Lacrimal Gland Epithelial Cells Shape Immune Responses through the Modulation of Inflammasomes and Lipid Metabolism

Abstract

Lacrimal gland inflammation triggers dry eye disease through impaired tear secretion by the epithelium. As aberrant inflammasome activation occurs in autoimmune disorders including Sjögren’s syndrome, we analyzed the inflammasome pathway during acute and chronic inflammation and investigated its potential regulators. Bacterial infection was mimicked by the intraglandular injection of lipopolysaccharide (LPS) and nigericin, known to activate the NLRP3 inflammasome. Acute injury of the lacrimal gland was induced by interleukin (IL)-1α injection. Chronic inflammation was studied using two Sjögren’s syndrome models: diseased NOD.H2^b compared to healthy BALBc mice and Thrombospondin-1-null (TSP-1^-/-) compared to TSP-1^WT C57BL/6J mice. Inflammasome activation was investigated by immunostaining using the R26^ASC-citrine reporter mouse, by Western blotting, and by RNAseq. LPS/Nigericin, IL-1α and chronic inflammation induced inflammasomes in lacrimal gland epithelial cells. Acute and chronic inflammation of the lacrimal gland upregulated multiple inflammasome sensors, caspases 1/4, and interleukins Il1b and Il18. We also found increased IL-1β maturation in Sjögren’s syndrome models compared with healthy control lacrimal glands. Using RNA-seq data of regenerating lacrimal glands, we found that lipogenic genes were upregulated during the resolution of inflammation following acute injury. In chronically inflamed NOD.H2^b lacrimal glands, an altered lipid metabolism was associated with disease progression: genes for cholesterol metabolism were upregulated, while genes involved in mitochondrial metabolism and fatty acid synthesis were downregulated, including peroxisome proliferator-activated receptor alpha (PPARα)/sterol regulatory element-binding 1 (SREBP-1)-dependent signaling. We conclude that epithelial cells can promote immune responses by forming inflammasomes, and that sustained inflammasome activation, together with an altered lipid metabolism, are key players of Sjögren’s syndrome-like pathogenesis in the NOD.H2^b mouse lacrimal gland by promoting epithelial dysfunction and inflammation.

1. Introduction

Dry eye disease affects millions of adults worldwide and can be divided into two major types: evaporative dry eye and aqueous-deficient dry eye (ADDE) [1,2]. ADDE is characterized by a reduced secretion or an altered composition of fluid from the lacrimal gland (LG) [3,4], the exocrine tubuloacinar gland responsible for secreting the aqueous layer of the tear film [5,6,7]. ADDE induces eye irritation and pain, and may lead to severe ocular surface disorders [8,9]. The leading cause of ADDE is the chronic inflammation of the LG triggered by aging or auto-immune diseases such as Sjögren’s syndrome (SS).

Dry eye disease involves inflammatory mechanisms and the production of several tear cytokines, including interleukins (IL)-1α and IL-1β, which correlate with clinical severity [10]. IL-1α/β are potent proinflammatory cytokines that function as key danger signals during infection or tissue damage. IL-1α/β binding to their receptor IL-1R1 promotes the transcription of genes involved in acute and chronic inflammation. In mice, reversible ADDE can be experimentally induced by a single injection of IL-1α into the LG [11,12]. Several studies demonstrated that IL-1α induces acute LG inflammation and destruction within the first two days after the injury [11,12,13,14]. The resolution of inflammation was noted on the third day after injury and was followed by cell proliferation and complete regeneration within 5–7 days [12,13].

Inflammasomes are large intracellular multiprotein complexes that play a central role in innate immunity [15,16]. The largest class of inflammasomes contain an apoptosis-associated speck-like protein containing a CARD (ASC, encoded by the gene Pycard) and pathogen/danger sensors, which recruit and activate the pro-caspase 1 (pro-CASP1). Each type of inflammasome is characterized by a particular sensor or receptor: PYRIN, the nucleotide-binding domain (NOD), leucine-rich repeat (LRR)-containing protein (NLR) family (e.g., NLRP1/3/6, NLRC4), or the pyrin and HIN domain-containing protein (PYHIN) family (e.g., AIM2 and IFI204, the murine homolog of human IFI16). Canonical inflammasomes cleave IL-1β and IL-18 precursors to generate the mature cytokines. Activated CASP1 and the non-canonical inflammasome formed by CASP4/11 can also cleave the pore-forming protein gasdermin D (GSDMD), which mediates interleukin secretion [17]. GSDMD is required for pyroptosis, an immunogenic form of cell death which enables the massive release of active IL-1α and IL-1β from dying cells [15,18]. Inflammasomes can be activated by a multitude of infectious and sterile stimuli, including microbiome-derived signals and host-derived signals, and were found in different cell types [19]. We previously showed that acute LG inflammation triggers the upregulation of inflammasome-related molecules: Casp4, Nlrp3, the purinergic receptors P2RX7 and P2RY2, the Pannexin-1 (Panx1) membrane channel glycoprotein—a key regulator of inflammasome assembling—, and numerous proinflammatory factors including IL-1β and IL-18 [14].

Whilst inflammasome signaling is critical for the initiation of a fast innate immune response to tissue damage or invading pathogens, aberrant inflammasome activation contributes to various pathologies, including autoimmune disorders, cardiometabolic diseases, cancer, and neurodegenerative diseases [20,21]. It has been shown that NLRP1, NLRP3, NLRC4, and AIM2 inflammasomes play a significant role in shaping immune responses and regulating the homeostasis of intestine and ocular surface immunity in several inflammatory diseases [22,23,24,25,26,27]. In human patients suffering from SS dry eye, conjunctival impression cytology demonstrated an upregulation of Nlrp3 and Casp1 [25]. Baldini and coauthors [28] showed that in the salivary glands of SS patients, the increased expression of NLRP3, CASP1, and P2RX7 was a marker of disease and correlated with the focus score evaluating the number of immune cell infiltrates in gland sections. Moreover, AIM2 inflammasomes are activated in the salivary epithelium of primary Sjögren’s syndrome (pSS) patients and correlate with the expression levels of type I interferon (IFN) signature genes [29]. IFNs are major regulators of the innate immune response and contribute to the activation of canonical and non-canonical inflammasomes [30], partly by inducing the guanylate-binding proteins (GBPs) that are dynamin-like GTPases [31,32]. Among them, GBP1 and GBP2 are upregulated in the biopsies of pSS salivary glands [33] and the latter was proposed as a biomarker for SS in saliva [34] and salivary glands [35]. Although the implication of inflammasome modulators in chronically inflamed LG remains unknown, their therapeutic potential is supported by our observation that the inhibition of Panx1 or Casp4 reduced the inflammation of LGs from thrombospondin-1-null (TSP-1^-/-) mice, a model for SS, and improved the epithelium repair through the increased engraftment of progenitor cells [14].

