Differential Regulation of ATP- and UTP-Evoked Prostaglandin E₂ and IL-6 Production from Human Airway Epithelial Cells

The airway epithelial cells (AECs) lining the conducting airways of the lung form the first line of defense against a variety of inhaled pathogens, allergens, and environmental irritants. AECs not only provide a physical barrier, but also actively orchestrate immune responses to inhaled substances through the production of a wide array of secreted factors that include alarmins, chemokines, cytokines, and eicosanoids (1). Among these alarmins, extracellular ATP, a damage-associated molecular pattern, is of growing interest due to its ability to drive airway inflammation (2–4). Elevated ATP in the bronchoalveolar lavage (BAL) fluid stimulates cytokine release, activates the inflammasome, and recruits immune cells (2–5). Moreover, elevated ATP in the BAL fluid is a key characteristic of many lung disorders, including asthma, acute respiratory distress syndrome (ARDS), and chronic obstructive pulmonary disease (COPD) (2, 6–9). In contrast, ATP signaling also plays important roles in physiological processes in the healthy lung related to wound healing, rejection of tumors, killing of bacterial pathogens, and mucociliary clearance (MCC) (3, 10–14). This latter response is also evoked by the uridine nucleotide, UTP, which is released in the lung airways and can stimulate numerous physiologically important airway functions, including MCC, ion transport, and mucin secretion (15). How these nucleotides elicit such broad effects on airway function is not well understood.

Two of the most prominent mediators produced by AECs are PGE₂ and IL-6. PGE₂ is synthesized through an enzymatic cascade tied to the activities of phospholipase A₂, cyclooxygenase 1 (COX-1) and COX-2 enzymes, and a terminal PGE synthase (16). PGE₂ stimulates dilation of the lung airways, thereby protecting the lung against severe bronchoconstriction that is a hallmark of diseases such as asthma (17, 18). PGE₂ inhibits several inflammatory processes, including the release of histamine and cysteinyl leukotrienes from mast cells (17), T cell migration (19), ILC2 function (20), and the production of proinflammatory TNF-α and IL-12β by dendritic cells (21), while also enhancing the synthesis of the anti-inflammatory cytokine IL-10 (21) and promoting wound healing (22). Further, growing evidence suggests that PGE₂ can drive reverse migration and removal of neutrophils from sites of inflammation to dissipate inflammation in vivo (23). Failure of resolution of inflammation can lead to inflammatory diseases such as COPD and ARDS. Thus, within the lung, PGE₂ exerts predominantly bronchoprotective and anti-inflammatory effects (24).

By contrast, IL-6 is a potent inflammatory cytokine that elicits a wide range of inflammatory effects, including T cell expansion, pulmonary neutrophil infiltration, airway mucus secretion, lung fibrosis, and the hyperplasia and hypertrophy of airway smooth muscle cells (25–28). Further, IL-6 is linked to numerous airway diseases from asthma to COPD (25–28), and recent studies in patients with severe coronavirus disease 2019 (COVID-19) have revealed massive increases in IL-6 in the lung airways, a feature thought to contribute to microvascular thrombosis in the lung and multiple organ dysfunction in these patients (29). Conversely, blockade of IL-6 signaling in mouse models of asthma protects against airway inflammation (26, 30), and early reports indicate that targeting IL-6 may be a viable therapy to decrease mortality in patients with severe COVID-19 (31).

The mechanisms by which ATP and UTP signaling regulates the synthesis of these distinct mediators (PGE₂ versus IL-6) in AECs are largely unclear. There is compelling evidence, however, that cytosolic Ca²⁺ elevations mediated by the opening of store-operated Ca²⁺ release-activated Ca²⁺ (CRAC) channels formed by the ORAI1 protein regulates both factors (32, 33). As a multifunctional second messenger, Ca²⁺ activates distinct genetic programs to regulate many cellular functions, including gene transcription, cytokine production, and activation of numerous enzymes (34). Although the necessity for Ca²⁺ elevations is suggestive of a unified mechanism driving the synthesis of both mediators, there are many unanswered questions. Are the upstream receptors driving ATP-mediated synthesis of PGE₂ and IL-6 identical? What are the signaling mechanisms downstream of receptor activation in their synthesis? And what is the precise role of CRAC channels in the synthesis of these mediators? In this study, we addressed these and other questions in primary human AECs. We find that ATP induction of PGE₂ and IL-6 requires different ATP concentrations and involves distinct upstream receptors (P2Y versus P2X receptors). In contrast to ATP, extracellular UTP acting via P2Y₂ receptors exclusively stimulated PGE₂ synthesis without concurrent induction of IL-6. Synthesis of PGE₂ involved activation of cytosolic phospholipase A₂ (cPLA₂) by local Ca²⁺ signals near CRAC channels and activity of COX-2, whereas IL-6 induction required activation of calcineurin/NFAT signaling by CRAC channels. These findings reveal the key signaling checkpoints that mediate the synthesis of distinct immunomodulatory products from the same ligand but working through distinct upstream receptors and involving Ca²⁺ signals through CRAC channels as the final common pathway.

Normal human bronchial epithelial (NHBE) cells were purchased from Lonza (catalog number CC-2540) and grown in bronchial epithelial growth medium (BEGM; CC-3170). All experiments that used media for the stimulation phase used Lonza bronchial epithelial basal medium (BEBM) supplemented with Ca²⁺ to bring the total concentration up to 2 mM. Cells were grown in 37°C and 5% CO₂. Ringer solutions were as follows: 2 mM Ca²⁺ Ringer solution: 155 mM NaCl, 4.5 mM KCl, 10 mM d-glucose, 5 mM HEPES, 2 mM CaCl₂, and 1 mM MgCl₂; and 0 mM Ca²⁺ Ringer solution: 155 mM NaCl, 4.5 mM KCl, 10 mM d-glucose, 5 mM HEPES, 1 mM EGTA, and 3 mM MgCl₂.

