Spatial multi-omics profiling uncovers metabolic heterogeneity in Sjögren’s syndrome and identifies PS(36:1) as a potential therapeutic target

Abstract

Background

Sjögren syndrome (SS) is a common autoimmune disease characterized by lymphocytic infiltration. Describing the transcriptional and metabolic features of the disease from a spatial perspective can enhance our understanding of the disease pathogenesis and treatment;

Methods

We collected eight human labial samples, including four labial gland samples from patients with SS and from four healthy controls. We integrated single-cell RNA sequencing, spatial transcriptomics, and spatial metabolomics techniques to generate SS-associated spatial gene expression maps and spatial metabolite profiles at a single-cell resolution. We also analyzed the characteristic metabolic and genetic changes of SS samples and infiltrated CD4⁺ T cells. Immunohistochemistry was used to detect the infiltration of CD4⁺ T cells in the salivary glands. In vivo experiments were conducted using NOD/ShiLtj mice to validate the therapeutic effects of targeting PS(36:1) on SS-like symptoms, as assessed by HE staining, salivary flow rate assays, and immunofluorescence experiments.

Results

Comprehensive data from spatial multi-omics identified the cell types and distributions within the immune microenvironment of salivary glands in SS. CCL19 was significantly increased in lymphocyte infiltration, while IGHG4 was elevated in glandular area. Linoleic acid metabolism undergoes reprogramming in SS, with alterations in lecithin, linoleic acid, 13(S)-hydroxyoctadecadienoic acid and dihomo-γ-linolenate. Furthermore, PS (36:1) was found to be abnormally enriched in lymphocyte focus, which may be related to the abnormal expression of CD74 and HLA-DRA. Also, CXCL13 corresponded to areas resembling high levels of PS(36:1) in infiltrated CD4⁺T cells. We further investigated the effect of CD4⁺ T cells on the reprogramming of glycerophospholipid metabolism in SS, including key regulatory genes and key metabolites. Among them, LYPLA2 and PS(36:1) exhibited the most representative results. In vivo, the PS(36:1) synthesis inhibitor alleviated dry mouth symptoms and reduced lymphocyte infiltration in the salivary gland tissues of NOD/ShiLtj mice, while also suppressing CD4⁺ T cell accumulation in the infiltrating foci.

Conclusions

The multi-omics analysis conducted in this study enhances our understanding of the key regulatory mechanisms driving the pathogenesis of SS and offers novel insights for its precision therapy. Furthermore, PS(36:1) emerged as a potential therapeutic target for future SS research and treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-07361-x.

Introduction

Sjögren’s syndrome (SS) is a common autoimmune connective tissue disease primarily affecting the exocrine glands, manifesting clinically as sicca symptoms such as dry mouth and eyes [1]. Histopathologically, typical features include lymphocytic infiltration of the salivary gland by T-, B-, and antigen-presenting cells, which is the diagnostic standard for labial salivary gland biopsy [2]. Dysregulation of autoimmune behaviors is a key contributor to the pathogenesis of SS. T and B cells play a major role in adaptive immunity, and B cell accumulation is more prominent in the later stage of SS, while the early stage of SS is marked by T lymphocyte infiltration [3, 4]. Further investigation of immune alterations in SS is crucial for determining effective therapeutic targets.

Metabolic reprogramming involves the modification of different metabolic processes that regulate the function and phenotype of immune cells to meet the demands of immune regulation and response. Metabolic reprogramming has been reported in various autoimmune diseases, such as vitiligo [5], rheumatoid arthritis [6], systemic lupus erythematosus [7], and SS [8]. The abnormal activation of CD4⁺ T lymphocytes is considered a key initiator of SS. Naive T cells have lower energy requirements; they take up small amounts of glucose and produce adenosine triphosphate (ATP), primarily through the tricarboxylic acid (TCA) cycle and oxidative phosphorylation (OXPHOS) [7]. During activation, T cells require extensive reprogramming, including upregulation of aerobic glycolysis, downregulation of oxidative phosphorylation, and upregulation of cholesterol biosynthesis, to meet their bioenergy and biosynthesis requirements [9]. Evidence also suggests that disordered phospholipid metabolism contributes to the pathogenesis of SS. A previous study reported that the abundance of triglyceride (TAG) and phosphatidylcholine (PC), which are linked to fatty acid oxidation and energy metabolism, respectively, changed significantly in the serum of SS patients [10]. Therefore, it is important to explore the metabolic reprogramming patterns of SS for effective treatment.

The characterization of spatial information is essential for the understanding of complex metabolic reprogramming in SS. Currently, relevant studies on SS have not explored the influence of the heterogeneity of spatial distribution and the interaction of tissue cells on metabolism. Single-cell RNA sequencing (scRNA-seq) can reveal different cell types in diseases and provide high-resolution transcriptome maps of SS at the single-cell level; however, this technique lacks information on cell spatial distribution. The 10× Genomics Visium platform for a spatial transcriptome (ST) can systematically identify cell types and gene expression at the structural and spatial levels of selected tissues to observe differences in gene expression in different functional regions [11]. Spatial metabolomics (SM) is a method that enables the determination of small-molecule metabolites in tissue sections with micron resolution and high throughput, thereby revealing changes in metabolic levels in the spatial dimension. However, because the sizes and arrangements of different cell types vary, the spatial resolution of ST and SM currently does not reach the single-cell level.

Spatial information is discarded in these scRNA-seq analyses due to tissue digestion and cell dissociation [12]; consequently, a critical gap remains in understanding how the spatial organization of specific cell types within the salivary gland microenvironment—particularly the localization of infiltrating CD4⁺ T cells relative to resident glandular cells—drives local metabolic reprogramming that cannot be discerned from dissociated cell analyses. Integrated analysis using scRNA-seq, ST, and SM facilitates the analysis of changes in the disease metabolic network at the single-cell and spatial distribution levels [13, 14]. We hypothesized that such spatial metabolic heterogeneity is a key driver of SS pathology. In this study, we applied an integrated scRNA-seq, ST, and SM approach to labial gland tissues of patients with SS and healthy controls. Our study provides spatially resolved gene expression and metabolic profiles of SS with single-cell resolution, specifically aiming to decipher how metabolic reprogramming within distinct spatial regions contributes to SS development, thereby laying the scientific foundation for the development of novel therapeutic targets.

