Interferon signaling gene expression, and DNA methylation interactions in Sjögren’s disease via mendelian randomization

Summary

Sjögren’s disease (SD) is an autoimmune disorder characterized by dysregulated interferon (IFN) signaling, but the causal genes and regulatory mechanisms remain unclear. We integrated transcriptomic, epigenomic, and genetic data using summary data-based Mendelian randomization (SMR) and colocalization analyses. A meta-analysis of three datasets (N = 124) identified 331 differentially expressed IFN-associated genes enriched in immune cells. SMR analysis of blood and minor salivary gland (MSG) expression quantitative trait locus (eQTL)/DNA methylation QTL (mQTL) data with the SD genome-wide association study (GWAS) identified five causal genes: SH2B3, LGALS9, CD40, GRB2, and DTX3L. DNA methylation at specific CpG sites regulated the expression of SH2B3 and LGALS9. Colocalization revealed that these genes interact with inflammatory cytokines, including C-C motif chemokine 19 (CCL19), interleukin-2 receptor subunit beta (IL-2Rβ), IL-10, and CCL4. Enzyme-linked immunosorbent assay (ELISA) validation in 16 patients with SD confirmed elevated serum levels. This study elucidates the epigenetic regulation of IFN-associated genes in SD pathogenesis and identifies potential therapeutic targets.

Graphical abstract

Highlights

• •Multi-omics SMR identified five IFN-associated causal genes for Sjögren’s disease
• •DNA methylation at specific CpG sites regulates SH2B3 and LGALS9 expression
• •Causal genes interact with inflammatory cytokines in SD pathogenesis

Health sciences

Introduction

Sjögren’s disease (SD) is a common systemic autoimmune disease that predominantly affects women.¹ Inflammation and tissue destruction of the salivary and lacrimal glands result in oral and ocular dryness, fatigue, and arthralgia.² In addition, extra-glandular manifestations occur in 30%–40% of patients, including interstitial lung disease, peripheral neuropathy, and interstitial nephritis.^3,4 Despite increasing recognition, the etiology of SD remains incompletely understood, and there are currently no curative treatments.⁵

Interferon (IFN) signaling has emerged as a central pathogenic pathway in SD and a promising therapeutic target.^6,7 Multiple genome-wide association studies (GWASs) have identified susceptibility loci within IFN-regulated genes, such as IRF5, STAT4, and TNFAIP3, implicating both innate and adaptive immune pathways in disease pathogenesis.^8,9,10,11 In parallel, epigenetic studies have revealed widespread hypomethylation of IFN-responsive genes in SD, correlating with increased expression and immune activation.^12,13 Moreover, elevated inflammatory cytokines further amplify IFN-induced immune responses, highlighting a complex regulatory network involving genetics, epigenetics, and immune signaling.^14,15

However, most previous studies have relied on correlative evidence and lacked causal inference. In particular, the interplay between IFN-associated gene expression, DNA methylation, and inflammatory cytokines in SD remains poorly understood. Importantly, while blood has been the predominant tissue of study, salivary glands—especially the minor salivary gland (MSG)—are the primary target organs in SD and may exhibit distinct regulatory mechanisms.

To address these gaps, we conducted a multi-omics summary data-based Mendelian randomization (SMR) study to explore the potential causality and interactions among IFN-associated gene expression, DNA methylation, inflammatory cytokines, and SD. Using the SMR approach, we integrated expression quantitative trait loci (eQTLs) and DNA methylation QTLs (mQTLs) with SD GWAS summary statistics, and we also analyzed MSG tissues eQTLs with SD GWAS to identify potential causal IFN-associated genes for SD in the blood and MSG. Furthermore, colocalization analysis was used to integrate eQTLs with inflammatory cytokine GWAS summary statistics, revealing potential interactions between causal IFN-associated genes and inflammatory cytokines. To strengthen the findings from MR analysis and provide robust clinical validation, we investigated the expression of proteins and cytokines associated with IFN-associated differentially expressed genes (DEGs) in serum samples from patients with SD and controls using enzyme-linked immunosorbent assay (ELISA).

Results

Differentially expressed IFN-associated genes

The overall study design and analytical workflow are summarized in Figure 1. To investigate the role of IFN signaling genes in SD, three transcriptomic datasets were utilized. Gene expression from patients with SD (n = 71) and healthy controls (HCs, n = 53) was compared using meta-analysis (Table S1). Subsequently, 711 IFN-associated genes were extracted from GeneCards (Table S2), and linear regression models yielded 331 IFN-associated DEGs (Figure 2A, Table S3). In addition, we performed cell type-specific expression analysis (CSEA) on the 331 IFN-associated DEGs, which were predominantly enriched in T cells, macrophages, monocytes, and natural killer cells (Figure 2B, Table S4).

Figure 1.

Research design

Figure 2.

Differentially expressed IFN-associated genes and cell-type-specific enrichment analysis

(A) Linear regression models and meta-analysis identified 331 differentially expressed IFN-associated genes between patients with SD and healthy controls (HCs). In the depicted volcano plot, meta-analytic effect sizes on the x axis, whereas −log10-transformed p values are shown on the y axis. Red markers signify the 331 significantly differentially expressed genes (DEGs), while black markers depict genes that are not significantly differentially expressed. A dashed line delineates the significance boundary (p < 0.05).

