Prostaglandin E2 Inhibits Histamine-Evoked Ca2+ Release in Human Aortic Smooth Muscle Cells through Hyperactive cAMP Signaling Junctions and Protein Kinase A

In human aortic smooth muscle cells, prostaglandin E2 (PGE2) stimulates adenylyl cyclase (AC) and attenuates the increase in intracellular free Ca2+ concentration evoked by activation of histamine H1 receptors. The mechanisms are not resolved. We show that cAMP mediates inhibition of histamine-evoked Ca2+ signals by PGE2. Exchange proteins activated by cAMP were not required, but the effects were attenuated by inhibition of cAMP-dependent protein kinase (PKA). PGE2 had no effect on the Ca2+ signals evoked by protease-activated receptors, heterologously expressed muscarinic M3 receptors, or by direct activation of inositol 1,4,5-trisphosphate (IP3) receptors by photolysis of caged IP3. The rate of Ca2+ removal from the cytosol was unaffected by PGE2, but PGE2 attenuated histamine-evoked IP3 accumulation. Substantial inhibition of AC had no effect on the concentration-dependent inhibition of Ca2+ signals by PGE2 or butaprost (to activate EP2 receptors selectively), but it modestly attenuated responses to EP4 receptors, activation of which generated less cAMP than EP2 receptors. We conclude that inhibition of histamine-evoked Ca2+ signals by PGE2 occurs through “hyperactive signaling junctions,” wherein cAMP is locally delivered to PKA at supersaturating concentrations to cause uncoupling of H1 receptors from phospholipase C. This sequence allows digital signaling from PGE2 receptors, through cAMP and PKA, to histamine-evoked Ca2+ signals.


Introduction
Ca 21 and cAMP are ubiquitous intracellular messengers that regulate most cellular behaviors. The versatility of these messengers depends on both the spatiotemporal organization of the changes in their concentration within cells (Cooper and Tabbasum, 2014) and on interactions between them [see references in Tovey et al. (2008)]. These interactions are important in many smooth muscles, where increases in intracellular free Ca 21 concentration ([Ca 21 ] i ) stimulate contraction, but receptors that stimulate formation of cAMP usually cause relaxation. The clinical importance is clear from the widespread use of b-agonists to provide symptomatic relief from asthma (Morgan et al., 2014). In vascular smooth muscle (VSM), too, cAMP attenuates the contractile responses mediated by many receptors that evoke Ca 21 signals (Morgado et al., 2012). This inhibition is assumed to be mediated by cAMP-dependent protein kinase (PKA) (Murthy, 2006), but there are also PKA-independent effects of cAMP (Spicuzza et al., 2001). At least some of these effects may be through exchange proteins activated by cAMP (EPACs), probably EPAC 1, which is abundant in blood vessels particularly within endothelial cells (Roscioni et al., 2011).
atherosclerosis (Gómez-Hernández et al., 2006), and restenosis (Sasaguri et al., 2005). We demonstrated previously that histamine, through H 1 receptors, stimulates an increase in [Ca 21 ] i in human aortic smooth muscle cells (ASMC). The initial response is mediated by Ca 21 release through inositol 1,4,5-trisphosphate receptors (IP 3 R) and it is followed by Ca 21 entry across the plasma membrane (Pantazaka et al., 2013). PGE 2 , acting largely through EP 2 receptors, both stimulates the activity of adenylyl cyclase (AC) and substantially attenuates the Ca 21 signals evoked by histamine. Here, we show that inhibition of histamine-evoked Ca 21 signals by PGE 2 is mediated by cAMP delivered within "hyperactive signaling junctions." The response does not require EPACs, but it is attenuated by inhibition of PKA. The effect of PGE 2 on histamine-evoked Ca 21 signals does not result from a decrease in IP 3 R sensitivity or from increased Ca 21 extrusion from the cytosol, nor does PGE 2 affect the Ca 21 signals evoked by stimulation of either endogenous type 1 protease-activated receptor (PAR1) or heterologously expressed muscarinic M3 acetylcholine receptors. We suggest that PKA uncouples H 1 histamine receptors from the guanine nucleotide-binding protein, G q/11 , and activation of phospholipase C (PLC). Our results establish that digital signaling from PGE 2 receptors, through cAMP and PKA, inhibits histamine-evoked Ca 21 signals.
