Sex-dependent differential increase of specialized pro-resolving mediators in extracellular vesicles secreted by human primary conjunctival goblet cells during allergic inflammation

Abstract

Aims

Conjunctival epithelium lines the inside of the eyelids and covers the sclera, thus providing stability to the eye surface. Goblet cells in conjunctival epithelium (CjGCs) are well known for their mucin-secretion function, which wet and protect the ocular surface, but other aspects are still not well understood. To expand our understanding beyond their mucin-secreting function, we investigated CjGC-secreted extracellular vesicles (EVs) and lipid mediators therein.

Materials and Methods

Using histamine-mediated allergic inflammation in human primary CjGCs (HCjGCs) as a disease model, we quantified using ELISA a proinflammatory mediator PGE2 and two specialized pro-resolving mediators (SPMs) LXA4 and RvD1 in EVs secreted during allergic inflammation.

Key findings

At 18 hr post histamine stimulation, the amount of LXA4 and RvD1 in EVs was notably higher compared to those in unstimulated. Interestingly, this increase was only observed in female EVs but not in males. The mean fold increase of LXA4 and RvD1 in female EVs was 3.9 and 3.4, respectively, but it was only 0.9 and 1.0 in male EVs. Supplying docosahexaenoic acid (DHA, the source of RvD1 and other SPMs) to the culture medium during the allergic inflammation resulted in even higher mean fold increase of 5.3 and 6.9 for LXA4 and RvD1 in female EVs, respectively, but it was only 0.5 and 0.8 in male EVs.

Significance

We conclude that HCjGCs show a clear sex difference in allergic response. Our results may also provide a new insight into the male predisposition to severe forms of allergic conjunctivitis and potential improvement in disease care in the clinic.

1. Introduction

The ocular surface is comprised of cornea and conjunctiva overlaid by the tear film [1]. The epithelium of the conjunctiva lines the inside of the eyelids and covers the sclera, thus providing stability to the transparent cornea and anterior segment of the eye. The goblet cells in the conjunctival epithelium (CjGCs) secrete mucins to wet and protect the ocular surface. Under- or over-secretion CjGC mucins due to inflammation causes ocular surface instability, visual disturbance, and disease [2,3].

Allergic conjunctivitis (AC) is an allergic inflammation developed in conjunctival epithelium, which is one of the most common ocular allergies affecting more than 40 % of the North American population [4]. Symptoms caused by AC include pruritis (itchy eyelid skin), excess tearing, conjunctival redness, foreign body sensation, mucous discharge, and eyelid swelling. Many individuals with AC also experience allergic rhinitis, asthma, eczema, food allergy, and eosinophilic esophagitis, which make daily life even more debilitating. One molecule that causes multiple types of allergic reactions is histamine. Released immediately after allergen sensitization by mast cells, histamine binds to histamine receptors (H1, H2, H3, and H4) on the mast cell surface to promote the production and secretion of proinflammatory mediators (PIMs) prostaglandins, leukotrienes, and thromboxanes, which collectively enhance the inflammatory response. Histamine binding to histamine receptors in CjGCs increases the intracellular Ca²⁺ concentration ([Ca²⁺]_i), activates ERK1/2, and induces mucin over-secretion in CjGCs [5].

At the peak of the inflammatory response, various specialized pro-resolving mediators (SPMs) start to sequentially appear at the site of inflammation. SPMs are a class of bioactive lipid mediators derived from omega-3 (ω−3) or ω−6 polyunsaturated fatty acids (PUFAs) that resolve inflammation. These SPMs include ω−3 PUFA-derived resolvins, maresins, protectins, and ω−6 PUFA-derived lipoxins [6]. SPMs provided by infiltrating immune cells (i.e., neutrophils, monocytes, and macrophages) collectively resolve inflammation by inhibiting apoptotic polymorphonuclear infiltration, neutrophil trans-epithelial migration, and proinflammatory cytokine release by epithelial cells, as well as by promoting macrophage efferocytosis. These extracellular SPMs also affect the function of CjGCs by counter-regulating histamine-stimulated mucin secretion in CjGCs [7–10].