(B) Cell type-specific expression analysis (CSEA) was performed to determine the cell types in which these IFN-associated DEGs were specifically enriched. The y axis represents cell types from blood and minor salivary gland (MSG) tissue. Statistical analysis: Linear regression adjusted for age, sex, and batch effects, followed by fixed-effects meta-analysis (metafor package). Multiple testing correction: Benjamini-Hochberg false discovery rate (FDR) < 0.05. Total N = 124 samples (71 patients with SD, 53 HCs) across three datasets (GSE40611, GSE84844, GSE23117). Red: FDR < 0.05; Black: FDR ≥ 0.05.

Potential blood IFN-associated SD-causal genes regulated by DNA methylation

Integrating eQTL gene data with SD GWAS identified 23 IFN-associated genes that met our significance criteria (P SMR multi < 0.05; heterogeneity in the dependent instrument (HEIDI) test p > 0.01; Cochran’s Q p > 0.05), as detailed in Table S5. Furthermore, by combining the same SD GWAS with mQTL data, we pinpointed 70 DNA methylation probes associated with 38 genes within a 1 Mb range (Table S6). A subsequent analysis that merged potential SD-causal cis-eQTLs and cis-mQTLs highlighted 11 DNA methylation probes potentially influencing five adjacent genes: SH2B3, LGALS9, CD40, GRB2, and DTX3L (Table S7). These findings, which passed our stringent significance and heterogeneity tests (P SMR multi < 0.05; HEIDI test p > 0.01; Cochran’s Q p > 0.05), offer new insights into the epigenetic regulation of genes potentially causal for SD.

The results of blood-based enrichment analysis of mQTL sites showed that DNA methylation sites were mainly enriched in primary monocytes (p = 0.00931), primary B cells (p = 0.00142), primary T cells (p = 0.00029), natural killer cells (p = 0.000126) from peripheral blood, and primary T cells from cord blood (p = 0.00681) (Table S8).

Blood SMR preferentially selected SH2B3, which encodes SH2B3 (also known as LNK, a lymphocyte adaptor protein) in the SH2B adaptor family of proteins, involved in IFN signaling through growth factors and cytokine receptors.^16,17 This study demonstrated significant single nucleotide polymorphism (SNP) signals linked to SH2B3 across eQTL, mQTL, and SD GWAS data. The DNA methylation probe cg26359133 is located in the 5′ UTR region of SH2B3, 3,668 bp from the gene start site. Methylation at this site negatively affected SH2B3 expression (β SMR = −0.44) and SD onset (β SMR = −0.23), whereas SH2B3 expression was positively correlated with SD (β SMR = 0.50). Similarly, ELISA measurements revealed significantly elevated serum SH2B3 levels in patients with SD compared to controls (p < 0.0001, Figure 3D), further supporting the SMR findings. Our findings propose a mechanism whereby low DNA methylation levels increases SH2B3 expression, thereby elevating SD risk (Figures 3A and 3B).

Figure 3.

DNA methylation regulation of SH2B3 expression and its interaction with inflammatory cytokines in SD

(A) SH2B3 locus plots reveal consistent genetic effects of SD GWAS, cis-mQTL, and cis-eQTL (p < 1 × 10⁻⁵).

(B) Three-step SMR analysis demonstrated causal relationships between DNA methylation-mediated SH2B3 gene expressions and the onset of SD (P SMR multi < 0.05; HEIDI test p > 0.01; Cochran’s Q p > 0.05).

(C) Locus comparisons for SH2B3 gene expression and inflammatory cytokines (CCL19, IL-2Rβ, PD-L1, and TGF-α) were performed via colocalization (PPH4 > 0.5).

(D) ELISA validation of serum levels of SH2B3, CCL19, IL-2Rβ, PD-L1, and TGF-α in 16 patients with SD and 16 controls. Statistical analysis: SMR: P_SMR_multi < 0.05, HEIDI p > 0.01, Cochran’s Q p > 0.05. Data sources: eQTLGen (N = 31,684),¹⁸ blood mQTL,¹⁹ and FinnGen R10.²⁰ Colocalization: coloc package, PPH4 > 0.5 indicates shared causal variants. ELISA: p < 0.0001 for SH2B3; Mann-Whitney U test or independent t test for cytokines, p < 0.001. Boxplots show the median and interquartile range (IQR). All samples were assayed in duplicate.

Another key example is LGALS9 (galectin 9), which modulates cell-cell and cell-matrix interactions and reflects hyperactivity of B cells and IFN signaling activation in SD.²¹ Our findings indicate a positive causal association between DNA methylation at probe cg06852032 and LGALS9 expression (β SMR = 3.10). Probe cg06852032 is located in the gene body and 3′ UTR region of LGALS9, 272 bp from the gene termination site. Consistent observations showed that elevated gene expression of LGALS9 (β SMR = 0.06) and methylation levels (β SMR = 0.34) may increase SD risk. Additionally, elevated levels of LGALS9 were observed in the serum of patients with SD (p = 0.0486, Figure 4D), as confirmed by ELISA analysis. Therefore, it is hypothesized that genetic variants increase SD risk by upregulating LGALS9 expression through DNA methylation modifications (Figures 4A and 4B).

Figure 4.

DNA methylation regulation of LGALS9 and its interaction with inflammatory cytokines in SD

(A) LGALS9 locus plots reveal consistent genetic effects across SD GWAS, cis-mQTL, and cis-eQTL datasets (p < 1 × 10−5).

(B) Three-step SMR analysis demonstrated causal relationships between DNA methylation-mediated LGALS9 gene expressions and the onset of SD (P SMR multi < 0.05; HEIDI test p > 0.01, and Cochran’s Q p > 0.05).

(C) Locus comparisons of LGALS9 gene expression and inflammatory cytokines (CD5, IL-10, and CST5) were performed via colocalization (PPH4 > 0.5).

