C57BL/6 (stock # 000664) mice were purchased from Jackson Laboratories (Bar Harbor, ME) and IL-14αTG mice were housed at the Christopher S. Bond Life Sciences Center Animal Facility of the University of Missouri (Columbia, MO). Animals were housed in vented cages with 12 h light/dark cycles and received food and water ad libitum. Age-matched female mice were used in all experiments and genotyping was performed by PCR, as previously described (16). Euthanasia was performed by terminal anesthesia with isoflurane followed by cervical dislocation, with efforts taken to minimize suffering. All experimental animal procedures were conducted in accordance with National Institutes of Health guidelines that were approved by the University of Missouri Animal Care and Use Committee (Protocol Number 38921).

For bright field microscopy, mouse SMGs and sublingual glands (SLGs) were excised, fixed in 4% (v/v) paraformaldehyde for 24 h and dehydrated by incubation in 70% (v/v) ethanol for 24 h at 4°C. Samples were sent to IDEXX BioAnalytics (Columbia, MO), where they were embedded in paraffin, sectioned and stained with hematoxylin and eosin. Stitched images of whole submandibular and sublingual glands were captured on a Leica DMI6000B inverted microscope using LAS X software.

For immunofluorescence analysis, SMGs and SLGs were excised, embedded in OCT compound, snap frozen in 2-methylbutane cooled with liquid nitrogen and cryosectioned on a Leica CM3050 cryostat at the University of Missouri Advanced Light Microscopy Core (UMALMC) Facility. Cryosections were adhered to slides and then fixed in 4% (v/v) paraformaldehyde for 20 min at room temperature or in acetone at −20°C for 10 min (for detection of GL7), washed in PBS and placed for 1 h at room temperature in blocking buffer containing 5% (v/v) goat serum in PBS with mouse BD Fc Block (1:250, BD Biosciences). Sections were then stained for 16 h at 4°C in blocking buffer containing the following conjugated primary antibodies: AlexaFluor 594 rat anti-B220 (1:200, Biolegend clone RA3-6B2), AlexaFluor 488 hamster anti-CD3 (1:200, ThermoFisher clone 145-2C11), Brilliant Violet 421 rat anti-CD169/Siglec-1 (1:200, Biolegend clone 3D6.112), AlexaFluor 594 hamster anti-CD11c (1:200, Biolegend clone N418) or FITC rat anti-GL7 (1:250, BD Biosciences clone GL7). For unconjugated primary antibodies, sections were stained for 16 h at 4°C in blocking buffer containing rat anti-CD45 (1:100, Biolegend clone 30-F11) and rabbit anti-Aquaporin 5 (1:100, MilliporeSigma #178615) antibodies, washed in PBS, then stained with AlexaFluor 594 goat anti-rat IgG (1:1,000, ThermoFisher #A-11007) and AlexaFluor 488 goat anti-rabbit IgG (1:1,000, ThermoFisher #A-11008) secondary antibodies for 1 h at room temperature. Slides were then washed in PBS and coverslips mounted using Fluoroshield with DAPI (MilliporeSigma). Stitched and high-magnification SMG and SLG images were captured on a Leica STELLARIS 5 confocal microscope using LAS X software.

SMGs were isolated, finely minced in digestion media [RPMI-1640 media containing 5% (v/v) fetal bovine serum, 2 mM EDTA, 2.5 mM CaCl₂ and 1 mg/ml collagenase D (MilliporeSigma)], and placed in a shaking incubator for 2 h at 37°C and 270 rpm. Dispersed SMGs and isolated spleens were then passed through a 40 µm cell strainer, washed in PBS and spleen homogenates were resuspended in red blood cell lysis buffer (Miltenyi Biotec) for 10 min in the dark. Following a wash in PBS, cells were pelleted by centrifugation at 500×g at 4°C, resuspended in PBS, counted and aliquoted for analysis.