Results

Spatial multi-omics maps revealed spatially specific genetic and metabolic alterations in the labial glands of patients with SS

To systematically characterize the cellular basis of SS, we began by constructing a scRNA-seq atlas of labial glands from patients and healthy controls. Dimension reduction was performed using principal component analysis, followed by visualization using the uniform manifold approximation and projection (UMAP) method. Cell annotation identified eight clusters: acinar cells, B plasma cells, duct cells, endothelial cells, fibroblasts, myeloid cells, smooth muscle cells, and T cells (Fig. 1A–C). A significant decrease in acinar cells and an increase in fibroblasts were observed in SS (p<0.01, t-test), suggesting glandular destruction and fibrotic repair. Subsequently, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed a close association between SS development and immune alterations (Fig. 1D). T, B, duct, endothelial, and acinar cells were examined (Fig. 1E–G, Figure S1). CD4⁺ T-lymphocytes played a crucial role in the onset and progression of SS. T cells were further classified as CD4⁺ memory T cells, CD4⁺ naïve T cells, CD4⁺ Treg cells, CD8⁺ cytotoxic GZMA⁺ cells, CD8⁺ cytotoxic GZMB⁺ cells, CD8⁺ cytotoxic GZMK⁺ cells, or gdT cells (Fig. 1E–G). Notably, the numbers of naïve CD4⁺ T cells and CD4⁺ Treg cells increased significantly in SS (p<0.001, t-test). KEGG analysis revealed that the upregulated genes in SS were mainly enriched in antigen processing and presentation as well as the IL-17 signaling pathway (Fig. 1H). Finally, the results of cell communication analysis were closely linked to changes in cell proportions. For instance, the interaction strength and number of interactions of CD4⁺ memory cells with other cells decreased, whereas those of CD4⁺ naïve T cells and CD4⁺ Treg cells with other cells increased (Fig. 1I–J) (p<0.01, t-test).

Fig. 1.

Single-cell sequencing analysis of Sjögren syndrome (SS). (A) Uniform manifold approximation and projection (UMAP) plot showing the eight cell subsets in the SS and control groups. (B) Proportion of the eight cell subsets in the SS and control groups. (C) Heat map of the top markers of the eight cell subsets. (D) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways of differentially expressed genes between the SS and control groups. (E) UMAP plot showing the seven cell subsets of T cells in the SS and control groups. (F) Proportions of the seven cell subsets of T cells in the SS and control groups. (G) Heatmap of the top markers of the seven cell subsets of T cells in the SS and control groups. (H) KEGG pathways of differentially expressed genes in T cells (SS vs. control). (I) Cell communication analysis of the eight cell subsets (SS vs. control). Red indicated increased intercellular communication, while blue represented decreased intercellular communication. The thicker the line, the more significant the trend. (J) Analysis of cell communication between T cells and other cells (SS vs. control). Red indicated increased intercellular communication, while blue represented decreased intercellular communication. The thicker the line, the more significant the trend

To spatially resolve the cellular alterations identified by scRNA-seq, we performed ST and SM analyses on tissue sections from patients with SS and healthy controls. After applying the UMAP method for dimensional reduction of the spatial transcriptomic data, eight clusters were identified. By integrating the scRNA-seq data, robust cell type decomposition (RCTD) analysis was performed to determine the possible cell types and proportions at each spot of the tissue section (Fig. 2A–C). Compared to healthy controls, T and B cells in SS exhibited specific clustering and distribution, indicating the presence of lymphocytic infiltrates. This suggests that exploring the mechanisms underlying the local aggregation of lymphocytic cells and decreasing the degree of lymphocytic infiltration may be crucial for treating SS. Cell communication analysis demonstrated a significantly greater number of inferred interactions in the SS group compared to the control group (p < 0.001, t-test; Fig. 2D). Furthermore, the analysis revealed a markedly enhanced communication network between CD4⁺ regulatory T cells (Tregs) and other cell types in SS (Fig. 2E), implying a potential key role of Tregs in the pathogenesis of the disease. Spatial metabolomic data were subjected to t-distributed stochastic neighbor embedding analysis. Fifteen clusters in the positive mode and 12 clusters in the negative mode were obtained (Fig. 2F, G, Figure S2). Orthogonal partial least squares discrimination analysis (OPLS-DA) was then applied to assess the group distribution, and the results indicated that the SS and control groups could be clearly separated in both modes (Fig. 2H–I).

Fig. 2.

Spatial transcriptome and metabolism analyses of SS. (A) UMAP plot showing the eight clusters in the SS and control groups. (B) Robust cell type decomposition (RCTD) analysis of each group. (C) Heatmap of spatial marker genes. (D, E) Cell communication analysis of spatial transcriptome and metabolism (SS vs. control). Red indicated increased intercellular communication, while blue represented decreased intercellular communication. The thicker the line, the more significant the trend. (F, G) UMAP reduction visualizing 15 clusters in the positive mode and 12 clusters in the negative mode. (H, I) Orthogonal partial least squares discrimination analysis (OPLS-DA) analysis in the SS and control groups in the positive and negative modes indicated that the SS and control groups could be clearly separated in both modes