Culture of Human Aortic Smooth Muscle Cells. Human ASMC from the American Tissue Culture Collection (Manassas, VA) or Dr. Trevor Littlewood (Boyle et al., 2002) were cultured as described (Pantazaka et al., 2013). Ethical approval for the latter was obtained from Addenbrooke's NHS Trust. Cells were derived from four Caucasian patients (males aged 23, 52, and 54, and a female aged 58), who died of causes unrelated to cardiovascular pathologies. Cells were used between passages two and six.
Measurements of Intracellular cAMP. Confluent cultures of ASMC grown in 24-well plates and labeled with 3 H-adenine were incubated under conditions that replicated those used for measurements of [Ca 21 ] i . Reactions were terminated by aspiration of medium and addition of ice-cold trichloroacetic acid (5% v/v, 1 ml). After 30 minutes on ice, 3 H-cAMP was separated from other 3 H-labeled adenine nucleotides (Pantazaka et al., 2013).
Expression of PKI and M3 Muscarinic Receptors. Plasmids encoding PKI (pRSV-PKI-v2) and its inactive form (pRSV-mut PKI-v2) were from Addgene (cat. no. 45066 and cat. no. 45067; Cambridge, MA) (Day et al., 1989); they were C-terminally tagged with mCherry. Plasmid encoding the human M3 muscarinic acetylcholine receptor was from the cDNA Resource Centre (cat. no. MAR0300000) (Ford et al., 2002). The three constructs were each recombined into BacMam pCMV-DEST. Bacmids were then prepared, and virus was produced from bacmid-infected Sf9 cells according to the manufacturer's instructions (Thermo Fisher Scientific, Runcorn, UK). ASMC were infected at a multiplicity of infection (MOI) of ∼50 and used after 96 hours.
Flash Photolysis of Caged IP 3 . Confluent cultures of ASMC grown on poly-L-lysine-coated imaging dishes (35-mm diameter with a 7-mm glass insert; MatTek Corporation, Ashland, MA) were loaded (45 minutes, 20°C) with a membrane-permeant form of caged IP 3 (ci-IP 3 PM, 1 mM) in HBS with probenecid (2.5 mM) and pluronic F127 (0.02% v/v). Fluo-4 AM (4 mM) was then added and after 45 minutes at 20°C, the medium was replaced with HBS containing only probenecid. After a further 45 minutes, this medium was replaced with HBS. Cells were illuminated with a 488-nm diodebased solid-state laser, and emitted fluorescence (500-550 nm) was captured with an EMCCD camera. Three UV flashes (each ∼1-millisecond duration; ,345 nm, 3000 mF, 300 V, ∼170 J) from a JML-C2 Xe flash-lamp (Rapp OptoElectronic, Hamburg, Germany) allowed photolysis of caged IP 3 (ci-IP 3 ). Responses are reported as F/F 0 , where F 0 and F are fluorescence intensities corrected for background recorded from the same region of interest immediately before (F 0 ) and after stimulation (F).
Measurements of IP 3 and PLC Activity. ASMC in 12-well plates were cultured until confluent. The medium was then supplemented with D-myo-[2-3 H]-inositol (10 mCi/ml) for 48 hours at 37°C. After washing, cells were incubated at 20°C in HBS with LiCl (10 mM) for 5 minutes before stimulation. Reactions were terminated by aspirating medium and adding cold HClO 4 (1 ml, 0.6 M) containing phytic acid (0.2 mg/ml). After 30 minutes, the acid-extract was removed, the cells were scraped into 50 mM Tris at pH 7 (400 ml), and the pooled extract and cells were centrifuged (10,000g, 2 minutes, 4°C). The supernatant was neutralized using K 2 CO 3 (1 M) with EDTA (5 mM). 3 H-inositol phosphates were separated using ion-exchange columns.
Quantitative PCR Analysis. QPCR was performed as described (Tovey et al., 2008) using primers specific for AC subtypes (Ludwig and Seuwen, 2002) and calibrated against expression of the housekeeping gene, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Pantazaka et al., 2013). Human BioBank cDNA pooled from a variety of tissues (Primerdesign, Southampton, UK) was used as a positive control for AC subtypes not expressed in ASMC.