Primary Abs and their sources were as follows: COX-2 (12282; Cell Signaling Technology), β-actin (3700; Cell Signaling Technology), phospho-ERK (4370; Cell Signaling Technology), total ERK (9102; Cell Signaling Technology), cPLA₂ (2832; Cell Signaling Technology), STIM1 (3917; Feske Lab), and α-tubulin (ab52866; Abcam). Pharmacological tools used in the study were: UTP (U6875; Sigma-Aldrich), SLIGKV-NH2 (4153; Tocris Bioscience), FK-506 (3631; Tocris Bioscience), BTP2 (203890; Sigma-Aldrich), ATP (A6419; Sigma-Aldrich), diclofenac (4454; Tocris Bioscience), apocynin (4663; Tocris Bioscience), NAC (106425; Sigma-Aldrich), U0126 (1144; Tocris Bioscience), ATPγS (4080; Tocris Bioscience), AACOCF3 (1462; Tocris Bioscience), AR-C 118925XX (4890; Tocris), NF546 (3892; Tocris Bioscience), S3QEL 2 (5735; Tocris Bioscience), ADPβS (A8016; Sigma-Aldrich), UDP (U4125; Sigma-Aldrich), NF157 (2450; Tocris Bioscience), Apyrase (A6410 Grade VI, High ATPase/ADPase activity; Sigma-Aldrich), TNP-ATP (2464; Tocris), 5-BDBD (3579; Tocris), A740003 (3701; Tocris Bioscience), Suramin (AC328540500; Acros Organics-Fisher), and PPADS (0625; Tocris Bioscience). CM4620 was a kind gift from CalciMedica.

On day 0, NHBE cells were plated onto 24-well plates in the morning in BEGM lacking antibiotics. Six to 8 h later, the cells were transfected using Lipofectamine RNAiMAX reagent and small interfering RNA (siRNA) of interest at final concentrations of 10 nM. On day 1, cells were given fresh BEGM lacking antibiotics and retransfected using identical conditions as day 0. On day 2, cells were given fresh BEGM lacking antibiotics and lacking hydrocortisone and retransfected using identical conditions as day 0. On day 3 (72 h after initial transfection), cells were either collected for analysis of protein levels or mRNA levels or stimulated with ATP for 2 h for PGE₂ analysis according to the standard PGE₂ stimulation protocol (see PGE₂ measurements section). siRNAs used included “siRNA Universal Negative Control #1” SICOO1 (Sigma-Aldrich), hereafter termed “siCon,” siSTIM1: SASI_Hs01_00107803 (Sigma-Aldrich), and siORAI1: 4392420, assay ID s228396 (Thermo Fisher Scientific). For short hairpin RNA (shRNA) experiments, cells were infected with multiplicity of infection of 10. Two days later, puromycin was added to the culture media at a final dose of 4 μg/ml, and selection was allowed to occur for 3 d. Nontransduced cells were always handled in parallel to confirm puromycin’s ability to induce selection. Cells were then plated for experiments and maintained in 1 μg/ml puromycin until the time of stimulation when puromycin was removed from the culture media, and cells were stimulated in BEBM without growth factors. Lentiviral particles expressing shRNA against P2RX4 were purchased from Sigma-Aldrich with the clone identification number TRCN0000044962.

NHBE cells were grown on poly-l-lysine–coated glass-bottom dishes purchased from MatTek. Cells were loaded with 2 μM Fura-2-AM (F1221; Thermo Fisher Scientific) in BEGM with 5% FBS added to increase loading. Cells were loaded for 40 min at room temperature in the dark. Cells were washed three times with 2 mM Ca²⁺ Ringer solution and then incubated for an additional 15 min in the dark before initiating Ca²⁺ imaging. Experiments were performed at room temperature. Dishes were mounted on the stage of an Olympus IX71 inverted microscope. Images were acquired every 6 s at excitation wavelengths of 340 nm and 380 nm and an emission wavelength of 510 nm. Image acquisition and analysis were performed using SlideBook software. For data analysis, regions of interest were drawn around individual cells, background fluorescence was subtracted, and the F₃₄₀/F₃₈₀ ratios were calculated for each time point. An increase in the ratio of F₃₄₀/F₃₈₀ indicates a rise in intracellular Ca²⁺ concentration ([Ca²⁺]_i). The F₃₄₀/F₃₈₀ ratios were then converted to an estimate of [Ca²⁺]_i through the equation: [Ca²⁺]_i = β × K_d × (R − R_min)/(R_max − R), where R is the F₃₄₀/F₃₈₀ ratio and the values of β, R_min, and R_max were determined from an in vitro calibration with Fura-2 pentapotassium salt. β is determined from the F_min/F_max ratio at 380 nm, and K_d is the dissociation constant of Fura-2 binding to Ca²⁺ (135 nM). The determined values were β = 23.152, R_min = 0.2092, and R_max = 6.954.

On day 0, NHBE cells were plated onto 24-well plates. On day 1, BEGM was replaced with BEGM lacking hydrocortisone. On day 2, cells were stimulated. If cells were pretreated with drugs, half of the BEGM lacking hydrocortisone was taken off the cells, the drug was added to that media at 2× final concentration, vortexed, and added back to the relevant wells. Cells were activated for either 15 min (for ERK1/2 measurements) or 2 h to overnight (for COX-2 upregulation measurements) as indicated in the figure legends. Following stimulation, cells were lysed using 1× Cell Signaling Lysis Buffer (9803S) containing protease and phosphatase inhibitors (PPIs) (78440; Thermo Fisher Scientific). Cell lysates were incubated on ice for 30 min and vortexed every 10 min. Samples were spun at 12,000 rpm for 15 min at 4°C, and supernatants were collected and analyzed via Western blotting.

