The miRNA Landscape of Lacrimal Glands in a Murine Model of Autoimmune Dacryoadenitis

Abstract

Purpose: To analyze the changes in the lacrimal gland (LG) miRNAome from male nonobese diabetic (NOD) mice with autoimmune dacryoadenitis compared with LG from healthy male BALB/c and dacryoadenitis-free female NOD mice.

Methods: LG from these mice were collected for small RNA sequencing to identify dysregulated miRNAs; hits were validated by RT-qPCR in male NOD and BALB/c LG. Dysregulation of validated species within immune cell–enriched cell fractions and epithelial-enriched cell fractions from LG was probed by RT-qPCR. Ingenuity pathway analysis identified putative miRNA targets, which were examined in publicly available mRNA-seq datasets. Western blotting and confocal imaging of immunofluorescence enabled validation of some molecular changes at the protein level.

Results: Male NOD LG exhibited 15 and 13 significantly up- and downregulated miRNAs, respectively. Dysregulated expression of 14 of these miRNAs (9 upregulated, 5 downregulated) was validated in male NOD versus BALB/c LG by RT-qPCR. Seven of the upregulated miRNAs were increased owing to their abundance in immune cell–enriched cell fractions, whereas four downregulated miRNAs were largely expressed in epithelial-enriched cell fractions. Ingenuity pathway analysis predicted the upregulation of IL-6 and IL-6–like pathways as an outcome of miRNA dysregulation. Increased expression of several genes in these pathways was confirmed by mRNA-seq analysis, whereas immunoblotting and immunofluorescence confirmed Ingenuity pathway analysis–predicted changes for IL-6Rα and gp130/IL-6st.

Conclusions: Male NOD mouse LG exhibit multiple dysregulated miRNAs owing to the presence of infiltrating immune cells, and decreased acinar cell content. The observed dysregulation may increase IL-6Rα and gp130/IL-6st on acini and IL-6Rα on specific lymphocytes, enhancing IL-6 and IL-6–like cytokine signaling.

Sjögren's syndrome (SS), the second most common autoimmune disease,^1,2 is associated with lymphocytic infiltration and secretory dysfunction of lacrimal glands (LG) and salivary glands (SG). Systemic symptoms including interstitial lung disease,^3,4 nephritis, peripheral neuropathy, chronic fatigue, and others^5,6 are also associated with disease. The molecular events involved in LG inflammation in SS are not well-understood. As a result, treatment options specific for ocular symptoms of SS that can fundamentally modify disease processes in the LG rather than provide symptomatic relief are lacking. 

miRNAs are 16 to 26 nucleotide (nt), short noncoding functional RNAs (ncRNA) that contain a 6- to 8-nucleotide long seed sequence partially complementary to over 60% of all mammalian mRNAs.⁷ Through translational repression and other mechanisms,⁸ they are master regulators of gene expression⁹ and implicated in disease pathogenesis.^10–14 We have previously isolated tear RNA from an SS disease model, the male nonobese diabetic (NOD) mouse, to identify dysregulated tear miRNAs that could represent diagnostic biomarkers for ocular manifestations of SS.¹⁵ These tear miRNAs may also act on cells of the ocular surface and in draining lymph nodes; thus, they provide insights into the mechanisms of ocular surface disease pathogenesis in SS. We hypothesized that a comparable analysis of the changes in miRNA expression in the diseased LG could identify other dysregulated miRNAs that might shed insights into disease mechanisms and progression in SS. To date, few studies have reported on miRNA dysregulation in LG^16,17 and fewer still have profiled LG miRNA in SS.¹⁸ 

The male NOD mouse is one of the most thoroughly investigated models of ocular symptoms of SS.^19–24 By 6 to 8 weeks of age, male NOD mice spontaneously develop SS-like LG pathology, including lymphocytic infiltration,^21,25 decreased basal and stimulated tear production,²¹ increased cysteine protease expression in tissue^26,27 and tears,^21,27,28 and remodeling of extracellular matrix.²⁰ The male BALB/c mouse is commonly used as a sex-matched control for healthy LG.^19,20,29,30 Female NOD mice develop SS-like pathology in SG but not LG even by 20 weeks of age.^22,25 The females thus represent a strain-specific control lacking ocular disease manifestations. We use both age-matched male BALB/c and female NOD mouse LG as controls to compare with male NOD mouse LG for small RNA sequencing (sRNAseq) evaluation of miRNA expression. 

Here, we present the changes in the NOD mouse LG miRNA transcriptome associated with autoimmune dacryoadenitis. Using ingenuity pathway analysis (IPA), we relate findings on miRNA dysregulation within immune and epithelial cell populations and their potential impact on signaling changes associated with establishment and progression of disease, using publicly available mRNA sequencing (mRNAseq) datasets as well as immune assays to validate changes in gene and protein expression. Our study presents a snapshot of miRNA expression in the healthy LG and identifies dysregulation of particular miRNAs associated with immune cell infiltration and loss or altered function of epithelia that may provide new insights into SS disease mechanisms. 