The onset of an allergic reaction and its resolution are relatively well studied from the infiltrating immune cell standpoint; thus, the current regimen for treating allergic conjunctivitis (i.e., ocular antihistamine agents, mast cell stabilizers, ophthalmic steroids, and non-steroidal anti-inflammatory drugs) is focused on regulating immune cell activity or mitigating the action of inflammatory mediators [4]. Despite advancements in our understanding and treatment options for allergic conjunctivitis, how conjunctival epithelial cells respond to allergic inflammation is still unknown. More specifically, there is no report on whether conjunctival epithelial cells only indirectly respond to the extracellular mediators provided by infiltrating immune cells, thus merely controlling the level of mucin secretion from CjGCs or take active measures to resolve inflammation, such as by biosynthesizing SPMs.

We targeted our investigation to lipid mediators in extracellular vesicles (EVs). EVs are cell-secreted nano-vesicles made up of a lipid bilayer and carry nucleic acids, proteins, and lipids [11–13]. Due to their ability to mediate cell-to-cell communication in an auto-, para-, and endocrine manner, EVs carry different molecules that reflect surrounding biological condition. This dynamism makes them an important biological entity to investigate and understand health and disease better.

We report here that human primary CjGCs (HCjGCs) secrete biosynthesized lipid mediators (PIMs and SPMs) into EVs, and at 18 hr under allergic conditions mimicked by histamine stimulation the amount of these lipid mediators significantly increased only in female EVs, but not in male EVs. Our result expands our understanding of CjGC physiology beyond mucin secretion. It proves the existence of active measures that these cells take during allergic inflammation and also shows a notable sex difference in allergic response between males and females.

2. Materials and Methods

2.1. Human primary conjunctival epithelial cell (HCjECs) cultures

Postmortem human conjunctival tissues, 4 males and 3 females, were obtained from eye banks in the US, including Eversight Eye Bank (Ann Arbor, MI), Lions Eye Bank of Wisconsin (Madison, WI), Eye-Bank for Sight Restoration (New York, NY), and Lions Eye Institute for Transplant & Research (Tampa, FL). The use of the tissue was reviewed by the Massachusetts Eye and Ear Human Studies Committee and determined to be exempt and did not meet the definition of research with human subjects. Human conjunctiva tissue explants (2 ~ 4 mm² per explant) were placed in 6-well tissue culture plates and incubated with RPMI-based complete medium (RPMI 1640, 10 % fetal bovine serum, 25 mM HEPES, 2 mM glutamine, 2 mM non-essential amino acids, 2 mM sodium pyruvate, and 100 μg/ml penicillin–streptomycin). Once the outgrowth of primary cells was visible (usually after 3 days), medium was replenished every other day. Cells were grown out for 2 weeks before each experiment.

2.2. Immunocytochemistry

Trypsinized primary cells were passaged onto sterile glass coverslips and left undisturbed for 24 ~ 48 hr. These first passage cells were then washed three times with 1x phosphate buffered saline (PBS) and fixed using 100 % MeOH at room temperature for 10 min. Fixed cells were washed three times using 1x PBS and blocked with 1 % bovine serum albumin (BSA) for 1 hr at room temperature. After blocking, cells were stained for 2 hr at room temperature with primary antibodies. After incubation, cells were washed three times using 1x PBS at 5 min intervals and then incubated with a mixture of secondary antibodies, DAPI (ThermoFisher #R37606), and HPA (Helix pomatia agglutinin)-AF647 (ThermoFisher #L32454, dilution 1:200) for 1 hr at room temperature. After incubation, stained cells were washed four times using 1x PBS at 5 min intervals. Fluorescence was visualized under the Nikon 80i fluorescent microscope or Leica SP8 confocal microscope. A list of antibodies used in this study is provided in the Supplementary Table 1.

2.3. RT-qPCR

Total RNA was extracted from HCjECs using RNeasy Mini Kit (Qiagen #74104). cDNA was synthesized by reverse transcription from 1 μg RNA using iScript Reverse Transcription Supermix (Bio-Rad #1708841). 3 μL of synthesized cDNA was used per 10 μL qPCR reaction (iTaq Universal SYBR Green Supermix, Bio-Rad #172–5121). The absolute copy number of a gene per 1 million GAPDH (reference gene) was calculated using the equation (2^(-ΔCt)) x 1,000,000 (Livak and Schmittgen, 2001). Primer sequences (Supplementary Table 2) were obtained from the PrimerBank (https://pga.mgh.harvard.edu/primerbank/) and their specificities were confirmed by NCBI Primer BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/).

