Female-Specific Downregulation of Tissue Polymorphonuclear Neutrophils Drives Impaired Regulatory T Cell and Amplified Effector T Cell Responses in Autoimmune Dry Eye Disease

The concept of polymorphonuclear neutrophils’ (PMNs’) single-track fate as terminally differentiated and short-lived primary effector cells of host defense and acute inflammation is rapidly evolving. PMNs are the most abundant effector leukocytes in the innate immune system, and recent discoveries have dramatically expanded their function from simply killing pathogens to regulating innate and adaptive immune responses (1–6). Compelling evidence defined distinct PMN phenotypes that have immune-regulatory and immune-suppressive functions. Nontraditional functions for PMNs were reported in the ocular surface where innate and adaptive immune responses are tightly regulated to preserve the delicate visual axis. Unlike the paradigm pro-inflammatory role of recruited PMNs, PMNs in the cornea are critical for wound healing of minor epithelial injuries and drive nerve regeneration (7–11). These unexpected roles of PMNs in corneal inflammatory/reparative responses were attributed, in part, to their key role in generating specialized proresolving mediators (SPMs), of which the eicosanoid lipoxin A₄ (LXA₄) is the most abundant. This SPM has wide-ranging anti-inflammatory, wound healing actions, and it promotes clearance of apoptotic PMNs by macrophages in acute inflammatory responses (12).

Peripheral blood PMNs highly express the conserved 5-lipoxygenase (LOX), which is the rate-limiting enzyme for generating leukotriene B₄ (LTB₄) (13), an eicosanoid that amplifies host defense and recruits and controls effector T cells (14, 15). The diverse role of PMNs is underscored by the fact that 5-LOX is also the rate-limiting enzyme for LXA₄ formation. In tissues, LXA₄ synthesis requires the coordinated interaction of two enzymes, 5-LOX and 15-LOX, which are both expressed in mouse PMNs and, in humans, can be induced to switch the function of recruited peripheral blood PMNs from host defense to resolving acute inflammation (12, 16, 17). Hence, depending on the state of activation or, potentially, phenotype, PMNs can either amplify or resolve acute inflammation (16). Limited studies investigated whether the SPM LXA₄, like its proinflammatory counterpart LTB₄, has direct actions on T cells (18). However, several reports demonstrated interactions of PMNs with B and T lymphocytes and PMN regulation of lymphocyte function (2–5); hence, PMNs are ideally situated as regulatory cells at the interface of innate and adaptive immune responses.

Because of the normal immune-privileged state of the cornea, the ocular surface has been studied extensively as a model to understand regulation and suppression of effector T cell activation. Aqueous tear–deficient dry eye disease, which is one of the most frequent ocular morbidities, is an immune-driven ocular disease whose pathogenesis is initiated by disrupting suppression of effector T cell activation (19–21). Key features of the pathogenesis are activation and homing of effector T cells to the lacrimal gland, conjunctival goblet cells, and cornea, which initiate chronic inflammation and epithelial defects that can lead to blindness. Treatment options are limited to artificial tears, corticosteroids, and the immunosuppressant cyclosporine A. The etiology of this common ocular immune disease is unknown as is the cause for its striking prevalence in women (22, 23). We recently reported on marked sex-specific differences in corneal inflammatory/reparative responses in both humans and mice (24, 25). More importantly, we identified estrogen downregulation of the resident corneal LXA₄ circuit as a mechanism for the female phenotype of delayed epithelial wound healing and altered PMN functional responses (24).

The aim of this study was to determine whether PMNs and LXA₄ have a role in sex-specific dry eye pathogenesis, which has not been investigated. We report on a tissue population of PMNs in normal draining lymph nodes, lacrimal glands, and the limbus of the cornea. These tissue-resident PMNs express a highly amplified LXA₄ pathway and generate LXA₄ in draining lymph nodes. A striking sex-specific difference was that desiccating stress triggered downregulation of the tissue PMN–LXA₄ circuit selectively in females, which induced amplified effector T cell activation. More importantly, we were able to recapitulate this female-specific dry eye response by depleting LXA₄-producing tissue PMNs, which provides evidence for a potential key checkpoint system for maintaining appropriate Treg function and preventing aberrant activation of Th1 and Th17 cells in females.

Age-matched (6–10-wk-old) C57BL/6J female and male mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were subjected to a 12-h day/night cycle and fed a standard diet ad libitum (rat/mouse diet LM-485; Harlan Tekland, Madison, WI). All animal studies were approved by the University of California Berkeley Animal Care and Use Committee and were performed according to the U.S. National Institutes of Health’s Guide for the Care and Use of Laboratory Animals.