Methods

Mice

Male NOD mice were used as a model for SS-associated autoimmune dacryoadenitis with controls as described previously.¹⁵ We used only three samples from the additional control—the female NOD mice—to optimize costs for RNA-seq and decrease the number of mice used. 8-week male and female NOD/ShiLtJ (001976), and male BALB/c J (000665) mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA), housed up to five mice per cage with ad libitum access to food and water, and aged to 13 weeks ± 3 days. All procedures involving mice adhere to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, were performed in compliance with the Guide for the Care and Use of Laboratory Animals,³¹ and approved by the USC Institutional Animal Care and Use Committee. 

Tissue Collection

We have published sRNAseq data from tears of male NOD mice compared with tears from female NOD and male BALB/c mice and identified dysregulated miRNAs.¹⁵ From these same 13-week-old mice, after the collection of stimulated tears through addition of topical carbachol as described,¹⁵ mice were euthanized. One LG from these mice was collected and processed immediately for RNA isolation using the Universal mini kit (Qiagen, Germantown, MD, USA) and sRNAseq. The contralateral LG was used for histology to quantify autoimmune dacryoadenitis as published.¹⁵ Datasets generated from these LG RNA samples form the basis of the current study. 

For sRNAseq, one sample was composed of RNA from five LG from each of five mice, with five samples for male NOD, five for male BALB/c and three for female NOD mice. Additional unstimulated LG from male NOD and male BALB/c mice were collected for RNA isolation and validation of dysregulated miRNAs identified by sRNAseq, with the left and right LG from each mouse pooled before RNA isolation using the Universal mini kit (Qiagen). Additional LG from male NOD mice were collected and pooled before isolation of cell populations as below, for RNA isolation and analysis of miRNA expression by RT-qPCR. 

LG Immune and Epithelial Cell Enrichment

Immune cells from LG of male NOD mice were separated from epithelia using protocols developed previously^32,33 with modifications. The epithelial cell fraction is enriched in acinar cells, which represent 85% of the mass of the LG. Briefly, LG were rinsed 3× in s-Ham's³² (Supplementary Methods), and digested using collagenase I (300 U/mL) in s-Ham's, by incubating at 37°C for 10 minutes 3× with repeated trituration in fresh medium. The supernatant was collected after each incubation and sequentially filtered through a 100-µm cell strainer stacked on a 10-µm cell strainer over a 50-mL tube to collect enriched immune cells, allowing retention of epithelia in the 10-µm cell strainer. The filtrate was spun down (300×g for 5 minutes) and the cell pellet resuspended 2× in 5 mL s-Ham's before sedimentation to concentrate the immune cell-enriched fraction (IEF). The epithelial-enriched fraction (EEF) was collected by inverting the 10 µm strainer and eluting with 30 mL of s-Ham's (Supplementary Fig. S1A). Enriched epithelia were spun down (200×g, 5 minutes) and resuspended 2× in 5 mL s-Ham's before subsequent centrifugation. IEF and EEF cell pellets were resuspended in 200 µL s-Ham's and viability assessed with 0.4% Trypan Blue using a Tc-10 Cell Counter (BioRad, Hercules, CA, USA). Cells were lysed with Qiazol (Qiagen) and 10 µL of β-mercaptoethanol was added to inhibit RNAse before isolation of total RNA using the miRNeasy mini kit (Qiagen). 

sRNAseq and Bioinformatics

For sRNAseq, total RNA was isolated from stimulated LG of the same mice used previously for tear collection and tear sRNAseq.¹⁵ The quality of the LG RNA samples was assessed using Agilent Tapestation, and RNA samples were outsourced to Qiagen for library preparation and sequencing. sRNA libraries were prepared using the QIAseq miRNA Library Kit (Cat 331502) and sequenced on NextSeq 550 (Illumina, San Diego, CA, USA) platform (75 base-pairs) single-end configuration, read depth of ∼25 million reads/sample). mRNAseq datasets generated by Ohno et al.¹⁹ were obtained from the European Nucleotide Archive (ENA: PRJDB9749, GEO GSE81621). Raw FASTQ reads were trimmed and filtered with fastp.³⁴ For mRNAseq, reads were aligned to GRCm39 using STAR³⁵ (v2.7.0a) and mapped to Gencode (GRCm39)³⁶ PRI assembly. Transcripts per million were computed in RStudio. For sRNAseq, reads were aligned with the noncoding (nc)RNA (Ensembl³⁷) and miRNAs transcriptomes (miRBase v22.01³⁸) using Bowtie³⁹ as well as miRGrepp.⁴⁰ Alignment parameters are in Supplementary Table S1. 

