Resolvin D2 elevates cAMP to increase intracellular [Ca²⁺] and stimulate secretion from conjunctival goblet cells

Abstract

Under physiologic conditions, conjunctival goblet cells (CGCs) secrete mucins into the tear film to preserve ocular surface homeostasis. Specialized proresolving mediators (SPMs), like resolvins (Rvs), regulate secretion from CGCs and actively terminate inflammation. The purpose of this study was to determine if RvD2 stimulated mucin secretion and to investigate the cellular signaling components. Goblet cells were cultured from rat conjunctiva. Secretion was measured by an enzyme-linked lectin assay, change in intracellular [Ca²⁺] ([Ca²⁺]_i) using Fura-2, and cellular cAMP levels by ELISA. RvD2 (10⁻¹¹–10⁻⁸ M) stimulated secretion, increased cellular cAMP levels and the [Ca²⁺]_i. RvD2-stimulated increase in [Ca²⁺]_i and secretion was blocked by Ca²⁺ chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis and the PKA inhibitor N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide dihydrochloride but not by the cAMP exchange protein inhibitor α-[2-(3-chlorophenyl)hydrazinylidene]-5-(1,1-dimethylethyl)-b-oxo-3-isoxazolepropanenitrile. Forskolin, 3-isobutyl-1-methylxanthine, and 8-bromo-cAMP (8-Br-cAMP) increased [Ca²⁺]_i. Increasing cAMP with 8-Br-cAMP inhibited the increase in [Ca²⁺]_i stimulated by the cAMP-independent agonist cholinergic agonist carbachol. In conclusion, RvD2 uses both cellular cAMP and [Ca²⁺]_i to stimulate glycoconjugate secretion from CGCs, but the interaction of cAMP and [Ca²⁺]_i is context dependent. Thus RvD2 likely assists in the maintenance of the mucous layer of the tear film to sustain ocular surface homeostasis and has potential as a novel treatment for dry eye disease.—Botten, N., Hodges, R. R., Li, D., Bair, J. A., Shatos, M. A., Utheim, T. P., Serhan, C. N., Dartt, D. A. Resolvin D2 elevates cAMP to increase intracellular [Ca²⁺] and stimulate secretion from conjunctival goblet cells.

The multilayered tear film overlies the ocular surface and is continuously exposed to the external environment. The innermost layer of the tear film consists of high MW glycoconjugates (HMWGs) synthesized and secreted by conjunctival goblet cells (CGCs) (1). One of the HMWG secreted by goblet cells is the gel-forming mucin MUC5AC that is critical for ocular surface homeostasis, clearance of microbes and allergens, and lubrication (1–3). Over- and undersecretion of mucins is associated with ocular surface inflammation (1, 2, 4).

Acute inflammation is actively terminated by specialized proresolving mediators (SPMs) (5) derived from polyunsaturated fatty acids. SPMs include lipoxins, resolvins (Rvs), protectins, and maresins. Production of proinflammatory arachidonic acid derivatives undergo a class switch to promote sequential biosynthesis of endogenous SPMs to terminate inflammation, induce homeostasis, and limit tissue damage and fibrosis (5, 6). Using SPMs to limit inflammatory responses represents a novel approach for treating ocular inflammatory disease.

Rvs were first isolated from inflammatory exudates (7). Both D-series and E-series Rv were identified. RvE1 has been well-studied in preventing multiple ocular surface inflammatory diseases in different experimental models (8–11). RvD1 improves corneal wound healing and has important functions in improving disease pathogenesis in the diabetic retina and uveitis (12–14). In the conjunctiva, RvD1 stimulates HMWG secretion from goblet cells and counterregulates histamine-induced secretion as in allergic conjunctivitis (1, 15). In a mouse model of severe allergic eye disease, topical RvD1 improves the ocular surface symptoms and returns tear and MUC5AC secretion to normal levels (16). A recent study shows that Rvs are present in human tears, with more types present in male than female tears (17) and confirms the important role of SPMs in ocular surface homeostasis and inflammatory disease.

The RvD family consists of 6 members: RvD1–RvD6. RvD1 is the most extensively studied RvD in the eye and especially in the ocular surface (1, 8, 12–15). However, endogenous RvD2 has been identified in human tears (18), serum, plasma (19), lymphoid tissue (20), adipose tissue (21), placenta (22), breast milk (23), lung (24), and peripheral blood of sepsis patients (25). The presence of RvD2 under normal conditions suggests physiologic functions in addition to resolution of inflammatory disease.

RvD2 activates the GPCR DRV2/GPR18 (5, 26). There is evidence that RvD2 works by activating the cAMP pathway (27, 28). In addition, in mouse macrophages, activation of DVR2/GPR18 by RvD2 increases activation of AKT, p38 MAPK, cAMP response element binding, S6, ERK 1/2, signal transducer and activator of transcription (STAT) 1, 3 or 4 with different kinetics peaking in activation between 1 and 30 min (28).

In the present study, we used cultured CGCs to determine the effect of RvD2 on HMWG secretion and the increase in the intracellular [Ca²⁺] ([Ca²⁺]_i). We also interrogated the cellular signaling pathways used. We demonstrated that RvD2 stimulates HMWG secretion in goblet cells by increasing cellular cAMP levels that activate PKA and by increasing [Ca²⁺]_i. Elevating the cellular cAMP level can increase [Ca²⁺]_i, but in the presence of a cholinergic agonist, activated PKA can counterregulate the muscarinic receptors to block the increase in [Ca²⁺]_i.