For antibody staining, 10⁶ cells were resuspended in 100 µl of PBS containing Zombie NIR viability dye (1:2,000) and mouse BD Fc Block (1:100), incubated for 15 min in the dark, and then washed in cytometry buffer [0.5% (w/v) bovine serum albumin, 2 mM EDTA in PBS]. Cells were then resuspended in cytometry buffer containing 1:100 dilutions of each antibody in either the B cell or T cell panel (Table 1) and incubated for 1 h at 4°C in the dark. Following a 5 min wash in cytometry buffer, cells were fixed and permeabilized using the eBioscience FoxP3/transcription factor staining buffer set (ThermoFisher) per the manufacturer's protocol. Cells were then resuspended in eBioscience permeabilization buffer containing 1:100 dilutions of antibodies targeting intracellular antigens (denoted with # in Table 1) and incubated for 16 h at 4°C in the dark. Cells were then washed twice and resuspended in cytometry buffer for analysis, with an aliquot of unstained sample reserved for autofluorescence control. Fluorescence minus one (FMO) control samples and single-stained compensation controls were similarly prepared using spleen cells for FMOs and UltraComp eBeads Plus compensation beads (ThermoFisher), respectively. Samples were analyzed on a Cytek Aurora spectral flow cytometer at the University of Missouri Cell and Immunobiology Core Facility and spectral unmixing was performed using Cytek SpectroFlo software. Representative gating strategies for each panel are shown in Supplementary Figures S1 and S2.

SMGs from 6- and 12-month-old IL-14αTG and C57BL/6 mice were excised, SLGs were removed, and SMG tissue was homogenized in 1 ml TRIzol reagent (ThermoFisher) using a handheld homogenizer. Next, 0.2 ml of chloroform was added, and samples were vortexed for 20 s and allowed to sit at room temperature for 15 min. Samples were centrifuged at 10,000×g for 18 min at 4°C and the resulting aqueous phase was passed through a genomic DNA Eliminator column from the RNeasy Plus Mini kit (Qiagen) before RNA isolation was performed per the manufacturer's protocol. Purified RNA concentration was calculated using a NanoDrop One spectrophotometer and 1 µg of RNA was used for sequencing library preparation using the Stranded mRNA Prep kit (Illumina) at the University of Missouri Genomics Technology Core (UMGTC) Facility. Total mRNA integrity was assessed on an Agilent 5200 Fragment Analyzer before whole transcriptome sequencing to a depth of at least 50 million reads per sample was performed on a NovaSeq 6000 sequencing system (Illumina).

Raw FASTQ data from mouse SMGs (n = 3/genotype/timepoint) and from human minor salivary gland biopsies of SjD patients and healthy volunteers [n = 9 healthy volunteers and 35 SjD patients; accessed through the National Center for Biotechnology Information's Database of Genotypes and Phenotypes (dbGAP) accession phs001842.v1.p1] (20) were processed by the University of Missouri Bioinformatics and Analytics Core (UMBAC) Facility. Sequence read quality was assessed using FastQC and reads were filtered and trimmed using fastp (32), then mapped against the Mus musculus reference genome GRCm39 or the Homo sapiens reference genome GRCh38 using STAR software (33). Data were normalized by variance stabilizing transformation using the DESeq2 R package (34) and sample variance was assessed by principal component analysis (PCA). Differential gene expression analysis was performed using DESeq2, where differentially expressed genes were defined as those having a greater than 2-fold change (Log2FC ≥ 1) and adjusted P value less than 0.05 [-Log10(padj) ≥ 1.3]. Volcano plots of DEGs were generated using VolcaNoseR (35) and DEG ontology, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway/biological process enrichment, and functional annotation analysis was performed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) (36, 37).

