Wild-type (C57BL/6) mice were purchased from Charles River Laboratories (Wilmington, MA). TSP-1^−/− mice were purchased from Jackson Laboratories (Bar Harbor, ME). They were then bred in-house at a pathogen-free facility at Boston University School of Medicine, Boston, MA. The experimental protocols strictly followed the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were in accordance with the institution’s guidelines.

Wild-type mice were treated topically with 10 ng flagellin (InvivoGen, San Diego, CA) bilaterally (5 µl/eye) for 7 days. Corneal barrier integrity was monitored for 4 weeks by fluorescein staining as described previously (12). Briefly, 1% sodium fluorescein (Sigma-Aldrich, St. Louis, MO, USA) was applied to each eye. Three minutes later, eyes were flushed with PBS to remove excess fluorescein, and corneal staining was evaluated using the slit-lamp microscope (Haag-Streit, USA) with cobalt blue light. Punctate staining was scored according to the standardized National Eye Institute grading system of 0–3 for each of the five areas of the cornea (13).

To prepare frozen sections of conjunctiva, eyes were enucleated with the lids intact, fixed in 4% formaldehyde in phosphate buffered saline (PBS) for 48 h, followed by cryoprotection in 10%–30% sucrose for 72 h before embedding in OCT. Tissue sections 7 µm in thickness were cut using a microtome and placed on slides to be stored at – 20°C until ready to use. Conjunctival explants were fixed in 4% formaldehyde before staining. Tissue sections and explants were blocked by PBS that contained normal goat, hamster, or donkey serum, 0.3%–1% Triton X-100 at RT for 1 h, and incubated overnight at RT with primary specific antibodies for CD11c (Biolegend, Dedham, MA), MUC5AC (Abcam, Cambridge, MA), MHC class II (BD Biosciences, San Jose, CA), or TSP-1 (Santa Cruz Biotech, Santa Cruz, CA). After a PBS rinse, fluorescence-conjugated secondary antibodies (Invitrogen, Thermo Fisher Scientific, Waltham, MA) were applied for 1 h in RT. Cell nuclei were counterstained with DAPI dye. Fluorescent staining in tissue sections was visualized using the FSX100 Olympus fluorescence microscope. Staining in conjunctival explants was visualized using a Zeiss LSM 700 confocal microscope equipped with a 63× oil objective. Images (z-stacks) were captured using the Zeiss ZEN software. Microscopic images were analyzed using NIH ImageJ software (14).

Goblet cells were obtained from mouse conjunctival pieces and grown in an organ culture as described previously (15). Briefly, conjunctival tissues were removed from 4- to 12-week-old male mice and kept in Hank’s based salt solution (Lonza, Walkersville, MD). Small pieces of conjunctival tissue were anchored onto a scratch at the bottom of a well in a 24-well plate. Culture medium (RPMI-1640) supplemented with 10% heat-inactivated fetal bovine serum, 10 mM HEPES, 100 µg/ml penicillin/streptomycin, 1 mM sodium pyruvate, and 1× nonessential amino acid mixture (Lonza, Walkersville, MD) was added to submerge the tissue pieces. These explants were fed every other day with RPMI-1640 medium and grown for 2 weeks at 37°C and 5% CO₂ to reach 85% confluence. Explants were later discarded, and goblet cell phenotype was confirmed by positive immunostaining for cytokeratin-7 and MUC5AC and negative staining for cytokeratin-4 (15). Goblet cells were cultured with heat-killed pathogenic Pseudomonas aeruginosa strain PA14 (kindly provided by Dr. Mihaela Gadjeva, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA) at a MOI of 60 or flagellin-PA (InvivoGen, San Diego, CA) at different concentrations (0–10 µg/ml) for a period of 24 h. No significant effect on cell viability was detected as determined using CCK-8 kit (Dojindo Molecular Technologies, Inc., Rockville, MD). Culture supernatants were collected to determine cytokine content and cells were used to harvest RNA.

Cultured goblet cells were stained with Fixable viability dye (eBioscience, San Diego, CA) and cells were immunostained with Alexa-647-conjugated anti-TLR5 antibody (Clone ACT5, Biolegend, Dedham, MA). Specificity of TLR5 staining was validated by confirming negative staining in RAW 264.7 cell line consistent with their minimal TLR5 transcript levels and functional responses to TLR5 agonist (16, 17). Fluorescence-labeled cells were analyzed using BD LSRII Flow Cytometer (BD Bioscience, San Jose, CA) at Boston University Flow Cytometry Core Facility. Further analysis of the data was performed using FlowJo software (Tree Star, Inc., Ashland, OR).