MATERIALS AND METHODS

Materials

Roswell Park Memorial Institute (RPMI) 1640 cell culture medium, penicillin/streptomycin, and l-glutamine were purchased from Lonza Group (Basel Switzerland). Fetal bovine serum was from Atlanta Biologicals (Flowery Branch, GA, USA). Fura-2- a cetoxymethyl ester and Amplex Red were from Thermo Fisher Scientific (Waltham, MA, USA). Pluronic acid F127, sulfinpyrazone, carbachol (Cch), 3-isobutyl-1-methylxanthine (IBMX), forskolin, and 8-bromo-cAMP (8-Br-cAMP) were from MilliporeSigma (Burlington, MA, USA). Vasoactive intestinal peptide (VIP) was purchased from AnaSpec (Fremont, CA, USA).

RvD2 (7S,16R,17S-trihydroxy-4Z,8E,10Z,12E,14E,19Z-docosahexaenoic acid, RvD2), RvD1 (7S,8R,17S-trihydroxy-4Z,9E,11E,13Z,15E,19Z-docosahexaenoic acid, RvD1), and Maresin-1 (7R,14S-dihydroxy-4Z,8E,10E,12Z,16Z,19Z-docosahexaenoic acid) were from Cayman Chemical Company (Ann Arbor, MI, USA). All compounds were dissolved in ethanol as provided by the manufacturer and stored at −80°C with minimal exposure to light. Annexin-A1 was purchased from Abcam (Cambridge, United Kingdom).

N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide dihydrochloride (H89), 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis (BAPTA/AM), α-[2-(3-chlorophenyl)hydrazinylidene]-5-(1,1-dimethylethyl)-b-oxo-3-isoxazolepropanenitrile (ESI-09), and 2-aminoethyl diphenylborinate (2-APB) were purchased from MilliporeSigma.

Agonists and inhibitors were diluted in RPMI 1640 medium for secretion experiments and Krebs-Ringer bicarbonate buffer with 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid [KRB-HEPES; 119 mM NaCl, 4.8 mM KCl, 1.0 mM CaCl₂, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 25 mM NaHCO₃, 10 mM HEPES, and 5.5 mM glucose (pH 7.45)] for [Ca²⁺]_i measurements. Scansite (https://scansite4.mit.edu/4.0/#home) was used to determine phosphorylation sites of receptors.

Animals

Four- to 8-wk-old male Sprague-Dawley rats (Taconic Biosciences, Rensselaer, NY, USA) were euthanized with CO₂ and decapitated. Conjunctiva was removed from both eyes. Experiments were performed according to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, Bethesda, MD, USA; publication No. 8023, revised 1978) and approved by Schepens Eye Research Institute Animal Care and Use Committee.

Cell culture

Rat CGCs were grown from explants in culture as previously described (29). Cells were grown in 6-well plates containing RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and 100 µg/ml penicillin-streptomycin. First-passage cells were used in all experiments. Goblet cells were seeded in glass-bottom petri dishes or 24-well plates. Cells were regularly stained with the lectin Ulex europaeus agglutinin-1 (UEA-1), cytokeratin 7, and/or MUC5AC to detect goblet cell secretory product, intermediate filaments, and MUC5AC, respectively (29).

HMWG secretion measurements

Passaged goblet cells seeded in 24-well plates were grown to 75% confluence and serum-starved for 2 h in RPMI 1640 medium supplemented with 0.5% bovine serum albumin. Thereafter, cells were incubated with inhibitors for 30 min and stimulated with agonists for 2 h. Secretion was measured using an enzyme-linked lectin assay with the lectin horseradish peroxidase–conjugated UEA-1 that binds HMWG (29). Medium was then collected, transferred to Nunc microplates (Thermo Fisher Scientific) and dried overnight at 60°C. UEA-1 was detected using Amplex Red and quantified using a fluorescence ELISA reader (model Synergy MX; BioTek Instruments, Winooski, VT, USA) with excitation and emission wavelengths of 530 and 590 nm, respectively (8). Total cell protein was determined by Bradford assay, and secretion was normalized to total protein and expressed as fold increase above basal that was set to 1. Standards were made from bovine submaxillary mucin.

[Ca²⁺]_i measurements

Goblet cells in glass-bottom petri dishes were incubated for 1 h at 37°C with KRB-HEPES containing 0.5% bovine serum albumin, 0.5 μM Fura-2/AM, 8 μM pluronic acid F127, and 250 μM sulfinpyrazone. Cells were washed in KRB-HEPES with sulfinpyrazone. [Ca²⁺]_i measurements were performed using a ratio imaging system (In Cyt Im2; Intracellular Imaging, Cincinnati, OH, USA) with excitation wavelengths of 340 and 380 nm and an emission wavelength of 505 nm. Cells were incubated with inhibitors 30 min prior to stimulation with agonists. At least 10 cells were selected for each condition, and experiments were repeated with cells from at least 3 separate animals. Data are presented as the actual [Ca²⁺]_i with time or as the change in peak [Ca²⁺]_i.. Change in peak [Ca²⁺]_i was calculated by subtracting the average of the basal value from the peak [Ca²⁺]_i. When no peak occurred, the maximum value in the time frame in which the agonist was present was used. If the response was negative, the value 15 s after addition of agonist was used.

cAMP measurements

Goblet cells seeded in 24-well plates were grown to 75% confluence. All cells were incubated with IBMX 10⁻³ M for 40 min total. IBMX 10⁻³ M was added to all wells. RvD2 10⁻⁸ M was added for 10–40 min to separate wells containing IBMX. VIP 10⁻⁸ M was added for 2 min to IBMX-containing wells. Cells were lysed in 0.1 M HCl. Total cell cAMP was assayed by direct cAMP ELISA kit following the manufacturer’s instructions (Enzo Life Sciences, Farmingdale, NY, USA). The acetylation protocol was used to increase sensitivity. Total cell protein was determined by Bradford assay, and cellular cAMP was normalized to total protein. cAMP levels are presented in real numbers.