Preliminary sRNAseq analysis showed that the LG of male NOD mice exhibited a significantly higher percentage of total reads aligning with ncRNA (P = 0.01) in general, and mature miRNA specifically, relative to sRNA from LG of male BALB/c or female NOD mice (Supplementary Fig. S2A, 2B). This increase was unique to LG, and not to tears of the same mice.¹⁵ A comparison of Tapestation and sRNA BioAnalyzer (Agilent) electropherogram peaks of RNA from male NOD LG IEF also revealed a higher percentage of small ncRNA and miRNA relative to total RNA as compared with the NOD LG EEF (Supplementary Fig. S2E, 2F); however, these differences were not statistically significant. We accounted for this confounder, which seemed to be caused by additional sRNA expression from infiltrating immune cells, by normalizing the data to miRNA-aligned reads and sequencing depth before analyzing dysregulated miRNA expression using DESeq2.⁴¹ 

RT-qPCR

For miRNA RT-qPCR, 200 ng of total RNA was used for cDNA synthesis using miRCury LNA RT kit (Qiagen #339340), and 1.3 ng of cDNA per sample was used for RT-qPCR using the miRCury LNA SYBR green qPCR kit (Qiagen #339345) and mature miRNA primers (Qiagen #339306). Normalization was done using spike-in controls, UniSp6 (Qiagen) and housekeeping sRNA Snord68. For validation of IEF and EEF, mouse mRNA primers were purchased from Thermo Fisher Scientific (Rockford, IL, USA). cDNA was synthesized from 1 µg of total RNA. We used 60 to 80 ng of cDNA for RT-qPCR using Taqman reagents (Applied Biosystems, Waltham, MA, USA). For mRNA primers, gene expression was normalized to gapdh. Relative expression was calculated using the ΔΔCt method. Primer catalog numbers are listed in Supplementary Table S2. 

Pathway Analysis

miRNAs expressed in IEF or EEF with their respective log2 fold-change (L2FC) and adjusted P values obtained using DESeq2 (Table 1) were uploaded in IPA (Qiagen) and potential gene targets were obtained. Using these, pathway enrichment analysis was run in Metascape to identify Gene Ontology biological processes or pathways most likely to be affected by miRNAs and assessed further in IPA to obtain predictions specific to expression changes in genes identified by enrichment analysis. For validation of IPA results, publicly available mRNAseq datasets were obtained from ENA (PRJDB9749). These were generated by Ohno et al.¹⁹ from LG RNA from an in-house colony of male NOD mice pre-onset of dacryoadenitis (4 weeks, Pre DO) and post-onset of dacryoadenitis (10 weeks, Post DO), along with age- and sex-matched BALB/c mice.¹⁹ 

Table 1.

 

Differentially Expressed miRNA in LG of Male NOD Mice

 

Table 2.

 

Comparison of mRNAseq Expression Analysis of Genes From the IL6-like Cytokine Signaling Pathway

 

Western Blotting

After euthanasia, LGs were collected from 13-week-old male NOD and BALB/c mice and lysed in RIPA buffer containing protease/phosphatase inhibitor cocktail (Cell Signaling Technology, Danvers, MA, USA) in 2 mL BeadBug tubes (Sigma-Aldrich Corporation, St. Louis, MO, USA). Protein concentration was measured using the Micro BCA Protein Assay Kit (Thermo Fisher Scientific). Lysates were incubated with 6X Laemmli SDS sample buffer (Thermo Fisher Scientific) containing β-mercaptoethanol for 5 minutes at 98°C and then 30 µg of total protein was loaded on 10% Tris-Glycine gels (#XP00105BOX, Thermo Fisher Scientific) before electrophoresis. Proteins on gels were transferred to nitrocellulose membranes (#IB23001, Thermo Fisher Scientific) and stained for total protein (#926-11016, Li-Cor, Lincoln, NE, USA) to enable normalization to protein loading by signal intensity. Membranes were incubated in blocking buffer (Rockland, Limerick, PA, USA), and then incubated overnight in primary antibodies—either rabbit anti-gp130/IL-6st (#3732, Cell Signaling Technology) or goat anti-IL-6Rα (AF1830, R&D Systems Minneapolis, MN, USA)—at a 1:1000 dilution at 4°C. After this, membranes were washed three times (Tris-buffered saline, 0.2% Tween, 5 minutes each) and incubated in 1:2000 dilution donkey anti-rabbit or donkey anti-sheep IR680 secondary antibodies (Li-Cor) at room temperature for 1 hour. After six washes (Tris-buffered saline, 0.2% Tween, 5 minutes each), membranes were imaged with an Odyssey Licor imaging system. 