Spatial transcriptomic analysis was carried out using the poly-(dT) capture primer-based Visium v1 spatial gene expression assay from 10X Genomics (Pleasanton, CA) according to the manufacturer's protocols. Briefly, SMGs and SLGs were excised from 12-month-old IL-14αTG and C57BL/6 mice, embedded in OCT compound, snap frozen in 2-methylbutane cooled with liquid nitrogen, and cryosectioned to 10 µm thickness on a Leica CM3050 cryostat. For tissue optimization, cryosections were adhered to a Visium spatial tissue optimization slide and the optimum SMG and SLG tissue permeabilization time was determined to be 18 min. For spatial gene expression analysis, SMG and SLG cryosections were adhered to a Visium spatial gene expression slide, stained with hematoxylin and eosin, and imaged on a Zeiss Axiovert 200M inverted microscope at the UMALMC Facility. Serial tissue cryosections were also collected for immunofluorescence analysis of spatial gene expression tissues. Next, mRNA libraries were prepared and sequenced to a depth of 50,000 reads/spot on an Illumina NovaSeq 6000 sequencing system at the UMGTC Facility and FASTQ files were processed using the Space Ranger 2.0.0 pipeline at the UMBAC Facility. Spatial gene expression visualization was carried out using Loupe Browser 8.1 (10X Genomics). Unsupervised Louvain clustering using PCA embedding and a resolution of 1 was performed using BioTuring Lens, and cell cluster annotation was performed using BioTuring Talk2Data and SpatialX modules.

Flow cytometry data were analyzed using GraphPad Prism version 10.2.0. Data were first assessed for normality using Shapiro–Wilk test and data with normal distribution were analyzed by one-way ANOVA followed by Sidak correction for multiple comparisons or unpaired two-tailed t test. Data that did not fit normal distribution were analyzed by Kruskal–Wallis test followed by Dunn's test for multiple comparisons or Mann–Whitney test.

The flow cytometry data presented in this study were deposited at ImmPort Shared Data database (accession SDY3118). Raw and normalized bulk RNAseq data from IL-14αTG and C57BL/6 mice are provided in Supplementary Table S2. Bulk RNAseq data from human SjD patients and healthy volunteers are available from dbGAP (accession phs001842.v1.p1). Spatial transcriptomic datasets were deposited at the National Center for Biotechnology Information's Gene Expression Omnibus (accession GSE298921).

From 6 to 18 months of age, IL-14αTG mice develop progressive inflammation of the submandibular and parotid glands while the sublingual glands are less affected (18). Hematoxylin and eosin staining of 6- and 12-month-old IL-14αTG mouse salivary glands revealed the presence of numerous large inflammatory foci in the SMG at 12 months of age, as compared to age-matched C57BL/6 control mouse SMG, but not in the SLG (Figure 1A). These inflammatory foci were each positioned around a central blood vessel and large excretory duct (Figure 1B) and stained positive for the pan-immune cell marker CD45, whereas aquaporin 5 (AQP5)⁺ salivary acinar cells were only present in the surrounding SMG tissue (Figure 1C). Immune cell type-selective markers identified B220⁺ B cells and CD3⁺ T cells that occupied distinct areas within each focus (Figure 1C, inset). CD11c⁺ dendritic cells were present throughout the focus and the surrounding SMG tissue, while CD169⁺ macrophages were present at the focus perimeter and throughout the surrounding SMG tissue (Figure 1C, inset). Some lymphoid structures in the IL-14αTG SMG also stained positive for the activated B and T cell antigen GL7 (Figure 1D), a marker for activated B and T cells present in germinal centers (GC) where B cells undergo antibody affinity maturation and clonal expansion (38, 39).