The standard model of desiccating environmental stress was used to induce immune-driven dry eye disease (20, 26–28). Briefly, mice were placed in cages with perforated screen walls and exposed to continuous airflow from fans in a low humidity (20–30%) cubicle. Lacrimal gland function was inhibited by injection with scopolamine hydrobromide (0.1 ml of 10 mg/ml, formulated in sterile saline; Sigma-Aldrich, St Louis, MO) for three or five consecutive days and a reduced dose (0.1 ml of 5 mg/ml) for 10 d. Scopolamine hydrobromide was injected s.c. three times per day (9 am, 2 pm, and 7 pm) into alternating hindquarters of mice. Age- and sex-matched untreated mice housed in a standard animal facility environment served as healthy controls. In selected experiments, mice were rendered neutropenic (29, 30) by i.p. injection of purified anti-Ly6g (1A8 clone, 200 μg; BD Pharmingen, San Diego, CA) 24 h prior to starting desiccating stress (first injection) and 2 d after induction of dry eye disease (second injection). Control mice received the same dose of serum-type IgG. Selected mice were treated topically (100 ng, three times a day) and systemically (1 μg, daily) with LXA₄ (Cayman Chemical, Ann Arbor, MI) or sterile saline alone (PBS [pH 7.4]) throughout 10 d of desiccating stress. Ethanol from LXA₄ stock solution was rapidly removed under a gentle stream of nitrogen, and autacoids were immediately resuspended in sterile saline and applied to the eye (5 μl/drop) (31, 32). Corneas with complete limbus, lacrimal glands, and cervical draining lymph nodes were excised surgically with sterile instruments and cleaned in ice-cold sterile PBS under a dissecting microscope. Lacrimal glands were weighed, and each draining lymph node (1.8–2.0 mm) was extracted. Isolated tissues were either snap-frozen for RNA/lipidomic/myeloperoxidase (MPO) analyses or immediately processed for flow cytometry/immunohistochemistry (IHC).

Clinical signs of dry eye were assessed by corneal fluorescein staining using 0.5 μl 2.5% Fluorescein Sodium (Bausch & Lomb), in accordance with the standard National Eye Institute scoring system (33, 34). Tear production was measured by the cotton thread test. Briefly, a phenol red thread (Zone-Quick; Showa Yakuhin Kako, Tokyo, Japan) was placed in the lateral canthus of the conjunctiva fornix of each eye for 30 s after excess tear had been removed for a standard time of 30 s, and tear distance (in millimeters) was read under a microscope (Carl Zeiss, Jena, Germany).

Tissue PMNs were isolated from lacrimal gland and cervical lymph nodes of normal female mice. Naive T cells and activated CD4⁺ T cells were isolated from cervical draining lymph nodes. Briefly, the tissues were minced and filtered through a 40-μm cell strainer (BD Falcon; BD Biosciences, San Diego, CA). After preparing a single-cell suspension, cells were negatively separated and isolated using neutrophil/CD4⁺ T cell isolation kits (Miltenyi Biotec, Bergisch-Gladbach, Germany), according to the manufacturer’s instructions. Inflammatory PMNs were collected from zymosan A–induced peritonitis exudates in C57BL/6 female mice. Briefly, normal female mice were injected i.p. with 1 mg zymosan A (Sigma-Aldrich) in 1 ml sterile HBSS. After 12 h, which is the peak of PMN infiltration in this model (35), peritoneal lavages that contained >90% PMNs were collected with sterile HBSS. Cells were stained with trypan blue and counted using light microscopy. The cell suspension was pelleted by centrifugation, followed by washing in RPMI 1640 with 5 % FBS. The cell pellet was resuspended (5 × 10⁵ PMN/ml) in 200 μl RPMI 1640 with 5 % FBS for histological analysis or was activated with calcium ionophore (37°C, 15 min, 5 μM) to establish endogenous lipid mediator formation.

Whole eyes and lymph nodes were removed and embedded in optimal cutting temperature compound (Sakura Finetek, Torrance, CA). The samples were allowed to set at −80°C for ≥2 h before being sectioned lengthwise into 5-μm-thick slices. Conventional smears on slides were prepared from isolated neutrophils. Sections and smears were stained with H&E for evaluating morphology to distinguish cell types.

Sections of whole eyes were processed according to conventional histologic techniques for periodic acid–Schiff staining. Briefly, histological sections were fixed in 4% paraformaldehyde, oxidized in 100 μl 0.5% periodic acid solution, and treated with 100 μl Schiff reagent. After computer capture through a 10× magnification setting via light microscopy (Carl Zeiss, Jena, Germany), goblet cell numbers were counted manually, and mucin area was assessed using ImageJ software by calculating area and density through intensity-threshold settings.

Immunofluorescence and deconvolution imaging were performed as described previously (36). In brief, corneas with complete limbus were fixed (2% formaldehyde), permeabilized (0.1% Triton X-100), and incubated with the following fluorescence-labeled mAbs: FITC- or PE-conjugated anti-Ly6g (1A8 clone; BD Pharmingen) for PMNs; FITC- or PE-conjugated anti-CD31 (MEC 13.3 clone; BD Pharmingen) for limbal vessel endothelium; PE-conjugated anti-CD3 (500A2 clone; BD Pharmingen) for T cells; and FITC- or allophycocyanin-conjugated anti-CD4 (RM4-5 clone; BD Pharmingen) for activated CD4⁺ T cells. Each step was followed by three washes with PBS. In all cases, controls using isotype- and species-matched Abs were negative. Radial cuts were made in the cornea so that it could be flattened under a coverslip, and the cornea was mounted in Celvol (Sekisui Specialty Chemical, Dallas, TX) containing 1 μg/ml DAPI (Sigma-Aldrich) to assess nuclear morphology. Image analysis and quantification of corneas were performed using a DeltaVision Elite deconvolution microscope (Applied Precision, Issaquah, WA). Whole mounts were evaluated using a 40× oil-immersion lens to assess each field of view across the diameter of the cornea (from limbus to limbus). Each field of view had a tissue diameter of 0.53 mm. The limbal region encompasses the limbal vessels, and the remaining regions include the avascular cornea. The graphical values were obtained by counting the total number of stained cells throughout the depth of the cornea from the epithelial to endothelial surfaces (a range ∼90 μm) in each of nine, 40× fields of view making up the diameter of a cornea (9, 36).