Statistical analysis

Results are presented as fold-increase above basal as means ± sem. Statistical analysis was performed using Student’s t test, and a value of P < 0.05 was considered statistically significant.

RESULTS

RvD2 stimulates HMWG that is dependent on cellular cAMP and [Ca²⁺]_i

When cultured rat CGCs were incubated with increasing concentrations of RvD2, RvD2 increased HMWG secretion (Fig. 1A). RvD2 10⁻¹¹ M increased secretion above basal by 1.9 ± 0.3-fold (P = 0.02); RvD2 10⁻¹⁰ M by 1.9 ± 0.1-fold (P = 0.0007); and RvD2 10⁻⁹ M by 2.0 ± 0.3-fold (P = 0.03). A maximum effect was caused by RvD2 at 10⁻⁸ M that increased secretion 2.4 ± 0.4-fold above basal (P = 0.008). As a control, the cholinergic agonist Cch increased goblet cell secretion 1.6 ± 0.06-fold (P = 0.00004) (unpublished results).

Figure 1.

RvD2 stimulates HMWG secretion from rat CGCs using PKA and [Ca²⁺]_i. Goblet cells were serum-starved for 2 h before stimulation with increasing concentrations of RvD2 (A) or preincubated for 30 min with H89 10⁻⁵ M (B) or BAPTA/AM (C). The amount of HMWG secretion is shown as fold increase above basal. Data are means ± sem from 4 individual experiments. *Significant difference from basal; ^#significant difference from RvD2, VIP, or Cch alone.

To compare the effects of RvD2 with other agonists known to stimulate HMWG secretion, CGCs were incubated with the Cch, VIP, the SPM RvD1 as well as RvD2. As shown in Table 1, all agonists significantly increased secretion above basal to the same extent as there was no significant difference between stimulated values.

TABLE 1.

RvD1 and RvD2 significantly increase glycoconjugate secretion

Agonist	Glycoconjugate secretion
Basal	1.0 ± 0.0
Cch 10−4 M	1.9 ± 0.3a
VIP 10−8 M	1.3 ± 0.1a
RvD1 10−8 M	1.8 ± 0.1a
RvD2 10−8 M	2.5 ± 0.4a

Data are means ± sem from 3 independent experiments. ^aIndicates significant difference from basal.

CGCs were incubated with the PKA inhibitor H89 10⁻⁵ M for 30 min prior to RvD2 (10⁻⁸ M) stimulation to determine if RvD2 activates PKA to stimulate secretion (Fig. 1B). H89 significantly blocked RvD2-induced HMWG secretion by 92.8 ± 17.6% (P = 0.04). As previously described, VIP stimulates glycoconjugate secretion by activating PKA (30). Thus, VIP was used as a positive control. VIP (10⁻⁸ M) alone significantly increased glycoconjugate secretion 2.2 ± 0.3-fold above basal (P = 0.007). The VIP-stimulated increase in secretion was significantly blocked 76.6 ± 7.8% (P = 0.03) by H89 (10⁻⁵ M).

Cultured goblet cells were incubated with the cell-permeable cAMP analog RpcAMPs 10⁻⁶ M for 30 min prior to RvD2 (10⁻⁸ M) stimulation (Supplemental Fig. S1A). RpcAMPs acts as a competitive antagonist of cAMP-induced activation of PKA. RvD2 (10⁻⁸ M) alone significantly increased HMWG secretion above basal by 1.9 ± 0.2-fold above basal (P = 0.003). RpcAMPs significantly blocked RvD2-induced secretion by 65.2 ± 9.5%. VIP 10⁻⁸ M alone significantly increased HMWG secretion 3.1 ± 0.8-fold above basal (P = 0.04). VIP-stimulated secretion was decreased by RpcAMPs (10⁻⁶ M) by 62.1 ± 14.8%, though the level did not reach significance (P = 0.4). These results show that RvD2 activates PKA to secrete mucins from goblet cells.

To investigate if RvD2 uses cellular Ca²⁺ to increase [Ca²⁺]_i, goblet cells were incubated with the intracellular calcium chelator BAPTA/AM for 30 min prior to RvD2 (10⁻⁸ M) stimulation (Supplemental Fig. S1B). We first measured the [Ca²⁺]_i over time to determine the concentration of BAPTA/AM that blocks the RvD2-stimulated increase in [Ca²⁺]_i. RvD2 at 10⁻⁸ M alone significantly increased [Ca²⁺]_i by 424.8 ± 20.4 nM above basal (P = 0.00003). BAPTA at 10⁻⁴ M significantly blocked the RvD2-induced increase in [Ca²⁺]_i by 93.0 ± 3.2% (P = 0.00008). Cch is known to increase [Ca²⁺]_i and stimulate secretion dependent on Ca²⁺. Hence, Cch 10⁻⁴ M was used as a positive control for [Ca²⁺]_i and secretion measurements with BAPTA/AM. Cch alone significantly increased [Ca²⁺]_i by 379.3 ± 88.6 nM above basal (P = 0.01). BAPTA at 10⁻⁴ M significantly blocked the Cch-induced response by 91.3 ± 2.7% (P = 0.02). Therefore, BAPTA/AM at 10⁻⁴ M was used for the secretion experiments.