(D) ELISA validation of serum levels of LGALS9, CD5, IL-10, and CST5 in 16 patients with SD and 16 controls. Statistical analysis: SMR criteria as in Figure 3. Data sources: eQTLGen and FinnGen R10. Colocalization: PPH4 > 0.5. ELISA: p = 0.0486 for LGALS9; ∗p < 0.05, ∗∗p < 0.01 versus controls. Boxplots show the median and IQR.

Blood IFN-associated SD-causal genes interacting with inflammatory cytokines

IFN-associated gene expression in the blood interacts with inflammatory cytokines in the pathogenesis of SD.^21,22 Therefore, we hypothesized that integrating IFN-associated cis-eQTL from blood with inflammatory cytokine GWAS data could identify new candidate targets for interactions between blood IFN-associated genes and inflammatory cytokines.

Our findings identified five blood IFN-associated genes with potential causal effects in SD. To further explore the interactions between candidate genes and inflammatory cytokines, we integrated inflammatory cytokines GWAS with cis-eQTLs through colocalization analysis, aiming to investigate whether common genetic effects exist between blood IFN-associated gene expression and inflammatory cytokines. Ultimately, a total of three candidate genes—SH2B3, LGALS9, and CD40—were detected to interact with inflammatory cytokines at a threshold of PPH4 > 0.5 (Table S11).

We found that SNPs regulating SH2B3 expression may also regulate the levels of C-C motif chemokine 19 (CCL19; PPH4 = 0.58), interleukin-2 receptor subunit beta (IL-2Rβ; PPH4 = 0.55), programmed cell death ligand 1 (PD-L1; PPH4 = 0.75), and transforming growth factor α (TGF-α; PPH4 = 0.52). We further confirmed that serum levels of CCL19, IL-2Rβ, PD-L1, and TGF-α were significantly elevated in patients with SD compared to controls (Figure 3D). Our research points to the possibility that genetic variants in SH2B3 may regulate both its expression and the levels of CCL19, IL-2Rβ, PD-L1, and TGF-α, potentially heightening the risk of SD (Figures 3C and 3D).

Further colocalization analyses revealed that LGALS9 expression correlates genetically with T cell surface glycoprotein CD5 (CD5; PPH4 = 0.72), IL-10 (PPH4 = 0.74), and cystatin D (CST5; PPH4 = 0.51), suggesting potential interactions with CD5, IL-10, and CST5 (Figure 3D). Furthermore, ELISA analysis revealed significant differences in the serum levels of CD5, IL-10, and CST5 between patients with SD and the control group (p < 0.0001, Figure 4D). Our findings indicate that genetic variations in LGALS9 could concurrently influence its expression and the levels of CD5, IL-10, and CST5, thereby elevating the risk of SD.

CD40 represents another potential causal gene for SD, implicated in the interactions between blood genes expression and inflammatory cytokines. We identified an inverse relationship between CD40 expression and the risk of SD (β SMR = −0.01). Two inflammatory cytokine pathways exhibit shared genetic effects with CD40 gene expression: CCL4 (PPH4 = 0.55) and IL-13 (PPH4 = 0.53) (Figure 5A). These findings were further validated in the serum of patients with SD (p < 0.0001, Figure 6). Therefore, we hypothesize that genetic variation regulates CD40 expression and interacts with CCL4 and IL-13, thereby contributing to SD pathogenesis in patients.

Figure 5.

SMR and colocalization analyses of CD40 and MSG IFN-associated genes in SD

(A) The left panel depicts SMR results showing the association between CD40 expression and SD GWAS (P SMR multi < 0.05; HEIDI test p > 0.01), whereas the right panel displays a locus comparison of CD40 cis-eQTL and CCL4 GWAS via colocalization (PPH4 > 0.5).

(B) SMR results showing the association between APOBEC3G expression and SD GWAS and locus comparisons of APOBEC3G cis-eQTL with Flt3L and DNER GWAS via colocalization.

(C) SMR results showing the association between IFI27L2 expression and SD GWAS and locus comparison of IFI27L2 cis-eQTL and NGF-β GWAS via colocalization.

(D) SMR results showing the association between TMEM50B expression and SD GWAS and locus comparison of TMEM50B cis-eQTL and IL-1α GWAS via colocalization. Statistical analysis: SMR significance criteria: P_SMR_multi < 0.05, HEIDI test p > 0.01. Data sources: blood eQTL (eQTLGen, N = 31,684), MSG eQTL (GTEx V8, N = 144), inflammatory cytokines GWAS, and SD GWAS (FinnGen R10). Colocalization: coloc R package, PPH4 > 0.5 indicates shared causal variants.

Figure 6.

ELISA validation of causal genes and inflammatory cytokines in patients with SD

Statistical analysis: Independent samples t test or Mann-Whitney U test based on normality testing (Shapiro-Wilk test). The box represents interquartile range (IQR), the line indicates the median, whiskers show the min-max within 1.5 × IQR, and dots represent individual samples. All samples were assayed in duplicate, with intra-assay coefficient of variation (CV) < 10%. Software: GraphPad Prism 10.0.

MSG IFN-associated SD-causal genes interacting with inflammatory cytokines

Integration of MSG tissue eQTLs and inflammatory cytokine GWAS data can identify new candidate targets for interactions between SG tissue genes and inflammatory cytokines. By combining MSG tissue eQTL data from genotype-tissue expression (GTEx; n = 144) with SD GWAS, SMR analysis indicated a potential causal link between SD and thirteen MSG-expressed genes (P SMR multi <0.05, HEIDI p > 0.01, and Cochran’s Q p > 0.05) (Tables S9 and S10).