In this study, we investigated inflammasome formation during acute and chronic LG inflammation using the R26^ASC-citrine mouse. This reporter mouse forms fluorescent ASC specks when inflammasomes are activated [36]. We show that LG epithelial cells can form inflammasomes upon sensing danger signals and inflammation. To identify inflammasome types involved in acute and chronic inflammation, we analyzed RNA-sequencing (RNA-seq) data from a previously published study on acute injury [13] and performed RNA-seq of LGs from NOD.H2^b mice [37]—a pSS mouse model. Our results indicate a strong activation of multiple inflammasome complexes during acute and chronic inflammation. Finally, we analyzed the RNA-seq data of the LGs during acute and chronic inflammation in terms of the biological pathways to identify candidate mechanisms that promote the resolution of inflammation. We demonstrated that lipid biosynthesis is activated during the resolution of inflammation/regeneration after acute injury, but that genes for fatty acid and cholesterol synthesis are, respectively, down- and upregulated during chronic inflammation. Moreover, chronic inflammation also downregulates the genes involved in the tricarboxylic acid (TCA) cycle and the β-oxidation of fatty acids in the mitochondria. Combined, our results show that during chronic inflammatory disease, LG epithelial cells have reduced lipid biosynthesis, accumulated cholesterol, and showed mitochondrial dysfunction. Altogether, these alterations likely induce cell damage, sustain inflammasome activation, and impair LG regeneration and function.

2. Methods

2.1. Mice

For the in vivo detection of the activated inflammasome complex, we employed the transgenic mouse expressing a mouse ASC-citrine fusion protein in the Rosa26 (R26) locus (R26^ASC-citrine, kind gift of Dr. Golenbock) [36]. R26^ASC-citrine mice were bred and maintained on the C57BL/6J background.

B6.129S2-Thbs1^tm1Hyn/J mice (TSP-1^-/-, RRID:IMSR_JAX:006141) were originally purchased from Jackson Laboratory (Sacramento, CA, USA) and were bred and maintained on the C57BL/6J background. For immunoblotting, TSP-1^-/- mice were compared to age-matched wild-type (WT) C57BL/6J mice. For immunofluorescence, we crossed R26^ASC-citrine x TSP-1^-/- mice to obtain TSP-1^-/-:R26^ASC-citrine mice that were compared to age-matched R26^ASC-citrine mice. The NOD.B10Sn-H2^b/J mice (NOD.H2^b, RRID:IMSR_JAX:002591) and their BALB/cJ controls (BALBc, RRID:IMSR_JAX:000651) mice were purchased from Jackson Laboratory (Sacramento, CA, USA).

The mice were housed under standard conditions of temperature and humidity, with a 12 h light/dark cycle and free access to food and water. All experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Guidelines for the Care and Use of Laboratory Animals, published by the US and National Institutes of Health (NIH Publication No. 85-23, revised 1996), and were pre-approved by TSRI Animal Care and Use Committee.

2.2. Induction of Inflammasome Formation with Lipopolysaccharide and Nigericin

R26^ASC-citrine mice were primed by intraglandular injection of lipopolysaccharide (LPS, 1 µg/mL). After 3 h, inflammasome activation was induced by the injection of nigericin (10 µM). Six hours after the last injection, the mice were sacrificed and LGs were dissected out and processed for frozen section preparation.

2.3. LG Acute Injury

For acute injury experiments, the R26^ASC-citrine and WT C57BL/6J females were used. LG inflammation in these mice was induced by the intraglandular injection of IL-1α, as previously described [11]. Briefly, 12 female mice (10 to 12 weeks old) were anesthetized, and the exorbital LG was injected with either saline (vehicle) or IL-1α (1 μg; PeproTech, Cranbury, NJ, USA) in a total volume of 2 μL using a Hamilton glass syringe (#300329, World Precision Instruments, Inc., Sarasota, FL, USA) and NanoFil 35G needle (#NF35BV-2, World Precision Instruments, Inc., Sarasota, FL, USA). The LGs from uninjected mice were used as additional controls. The LGs were harvested 6 and 12 h after injection and processed for immunohistochemistry and RNA extraction.

2.4. Frozen Section Preparation and Immunostaining

The dissected LGs were fixed with 2% paraformaldehyde in PBS (pH 7.4) for 45 min and frozen in 2-methylbutane cooled by liquid nitrogen, and 10-μm cryosections were prepared using Hacker/Bright OTF5000-LS004 Cryostat. The sections were blocked with 1% bovine serum albumin in Tris-buffered saline containing 0.05% Tween 20. The following primary antibodies were used for immunostaining overnight at 4 °C: mouse monoclonal α-smooth muscle actin antibody (1/200, clone 1A4; #A2547, RRID:AB_476701, Millipore-Sigma, Rocksville, MD, USA) was used to label the myoepithelial cells (MECs) and pericytes (contractile cells around the endothelial cells), rat monoclonal CD31 antibody (1/100, #553370, RRID:AB_394816, BD Biosciences, Franklin Lakes, NJ, USA) was specific to blood vessels, mouse monoclonal E-Cadherin antibody (1/200, #610182, RRID:AB_397581, BD Biosciences) labeled the epithelial cells, and the rabbit polyclonal AIM2 antibody (1/100, #63660, RRID:AB_2890193, Cell Signaling Technology, Danvers, MA, USA) was used to detect AIM2-inflammasomes.

Appropriate fluorochrome-conjugated secondary antibodies were obtained from Invitrogen (Waltham, MA, USA) and nuclei were counterstained with DAPI. The formation of inflammasome complexes following bacterial and sterile stimuli in WT mice and during chronic inflammation in TSP-1^-/- mice was detected using the ASC-citrine fusion protein that is constitutively and ubiquitously expressed in R26^ASC-citrine mice and forms fluorescent specks upon inflammation. Images were taken using a LSM 880 laser scanning confocal microscope (Zeiss, Oberkochen, Germany) at the microscopy core of the Scripps Research Institute (La Jolla, CA, USA). Three different fields were analyzed per animal. Inflammasomes were counted using Imaris Spots software, which allows for the detection and counting of small particles. The number of specks detected in the inflamed LGs was adjusted by subtracting the number of specks found in their respective controls.