Immunofluorescence

Unstimulated LG were collected and fixed in 4% paraformaldehyde. OCT embedded LG were cut into 5-µm thin sections and mounted on Superfrost Plus microslides (VWR, Radnor, PA, USA). The cryosections were quenched with 50 mM NH₄Cl in PBS for 5 minutes and permeabilized with either 0.1% Triton X-100 (for gp130/IL-6st) or 0.3% Triton X-100 (for IL-6Rα) for 30 minutes. Sections were blocked in 5% BSA with either 0.1% or 0.3% Triton X-100 for 3 hours at RT and incubated with the primary antibodies as described (Western blotting section) at 1:50 dilution overnight at 4°C. After three washes in PBS, slides were incubated with secondary antibodies (1:200 donkey anti-goat AF 567, 1:200 donkey anti-rabbit AF 594) (Invitrogen, Grand Island, NY, USA), as well as DAPI (Invitrogen) and FITC-phalloidin (Invitrogen) for 1 hour at room temperature, washed three times in PBS and cover-slipped after application of 1 drop of Prolong Gold Antifade mounting medium (Invitrogen). Imaging used a 63× objective (Plan-Apochromat 1.4 Oil DIC M27, Zeiss, Jena, Germany). Single images of 2048 × 2048 pixels were generated with a unidirectional scan at zoom 1.3 (82 µm × 82 µm, step size: 0.04 µm) with a scanning rate of 100 Hz using an LSM 800 Laser scanning microscope (Zeiss) in AiryScan mode. 

Statistical Analyses

sRNAseq/mRNAseq data were analyzed by DESeq2⁴¹ or EdgeR.⁴² Data with more than two groups or factors, and more than five comparisons were analyzed using two-way ANOVA, with repeated measures along rows (i.e., genes) for unpaired data; or along rows (i.e., genes) and columns (such as cell populations, disease state) for paired data. Correction for multiple comparisons and q-values (analogous to adjusted P values) were calculated using the Benjamini–Hochberg fdr procedure in RStudio or GraphPad v7.0a. Data with fewer than five multiple comparisons were analyzed using one-way ANOVA or Student's t test and were corrected for multiple comparisons (wherever needed) with Tukey's honestly significant difference test in RStudio or GraphPad and adjusted P values are reported. 

Results

Differentially Expressed miRNA in the Male NOD LG

sRNAseq analysis was initiated with LG topically stimulated with the muscarinic agonist carbachol using methods described previously.^27,43–45 Topical carbachol stimulates the exocytosis of secretory vesicles containing encapsulated tear proteins⁴⁶ and lipids, as well as sRNAs. We have described the changes in stimulated tear miRNA expression in male NOD mice previously,¹⁵ and LG from the same mice were used for sRNAseq herein. This choice enabled us to identify and focus on miRNA species that were expressed and retained in the LG, as well as to maximize the use of these mice. By 13 weeks, LG lymphocytic infiltration in the contralateral LG of the same male NOD mice was approximately 8% of the total area of the LG.¹⁵ This value was significantly higher than age-matched female NOD or male BALB/c mice, both of which exhibited no discernable infiltration.¹⁵ Evaluation of miRNA expression differences showed that stimulated NOD male LG expressed approximately 700 unique miRNAs compared with 614 in male BALB/c (P_adj = 1.50 × 10⁻⁶) and 630 in female NOD mouse LG (P_adj = 3.44 × 10⁻⁵) (Fig. 1A). We identified 56 miRNAs in male NOD that were not expressed in male BALB/c or female NOD mice LG (Fig. 1B), although many were of low abundance with normalized mean expression of less than 50 reads. 

Figure 1.

 

miRNA expression from sRNAseq of stimulated LG from NOD and BALB/c mice. (A) Boxplot of the number of unique miRNAs per group.^† Red dots indicate the mean, and the other dots indicate the individual data points. (Constraints: at least 1 read detected per each miRNA per sample.) (B) Venn diagram of common and distinct miRNAs in LG from male (M) NOD, M BALB/c and female (F) NOD mice. (Constraints: miRNAs are detected in at least 50% of the biological replicates of a group.) (C) Barplot of the most highly expressed miRNAs^†† in each strain. ^†One-way ANOVA with Tukey's-HSD for multiple correction. ***P_adj < 10⁻⁴, ****P_adj < 10^{−5 ††}Average DESeq2 normalized counts >10⁴; error bars = standard error of the mean.

Differential gene expression analysis using DESeq2 identified 28 miRNAs that were statistically significantly altered (P < 0.05) in stimulated LG of male NOD mice but had comparable expression in stimulated LG of male BALB/ and female NOD (Table 1). Of these, we shortlisted miRNA hits based on (i) a mean expression in male NOD LG of more than 1000 normalized read counts, (ii) a raw fold change of more than |2|, (iii) comparable expression in LG of the two control groups, and (iv) greater than 95% sequence similarity with the human miRNA homolog. Of the 28 miRNAs (Table 1), 14 met our criteria of which 9 miRNAs were significantly upregulated (Fig. 2A), and 5 miRNAs were significantly downregulated (Fig. 2B) in LG from male NOD as compared with LG of control groups. Some of the most differentially expressed miRNA were also the most highly expressed in the LG (let-7c-5p, miR-200c-3p, miR-375-3p, and miR-150-5p) (Fig. 1C). let-7c-5p accounted for 7.08% of all miRNA reads detected in the male NOD LG. Downregulated miR-200c-3p and miR-375-3p accounted for 1.36% and 1.11%, respectively, whereas upregulated miR-150-5p accounted for 4.07% of the total miRNA reads detected (Table 1). The nine upregulated miRNAs accounted for an average 8.03% of all mature miRNA reads detected in NOD LG (Table 1) but made up less than 0.5% of all miRNAs detected in either male BALB/c or female NOD LG. The five downregulated miRNAs accounted for more than 15.5% of total mature miRNA reads in the control groups but made up only 9.5% of all reads in male NOD LG (Table 1). 