BAPTA/AM (10⁻⁴ M) pretreatment for 30 min significantly blocked 10⁻⁸ M RvD2-induced HMWG secretion by 76.4 ± 11.2% (Fig. 1C, P = 0.04). The positive control Cch 10⁻⁴ alone significantly increased glycoconjugate secretion by 1.6 ± 0.06-fold (P = 0.00005) above basal. Cch-induced secretion was completely and significantly blocked (P = 0.001) by BAPTA. These data indicate that RvD2 increased [Ca²⁺]_i, which led to glycoconjugate secretion. As inhibition of either PKA activation or an increase in [Ca²⁺]_i blocks RvD2-stimulated secretion, these 2 pathways must interact at the level of the release of Ca²⁺ from intracellular stores so that both active PKA or its phosphorylated target and inositol trisphosphate (IP₃)/IP₃ receptor (IP₃R) binding converge to release [Ca²⁺]_i.

RvD2 increases cellular cAMP levels in CGCs

We investigated whether RvD2 increases cellular cAMP in CGCs by measuring cellular cAMP after stimulation for 10–40 min (Fig. 2). All samples contained IBMX 10⁻³ M to prevent breakdown of formed cAMP. IBMX alone served as the basal level and produced 1.6 ± 0.6 pg cAMP/µg total protein after 40 min. RvD2 at 10⁻⁸ M increased cellular cAMP to 5.3 ± 1.4 pg/µg protein after 40 min, a value significantly higher than basal IBMX alone (P = 0.04). RvD2 at the other time points did not significantly increase cAMP. The positive control VIP 10⁻⁸ M incubated for 2 min increased cAMP. Thus, RvD2 increases cellular cAMP in goblet cells but with a long time dependency.

Figure 2.

RvD2 increases cellular cAMP in rat CGCs. Amount of cellular cAMP measured as picogram per microgram total protein. VIP 10⁻⁸ M or RvD2 10⁻⁸ M was added to wells containing IBMX for 0–40 min, and cAMP was measured. Data are means ± sem from 8 individual animals. *Significant difference from IBMX alone.

RvD2 activates PKA to increase [Ca²⁺]_i in CGCs

To date, there are 2 different mechanisms by which cAMP can cause its effects. One is by activating PKA and the second is by activating the cAMP exchange protein (Epac). Activation of PKA causes mucin secretion (Fig. 1B), so to determine whether RvD2 activates PKA to increase [Ca²⁺]_i, CGCs were incubated with the PKA inhibitor H89 at (10⁻⁷–10⁻⁵ M) for 30 min before addition of RvD2 (10⁻⁸ M). RvD2 alone increased peak [Ca²⁺]_i by 290.5 ± 92.9 nM (P = 0.02) above basal (Fig. 3A, B). The positive control VIP increased [Ca²⁺]_i by 392.6 ± 68.9 nM above basal (P=0.001). Treatment with H89 blocked the RvD2-stimulated increase in [Ca²⁺]_i that was significantly decreased by 89.6 ± 1.8% (P = 0.03) at 10⁻⁵ M. As expected, H89 at 10⁻⁷ M and 10⁻⁶ M blocked the VIP 10⁻⁸ M–stimulated increase in [Ca²⁺]_i by 64.9 ± 21.8% (P = 0.01) and 48.5 ± 19.9% (P = 0.03), respectively (Fig 3C). We concluded that RvD2 activates PKA to increase [Ca²⁺]_i in rat CGCs.

Figure 3.

A) RvD2 activates PKA to increase [Ca²⁺]_i in rat CGCs. [Ca²⁺]_i over time for RvD2 10⁻⁸ M in the presence of H89 at different concentrations. B) [Ca²⁺]_i over time for the VIP control 10⁻⁸ M in the presence of H89 at different concentrations. C) Change in peak [Ca²⁺]_i for all conditions. Data are means ± sem from 4 individual animals. *Significant difference from basal; ^#significant difference from RvD2 or VIP alone.

RvD2 does not activate Epac to increase [Ca²⁺]_i

To determine whether RvD2 activates Epac, CGCs were incubated with the Epac-inhibitor ESI-09 for 30 min before stimulation with RvD2 10⁻⁸ M or VIP 10⁻⁸ M. [Ca²⁺]_i was measured over time. RvD2 increased peak [Ca²⁺]_i by 295.4 ± 52.7 nM (P = 0.001) (Fig. 4A, B). Whether VIP activates Epac has not previously been investigated in CGCs. VIP 10⁻⁸ M increased [Ca²⁺]_i by 207.8 ± 29.6 nM (P = 0.0004) above basal (Fig. 4B, C). ESI-09 at 10⁻⁷, 10⁻⁶, or 10⁻⁵ M concentrations did not alter the RvD2-stimulated increase in [Ca²⁺]_i.

Figure 4.

RvD2 does not use Epac to increase [Ca²⁺]_i. A, B) Rat CGCs were preincubated with ESI-09 for 30 min before stimulation with RvD2 10⁻⁸ M or VIP 10⁻⁸ M. [Ca²⁺]_i over time for RvD2 10⁻⁸ M (A) and VIP 10⁻⁸ M (B) in the presence of ESI-09 is shown. C) Change in peak [Ca²⁺]_i for all conditions. Data are means ± sem from 4 individual animals. *Significant difference from basal; ^#significant difference from RvD2 or VIP alone.