Given the pathological hallmark of focal lymphocyte infiltration, we next segmented the tissue into the gland areas (Zone A) and lymphocyte infiltration areas (Zone B) to pinpoint spatially restricted molecular changes. The pathological diagnostic criteria for Sjögren’s syndrome were the formation of lymphocyte infiltration foci and the atrophy of glandular tissue. Therefore, these two regions were of significant importance for the treatment of SS. In this study, areas containing glandular cells were selected as gland areas (Zone A), while regions with lymphocyte aggregation forming infiltration foci were designated as lymphocyte infiltration areas (Zone B). The control group samples contained only Zone A, whereas the SS group samples contained both Zone A and B (Fig. 3A). CCL19 was a chemokine that regulated T-cell migration and played a critical role in immune-related diseases. IGHG4 was a key member of the immunoglobulin heavy-chain gene family and was involved in immune regulation. Spatial transcriptomic analysis demonstrated a significantly (p<0.01, t-test) higher expression of the chemokine CCL19 within the lymphocyte infiltration areas compared to the glandular regions in SS tissues (Fig. 3B, C), positioning it as a potential key factor driving lymphocyte migration and focal aggregation. The gland area in the SS group exhibited a significant (p<0.001, t-test) increase in IGHG4 expression (Fig. 3D, E), suggesting that IGHG4 is associated with gland destruction or fibrosis in patients with SS. Furthermore, we performed dimension reduction based on the gene expression signatures and visualized the spatial features of CCL19 and IGHG4 in each sample (Fig. 3F, G). Subsequently, based on Zones A and B, t-distributed stochastic neighbor embedding dimension reduction and heatmap analysis were conducted on the spatial metabolomic data (Fig. 3H, I). We integrated differentially expressed genes (DEGs) and metabolites (DEMs) along with their spatial features in ST and SM data using a “point-to-point” analysis, establishing a metabolome-transcriptome correlation network. This approach allowed us to uncover spatial relationships between metabolite distribution and gene expression in different tissue compartments (Fig. 3J).

Fig. 3.

Combined analysis of the spatial transcriptome and metabolism. (A) The tissue of eight samples was divided into two sections based on the histopathological features, including gland and lymphocyte infiltration sections. (B) Volcano plot of differentially expressed genes between the SS-gland and SS-lymphocyte infiltration sections. (C) Violin plot showed that CCL19 was significantly higher in the lymphocyte infiltration area compared with the gland area in SS. (D) Volcano plot of differentially expressed genes between the ctrl- and SS-gland infiltration sections. (E) Violin plot of IGHG4 exhibited a significant increase in Ctrl-gland sections compared with SS-gland sections. (F, G) Spatial feature plot of CCL19 and IGHG4 expression. (H) T-SNE reduction plot for all spatial metabolism sections. (I) Heatmap of the metabolites in each section. (J) Sankey diagram showing distribution of marker genes and metabolites in each section

Linoleic acid metabolism reprogramming in SS

Our spatial metabolomic profiling identified linoleic acid metabolism as a key pathway undergoing significant reprogramming in SS tissues. The OPLS-DA results indicated high discrimination between the two groups based on DEMs (Fig. 4A). Subsequently, metabolic pathway enrichment analysis was performed, and the identified DEMs were enriched in linoleic acid metabolism, ascorbate and aldarate metabolism, pyrimidine metabolism, and the biosynthesis of unsaturated fatty acids (Fig. 4B). Unlike previous reports, glucose metabolism was not enriched in our study [8, 9]. Among the four DEMs enriched in linoleic acid metabolism, lecithin (C00157) was downregulated in SS, while linoleic acid (C01595), 13(S)-hydroxyoctadecadienoic acid (HODE) (C14762), and dihomo-γ-linolenate (C03242) were upregulated compared to the control group. The levels of these four metabolites were strongly correlated (Fig. 4C). The spatial expression patterns of the four metabolites in the SS samples are shown in Fig. 4D. Linoleic acid, 13(S)-HODE, and dihomo-γ-linolenate exhibited nearly identical spatial distributions, while lecithin exhibited an opposing spatial expression pattern. This suggests that a substantial amount of lecithin is consumed in SS, promoting the generation of linoleic acid and its downstream metabolites. However, differential expression of genes encoding enzymes involved in this pathway was not observed in the present study. For example, PLA2G and LOX2S were not detected in the scRNA-seq and ST data. The ALOX15 expression in the SS and control groups did not show a statistically significant difference. This intriguing disconnect between metabolite levels and the expression of canonical biosynthetic enzymes prompts us to consider alternative mechanisms. First, non-enzymatic oxidation, driven by reactive oxygen species (ROS) which are often elevated in inflammatory environments like SS, could contribute to the generation of oxidized lipids such as 13(S)-HODE. Second, the metabolites may be produced through enzymatic activity in cell types that were under-sampled in our sequencing data or by isoforms not captured by our assays. Third, technical differences between the platforms (e.g., the sensitivity of metabolomics vs. transcriptomics) could also be a factor. While our current data cannot definitively distinguish these possibilities, the clear spatial co-localization of the metabolites strongly suggests active local metabolism. Therefore, we have toned down the conclusion to focus on the observed metabolic shift itself.

Fig. 4.

Reprogramming of linoleic acid metabolism in SS. (A) OPLS-DA of differentially expressed metabolites between the SS and control groups indicated high discrimination (B) Pathway analysis of differentially expressed metabolites between the SS and control groups. The identified DEMs were siginificantly enriched in linoleic acid metabolism. (C) Correlation analysis of differentially expressed metabolites between the SS and control groups. (D) MSI images showing the abundance of representative differential metabolites of linoleic acid metabolism in the SS samples

Metabolic alterations in lymphocyte infiltrates in SS

A comparative analysis of the spatially defined regions revealed that the lymphocyte infiltrates (SS-B) possess a unique metabolic profile, distinct from both the patient’s own glandular areas (SS-A) and healthy tissues (Ctrl-A). OPLS-DA results revealed high discrimination of DEMs between the ctrl-A and SS-A regions. The DEMs between regions A and B in the SS samples also exhibited high discrimination, indicating metabolic reprogramming in the affected tissues, particularly in the lymphocyte infiltrates of SS (Fig. 5A, C). KEGG analysis revealed that the DEMs in the ctrl-A region versus the SS-A region and the SS-A region versus the SS-B region were enriched in glycerophospholipid metabolism and glycine, serine, and threonine metabolism pathways (Fig. 5B, D). A metabolite with a mass-to-charge ratio of 788.54338 was identified as phosphatidylserine (PS) (36:1), and its expression levels significantly increased in SS-A regions compared to Ctrl-A regions(p < 0.001, t-test), with further elevation observed in SS-B regions relative to SS-A regions (Fig. 5E). PS is the most abundant negatively charged phospholipid in the inner leaflet of the cellular membrane and is associated with apoptosis, suggesting that apoptosis may be active in the lymphocyte infiltrates of patients with SS [15]. CD74 was a non-polymorphic type II transmembrane glycoprotein and was highly expressed in activated CD4⁺ T cells [16]. HLA-DRA was relevant for the formation of anti-Ro/La autoantibodies in SS [17]. Furthermore, we observed that the expression levels of CD74 and HLA-DRA increased sequentially (p<0.01, t-test) in the ctrl-A, SS-A, and SS-B regions, which is consistent with the distribution of PS (36:1) (Fig. 5F–I).