Lysates were boiled for 5 min in 1× Laemmli Sample Buffer (1610747; Bio-Rad) containing 2-ME. Samples were then subjected to SDS-PAGE. Transfer occurred at 4°C for 1.5 h and at 100 V. Polyvinylidene difluoride membranes were used for the transfer. Following transfer, membranes were washed in TBST (0.1% Tween 20), blocked for 1 h at room temperature using 5% BSA dissolved in TBST, and then incubated overnight at 4°C with primary Abs. Dilutions for primary Abs were as follows: anti-STIM1: 1:1000; anti–COX-2: 1:1000; anti–phospho-ERK1/2: 1:2000; anti-total ERK1/2: 1:1000; anti–β-actin: 1:2000; anti-cPLA2: 1:1000; anti-histone H3: 1:1000; and anti–α-tubulin: 1:5000. The following day, membranes were washed three times for 5 min each using TBST and incubated with secondary Abs for 1 h at room temperature in 5% BSA dissolved in TBST. Li-Cor secondary Abs were used (IRDye 800CW or IRDye 680RD) at dilutions of 1:10,000. Membranes were washed three times for 5 min each using TBST and immediately imaged using an Odyssey CLx imaging system.

On day 0, NHBE cells were plated onto six-well plates. On day 1, BEGM was replaced with BEGM lacking hydrocortisone. On day 2, cells were stimulated. If cells were pretreated with drugs, half of the BEGM lacking hydrocortisone was taken off the cells, and the drug was added to that media at 2× final concentration, vortexed, and added back to the relevant wells. This protocol was adapted from Chang et al. (35). Cells were stimulated for 5 min in BEBM lacking all growth factors. Media was immediately aspirated, cells were placed on ice, and cells were scrapped in hypotonic lysis buffer containing PPIs (78440; Thermo Fisher Scientific) and collected into prechilled Eppendorf tubes. Lysates from two wells were combined into one tube to ensure sufficient protein content. Samples were incubated on ice for 10 min. Samples were then spun at 1000 rpm for 4 min at 4°C. The supernatant was collected as the crude cytosolic fraction, and the pellet was resuspended in hypertonic lysis buffer containing PPIs. Both the crude cytosolic fractions and the resuspended pellet were incubated on ice for another 30 min and periodically vortexed every 10 min to ensure full lysis. Next, both crude cytosolic fraction samples and the resuspended pellet samples were centrifuged at 12,000 rpm for 10 min at 4°C, and the supernatants were collected and termed “cytosolic fraction” and “nuclear fraction.” The hypotonic buffer consisted of: 10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl₂, 0.5 mM EDTA, and PPIs added fresh. The hypertonic buffer consisted of: 20 mM HEPES (pH 7.9), 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 25% glycerol, and PPIs added fresh.

On day 0, NHBE cells were plated onto 24-well plates. On day 1, BEGM was replaced with BEGM lacking hydrocortisone. On day 2, cells were stimulated. If cells were pretreated with drugs, half of the BEGM lacking hydrocortisone was taken off the cells, and the drug was added to that media at 2× final concentration, vortexed, and added back to the relevant wells. At the time of stimulation, media was removed, and cells were stimulated for 2 h in BEBM lacking all growth factors. Following this stimulation phase, the supernatants were collected, spun at 300 × g for 4 min to remove cellular debris, and the supernatant from this sample was then collected and stored at −80°C until the time of analysis. To perform analysis of PGE₂ levels, the Cayman Chemical kit (514010) was used, and the manufacturer’s protocols were followed. All samples were diluted at least 10-fold in ELISA buffer prior to analysis. All samples that were compared statistically had the same concentration of organic solvents (such as DMSO and ethanol) as organic solvents can interfere with the assay.

On day 0, NHBE cells were plated onto 24-well plates. On day 1, BEGM was replaced with BEGM lacking hydrocortisone. On day 2, cells were stimulated. If cells were pretreated with drugs, half of the BEGM lacking hydrocortisone was taken off the cells, and the drug was added to that media at 2× final concentration, vortexed, and added back to the relevant wells. At the time of stimulation, media was removed, and cells were stimulated overnight (16 h) in BEBM lacking all growth factors. Following this stimulation phase, the supernatants were collected, spun at 300 × g for 4 min to remove cellular debris, and the supernatant from this sample was then collected and stored at −80°C until the time of analysis. To perform analysis of IL-6 levels, the RayBiotech Human IL-6 ELISA kit (ELH-IL6-1) was used, and the manufacturer’s protocols were followed. All samples were diluted typically 5-fold in ELISA buffer prior to analysis. Samples that were compared statistically had the same concentration of organic solvents (DMSO).

Total RNA was extraction was performed using RNeasy Plus Mini Kit (74134; Qiagen). cDNA generation was performed using iScript Reverse Transcription Supermix for RT-qPCR (1708841; BioRad). Quantitative PCR (qPCR) was performed using PowerUp SYBR Green Master Mix (A25741). For qPCR, the final concentration of primers was 500 nM, and cDNA was used at 6 ng/well. Primer sequences were as follows: ORAI1 forward, 5′-GATGAGCCTCAACGAGCACT-3′, ORAI1 reverse, 5′-ATTGCCACCAT-GGCGAAGC-3′; P2RY2 forward, 5′-CCGCTTCAACGAGGACTTCAA-3′, P2RY2 reverse, 5′-GCGGGCGTAGTAATAGACCA-3′; P2RX4 forward, 5′-CTACCAGGAAACTGACTCCGT-3′, P2RX4 reverse, 5′-GGTATCACATAATCCGCCACAT-3′; HPRT1 forward, 5′-ACCCTTTCCAAATCCTCAGC-3′, HPRT1 reverse, 5′-GTTATGGCGACCCGCAG-3′; RPLP0 forward, 5′-AGCCCAGAACACTGGTCTC-3′, RPLP0 reverse, 5′-ACTCAGGATTTCAATGGTGCC-3′; PPIA forward, 5′-CCCACCGTGTTC-TTCGACATT-3′, PPIA reverse, 5′-GGACCCGTATGCTTTAGGATGA-3′; P2RY11 forward, 5′-GTAGCAGACACAGGCTGA-3′, P2RY11 reverse, 5′-CCTGGAACCCACTGAGTTTG-3′; P2RX1 forward, 5′-CGTTATCTTCCGACTGATCCAG-3′, P2RX1 reverse, 5′-CACAGAGACACTGCTGATGAG-3′; P2RX2 forward, 5′-GAGGTGTTCGGCTGGTG-3′, P2RX2 reverse, 5′-GGTAGTGGATGCTGTTCTTGA-3′; P2RX3 forward, 5′-CAACATCATC-CCCACCATCA-3′, P2RX3 reverse, 5′-CTCATTCACCTCCTCAAACTTCT-3′; P2RX5 forward, 5′-GGAAGCAGCAGTCAGAAGG-3′, P2RX5 reverse, 5′-AAAGGCATGGGATCACTGG-3′; P2RX6 forward, 5′-TGGCCTCACTACTCCTTCC-3′, P2RX6 reverse, 5′-ATGTCGAAGCGGATTCCATAG-3′; and P2RX7 forward, 5′-CCCTGTGTGTGGTCAACGAAT-3′, P2RX7 reverse, 5′-TGCAGACTTCTCCCTAGTAGC-3′.