The increase in [Ca²⁺]_i was not altered by ESI-09 10⁻⁷ M (P = 0.2) (Fig 4B, C). ESI-09 10⁻⁶ and 10⁻⁵ M significantly blocked the VIP 10⁻⁸ M–stimulated [Ca²⁺]_i increase by 49.1 ± 20.1% (P = 0.04) and 49.3 ± 13.5% (P = 0.03), respectively. Thus, VIP but not RvD2, activates Epac to increase [Ca²⁺]_i in cultured goblet cells.

Raising cellular cAMP levels increases [Ca²⁺]_i in CGCs

To investigate whether raising cellular levels of cAMP independent of receptor activation increases [Ca²⁺]_i, we stimulated CGCs with compounds that are known to increase cellular cAMP levels and measured [Ca²⁺]_i over time. Forskolin increases cAMP by activating adenylyl cyclase (31, 32). Forskolin at 10⁻⁶, 10⁻⁵, and 10⁻⁴ M significantly increased peak [Ca²⁺]_i by 349.6 ± 40.7 nM (P = 0.001), 402.3 ± 25.9 nM (P = 0.0001), and 238.7 ± 34.9 nM (P = 0.0001), respectively (Fig. 5A, B).

Figure 5.

Increasing cellular cAMP elevates [Ca²⁺]_i in rat CGCs. A, B) [Ca²⁺]_i over time (A) and change in peak [Ca²⁺]_i after forskolin stimulation (B). C, D) [Ca²⁺]_i over time (C) and change in peak [Ca²⁺]_i (D) after 8-Br-cAMP stimulation. E, F) [Ca²⁺]_i over time (E) and change in peak [Ca²⁺]_i (F) after IBMX stimulation. Data are means ± sem. Data are from 3 (A, B, E, F) or 5 (C, D) individual animals. *Significant difference from basal.

The membrane-permeable cAMP analog 8-Br-cAMP at 10⁻⁶, 10⁻⁵, and 10⁻⁴ M significantly increased peak [Ca²⁺]_i by 248.0 ± 41.2 nM (P = 0.0003), 270.3 ± 57.1 nM (P = 0.001), and 295.7 ± 799 nM (P = 0.006), respectively (Fig. 5C, D).

IBMX inhibits the enzyme phosphodiesterase that breaks down cAMP (32, 33). IBMX at 10⁻⁶–10⁻³ M significantly increased peak [Ca²⁺]_i by 300.3 ± 67.0 nM (P = 0.01), 311.6 ± 36.2 nM (P = 0.001), 171.1 ± 15.1 nM (P = 0.0003), and 188.4 ± 34.2 nM (P = 0.005), respectively (Fig. 5E, F). Thus, increasing cellular cAMP levels by 3 different receptor-independent mechanisms in CGCs increases [Ca²⁺]_i.

RvD2 and VIP, but not RvD1, maresin-1, nor annexin-1, increase cellular cAMP and activate PKA to increase [Ca²⁺]_i in CGCs

To determine if other SPMs activate cAMP, CGCs were incubated with H89 10⁻⁵ M before select SPMs were added and [Ca²⁺]_i was measured over time. Annexin-1 at 10⁻⁹ M increased peak [Ca²⁺]_i by 484.3 ± 151.9 nM (P = 0.01); maresin-1 (10⁻⁸ M) by 382.4 ± 97.3 nM (P = 0.007); and RvD1 (10⁻⁸ M) by 489.9 ± 170.5 nM (P = 0.004) above basal (Supplemental Fig. S2A). H89 did not significantly alter any of these responses. In contrast, RvD2 at 10⁻⁸ M increased peak [Ca²⁺]_i by 510.0 ± 158.2 nM (P = 0.002) that H89 blocked by 73.8 ± 7.3% (P = 0.03). As a positive control, VIP at 10⁻⁸ M increased peak [Ca²⁺]_i by 632.5 ± 144.8 nM (P = 0.002). H89 blocked this response by 44.7 ± 9.5% (P = 0.05).

We measured cellular cAMP in CGCs after a 40-min stimulation with the same SPMs as in Supplemental Fig. S2A. IBMX 10⁻³ M was added to all samples to inhibit the breakdown of cAMP. In the presence of IBMX alone, cellular cAMP levels were 0.8 ± 0.2 pg/µg total protein. Compared with basal IBMX values, neither annexin-1 at 10⁻⁹ M, maresin-1 at 10⁻⁸ M, nor RvD1 at 10⁻⁸ M increased cellular cAMP levels (Supplemental Fig. S2B). As a positive control, RvD2 at 10⁻⁸ M increased cAMP levels to 1.8 ± 0.3 pg/µg, which is significantly higher than basal IBMX (P = 0.02). Similarly, VIP at 10⁻⁸ M increased cAMP levels to 1.9 ± 0.4 pg/µg, which is significantly higher than basal IBMX (P = 0.03). These results indicate that RvD2 and VIP increase cellular cAMP to activate PKA in goblet cells, whereas annexin-1, maresin-1, and RvD1 do not.

cAMP and Ca²⁺ interact and use similar intracellular calcium stores

The IP₃R antagonist, 2-APB, was used to investigate if 8-Br-cAMP used an intracellular source of Ca²⁺. Cells were incubated for 30 min with 2-APB prior to stimulation with agonist (Supplemental Fig. S3A). The 8-Br-cAMP–induced increase in [Ca²⁺]_i was significantly blocked by 50.8 ± 17.0% (P = 0.05) of 2-APB. Cch was the positive control as Cch does not activate PKA, and the increase in [Ca²⁺]_i is independent of cAMP. In addition, Cch uses both intracellular and extracellular Ca²⁺ stores to increase [Ca²⁺]_i (34). 2-APB blocked the Cch 10⁻⁴ M–stimulated increase in [Ca²⁺]_i by 65.4 ± 7.1% (P = 0.004).