2.5. Western Blotting Analysis

For protein extraction, the dissected LGs were rinsed in cold PBS and transferred into 2 mL tubes pre-filled with 2.8 mm ceramic beads (#19-628, Omni, Inc., Kennesaw, GA, USA) containing 500 µL of ice-cold RIPA buffer without detergents (50 mM Tris-HCl + 150 mM NaCl + 1 mM EGTA) and supplemented with protease/phosphatase inhibitor cocktail (#5872, Cell Signaling, Danvers, MA, USA). The tissue was homogenized with the Omni Bead Ruptor 4 (# 25-010, Omni, Inc., Kennesaw, GA, USA; 2 cycles: Speed 5, 40 s each) and lysate was kept on ice. An appropriate volume of complete RIPA buffer supplemented with each detergent five-times concentrated was added to the sample (final concentration: 1% Nonidet P-40 + 0.5% sodium deoxycholate + 0.1% SDS + 1 mM EDTA) before incubation on ice for at least 30 min. Then, lysate was centrifuged (15 min, 15,000× g, 4 °C) and the supernatant was collected for protein quantitation using Pierce BCA protein Assay kit (#23225, Thermo Fisher Scientific, Waltham, MA, USA). After denaturation (with NuPAGE LDS + β-mercaptoethanol at 70 °C for 10 min), 30 µg total protein from each sample was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (NuPAGE Novex Bis Tris gels, ThermoFisher Scientific, Waltham, MA, USA) and transferred to polyvinylidene difluoride membranes using the iBlot2 system (ThermoFisher Scientific, Waltham, MA, USA). The transfer membranes were stained for total protein with the No-Stain Protein Labeling Reagent (#A44449, ThermoFisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions. The membranes were then blocked with TBS + 0.1% Tween + 5% milk for 1 h at RT before incubation with the appropriate primary antibody at 4 °C overnight: anti-GSDMDC (1/1000, #sc-393656, RRID:AB_2728694, Santa Cruz Biotechnology, Dallas, TX, USA,), anti-CASP1 (1/1000, #22915-1-AP, RRID:AB_2876874, ProteinTech, San Diego, CA, USA), or anti-IL-1β (1/1000, #A16288, RRID:AB_2769945, Abclonal, Woburn, MA, USA). After 3 washes, the blots were incubated for 1 h with the appropriate horseradish peroxidase-linked secondary antibody and processed for chemiluminescence. The signal was detected using the iBright (ThermoFisher Scientific, Waltham, MA, USA) or ChemiDoc MP (Bio-Rad, Hercules, CA, USA) imaging systems. The quantification of the relative protein abundance was performed using Image Lab software (v6.1, Bio-Rad, Hercules, CA, USA). Background-adjusted band volumes were corrected with the normalization factor (calculated using total protein stain for BALBc/NOD.H2^b samples and β-actin for TSP-1^-/- samples) and normalized to a reference volume corresponding to the average of the biological replicates in the control condition.

2.6. qRT-PCR

The total RNA was isolated using Trizol and converted into cDNA using RT² First Strand Kit (#330404, Qiagen, Germantown, MD, USA). The gene expression was assessed using the SYBR Green kit, according to the manufacturer’s instructions (Applied Biosystems, Waltham, MA, USA) using a 7300 Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). The sequences for qRT-PCR primers are listed in

Table 1. Sequences of primers used for qRT-PCR.

Target	Primer Sequence
Nlrp3	Forward: 5′-AGAAGAGACCACGGCAGAAG-3′
	Reverse: 5′-CCTTGGACCAGGTTCAGTGT-3′
Nlrc4	Forward: 5′-CTTGGCCAGGAGAGCCTTG-3′
	Reverse: 5′-GGGCTCGTCTGTTGTTCCTT-3′
Aim2	Forward: 5′-GATTCAAAGTGCAGGTGCGG-3′
	Reverse: 5′-TCTGAGGCTTAGCTTGAGGAC-3′
Casp1	Forward: 5′-GCGAAGCATACTTTCAGTTTC-3′
	Reverse: 5′-TCTCCTTCAGGACCTTGTCG-3′
Casp4(11)	Forward: 5′-AGGAGCCCACTCCTACAGAG-3′
	Reverse: 5′-AAGGTTGCCCGATCAATGGT-3′
Actb	Forward: 5′-AGAGGGAAATCGTGCGTGAC-3′
	Reverse: 5′-CAATAGTGATGACCTGGCCGT-3′
Gapdh	Forward: 5′-CGTCCCGTAGACAAAATG GT-3′
	Reverse: 5′-TTGATGGCAACAATCTCC AC-3′

. qPCR data were analyzed using the comparative C_t (ΔΔC_t) method. Actb and Gapdh were used as internal reference genes.

2.7. RNA-Sequencing (RNA-seq) Data Analysis

For acute injury experiments, we used the previously published data available from the Gene Expression Omnibus (GEO) database (Accession number: GSE99093) [13]. To study chronic inflammation, we mined our RNA-seq data of BALBc and NOD.H2^b male LGs processed at 2, 4, and 6 months of age, deposited under GSE210332. Data were analyzed using ROSALIND^®® software (https://rosalind.bio/ accessed on 19 February 2023) developed by ROSALIND, Inc. (San Diego, CA, USA). As previously described, ROSALIND uses DESeq2 R library to normalize read counts and calculate fold-changes along with the corresponding p-values [37].

For all projects, differentially expressed genes (DEGs) were selected based on a fold-change (FC) cut-off equal to log₂(FC) ≈ ±0.585 (corresponding to FC = 1.5) and p-adj < 0.05. On the figures, the statistical significance of log₂(FC) compared to the respective control (uninjured LG for acute injury; age-matched BALBc for NOD.H2^b LGs) is shown with: * p-value adj. < 0.05; ** p-value adj. < 0.01; *** p-value adj. < 0.001.

For acute injury data, a pathway enrichment analysis was conducted using Metascape [38] using default parameters (min. overlap = 3, p-value cut-off = 0.01, min. enrichment = 1.5) and Gene Prioritization by Evidence Counting (GPEC), an algorithm identifying the subset of input genes that are more likely to be true hits, and by interrogating WikiPathways and Gene Ontology Biological Process databases. The dendrogram was created with the web-based tool Clustergrammer (https://maayanlab.cloud/clustergrammer/ accessed on 25 May 2022) using complete-linkage clustering with Euclidian distances.