PMNs were quantified in lacrimal glands and lymph nodes at the indicated time points by measuring MPO activity as a specific and quantitative index of tissue PMN infiltration (7, 31, 32). In brief, tissues were homogenized in 450 μl 50 mM potassium phosphate buffer containing 0.5% hexadecyltrimethylammonium bromide (pH 6), followed by sonication, cycles of freeze-thaw. After centrifugation, MPO activity in supernatants was measured by spectrophotometry using o-dianisidine dihydrochloride oxidation as a colorimetric indicator. Calibration curves for conversion of MPO activity to PMN number were established with PMNs collected from zymosan A–induced peritonitis in C57BL/6J mice.

Lacrimal glands were digested with 2 mg/ml collagenase D and 0.5 mg/ml DNase I (both from Roche Diagnostic) in FCS-containing RPMI 1640 for 1 h at 37°C. Single-cell suspensions from the digested samples and draining lymph nodes were prepared with a 40-μm cell strainer (BD Falcon; Becton-Dickinson, Franklin Lakes, NJ). FcRs were blocked with anti-FcR mAb (BD Pharmingen), and cells were incubated with titrated amounts of fluorescent-labeled Abs: FITC-conjugated anti-Ly6g (1A8 clone; BD Pharmingen) for PMNs; PE-conjugated anti-CD45 (MEC 13.3 clone; BD Pharmingen) for leukocytes; PE-conjugated anti-CD3 (500A2 clone; BD Pharmingen) for T cells; FITC-conjugated anti-CD4 (RM4-5 clone; BD Pharmingen) for activated CD4⁺ T cells; allophycocyanin-conjugated anti–IFN-γ (XMG1.2; Tonbo Biosciences); FITC- or allophycocyanin-conjugated anti–IL-17 (eBio17B7; eBioscience); and allophycocyanin-conjugated anti-Foxp3 (3G3; Tonbo Biosciences). Isotype control was stained with the appropriately matched Abs. Data were analyzed with FlowJo software (TreeStar, Ashland, OR). The percentage of stained cells in the samples was calculated with respect to isotype control staining. Cell sorting was performed using a high-speed cell sorter (MoFlo; SX DakoCytomation, Fort Collins, CO). Each flow cytometry experiment was performed at least three times.

RNA from PMNs, naive T cells, and CD4⁺ T cells was isolated using an RNeasy Mini Kit (QIAGEN Sciences, Germantown, MD) and quantified by spectrophotometry. RNA was reverse transcribed using a High-Capacity cDNA Kit (Applied Biosystems, Foster City, CA). β-Actin was used as a reference gene. Real-time PCR was performed using SYBR Green Master Mix with a StepOne Plus qPCR system (both from Applied Biosystems), as described previously (24, 32, 37). Amplifications were run in duplicates, and efficiency curves for all primers (Supplemental Table I) were established. Comparative quantification of gene expression was performed by StepOne software (Applied Biosystems). Relative expression was expressed as fold change from a mouse reference universal RNA control that was generated by pooling mRNA from C57BL/6J mouse spleen and kidney.

Liquid chromatography–tandem mass spectrometry (LC/MS/MS)-based lipidomic analysis was used to identify and quantify lipid mediators, as previously described (32, 37–39). Class-specific deuterated internal standards (PGE₂-d4, LTB₄-d4, 15-hydroxyeicosatetraenoic acid [HETE]–d8, LXA₄-d5, DHA-d5, AA-d8) were used to calculate work-up and extraction recovery. The analysis was carried out with an Agilent 1200 Series HPLC, Shiseido Capcell Pak C18 column, and AB Sciex QTRAP 3200 mass spectrometer. Mobile phases used were A (71.9% water/28% acetonitrile/0.1% acetic acid [v/v/v]) and B (60% isopropanol/40% acetonitrile [v/v]) that were run as a gradient. Tandem mass spectrometry analyses were conducted in negative ion mode, and lipids were quantitated using scheduled multiple reaction monitoring mode using specific and established transition ions. Calibration curves were prepared from synthetic standards (Cayman Chemical).

All data are presented as mean ± SEM. The one-tailed, unpaired Student t test was used to evaluate the significance of differences between two sex-specific groups. One-way ANOVA with the Tukey post hoc test was used for overall statistical comparisons. The p values <0.05 were considered statistically significant. These tests were performed with GraphPad Prism software (GraphPad, San Diego, CA) and SigmaPlot (Systat Software, San Jose, CA).

Because of the female prevalence of dry eye in humans, all dry eye animal models are carried out in females. However, no study had investigated whether dry eye disease pathogenesis in animal models or humans exhibits sex-specific differences. A key feature of advanced aqueous tear–deficient dry eye is damage to the corneal epithelial layer. Hence, we investigated whether dry eye induces sex-specific ocular surface pathogenesis if matched male and female mice are exposed to the standard mouse model of aqueous tear–deficient dry eye disease (19, 26, 28). Desiccating stress induced significantly greater epithelial defects in females at 3 and 5 d compared with males using a standard clinical fluorescent scoring method (Fig. 1A). Extended desiccating stress eventually led to identical and marked increases in epithelial defects after 10 d in both males and females. Epithelial defect correlated with a sex-specific decrease in tear production in response to desiccating stress, as evidenced by significantly lower tear production in females after 3–10 d when directly compared to males (Fig. 1B). Another feature of this mucosal immune disease is loss of goblet cells in the conjunctiva. Conjunctival goblet cells (Fig. 1C) were counted manually, and total mucin area was quantified in eyelids of females and males as a marker of mucin secretion. Healthy male mice had significantly higher numbers of goblet cells compared with females, whereas no significant sex-specific difference was observed in total mucin area in the healthy conjunctiva. Desiccating stress reduced goblet cell numbers in both males and females after 10 d of dry eye. Consistent with sex-specific differences in corneal epithelial defects, desiccating stress also induced a greater decrease in the number of goblet cells and a 49 ± 11% greater decrease in mucin production in females versus males (Fig. 1C).