To further explore the interactions between MSG IFN-associated gene expressions and inflammatory cytokines, we integrated GWAS summary statistics for inflammatory cytokines with potential causal cis-eQTLs using colocalization analysis. The aim was to assess whether genetic influences on MSG gene expression overlap with those affecting inflammatory cytokines. SNPs within the human major histocompatibility complex (MHC) region were excluded due to their high linkage disequilibrium (LD), which could potentially bias subsequent analyses.²³ The analysis revealed four gene-cytokine pathway pairs with PPH4 > 0.5, including apolipoprotein B mRNA editing enzyme catalytic subunit 3G (APOBEC3G), interferon alpha inducible protein 27 like 2 (IFI27L2), SH2B3, and transmembrane protein 50B (TMEM50B) (Table S12).

Through SMR and colocalization analysis, we identified APOBEC3G as a candidate IFN causal gene interacting with inflammatory cytokines in SD MSG tissues (Figure 5B). The research demonstrated a negative correlation between APOBEC3G expression and SD development (β SMR = −0.04). Moreover, Fms-related tyrosine kinase 3 ligand (Flt3L; PPH4 = 0.58) and Delta and Notch-like epidermal growth factor-related receptor (DNER; PPH4 = 0.64) shared genetic effects with APOBEC3G expression. Serum measurements from patients with SD confirmed that APOBEC3G concentrations were reduced, while Flt3L and DNER levels were elevated compared to controls (p < 0.0001, Figure 6). Our results indicate that genetic variations in APOBEC3G could concurrently influence its gene expression and the levels of Flt3L and DNER, thereby contributing to SD pathogenesis in patients.

IFI27L2 is involved in interactions between MSG gene expression and inflammatory cytokines, with its expression positively linked to SD risk (β SMR = 0.05) (Figure 5C). Furthermore, IFI27L2 expression shares genetic effects with nerve growth factor-beta (NGF-β) (PPH4 = 0.55). These findings were validated in serum ELISA tests in both patients with SD and controls (Figure 6). This suggests that genetic variations in IFI27L2 may interact with NGF-β through the IFN signaling pathway, contributing to SD pathogenesis.

TMEM50B is another IFN-associated potential causal gene (Figure 5D). This study found TMEM50B expression to be negatively associated with SD risk (β SMR = −0.05). Furthermore, TMEM50B expression shares genetic effects with IL-1α (PPH4 = 0.72), suggesting a possible interaction with IL-1α. We further confirmed that serum levels of TMEM50B and IL-1α were significantly elevated in patients with SD compared to controls (Figure 6). Therefore, genetic variations may influence SD pathogenesis by regulating TMEM50B expression and IL-1α levels.

Discussion

This study established a potential causal relationship between IFN signaling gene expression and SD. Using multi-omics data, we elucidated the interactions and molecular mechanisms linking IFN-associated causal gene expression, DNA methylation, and inflammatory cytokines. We identified 331 IFN-associated DEGs and integrated SD GWAS data with the eQTLs and mQTLs of DEGs to identify five IFN-associated SD-causal genes: LGALS9, SH2B3, CD40, GRB2, and DTX3L. Meanwhile, colocalization analyses integrating inflammatory cytokine GWAS data identified common genetic effects between SH2B3, LGALS9, and CD40 expression and inflammatory cytokines. In addition, using the same approach of integrating MSG eQTLs with SD and inflammatory cytokine GWAS data, four MSG tissue IFN-associated SD-causal genes, APOBEC3G, IFI27L2, TMEM50B, and SH2B3, were identified as being involved in the interaction between MSG gene expression and inflammatory cytokines. The clinical validation part of this study verified the reliability and clinical relevance of the pre-gene association analysis and MR through serum protein and cytokine content assays, reinforcing the potential role of IFN-associated genes in the pathogenesis of SD and providing a new research perspective for the identification of future targets for intervention.

In the present study, we confirmed that hypomethylation levels upregulated SH2B3 gene expression, which potentially interacted with CCL19, IL-2Rβ, PD-L1, and TGF-α, thereby increasing the risk of SD. SH2B3 (also LNK) was involved in IFN signaling activities through growth factors and cytokine receptors.^16,17 Measurable differences in SH2B3 across phenotypes of SD, particularly in patients with HSB (high symptom burden) or DDF (dryness dominant with fatigue),²⁴ may be associated with clinical heterogeneity in SD by influencing different inflammatory processes. Several studies have found that CCL19, associated with a variety of autoimmune diseases,^25,26,27,28 is highly expressed in SD blood and salivary glands.^29,30 CCL19 attracts dendritic cells and antigen-binding B cells by binding to CCR7,^31,32 causing inflammation in the blood and salivary glands. Furthermore, it has also been shown in large cohorts to correlate with anti-Sjögren’s syndrome-related antigen A (SSA) antibody and IgG levels in patients with SD.²⁹ IL-2Rβ (CD122, IL-2Rβ), together with CD25 and CD132, forms the heterotrimeric receptor (IL-2R).^33,34 By binding to IL-2R, IL-2 regulates T cell activation and cellular immune responses.³⁵ However, elevated levels of IL-2R and impaired IL-2/IL-2R signaling lead to diminished immunosuppression of regulatory T cells (Tregs) in patients with SD and are correlated with the severity of SD.³⁶ Enhancement of Tregs by low-dose IL-2 restores the Th17/Treg balance,^37,38 which is effective and well tolerated in patients with SD.³⁹ PD-L1, a transmembrane protein, is expressed in myeloid cells and regulated by IFN signaling pathways.⁴⁰ In patients with SD, PD-L1 is upregulated by IFN/Janus kinase (JAK)/signal transducer and activator of transcription (STAT) in salivary gland epithelial cells (SGECs). It serves a dual function, being associated with epithelial cell survival and resistance to IFN-mediated apoptosis.⁴¹ PD-L1 reduces the activation of T cells and IFN-γ production, while also contributing to the activation of SGECs and the persistence of the disease.⁴²