2.4. Western blotting analysis

Crude cell lysates collected from HCjECs in RIPA buffer (ThermoFisher #89901) were spun down at 12,000 x g for 20 min in 4 °C. Supernatant was transferred to a clean tube and measured for a protein concentration using Pierce BCA Protein Assay Kit (ThermoFisher #23227). 30 μg cell lysate was separated in 12 % Mini-PROTEAN TGX Precast protein gel (Bio-Rad #4561003) and transferred into a nitrocellulose membrane using Trans-Blot Turbo Mini Transfer Packs (Bio-Rad #1704158). The membrane was blocked using 5 % BSA solution prepared in 1x TBST (Tris-buffered saline + 0.1 % Tween 20) for 1 hr at room temperature. Then, the membrane was incubated with primary antibody for 2 hr at room temperature on an orbital shaker. Unbound primary antibody was washed out 3 times with 1x TBST at 5 min interval. Washed membrane was incubated with secondary antibody for 1 hr at room temperature on an orbital shaker. After removing the secondary antibody, the membrane was washed 4 times with 1x TBST at 5 min intervals. Washed membrane was developed using Pierce ECL Western Blotting Substrate (ThermoFisher #32209) for chemiluminescence imaging.

2.5. Measurement of mucin 5AC (MUC5AC) secretion

Trypsinized primary HCjGCs were passaged onto 24-well plate at a density of 0.3 × 10⁶ cells/well. After 24 hr, cells were serum-starved for 2 hr and then stimulated with either 1x PBS (vehicle control) or histamine (Sigma-Aldrich #53300) for 4 hr. The media collected after 4 hr were cleared of live (300 × g, 5 min) and dead cells (2,000 × g, 20 min) by sequential centrifugation. Cleared supernatant was subjected to enzyme-linked immunosorbent assay (ELISA) to quantify secreted MUC5AC (Novus Biologicals # NBP2–76703).

2.6. Measurement of intracellular calcium concentration

Trypsinized HCjGCs were passaged onto glass bottom dishes and incubated for 24 hr at 37 °C. Cells were washed with Hank’s balanced salt solution (HBSS) twice and then incubated with 1 μM Fura-2 AM (Invitrogen #F1201) prepared in HBSS under dark for 1 hr at RT. After washing twice using fresh HBSS, cell fluorescence was recorded in real-time using a microscope. Fluorescence was then converted into Ca²⁺ concentration based on precalibrated values. Twenty-five cells were analyzed and averaged per tissue for total of three tissues. Changes in peak intracellular calcium concentration (Δ[Ca²⁺]_i) was calculated from the concentration before and after histamine stimulation.

2.7. Biophysical characterization of extracellular vesicles (EVs)

HCjGCs were cultured in regular RPMI complete media for one week and then cultured in RPMI medium supplemented with exosome-depleted FBS (System Biosciences #EXO-FBS-50A-1) for another one week. Collected medium was cleared of live (300 × g, 5 min) and dead cells (2,000 × g, 20 min) by sequential centrifugation. EVs were isolated from cleared supernatant using Total Exosome Isolation Reagent (ThermoFisher #4478359) following manufacturer’s protocol. Culture media were not concentrated before applying the EV isolation reagent. Culture media from each donor were not mixed with other donor’s but processed and analyzed individually. EVs were split into three aliquots for single vesicle morphology analysis, EV size measurement, and western blotting analysis. Transmission electron microscopy (TEM) was used for single vesicle analysis and nanoparticle tracking analyzer ZetaView (Particle Metrix GmbH, Germany) was used for the EV size measurement.