Thapsigargin inhibits the Ca²⁺-ATPase activity in the endoplasmic reticulum, depleting the [Ca²⁺]_i stores that then increases entry of extracellular Ca²⁺ and thus increases [Ca²⁺]_i independently of IP₃ (35, 36). Thapsigargin 10⁻⁶ M significantly increased [Ca²⁺]_i above basal by 787.1 ± 51.7 nM (P = 0.0000003) (Supplemental Fig. S3B). 8-Br-cAMP 10⁻⁴ M alone significantly increased [Ca²⁺]_i by 476.7 ± 95.6 nM above basal (P = 0.001). In the next experiments, cells were treated with thapsigargin 10⁻⁶ M for 15 min before stimulation with 8-Br-cAMP. The 8-Br-cAMP–stimulated increase in [Ca²⁺]_i was completely blocked (P = 0.002). Thus, cAMP mobilizes Ca²⁺ from the endoplasmic reticulum to increase [Ca²⁺]_i in goblet cells.

To determine if 8-Br-cAMP used extracellular Ca²⁺, we removed CaCl₂ from the incubation medium. CGCs stimulated with 8-Br-cAMP 10⁻⁴ M significantly increased [Ca²⁺]_i by 374.7 ± 85.7 nM (P = 0.001) (Supplemental Fig. S3C). Removing extracellular Ca²⁺ significantly blocked the 8-Br-cAMP–stimulated response by 66.8 ± 5.7% (P = 0.01). As we previously showed, the Cch 10⁻⁴ M–induced increase in [Ca²⁺]_i was blocked after removal of extracellular Ca²⁺ (34). Cch was used as the positive control. Herein, Cch 10⁻⁴ M alone increased [Ca²⁺]_i significantly by 356.8 ± 67.4 nM (P = 0.0004). This response was blocked by 84.0 ± 41% (P = 0.001) after removal of extracellular Ca²⁺. Thus, 8-Br-cAMP, similarly to Cch, releases Ca²⁺ from intracellular stores that stimulate the influx of extracellular Ca²⁺ to increase [Ca²⁺]_i.

Next, we determined if elevation of cAMP levels without activation of a receptor stimulates HMWG mucin secretion. Conjunctival goblet cells were incubated with the cell permanent cAMP analog 8-Br-cAMP and HMWG secretion was measured. 8-Br-cAMP increased secretion significantly to 1.8 ± 0.2-fold above basal. In cells from the same animal, RvD2 increased secretion 1.6 ± 0.1-fold above basal (Fig. 6). 8-Br-cAMP–stimulated secretion was blocked by chelation of [Ca²⁺]_i with a pretreatment of the Ca²⁺ chelator BAPTA and was 0.5 ± 0.2-fold above basal (Fig. 6). Similarly, RvD2-stimulated secretion was also blocked by chelation of Ca²⁺ and was 0.7 ± 0.1-fold above basal. These values were significantly reduced from either 8-Br-cAMP and RvD2 alone.

Figure 6.

Elevating cAMP stimulates HMWG secretion from rat CGCs using [Ca²⁺]_i. Goblet cells were serum-starved for 2 h before stimulation before a 30 min preincubation with vehicle alone or BAPTA/AM 10⁻⁴ M. Cells were then stimulated with 8-Br-cAMP (10⁻⁴ M) or as a control RvD2 10⁻⁸ M. Data are means ± sem from 4 individual animals. *Significant difference from basal; ^#significant difference from 8-Br-cAMP or RvD2 alone.

Whether cAMP and Cch interact to alter each other’s [Ca²⁺]_i response was investigated. Goblet cells were stimulated with Cch 10⁻⁴ M and 8-Br-cAMP 10⁻⁴ M alone or together. Cch 10⁻⁴ M alone significantly increased [Ca²⁺]_i by 425.7 ± 103 nM above basal (P = 0.003) (Fig. 7A). 8-Br-cAMP 10⁻⁴ M alone significantly increased [Ca²⁺]_i by 370.9 ± 120.6 nM above basal (P = 0.02). When added at the same time, Cch 10⁻⁴ M and 8-Br-cAMP 10⁻⁴ M increased [Ca²⁺]_i by 239.3 ± 50.7 nM above basal (P = 0.002). The theoretical additive response, calculated by adding the responses from each Cch and 8-Br-cAMP experiment, was 796.6 ± 192.1 nM above basal (P = 0.002). The [Ca²⁺]_i response obtained when Cch and 8-Br-cAMP were added at the same time (P = 0.02) was significantly lower than the experimental value (Fig. 7B). Thus, 8-Br-cAMP and Cch use the same Ca²⁺ stores to increase [Ca²⁺]_i and potentially inhibit each other’s response.