2.8. Statistical Analysis

Prism9 v9.1.2 (GraphPad software Inc, La Jolla, CA, USA) was used to plot the results and test their statistical significance. First, a Shapiro-Wilk normality test was performed to evaluate if the data follow a normal distribution. If data passed the normality test, the statistical significance between the two conditions was assessed with an unpaired t-test and the results were represented as mean ± standard deviation (SD). To compare more than two groups, a one-way ANOVA was used. Otherwise, non-parametric tests were used (Mann-Whitney test for two groups, and Kruskal-Wallis otherwise) and the plots showed the median ± interquartile range (IQR), as specified in the legends. The significant differences are represented as * if p value p < 0.05, ** if p < 0.01, and *** if p < 0.001.

3. Results

3.1. Epithelial Cells Can Sense Microbial/Sterile Inflammatory Stimuli

In contrast to non-canonical inflammasomes formed by CASP4, most types of canonical inflammasomes require the assembly of the adaptor protein ASC for the activation of the inflammasome cascade leading to IL-1β/IL-18 maturation, and eventually release through the GSDMD pores, also leading to cell death (Figure 1A). Thus, for the in vivo detection of the canonical inflammasome complexes, we used a transgenic mouse constitutively expressing the ASC-citrine fusion protein (R26^ASC-citrine) (see enlarged micrograph in Figure 1A) [36]. To test whether inflammasomes could be formed in the LG, we first mimicked a bacterial infection in the R26^ASC-citrine reporter mouse, as previously described [36]. In brief, LGs were injected with lipopolysaccharide (LPS, typical pathogen-associated molecular pattern, or PAMP), which constitutes the priming signal inducing Il1b transcription, and with nigericin, a pore-forming bacterial toxin which activates the NLRP3 inflammasome (Figure 1A) [39]. An anti-α-SMA antibody was used to identify MECs that surround the secretory units formed by acinar cells and the pericytes wrapped around blood vessels (blood vessels were labeled by antibody to CD31). Therefore, the tubular structures negative for CD31 and α-SMA staining were identified as ducts. In control LGs, we observed just a few ASC specks (Figure 1B and Figure S1A). By contrast, numerous ASC specks were formed 6 h after LPS-nigericin stimulation, compared to the vehicle-injected LGs (Figure S1C). Many of the ASC specks were found in the LG epithelium (Figure 1C and Figure S1B).

To test whether LG cells could also form inflammasomes upon sterile stimuli, we induced acute injury by injecting IL-1α in the LG on one side of the R26^ASC-citrine mouse and injected the other LG with a saline (vehicle control) (Figure 1D,E). Similar to LPS-nigericin stimulation, ASC specks were detected 6 h after IL-1α-injection (Figure 1E). They were primarily formed in acinar cells and MECs (Figure 1E,F and Figure S1D). At 12 h after injury, ASC complexes were also detected within ducts and infiltrating immune cells (Figure 1G). The proportion of cells forming inflammasomes and the diameter of ASC specks significantly increased over time in injured LGs (Figure 1H,I). Following inflammasome formation (Figure 1J), we also observed higher cell death rates with characteristic nuclear fragmentation in IL-1α-injected LGs (Figure 1K,L).

Altogether, these results show that LG epithelial cells form inflammasomes in response to microbial or sterile pro-inflammatory stimuli and may eventually undergo pyroptosis prior to immune infiltration.

3.2. Inflammasome Activation during Chronic Inflammation

We then tested whether inflammasomes are activated in the LG epithelial cells during chronic inflammation. First, we analyzed NOD.H2^b males, which develop robust lymphocytic infiltrates in the LG at 4–6 months of age [37] but do not have autoimmune diabetes [40,41]. To determine whether inflammasomes are active, we performed Western blotting to study the proteolytic maturation of downstream targets in the LGs of 6-month-old (6M) NOD.H2^b and control BALBc males (Figure 2A,B and Figure S1E). Although there was no significant difference in the abundance of the full length CASP1 (Pro-CASP1, p46) between the LGs of NOD.H2^b and BALBc mice, it was detected at a slightly higher molecular weight in NOD.H2^b mice (Figure 2A). This suggests a post-translational modification (ubiquitination, phosphorylation) that could modulate the inflammasome activity [42]. Moreover, in NOD.H2^b LGs, we detected an increased amount of the cleaved form of CASP1 p33, which forms the active species on the inflammasome hub with the p10 subunit of CASP1 (Figure 2A). This complex is able to process many different substrates [43]. Consistent with inflammasome/CASP1 activation, we found that both the precursor (31 kDa) and the mature form of IL-1β (17 kDa) were significantly increased in the diseased LGs, and that the cleaved form of GSDMD was approximately five times more abundant in NOD.H2^b as compared to BALBc LGs (Figure 2B).

To determine if the activation of inflammasome signaling is not specific to NOD.H2^b mice, we also analyzed the TSP-1-null (TSP-1^-/-) mouse—another model of pSS [44,45]. We generated the TSP-1^-/-:R26^ASC-citrine mice and used the R26^ASC-citrine mice as controls. While the control mice had only a few ASC specks in the LG, we found significantly more ASC specks in epithelial cells of the TSP-1^-/-:R26^ASC-citrine mice at 2M and 6M, respectively (Figure 2C–F). The number of ASC specks increased with disease progression in the LGs of TSP-1^-/-:R26^ASC-citrine mice (Figure 2G) and correlated with the increased abundance of pro-IL-1β and IL-1β (Figure 2H).

In summation, these results show that inflammasome complexes are constitutively formed in the LG of two murine pSS models of different background strains. Inflammasome activation gradually increases with the progression of the disease, suggesting that it could play a significant role in the pathogenesis through the secretion of inflammatory cytokines.

3.3. Acute Injury and Chronic Inflammation Upregulate Several Types of Inflammasomes

Key components of the inflammasome machinery are highly expressed during inflammation in different tissues [24,46,47,48]. To identify the signaling pathways promoting their transcription, we analyzed a previously published RNA-seq data (GSE99093) obtained for LG acute injury/regeneration in BALBc mice [13]. Similar to our experiments, the LGs of these mice were injured by the intraglandular injection of IL-1α and the bulk RNA-seq was performed at days 0 (uninjected control), 1, 2, 3, 4, 5, 7, and 14 after injury.