Statistical Analysis. Concentration-effect relationships were individually fitted by nonlinear curve-fitting to Hill equations (GraphPad Prism, La Jolla, CA). The absolute sensitivities and amplitudes of the responses to histamine and PGE 2 varied between patients and with cell passage. Results are, therefore, often presented as normalized values (e.g., as percentages of a maximal response) or as differences between paired comparisons (e.g., DpIC 50 ). Two-tailed paired or unpaired Student's t test, or oneway analysis of variance followed by Bonferroni's test, were used as appropriate.

Results
Cyclic AMP Mediates Inhibition of Histamine-Evoked Ca 21 Signals by PGE 2 . Figure 1A demonstrates that PGE 2 inhibits the Ca 21 signals evoked by histamine in human ASMC. Most experiments were performed at 20°C to minimize loss of the cytosolic Ca 21 indicator. However, in parallel analyses we confirmed that a maximally effective concentration of PGE 2 (10 mM) caused indistinguishable inhibition of the Ca 21 signals evoked by histamine (100 mM) whether the analyses were performed at 20°C (47% 6 12% inhibition, n 5 4) or 37°C (45% 6 6%). In parallel measurements of the effects of PGE 2 on intracellular cAMP and histamine-evoked Ca 21 signals, the cAMP response [negative logarithm of the half-maximally effective concentration (pEC 50 ) 5 6.76 6 0.09, n 5 4) was ∼140-fold less sensitive to PGE 2 than were the Ca 21 signals [negative logarithm of the half-maximally inhibitory concentration (pIC 50 ) 5 8.90 6 0.10, n 5 6] (Fig. 1B). This relationship is consistent with PGE 2 evoking formation of more cAMP than needed to maximally inhibit Ca 21 signals, and with cAMP lying upstream of the inhibition of Ca 21 signaling (Strickland and Loeb, 1981).
Forskolin and its water-soluble analog, NKH 477, directly activate eight of the nine membrane-bound forms of AC (AC1-8) (Seifert et al., 2012). Pretreatment of ASMC with NKH 477 caused a concentration-dependent reduction in both the peak amplitudes of the Ca 21 signals evoked by histamine and their sensitivity to histamine ( Fig. 1C; Table 1). Similar results were obtained with forskolin ( Fig. 1D). Maximally effective concentrations of PGE 2 and NKH 477 caused indistinguishable inhibition of histamine-evoked Ca 21 signals, and their combined maximal effects were not additive (Fig. 1E). These results are consistent with reports showing that forskolin and NKH 477 attenuate the Ca 21 signals evoked by receptors, including H 1 receptors that stimulate PLC in VSM (Yang et al., 1999, and references therein) and other smooth muscles. There has, however, been no prior demonstration that activation of AC inhibits PLC-evoked Ca 21 signals in human ASMC.
High concentrations of 8-Br-cAMP also inhibited histamineevoked Ca 21 signals by reducing the maximal response and the sensitivity to histamine ( Fig. 2A). 8-Br-cAMP had no effect on the Ca 21 content of the intracellular stores (Fig. 2B). The effects of 8-Br-cAMP were mimicked by Sp-cAMPS, which activates PKA and EPACs, and by 6-Bnz-cAMP and 8-CPT-6-Phe-cAMP, which activate PKA but not EPACs ( The negative result with 8-pCPT-29-O-Me-cAMP is important because this analog is more membrane-permeable than 8-Br-cAMP, and it both binds with greater affinity than cAMP to EPACs and more effectively activates them (Gloerich and Bos, 2010). Furthermore, antagonists of EPACs 1 and 2, HJC0197 (Chen et al., 2012), and ESI-09 (Almahariq et al., 2013) (10 mM, 20 minutes) did not prevent the inhibition of histamine-evoked Ca 21 signals by PGE 2 (Fig. 2E). Higher concentrations (50 mM) of either antagonist abolished the Ca 21 signals evoked by histamine (data not shown). We have not explored this effect further, although the antagonists caused similar inhibition of carbachol-evoked Ca 21 signals in human embryonic kidney 293 cells (Meena et al., 2015). Others have also reported nonspecific effects of these EPAC antagonists (Rehmann, 2013).