Figure 7.

cAMP and Ca²⁺ interact and use the same [Ca²⁺]_i stores. Rat CGCs were stimulated with Cch 10⁻⁴ M or 8-Br-cAMP 10⁻⁴ M alone or at the same time. [Ca²⁺]_i over time (A) and change in peak [Ca²⁺]_i (B). Theoretical additivity was calculated from values obtained by adding the response from Cch and 8-Br-cAMP by themselves subtracting the basal value (B). Data are means ± sem from 5 individual animals. *Significant difference from 0; ^#signficant difference from theoretical value.

cAMP counterregulates cholinergic receptors

We next investigated whether an increase in cAMP activates PKA to counterregulate the Cch-stimulated [Ca²⁺]_i response. Goblet CGCs were stimulated with Cch alone, Cch after preincubation with H89 for 45 min, Cch after addition of RvD2 10⁻⁸ for 30 min, or Cch after pretreatment with H89 for 15 min followed by addition of RvD2 for 30 min. Cch 10⁻⁴ M alone significantly increased [Ca²⁺]_i by 428.9 ± 88.8 nM above basal (P = 0.008) (Fig. 8). Preincubating cells with RvD2 10⁻⁸ M significantly blocked the Cch 10⁻⁴–stimulated increase in [Ca²⁺]_i by 91.9 ± 7.5% (P = 0.01). Preincubating cells with H89 10⁻⁵ M did not alter the [Ca²⁺]_i compared with Cch 10⁻⁴ M alone (P = 0.7). When cells were incubated with H89 10⁻⁵ M followed by RvD2 10⁻⁸ M, the Cch 10⁻⁴ M–stimulated increase was 231.8 ± 33.3 nM above basal and not significantly different from Cch 10⁻⁴ M alone (P = 0.1). Thus, an RvD2-stimulated increase in cellular cAMP activates PKA to counterregulate the Cch-stimulated increase in [Ca²⁺]_i.

Figure 8.

RvD2 activates PKA to inhibit cholinergic agonist-stimulated increase in [Ca²⁺]_i. Rat CGCs were stimulated with Cch 10⁻⁴ M alone, with RvD2 10⁻⁸ M for 30 min before stimulation with Cch, with H89 10⁻⁵ M for 45 min before stimulation with Cch, and with H89 10⁻⁵ M for 15 min plus RvD2 10⁻⁸ M for 30 min before stimulation with Cch 10⁻⁴ M. Data are means ± sem from 3 individual animals. *Significant difference from 0; ^#significant difference from Cch alone.

DISCUSSION

In this study, we demonstrate that RvD2 added to cultured rat CGCs stimulates mucin secretion (1, 8, 15, 37). RvD2 uses its receptor DRV2/GPR18 to activate 2 pathways (Fig. 9). First, RvD2 increases the activity of AC that catalyzes the conversion of ATP to cAMP. cAMP in turn activates PKA but not Epac. As demonstrated in this study, RvD2 increases cAMP to activate PKA that, in turn, increases [Ca²⁺]_i and induces mucin secretion. The second pathway activated is the PLC pathway. PLC is hydrolyzed into IP₃ and diacylglycerol. IP₃ binds to its receptors on the endoplasmic reticulum to release Ca²⁺ from intracellular stores. The increase in [Ca²⁺]_i leads to mucin secretion (Fig. 9). cAMP/PKA and IP₃ interact to activate the IP₃R and release Ca²⁺ from the same thapsigargin-dependent cellular stores. The effect of RvD2 on cAMP and Ca²⁺ in goblet cells is the same as activating the β2-adrenoceptor in human embryonic kidney cells where activating the β2-adrenoceptor mediates G_s-protein–stimulated activation of adenylate cyclase to produce cAMP. This activates PLC to produce IP₃ and release Ca²⁺ from intracellular stores (38, 39).

Figure 9.

Schematic diagram of signaling pathways activated by RvD2. RvD2 uses its receptor DVR2/GPR18 to stimulate adenylate cyclase (AC) and PLC. Activation of AC leads to increased cellular cAMP levels and stimulation of PKA. This leads to release of Ca²⁺ from the endoplasmic reticulum. PLC increases the amount of IP₃. IP₃ binds to IP₃R on the endoplasmic reticulum to release [Ca²⁺]_i. Emptying of the Ca²⁺ store leads to influx of Ca²⁺ to refill the stores. Increase in [Ca²⁺]_i leads to mucin secretion.

RvD2 is different from the other SPMs tested in goblet cells as it is the only SPM tested that increased cAMP and activated PKA. RvD2 increases both cellular cAMP levels and [Ca²⁺]_i. Mucin secretion is directly related to an increase in intracellular cAMP levels and [Ca²⁺]_i as use of the PKA inhibitor H89 or chelation of Ca²⁺ with BAPTA/AM each blocked RvD2-stimulated secretion. Thus, the cAMP- and Ca²⁺-dependent pathways activated either by RvD2 or by increasing cellular cAMP levels independent of the receptor interact at a common shared step. This step is probably at the level of the IP₃R on the thapsigargin-sensitive [Ca²⁺]_i stores. Of the SPMs tested, only RvD2-stimulated secretion was blocked by H89. In contrast, chelation of Ca²⁺ with BAPTA/AM blocks RvD2-stimulated secretion and is similar to mucin secretion stimulated with RvD1, RvE1, lipoxin A₄, and cholinergic agonists (9, 15, 37). As multiple agonists use [Ca²⁺]_i to stimulate secretion, it is likely that an increase in [Ca²⁺]_i is a common mechanism to stimulate mucin secretion, but elevation in cellular cAMP levels is not required.