Because the sublingual glands lacked immune cell foci and are the least affected salivary glands in SjD patients (40), flow cytometry was performed on immune cells isolated from SMGs at the 12-month time point and the results reflected histological findings, where a significant increase in CD45⁺ immune cells was observed in IL-14αTG mice as compared to age-matched C57BL/6 mice (Figure 2A). Previous studies identified alterations in innate B1 and conventional B2 cell subpopulations in IL-14αTG vs. littermate control mice (16); therefore, we compared B lymphocyte subpopulations and initially stratified cells based on CD19, B220 and CD5 expression. B1 cells, which have long been implicated in the development of autoimmunity and production of autoantibodies (41), were identified in both IL-14αTG and C57BL/6 mouse SMGs; however, no significant differences were observed between genotypes with regards to either B1a (CD19⁺, B220⁻, CD5⁺) or B1b (CD19⁺, B220⁻, CD5⁻) cell subsets (Figure 2B). In the conventional B2 cell compartment (CD19⁺, B220⁺), cell fate decisions determining differentiation into recirculating short-lived follicular B cells (CD23⁺) or long-lived static marginal zone (MZ) B cells (CD21/35⁺) are dictated by the strength of B cell receptor (BCR) signaling (42), and both follicular and MZ B cells were identified in IL-14αTG and C57BL/6 mouse SMGs (Figure 2C). Follicular B cells represented ∼1% of total immune cells in both genotypes and levels were not significantly different; however, the proportion of MZ B cells was significantly increased 5-fold in IL-14αTG mouse SMG as compared to controls (Figure 2C). GC B cells (GL7⁺) were present in the IL-14αTG mouse SMG, confirming observations of GL7 immunofluorescence staining (Figure 1D), and we observed similar levels of GL7⁺ B cells in C57BL/6 mouse control SMGs (Figure 2C). In both genotypes, long-lived memory B cells (CD38⁺, CD273⁺) represented the least abundant B cell population (∼0.6%; Figure 2C) and antibody producing plasma cells (CD19⁻, CD138⁺) were the most abundant population, representing 30%–35% of all CD45⁺ SMG-infiltrating immune cells (Figure 2D). However, no statistical differences between genotypes were observed in either population.

We next examined T cell subpopulations in the SMG and found no significant differences in the proportion of total CD3⁺ T cells (Figure 2E) or CD4⁺ T helper (TH) cells (Figure 2F) between IL-14αTG and C57BL/6 mice. Further analysis of CD4⁺ TH subsets, including TH1 (CXCR3⁺), TH2 (GATA3⁺), TH17 (RORγt⁺) and TREG (FoxP3⁺, CD25⁺) cells, did not reveal significant differences between IL-14αTG and C57BL/6 mouse SMG (Figure 2G). In both genotypes, TH17 cells made up the highest proportion of total CD45⁺ immune cells among all TH cell subsets (∼4%), whereas TH2 cells made up the lowest proportion (< 0.1%). Additionally, all SMG-infiltrating TREG cells in both genotypes were found to express the activation marker CD69 that enhances the immunomodulatory capacity of these cells (Supplementary Figure S3) (43). In contrast to the CD4⁺ TH cell populations, we observed a significant increase in the proportion of CD8⁺ T cells in C57BL/6 as compared to IL-14αTG mouse SMG (Figure 2F). Within the CD8⁺ T cell compartment, we further analyzed the proportion and activation state of the non-recirculating long-lived T resident memory (TRM) cells (44) using the TRM cell marker CD69 and the exhausted T cell marker programmed death-1 (PD-1), respectively, but found no significant differences in TRM levels nor exhaustion state between genotypes (Figure 2H).