Using a combined approach of IHC and flow cytometry, we next assessed the presence of PMNs in uninjured corneas and during desiccating stress (3–10 d) in both males and females. IHC confirmed no significant or consistent PMN presence in healthy corneas or stressed corneas at any of the time points. However, we detected a population of CD45 (a pan leukocyte marker) and Ly6g (a selective PMN marker)-positive cells that exhibited the morphology and characteristic polymorphonuclear shape of PMNs in the corneal limbal region of healthy uninjured eyes (Fig. 2A–D). This small, but significant, population of tissue PMNs was located near, but not in, blood vessels throughout the limbus (Fig. 2C). Quantification by manual cell counting in a defined whole-mount grid (Fig. 2C) and flow cytometry analysis (Fig. 2D) established the presence of this basal tissue PMN population in healthy male and female eyes, with no sex-specific differences. Unexpectedly, desiccating stress, which is a well-defined trigger for ocular surface inflammation and immune responses, induced a marked decrease in the tissue PMN population at the corneal limbus. More importantly, this decrease was significantly amplified in females (70 ± 4%) compared with males (49 ± 4%) at 3 d, and the tissue PMN level remained depressed for 3–10 d based on IHC analysis (Fig. 2B, 2C). Analysis of the cornea with attached limbus by flow cytometry confirmed this and was consistent with limbus-alone IHC results (Fig. 2D), with the exception of day 10 for which flow cytometry did not detect a sex-specific difference.

Activation of effector T cells and the subsequent targeting of CD4⁺ T cells to the ocular surface and lacrimal gland are key features of immune pathology in human autoimmune dry eye disease and the desiccating stress mouse model (20, 26, 27). A low number of CD4⁺ T cells was present in the limbus of healthy eyes in both males and females; the number increased with the duration of desiccating stress, with a sharp increase (10–15-fold) in CD4⁺ T cells after 10 d (Fig. 2E, 2F). IHC and flow cytometry analysis both documented a marked sex-specific difference in the number of activated CD4⁺ T cells that was 64 ± 22% and 70 ± 17% higher, respectively, in females compared with males at 10 d. Analysis by flow cytometry captured and measured T cell populations in both cornea and limbus. The analysis established a continuous increase in CD4⁺ T cells from 3 to 10 d and higher levels of CD4⁺ T cells in healthy/uninjured corneas of females and at every time point during desiccating stress compared with matched males (Fig. 2F).

Ocular surface and lacrimal gland function as an integrated unit linked by sensory and autonomic nerves and the tear film. The lacrimal gland is a primary target for immune-driven dry eye disease in humans and mice (19, 20, 27). Hence, we investigated whether desiccating stress induces sex-specific CD4⁺ T cell activation and whether PMNs are presented in the lacrimal glands during the course of dry eye disease. Lacrimal glands from matched healthy males and females contained significant numbers of PMNs (CD45^highLy6g^high) whose identity was established by flow cytometry (Fig. 3A) and confirmed by MPO activity (Fig. 3B), a specific enzymatic marker of PMNs. Lacrimal glands from females contained 161% higher levels of tissue PMNs (78,269 ± 20,812) compared with males. Desiccating stress triggered a pronounced and early influx of PMNs (159,878 ± 25,862) into lacrimal glands of females, as evidenced by a 2-fold increase in lacrimal gland PMNs at day 3. The PMN response to desiccating stress in females was 1.57-fold higher than was the PMN response (101,1878 ± 17, 958) in males at day 3. With the exception of day 10, female mice consistently had significantly higher numbers of PMNs in lacrimal glands throughout the course of dry eye disease. Both females and males had low, but significant, levels of naive T cells (CD3^high) and activated CD3^highCD4^high T cells in healthy unstressed lacrimal glands (Fig. 3C). Coincident with elevated levels of tissue PMNs, the lacrimal glands of females also contained higher numbers of activated CD4⁺ T cells. Consistent with the pathogenesis of immune-driven dry eye disease, desiccating stress triggered a marked increase in lacrimal gland CD4⁺ T cells. Activation of T cells was amplified in females early in the time course (3 d) and at the peak of dry eye disease (10 d), as evidenced by 53 ± 19% and 69 ± 18% higher numbers of CD4⁺ T cells in females, respectively, compared with males.