All bar graphs summarizing data are represented as mean ± SEM. Individual points are always indicative of biological replicates. For data sets involving more than two groups, an initial one-way ANOVA was performed followed by the Tukey multiple-comparison test. For data sets with only two groups, a two-tailed unpaired Student t test was performed. Dose-response curves were created using a four parameter (variable slope) nonlinear regression. Extra sum-of-squares F test was used to statistically compare the PGE₂ and IL-6 dose-response curves. Statistical analysis and data analysis were performed using Prism 8 (GraphPad Software). When statistical comparisons are made: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

All data needed to evaluate the conclusions in the study are present in the article or Supplemental Figs. 1, 2, 3 and 4. CM4620 was obtained via a material transfer agreement from CalciMedica.

ATP acts on two broad classes of membrane receptors: G-protein–coupled P2Y receptors and ionotropic P2X receptors (36, 37). Previous studies have shown that AECs express the G_q/phospholipase C–coupled P2Y₂ and P2Y₆ receptors as well as several classes of P2X receptors (38, 39). Because both P2X and P2Y receptors can initiate Ca²⁺ signaling, we first investigated the identity of the purinergic receptors contributing to Ca²⁺ signals evoked by extracellular ATP in primary NHBE cells using the ratiometric dye, Fura-2, as previously described (32, 33). These experiments revealed that administration of a saturating dose of ATP (100 µM) evoked a biphasic elevation in [Ca²⁺]_i consisting of an initial rapid phase of [Ca²⁺]_i elevation followed by a sustained phase of [Ca²⁺]_i elevation (Fig. 1A). Removing extracellular Ca²⁺ resulted in the selective loss of the sustained [Ca²⁺]_i rise with virtually no effect on the initial [Ca²⁺]_i signal (Fig. 1B), indicating that the early, transient increase of [Ca²⁺]_i results from release of Ca²⁺ from intracellular stores, whereas the sustained late phase requires Ca²⁺ entry across the plasma membrane. This kinetic signature of the ATP-induced Ca²⁺ response is consistent with nucleotide-evoked G_q/phospholipase C–inositol 1,4,5-triphosphate signaling and store-operated calcium entry (SOCE) that we and others have previously described in AECs and other cell types (32, 40–42). In agreement with this interpretation, UTP, a uridine nucleotide ligand that is selective for the metabotropic P2Y purinergic receptors (P2Y₂ and P2Y₄), also induced Ca²⁺ elevations with amplitude and kinetics similar to Ca²⁺ elevations evoked by ATP (Fig. 1C). These results indicate that the P2Y receptors are important mediators of the ATP-mediated [Ca²⁺]_i response.

Among the various P2Y receptors, P2Y₂ is highly expressed by AECs and shares a similar potency and efficacy for ATP and UTP (39, 43). To examine the role of P2Y₂ in mediating the [Ca²⁺]_i rises, we applied the selective P2Y₂ receptor antagonist AR-C 118925XX (44) (henceforth referred to as “AR-C”). Pretreatment with AR-C abolished the ATP and UTP-mediated Ca²⁺ signals (Fig. 1D–F). By contrast, the Ca²⁺ elevation elicited by activation of PAR2, an unrelated G_q receptor, which also evokes SOCE (32), was unaffected (Fig. 1D, 1E). These results indicate that the nucleotide triphosphate agonists ATP and UTP stimulate Ca²⁺ signaling in human bronchial epithelial cells via activation of Gq-coupled P2Y₂ receptors, reminiscent of SOCE activated by ATP and UTP in astrocytes (40).

The finding that P2Y₂ receptor activation by ATP and UTP elicits a sustained phase of Ca²⁺ entry (Fig. 1A) led to us to next consider whether this secondary phase requires the activation of CRAC channels. Previous work has demonstrated ORAI1 and STIM1, components of CRAC channels, are required for PAR2-mediated SOCE in AECs (32, 33, 41). To address the molecular identity of the channels responsible for SOCE following P2Y₂ receptor activation, we examined the consequences of siRNA-mediated knockdown of ORAI1 and STIM1 on ATP-mediated Ca²⁺ rises. The effectiveness of knockdown was assessed to be >90% based on Western blots for STIM1 and by qPCR for ORAI1 (Supplemental Fig. 1A, 1B). Knockdown of either ORAI1 or STIM1 significantly decreased the sustained Ca²⁺ entry phase but did not substantively affect the initial Ca²⁺ release from intracellular stores (Fig. 2A, 2B). Moreover, the sustained phase of the Ca²⁺ rise was abolished by the CRAC channel inhibitors, BTP2 and CM4620 (Fig. 2C, 2D) (45–49). Ca²⁺ add-back experiments following depletion of endoplasmic reticulum Ca²⁺ stores in Ca²⁺-free Ringer solution revealed robust SOCE in NHBE cells, which was completely abolished by the ORAI1 inhibitor CM4620 (Supplemental Fig. 1C, 1D). Together, these results, summarized in (Fig. 2E, indicate that CRAC channels composed of ORAI1 and STIM1 are essential for SOCE that is activated by stimulation of P2Y₂ receptors in bronchial epithelial cells.