Figure 2.

 

Dysregulated miRNAs in LG from male NOD mice. miRNA that are (A) upregulated or (B) downregulated in LG from 13-week diseased male NOD mice relative to LG of age-matched female NOD and male BALB/c mice. Data are plotted as Log10 normalized miRNA counts as calculated by DESeq2 for each group. N = 5 samples for male NOD and BALB/c and N = 3 samples for female NOD mice; n = 1 LG from each of 5 mice per sample. miRNAs were considered differentially expressed if they were upregulated or downregulated in the male NOD versus male BALB/c comparison as well as in the male NOD versus female NOD comparison, had a mean expression value of ≥1000, had a significant unadjusted P value in at least one of the two comparisons; and showed a >95% sequence similarity with human miRNA. (*P < 5 × 10⁻², **P < 10⁻³, ***P < 10⁻⁴, ****P < 10⁻⁵, adjusted P values - DESeq2.)

Validating Differentially Expressed miRNA in Male NOD LG

RT-qPCR was used to validate the 14 dysregulated miRNAs that met the defined criteria in both stimulated and unstimulated LG of male NOD mice (Fig. 2). The expression of the miRNA hits by RT-qPCR was validated in RNA isolated from topically-stimulated pooled LG from male NOD mice relative to strain and sex controls. miRNAs miR-150-5p, miR-155-5p, miR-142a-3p, miR-142a-5p, miR-10a-5p, miR-146a-5p, miR-342a-3p, and miR-34a-5p were significantly increased (P < 0.0001) in the LG of male NOD mice as compared with LG from male BALB/c or female NOD mice (Supplementary Fig. S3A). miR-148a-5p, miR-200c-3p, miR-375-3p, and let-7c-5p were likewise significantly decreased in LG of NOD male (P < 0.001) (Supplementary Fig. S3B). 

In additional samples from unstimulated male NOD mouse LG relative to unstimulated male BALB/c mouse LG, we validated the upregulation of miR-150-5p, miR-155-5p, miR-142a-3p, miR-142a-5p, and miR-10a-5p (P < 0.0001), all showing a greater than 13-fold change. miR-146a-5p and miR-342-3p were also validated as significantly upregulated, whereas miR-146b-5p was modestly but not significantly increased (Fig. 3A). miR-148a-5p, miR-365-3p, miR-200c-3p, miR-375-3p, and let-7c-5p were decreased significantly in the LG of male NOD mice (P < 0.01) (Fig. 3B). miR-34a-5p was significantly decreased in male NOD LG (Fig. 3B), as opposed to its apparent upregulation estimated by sRNAseq Figure 2A, and validated by qPCR in Supplementary Figure S3A in stimulated LG. 

Figure 3.

 

RT-qPCR validation of miRNA changes in unstimulated LG. Barplots showing log2 fold change in miRNA that are either (A) upregulated or (B) downregulated in unstimulated LG from 13-week-old male NOD mice relative to unstimulated LG from age-matched male BALB/c mice. miRNA Ct values were normalized to the endogenous control, snord68; ΔCt values for a given miRNA were then normalized to average ΔCt of that miRNA's expression in the healthy control group (male BALB/c). (Data are plotted as mean ± SEM. n = 3 LG from 3 mice/strain. *q < 0.05, **q < 0.01, ***q < 0.001, ****q < 0.0001, two-way ANOVA with fdr controlled at q = 0.05 for multiple comparisons correction.)

We compared the expression L2FC of miRNA hits in the LG with and without topical carbachol treatment in male NOD versus BALB/c mice to see if carbachol treatment affected miRNA expression. For upregulated miRNAs—miR-155-5p, miR-150-5p, miR-142-3p, miR-142-5p, and miR-10a-5p—there was little or no difference in L2FC before and after carbachol stimulation in the LGs of male NOD versus BALB/c mice (Supplementary Fig. S3C). In the case of the upregulated miRNAs—miR-146a-5p and miR-146b-5p—we observed that the NOD versus BALB/c L2FC increased from 1.83 and 0.56 in unstimulated LG (Fig. 3A) to 3.07 and 1.37 in carbachol-simulated LG, respectively (Supplementary Fig. S3A). This outcome could be due to an increase in expression of the miRNA genes or a relative decrease in the secretion of these miRNAs in the NOD mouse LG. In contrast, for downregulated miRNAs—miR-148a-5p, miR-365-3p, miR-200c-3p, let-7c-5p, and miR-375-3p—there was a decrease in the NOD versus BALB/c L2FC after carbachol stimulation (Supplementary Fig. S3C). This may be due to the enhanced secretion of these miRNAs in BALB/c mouse LG, either into tears or the interstitium. Only in the case of miR-34a-5p was the direction of the fold change expression flipped from an L2FC of –0.86 in unstimulated LG (Fig. 3A) to 1.61 after carbachol stimulation (Supplementary Fig. S3A). Initially, the sRNAseq data analysis and RT-qPCR validation found it to be upregulated in the LG male NOD (Fig. 2A, Supplementary Fig. S3A). Again, this outcome may be due to preferential release of this miRNA species from cells in BALB/c mouse LG. 