Formation of eicosanoids or LXA₄ in lacrimal glands has not been determined, but the high number of tissue PMNs in the lacrimal gland sets in place a prominent resident biosynthetic route. LC/MS/MS-based lipidomic analysis demonstrated endogenous and high levels of LXA₄ formation (Fig. 3D) in healthy lacrimal glands and during the time course of dry eye disease in females (0.74–2.39 ng/10 mg) and males (0.44–1.21 ng/10 mg). Levels of LXA₄ formation (Fig. 3D) correlated with a sex-specific difference in lacrimal gland PMN content (Fig. 3A, 3B); accordingly, LXA₄ levels were markedly higher in females than in males in both uninjured lacrimal glands and during desiccating stress (3–5 d). Both LXA₄ formation and PMN number in the lacrimal gland peaked 3 d after initiating dry eye disease in females. Lipidomic analysis also demonstrated higher basal levels of PGE₂ (Fig. 3E) in the healthy lacrimal gland of females (179 ± 71 pg in females versus 46 ± 9 pg in males) and marked induction (3.4-fold) of PGE₂ formation in females at 3 d. PGE₂ is a pleiotropic regulator of inflammatory/immune responses and, in both males and females, its level remained elevated 3–5 d after initiating dry eye disease and returned to baseline by 10 d.

Cervical lymph nodes, which drain the head region, including the eyes, are critical sites for the induction of ocular surface immune responses (i.e., effector T cell activation). The presence or role of a PMN–LXA₄ circuit in lymph nodes has not been explored, even though cross-talk of PMN–T cells (2–4) has emerged as an important new concept, and direct regulation of T cell function by LXA₄ was reported (18, 40). Routine IHC analysis identified a population of Ly6g⁺ polymorphonuclear cells in cervical draining lymph nodes of healthy and normal male and female mice (Fig. 4A). PMN morphology of these cells was confirmed by H&E staining of lymph node sections (Fig. 4B). For further identification, lymph node PMNs were isolated by a magnetic cell separation and isolation system or FACS. The isolated lymph node PMNs expressed CD45^highLy6g^high and demonstrated characteristic PMN morphology (Fig. 4C). To quantify changes in lymph node PMN numbers, we employed two independent approaches using both established surface markers and selective Abs (CD45, Ly6g) for flow cytometry analysis and an MPO assay as a PMN-selective enzymatic marker. Flow cytometry and MPO analysis provided equivalent results (Fig. 4D, 4E). PMN content in cervical draining lymph nodes from healthy female mice was 4.2 ± 0.3-fold and 1.8 ± 0.1-fold higher than in matched males based on MPO and flow cytometry quantification, respectively.

Desiccating stress triggered a marked increase in lymph node PMNs in males that peaked at 3 d and demonstrated a 10.7 ± 0.2-fold and 2.1 ± 0.1-fold increase over baseline, as measured by MPO assay and flow cytometry, respectively. A striking female-specific response to desiccating stress was the marked decrease in lymph node PMNs (by 64 ± 7% and 76 ± 2%, as measured by MPO and flow cytometry analysis, respectively). In sharp contrast, in male lymph nodes, PMN numbers remained significantly elevated throughout the desiccating stress, whereas in female lymph nodes, PMN levels remained depressed and failed to return to baseline. Desiccating stress is a mouse model of immune-driven dry eye diseases (19, 20, 26); consistent with the model, CD4⁺ T cells increased in draining lymph nodes in both males and females, as assessed by flow cytometry analysis (Fig. 4F). However, a significant increase in CD4⁺ T cells was not observed until day 10 in males, whereas CD4⁺ T cells increased significantly as early as day 3 in females. The magnitude of CD4⁺ T cell activation in female lymph nodes was 48–94% higher at all time points compared with matched males. Coincident with the 64 ± 1% decrease in tissue PMNs in female lymph nodes, CD4⁺ T cells also increased by 80 ± 2% as early as 3 d after initiating dry eye disease (Fig. 4E, 4F). In contrast, lymph node PMN levels peaked at 3 d in males, and there was no change in CD4⁺ T cell numbers compared with baseline.

CD4⁺ T cells include autoreactive IFN-γ–producing Th1 and IL-17–producing Th17 effector T cells, as well as Tregs that suppress effector T cells cell activation. Dysregulation of Th1 and Th17 effector cells and impaired Treg function were implicated in driving immune pathogenesis in dry eye disease (41–45). Hence, we sought to determine whether desiccating stress induces sex-specific effector T cell activation, as well as Treg responses. Analysis by flow cytometry documented a marked sex-specific difference in the number of Th1 cells (Fig. 5B) and Th17 cells (Fig. 5C) that was 206 ± 41% and 89 ± 7% higher, respectively, in females compared with males. Desiccating stress also induced a 26 ± 1% decrease in Tregs (Fig. 5A) in female lymph nodes. In sharp contrast, Treg levels increased by 31 ± 8% in males compared with baseline, which directly correlated with a lower amplitude of effector T cell activation.