There is considerable crosstalk between the cAMP and Ca²⁺ pathways. Activated PKA phosphorylates target proteins with a variety of cellular functions [e.g., the IP₃R directly to release [Ca²⁺]_i (30, 40)]. cAMP can also activate proteins such as IP₃R, independent of PKA (41) and PLC, through activation of the small GTPase Rap2B (38). In support of a direct role of cAMP in mucin secretion in goblet cells, we found that elevating cellular cAMP using forskolin, IBMX, or 8-Br-cAMP increased [Ca²⁺]_i. The 8-Br-cAMP–induced increase in [Ca²⁺]_i was inhibited by an IP₃R inhibitor, suggesting that the cAMP-induced increase in [Ca²⁺]_i is a result of a PKA and/or cAMP interaction with the IP₃R to release Ca²⁺ from the endoplasmic reticulum. The increased [Ca²⁺]_i would then stimulate secretion.

cAMP can activate PKA and/or Epac (40), and in goblet cells, RvD2 activates PKA to increase [Ca²⁺]_i and stimulate mucin secretion. Unlike VIP, RvD2 does not activate Epac to increase [Ca²⁺]_i. RvD2 is the only known SPM to date that works through the cAMP pathway to increase [Ca²⁺]_i. That the increase in cellular cAMP levels causes a rise in [Ca²⁺]_i in CGCs was demonstrated with cAMP analog 8-Br-cAMP.

RvD2 both activates PKA and increases [Ca²⁺]_i. In this setting, PKA and Ca²⁺ interact cooperatively to stimulate secretion. In contrast, when goblet cells are stimulated by 8-Br-cAMP that is independent of the RvD2 receptor activation and a cholinergic agonist that does not use cAMP, cAMP inhibits the increase in [Ca²⁺]_i. We had expected that 8-Br-cAMP would potentiate the Cch-stimulated increase in [Ca²⁺]_i in CGCs as occurred in human embryonic kidney cells using 8-Br-cAMP at a high concentration (41). This discrepancy in interaction could be due to the difference in concentration of 8-Br-cAMP used [8-Br-cAMP was used at 3 × 10⁻² M by Tovey et al., (41), whereas this study used 8-Br-cAMP at 10⁻⁴ M]. We hypothesize that 8-Br-cAMP and Cch use the same [Ca²⁺]_i stores, and depletion of these stores by one agonist prevents the increase by the second agonist. Alternatively, PKA may phosphorylate sites to counterregulate one or more of the muscarinic receptor subtypes. Indeed, M1 and M3 muscarinic subtypes contain phosphorylation sites for PKA as demonstrated using Scansite. M1 and M3 muscarinic receptors are present in goblet cells and stimulation of each subtype causes mucin secretion (42, 43).

We found earlier that VIP increases cellular cAMP levels to release Ca²⁺ and stimulate secretion (30). In the present study, we show that VIP, unlike RvD2, activates Epac, a PKA-independent signal transduction pathway stimulated by cAMP (44, 45). Epacs can release [Ca²⁺]_i by interacting with IP₃R, activating Ca²⁺ calmodulin-dependent protein kinase II, or interacting with the ryanodine receptor (39, 46, 47). Increase in [Ca²⁺]_i stimulated by Epac could lead to mucin secretion from goblet cells. We did not, however, investigate whether the Epac inhibitor ESI-09 blocked VIP-stimulated secretion, only the increase in [Ca²⁺]_i.

We conclude that in CGCs, RvD2 activates cAMP/PKA-dependent as well as [Ca²⁺]_i-dependent signaling pathways to up-regulate mucin secretion under physiologic conditions. In addition, RvD2 activation of PKA releases [Ca²⁺]_i from [Ca²⁺]_i stores. The 2 messengers cAMP and Ca²⁺ can activate separate pathways to increase [Ca²⁺]_i that converges at the level of release of [Ca²⁺]_i that, in turn, stimulates mucin secretion. These 2 pathways also show complex interaction in goblet cells that is inhibitory depending on the agonists used. Thus, activation of the cAMP and Ca²⁺ pathways by RvD2 in rat CGCs results in secretion of mucins into the tear film to help maintain homeostasis and prevent ocular inflammatory diseases.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

[Ca²⁺]_i

intracellular [Ca²⁺]

2-APB

2-aminoethyl diphenylborinate

8-Br-cAMP

8-bromo-cAMP

BAPTA/AM

1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis

Cch

carbachol

CGC

conjunctival goblet cell

Epac

cAMP exchange protein

ESI-09

α-[2-(3-chlorophenyl)hydrazinylidene]-5-(1,1-dimethylethyl)-b-oxo-3-isoxazolepropanenitrile

H89

N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide dihydrochloride

HMWG

high MW glycoconjugate

IBMX

3-isobutyl-1-methylxanthine

IP3

inositol trisphosphate

IP₃R

IP₃ receptor

KRB-HEPES

Krebs-Ringer bicarbonate buffer with 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

RPMI

Roswell Park Memorial Institute

Rv

resolvin

SPM

specialized proresolving mediator

STAT

signal transducer and activator of transcription

UEA-1

Ulex europaeus agglutinin-1

VIP

vasoactive intestinal peptide

AUTHOR CONTRIBUTIONS

N. Botten and R. R. Hodges performed experiments, analyzed data, and wrote the manuscript; D. Li, J. A. Bair, and M. A. Shatos performed experiments and analyzed data; and T. P. Utheim, C. N. Serhan, and D. A. Dartt analyzed data and wrote the manuscript.