As described above, a major mediator produced by AECs that limits the immune–inflammatory response and promotes resolution of inflammation is PGE₂ (24). PGE₂ induction is stimulated by ATP (50), but the purinergic receptors mediating this effect are unclear. We therefore sought to identify the ATP receptors driving synthesis of PGE₂. Administration of ATP resulted in rapid synthesis of PGE₂ (Fig. 3A), and analysis of the dose-response relationship indicated an EC₅₀ of ∼7 μM (Fig. 3B). Likewise, UTP, which stimulates P2Y₂ and P2Y₄ receptors, also induced robust and rapid PGE₂ synthesis within 2 h (Fig. 3C). The agonist efficacies of ATP and UTP were similar (Fig. 3C), suggesting that both ligands act on a common membrane receptor. Consistent with this interpretation, pretreatment with the selective P2Y₂ receptor antagonist AR-C completely blocked the PGE₂ induction by both ATP and UTP (Fig. 3C). Further, siRNA-mediated knockdown of P2RY2 mRNA (Supplemental Fig. 2A) resulted in complete abrogation of the ATP-induced PGE₂ response, confirming the involvement of P2Y₂ receptors (Fig. 3D). AR-C and knockdown of P2RY2 also inhibited basal PGE₂ synthesis (Fig. 3E, 3F), indicating that activation of P2Y₂ receptors by AEC-derived nucleotides drives the constitutive PGE₂ synthesis in resting cells. Collectively, these results demonstrate that nucleotide-driven PGE₂ synthesis in AECs occurs via activation of P2Y₂ receptors, implicating epithelial P2Y₂ receptors in the beneficial effects of PGE₂ in the lung airways.

As noted above, in addition to its ability to stimulate PGE₂ synthesis, extracellular ATP is also linked to the production of the proinflammatory cytokine IL-6 (51). To begin to understand how signaling underlying induction of IL-6 contrasts with that of PGE₂, we examined the pathways regulating ATP-mediated induction of IL-6. Administration of ATP to AECs induced IL-6 secretion, albeit with slower kinetics than PGE₂ (compare (Figs. 4A and (3A). Moreover, ATP-mediated synthesis of IL-6 occurred with an EC₅₀ of ∼16 μM (Fig. 4B), ∼2-fold higher than the EC₅₀ required for PGE₂ induction (∼7 μM), raising the possibility that divergent signaling mechanisms are involved in ATP-mediated PGE₂ and IL-6 induction. ATPγS, an ATP variant that is resistant to degradation by ecto-ATPases, also significantly induced IL-6 ruling out ATP metabolites such as ADP or adenosine in inducing the IL-6 response (Fig. 4C). Additionally, preincubating ATP with apyrase, an enzyme that degrades ATP, abolished IL-6 induction, indicating ATP is necessary for the IL-6 release (Fig. 4D). Together, these results indicate that ATP-induced IL-6 production occurs at higher ligand doses and requires longer periods of cell stimulation than the synthesis of PGE₂.

Several lines of evidence indicated that ATP-mediated induction of IL-6 is not mediated by activation of P2Y receptors. First, unlike the phenotype for PGE₂ synthesis, ATP-evoked IL-6 secretion was not affected by the P2Y₂ antagonist AR-C (Fig. 4E). Second, UTP, which acts on P2Y₂ and P2Y₄ receptors, failed to induce IL-6 (Fig. 4E). This is in striking contrast with UTP’s ability to strongly induce PGE₂ (Fig. 3C). Third, ADPβS, a ligand for P2Y₁, P2Y₁₂, and P2Y₁₃ receptors, also had no effect (Supplemental Fig. 2B). Fourth, UDP and NF546 (agonists for P2Y₆, P2Y₁₁, and P2Y₁₄ receptors) were also unable to drive IL-6 secretion (Supplemental Fig. 2C, 2D). Puzzlingly, NF157, a P2Y₁₁ receptor antagonist, significantly enhanced the potency of ATPγS-induced IL-6 synthesis (Supplemental Fig. 2E). We did not further investigate the mechanism of this synergism but continued our focus on the receptor subtype that mediates ATP-driven IL-6 induction. Collectively, these results indicate that in contrast to PGE₂ synthesis, ATP-driven IL-6 synthesis is not mediated by P2Y₂ receptors.

By contrast, the broad-spectrum P2X antagonists suramin and PPADS (36) both blocked ATPγS-mediated IL-6 production (Fig. 4F). mRNA analysis by real-time PCR revealed significant expression of the P2X receptor subtype transcripts P2X₄, P2X₅, and P2X₇, which agrees well with previous reports (52, 53) (Supplemental Fig. 2F). In an attempt to further narrow down the receptor subtypes involved, we next examined the effects of low doses of TNP-ATP, which inhibits P2X₁, P2X₃, and P2X_2/3 heteromeric receptors, on IL-6 production (36). However, TNP-ATP had no effect on IL-6 production (Fig. 4G). Likewise, the selective P2X₄ and P2X₇ receptor antagonists 5-BDBD and A740003 also did not inhibit ATPγS-induced IL-6 induction (Fig. 4H, 4I). shRNA-mediated knockdown of P2RX4 also failed to inhibit of IL-6 production (Supplemental Fig. 2G, 2H) consistent with the above pharmacological results. Endogenous P2X receptors are known for their ability to oligomerize to form channels with biophysical and pharmacological properties that are distinct from homomeric P2X receptor subunits (36). For example, a previous characterization of the biophysical properties of P2X receptors in AECs has concluded that P2X channels have unique features potentially indicative of heteromers of P2X₄/P2X₇ subunits (54). Thus, although our tests did not implicate specific P2X receptor subtypes, the failure of P2Y receptor agonists and antagonists to affect IL-6 production, the strong expression of P2X receptor subtypes in AECs, and the ability of the broad-spectrum P2X receptor inhibitors suramin and PPADS to block IL-6 production implicates P2X receptors in driving ATP-mediated IL-6 production.