Fig. 5.

Combined analysis of marker genes and metabolites in each section. (A) OPLS-DA of the differentially expressed metabolites between the Ctrl-A and SS-A groups. (B) KEGG analysis of differentially expressed metabolites between the Ctrl-A and SS-A groups. (C) OPLS-DA of differentially expressed metabolites between the SS-A and SS-B groups. (D) KEGG analysis of differentially expressed metabolites in the SS-A and SS-B groups. (E) MSI and intensity analysis of phosphatidylserine (PS)(36:1) (m/z 788.54338) by t-test. The expression level of PS(36:1) significantly increased in SS-A regions compared to Ctrl-A regions, with further elevation observed in SS-B regions relative to SS-A regions. ***: p < 0.001. (F, G) Spatial expression and distribution of CD74 and HLA-DRA in each sample. (H-I) Violin plots of CD74 and HLA-DRA expression in each section

Metabolic changes of CD4⁺ T cells in SS

To directly link the observed metabolic changes to a key immune population, we zoomed in on CD4⁺ T cells. Analysis of labeled spots representing CD4⁺ T cells in SS samples was performed using RCTD (Fig. 6A). Immunohistochemical staining in SS patients confirmed the pivotal role of CD4⁺ T cells within lymphocyte foci (Fig. 6B). OPLS-DA revealed a distinguishable separation of CD4⁺ T cell metabolites between the SS-A and SS-B regions (Fig. 6C). Subsequently, the DEMs and genes in CD4⁺ T cells between the SS-A and SS-B regions were visualized using volcano plots (Fig. 6D, E). However, KEGG analysis did not identify any significantly enriched metabolic pathways (Figs. 6F, S4, and S5). CXCL13 has been previously associated with various autoimmune diseases, such as rheumatoid arthritis, multiple sclerosis, systemic lupus erythematosus, and primary SS [18]. In this study, we found that the expression levels of CXCL13 in CD4⁺ T cells from the Ctrl-A, SS-A, and SS-B groups increased progressively (Fig. 6G), suggesting its potential involvement in the local migration and aggregation of CD4⁺ T cells, or the induction of the chemotactic function of aggregated CD4⁺ T cells (p < 0.001, t-test). Upon mapping these genes onto metabolic images, CXCL13 corresponded to areas resembling high levels of PS(36:1) (m/z 788.54338). Furthermore, PS(36:1) exhibited higher expression in the SS-B region than in the SS-A region (Fig. 6H). The relationship between CXCL13 and PS(36:1) has not been reported. In this study, the spatial distribution and expression trend analyses of CXCL13 and PS(36:1) reveal a significant correlation. This spatial association may suggest shared regulatory inputs or functional interplay within the lymphocyte foci, possibilities that require further experimental validation.

Fig. 6.

Combined analysis of marker genes and metabolites of CD4⁺ T cells in each section. (A) Robust cell type decomposition (RCTD) analysis of CD4⁺ T cells in SS samples. (B) OPLS-DA analysis of the differentially expressed metabolites of CD4⁺ T cells in the SS-A and SS-B groups. (C) Volcano plot of differentially expressed genes in CD4⁺ T cells between the SS-A and SS-B groups. (D, E) Volcano plot of differentially expressed metabolites and genes in CD4⁺ T cells between the SS-A and SS-B groups. (F) KEGG analysis of differentially expressed metabolites in CD4⁺ T cells between the SS-A and SS-B groups. (G) Spatial expression of CXCL13 and its distribution in each sample. (H) MSI images and intensity analysis of PS(36:1) (m/z = 788.54338) by t-test. The expression level of PS(36:1) in CD4⁺ T cells significantly increased in SS-B regions relative to SS-A regions. ***: p < 0.001

We further investigated the extent of CD4⁺ T cell influence on local glycerophospholipid metabolism. Significant changes in glycerophospholipid metabolism have been previously reported in the serum [19]. In this study, we found enriched DEMs in the glycerophospholipid metabolism pathway in SS salivary glands (Fig. 5B, D). We performed Euclidean distance analysis between the spatial expression distribution of key regulatory genes (CRLS1, ETNK1, ETNK2, GPD1, GPD1L, GPD2, LPGAT1, LYPLA2, MBOAT7, PEMT, PGS1, PISD, PNPLA6, PNPLA7, PTDSS1, PTDSS2, and TAZ) in this pathway and the spatial location of CD4⁺ T cells (Fig. 7A). The expression levels of glycerophospholipid metabolism regulatory genes in the salivary gland tissue increased gradually as the distance from the CD4⁺ T cells decreased. Next, we examined the spatial abundance distribution of key metabolites [dimethylethanolamine, glycerophosphocholine, glycerylphosphorylethanolamine, LPC(22:6), PC(32:0), PC(33:1), PC(33:2), PC(34:1), PC(34:2), PC(36:1), PC(42:2), PE(36:4), PE(38:5), and PS(36:1)] in this pathway and the spatial location of CD4⁺ T cells (Fig. 7B). The abundance of glycerophospholipid metabolites increased as the distance from CD4⁺ T cells decreased. Among them, LYPLA2 and PS(36:1) showed the most representative changes, consistent with the overall trend of glycerophospholipid metabolism. We performed a co-expression analysis of LYPLA2 and PS(36:1) with the spatial distribution of CD4⁺ T cells in the spots (Fig. 7C, D).