Influence of Lymphocytic Infiltration on Differentially Expressed miRNA

Given the large number of miRNAs detected in male NOD mouse LG compared with controls (Supplementary Fig. S1A), we surmised that infiltrating lymphocytes might contribute to the miRNA milieu. To test this, we separated cells from male NOD mouse LG into IEF and EEF. RT-qPCR was used to probe the expression of specific markers of immune and epithelial cells, finding the epithelial-specific aquaporin 5^47–49 and Rab3D^48,49 to be significantly expressed in EEF and immune-enriched genes^48,49 IL-4, IL-12A, IL-17A, and IL17R to be significantly expressed in IEF (Supplementary Fig. S1B). LG and SG epithelia can express cytokines and cytokine receptors in autoimmune dacryoadenitis and sialadenitis, explaining the low expression of some immune-enriched genes in EEF. Using RT-qPCR, we found that miR-150-5p, miR-155-5p, miR-142a-3p, miR-142a-5p, miR-10a-5p, miR-146a-5p, and miR-342-3p were highly expressed in IEF, suggesting their enrichment in infiltrating immune cells (Fig. 4A). miR-148a-5p, miR-365-3p, miR-200c-3p, and miR-375-3p were more highly expressed in EEF (Fig. 4B), suggesting that their depletion in male NOD mouse LG could be due to the loss or altered functioning of epithelia. miR-146b-5p and miR-34a-5p showed only a modest increase in IEF and EEF, respectively, whereas let-7c-5p showed no cell type specificity in expression. 

Figure 4.

 

miRNA expression in immune-enriched (IEF) and epithelia-enriched (EEF) cell fractions from male NOD mouse LG. Barplots comparing log2 relative expression of miRNAs that are either (A) upregulated or (B) downregulated in IEF of 13-week male NOD mice relative to their expression in EEF. Ct values were normalized to the endogenous control, snord68; ΔCt values for IEF were then normalized to average ΔCt of the EEF. (n = 5 mice/group, data are plotted as mean log2 fold change ± SEM. *q < 0.05, **q < 0.01, ***q < 0.001, ****q < 0.0001. ANOVA with repeated measures and fdr (q) controlled at 0.05 for multiple comparisons correction.)

IPA Analysis

IPA identified 584 experimentally validated gene targets of IEF miRNA hits (i.e., dysregulated miRNA in immune cells from male NOD mouse LG). Pathway enrichment analysis on these genes using Metascape indicated that 85 genes belong to the (Gene Ontology) GO-(biological processes) BP—Regulation of cytokine production—predicting it to be the most like BP affected by dysregulated miRNAs (P_adj = 10⁻²³). Among the cytokines, Regulation of IL-6 production was predicted to be the most likely to be affected (P_adj = 10⁻¹⁰) (Supplementary Fig. S4). IPA identified 32 genes in the IL-6 signaling pathway targeted by IEF miRNA hits. 

In IEF, upregulated-miR-155-5p can target SOCS1/3 (Suppressor of Cytokine Signaling), resulting in near-constitutive activation of JAK2/STAT3 and MAPK/ERK pathways (Fig. 5A). Downregulated miRNAs target expression of receptors including IL-6Rα (miR-34a-5p), gp130/IL-6st (let-7c-5p, miR-34a-5p), and its effectors SOS and Ras (let-7c-5p). RT-qPCR of total RNA from LG verified that gene expression of gp130/IL-6st is increased significantly in male NOD mouse LG (Fig. 5B). In EEF, the downregulated miRNAs, miR-148a-3p and miR-200c-3p, were predicted to affect signaling of IL-6 and IL-6-like cytokines (IL-11, LIF, OSM, CNTF, CLC, and CT1) and to upregulate the JAK2/STAT1 pathway (Fig. 6A). Bioinformatics analysis of publicly deposited mRNA-Seq data from Ohno et al.¹⁹ (GEO GSE81621)—of male NOD mouse LG, before and after onset of autoimmune dacryoadenitis verified that several genes from this pathway (e.g., IL-6st, STAT1, and JAK2) were upregulated significantly after the onset of autoimmune dacryoadenitis symptoms (Fig. 6B, Table 2), supporting the predicted effects of dysregulated miRNAs by IPA. 

Figure 5.