2.8. Transmission electron microscopy (TEM)

Every step was performed at room temperature under dry conditions unless specified. The protocol employed to negatively stain EVs was based on the protocol published by Théry et al [14]. EVs were fixed using equal volume of 4 % paraformaldehyde (PFA) prepared in 0.1 M sodium phosphate buffer for 30 min (2 % PFA final). Fixed EVs were then adsorbed onto a Ni coated EM grid for 20 min. Using forceps, the EV adsorbed grid was rinsed with 1X PBS for 8 times at 2 min interval by dipping the grid into a 100 μL drop. The rinsed grid was dipped into 50 μL of 1 % glutaraldehyde prepared in 0.1 M sodium phosphate buffer for 5 min. The grid was washed with deionized water for 8 times at 2 min interval by dipping the grid into a 100 μL drop. The washed grid was then dipped into 50 μL uranyl acetate-oxalate solution for 5 min followed by 50 μL methyl cellulose-uranyl acetate solution for 5 min on ice. The grid was suspended using a stainless-steel loop, removed of excess liquid using a soft tissue, and air-dried with the EV side facing up. EVs on the grid were observed under the TEM (FEI Tecnai G2 Biotwin Spirit TEM with a AMT Camera) at 100 kV.

2.9. Quantification of lipid mediators in EVs

Method used for Figure 3D:

Figure 3. HCjGCs have lipid synthesizing enzymes.

(A) Total RNAs extracted from HCjGCs were reverse transcribed and quantified (RT-qPCR) for six enzymes involved in synthesizing SPMs and PIMs. CK7 served as a positive control for HCjGCs. Mean ± SEM. (B) A subset of male and female HCjGCs used for RT-qPCR were fixed and fluorescently probed for 5-LOX, 12-LOX, 15-LOX, COX-1, COX-2, and FLAP. HPA staining served as a marker for HCjGCs. Scale bar = 20 μm.

Primary HCjGCs cultured in RPMI-based complete medium for about 2 weeks were washed two times using HBSS and incubated with HBSS for 2 hr. After 2 hr of serum starvation, fresh HBSS was added and incubated at 37 °C for 4 hr.

Method used for Figure 4:

Figure 4. Biophysical characterization of EVs secreted by HCjGCs.

(A) EVs were visualized under the transmission electron microscope (TEM) for a single vesicular analysis. (B) A nanoparticle tracking analyzer (NTA) is used to measure the size of EVs. Representative analysis from n=3 is shown. (C) EVs were positive for CD9, ALIX, CD81, and CD63 but negative for GM130 and calnexin. 20 μg proteins per lane. M = molecular marker. It should be noted that CD63 staining can appear above its expected 20 ~ 60 kDa range depending on the cell type, which is validated by the manufacturer. (D) HCjGC-secreted EVs carry lipid mediators, such as RvD1 (resolvin D1) and PGE2 (prostaglandin E2). n = 3 females, Mean ± SEM

Primary HCjGCs serum-starved for 2 hr in HBSS were given fresh HBSS containing 0.35 % of lipid-free bovine serum albumin (BSA), +/− 10 ng docosahexaenoic acid (Cayman chemical #90310), and +/− 10⁻⁴ M histamine. Cells were then incubated at 37 °C for 18 hr.

Method used for both figures:

Collected HBSS media after 4 hr or 18 hr incubation were cleared of live (300 × g, 5 min) and dead cells (2,000 × g, 20 min) by sequential centrifugation. EVs were isolated from the cleared supernatants using Total Exosome Isolation Reagent (ThermoFisher #4478359) following manufacturer’s protocol. Isolated EVs were then subjected to the ELISA against RvD1 (Cayman chemical #500380), LXA4 (Cayman chemical #590410), and PGE2 (Cayman chemical, #514010). Total protein amount in EVs measured using CBQCA Protein Quantitation Kit (ThermoFisher #C6667) was used to normalize the ELISA results.

2.10. Statistical analysis

Data are presented as Mean ± SEM. Student’s t-test and one-way ANOVA was performed using GraphPad Prism (Arlington, MA, USA). A p-value less than 0.05 was considered a significant difference. *p<0.05, **p<0.01.

3. Results

3.1. Validation of functional human primary conjunctival goblet cells (HCjGCs)

There are three major types of epithelial cells in the conjunctival epithelium, goblet cells, stratified squamous cells, and undifferentiated cells. To demonstrate that HCjGCs were the only cell types in the cultures, three molecular markers were detected using antibodies or lectin: the secretory product MUC5AC using anti-MUC5AC antibody, non-specific secretory products using the lectin HPA, and the cell body using anti-cytokeratin 7 (CK7) antibody. CjGCs’ unique function is to secrete mucins, such as MUC5AC and other highly glycosylated proteins [3]. These are packaged within secretory granules densely packed in the cytoplasm of the CjGCs. Once secreted, they become part of the muco-aqueous layer of the tear film and protect the ocular surface. CK7 is a type of cytoskeleton protein abundantly expressed in CjGCs. All these three have been consistent markers for HCjGCs as they are not observed in other two cell types in conjunctiva [15,16]. In addition to these three HCjGCs markers, cytokeratin 4 (CK4) was used as a marker for stratified squamous cells. ABCG2, predominantly found in cells with stem cell-like features in the conjunctiva, was used alongside other widely accepted ocular surface epithelial stem cell markers p63α, ΔNp63α, and Pax6 to identify and determine the abundance of undifferentiated cells in the culture [17].