Bulk RNAseq analysis of SMGs from 6-month-old and 12-month-old IL-14αTG and C57BL/6 mice identified upregulated and downregulated DEGs in multiple transcriptome comparisons. In the mouse SMG, dramatic changes in the transcriptomic landscape occurred during aging, where the 12-month-old vs. 6-month-old IL-14αTG mouse SMG comparison identified 1,653 total DEGs and 12-month-old vs. 6-month-old C57BL/6 mouse SMG comparison identified 508 DEGs (Table 2; Supplementary Figure S4). Comparison of age-matched IL-14αTG and C57BL/6 mouse SMGs at the 6-month and 12-month timepoints yielded 25 DEGs and 406 DEGs, respectively. We further analyzed human DEGs from bulk RNAseq analysis of minor salivary gland (MSG) biopsies from SjD patients and healthy volunteers (accessed through dbGAP accession phs001842.v1.p1) (20) and found 1,339 DEGs (Table 2). Principal component analysis of mouse SMG RNAseq data showed that IL-14αTG and C57BL/6 mouse SMGs grouped tightly at the 6-month time point, but diverged at the 12-month time point, with PC1 accounting for 79% of variance between samples compared to 7% of the variance for PC2 (Figure 3A). PCA of human minor salivary gland samples demonstrated a similar magnitude of variance across both PC1 and PC2 axes and diffuse grouping (Figure 3B). Volcano plots of DEGs in 12-month-old IL-14αTG vs. C57BL/6 mouse SMGs (Figure 3C) and SjD patient vs. healthy volunteer (HV) MSGs (Figure 3D) show the significance (-Log₁₀ padj) and magnitude (Log₂-fold change) of up- and downregulated DEGs, with annotated dots denoting the top 10 DEGs based on Manhattan distance from origin. Gene ontology analysis of all upregulated DEGs revealed a significant enrichment of KEGG pathways and biological processes associated with immune responses, autoimmunity, and viral infection with substantial overlap noted between IL-14αTG SMG (Figure 3E) and SjD patient MSG biopsies (Figure 3F). In total, 66 biological processes and 22 KEGG pathways were significantly enriched [False Discovery Rate (FDR) < 0.05] in mouse SMG and 169 biological processes (BP) and 68 KEGG pathways were significantly enriched in human MSG (Supplementary Table S1). Gene ontology analysis of downregulated DEGs did not identify any significantly enriched pathways or processes. Interestingly, gene ontology analysis of DEGs associated with aging (i.e., 12-month vs. 6-month) in either C57BL/6 or IL-14αTG mouse genotypes showed enrichment of inflammatory KEGG pathways and biological processes that mirrored those seen in diseased vs. control samples (Supplementary Table S1), corroborating recent reports on the presence of chronic inflammatory pathologies in aged C57BL/6 mouse SMG and lacrimal glands (38, 45).

Next, we compared DEGs in 12-month-old vs. 6-month-old IL-14αTG mouse SMGs and 12-month-old vs. 6-month-old C57BL/6 mouse SMGs and removed genes that were differentially expressed in both genotypes, reasoning that these genes would be associated with aging and the unique IL-14αTG mouse DEGs would be disease model-specific. In total, 457 DEGs were common between both genotypes, leaving 1,196 DEGs unique to the IL-14αTG mouse SMG disease model (Supplementary Table S1). When grouped by gene biotype, protein coding genes made up the majority of IL-14αTG mouse SMG unique DEGs (∼79%) followed by long non-coding RNA (lncRNA) and immunoglobulin genes (Figure 4A). The top 10 unique IL-14αTG DEGs based on Manhattan distance were identified by volcano plot and included the IgH constant and variable chain genes Ighg3 and Ighv1-64 (homologous to human Ighv1-24), the B cell genes Cd19 and Cd79a, the T cell genes Trac and Cd8a, the IgM receptor gene Fcmr, and the tumor necrosis factor family member lymphotoxin-β gene (Ltb) that functions in GC formation (46), all of which are DEGs in SjD vs. HV MSGs (Figure 4B and Supplementary Table S1). Also identified was the upregulation of the lncRNA gene Gm15987 and Fam169b and the downregulation of the transcription factor Dbp, which were not identified as DEGs in SjD vs. HV MSGs. Gene ontology analysis of upregulated unique IL-14αTG mouse SMG DEGs identified 215 significantly enriched biological processes and 55 KEGG pathways in total, with 81 biological processes and 44 KEGG pathways shared in IL-14αTG mouse and human SjD samples (Figure 4C and Supplementary Table S1). In contrast to human and mouse diseased vs. control transcriptome comparisons, the unique IL-14αTG mouse SMG downregulated DEGs showed significant enrichment of the circadian rhythm and rhythmic process KEGG pathway and biological processes (Figure 4D).