Depending on the state of activation, PMNs are a rate-limiting factor for LXA₄ formation at sites of innate immune responses. Hence, we assessed whether LXA₄ is formed in draining lymph nodes and how it correlates with tissue PMN content in females. Lipidomic analysis (Fig. 6A) established endogenous formation of LXA₄, as well as biosynthetic pathways markers for 5-LOX and 15-LOX, namely 5-HETE and 15-HETE, respectively. 5-LOX and 15-LOX synthesis of LXA₄ is the primary pathway for LXA₄ formation in tissues. During the coordinated synthesis, the 15-LOX product 15-HETE is a substrate for 5-LOX, which converts 15-HETE to LXA₄. LXA₄ levels and 5-LOX and 15-LOX activity in lymph nodes of healthy and uninjured females was 2.8–5.0-fold higher compared with males (Fig. 6B). The amplified LXA₄ pathway in females correlated with a 2–3-fold higher number of PMNs in female lymph nodes compared with those in males (Fig. 6C). Desiccating stress abrogated LXA₄ formation in lymph nodes starting at day 3 in females (Fig. 6B). Abrogated LXA₄ formation in females correlated with a marked decrease in basal lymph node PMN content (Fig. 6C) and decreased levels of the biosynthetic pathway markers 5-HETE and 15-HETE (Fig. 6B). In sharp contrast, 5-HETE and 15-HETE levels peaked at day 3 in lymph nodes from males, which correlates with the peak of PMN infiltration in male lymph nodes. Independently of PMN levels, LXA₄ formation decreased with induction of desiccating stress in male lymph nodes. PGE₂, a PG with pleiotropic inflammatory and immune regulatory action, was formed in lymph nodes of healthy uninjured males and females. PGE₂ levels markedly increased and peaked at 3 and 5 d in males and females, respectively. Analysis of lymph node PMN content and LXA₄ levels revealed a direct correlation that suggested PMN-dependent formation of LXA₄ in female lymph nodes (Fig. 6C), which also was observed in female lacrimal glands (Fig. 6D).

To assess whether tissue PMNs in corneal limbus, lacrimal gland, and cervical lymph nodes impact immune pathogenesis of dry eye in females, namely effector CD4⁺ T cell activation, we used an Ab approach to globally deplete PMNs (10, 30, 46). To ensure sustained neutropenia and to capture early immune activation, we selected 3 d of desiccating stress as the time point for these experiments. Peritoneal injection of a specific and established Ly6g (1A8) mAb reduced tissue PMN content (Fig. 7A–C) by 64% ± 5%, 90 ± 6%, and 52 ± 8% in the limbus, lacrimal gland, and cervical lymph nodes, respectively. Tissue neutropenia inversely correlated with a marked increase in effector CD4⁺ T cells of 277 ± 84%, 97 ± 19%, and 53 ± 17% in the corneal limbus, lacrimal gland, and cervical lymph nodes, respectively, indicating that tissue PMNs are a determinant for CD4⁺ T cell activation in the ocular surface. We next determined whether PMN depletion impacts ocular surface pathogenesis. Consistent with an amplified CD4⁺ T cell response, tissue neutropenia markedly increased dry eye–induced corneal defects by 75 ± 13% (Fig. 7D) and significantly decreased conjunctiva goblet cell numbers in females (Supplemental Fig. 1).

Because tissue neutropenia correlates with a marked expansion of CD4⁺ T cells in draining lymph nodes, we determined the differential effect of PMN depletion on Th1, Th17, and Treg populations in dry eye diseases. Tissue neutropenia caused a marked 123 ± 10% increase in Th1 effector T cells (Fig. 8B) and a 32 ± 5% increase in Th17 effector T cells (Fig. 8C) in the draining lymph nodes compared with control females that were subjected to 10 d of desiccating stress. The amplified Th1 and Th17 responses in neutropenic mice correlated with a 26 ± 5% decrease in regulatory T cells in draining lymph nodes (Fig. 8A). Taken together, these findings identify a previously unknown function for PMNs in regulating effector T cell responses triggered by ocular surface desiccating stress.

To determine whether PMN depletion and subsequently increased CD4⁺ T cell activation correlate with impaired tissue LXA₄ formation, we carried out lipidomic analysis of lacrimal gland and cervical lymph nodes because they are the primary sites for effector CD4⁺ T cell targeting and/or activation in aqueous tear–deficient dry eye disease. Three days of desiccating stress was selected for the lacrimal gland and 0 d (uninjured/healthy) was selected for lymph nodes because these time points in females represent the peak PMN content in these two tissues. Consistent with our hypothesis, PMN depletion caused a 41 ± 13% decrease in LXA₄ formation in the lacrimal gland, as well as the LXA₄ and PMN biosynthetic pathway markers 5-HETE (5-LOX) and 15-HETE (15-LOX) (Fig. 9A). In the draining lymph nodes, PMN depletion completely abrogated LXA₄ formation and significantly decreased 5-HETE and 15-HETE levels by 37 ± 14% and 64 ± 14%, respectively (Fig. 9B). In contrast, PMN depletion did not significantly change PGE₂ levels in female lymph nodes.