The primary cells expanded from postmortem human conjunctival tissue explants showed a similar morphology to the primary goblet cells previously observed in human conjunctival tissues and those of rabbits and rats [16,18,19] (Figs. 1A). By immunofluorescence microscopy, cells positive for MUC5AC, HPA-labeled secretory granules, and CK7 were negative for CK4 (Fig. 1B & 1C). Although it was not detected fluorescently, a minute amount of CK4 was detected by Western blotting; however, it is highly likely transiently expressed in undifferentiated cells before they differentiate into HCjGCs (Fig. 1D). Some cells were positive for ΔNp63α, p63α, ABCG2, and Pax6, indicating existence of undifferentiated cells (Figs. 1B, 1E) [17]. From these, we concluded that the primary cultures are a mixture of undifferentiated cells, cells gradually differentiating into HCjGCs, and majority of fully differentiated HCjGCs, of which undifferentiated cells only differentiate into HCjGCs.

Figure 1. Validating molecular signature of human primary conjunctival goblet cells (HCjGCs).

(A) Primary cells propagated from postmortem human conjunctival tissue explant (black piece at the center, within the yellow outline) were visualized using brightfield microscopy. The yellow outline defines the outermost edge of the cell layer. (B) Cultured primary cells were positive for CjGC marker MUC5AC and undifferentiated cell marker ΔNp63. (C) The same set of cells used in (B) was positive for CK7 and Pax6. (D) Cell lysates contained a negligible amount of CK4 but were abundant in CK7, showing a lack of stratified squamous cells in the culture. (E) Positivity for HPA (Helix pomatia agglutinin), ABCG2, and p63α indicates that the cells in cultures are a mixture of undifferentiated cells, cells gradually differentiating into CjGCs, and fully differentiated CjGCs. Unless stated otherwise, scale bar = 20 μm.

The HCjGCs were fully functional. Under serum-free condition, HCjGCs constantly secreted MUC5AC to the extracellular space at a rate of 6 ng per mg cell lysate. (Fig. 2A). MUC5AC secretion was elevated in response to histamine (allergic mediator) stimulation to a similar level that others have observed [20] (Fig. 2B). Next, HCjGCs were tested for its calcium mobilization capacity upon histamine stimulation as calcium signaling is central to diverse physiological actions in cells, including biogenesis and release of extracellular vesicles [21–23]. HCjGCs properly increased intracellular calcium concentration immediately upon histamine stimulation (Fig. 2C), which was significantly higher than mock treated cells (Fig. 2D).

Figure 2. Validation of functional human primary conjunctival goblet cells (HCjGCs).

Cultured primary cells (A) constantly secrete MUC5AC to the extracellular space in resting conditions, and (B) secrete more MUC5AC in response to histamine stimulation. These indicate that primary cells expanded from tissue explants are functional HCjGCs. n=6 for (A) and n=3 for each concentration in (B). (C) Real-time recording of fluorescent Ca²⁺ indicator Fura-2 AM within cells stimulated with and without histamine. Representative trace is shown from n=3. (D) Histamine immediately induced a significant increase in intracellular calcium concentration. n=3. Mean ± SEM. **p<0.01

3.2. Presence of lipid synthesizing enzymes in HCjGCs

Arachidonic acid (AA), docosahexaenoic acid (DHA), or eicosapentaenoic acid (EPA) released from the membrane bilayer by phospholipase A2 (PLA2) are subjected to the enzymatic activity of cyclooxygenases (COX-1 and COX-2) and lipoxygenases (5-LOX, 12-LOX, and 15-LOX) producing various lipid mediators [24,25]. These lipid mediators are then secreted into the extracellular space. Given that immune cells secrete lipid mediators, it is logical that part of the SPMs in tears might have come from them [6]; however, it is not well known whether ocular surface epithelial cells can also provide SPMs to tears. The existence of COXs and LOXs in male and female HCjGCs was thus determined by RT-qPCR and immunocytochemistry.