 

miRNAs targeting IL-6 signaling are dysregulated in male NOD LG IEF. (A) Arrows indicate direction of miRNA change (Created in BioRender). (B) RT-qPCR confirmation of gene expression changes in gp130/IL-6st , an IL-6 coreceptor, in 13-week-old male NOD relative to age-matched male BALB/c mouse LG. n = 3 LG from 3 separate mice, data are plotted as mean fold change ± SEM, *P = 0.05, unpaired t test.

Figure 6.

 

Dysregulated miRNAs target IL-6-like cytokines and their downstream effectors in EEF. (A) IL-6–like cytokine signaling pathway with gene expression as predicted by IPA owing to the presence of dysregulated miRNA in EEF. Color-coded genes in orange are IPAs estimated predictions with color saturation proportional to the intensity of predicted upregulation. (B) Heatmap showing levels of gene expression in LG of NOD male mice before (pre-dacryoadenitis onset, DO) and after onset of dacryoadenitis (Post DO) as compared with LG of age and sex matched BALB/c. (Bulk RNA-seq data generated by Ohno Y et. al. and raw data obtained from ENA Accession PRJDB9749¹⁹.)

Additionally, Western blotting of LG lysates showed a significant increase in protein expression levels of IL-6 receptor subunits, gp130/IL-6st (P = 0.0134) (Fig. 7A), and IL-6Rα (P = 0.0004) (Fig. 7B) in samples from male NOD mice compared with BALB/c mice. Imaging of immunofluorescence associated with these proteins in the LG showed increased expression of gp130/IL-6st in the acinar cells of male NOD LG when compared with BALB/c mouse LG acini (Fig. 7C). Immunofluorescence analysis also showed an increased expression of IL-6Rα concentrated near or within the lumena of male NOD mice LG acini, but not in acini from BALB/c mice. We also detected a subpopulation of lymphocytes expressing IL-6Rα on their cell surface (Fig. 7D). 

Figure 7.

 

Protein expression of IL-6 receptor subunits is altered in male NOD LG. Western blots showing protein expression of (A) gp130/IL-6st and (B) IL-6Rα in LG lysates from 13-week-old male BALB/c and NOD mice.^† Representative images of 5-µm LG sections showing immunofluorescence labeling of the IL-6 receptor subunits (C) gp130/IL-6st and (D) IL6-Rα in LG from male NOD and BALB/c mice. In (C), arrows point to acini expressing gp130/IL-6st. In (D), arrows point to lymphocytes expressing IL-6Rα, whereas the arrowheads point to an acinar lumen showing enhanced accumulation of IL-6Rα. Bars, 10 µm. ^†n = 5 LG from 5 separate mice, data are plotted as total protein normalized mean signal intensity ± SEM, *P < 0.05, ***P < 0.001, unpaired Student's t test.

Discussion

Although miRNA involvement in SS-associated SG disease^50–55 and systemic autoimmunity^16,52 has been investigated, their role in SS-associated LG disease is understudied. Here we present the miRNAome from the LG of the male NOD mouse and identify species dysregulated in parallel with the development of autoimmune dacryoadenitis. For initial NGS sequencing, we identified dysregulated species relative to healthy sex-matched and strain-specific controls, whereas subsequent validation and expansion studies with RT-qPCR used sex-matched healthy control male BALB/c mice. Nine high-expression miRNAs were significantly upregulated in male NOD mouse LG, 7 of which were validated by RT-qPCR, whereas five high-expression miRNAs were significantly downregulated in male NOD mouse LG, all of which were validated by RT-qPCR. 

Some of our identified miRNAs have been reported as dysregulated in tissues from patients with SS or SS animal models. sRNAseq of LG in an SS rabbit model found miRNAs miR-150-5p, miR-142-3p, and miR-142-5p to be upregulated significantly.¹⁸ A microarray comparing the expression of 534 human and viral miRNAs from minor SG of patients with SS versus healthy controls found miR-150-5p, miR-155-5p, miR-142a-3p, miR-142a-5p, miR-10a-5p, miR-342-3p, and miR-146b-5p to be significantly upregulated as well and miRNAs miR-148a-5p, miR-200c-3p, and miR-375-3p to be downregulated significantly.⁵⁵ However, our finding of downregulated miR-365-3p in the LG with disease is a novel finding in the context of SS. 

