Sjögren’s syndrome symptoms become clinically significant [52
Our findings are consistent with the hypothesis that signaling interactions with epithelial cells in all levels of the acinus duct axis, as well as with cells outside the acinus duct axis, influence formation of the immune cell clusters that evade peripheral tolerance mechanisms. Acinar cells maintain high baseline levels of APRIL. Acinar and ductal cells appear to contribute CCL2, CCL4, CCL28, BAFF, IL-6, and IL-10. Immune cell clusters appear to contribute CXCL13, CCL21, MHC II, and MMP-9 [53
], as well as additional BAFF. The levels at which mRNAs for these proteins are expressed vary coordinately, and through a wide range, as indicated by the range of gland projections with respect to PC1 (Figure 2
Because mRNA for MHC II was predominantly expressed in immune cell clusters, it appears that induction of epithelial HLA class II molecule expression may be a downstream event in Sjögren’s pathogenesis, rather than a triggering event in Sjögren’s etiology. Notably, however, acinar epithelial cells expressed mRNAs for MHC I, CD1d, CD80, and CD86. Therefore, lacrimal gland epithelial cells might contribute to Sjögren’s etiology not only by expressing chemokines that recruit T cells, B cells, and professional antigen-presenting cells, but also by expressing mitogenic factors that support immune cell survival and CD4+ cell proliferation in response to MHC II-restricted epitopes presented by the professional antigen-presenting cells. They might also by displaying costimulatory signals, presenting MHC I-bound autoantigen epitopes to CD8+ T cells, and presenting CD1d-bound glycolipids to invariant α-chain NK T cells. Presumably, the antigen receptors of CD4+ or CD8+ T cells that participate in such interactions would have autoantigen epitope affinities strong enough to support positive selection during thymic maturation, but too weak to determine activation-induced deletion.
A corollary to this hypothesis is that, in addition to the coreceptors, CD80 and CD86, epithelial cell also express additional coreceptors and receptors, typically expressed by T cells, which mediate immune cell-to-epithelial cell paracrine signaling, as well as epithelial cell autocrine signaling and epithelial cell-to-epithelial cell paracrine signaling. As indicated in Figure 5
, epithelial cells expressed mRNA for CD25, the α-subunit of the IL-2 receptor; mRNA for the CCL4 receptor, CCR5; mRNAs for CD8 and CD28 (Figure 6
); and mRNA for IL-2 [47
]. These transcripts all contributed to the PC1-negative loading transcript cluster. These findings are not unprecedented, as CD8 is known to be expressed by NK cells, dendritic cells, and cortical thymocytes, as well as by T cells; CD28 is known to be expressed by plasmacytoid dendritic cells, plasmacytes, NK cells, eosinophils, and neutrophils, as well as by T cells [54
]; and IL-2 receptors are known to be expressed by renal tubular epithelial cells [55
]. Their expression by epithelial cells in the lacrimal gland might contribute to the positive feedback loop which coordinates epithelial cell expression and immune cell expression of the transcripts of the PC1-negative loading cluster.
The resemblance between the transcript expression profiles of immune cell clusters and Sjögren’s foci is likely even greater than revealed by the present microdissection data, as PC1 also receives strong negative loadings from additional transcripts that also are expressed in Sjögren’s foci. The transcripts that contribute strong negative loadings to PC1 include CD40L and CD40 [49
], a cognate coreceptor pair expressed respectively by T cells and by B cells and professional antigen-presenting cells. They also include mRNAs for CD19 and CD72 [49
], expressed by resting B cells; mRNAs for IL-2, IL-4, IL-5, IL-7, IL-13, IL-17A [49
], which contribute to B-cell activation; and mRNA for IL-21 [49
], which supports T-cell survival in T-cell zones and dark zone bases.
Several transcripts that are particularly associated with active lymphoid follicles notably do not contribute strong negative loadings to PC1. These include mRNA for lymphotoxin beta (LT-β), expressed by lymphoid tissue-organizing cells; mRNA for CXCL12, which recruits B cells to germinal center dark zones; mRNA for amyloid precursor protein intracellular domain (AICD), which mediates immunoglobulin gene somatic hyper mutation; mRNA for CD22, expressed by mature B cells; and mRNA for CD138, expressed by terminally differentiated plasmacytes [49
]. Each of these transcripts contributes a strong or moderately strong negative loading to PC2. Gland P.G5.06.OS had a positive projection with respect to PC2, but gland P.G5.03.OD, another subgroup P.G5.A gland that had an even larger negative projection with respect to PC1, had a large negative projection with respect to PC2. Our finding (Figure 7
) that immune cell cluster samples from gland P.G5.06.OS had significantly higher abundances of transcripts that contributed strong loadings to other PCs is informative in this context, as it indicates that transcripts which contribute to other PCs can be expressed in the same immune cell clusters as the PC1-negative loading transcripts. In other words, immune cell clusters in gland P.G5.03.OD may have been relatively further along on a trajectory toward manifestation as recognizable ectopic lymphoid structures with germinal centers. Moreover, because autoreactive B cells escape self-recognition checkpoints when they are activated in ectopic lymphoid structures [57
], one might ask whether B cells in the immune cell clusters we described contribute autoantibodies to the circulation even before germinal centers are formed.
The remarkably broad distribution of lacrimal glands’ projections with respect to PC1 as depicted in Figure 2
may offer a hint for answering the question of why the prevalence of primary Sjögren’s syndrome is relatively low (0.02–0.1%) [60
] That is, immune cell clusters are only likely to expand into Sjögren’s foci in the glands in which they are already well developed, i.e., in glands with the largest negative projections with respect to PC1. In view of the broad distributions of lacrimal glands’ projections with respect to PC2 and PC3, and the broad distributions of subgroup P.G5.A glands’ projections with respect to PC4 and PC5 as depicted in Figure 2
, the conclusion that immune cell clusters express transcripts from multiple transcript clusters may have general implications for understanding the heterogeneities of both primary and secondary Sjögren’s syndrome phenotypes.
In rabbits, the likelihood that a gland will achieve a large negative PC1 projection is increased by exposure to high degrees of environmental dryness, as well as by the hormonal environment of pregnancy (Figure 3
). Environmental dryness is a risk factor for dry-eye disease in humans [61
]; to our knowledge, it is yet to be reported as a risk factor for Sjögren’s syndrome. However, the findings in Figure 2
, Figure 3
and Figure 4
demonstrate that responses to both environmental dryness and the hormonal environment of pregnancy are subject to very high levels of stochasticity.
In addition to the genetic risk factors we discussed in Section 1
of this paper, mechanisms related to viral infections were proposed [62
]; these include mimicry of autoantigens by viral proteins, apoptosis-associated release of autoantigens, apoptosis-associated release of ligands for molecular pattern recognition receptors, and expression of proliferation factors and chemokines in latently infected T cells or B cells [67
]. Occupational exposure to organic solvents [69
] and exposure to psychosocial stress [70
] were also proposed to be risk factors. Our findings indicate that epithelial cells in histologically normal lacrimal glands are able to recruit immune cells, provide mitogenic support for them, and shape their activities. These abilities may contribute to the responses to such risk factors, and the inherent stochasticity of the development of immune cell/epithelial cell networks would contribute to the stochasticity of responses to the risk factors.