Consistent with our previous study [14], we found a strong induction of Il1b and Il18 transcription on days 1 and 2 (Figure 3A, Table S1). Therefore, we investigated the expression of all genes listed in the Gene Ontology Biological Process entitled “Interleukin-1 production” (GO:0032612) that were significantly enriched on days 1 and 2 after injury (Figure 3B, Table S1). Thus, the gene expression heatmap demonstrates that LG samples from day 1 and 2 after injury cluster together, showing an upregulation of many genes of this pathway, while data from day 3 after injury display a significant decrease in the set of genes involved in inflammasome activation and interleukin processing (Figure 3B). In agreement with the original study [13], uninjected (day 0) and saline-injected controls (day “14S”) were highly similar (Figure 3B) and, thus, we retained the uninjected LGs as controls to study the gene expression changes. The main cluster of genes upregulated during the inflammatory phase (lower part of the heatmap) included the genes of the toll-like receptor (TLR)-myeloid differentiation primary response 88 (MyD88)-nuclear factor kappa B (NFκB) axis that promotes the transcription of Il1b and inflammasome components [49,50], and its partners such as F2rl1 that acts synergistically with TLR2 and TLR4 [51,52], both upregulated in this dataset (see selected genes shown in frame on Figure 3B, Table S1). The inflammasome sensors Ifi204 (murine ortholog of human IFI16 [49]), Nlrp3, Aim2, Mefv (encoding PYRIN), and Naip5 (that forms hetero-oligomeric inflammasomes with NLRC4 upon recognition of bacterial fragments [53]) were significantly upregulated on days 1 and/or 2 after acute injury, and decreased to basal levels on day 3 (Figure 3B,C, Table S1). By contrast, there was no significant change in the expression of Nlrp6, Nlrc4, and Nlrp1b (Figure 3C). Nlrp1a was not expressed at any time (Figure 3C), which was expected, as the BALBc strain lacks Nlrp1a expression [54]. Casp1 and Casp4 were significantly increased during the first two days after injury, while the expression of both caspases returned to basal levels on day 3 after injury (Figure 3B and Figure S2A, Table S1). There was a modest increase in the Casp4 mRNA level at days 4 and 5 but, from day 7, its expression level was not different from the uninjected controls. In addition, we detected the upregulation of the NLRP3-activator Gbp5, and the mediators of pyroptosis Panx1 and Gsdmd (Figure 3B, Table S1).

The results obtained by qRT-PCR performed in our lab after acute injury of LG in C57BL/6J mice were highly similar to the RNA-seq data from BALBc mice (Figure S2B). This demonstrates that the experimental model of LG acute injury provides mechanistically robust and reproducible results, even when different mouse strains are used. Therefore, these findings suggest that several types of canonical inflammasomes (NLRP3, AIM2, IFI204, PYRIN, NAIP5/NLRC4, and possibly non-canonical inflammasomes) are transiently activated by IL-1α-injection.

We also analyzed the transcriptomic changes leading to inflammasome priming and activation during the development of chronic inflammation. To do this, we mined our previously published RNAseq data of LGs from 2M, 4M, and 6M NOD.H2^b (diseased) and BALBc (control) males (GSE210332) [37]. In this study, we showed that, although there is a major shift in the gene expression between 2M and 4M/6M mice, 2M NOD.H2^b LG already features many alterations at the transcriptomic level compared to BALBc controls. We also found that B and T cell infiltrates appeared as early as 2M (early stage of the disease) in NOD.H2^b males—although not to the same extent as 6M males (clinical stage).

Consistent with our previous observations, the expression of Il1b and Il18 was significantly increased with the disease progression (Figure 4A) and genes for the “Interleukin-1 production” (GO:0032612) pathway were significantly enriched in the list of differentially expressed genes (DEGs) between all diseased and control animals. Our data show a set of genes that are upregulated in NOD.H2^b mice (Figure 4B, Table S2). Among these genes, we found several pro-inflammatory factors, including TLRs, Tnf, and Nod2, as well as Ifng, (coding for IFN-γ) that participate in NF-κB activation (Figure 4B and Figure S3A, Table S2). IFN-γ not only promotes the transcription of inflammasome components, but also their assembly through the activation of GBPs [55,56]. In NOD.H2^b mice, the expression of genes for inflammasome sensors Nlrp3, Nlrc4, Naip5, Ifi204, Aim2, and Mefv were significantly increased from 2 months of age, while no significant changes were detected for Nlrp1b and Nlrp6 mRNAs (Figure 4C, Table S2). Although Nlrp1a is not expressed in BALBc mice, we noticed that its mRNA expression significantly increased with age in NOD.H2^b LGs. Importantly, the upregulation of inflammasome sensors was associated with the increased transcription of Casp1 and Casp4 (Figure 4B and Figure S3B), along with Gsdmd and Panx1 (Figure 4B, Table S2). While most of inflammasome-related genes were significantly upregulated at 2M compared to the healthy controls, the expression level of some of them (including Aim2, Nlrc4, Nlrp1a, Casp1, Panx1, and Gsdmd) increased further in NOD.H2^b LGs at 4M (Figure S3C). By contrast, none of the genes from the “Interleukin-1 production” pathway passed our thresholds when comparing 4M and 6M NOD.H2^b LGs. No DEGs were found for the 2M/4M and 4M/6M comparisons in the control BALBc LGs. This shows that, in NOD.H2^b LGs, the activation of the inflammasome pathway amplifies up to 4M, the age at which the chronic inflammation is established [37].

Therefore, during chronic inflammation, several transcriptional activators of inflammasome components are upregulated from the early stages of the disease. Compared to acute injury, which leads to the upregulation of Nlrp3, Aim2, Ifi204, Mefv, and Naip5, chronic inflammation also increased Nlcr4 expression, suggesting that most of the identified sensors can be involved in both acute and chronic inflammation. By immunostaining, we confirmed an increased signal for AIM2 expression in NOD.H2^b compared to BALBc LGs (Figure S3D,E). In NOD.H2^b LGs, AIM2 complexes were found in the perinuclear areas of epithelial cells, and eventually in other compartments, such as blood vessels (Figure S3F). The constitutive activation of inflammasomes in NOD.H2^b LGs could be also facilitated by the upregulation of the gene encoding SYK (spleen-associated tyrosine kinase) (Figure S3G), which induces ASC phosphorylation and oligomerization that is essential for the assembly of NLRP3 inflammasome and CASP1 activation [57]. Finally, the upregulation of Casp4 and members of IFN-γ signaling suggests the activation of the non-canonical inflammasomes.