Results from spatial transcriptomic analysis validated immune cell marker localization observed by immunofluorescence analysis and provided insights on tissue localization of DEGs identified by RNAseq analysis. In 12-month-old IL-14αTG mouse SMG and SLG, localization of mRNA encoding the salivary acinar cell water channel aquaporin 5 (Aqp5) and the pan-immune cell marker CD45 (Ptprc) (Figure 5) largely mirrored observations from immunofluorescence analysis of a serial IL-14αTG mouse salivary gland section using anti-AQP5 and anti-CD45 antibodies (Figure 1C). Aqp5 and Ptprc expression delineated salivary acinar tissue and immune cell foci, respectively. However, Aqp5 expression was also observed within the immune cell foci and immune cells did not ubiquitously express Ptprc in contrast to immunofluorescence observations. The Aqp5 localization in immune cells may reflect the presence of acinar cells localized above or below the focal plane during confocal microscopy and the lack of Ptprc expression in immune cells may be due to low-level mRNA expression that was below the sequencing threshold. Visium v1 uses a grid of barcoded probe capture areas 55 µm in diameter that bind polyadenylated mRNA, resulting in each cluster being a representative mRNA pool from multiple cell types in the capture area. When examining localization of immune cell marker genes, many immune cell clusters expressed gene markers for multiple cell types. Because the B cell marker antibody B220 (CD45R) detects a splice variant of Ptprc, we instead used CD19 expression to identify B cells and CD3e expression to identify T cells and observed distinct expression patterns within each focus (Figure 5). Localization of Siglec1 (CD169) and Itgax (CD11c) resembled immunofluorescence analysis, where CD169⁺ macrophages localized to the focus periphery and CD11c⁺ dendritic cells were present throughout the focus (Figure 1C). The GL7 antibody used for the detection of activated B and T cells in germinal centers binds a sialic acid glycan (47), so we instead observed the localization of Ltb and Cxcl13 that are essential, but not exclusive, markers for GC development (46, 48) and found robust expression of both genes in IL-14αTG mouse salivary gland immune cell foci (Figure 5).