Fig. 7.

Reprogramming of glycerophospholipid metabolism in CD4⁺ T cells. (A) Euclidean distance analysis from CD4⁺ T cells to all other points showing the expression of glycerophospholipid metabolism marker genes. (B) Euclidean distance analysis from CD4⁺ T cells to all other points showing the abundance of marker metabolites involved in glycerophospholipid metabolism. (C) Co-expression analysis of LYPLA2 and CD4⁺ T cells. (D) Co-expression analysis of PS(36:1) and CD4⁺ T cells

Targeting PS(36:1) synthesis ameliorates SS-like symptoms in vivo

Based on our human spatial multi-omics findings, we hypothesized that PS(36:1) represents a novel therapeutic target for SS. Phosphatidylserine synthase 1 (PTDSS1), the rate-limiting enzyme in PS biosynthesis, played a pivotal role in phospholipid metabolism. To investigate this potential therapeutic avenue, we utilized the selective PTDSS1 inhibitor DS55980254, as an inhibitor of PS(36:1) synthesis, to pharmacologically modulate PS production. Following a 28-day therapeutic regimen, quantitative analysis revealed improvements in both salivary secretion rates and histopathological parameters in NOD/ShiLtj mice. As illustrated in Figs. 8A-C, PTDSS1 inhibition markedly (p<0.05, t-test) reduced the formation of lymphocytic foci in salivary gland tissue while effectively restoring salivary flow, thereby ameliorating xerostomia symptoms. Immunofluorescence analysis further elucidated the mechanism of action, demonstrating substantial reductions (p<0.05, t-test) in both total T cell (CD3⁺) infiltration and CD4⁺ T cell populations within treated specimens (Figs. 8D-E).

Fig. 8.

PTDSS1 inhibitor inhibited CD4⁺ T cells and alleviated SS-like responses in NOD/ShiLtj mice. (A) Histopathological analysis of the submandibular glands of 12-week-old NOD/ShiLtj mice from the indicated groups (five mice per group) by H&E staining. (B) Quantitative assessment of lymphocytic foci in the indicated groups. (C) Salivary flow assay for the indicated groups. (D) Immunofluorescence analysis of T cell infiltration in submandibular glands (five mice per group). CD4⁺ T cells (red) and total T cells (CD3⁺, green) are shown in representative images. (E) Quantification of CD3⁺ and CD4 + T cell populations in the indicated groups. *vs NOD p<0.05

Discussion

SS is a systemic autoimmune disease that primarily affects exocrine glands, particularly the salivary and lacrimal glands. Previous studies have used scRNA-seq of peripheral blood and labial salivary gland tissue from patients with SS [20, 21] to characterize the immune composition and transcriptional characteristics of the salivary glands. Additionally, modifications in various metabolic pathways, including glucose and lipid metabolism, have significant effects on the activation and function of immune cells and are important features of autoimmune diseases. However, the spatial features of transcription and metabolic reprogramming in SS salivary gland tissues have not yet been reported. In this study, we integrated scRNA-seq, ST, and SM to delineate the spatial transcriptomic and metabolic patterns in the salivary gland tissue of patients with SS at single-cell resolution.

ScRNA-seq revealed shifts in the composition of labial salivary gland cell populations. For instance, the decrease in acinar cells and the increase in fibroblasts in patients with SS indicate glandular damage and fibrotic repair. CD4⁺ T-lymphocytes play a crucial role in the development and occurrence of SS. CD4⁺ T cells play a crucial role in the production of autoantibodies, lymphocytic infiltration, and destruction of salivary gland epithelial cells through the release of various cytokines [22]. Additionally, patients with SS exhibit a decrease in the numbers of CD4⁺ T cells in the peripheral blood and a significant increase in CD4⁺ T cell infiltration in the affected salivary glands, supporting the hypothesis that the decrease in the numbers of peripheral lymphocytes is facilitated by the migration of CD4⁺ T cells into the tissues [23]. Further T cell subtyping underscored the importance of T cells and their subtype interactions in SS development. In the present study, through ST and SM analyses, we further elucidated the spatial characteristics of gene expression and metabolite abundance changes in the labial salivary gland tissue of patients with SS. T cells exhibited specific clustering in SS, i.e., we observed an increased number of Treg cells and significant interactions with other cells, further confirming their critical role in SS. Based on pathological features of SS tissues, the samples were segmented into gland areas (Zone A) and lymphocyte infiltration areas (Zone B) in the sliced sections. CCL19 was significantly upregulated in SS-B zone compared to SS-A zone. CCL19, located on the p-arm of chromosome 9, encodes a chemokine that recruits and activates immune cells and induces the formation of tertiary lymphoid structures [24]. Upregulation of CCL19 and its receptor CCR7 has been linked to focal accumulation of dendritic, B-, and CCR7⁺ central memory T cells [25]. It is also associated with the presence of anti-SS-A antibodies and increased IgG levels in SS. Moreover, the frequency of CCR7⁺ CD4⁺ T cells increased in patients with SS and was closely correlated with the EULAR SS Disease Activity Index, suggesting that CCL19/CCR7 may participate in the development of SS by mediating the migration of CD4⁺ cells [26]. IGHG4 was significantly upregulated in the SS-A zone, compared to the case in the ctrl-A zone; this may be related to changes in the autoantibody profile of SS.