Taken together, these results demonstrate that the expression/activation of multiple inflammasome complexes in the LG is transient during acute inflammation but is sustained during chronic inflammation. Both acute and chronic inflammasome activities likely lead to epithelial cell damage, the induction of adaptive immune response, and the formation of lymphocytic infiltrates. Therefore, we investigated whether molecular pathways promoting the resolution of inflammation and inflammasome inhibition after acute injury are altered during chronic inflammation.

3.4. The Resolution of Inflammation and LG Regeneration following Acute Injury Is Concomitant with the Activation of Lipid Metabolism

An analysis of the transcriptome of regenerating LGs determined day 3 as the critical time point for the switch between inflammatory and regenerative processes [13]. This time point is characterized by a significant decrease in the CD45⁺ (immune) cells in the LG, particularly neutrophils and monocytes as previously reported [13], and a decrease in the inflammasome components and a reduction in Il1b and Il18 expression down to basal levels. Therefore, we hypothesized that the genes involved in the resolution of inflammation and inhibiting the inflammasome signaling are upregulated on day 3 after LG injury.

We identified 337 differentially expressed genes (DEGs) on day 3 after the injury (relative to the uninjured control, log₂(FC) cutoff = ±log₂(1.5) and p-adj < 0.05), including 219 upregulated genes. An analysis of these 219 upregulated genes with Metascape demonstrated that most of the top 30 significant biological pathways were related to the activity of immune cells and inflammatory responses (Figure 5A). The protein-protein interaction enrichment analysis carried out with these 219 genes (Figure S4) identified 6 densely connected network components related to: (1) phagocytosis and tumor necrosis factor (TNF) production, (2) mitosis, (3) sterol biosynthesis, (4) leukocyte/myeloid activation, (5) inflammatory response, and (6) memory. The heatmap comparing the enrichment p-values between the first three days after injury revealed that the inflammatory pathways belong to the primary response highly activated on days 1 and 2. Indeed, these pathways become much less significant on day 3, meaning that many of these genes were not upregulated anymore (Figure 5A). Hierarchical clustering showed a group of pathways related to the metabolism and localization of lipids that were significantly enriched on day 3, but not significant on days 1 and 2. Nonetheless, some of the genes that were involved in sterol biosynthesis (Acat2), inert fat breakdown (Lipa), and lipid transport/signaling (Abcg1, Apobec1, Apoe, Lrp1, Ldlr, Pltp, Ptafr) were maximally expressed on day 2—except for Ldlr, whose expression level reached a peak on day 3 (Figure S5).

To exclude the genes belonging to the primary inflammatory response initiated on day 1 and to identify the mechanism(s) specifically promoting the switch to LG repair on day 3 after the injury, we selected genes specifically upregulated on day 3 after the LG injury (53 genes) and genes upregulated on day 2 with the expression level maintained or increased from day 2 to day 3 (13 genes) (Figure S6). Based on these criteria, we generated a list of 66 genes (Table S3) and performed gene set enrichment analysis with Metascape to interrogate several ontology sources (Gene Ontology, KEGG Pathway, Reactome, WikiPathways) (Figure 5B). The enriched ontology terms were grouped by similarity into clusters and named after the most significant pathway. Most of the resulting biological processes were interconnected and related to the biosynthesis and metabolism of lipids (Figure 5C). The most significant cluster with 18 differentially activated genes was “Cholesterol metabolism with Bloch and Kandutsch-Russell pathways” (Figure 5B). Among these genes, we identified Srebf1 reaching its highest level on day 3 after IL-1α injury (Figure 6A), and its inhibitors Insig1 and Prkaa2 (coding for a subunit of AMPK) (Figure S7A) [58,59]. Srebf1 encodes the transcription factor sterol regulatory element-binding protein 1 (SREBP-1), a master regulator of lipid homeostasis (Figure 6B). The translocation of SREBP-1 into the nucleus activates the transcription of genes regulating the biosynthesis and uptake of fatty acids and cholesterol [59] (Figure 6B). Consistent with SREBP-1 activation, many enzymes of the pathways processing acetyl-CoA for the biosynthesis of fatty acids and cholesterol reached their highest expression level on day 3 after injury (Figure S7B,C), including the rate-limiting enzymes acetyl-CoA carboxylase beta (ACACB) and squalene epoxidase (SQLE) (Figure 6A,B). Similarly, the gene encoding Acyl-CoA synthetase short-chain family member 2 (ACSS2) that catalyzes the formation of acetyl-CoA and the activation of fatty acids into fatty acyl-CoA (Figure 6B) was upregulated only on day 3 (Figure 6A). We also found an upregulation of the predicted lipase gene Gm8978 and other lipid-related genes: Aldh3b2, coding for the aldehyde dehydrogenase 3 family member B2 (ALDH3B2) protein that removes toxic aldehydes from lipid droplets, and Dgkg, encoding diacylglycerol kinase gamma (DGKγ), which produces phosphatidic acid (PA) by phosphorylating the second messenger diacylglycerol (DAG) (Figure S7D).

Taken together, these results confirmed that the inflammatory pathways significantly enriched among the 219 upregulated DEGs outlined the primary immune response that occurred within days 1 and 2. Furthermore, in contrast to cholesterol conversion and transport pathways that increased quickly after injury, most of the genes involved in lipid biosynthesis were upregulated only on day 3, suggesting that this transcriptional program may specifically control the resolution of inflammation and gland regeneration. The activation of lipid metabolism most likely promotes the synthesis of cell membranes (that is composed of triglycerides, phospholipids, cholesterol), adenosine triphosphate (ATP) production (through β-oxidation and the use of acetyl-CoA by the TCA cycle), and the generation of second messengers (i.e., PA and DAG) regulating various cell processes (Figure 6A), and may also repress the inflammasome pathway.

3.5. Lipid Metabolism Is Altered in Chronically Inflamed LG of NOD.H2^b Mice

We hypothesized that the mechanism(s) promoting the resolution of inflammation after acute injury might be altered in NOD.H2^b mice. We have recently reported that the downregulation of genes involved in fatty acid biosynthesis, TCA cycle, and fatty acid β-oxidation was associated with disease progression in NOD.H2^b LGs [37]. Since these pathways are interconnected and participate in lipid homeostasis (Figure 6B), we analyzed the genes of relevant pathways that were significantly enriched in the meta-analysis of BALBc/NOD.H2^b comparisons, according to Gene Ontology (p-elim < 0.05) and WikiPathways (p-adj < 0.05) (Table S4).