To determine whether isolated tissue PMNs have the capacity to generate LXA₄, we isolated PMNs from draining lymph nodes and lacrimal glands of normal, healthy female mice and activated them with calcium ionophore. The capacity to generate LXA₄ was compared with recruited inflammatory PMNs from female mice in which peritonitis was induced by the yeast Ag zymosan A. Tissue PMNs had a significantly higher capacity to generate LXA₄ than did peritoneal inflammatory PMNs (Fig. 10A), because LXA₄ formation was 313 ± 63% and 128% ± 31% higher in tissue PMNs from lacrimal glands and cervical lymph nodes, respectively. The functional upregulation of the LXA₄ circuit in tissue PMNs directly correlated with a marked difference in RNA expression of 5-LOX and 15-LOX (Fig. 10B). Expression of 5-LOX was 9–21-fold higher in inflammatory PMNs compared with tissue PMNs, whereas expression of 15-LOX, a marker of leukocyte housekeeping function (16), was 25–656-fold higher in tissue PMNs than in inflammatory PMNs. To determine whether T cells are able to respond directly to LXA₄, we assessed expression of the LXA₄ receptor ALX (mFPR-rs1 and mFpr-rs2) in isolated naive CD3⁺CD4⁻ or activated CD3⁺CD4⁺ T cells from draining lymph nodes (Fig. 10C). Quantitative PCR analysis demonstrated robust expression of both receptors in T cells, and expression of ALX2 (mFpr2, mFPR-rs2) was upregulated 4.4-fold in activated CD4⁺ T cells. To test the hypothesis that LXA₄ can regulate T cell function in vivo, we treated female mice with a combined topical (100 ng, three times a day) and systemic (1 μg, every day) dose of LXA₄ for 10 d after initiating dry eye disease (Fig. 10D–F, Supplemental Fig. 2). Consistent with the hypothesis that reduced levels of LXA₄-producing tissue PMNs are a significant factor in CD4 T cell activation, treatment with LXA₄ significantly inhibited CD4⁺ T cells in draining lymph nodes by 38 ± 10% (Supplemental Fig. 2A). Regulation of CD4 T cells by LXA₄ treatment correlated with attenuated dry eye pathogenesis, as evidenced by a significant increase in goblet cells and mucin secretion in the conjunctiva (Supplemental Fig. 2B). To define which CD4⁺ T cell types are regulated by LXA₄ treatment, we quantified Th1 cells, Th17 cells, and Tregs in draining lymph nodes by flow cytometry after 10 d of desiccating stress. LXA₄ treatment almost completely abrogated Th1 cells, as evidenced by 88 ± 4% decreases, and it decreased Th17 cells by 27 ± 4% (Fig. 10E, 10F). The marked decrease in effector T cells caused by LXA₄ treatment correlated with an increase in Tregs by 96 ± 22% (Fig. 10D).

A key feature of immune-driven aqueous tear–deficient dry eye diseases is activation of effector T cells and their autoimmune targeting of the ocular surface and lacrimal gland. The reason why this ocular surface disease has such a high prevalence in females remains poorly defined nor is it clear what causes the misguided activation of effector T cells (23, 47, 48). The ability of estrogen, and especially testosterone, to regulate lacrimal gland function and the inflammatory pathway in epithelial cells in vitro and in vivo implicates sex steroids as a likely factor in the pathogenesis of dry eye disease in females (47, 49–51). However, limited studies investigated whether there are sex-specific differences in the initiation, regulation, or outcome of inflammatory, immune, or injury responses in the eye. Using the established mouse model of desiccating stress that recapitulates key features of effector T cell–driven human aqueous tear–deficient dry eye disease (20, 26–28), we discovered marked sex-specific differences in ocular surface pathogenesis, activation of effector T cells and Tregs, and regulation of a previously unknown LXA₄-producing tissue PMN population. The standard mouse model of desiccating stress uses scopolamine, a nonspecific acetylcholine receptor antagonist, to reduce tear production in the lacrimal gland. A cholinergic–anti-inflammatory pathway has been implicated as a potential counterregulatory circuit in inflammatory diseases, and nicotinic acetylcholine receptors are expressed in lymphocytes, macrophages, and PMNs (52). No experiment has addressed whether acetylcholine regulates the LXA₄ pathway and explored whether scopolamine disrupts that counterregulatory circuit, which could be a factor in the disease pathogenesis.

In human keratoconjunctivitis sicca, infiltration of the lacrimal gland by autoreactive Th1 and Th17 cells leads to deficient tear production, which as a downstream effect can cause corneal epithelial defects (20). Despite subjecting matched male and female mice to identical desiccating stress and pharmacological inhibition of lacrimal gland function, we observed subtle, but robust, increased corneal epithelial defects, decreased numbers of goblet cells, and lower mucin and tear production in females. These findings suggest a female-specific epithelial response to desiccating stress.

Resolution of minor epithelial defects depends on the time course for restoring the epithelial barrier; we demonstrated previously that acute epithelial wound healing is delayed in female mice in response to repeated mild corneal abrasion injuries (24) or in human female patients treated for corneal fungal ulcers (25). The female-specific wound-healing response can be induced in males by activation of the epithelial estrogen receptor ERβ and is mediated, in large part, by functional downregulation of an epithelial LXA₄ circuit (24). Because epithelial injury responses are tightly linked to inflammation, it is important to note that activation of ERβ also reduces recruitment of PMNs in the healing cornea. However, these corneal PMNs, unlike in most other tissues, drive wound healing and nerve regeneration (7, 9, 10) and, thus, are critical to the resolution of injury responses. The unusual function of corneal PMNs in mild injury responses also suggests that the ocular surface, through unknown factors, promotes or recruits a novel functional phenotype of PMNs. It is likely that these regulatory PMNs are a significant factor in the female-specific corneal injury response. This hypothesis is strongly supported by our current findings that demonstrate striking sex-specific differences in tissue PMNs in the lacrimal gland and draining lymph nodes.

Normal lacrimal glands and draining lymph nodes in females contained two to four times as many PMNs as did those of matched males. More importantly, desiccating stress triggered an early female and tissue-specific PMN response. The most striking sex-specific difference in tissue PMNs was observed in draining lymph nodes. Desiccating stress triggered a 2–10-fold increase in male lymph node PMNs, which, in sharp contrast, resulted in a diametric response in females: a marked decrease in lymph node PMNs that remained depressed throughout the entire dry eye time course. Depressed lymph node PMN number correlated directly with an early and female-specific increase in activated CD4⁺ effector T cells (Th1 and Th17) and a decrease in Tregs, which suggest that these tissue-resident PMNs have a regulatory role in immune-driven dry eye disease.