Spatial metabolomics revealed distinct metabolite distributions among the SS-A, SS-B, and ctrl-A areas. Subsequently, a metabolic-transcriptomic correlation network was constructed, and multi-omics integration analysis was performed. In SS, linoleic acid metabolism was significantly reprogrammed, with downregulation of lecithin expression and upregulation of linoleic acid, 13(S)-HODE, and dihomo-γ-linolenate. Lecithin promotes adipocyte differentiation and activates PPARA [27], while Chang et al. [28] found that PPARA activation inhibits Th17 differentiation through the IL-6/STAT3/RORγt pathway. Linoleic acid can drive Stat1 phosphorylation at Ser727 by specifically binding to AHR, inhibiting IL-17, and enhancing Foxp3 expression, thereby suppressing Th17 differentiation and promoting Treg cell differentiation [29]. Linoleic acid generates 13(S)-HODE and dihomo-γ-linolenate. Furthermore, 13(S)-HODE enhances NF-κB activity by phosphorylating I-κ kinase-β-IκB [30]. The development of SS is associated with activation of the NF-κB pathway in acinar cells, ductal epithelial cells, and lymphocytes [31]. In mice, activation of the NF-κB pathway can result in reduced saliva production and entry of a T cell-rich infiltrate into the salivary glands [32]. Therefore, it is speculated that linoleic acid metabolic reprogramming contributes to lymphocyte infiltration and Th17/Treg imbalance in SS.

Lymphocytic infiltration in salivary gland tissue is a defining pathological feature of SS. In this study, we observed progressively elevated levels of PS(36:1), CD74, and HLA-DRA from ctrl-A to SS-A to SS-B regions. PS is a glycerophospholipid conventionally recognized as a signal for the removal of apoptotic cells, indicating active apoptosis of lymphocytic infiltrates. PS can act as a negative regulator of T cell function by inducing CD4⁺ T cell apoptosis mediated by T cell immunoglobulin and mucin domain (TIM)-3 [33]. In contrast, the expression of TIM-3 in CD4 + T cells in SS decreased significantly; this may account for their excessive activation and proliferation in the salivary glands [34]. Therefore, the role of elevated PS(36:1) in lymphocytic infiltration requires further investigation. CD74 is a non-polymorphic type II transmembrane glycoprotein that plays important roles in antigen presentation and triggers autoimmune diseases, such as systemic lupus erythematosus and SS [35]. In SS, increased expression of HLA-DRA in salivary gland epithelial cells suggests their involvement as nonprofessional APCs in the immune response [36]. In this study, the increased expression of PS(36:1) was mapped to the distribution of CD74 and HLA-DRA, providing objective evidence linking these three factors.

CD4⁺ T cell dysregulation is central to the immunopathogenesis of SS. The abnormal activation of CD4⁺ T cells in SS is closely associated with metabolic reprogramming. Previous studies have shown that glutamine metabolism is closely linked to the activation and proliferation of CD4⁺ T cells, particularly Th17 and Th1 cells [37]. Fu et al. [38] reported the metabolic reprogramming of CD4⁺ T cells in SS, specifically alterations in glucose metabolism, supporting enhanced metabolic demands following activation. It is noteworthy that our spatial metabolomic analysis did not identify significant enrichment of glucose metabolism pathways, which differs from some prior reports. This discrepancy may stem from several factors. Previous studies highlighting glucose metabolism changes often focused on peripheral blood mononuclear cells (PBMCs) or in vitro stimulated T cells [38], whereas our study analyzed whole labial salivary gland tissue using spatial omics. Within this complex microenvironment, infiltrating immune cells (e.g., T cells) may undergo metabolic reprogramming as a result of intercellular crosstalk with resident glandular cells (e.g., acinar and ductal cells). Owing to limitations in sample size and the current resolution constraints of ST and SM platforms, subtle yet biologically meaningful changes in glucose metabolism may not have surpassed our stringent statistical thresholds. Nevertheless, such alterations could still exert functional effects within specific cellular subsets. In vitro inhibition of these metabolic pathways alleviates SS-like symptoms [39], highlighting the therapeutic potential of targeting CD4⁺ T cell metabolism.

In this study, we analyzed changes in the spatial expression of genes and metabolites in CD4⁺ T cells using RCTD. Owing to the limited number of samples, KEGG analysis did not significantly enrich the characteristic metabolic pathways. We found that CXCL13 expression was elevated in CD4⁺ T cells from the SS-B region, compared to the SS-A region. CXCL13 was initially identified as a B-cell chemokine because of its strong chemotactic effects on B-cells. Previous studies have shown that CXCL13 is overexpressed in lymphocytic infiltrates in the salivary glands of patients with SS, further promoting B cell expansion, local antibody production, and lymphoma formation [18]. The expression levels of CXCL13 and its receptor CXCR5 increase with disease progression [40], and CXCR5⁺ B cells can form ectopic germinal centers in SS [18]. The present study reveals that heightened PS(36:1) expression in tissue-infiltrating CD4 + T cells may facilitate their activation and expansion, consistent with mechanisms described previously [33, 34]. The prevalence of Tregs among CD4⁺ T cells in our RCTD data aligns with their reliance on fatty acid oxidation, underscoring the pathogenic relevance of Treg metabolism in SS.

Glycerophospholipid metabolism has been reported to be reprogrammed in various autoimmune diseases. In rheumatoid arthritis, the dysregulation of glycerophospholipid metabolism is strongly negatively correlated with C-reactive protein levels [41]. Significant alterations in glycerophospholipid metabolism have also been observed in the sera of patients with SS [19]. Here, we investigated the effect of CD4⁺ T cells on the reprogramming of glycerophospholipid metabolism in SS tissues at both the transcriptional and metabolic levels. The results revealed that key regulatory genes and metabolites in this pathway exhibited increased expression with a reduction in the distance from CD4⁺ T cells, indicating that CD4⁺ T cells upregulate glycerophospholipid metabolism in the surrounding cells, particularly regarding LYPLA2 and PS (36:1).

We found that PS (36:1) exhibited significantly elevated expression levels not only in infiltrating immune cells (particularly CD4⁺ T cells) within the salivary glands of Sjögren’s syndrome (SS) patients, but also manifested a distinct reprogramming phenotype in glandular epithelial cells adjacent to lymphoid foci. In vivo inhibition of PS(36:1) synthesis using DS55980254 in NOD/ShiLtj mice ameliorated SS-like symptoms, including dry mouth and lymphocyte infiltration in the salivary glands. Furthermore, treatment with DS55980254 significantly reduced CD4⁺ T cell infiltration. These results suggest that PS(36:1) may serve as a promising therapeutic target for SS intervention.