The results presented thus far indicate that ATP stimulates the production of bronchoprotective PGE₂ and proinflammatory IL-6 through distinct membrane receptors. Consistent with previous evidence indicating that PGE₂ synthesis is a Ca²⁺-regulated process (50), siRNA-mediated knockdown of ORAI1 or STIM1 abrogated the ATP-evoked PGE₂ synthesis (Fig. 5A). Likewise, inhibiting SOCE with the CRAC channel inhibitors CM4620 and BTP2 effectively blocked both ATP- and UTP-induced PGE₂ synthesis (Fig. 5B, 5C). These results indicate that CRAC channels are essential for ATP-evoked PGE₂ synthesis. Similarly, pharmacological inhibition of CRAC channels with BTP2 or CM4620 inhibited ATPγS-evoked IL-6 induction (Fig. 5D). CM4620 has recently been shown to improve outcomes in hospitalized patients with severe COVID-19 infections (55). Thus, the ability of CM4620 to block IL-6 induction suggests that CRAC channel blockade may alleviate IL-6–mediated inflammation in the lung. Taken together, these results indicate that activation of CRAC channels is essential for both ATP-mediated PGE₂ and IL-6 production in AECs.

To further understand the mechanism by which CRAC channels regulate PGE₂ production, we examined the role of local Ca²⁺ signaling by using Ca²⁺ buffers EGTA and BAPTA. Local Ca²⁺ signals in the vicinity of CRAC channels are implicated in many downstream signaling events, including the synthesis of the arachidonic acid metabolite leukotriene C₄ in mast cells (56), calcineurin/NFAT activation in neural stem cells (57), and gliotransmitter release in astrocytes (40). We found that BAPTA, which binds Ca²⁺ with rapid kinetics, abolished the P2Y₂ receptor–mediated increases in PGE₂ synthesis, but EGTA, a Ca²⁺ chelator with comparable Ca²⁺ affinity to BAPTA but with 100-fold slower binding kinetics (58), was far less effective ((Fig. 5E). BAPTA also inhibited basal PGE₂ synthesis, whereas EGTA had no effect (Fig. 5F). Attempts to examine EGTA/BAPTA regulation of IL-6 production were unsuccessful due to high cell death in buffer-loaded cells over the 16-h measurement period for IL-6. Collectively, these results indicate that CRAC channels are spatially coupled to the PGE₂ synthesis machinery (likely cPLA₂, see below), analogous to the functional coupling described in other cells between CRAC channels and calcineurin/NFAT signaling and gliotransmitter vesicles (40, 57).

To further determine how P2Y₂–CRAC channel signaling stimulates PGE₂ synthesis, we next examined signaling downstream of P2Y₂ receptor activation. Previous work in the alveolar type II cancer cell line A549 has indicated that extracellular ATP and UTP activate and increase the expression of the inducible COX isoform COX-2 (59). However, whether primary human bronchial epithelial cells share the same mechanism is unknown. We found that low doses of FR122047 and celecoxib, which are selective inhibitors of the COX-1 and COX-2 enzymes, respectively, abolished ATP-mediated PGE₂ synthesis (Supplemental Fig. 3A), indicating that COX enzyme activity is essential for ATP-mediated PGE₂ synthesis. However, examination of COX-2 protein levels revealed only a very modest increase in COX-2 protein levels following ATP stimulation at 2 h, which returned to baseline levels by 6 h (Supplemental Fig. 3B). This slight increase in COX-2 expression was inhibited with the P2Y₂ receptor antagonist AR-C (Supplemental Fig. 3C, 3D), but was not affected by the CRAC channel inhibitors CM4620 and BTP2 (Supplemental Fig. 3E). Thus, P2Y₂ receptor–CRAC channel–mediated synthesis of PGE₂ requires COX-2 activity but does not substantively involve increased expression of COX-2 protein.

A key molecule implicated in the induction of PGE₂ by growth factors and cytokines is ERK1/2 (60, 61). In this signaling, ERK1/2 phosphorylates the enzyme cPLA₂, and this step has been shown to be essential for the synthesis of PGE₂ by epidermal growth factor, IL-1β, and PAR2 ligands (60, 62). ERK1/2 activation is regulated by its own phosphorylation by the kinase MEK1/2 (63). Thus, to examine the necessity of ERK1/2 involvement for ATP-induced PGE₂ synthesis, we probed the phosphorylation status of ERK1/2 following ATP administration. These experiments showed that extracellular ATP rapidly induced ERK1/2 activation within 15 min (Fig. 6A, 6B). The P2Y₂ antagonist AR-C completely blocked this step (Fig. 6A, 6B). Consistent with a requirement for MEK1/2-mediated phosphorylation, the selective MEK1/2 inhibitor U0126 (20 μM) abrogated ERK1/2 activation (Supplemental Fig. 3F). Importantly, ATP- and UTP-induced PGE₂ synthesis was strongly inhibited by MEK1/2 inhibition (Fig. 6C, 6D), reaffirming the importance of ERK1/2 activation for P2Y₂-induced rapid PGE₂ synthesis. Further, MEK1/2 inhibition did not affect the modest upregulation of COX-2 expression (Fig. 6E, 6F), suggesting that COX-2 upregulation is not a major contributor to ATP-induced PGE₂ synthesis. Interestingly, the CRAC channel inhibitors BTP2 or CM4620 failed to attenuate ERK1/2 activation (Supplemental Fig. 3G). Thus, although both CRAC channels and ERK1/2 activation are required for PGE₂ synthesis by ATP and UTP, they appear to be independent pathways converging on enzymes in the PGE₂ synthesis pathway, likely cPLA₂ (60). Finally, inhibiting the MEK1/2-ERK1/2 kinase cascade with U0126 also impaired ATPγS-induced IL-6 induction (Fig. 6G). Thus, MEK1/2-ERK1/2 kinase signaling appears to be necessary for the synthesis of both PGE₂ and IL-6 by extracellular ATP.