Active TGFβ content of culture supernatants was determined using fibroblast cell line, MFB-11, derived from TGFβ knockout mice and stably transfected with SBE-SEAP reporter (18). Cells in complete DMEM were seeded into flat-bottomed 96-well plates (3 × 10⁴ per well) and incubated overnight at 37°C, 5% CO₂. Cells were then washed twice with PBS and incubated with 50 µl of serum-free DMEM for 2 h before adding culture supernatants. Standard curve was set up using recombinant TGFβ2 (R&D Systems, Minneapolis, MN). After 24 h incubation, culture supernatants from each well were tested for SEAP activity using Great EscApe SEAP Reporter system 3 (Clontech, Mountain View, CA) according to the manufacturer’s instructions. Absorbance was measured using a Synergy H1 microplate reader (Biotek, Winooski, VT). Each sample was analyzed in triplicate.

Total RNA was isolated using TRIzol Reagent (Life Technologies, Carlsbad, CA) according to the manufacturer’s directions and reverse transcribed to generate cDNA using the Superscript VILO cDNA kit (Life Technologies, Carlsbad, CA). Real-time PCR was performed to determine relative quantitative expression levels of TSP-1, TGFβ2, and IL-6 on 7200 Real-Time System (Applied Biosystems, Carlsbad, CA) using SYBR Green PCR Master Mix (Life Technologies, Carlsbad, CA). Amplification reactions were set up in quadruplicate using specific primer sets with the thermal profile of 95°C for 3 min, 40 cycles of 95°C for 20 s, 55°C for 30 s, and 72°C for 40 s. A melting curve analysis was performed to verify the specificity of amplification reactions. Analysis of fluorescence signal generated at each cycle with system software generated threshold cycle values, and quantification of gene expression was determined relative to that of the reference gene Glyceraldehyde-3-phosphate dehydrogenase. Primer sequences used were TGFβ2 (F-5’-AGGCGAGATTTGCAGGTATTGA-3’ and R-5’-GTAGGAGGGCAACAACATTAGCAG-3’), TSP-1 (F-5’-AAG AGG ACC GGG CTC AAC TCT ACA and R-5’-CTC CGC GCT CTC CAT CTT ATC AC), IL-6 (F-5’-AGT CAA TTC CAG AAA CCG CTA TGA and R-5’-TAG GGA AGG CCG TGG TTG T), and GAPDH (F-5’-CGA GAA TGG GAA GCT TGT CA and R-5’-AGA CAC CAG TAG ACT CCA CGA CAT).

Differences between mean values were compared using Student’s t-test. A difference with p < 0.05 is considered statistically significant. The standard error of the mean is represented with error bars in figures. For corneal fluorescein staining scores, normal distribution of the pooled data for independent groups from a series of experiments was confirmed using D’Agostino and Pearson test supporting the assumption that data in this study are drawn from a normally distributed population. Homogeneity of variances among experimental and control groups are confirmed using F-test. Data analysis was performed using Excel (Microsoft Office) and GraphPad Prism software (GraphPad software Inc., San Diego, CA).

Similar to other mucosal surfaces, the presence of CD11c+ DC subsets have been reported in mouse ocular mucosa (19). These were identified by flow cytometry. In the intestinal mucosa, DC in the lamina propria extend trans-epithelial dendrites in response to epithelial cell TLR engagement (20). To evaluate the location and morphology of DC in the context of GC in the ocular mucosa, we evaluated conjunctiva explants from TSP-1^−/− mice, previously reported to harbor increased microbial frequency at the ocular surface (11), and from C57BL/6 mice as controls. A representative 3D reconstruction of confocal images of immunostained conjunctiva from TSP-1^−/− mice is shown in Figure 1 . Positively stained cell bodies of DC were detected below the epithelial layer containing MUC5AC+ GC ( Figure 1A ). While several DC were located in close proximity of GC, many DC extensions were visible on the ocular surface side of the tissue ( Figure 1B ). Some of these extensions had globular structure at the tip similar to that described by others in trans-epithelial dendrites (TEDs) of DC subsets in intestinal mucosa (20–22). In C57BL/6 conjunctiva, very few positively stained DC were located in the subepithelial region without clearly detectable trans-epithelial projections ( Supplemental Figure 1 ). The presence of TEDs in TSP-1^−/−, but not C57BL/6, conjunctiva correlates with the known increased microbial colonization of the ocular surface in TSP-1^−/− mice. These results demonstrate that DC in ocular mucosa can be located in close proximity of GC and extend trans-epithelial dendrites towards the ocular surface presumably to sample microbes.