Among these genes, we found key regulators of lipid metabolism downregulated in the LG of NOD.H2^b mice as early as 2 months of age (Table S4). Most altered were the genes coding for peroxisome proliferator-activated receptor alpha (PPARα, encoded by Ppara, Figure 7A) and its partner retinoid X receptor alpha (RXRα, Rxra) that regulate fatty acid transport, oxidation ketogenesis, and promote SREBP-1 activity [60,61]. There was also a downregulation of Agt coding for the angiotensin precursor ANGT that promotes the transcription of lipogenic genes, such as Srebf1. Srebf1 itself was reduced (Figure 7A) along with its synergic partner ChREBP (encoded by Mlxipl, Table S4) that, together with SREBP-1, activates the transcription of genes involved in fatty acid biosynthesis.

Consistent with the downregulation of the PPARα-dependent transcriptional program, the expression levels of enzymes involved in mitochondrial fatty acid β-oxidation were also reduced in the LGs of the NOD.H2^b mice (Table S4, Figure 7B). Similarly, we noted the downregulation of enzymes mediating ketone catabolism such as Oxct1 (Table S4). Altogether, this suggests a reduced mitochondrial pool of acetyl-CoA available for the TCA cycle. The decreased TCA cycle activity most likely reduces ATP production and also the amount of citrate that can be used for acetyl-CoA synthesis in the cytosol (Figure 6A). We also found the downregulation of enzymes catalyzing the synthesis of acetyl-CoA from citrate, acetate, or pyruvate (Table S4), including ACSS2 (Figure 7C), which also activates fatty acids by the reaction with acetyl-CoA to form fatty acyl-CoA (Figure 6A).

Since acetyl-CoA is the primary substrate for de novo lipid biogenesis, we hypothesized that this process is altered in chronically inflamed LGs. In addition to reduced Srebf1 levels, many enzymes for fatty acid biosynthesis such as Acacb were indeed downregulated in diseased glands, thus suggesting a reduced generation of free fatty acids and activated acyl-CoA (Figure 7C,D and Figure 8, Table S4). The number of significant DEGs in NOD.H2^b LGs and the extent of the alterations in their expression level compared to their respective BALBc controls increased with age, especially between 2M and 4M (Figure 7C,D, Table S4). By contrast, the enzymes catalyzing the elongation of long-chain fatty acids (LCFAs) into very long-chain fatty acids (VLCFAs) (Elovl5, Elovl6, Hacd4) were upregulated in diseased LGs (Figure 7D, Table S4). VLCFAs serve as precursors for eicosanoids such as inflammatory prostaglandins, thus suggesting that eicosanoid metabolism was also altered in NOD.H2^b mice.

In addition, many enzymes of the pathways leading to cholesterol synthesis were upregulated during chronic inflammation, including the rate-limiting enzymes 3-Hydroxy-3-Methylglutaryl-CoA Reductase (HMGCR) and SQLE (Figure 7E,F, Table S4). We also noticed the downregulation of Cyp27a1 and Cyp46a, which convert free cholesterol into secreted metabolites. Finally, there was a significant increase in the expression level of Soat1 and Soat2 catalyzing the formation of cholesterol esters (Figure 7E,F, Table S4). Of note, the expression level of Srebf2, which preferentially promotes cholesterol biosynthesis, was not altered by chronic inflammation (Figure S8A). This supports our observation that fatty acids, but not cholesterol synthesis, are reduced during chronic inflammation.

Some of the genes altered during chronic inflammation were in the list of lipogenic genes upregulated on day 3 after injury so we analyzed the expression of the entire gene set activated at the beginning of the regenerative phase after acute injury (Table S4, Figure S8B). As expected, we found a significant downregulation of enzymes responsible for acetyl-CoA and fatty acid biosynthesis contrasting with the upregulation of genes related to cholesterol biosynthesis, transport, and lipases.

All these observations indicate an alteration of mitochondrial metabolism—with decreased acetyl-CoA and ATP production—and of de novo lipid biosynthesis using these metabolites, due to the downregulation of PPARα/SREBP-1-dependent transcriptional programs (Figure 8). It is possible that enzymes involved in eicosanoid metabolism might promote the synthesis of pro-inflammatory lipids over anti-inflammatory metabolites (Table S4, Figure 8). Considering the overall upregulation of enzymes of the mevalonate and lanosterol pathways (Figure 8), the chronically inflamed glands may accumulate cholesterol, as previously reported in NOD mice [62]. Together with the impaired mitochondrial and fatty acid metabolism, this may induce epithelial cell stress and promote inflammasome activation in the LGs of NOD.H2b mice. Few of these alterations were significantly aggravated between 2M and 4M/6M NOD.H2^b mice (Table S4). Only Acacb was further downregulated at each time-point, while it was not affected in BALBc mice. Altogether, this suggests that changes in lipid metabolism are one of the earliest mechanisms of disease development.

4. Discussion

In this study, we showed that several types of inflammasomes could be activated in LG epithelial cells during acute and chronic inflammation. This suggests that, similar to corneal epithelial cells [63], LG epithelial cells function as sentinel cells [64]. We also found that inflammasome activation precedes epithelial cell death after IL-1α-induced acute injury and discovered increased GSDMD cleavage in chronically inflamed LGs of NOD.H2^b mice. Studies in the salivary gland suggest that inflammasome activation and pyroptosis exacerbate inflammation through immune cell infiltrations in SS patients and promote salivary gland dysfunction [28,29]. To our knowledge, there is no such report about pyroptotic events in the lacrimal gland. However, topical administration of anakinra (IL-1 receptor antagonist) to the cornea improved the ocular surface integrity and tear secretion in Aire-deficient mice with SS-like disease [65], suggesting that secretion of IL-1β by epithelial LG cells during chronic inflammation could participate in corneal damage.