Previous work has shown that, in addition to Ca²⁺ signaling, the induction of PGE₂ and IL-6 is regulated by numerous other signaling pathways, including reactive oxygen species (ROS) (59, 64, 65). The two primary sources of ROS in most cells are the plasma membrane–associated NADPH enzymes and mitochondria (66). To examine the potential involvement of these ROS sources for ATP-induced synthesis of PGE₂ and IL-6, we tested the mitochondrial complex III ROS inhibitor S3QEL-2 and the NADPH oxidase inhibitor apocynin. These experiments showed that apocynin abolished nucleotide-driven PGE₂ synthesis, whereas S3QEL-2 had no effect (Supplemental Fig. 4A, 4B). NADPH oxidase inhibition did not attenuate ERK1/2 activation downstream of P2Y₂ (Supplemental Fig. 4C, 4D). Thus, production of ROS by NADPH oxidase, but not mitochondria, is necessary for PGE₂ synthesis. By contrast, both S3QEL-2 and apocynin decreased ATPγS-induced production of IL-6, indicating that both sources of ROS regulate IL-6 induction (Supplemental Fig. 4E). These results reveal an essential role of ROS for both PGE₂ synthesis and IL-6 secretion by ATP, but suggest that the sources of ROS differ for the induction of the two mediators.

We next interrogated the role of Ca²⁺-dependent enzymes in purinergic-driven PGE₂ synthesis and IL-6 secretion. A key Ca²⁺-dependent enzyme involved in agonist-driven transcriptional responses is calcineurin. To examine a role for calcineurin/NFAT signaling in the ATP-mediated IL-6 response, we used FK-506, a widely used small-molecule inhibitor of calcineurin/NFAT signaling. FK-506 abrogated ATPγS-induced IL-6 secretion (Fig. 7A), indicating that the ATP-induced IL-6 response is mediated by NFAT-dependent gene transcription, as previously shown for PAR2-mediated induction of IL-6 (32).

By contrast, FK-506 did not inhibit ATP-induced PGE₂ synthesis (Fig. 7B), consistent with the evidence presented above indicating that acute PGE₂ production is not substantively influenced by calcineurin (Supplemental Fig. 3A–E). Another key Ca²⁺-dependent enzyme linked to PGE₂ synthesis is cPLA₂, which liberates arachidonic acid from lipid membranes and is regulated independently both by ERK1/2 phosphorylation and Ca²⁺ binding to its C2 domain (67, 68). ATP-induced PGE₂ production was blocked by the cPLA₂ inhibitor AACOCF₃, whereas ATPγS-induced IL-6 secretion was not inhibited (Fig. 7C, 7D). Upon Ca²⁺ binding, cPLA₂ translocates to intracellular organelle membranes of the endoplasmic reticulum, Golgi, and nucleus, where it liberates arachidonic acid from the resident phospholipids (67, 68). Consistent with this signaling, cell fractionation analysis (35) revealed that ATP stimulation of AECs induced rapid enrichment of cPLA₂ in the nuclear fraction (Fig. 7E, 7F), a response that was blocked by the P2Y₂ receptor antagonist AR-C (Fig. 7E 7F). The CRAC channel inhibitors CM4620 and BTP2 abrogated the P2Y₂ receptor–mediated recruitment of cPLA₂ to the nuclear fraction (Fig. 7G–J). Collectively, these data indicate that P2Y₂ receptors stimulate PGE₂ synthesis via CRAC channel–mediated activation of cPLA₂ (Fig. 8).

In the airways, ATP and UTP signaling is linked to a wide range of immune effects. High concentrations of extracellular ATP are linked to the production of inflammatory cytokines, leading to immune cell infiltration, airway remodeling, and hyperresponsiveness (2, 4, 5, 51). Patients with asthma, COPD, and ARDS show increased airway ATP levels in the BAL fluid, and elevated extracellular ATP is thought to contribute to disease pathology (2, 8, 9). However, although high concentrations of extracellular ATP are considered proinflammatory, ATP (and UTP) can also stimulate many anti-inflammatory (69) and physiologically beneficial effects in the lung, including MCC, PGE₂ synthesis, and wound healing (10–12, 14, 50). The differential cellular pathways mediating these potentially beneficial and harmful effects are not well understood.

In this study, we describe the cell signaling mechanisms underlying the divergent effects of ATP and UTP on PGE₂ and IL-6 production from AECs involving distinct purinergic receptor subtypes and downstream effectors (Fig. 8). Our results indicate that ATP/UTP signaling through P2Y₂ receptors activates CRAC channels and ERK1/2 and subsequently cPLA₂, resulting in rapid synthesis of PGE₂, which is known to evoke bronchoprotective and immune-suppressive effects (17–21, 24, 70). To our knowledge, this is the first identification of AEC P2Y₂ receptors in stimulating protective PGE₂ synthesis. By contrast, ATP-mediated induction of IL-6, which is implicated in proinflammatory responses in the airways (25–28, 30, 31), appears to occur through stimulation of P2X receptors through a process requiring CRAC channels and ERK1/2 activation, but also distinctly involving calcineurin/NFAT activation (Fig. 8).

Several key features both upstream and downstream of CRAC channels underscore the differences in the two pathways (Fig. 8). First, although UTP strongly induces PGE₂, it is completely ineffective in evoking IL-6 synthesis (Fig. 3C versus (Fig. 4E). ATP by contrast evokes both PGE₂ and IL-6 synthesis, albeit at different potencies (7 μM versus ∼16 μM; (Fig. 4B) and over significantly different time courses (Fig. 3A versus (Fig. 4A). As noted above, this difference is related to the different receptors (P2Y₂ versus P2X receptors) involved in stimulating PGE₂ and IL-6. Second, the Ca²⁺-dependent signaling pathways downstream of receptor activation significantly differ: P2Y₂ receptor–driven PGE₂ synthesis involves Ca²⁺ activation of cPLA₂, whereas P2X-driven IL-6 secretion requires Ca²⁺-dependent transcription via the calcineurin/NFAT pathway. As well, distinct sources of ROS appear to be involved in regulating the production of these mediators (Supplemental Fig. 4).