Microbial keratitis caused by P. aeruginosa is the most common vision-threatening infection among contact lens wearers (23) and a relatively common but potentially serious cause of conjunctivitis in hospitalized preterm infants with significant morbidity and in some cases death due to systemic complications (24). Potential complications of keratitis include chronic inflammation and corneal scarring. To determine if GC responses may contribute to these complications, we examined their responses to pathogenic strain of P. aeruginosa (PA14). We stimulated primary GC cultures, derived from C57BL/6 conjunctiva explants, with heat-inactivated PA14 and detected significantly increased levels of IL-6 in culture supernatants ( Figure 2A ). As GC respond to LPS (TLR4 ligand) by increasing their secretion of active TGFβ, we determined if PA14 induces a similar response in GC. As shown in Figure 2B , GC stimulated with PA14 produced significantly reduced levels of active TGFβ as detected in a bioassay described in Materials and Methods.

Conjunctival GC are known to respond to TLR2- and TLR4-mediated signals (7, 8). To determine their TLR5 expression, cells from primary GC cultures were stained with anti-TLR5 antibody and analyzed by flow cytometry. Consistent with reported expression of TLR5 in human conjunctival epithelial cells (25, 26), GC expressed TLR5 ( Figure 2C ). To determine if reduced TGFβ secretion induced by PA14 resulted from combined TLR4/5 signaling or if it can be attributed to TLR5-mediated signaling in GC, we stimulated these cells with different concentrations of TLR5 ligand, flagellin derived from P. aeruginosa. As shown in Figure 3 , we detected dose-dependent increased expression of IL-6 message by real-time PCR ( Figure 3A ) in flagellin-stimulated GC that was accompanied by increased IL-6 protein detected in culture supernatants by ELISA ( Figure 3B ). Similarly, flagellin also reduced secretion of active TGFβ by GC in a dose-dependent manner ( Figure 3C ). Together, these results confirm the expression of functional TLR5 on conjunctival GC that, in contrast to TLR4-mediated signals, reduces GC-derived active TGFβ.

The ability of GC to secrete TGFβ and modulate DC phenotype implicates a potential role of GC in maintenance of immune homeostasis (8). To determine if flagellin-mediated reduced secretion of active TGFβ and increased secretion of IL-6 by GC disrupt ocular mucosal homeostasis and lead to the development of chronic ocular surface inflammation, we topically administered flagellin (10 ng/mouse) in mice daily for 1 week. Chronic inflammation of the conjunctiva is associated with the development of corneal epitheliopathy resulting from a disrupted corneal barrier. Therefore, we assessed corneal barrier integrity by determining corneal fluorescein staining score before initiating flagellin application (baseline) and comparing it to scores determined up to 4 weeks post-flagellin application. As shown in Figure 4A , a gradual increase in corneal fluorescein score led to a significant increase by 4 weeks after completion of flagellin application as compared to the baseline score. Consistent with this result, clinical signs of conjunctival inflammation were noted in mice. These included edema and hair loss around the eye caused by excessive grooming ( Figure 4B ). Figure 4C shows representative corneal fluorescein staining in untreated and flagellin-treated mice. These clinical signs correlated with inflammatory infiltrates observed in histological evaluation of conjunctiva from flagellin-treated mice ( Figure 4D ). Also, GC density appeared to be lower in flagellin-treated conjunctiva relative to that seen in untreated tissue. Together, these findings indicate development of chronic ocular surface inflammation in mice treated with topical application of flagellin.

We have reported previously that GC depend on their endogenously expressed TSP-1 to activate their latent TGFβ (8). Moreover, TSP-1 deficiency in mice results in spontaneous development of chronic ocular surface inflammation (9). Therefore, we evaluated change in TSP-1 expression in primary cultures of GC exposed to different concentrations of flagellin. As shown in Figure 5A , significantly reduced levels of TSP-1 message were detected by real-time PCR in flagellin-treated GC. This in vitro finding was also confirmed in vivo, by assessing TSP-1 immunostaining in conjunctiva from flagellin-treated mice. Figure 5B shows reduced TSP-1 immunoreactivity in conjunctival epithelium after flagellin application.

Immature or tolerogenic state of DCs in tissues under steady-state homeostatic conditions is characterized by their low expression of MHC class II. To determine if flagellin-induced reduced TSP-1 and active TGFβ expression by GC alter ocular mucosal homeostasis, we examined MHC class II expression of CD11c+ DC in flagellin-treated conjunctiva. As seen in Figure 6 , while CD11c+ DC are detectable in untreated conjunctiva, MHC class II expression of these cells was stronger in flagellin-treated conjunctiva. These results indicate loss of homeostasis in flagellin-exposed conjunctiva is consistent with reduced TSP-1 expression in this tissue.