Furthermore, our study confirms that DNA methylation upregulates LGALS9 expression, and elevated transcript levels of LGALS9 may affect related cytokines (CD5, IL-10, and CST5) and increase the risk of SD. LGALS9, also known as galectin-9, is part of the galectin family and binds beta-galactoside. It is induced by IFNα signaling, upregulated in SD, regulates B cell activation as a feedback mechanism, and monitors SD disease activity.^21,43 Additionally, LGALS9 shares genetic effects with cytokines (CD5, IL-10, and CST5), collectively contributing to the pathogenesis of SD. For instance, T cell surface glycoprotein CD5, a negative co-stimulator in T cell receptor (TCR) signaling,⁴⁴ shows decreased expression and functionality on peripheral blood lymphocytes of patients with SD,⁴⁵ potentially increasing SD risk.²² Galectin-9 regulates B cell activation by increasing colocalization with the inhibitory co-receptor CD5, thereby modulating TLR4 signal transduction.⁴⁶ It is well recognized that a significant number of cells within SD lesional tissue express the CD5 cell surface molecule.⁴⁷ CD5⁺ Breg cells are crucial in restraining the effector Tfh cell response by producing IL-10 during SD development.⁴⁸ IL-10, a critical anti-inflammatory mediator,⁴⁹ is significantly elevated in the peripheral blood of patients with SD,⁵⁰ with IL-10-producing Breg cells helping to restrain the effector Tfh cell response in SD.⁴⁸ Additionally, a positive correlation exists between IL-10 levels and EULAR Sjögren's Syndrome Disease Activity Index (ESSDAI), which is useful for monitoring SD disease activity.⁵¹ Previous research indicates that the Tim-3/galectin-9 signaling pathway may regulate the function of CD4⁺ T cell subsets, impacting IL-10 levels.⁵²

The present study confirms that the interaction of CD40 expression with inflammatory cytokines (CCL4 and IL-10) is involved in the pathogenesis of SD. CD40, belonging to the tumor necrosis factor (TNF) gene superfamily, encodes proteins that serve as receptors on antigen-presenting cells of the immune system⁵³ and are essential in SD for mediating a variety of immune and inflammatory responses.^54,55 For example, CD40⁻CD154 mediated T-B cell interactions result in aberrant lymphocyte activation, SG inflammation, and subsequent tissue damage.^56,57 CD40 gene expression correlates with CCL4, as demonstrated in previous studies.⁵⁸ CCL4 levels are significantly elevated in the MSG, submandibular gland, and lacrimal gland, suggesting that it is closely involved in the pathogenesis of SD.^59,60,61,62 In addition, IL-13 release from B cells requires CD40 cross-linking and cytokine signaling.⁶³ IL-13 regulates the function of glandular tissue and the recruitment of mast cells to the gland, contributing to the progression of SD immunopathology.⁶⁴

SD is a chronic inflammatory autoimmune disease characterized by lymphocytic infiltration of exocrine glands, with a predominance in the salivary glands. Using MSG eQTLs (the most relevant tissue type) to study their genetic influence on IFN gene expression may be more relevant than studies conducted in blood. We identified APOBEC3G, IFI27L2, TMEM50B, and SH2B3 as potential IFN-associated SD causal genes in MSG tissues based on SMR analyses, with SH2B3 also identified as a relevant gene in blood. Nonetheless, the relationship between SH2B3 expression and SD in blood differed from that in MSG, indicating a tissue-specific influence on the pathogenesis of SD.⁶⁵ In addition, colocalization analyses of gene expression with inflammatory cytokines confirmed the presence of common genetic effects between these gene expression (APOBEC3G, IFI27L2, TMEM50B, and SH2B3) and inflammatory cytokines.

Our study hypothesized that the genetic variant regulating APOBEC3G might simultaneously regulate its gene expression as well as the levels of Flt3L and DNER, thereby participating in the pathogenesis of SD. APOBEC3G, one of the APOBEC3 genes, is a member of the large apolipoprotein B mRNA editing enzyme catalytic polypeptide-like family, which plays a critical role in innate immunity.⁶⁶ The APOBEC3 family is highly expressed in macrophages, lymphoid cells, and dendritic cells, and dysregulated APOBEC3 activity has been implicated in genome mutagenesis in cancer.^66,67 APOBEC3G was notably upregulated in MSG tissues obtained from patients with SD lymphoma,⁶⁸ exhibiting a more widespread distribution across myeloid, B-lymphocyte, and T-lymphocyte clusters.⁶⁹ Furthermore, our research colocalized the genetic influences on APOBEC3G expression with those of inflammatory cytokines. Both Flt3L and DNER exhibited shared genetic effects with APOBEC3G expression, indicating possible interactions among this gene, Flt3L, and DNER. Interestingly, serum levels of Flt3L are elevated in patients with SD and are associated with high disease activity scores, aberrant B cell distribution, and an increased risk of developing lymphoma.^70,71 Flt3L serves as a predictor of SD lymphoma⁷² and may prove valuable as a predictive marker for lymphoproliferative disorders in SD.⁷³ Elevated levels of Flt3L preferentially enhance the proliferation of type I IFN-producing plasmacytoid dendritic cells.⁷⁴ Increased levels of Flt3L may trigger autoimmune diseases.⁷⁵ For example, patients with rheumatoid arthritis (RA) exhibit increased Flt3L levels in both serum and the synovial fluid of inflamed joints.⁷⁶ Moreover, DNER is associated with inflammation of articular cartilage.⁷⁷