We utilized BioTuring software to perform clustering of mRNA capture areas and identify areas spatially enriched in specific cell types (Figure 6). Unsupervised clustering of capture areas identified 9 distinct cell clusters for the 12-month-old IL-14αTG mouse SMG and SLG and 7 clusters for the C57BL/6 mouse salivary glands (data not shown). However, two IL-14αTG and three C57BL/6 mouse clusters were enriched in salivary gland acinar genes and appeared otherwise indistinguishable. Therefore, these acinar-enriched clusters were combined for each sample and the uniform manifold approximation and projection (UMAP) dimensionality reduction revealed a total of 8 IL-14αTG (Figure 6A) and 5 C57BL/6 (Figure 6B) mouse salivary gland cell clusters. Color-coded, cell type-enriched clusters were overlayed on the hematoxylin and eosin-stained salivary gland sections (Figures 6C,D) to reveal spatially resolved cell type enrichment (Figures 6E,F). Due to the low resolution of Visium v1 resulting from 55 µm diameter probe capture areas, identification of individual cell types was not possible, and cluster annotations indicate enrichment signatures for the indicated cell types rather than homogeneous cell populations. For cluster annotation, we first utilized the cluster-defining genes from a previously published mouse SMG scRNAseq analysis (25). However, many of the cell type gene markers used for annotation, such as acinar cell secretoglobins (i.e., Scgb2b26, Scgb1b27 and Scgb2b27), were ubiquitously expressed across all clusters (Supplementary Figure S5). Furthermore, the scRNAseq dataset lacked representative sublingual gland, B cell, T cell and neural cell populations. Instead, clusters were annotated using published mouse salivary and lacrimal gland scRNAseq and spatial transcriptomic datasets (25, 26, 49) and the cell type prediction tool of BioTuring SpatialX that uses a reference database of all annotated scRNAseq datasets in the Talk2Data module. In both IL-14αTG and C57BL/6 mouse tissues, sublingual glands formed a single cluster marked by Dcpp1-3, A2ml1 and Syt7, whereas submandibular glands contained multiple cell clusters enriched in salivary epithelial cell markers and/or immune cell markers (Figures 6G,H). SMG acinar cell markers included Lpo, Slc12a2 and Nupr1 while SMG duct cell markers included Crisp3 and various kallikrein serine proteases (i.e., Klk1b26, Klk1b5 and Klk1b11). In the SMG of both genotypes, a population of histologically epithelial cell clusters located near the sublingual gland were enriched in Bpifa2, Dcpp1-3 and Muc19 compared to SMG acinar- and SMG duct-enriched clusters and these likely represent Bpifa2⁺ serous-like acinar cells previously identified in the adult mouse SMG (25). In the C57BL/6 mouse SMG, one cell cluster was enriched in immune cells including the B cell markers Jchain, CD79A and CD79B and the macrophage markers C1Qa/b, Tyrobp and H2-Aa, with lower expression of the T cell marker CD3e. In the IL-14αTG mouse SMG, 3 cell clusters were enriched in immune cell gene markers including the B cell-enriched cluster found in C57BL/6 mouse SMG, an activated B and T cell cluster with increased expression of CD3e and the GC markers Cxcl13 and Ltb, and a mixed immune/epithelial cell cluster located at the periphery of the immune cell foci. The latter cell cluster appears to be SMG acinar and ductal cells histologically, but also expresses higher levels of immune cell markers compared to SMG acinar-enriched and duct-enriched cell clusters. Lastly, cell clusters located near the main excretory duct in the IL-14αTG mouse SMG expressed the neuronal microtubule component gene Tubb3, the voltage-gated potassium channel-interacting protein gene Kcnip4, and the dendritic notch signaling regulator gene Dner, thusly defining a population of neural cells that likely represent the SMG-innervating facial nerve that is closely associated with the Wharton's excretory duct and blood vessels that vascularize the SMG (50).

Flow cytometry results indicated that plasma cells represented the largest population of CD45⁺ immune cells in both IL-14αTG and C57BL/6 mouse SMG (Figure 2D) and these cells abundantly express Jchain, a linker polypeptide for the multimeric assembly of IgM and IgA from antibody-secreting cells (51). Spatial transcriptomic analysis showed high Jchain expression localized in and around the SMG immune cell foci of both genotypes and violin plots demonstrated enriched expression in activated B and T cell, B cell-enriched, and mixed immune/epithelium cell clusters, with lower expression in non-immune cell clusters (Figure 7A). In the IL-14αTG mouse salivary gland, higher level Jchain expression was observed at the periphery of immune cell foci, with the core showing lower or no expression. The proportion of marginal zone B cells was significantly increased in IL-14αTG, as compared to C57BL/6 mouse SMG (Figure 2C), and MZ B cells abundantly express Mzb1, which encodes the endoplasmic reticulum protein pERp1 that functions in Ca²⁺ homeostasis and immunoglobulin assembly (52). In the SMG, low levels of Mzb1 were expressed in IL-14αTG mice with spatial localization in immune cell-enriched clusters, but Mzb1 was undetectable in C57BL/6 mice (Figure 7B). Similarly, mRNA capture areas co-expressing the MZ B cell markers CD19 and Cr2 (i.e., complement receptor 2; CD21/35) were identified in IL-14αTG mouse SMG, but were absent in C57BL/6 mouse SMG (Supplementary Figure S6). While the proportion of CD8⁺ T cells was significantly increased in C57BL/6 mouse SMG, as compared to IL-14αTG mice (Figure 2F), absolute expression of the CD8 alpha subunit gene CD8a was greater in IL-14αTG mouse SMG and violin plots show localization primarily in the activated B and T cell cluster with low expression in all other cell clusters (Figure 7C). All four immune cell marker genes Jchain, Mzb1, Cr2, and CD8a were upregulated DEGs in RNAseq analyses of 12-month-old IL-14αTG vs. C57BL/6 mouse SMG (Figure 7D) and Mzb1, Cr2 and CD8a were upregulated DEGs in human SjD vs. healthy volunteer MSG (Figure 7E).