The amount of RNA transcripts of 5-LOX, 12-LOX, 15-LOX, and FLAP (5-LOX-activating protein) was similar between sexes (Fig. 3A). There was a non-statistically significant increase in COX-1 and COX-2 transcripts in males compared to females. The quantification of the CK7 transcript served as a positive control to verify the HCjGC’s identity. The proteins of these transcripts were also fluorescently visualized in HCjGCs (Fig. 3B). All enzymes were detected in the cytoplasm except 5-LOX, which was found in the nucleus. 12-LOX, 15-LOX, and COX-2 showed a relatively lower immunofluorescence signal than 5-LOX, FLAP, or COX-1, suggesting their low abundance in HCjGCs. HPA staining was used to confirm the HCjGC identity. From these results, we validated the existence of lipid mediator synthesizing machinery in HCjGCs.

3.3. Biophysical characterization of extracellular vesicles (EVs) secreted from HCjGCs

Lipid mediators become a tear component as soon as they are secreted into the extracellular space. Some may find their target immediately but a portion of them may spend hours on the ocular surface before reaching their destination. In case of the latter, a protective mechanism may be necessary to ensure their structural integrity. One of the mechanisms would be EVs, a cell-made nanoparticle. Although HCjGCs are assumed to secrete EVs as almost all cells do, given that there are no published reports, HCjGC-secreted EVs were collected and biophysically characterized. When visualized under a transmission electron microscope (TEM), EVs were found as cup-shaped or intact spheres (Fig. 4A). Molecularly, EVs were positive for CD9, ALIX, CD63, and CD81 (Fig. 4B), but negative for GM130 (a marker for Golgi apparatus) and calnexin (a marker for endoplasmic reticulum), which are widely accepted molecular markers of EVs [26]. In nanoparticle tracking analyzer (NTA), EV’s median and mean hydrodynamic diameter were observed at 151.9 ± 23.8 nm and 175.0 ± 21.5 nm (n=3), respectively (Fig. 4C). Blank medium that underwent an identical EV isolation procedure did not show any nano-sized entities; hence, EVs were exclusively from HCjGCs. Analyses using TEM, NTA, and Western blotting confirmed that the EVs secreted by HCjGCs share similar size, molecular markers, and morphology to EVs secreted by other ocular surface cells, such as corneal epithelial cells, corneal myofibroblasts, and conjunctival mesenchymal stromal cells [23,27,28]. We chose to detect two representative lipid mediators in EVs that play a crucial role in inflammation and resolution: PGE2 (a type of PIM) and RvD1 (resolvin D1, a type of SPM). Both of them were detected in the EV samples, which confirmed that EVs are a good source to study lipid mediators (Fig. 4D).

3.4. Only female EVs show significant change in the amount of SPMs after histamine stimulation

From lipidomic approach, we recently found out that human tears contain an array of SPMs (including LXA4, RvD1, RvD2, RvD5, and PD1) as well as multiple types of PIMs (such as PGE2 and LTB4) with a notable sex difference [29]. For example, while PGE2 and LTB4 were found in tears of all six males, only three (out of six) female tears had these. Some males had RvD1, RvD5, and PD1 in their tears, whereas none of them are found in female tears. Since any molecule that is secreted from the ocular surface epithelial cells becomes a tear component, we tested whether secretion of lipid mediators from HCjGCs shows a sex-based difference. We chose two SPMs, resolvin D1 (RvD1, one of a D-series resolvin) that is biosynthesized from DHA and lipoxin A4 (LXA4) biosynthesized from arachidonic acid (AA), and one PIM prostaglandin E2 (PGE2) also biosynthesized from AA.