To delineate the cellular origin of the dysregulated miRNAs, we isolated LG cells into IEF and EEF. RT-qPCR analysis showed that that all seven validated upregulated miRNAs from Figure 3 showed high expression in IEFs; conversely, four of the five validated downregulated miRNAs from Figure 3 were more highly expressed in EEF and minimally expressed in IEF. Thus, the upregulation of most miRNAs in male NOD mouse LG seems to be due to their expression by infiltrating immune cells, whereas the epithelial cell specificity of most downregulated miRNAs suggests either a decrease in epithelial cells or altered epithelial cell function with disease. Consistent with our findings, miR-200b-5p is expressed in SS patient-derived SG epithelial cells, but not in the peripheral blood mononuclear cells,⁵² whereas miR-375-3p expression is epithelial specific and not detected in immune cells.⁵⁶ miR-146a was previously reported as increased in both LG and SG from the NOD.Aec1Aec2 mouse model of SS¹⁶ and in the peripheral blood mononuclear cells of patients with SS.^16,57,58 

Of the upregulated miRNAs in our study, miR-142a-5p, a lymphoid tissue–specific miRNA,⁵⁹ is required for hematopoiesis of T cells. miR-200b and let-7b are predicted to target expression of SS autoantigens—Ro/SSA (both TRIM21 and TROVE2 subunits) and La/SSB.^52,60 Serum autoantibodies against these antigens constitute diagnostic biomarkers for SS.⁶¹ Additionally, gene transcripts for Rgs16 and ccl22, part of the IL-17 pathway, are upregulated in Post DO male NOD LG from data deposited by Ohno et al.¹⁹ Rgs16 is involved in autoantibody production and is targeted by let7c. Thus, the downregulation of miR-200b-3p and let-7c could increase expression of Ro/SSA and La/SSB, increasing their propensity to elicit an autoantibody response. 

Several IEF-specific upregulated miRNAs were also significantly increased in tears of male NOD mice relative to strain controls-miR-142-5p (FC = 2.5, P = 0.017, DESEq2) and miR-155-5p (FC = 1.4, P = 0.012, DESEq2). In addition, the EEF-specific downregulated miRNA miR-200c-3p-was also decreased significantly in the tears of male NOD mice relative to strain controls (FC = −1.3 P = 0.04). Let-7c-5p was the most abundant miRNA in tears and LG overall; concurrent with its decreased abundance in male NOD LG, it was also significantly decreased (FC = −1.45, P = 0.01) in the tears of male NOD mice compared with the tears of male BALB/c mice. These relationships suggest that these miRNAs may reach the tear film through acinar endocytic uptake and transcytotic transport from the LG interstitium where they are secreted by the cells of origin. However, some upregulated LG miRNAs from this study are significantly decreased in the tears of male NOD mice—miR-342-3p (FC = −3.06, P = 10⁻⁷), miR-146a-5p (FC = −1.74, P = 0.001), and miR-322-5p (FC = −1.6, P = 0.04)—compared with the tears of male BALB/c mice, indicating an inverse relationship. This finding may reflect the profound changes in cellular composition of the LG associated with autoimmune dacryoadenitis, with a decreased acinar content and thus perhaps decreased secretion of some miRNAs and greatly increased lymphocytes. 

The major cytokine pathway predicted as targeted by dysregulated miRNAs is that mediated by IL-6. IL-6 has pro- and anti-inflammatory effects, with the former elicited through the trans-pathway via soluble IL-6Rα, and the latter via the membrane-bound gp130/IL-6st-IL-6Rα complex.⁶² IL-6Rα is reported to be present largely on immune cells (specifically certain types of lymphocytes), whereas gp130/IL-6st is expressed more ubiquitously,^48,49,63,64 including in epithelia, and is also involved in IL-6–like cytokine signaling⁶⁵ (Fig. 6A). Imaging of gp130/IL-6st immunofluorescence in the LG showed more signal in acinar and ductal cells in LG from both strains, although the signal was greater in NOD mouse LG. There was minimal gp130/IL-6st signal in infiltrating immune cells, whereas IL-6Rα showed pronounced expression on subpopulations of infiltrating lymphocytes in male NOD mouse LG. A different milieu of dysregulated miRNAs in immune versus epithelia cells may differentially affect components of the IL-6 and IL-6–like pathways. The downregulation of key miRNAs that target il6 and il6rα mRNA may make the proinflammatory IL-6 pathway more dominant in epithelial cells. This is also evident from immunofluorescence images, which show increased expression of IL-6Rα in the lumena of male NOD LG acini. Previous use of tocilizumab (humanized IL-6Ra recombinant antibody) was not successful in patients with SS.⁶⁶ This outcome could be due to its nonspecific and dual targeting of soluble and membrane-bound IL-6Ra. It is possible that selective targeting of the trans-pathway to sequester sIL-6Ra and the sIL6Ra-il6 complex,⁶³ with a recombinant soluble gp130/IL-6st-Fc protein (such as olamkicept⁶⁷) may have more usefulness in SS, particularly if administered locally. 

Acknowledgments

Supported by National Institutes of Health funding from the National Eye Institute, grant EY011386 to SHA. The research reported in this publication was also supported by P30EY029220 from the National Eye Institute and P30CA014089 from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. An unrestricted grant to the Department of Ophthalmology from Research to Prevent Blindness, New York, New York, also supported this research. Finally, the authors acknowledge the support of the Translational Research Laboratory at the USC School of Pharmacy. 

Disclosure: S. Singh Kakan, None; X. Li, None; M.C. Edman, Oyster Point (C); C.T. Okamoto, None; B.E. Hjelm, None; S.F. Hamm-Alvarez, Oyster Point (C)