This study still has certain limitations. Most importantly, the sample size (n = 4 per group) limits the statistical power and generalizability of our findings. While our multi-omics approach provided deep insights, the cohort size may have been insufficient to detect more subtle or heterogeneous alterations, such as the previously reported changes in glucose and amino acid metabolism which were not observed here. Consequently, the prevalence and effect sizes of the identified lipid metabolism changes should be interpreted with caution and require validation in larger, independent cohorts. Future large-scale studies are essential to confirm and extend our observations on the metabolic and transcriptomic landscape of SS.

In conclusion, spatial multi-omics profiling uncovered distinct transcriptomic and metabolic characteristics of the salivary glands in SS. We examined the transcriptomic and metabolic profiles of CD4⁺ T cells, which play crucial roles in SS pathogenesis. We found that different cell types exhibited specific spatial distributions within the salivary glands and that immune cells, such as CD4⁺ T cells, were key players in the development of SS. Our metabolomic analysis revealed dysregulated linoleic acid metabolism and aberrant lipid metabolite profiles in SS patients. Notably, we identified PS(36:1) as a differentially expressed phospholipid that may represent a promising therapeutic target for SS intervention. These findings provide new insights into the development of novel therapeutic strategies and targeted drugs for the treatment of SS.

Materials and methods

Human labial gland samples

The study included eight human labial gland samples: four from patients with SS and four from healthy controls. Labial gland biopsy is the gold standard for diagnosing SS. The SS labial glands were diagnosed based on clinical and histopathological criteria, which required the presence of lymphocytic infiltrates with a minimum of 50 lymphocytes per high-power field. The control group consisted of healthy labial glands obtained from the excision of labial gland cysts. Upon isolation, acinar tissue was removed from the scRNA-seq samples of one SS patient and one healthy control. The samples were subsequently washed in physiological saline to eliminate residual blood. Subsequently, they were stored in a tissue preservation solution at 4 °C and prepared as single-cell suspensions within 12 h. Labial gland samples obtained from four SS patients and four matched healthy controls were embedded in OCT compound and stored at -80 °C for subsequent ST and SM analyses. The study of human labial gland samples was performed according to the Declaration of Helsinki and Good Clinical Practice and was approved by the Ethics Committee of Shanghai East Hospital (grant no. 2023 − 132). Written informed consent was obtained from all participants.

ScRNA-seq

Detailed steps for the scRNA-seq of labial gland tissues can be found on the 10X Genomics website (https://www.10xgenomics.com/). Fresh labial gland samples were extracted from one patient with SS and one healthy control, and single-cell suspensions were prepared. After quality control, single-cell gel bead emulsions were generated using microfluidic technology, and cDNA libraries were prepared. The constructed libraries were subjected to high-throughput sequencing on an Illumina Nova 6000 PE150 platform.

Spatially resolved transcriptomics

The embedded labial gland tissues were sliced into 10-µm-thick consecutive frozen sections using a cryostat (Leica CM 1950, Leica Microsystem, Germany) at low temperature. The sections were placed on an ST microarray. After dehydration with isopropanol for 1 min, hematoxylin and eosin (H&E) staining was performed. Bright-field images were captured using a 3D HISTECH Pannoramic MIDI FL whole-slide scanner at a resolution of 20×. Library preparation slides were purchased from the Spatial Transcriptomics team at 10X Genomics (https://www.10xgenomics.com/). Each spot printed on the array had a diameter of 55 μm, with a center-to-center distance of 100 μm between two spots, covering an area of 6.5 × 6.5 mm². Each slide included four capture areas with approximately 5,000 unique gene expression points in each capture area. The tissue sections were placed on the corresponding capture areas of the visible tissue optimization slides. After being placed on the gene expression slides, tissue sections were fixed, stained, and permeabilized. Subsequently, cellular mRNA was captured by primers on the spots. All cDNA generated from the mRNA captured at specific spots had a common spatial barcode. Libraries were generated from the cDNA, followed by sequencing. Spatial barcodes were then used to map the sequencing reads back to their specific locations on the tissue section images, enabling the generation of spatial gene expression maps.

AFADESI-MSI

The embedded labial gland tissues were sliced into 10-µm-thick consecutive frozen sections using a cryostat (Leica CM 1950, Leica Microsystem, Germany) at low temperature and then placed on positively charged glass slides (Thermo Scientific, USA). MSI scan analysis was performed in accordance with the methods described by Luo et al. [42] For the experiments, an AFADESI-MSI platform (Beijing Victor Technology Co., LTD, Beijing, China) coupled with a Q-Orbitrap mass spectrometer (Q Exactive, Thermo Scientific) was used. The spray solvents for the negative and positive ion modes were acetonitrile/water (8:2) and acetonitrile/water (8:2, 0.1% formic acid), respectively. During the MSI scan, the scanning speed in the X direction was 0.2 mm/s, and the scan in the Y direction proceed with a step size of 40 μm.

Single-cell transcriptome and ST quality control and data preprocessing

Library preparation, sequencing, and data analysis were conducted by Shanghai OE Biotech Co., Ltd. The raw data generated from the high-throughput sequencing were in the FASTQ format. The official software Cell Ranger (version 7.0.1) from 10X Genomics was used to perform quality statistics on the raw data. Further quality control was performed using the “Seurat” package (version 4.0.0) for downstream analysis. Once the quality control was completed, the “NormalizeData” function from the “Seurat” package was used to standardize the single-cell transcriptomic data. For spatial transcriptomic data, the “sctransform” function was used for normalization, detection of highly variable features, and storage of data in the SCT matrix.