Although 10⁻⁵ M histamine was enough to test the functionality of HCjGCs (Fig. 2), we stimulated HCjGCs with 10⁻⁴ M histamine for this experiment because 10⁻⁵ M caused a high variation in results, highly likely due to the tissue donors’ diverse medical background. At 18 hr post stimulation, there was a notable change in the amount of lipid mediators in EVs. In response to histamine, the amount of RvD1, LXA4, and PGE2 increased in female EVs (EVs secreted from female HCjGCs) in general, but no trend was observed in male EVs (EVs secreted from male HCjGCs) (Fig. 5A, grey bars 0 vs. 10⁻⁴). We then tested if supplementing docosahexaenoic acid (DHA) can increase the amount of SPMs in male EVs. DHA is a type of ω−3 PUFA, a precursor to three families of SPMs: resolvins (includes RvD1), protectins, and maresins [30]. In female EVs, the addition of DHA resulted in a higher amount of RvD1 at both basal and histamine stimulation, whereas males show no difference in RvD1 (Fig. 5A, white bars vs. grey bars in RvD1). The addition of DHA decreased LXA4 (not a derivative of DHA) amount in female EVs under basal conditions (likely at the expense of increased RvD1 amount), but its amount was appropriately increased upon histamine stimulation (Fig. 5A, white bars vs. grey bars in LXA4). The addition of DHA elevated LXA4 in male EVs under basal conditions, but it did not increase further by histamine stimulation. The addition of DHA affected the amount of PGE2 in both male and female EVs. In male EVs, there was no difference in the amount of PGE2 between basal and histamine stimulation without DHA, but the amount of PGE2 was increased in general and showed a difference between basal and histamine stimulation in the presence of DHA. This trend was also observed in female EVs (Fig. 5A, white bars vs. grey bars in PGE2).

Figure 5. The amount increase of SPMs in EVs or its fold increase in basal condition and histamine stimulation is more prominent in female EVs compared to male EVs.

(A) Addition of DHA results in enhanced pro-resolution capability in female EVs but not in male EVs at 18 h post-stimulation. (B) The amount of lipid mediators in EVs from male HCjGCs did not change, while EVs from females showed increased RvD1 and LXA4 in response to histamine stimulation. Addition of DHA makes the amount change of lipid mediators in female EVs more prominent, but not in EVs from males. n=3~4, Mean ± SEM. *p<0.05

Since the HCjGCs were expanded ex vivo from deceased donors with very different medical conditions, it would be more intuitive to visualize the quantification in fold change. Without DHA, RvD1 and LXA4 amounts were 3~4 fold higher in female EVs, but no change was observed in male EVs (Fig. 5B, dark grey bar vs. white bar). In the presence of DHA, female EVs showed even higher RvD1 (6~7 fold) and LXA4 (5~6 fold), suggesting enhanced pro-resolving capability in females (Fig. 5B, light grey bar vs dark grey bar). This fold increase was observed only in female EVs but not in males. There was almost no fold change in male EVs except a significant decrease in LXA4.

4. Discussion

Using histamine-mediated allergic inflammation as a disease model, we observed a sex difference in the degree of increase of LXA4 and RvD1 in EVs secreted by HCjGCs. The amount of LXA4 and RvD1 in EVs secreted by female HCjGCs dramatically increased upon histamine stimulation, but there was no notable change in EVs from males. The presence of extra DHA during allergic inflammation resulted in more SPMs in EVs, but again, only in females but not in males. From these results, we conclude that male and female HCjGCs respond differently to allergic inflammation in terms of EV-mediated lipid mediator secretion. Because of a high variation in the amount of lipid mediators that each set of HCjGCs produces — obviously due to naturally existing interpersonal diversity and various pre-existing medical conditions in each individual — the absolute quantity of lipid mediators between males and females at basal condition was not significantly different. However, without exception, female HCjGCs were much more responsive than males in secreting more SPMs via EVs upon induction of allergic inflammation.

Our observation is in line with others. Troisi et al. have shown that the PIM formation in male mouse peritoneal exudates during the first 24 h of peritonitis, especially COX-derived PIMs (sum of PGE2, PGD2, PGF2a, and TXB4) and 5-LOX/FLAP-derived PIMs (sum of LTB4, t-LTB4, and the precursors of these) was higher compared to females [31]. One of the recent clinical trials reported by Rathod et al. showed that females having a higher amount of D-series resolvins (RvDs) and a lower amount of LTB4 in blister exudates 24 h after cantharidin (a substance that incurs blistering, pain, itching, and swelling at the treated area) treatment than males resulted in an accelerated resolution of inflammation in females compared to males [32]. Our data presented in this manuscript aligns with these reports and is consistent with more males suffering from severe types of conjunctivitis than females.