Our study found that IFI27L2 expression was positively associated with the risk of developing SD and shared genetic effects with NGF-β. IFI27L2 (also ISG12B), a member of the IFN-stimulated gene (ISG)12 family, has an important role in the apoptotic properties induced by type 1 IFN. Notably, IFI27 was selected as prospective biomarkers for SD.⁷⁸ NGF-β has also been reported to have increased expression in SD SGECs, correlating with MSG inflammatory grade.⁷⁹ Thus, we hypothesize that genetic variation regulates IFI27L2 expression and interacts with NGF-β via the IFN signaling pathway, thereby promoting SD pathogenesis.

Additionally, this study highlighted a potential interaction between TMEM50B gene expression, which is negatively correlated with SD risk, and IL-1α in SD pathogenesis. IL-1α, one of the IL-1 family member, is a widely distributed and pivotal pro-inflammatory cytokine.⁸⁰ Inflammation and damage in SGECs are marked by the release of IL-1α and IL-1β, leading to progressive SG damage and dysfunction.⁸¹ In summary, this study provides mechanistic insights into SjD by integrating transcriptomic, epigenomic, and genetic data through a multi-omics SMR and colocalization framework. By identifying tissue-specific IFN-causal genes and their interactions with inflammatory cytokines, we demonstrate how dysregulated IFN signaling contributes to SD pathogenesis. Clinical validation using patient sera further supports the robustness of these findings. Together, our results advance the understanding of SD pathophysiology and highlight prioritized IFN-associated targets with potential therapeutic relevance.

Limitations of the study

This study has several limitations that should be acknowledged. First, although we utilized GWAS data from FinnGen and UK Biobank, these datasets may not fully capture the genetic and epigenetic diversity of the global population, which restricts the generalizability of our findings to all patients with SD, particularly those from non-European ethnic backgrounds. Second, the cross-sectional nature of the genetic and epigenetic data does not adequately reflect the dynamic changes in gene expression and methylation status that occur during the chronic progression of SD. Third, while we validated key findings at the individual level using ELISA in clinical samples (N = 16 patients with SD and 16 controls), the relatively small sample size limits statistical power, and larger independent validation cohorts are needed to confirm the clinical relevance of the identified biomarkers. Fourth, although Mendelian randomization provides evidence for causal relationships, it relies on specific assumptions (relevance, independence, and exclusion restriction) that cannot be fully verified, and pleiotropy and population stratification may introduce bias despite our sensitivity analyses using HEIDI tests and Cochran’s Q statistics. Fifth, we did not explore the functional consequences of identified genetic and epigenetic modifications through cellular or animal models, and functional validation experiments examining how these variants directly impact cellular phenotypes in SD would significantly strengthen the biological plausibility of our findings. Finally, the MSG eQTL data from GTEx were derived from a limited number of samples and may not fully represent the diseased tissue microenvironment in patients with SD. Future studies incorporating diverse populations, longitudinal designs, larger sample sizes, and functional validation experiments would address these limitations and provide more definitive evidence for tissue-specific mechanisms in SD pathogenesis.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Xiaopo Tang (tangxiaopo@163.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •
  Data: All data supporting the findings of this study are publicly available:

  • ◦
    Transcriptome datasets: Gene Expression Omnibus (GEO) under accession nos. GSE40611, GSE84844, and GSE23117
  • ◦
    SD GWAS summary statistics: FinnGen consortium (Release R10) and UK Biobank
  • ◦
    Blood eQTL data: eQTLGen consortium (https://www.eqtlgen.org/)
  • ◦
    Blood mQTL data: Nat Commun (https://doi.org/10.1038/s41467-018-03371-0)
  • ◦
    MSG eQTL data: GTEx Portal V8 (https://gtexportal.org/)
  • ◦
    Inflammatory cytokine GWAS data: 91 inflammatory markers from a Nordic population
  • ◦
    The ELISA validation data are available from the corresponding author upon reasonable request.
• •Code: Analysis code is available upon reasonable request to the corresponding author.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported by the Science and Technology Innovation Projects for Graduate Students at China Academy of Chinese Medical Science (KC2025017), the High Level Chinese Medical Hospital Promotion Project (No. HLCMHPP2023002), the National Natural Science Foundation of China (82374285), and the Scientific and technological innovation project of China Academy of Chinese Medical Sciences (No. CI2021A01502).

Author contributions

Research design and guidance: J.H., C.X., X.Z., and X.T. Data collection and analysis: Z.G. and J.H. Manuscript writing and revision: J.H. Suggestions for research content and revision: C.X., Z.G., X.Z., and X.T. Manuscript revision and editing: C.X., Z.G., and J.H. All authors have endorsed the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Serum from SD patients	Guang'anmen Hospital	N = 16
Serum from healthy controls	Guang'anmen Hospital	N = 16
Critical commercial assays
Human CD40 ELISA Kit	RUIXIN BIOTECH	RX106392H; RRID:AB_10584325
Human CCL4 ELISA Kit	RUIXIN BIOTECH	RX105348H; RRID:AB_2831264
Human APOBEC3G ELISA Kit	mIbio	ml221663; RRID:AB_2258442
Deposited data
SD transcriptome GEO: GSE40611	GEO	https://www.ncbi.nlm.nih.gov/geo/
SD GWAS	FinnGen	https://www.finngen.fi/ (R10)
Blood eQTL	eQTLGen	https://www.eqtlgen.org/
MSG eQTL	GTEx V8	https://gtexportal.org/
Software and algorithms
R	R Core Team	https://www.r-project.org/ (v4.4.3)
metafor	Viechtbauer82	CRAN
SMR	Zhu et al.83	https://yanglab.westlake.edu.cn/software/smr/
GraphPad Prism	GraphPad	Version 10.0