During acute and chronic inflammation, various types of inflammasome sensors are upregulated in the LG, thereby illustrating the complexity of the innate inflammatory response in this organ. Surprisingly, both acute and chronic inflammation activated similar inflammasome sensors: (1) AIM2 and IFI204 that are activated by DNA released from damaged mitochondria or dead cells; (2) NLRP3 that is activated by a wide variety of stimuli, including oxidative stress and stress-induced lipid signaling; (3) PYRIN, whose loss-of-function mutation causes the monogenic autoinflammatory disease familial Mediterranean fever [66]; and (4) NLRC4 that, if mutated, causes constitutive CASP1 cleavage in cells leading to severe autoinflammatory syndromes in humans [67]. In the LG, Nlrc4 was upregulated only by chronic inflammation but the transcription of its partner NAIP5 was increased during both acute and chronic inflammation. Our data also show that several signaling pathways essential for NF-κB/inflammasome activation, including TLR/MyD88, were upregulated by acute and chronic inflammation. MyD88-deficiency was shown to significantly dampen disease development in NOD.H2^b mice [68], thus supporting a critical role in SS pathogenesis for this signaling pathway. In our study, we also noted the robust upregulation of GBPs and SYK that were recently described as modulators of inflammasome signaling [69]. SYK controls the activation of AIM2 and NLRP3 inflammasomes by phosphorylating ASC [70], thereby promoting its oligomerization and the recruitment of pro-CASP1 [71]. GBPs are part of the interferon signature that is involved in the pathogenesis of SS [72] and may play a role in SS and other autoimmune diseases by regulating inflammasome activation. Taken together, we propose that the sustained activation of inflammasome pathways observed in NOD.H2^b and TSP-1^-/- mice contribute to LG chronic inflammation.

In the search for molecular suppressors of inflammasome signaling following IL-1α injection, we discovered that Srebf1 and the genes involved in lipid biosynthesis/transport were upregulated at the resolution of inflammation. As for now, the link between lipid metabolism and inflammasome activity was mostly studied in immune cells. During the acute phase of inflammation, macrophages accumulate cholesterol, which activates inflammasomes through TLR signaling [73]; for example, by forming TLR4-inflammarafts that upregulate Il1b as shown in microglia [74]. Oishi and co-authors elegantly showed that in macrophages, SREBP-1 is first inhibited and then induced by TLR4/MyD88 at the later stages of the inflammatory response to promote the synthesis of anti-inflammatory fatty acids that promote the resolution of inflammation [75]. Similarly, the SREBP-dependent lipogenic program is induced by the activation of CASP1 by NLRP3 and NLRC4 inflammasomes in vitro [76]. Whether similar crosstalk between SREBP-1 signaling and inflammasomes occurs in LG epithelial cells remains to be determined.

By contrast, we showed that reduced mitochondrial metabolism (fatty β-oxidation, TCA cycle) and fatty acid metabolism were associated with disease progression in NOD.H2^b mice and mainly affected LG epithelium [37]. NOD.H2^b LGs might display similar mitochondrial alterations as diabetic NOD mice [77] and the mitochondrial damage itself could enhance inflammasome signaling [78]. A decrease in acetyl-CoA pools may also have a profound effect on gene expression through protein acetylation [79,80] and impair de novo lipid biosynthesis. Our data indeed showed the downregulation of PPARα/SREBP-1 signaling and of downstream genes involved in fatty acid metabolism, while genes promoting the generation and transport of cholesterol were upregulated at the early stages of the disease. Thus, the activation of cholesterol biosynthesis could be SREBP-1-independent and/or higher free cholesterol levels would exert negative feedback on Srebf1 expression. We also found the downregulation of Cyp27a1 and Cyp47a1 catalyzing the conversion of cholesterol into the secreted form 25-Hydroxycholesterol that reduces Il1b transcription and CASP1 activation [81].

In agreement with our study, Wu et al. (2009) showed altered lipid homeostasis in the LGs of diabetic NOD males [62]. In this model, cholesterol ester accumulation preceded lymphocytic infiltration and was not a consequence of dacryoadenitis. Altered lipid homeostasis is not restricted to mouse models of SS, since fat deposition in the LG is a feature of SS patients [82]. We previously reported large cytoplasmic vacuoles in LG acinar cells of NOD.H2^b mice, suggesting they accumulate lipid droplets [37]. The activation of the lanosterol/mevalonate pathway could be a compensatory mechanism aimed at increasing other non-steroid products, for example, to rescue mitochondrial function [83]. The resulting cholesterol accumulation in epithelial cells might in turn downregulate SREBP-1-dependent lipid biosynthesis, promote lipotoxic damage, and activate inflammasome signaling.

Consistent with our findings, PPARα is also downregulated in experimental mouse models of LG inflammation induced by high-fat diets [84] or obstructive sleep apnea [85] that lead to lipid accumulation and dry eye symptoms. The latter can be alleviated by fenofibrate [84,85], an FDA-approved PPARα activator. Fenofibrate is a hypolipidemic drug that is used to treat the symptoms of high cholesterol and triglycerides in human. Recently, Guo and co-authors showed that fenofibrate improved tear production and corneal surface state and reduced lymphocytic infiltrates in the LGs of NOD/ShiLtJ mice through the modulation of Th17/T_reg cell differentiation [86]. Thus, promoting fatty acid β-oxidation and biosynthesis through the activation of PPARα/SREBP-1 signaling could inhibit pro-inflammatory pathways in the LGs of NOD.H2^b mice and promote regenerative processes.

The crosstalk between epithelial cells and lymphocytes plays a key role in SS development [87,88]. In fact, anti-inflammatory drugs such as Rituximab (B-cell depleting agent) and Anakinra (anti-IL-1) had only transient or no effects on SS patients [87,89]. We thus believe that anti-inflammatory molecules, combined with drugs that restore epithelial cell homeostasis, may lead to better outcomes. Therefore, one possible avenue of research is the combination of fenofibrate with iguratimod (anti-rheumatic drug inhibiting TNF-α, IL-1, IL-6, BAFF-R, CD38 signaling) or necrosulfonamide (inhibitor of GSDMD-pore formation) to decrease inflammation, reduce deleterious effects of inflammasome activation, and durably improve the epithelial function in SS.

5. Conclusions

In summary, our work shows that, in addition to secreting antimicrobial and immunoregulatory factors into the tear fluid, the lacrimal gland epithelium plays a pivotal role in the innate immune response by activating inflammasome signaling in response to exogenous or endogenous stimuli. The dysregulation of this protective defense mechanism during chronic inflammation is associated with an imbalance between the metabolic pathways producing fatty acids and cholesterol. These alterations contribute to pSS pathogenesis and thus represent a new avenue for therapeutics development.