Our result showing the positive correlation between DHA amount and higher SPM quantity in EVs supports the positive effect of ω−3 PUFA supplementation on ocular surface health. Supplementation of omega-3 (ω−3) fatty acids generally promotes ocular surface health [33–35]. A recently published meta-analysis of seventeen randomized clinical trials involving more than 3300 patients revealed that both systemic and local ω−3 PUFA supplementation significantly improved the symptoms and signs of dry eye disease [36,37]. Among pro-resolution mechanisms further activated by DHA, the increased amount of SPMs on the ocular surface - as seen in our result – will be an important factor potentiating the resolution process. Adding DHA also does not support HCjGC production of SPMs in males. It is possible that there is an unknown mechanism that interferes the activity of lipid mediator synthesizing enzymes and that is only activated in males but in females.

Our analysis of lipid mediators was confined to those in EVs secreted by HCjGCs. Serhan and coworkers showed that the microparticles collected from inflammatory exudates in mice contained 17-HDHA and 14-HDHA, which are the precursors of D-series resolvins (RvDs) and maresins (MaRs) [38]. Based on their findings, we speculated that EVs may also have fully synthesized lipid mediators. Therefore, we first tested the existence of lipid mediator synthesizing enzymes in HCjGCs and the HCjGC-secreted EVs. After confirming their existence, we quantified three lipid mediators, PGE2, LXA4, and RvD1, in isolated EVs. Finding these three lipid medaitors in HCjGC-secreted EVs show that HCjGCs contain functional enzymes that synthesize lipid mediators, these synthesized mediators are actively secreted via EVs during health and disease, and the degree of LXA4 and RvD1 increase in EVs during allergic inflammation is sex dependent.

As it was our first time to quantify lipid mediators in EVs, we did not pool samples but analyzed individual samples separately because we did not want to be biased by one outlier and also wanted to know how many human samples are required to capture the significant difference if the difference exists. We limited our measurement to three lipid mediators: RvD1, LXA4, and PGE2. This is because the amount of EVs collected from primary cultures derived from one individual, mostly over 60 years of age at the time of death with various pre-existing moderate to severe medical condition, was limited. For the EV purification method, we used an EV precipitating reagent commercially available. Although there are methods that give better EV purity, we proceeded with one-step precipitation method as there was no difference in EV particle amount and lipid mediator quantification result between EVs prepared using precipitation method and EVs prepared using density gradient followed by ultracentrifugation (DG+UC) (Supplementary Fig. 1) [39]. The nanoparticle tracking analyzer unexpectedly become unavailable at the time of study was an external factor that we decided to proceed with precipitation method for consistency. This also led us to normalize our quantitation to the total protein amount in EVs, which unequivocally captured the difference between experimental group and control group at a similar degree to normalizing to other widely accepted factors, such as EV particle number or total protein amount in cell lysates (Supplementary Fig. 2).

To conclude, HCjGCs secrete synthesized lipid mediators via EVs, of which amount notably changes during histamine-mediated allergic inflammation. This change was only observed in female EVs but not in males. Our data showing the response of HCjGCs against allergic inflammation from a lipid mediator and EV standpoint provide a groundwork for a better understanding of CjGC physiology beyond mucin secretion. We also hope that our result gives insights into explaining the male predisposition (male-to-female ratio 3:1) to severe forms of conjunctivitis, i.e., vernal keratoconjunctivitis and atopic keratoconjunctivitis, observed in the clinic and the better treatment.

Highlights:

• Functional human primary CjGCs (HCjGCs) contain the biosynthetic machinery that can produce proinflammatory mediators (PIMs including leukotrienes and prostaglandins) and specialized pro-resolving mediators (SPMs including lipoxins, resolvins, maresins and protectins).

• Histamine stimulated HCjGCs secrete significantly higher amount of SPMs via EVs after 18 hr.

• Increase in SPM amount was only observed in EVs secreted from female HCjGCs but not in EVs of males, showing a clear sex difference in allergic response.

• Addition of docosahexaenoic acid (DHA) during histamine-mediated allergic inflammation further increased SPM amount in female EVs but did not affect the amount in male EVs.

Data statement

Raw and analyzed data will be provided upon request.