Experimental model and study participant details

Human subjects

This study was approved by the Ethics Committee of Guang'anmen Hospital, China Academy of Chinese Medical Sciences (approval number: 2022-175-KY-01) and registered through the International Traditional Medicine Clinical Trial Registry (ITMCTR) (registration number: ITMCTR2025000593, registered on 2025-03-26; http://itmctr.ccebtcm.org.cn/en-US/Home/ProjectView?pid=0bd4c245-d023-4320-9170-f651e8d6a278). All experiments were conducted in accordance with relevant regulatory standards in China and the Declaration of Helsinki. All participants provided written informed consent prior to enrollment.

Patient characteristics∗∗

The patient cohort included 15 females and 1 male, with a mean age of 52.92 ± 11.65 years. All SD patients fulfilled the ACR-EULAR 2016 classification criteria. The control group included 15 females and 1 male, with a mean age of 52.67 ± 12.18 years, matched for age and sex.

Inclusion criteria

Fulfilled ACR-EULAR 2016 classification criteria

Age ≥ 18 years

Exclusion criteria:

• (1)Patients with comorbid rheumatoid arthritis, systemic lupus erythematosus, or other connective tissue diseases;
• (2)Patients with other severe conditions, such as malignancies, infections, psychiatric disorders, etc., that may affect study outcomes;
• (3)Patients who have used antibiotics, biologics, or similar medications within the past month.

Serum samples were collected from 16 SD patients and 16 age-sex matched controls. All participants provided written informed consent prior to enrollment. The inclusion criteria for the SD group adhered to the ACR-EULAR 2016 classification criteria, ensuring a consistent and reliable diagnosis, while the control group included healthy individuals without autoimmune disorders. We utilized ELISA to quantify the expression levels of proteins encoded by IFN-related DEGs, as well as key inflammatory cytokines in the serum samples. Commercially available ELISA kits with established sensitivity and specificity were selected based on a thorough review of relevant literature and manufacturer recommendations. The assays were conducted according to standardized protocols, ensuring high reproducibility and minimizing technical variability.

Method details

IFN-associated gene extraction and data sources

We extracted IFN-associated genes from the GeneCards database using the keyword “interferon” and a relevance score ≥ 3. From the Gene Expression Omnibus (GEO) database, we obtained three public transcriptome datasets from SD patients and healthy controls (GEO: GSE40611, GEO: GSE84844, and GEO: GSE23117).^84,85,86 Linear regression models and the transcriptome meta-analysis identified IFN-associated DEGs in SD. GWAS for SD were sourced from FinnGen and UK Biobank.²⁰ The data for inflammatory cytokines were sourced from 91 cytokines and growth factors in a Nordic population (Table S10). Blood eQTL were sourced from eQTL Gen, encompassing genetic data of 31,684 individuals.¹⁸ Additionally, blood mQTL summary data originated from two European cohorts.¹⁹ MSG tissue eQTL data were acquired from the Genotype-Tissue Expression (GTEx) project.⁸⁷ This study specifically focused on cis-eQTLs and cis-mQTLs, denoting single nucleotide polymorphisms (SNPs) located within a 1-Mb distance from both the beginning and end of a gene. SNPs were filtered based on standard quality control criteria: minor allele frequency >0.01, Hardy-Weinberg equilibrium P >1×10^-6, and linkage disequilibrium pruning with r² <0.1.

Differential expression and cell-type enrichment analysis

We analyzed IFN-associated DEGs using linear regression models under the assumptions of linearity, independence of observations, and homoscedasticity of residuals. Models were adjusted for age, sex, and dataset-specific batch effects, as these covariates are known to influence transcriptomic variation and autoimmune disease-related heterogeneity in SjD. Fixed-effects meta-analyses of DEGs across the three datasets were conducted using the R package metafor, as the effect directions were generally consistent. Heterogeneity was evaluated using Cochran’s Q and I² statistics to confirm the appropriateness of the fixed-effects model, and the detailed results of heterogeneity measures are provided in Supplementary Tables. Cell Type-Specific Expression Analysis (CSEA) was utilized to pinpoint the specific cell types that are enriched by IFN-associated DEGs. Given the disease characteristics of SD, our research concentrated on cell types found in blood and MSG tissues. Additionally, we examined the regulatory aspects of DNA methylation sites. For these analyses, we utilized tools such as eFORGE (available at http://eforge.cs.ucl.ac.uk/) and http://grch37.ensembl.org/.