Immune-driven dry eye pathogenesis has been largely attributed to aberrant activation of Th1 and Th17 and impaired suppression of these effector T cells by Tregs. However, the mechanism for the misguided activation of autoreactive T cell remains obscure. In view of the important interplay of innate and adaptive immune effector cells in initiating adaptive immune responses, it is likely that dysregulated interaction of effector cells in the ocular surface is a factor in aberrant activation of autoreactive T cells. Recent studies have begun to explore the role of macrophages in autoimmune dry eye syndrome (53). However, PMNs, which are the most abundant innate immune effector cell, have largely been ignored as a significant cell type in the pathogenesis of ocular immune responses. PMNs’ presumed role in the eye is host defense, and because of their proinflammatory actions are considered downstream and recruited effector cells in inflammatory diseases that cause collateral tissue damage. Hence, our identification of a population of tissue PMNs in the limbus, lacrimal gland, and cervical draining lymph nodes in normal healthy mice was unexpected. The fact that loss of tissue PMNs by either the female-specific response to desiccating stress or by Ab depletion correlated with amplified effector T cell activation in all ocular tissues strongly implicates tissue PMNs as relevant suppressor cells.

We used independent and established approaches/markers to confirm the identity of tissue PMNs that included cell morphology, MPO activity, the specific Ly6G (clone 1A8) Ab (3), flow cytometry, and IHC. The concept of a single primitive PMN cell type has evolved rapidly, and distinct populations of PMN-like effector cells have emerged that include myeloid-derived suppressor cells, tumor-associated neutrophils, B cell helper neutrophils, and suppressor neutrophils (3, 5, 6). Although there are numerous markers for the emerging PMN populations, no single or combined surface markers can clearly identify subpopulation of PMNs, especially in mice, and classification is often based on function (3). The relatively low number of ocular tissue PMNs is a hurdle for an extensive surface marker analysis. However, a unique feature of these tissue PMNs in the lacrimal gland and cervical lymph nodes of females is their 3–5-fold higher capacity to generate LXA₄ compared with recruited inflammatory PMNs. The amplified endogenous formation of LXA₄ by tissue PMNs is matched by the high expression of 15-LOX and the markedly lower expression of 5-LOX. It is important to note that 15-LOX expression is also associated with wound healing and regulatory function in macrophages, because it is a differentiation marker of the alternatively activated M2 phenotype but not inflammatory macrophages, and it is highly expressed in tissue-resident macrophages where it was identified as a critical pathway for maintaining immune tolerance (54–56). More importantly, induction of 15-LOX expression induces proresolving/anti-inflammatory function in recruited human peripheral blood PMNs (16). Multiple lines of evidence in our study provide strong proof of concept that LXA₄-producing tissue PMNs have regulatory functions in draining lymph nodes: isolated naive and CD4⁺ lymphocytes from draining lymph express ALX receptors, systemic treatment with LXA₄ inhibits activation of Th1 and Th17 effector T cells and upregulates Tregs, endogenous levels of LXA₄ in draining lymph nodes from females directly correlate with PMN numbers, and PMN depletion abrogates LXA₄ formation, amplifies effector T cells, and downregulates Tregs in lymph nodes. The impaired Treg responses in females and their regulation by tissue PMNs are of particular interest because the balance of Treg and effector T cells in regional draining lymph nodes is critical to execute healthy immune responses and avoid aberrant autoimmunity. Impaired Treg suppression of Th17 cells was identified as a key factor in inducing autoimmune dry eye diseases in mice (44). The factors that control Treg numbers and function in lymph nodes are unknown. Hence, our findings that LXA₄ treatment can markedly increase Treg numbers and rescue females from amplified Th17- and Th1-driven dry eye disease identify a novel immune-regulatory mechanism.

LXA₄ formation in lymph nodes from males, unlike females, did not correlate with changes in PMN numbers. These findings suggest that, in males, tissue-resident PMNs and PMNs that are recruited to tissues after initiating desiccating stress likely constitute distinct cell types.

The detailed mechanism for LXA₄ regulation of effector T cells or Tregs is of considerable interest and under investigation. However, direct regulation of lymphocytes by LXA₄ is strongly supported by recent reports (18, 57) that demonstrated that innate lymphoid cells, NK cells, and B cells also express the ALX receptor and are regulated by LXA₄. Several mechanisms identified for the direct suppressive activity of PMNs include arginase 1, ROS, and MAC1. It is important to note that a primary function and released product of activated PMNs are eicosanoids, and specific receptors for early-response lipid signals are expressed in many lymphocyte populations. A recent study (58) demonstrated that PMN recruitment to lymph nodes after immunization is responsible for the spread of the immune response by producing thromboxane A₂. In addition, LTB₄, another primary product of inflammatory PMNs, was established as a key factor in recruiting effector T cells (14, 15). The PMN profile of regulatory eicosanoids likely depends highly on the state of activation or the specific PMN phenotype. We discovered a tissue-resident PMN phenotype that generates LXA₄ in draining lymph nodes and regulates Th1, Th17, and Treg responses to ocular surface stress. Sex-specific downregulation of this resident PMN–LXA₄ circuit identifies a new regulatory system for initiating and preventing aberrant effector T cell activation and subsequent immune-driven ocular surface disease in females.

The authors have no financial conflicts of interest.