Single-cell transcriptome and ST data analysis

The “FindVariableGenes” function in the “Seurat” package was used to identify highly variable genes, and principal component analysis was performed for dimensionality reduction. The results were visualized in two-dimensional space using UMAP, a nonlinear dimensionality reduction technique. Marker genes were identified using the “FindAllMarkers” function in the “Seurat” package, and the results were visualized using the “VlnPlot” and “FeaturePlot” functions. The “FindMarkers” function from the “Seurat” package was used to identify DEGs. Genes with a p-value < 0.05 and fold change > 1.5 were selected as significantly DEGs. Gene Ontology and KEGG enrichment analyses were performed on significantly DEGs using hypergeometric distribution testing.

For single-cell transcriptomics, the “SingleR” package (version 1.4.1) was used to identify the cell types based on a common reference dataset. The “CellPhoneDB” package (version 4.1.0) was used to identify ligand-receptor interaction pairs in the single-cell transcriptomic data. Cellular expression profiles were used to detect these interactions. The R packages “Igraph” and “Circlize” were used to graphically visualize the cell communication network. For spatial transcriptomics, the “RCTD” package (version 1.1.0) was used in conjunction with the single-cell transcriptomic data to infer the cellular composition at each spot location. Cell communication was analyzed using the “StLearn” package (v0.4.11). Additionally, a reference point containing CD4⁺ cells was selected for spatial regions. The Euclidean distance was calculated from this reference point to all other points, and density curves and spatial distribution maps were generated based on this distance [43, 44].

Spatial metabolome data analysis

The raw mass spectrometry data in “.raw” format were converted into the “.imzML” format using imzMLConverter software. After conversion, the Cardinal software package was used to perform background subtraction and reconstruct ion images. After ion image reconstruction, DEMs were selected using OPLS-DA. The variable importance in projection values were used to assess the overall contribution of each metabolite to the differentiation between groups. Metabolites with variable importance in projection values > 1.0 and p-values < 0.05 were considered DEMs.

The ions detected using AFADESI were annotated using the pySM SM annotation framework along with the local metabolite database SmetDB (Lumingbio, Shanghai, China). The pySM framework provides tools and algorithms for spatial metabolomic analysis, including metabolite annotation. By leveraging information from the local metabolite database SmetDB, the detected ions were matched to known metabolites based on their mass-to-charge ratios (m/z) and other characteristics. Using PySM and SmetDB, the detected ions were annotated with metabolite identities, providing valuable information regarding the chemical composition and potential biological relevance of the detected species.

Animals

Eight-week-old female NOD/ShiLtj mice and wild-type ICR mice were obtained from the Model Animal Research Center of Nanjing University (China). Three experimental groups (five mice per group) were established for in vivo studies: group 1 (wild-type ICR mice receiving solvent treatment as vehicle control), group 2 (NOD/ShiLtj mice receiving solvent treatment as disease control), and group 3 (NOD/ShiLtj mice treated with 1 µg/g DS55980254). All groups received intraperitoneal injections every 48 h for 28 days. Following anesthsia induction with 0.24 mg/g 2,2,2-tribromoethanol (T48402; Sigma-Aldrich, St. Louis, MO, USA), salivary flow rates were measured. Subsequently, mice were euthanized for submandibular gland collection. All animal procedures strictly complied with the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health and followed the guidelines of the Animal Welfare Act. This study was approved by the Ethics Committee of Shanghai East Hospital (grant no. 2023 − 132).

Data registration of ST and SM

Both the barcode spatial information of the ST and the pixel points information of the SM were transformed into unified spatial information identifiers. Then, the transformed pixel points of the SM were aligned to the spots from the ST via a summation operation on the ion data of each pixel point corresponding to each spot. Finally, This process resulted in reintegrated SM data that were spatially aligned with the ST data.

HE staining, immunohistochemistry, and immunofluorescence staining

HE staining, immunohistochemistry, and immunofluorescence staining were performed using 4-µm sections. For immunohistochemistry, the primary antibody used was anti-CD4 (1:200 dilution, catalog number GB13588, Wuhan Servicebio Technology, Wuhan, China). For immunofluorescence staining, the following primary antibodies were employed: anti-CD3 (1:100 dilution, catalog number GB13014, Wuhan Servicebio Technology, Wuhan, China) and anti-CD4 (1:200 dilution, catalog number GB13064-2, Wuhan Servicebio Technology, Wuhan, China). The experimental procedures were strictly conducted in accordance with the manufacturer’s instructions.

Salivary flow assay

Mice were anesthetized by intraperitoneal administration of 2,2,2-tribromoethanol (diluted in phosphate-buffered saline). Following anesthesia, pilocarpine (0.125 mg/kg; Sigma-Aldrich, St. Louis, MO, USA) was administered intraperitoneally. After 10 min of pilocarpine stimulation, saliva was collected from the oral cavity over a 5-minute period using sterile dry cotton balls. Saliva production was quantified by weighing the collected samples, and the values were normalized to the animals’ body weight (approximately 25 g).

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Yanxiong Shao, Ninging Cao, and Fei Qian conducted the experiments, analysed the data, and drafted the manuscript. Diqing Wu, Yuting Yang, Yichao Xia, Zongze Shen, Lijuan Zhu, Jiajia Li, Yeping Lu, Chaoran Li, Jingwen Yang, Junqin Lu, Zhenrong Hu and Xudong Wang provided technical and data analysis assistance. Ying Song, Tianyu Xiao and Tianlin Lu provided the clinical samples. Jie Zhang and Yubo Xu conceived and designed the study. All authors have read and approved the article.

Funding

This work was supported by Shanghai East Hospital Youth Scientific Research Cultivation Fund (DFPY2022005), Shanghai East Hospital Scientific Research Start-Up Fund for the Introduction of Talent (DFRC2021006).

Data availability

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2021) in National Genomics Data Center (Nucleic Acids Res 2022), China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA010203 and HRA010153) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human.

Declarations

Ethics approval and consent to participate

The study of human labial gland samples was performed according to the Declaration of Helsinki and Good Clinical Practice and was approved by the Ethics Committee of Shanghai East Hospital (grant no. 2023 − 132). Written informed consent was obtained from all participants.

Consent for publication

All authors consent to the submission and publication of this article.

Competing interests

The authors declare no competing interests.