What are the functional implications of these findings, and under what conditions could these differing outcomes become apparent? Although speculative, we can envision two scenarios. First, the observation that PGE₂ synthesis by ATP occurs at lower ATP doses than that required for IL-6 suggests that protective versus proinflammatory outcomes for ATP may be dictated in part by the concentration and duration of ATP signaling in the airways. In the healthy lung, low micromolar levels of ATP that are naturally released from cells may activate tonic baseline P2Y₂ signaling conducive to selective PGE₂ induction and its bronchoprotective effects. By contrast, higher levels of ATP occurring over longer durations, such as those found under conditions of severe damage and/or cellular necrosis, would be predicted to elicit strong IL-6 synthesis and IL-6–mediated inflammation. High-dose ATP is also known to engage the NLRP3 inflammasome, leading to additional pathways for inflammation (71). Second, extracellular UTP secretion would be predicted to exclusively induce PGE₂. Growing evidence points to UTP as a physiologically relevant signaling molecule linked to numerous cellular processes, including ion transport, ciliary beat frequency, and mucin release (15). Although regulated secretion of UTP is very poorly understood likely due to lack of tools to easily detect extracellular UTP (72), release of UTP in vitro following mechanical stress, apoptosis, and solution exchange has been demonstrated using HPLC from a number of tissues, including AECs (73–76). UTP release under these conditions would be expected to preferentially evoke bronchoprotective responses in the lung. Future in vivo studies using mouse models in which P2Y₂ receptors are selectively deleted in the lung epithelium could shed light on this important question.

An interesting feature of PGE₂ synthesis is its exquisite sensitivity to blockade by the rapid Ca²⁺ buffer BAPTA but not the slower buffer EGTA (Fig. 5E, 5F). This finding implies that PGE₂ synthesis relies on local Ca²⁺ signaling likely through Ca²⁺ microdomains near CRAC channels that are functionally linked to cPLA₂ activation. Previous studies have shown that Ca²⁺ microdomains arising from CRAC channels can stimulate arachidonic acid release and leukotriene production in mast cells (35). Ca²⁺ microdomains around CRAC channels are also linked to NFAT-dependent gene transcription and exocytosis in neuronal stem cells and astrocytes (40, 57). Thus, the finding that local Ca²⁺ signals around CRAC channels are essential for cPLA₂-mediated synthesis of PGE₂ broadens the role of Ca²⁺ microdomains linked to enzyme activation in different cell types.

Two important mechanistic questions raised by our study that remain to be addressed relate to the nature of the Ca²⁺ signal activated by CRAC channels that drives IL-6 induction and the mechanism of how P2X receptor stimulation is coupled to activation of CRAC channels and IL-6 synthesis. The data indicate that antagonism of P2Y₂ receptors blocks both ATP- and UTP-evoked Ca²⁺ elevations to similar extents (Fig. 1F). Yet, unlike ATP, UTP is completely ineffective in inducing IL-6 (Fig. 4E). Because CRAC channel blockade strongly inhibits induction of IL-6 by ATP (Fig. 5D), this result suggests that ATP-mediated induction of IL-6 requires Ca²⁺ influx reliant on CRAC channels that is not readily detected by Fura-2. We speculate that P2X receptors may cause downstream activation of CRAC channels over longer time periods than was evaluated in our Ca²⁺ imaging experiments, and this step is sufficiently local so as to evade detection by the bulk Ca²⁺ indicator, Fura-2. In this scenario, the functional coupling between P2X receptors and CRAC channels could involve localized ryanodine receptor–mediated calcium-induced calcium release events of the type that have been described in skeletal muscle and T cells (77, 78). Additional mechanistic studies using low-affinity and membrane-tethered Ca²⁺ indicators and genetic tools to manipulate specific P2X receptors are needed to help address the unknown links among P2X receptors, CRAC channels, and IL-6 induction.

Although P2X receptors have been widely investigated in many physiological contexts, their physiological roles and effector signaling mechanisms in AECs are not well understood. One study described a key role for P2X receptor activation in inducing IL-8 production via Ca²⁺ signaling and NF-κB (52), suggesting that P2X receptors may have an important role in AECs to drive proinflammatory responses. Our finding that the P2X receptor antagonists suramin and PPADS strongly suppress ATP-evoked synthesis of IL-6 in AECs is in agreement with this suggestion and expands the potential roles of P2X receptors in the airways to include the inflammatory cytokine IL-6.

Finally, airway P2Y₂ receptors and CRAC channels have attracted significant therapeutic interest in recent years. A stable P2Y₂ receptor agonist (denufosol) was investigated in patients with cystic fibrosis based on its ability to stimulate MCC (79). Unfortunately, long-term treatment with denufosol showed no improvement of lung function in patients with cystic fibrosis (80). CRAC channel inhibitors have also shown efficacy in preclinical models of asthma (81–83), and a new-generation CRAC channel inhibitor, CM4620, is currently being tested in human patients for relieving the cytokine storm in seriously ill patients with COVID-19 (55). Our results showing that CM4620 is very effective in occluding IL-6 production may provide the mechanistic explanation for the benefits of this small molecule in improving patient survival. Although CRAC channels also play a key role in agonist-evoked PGE₂ synthesis, this may be counterbalanced under chronic inflammatory conditions in vivo due to its role in driving proinflammatory cytokine production, thus providing a therapeutic window to dampen chronic inflammation in the lung airways. More studies are needed to address these scenarios, but the results of this study provide a framework for testing these and other models.

This work used resources of the Northwestern University Structural Biology Facility at the Robert H. Lurie Comprehensive Cancer Center. We thank members of the Prakriya laboratory for helpful discussions and Michaela Novakovic and Mohammad Mehdi Maneshi for technical advice.

K.A.S. is an employee of, and holds stock in, CalciMedica, Inc. The other authors neither received payment from CalciMedica for participation as investigators in the study nor for the development of the present manuscript.