Flow cytometry analysis identified GL7⁺ GC B cells in both 12-month-old IL-14αTG and C57BL/6 mouse SMG (Figure 2C) and we observed expression of the GC gene markers Ltb and Cxcl13 in immune cell foci in IL-14αTG mouse SMG (Figure 5). Spatial transcriptomic analysis revealed abundant expression of Ltb in IL-14αTG compared to C57BL/6 mouse SMG that was primarily localized to activated B and T cell and B cell-enriched clusters, with lower expression observed in mixed immune/epithelium, neural cell-enriched and sublingual gland clusters (Figure 8A). Similarly, Cxcl13 was more abundantly expressed in IL-14αTG compared to C57BL/6 mouse SMG and expression was almost exclusively localized to activated B and T cell and B cell-enriched clusters (Figure 8B). RNAseq analysis of 12-month-old IL-14αTG vs. C57BL/6 mouse SMG indicated that Glycam1, a marker for high endothelial venules (HEVs), was one of the most highly upregulated DEGs based on Manhattan distance from origin (Figure 3C). Whereas GCs serve as sites for B cell maturation and clonal expansion, HEVs support the extravasation of lymphocytes from the blood stream into inflamed tissues (53). Glycam1 was abundantly expressed in the activated B and T cell and B cell-enriched clusters of IL-14αTG mouse SMG, but was nearly undetectable in C57BL/6 mouse SMG (Figure 8C). Similarly, Ccl21a encoding the potent lymphocyte chemokine expressed in HEV endothelial cells (54), was highly expressed in activated B and T cell and B cell-enriched clusters of IL-14αTG mouse SMG, with expression also observed in the neural cell-enriched cluster (Figure 8D). RNAseq analysis further identified Ltb, Cxcl13 and Ccl21a as upregulated DEGs in IL-14αTG vs. C57BL/6 mouse SMG (Figure 8E) and in human SjD vs. HV MSG (Figure 8F). While Glycam1 is a non-expressed pseudogene in humans, another HEV marker involved in lymphocyte migration, Fut7 (fucosyltransferase 7) (54), was also significantly upregulated in SjD MSG (Figure 8F).

The most significantly upregulated and downregulated DEGs identified by RNAseq of 12-month-old IL-14αTG mouse SMG were the microtubule-associated protein gene Mid1 and the mitogen-activated protein kinase gene Mapk9, respectively (Figure 9). Both Mid1 and Mapk9 were also identified as DEGs in IL-14αTG mouse SMG at the 6-month timepoint (Supplementary Table S1) and Mid1 was identified as a unique SMG DEG that was differentially expressed with age in IL-14αTG, but not in C57BL/6 mouse SMG. Spatial transcriptomic analysis supported RNAseq findings, with broad Mid1 expression observed in all cell clusters and enriched in activated B and T cell, SMG duct cell-enriched and sublingual gland cell clusters in the IL-14αTG mouse, whereas Mid1 expression in C57BL/6 mouse SMG and SLG was sparse in all cell clusters (Figure 9A). Mapk9 expression was abundant in C57BL/6 mouse SMG and SLG and localized to all cell clusters, whereas Mapk9 localization in IL-14αTG salivary glands was noted in all clusters, but was most abundant in neural cell-enriched and sublingual gland cell clusters (Figure 9B). Neither Mid1 nor Mapk9 were identified as a DEG in human SjD MSG (Supplementary Table S1).