Mendelian randomization analysis

We used SMR multi-tools because they allow for the integration of eQTL, mQTL, and GWAS summary statistics to identify causal relationships between gene expression, DNA methylation, and SD risk. The inclusion of the HEIDI-outlier test helps to distinguish pleiotropy from confounding due to linkage disequilibrium, thereby increasing the robustness of causal inference. This makes SMR particularly suitable for multi-omics integrative analysis in SD.⁸⁸

Using a three-step SMR approach to determine the causal relationship between blood IFN-associated genes, DNA methylation and SD. Step one: expressions of blood IFN-associated genes served as the exposure, SNPs were instruments, and SD was considered the outcome. Step two: blood DNA methylation was exposure, SNPs were instruments, and SD was the outcome. Based on the results of the first two steps, Step three: blood DNA methylation was exposure, SNPs were instruments, and blood IFN-associated gene expressions were the outcome. Ultimately, candidate blood IFN-associated genes for SD need to fulfil the following conditions: P SMR multi < 0.05, genome-wide suggestive significance in eQTLs/mQTLs, and GWAS (P < 1 × 10^-5), and heterogeneity in the dependent instrument (HEIDI) test (P > 0.01).⁸³

Using the same SMR approach as above, MSG tissue analysis was performed to determine the MSG IFN-associated causal genes in SD. One step SMR is needed: expressions of MSG IFN-associated genes served as the exposure, SNPs were instruments, and SD was considered the outcome. Sensitivity analyses were employed to evaluate heterogeneity using Cochran's Q statistic in the MR-Egger and inverse variance weighting (IVW) methods, with P > 0.05 indicating no heterogeneity.

Colocalization analysis

Colocalization analysis serves as an effective method for identifying overlaps between causal variants in molecular and disease phenotypes, offering valuable insights into the molecular pathways of complex diseases.⁸⁹ Due to limitations in assessing causal relationship between SD and inflammatory cytokines,²² colocalization analysis was applied to reveal potential interactions between expression of SD-causal gene and inflammatory cytokines, utilizing the coloc R package, with a threshold of PPH4 > 0.5.

ELISA validation

To evaluate differences in protein and cytokine levels between the SD and control groups, statistical analyses were performed using appropriate methods based on data distribution. Independent t-tests or Mann-Whitney U tests were applied depending on whether the data followed a normal distribution. To visualize the differences in protein and cytokine levels between the SD and control groups, we used GraphPad Prism 10 for all graphical presentations. Box-and-Whisker Plots were used to display the distribution of protein and cytokine levels in each group, showing the median, interquartile range (IQR), and potential outliers.The following proteins and cytokines were measured using ELISA kits (see key resources table): Blood IFN-associated genes: SH2B3, LGALS9, CD40. MSG IFN-associated genes: APOBEC3G, IFI27L2, TMEM50B. Inflammatory cytokines: CCL4, CCL19, IL-2Rβ, PD-L1, TGF-α, CD5, IL-10, CST5, IL-13, Flt3L, DNER, NGF-β, IL-1α. All assays were performed in duplicate according to manufacturer's instructions.

Quantification and statistical analysis

The study design workflow is shown in Figure 1.

All statistical analyses were performed using R version (v4.4.3). Sample sizes and statistical tests are indicated in figure legends. P < 0.05 was considered statistically significant. Data are presented as mean ± SEM.

Differential expression analysis used linear regression models adjusted for age, sex, and batch effects. Fixed-effects meta-analyses across three datasets (GEO: GSE40611, GEO: GSE84844, GEO: GSE23117) were conducted using the metafor package (version 3.8-1) with Benjamini-Hochberg FDR < 0.05. Heterogeneity was assessed using Cochran's Q and I² statistics.

Mendelian randomization analyses were performed using SMR multi-tools (version 1.3.1). Significance thresholds: P_SMR_multi < 0.05, HEIDI test P > 0.01, and Cochran's Q P > 0.05. Colocalization analysis used the coloc R package (version 5.2.3) with PPH4 > 0.5 threshold.

For ELISA validation (N = 16 per group), independent t-tests or Mann-Whitney U tests were used based on normality testing. Samples were assayed in duplicate with intra-assay CV < 10%. GraphPad Prism 10.0 was used for analysis and visualization. Significance: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

Additionally, three transcriptomic datasets for SD from GEO were utilized to detect IFN-associated DEGs in SS. These DEGs underwent cell CSEA to ascertain the cell types where they were predominantly expressed. Using a three-step SMR method, an integrated examination of SD GWAS, along with eQTLs and mQTLs in blood, was carried out to delineate causal relationships among blood IFN-associated genes, DNA methylation, and SD pathogenesis (P SMR multi < 0.05; HEIDI test P > 0.01, and Cochran’s Q P > 0.05). Subsequently, blood cis-eQTLs were integrated with GWAS data on inflammatory cytokines, identifying interactions through colocalization analysis (PPH4 > 0.5). Further, cis-eQTLs from MSG tissue sourced from the GTEx database were analyzed and integrated with SD GWAS data, to identify IFN-associated causal genes in the MSG (P SMR multi < 0.05; HEIDI test P > 0.01; Cochran’s Q P > 0.05). Then, cis-eQTLs of IFN-associated causal genes in the MSG were then integrated with GWAS data on inflammatory cytokines, assessing potential interactions between MSG IFN-associated genes and inflammatory cytokines via colocalization analysis (PPH4 > 0.5). Sensitivity analyses were employed to evaluate heterogeneity using Cochran's Q statistic in the MR-Egger and IVW methods, with P > 0.05 indicating no heterogeneity. Serum from 16 SD patients and 16 healthy controls were collected and subjected to ELISA for the detection of causative gene-expressed proteins and related cytokines. IFN, interferon; SD, Sjögren’s disease; GEO, Gene Expression Omnibus; DEGs, differentially expressed genes; CSEA, cell type-specific expression analysis; SMR, summary data-based Mendelian randomization; GWAS, genome-wide association study; eQTL, expression quantitative trait loci; mQTLs, DNA methylation quantitative trait loci; MSG, minor salivary gland; GTEx, genotype-tissue expression; IVW, inverse variance weighting; ELISA, enzyme-linked immunosorbent assay.

Published: February 7, 2026