Maximally effective concentrations of PGE 2 , forskolin, NKH 477, or 8-Br-cAMP similarly attenuated the Ca 21 signals evoked by histamine, and combinations of the treatments were not additive (Fig. 2D). These results establish that inhibition of histamine-evoked Ca 21 signals by PGE 2 is mediated by cAMP (Fig. 2F) and does not require EPACs.
Inhibition of Histamine-Evoked Ca 21 Signals by PGE 2 is Not Mediated by cGMP-Dependent Protein Kinase. Cyclic AMP may directly activate PKG in arterial smooth muscle, although this is contentious [see references in Morgan et al. (2014)]. Cyclic AMP could, however, increase the concentration of cGMP by competing with it for degradation by cyclic nucleotide phosphodiesterases (PDEs). Stimulation of PKG might then attenuate IP 3 -evoked Ca 21 release (Masuda et al., 2010). There is evidence, however, that expression of Inhibition of Ca 21 Signals by cAMP Signaling Junctions proteins involved in PKG signaling in VSM are downregulated in culture (Lincoln et al., 2006, and references therein). 8-Br-cGMP (pIC 50 5 4.50 6 0.29, n 5 5) partially inhibited Ca 21 signals evoked by a submaximal concentration of histamine, but the maximal inhibition was less than half that evoked by PGE 2 or 8-Br-cAMP (Fig. 3, A and B). Furthermore, and in contrast to the effects of 8-Br-cAMP (Fig. 2C), 8-Br-cGMP did not inhibit the Ca 21 signals evoked by a maximal histamine concentration (Fig. 2D). Prolonged incubation with IBMX (20 minutes, 1 mM), a nonselective inhibitor of PDEs, inhibited histamine-evoked Ca 21 signals, but the inhibition (33% 6 3%, n 5 4) was less than that caused by PGE 2 (56% 6 3%) (Fig. 3C). More importantly, a maximal concentration of PGE 2 similarly inhibited histamine-evoked Ca 21 signals in Fig. 1. Inhibition of histamine-evoked Ca 2+ signals by PGE 2 is mediated by cAMP. (A) Ca 2+ signals evoked by histamine (3 mM, black bar) alone or with PGE 2 (10 mM, added 5 minutes before and then during stimulation with histamine). Results show means 6 range from two wells on a single plate; they are typical of results from at least four independent experiments. (B) Effects of PGE 2 on cAMP accumulation (measured after 5 minutes) and inhibition of the peak Ca 2+ signals evoked by histamine (3 mM). Results, as percentages of maximal inhibition (Ca 2+ ) or stimulation (cAMP), are means 6 S.E.M. from six and four experiments, respectively. This panel includes some data for cAMP measurements that were published previously (Pantazaka et al., 2013). (C) Effect of NKH 477 (100 mM) on the peak Ca 2+ signal evoked by histamine. (D) Concentration-dependent effects of forskolin on the peak Ca 2+ signals evoked by histamine (1 mM). (E) Effect of PGE 2 (10 mM), NKH 477 (100 mM), or both on the peak Ca 2+ signals evoked by histamine (as percentages of the maximal response). NKH 477, forskolin, or PGE 2 were added 5 minutes before and then during stimulation with histamine. Results show means 6 S.E.M. from four (C and D) or three (E) independent plates with one to three wells analyzed from each. Ct denotes control.

Inhibition of Ca 21 Signals by cAMP Signaling Junctions
The latter cannot mediate the effects of cAMP on IP 3 -evoked Ca 21 release, nor are EPACs responsible. We therefore assessed the role of PKA.
Immunoblotting with an antiserum that recognizes sequences phosphorylated by PKA showed that maximally effective concentrations of PGE 2 or 8-Br-cAMP stimulated similar levels of protein phosphorylation in ASMC and their effects were nonadditive (Fig. 4A). The phosphorylation was mimicked by 6-Bnz-cAMP but not by the EPAC-selective analog 8-pCPT-29-O-Me-cAMP (Fig. 4A). PGE 2 -evoked protein phosphorylation was attenuated by inhibition of either PKA (with H89) or AC [with 1 mM SQ 22536 with 200 mM DDA (SQ/DDA)] (Fig. 4B).
Maximal concentrations of PGE 2 and 8-Br-cAMP caused phosphorylation of the same proteins (Fig. 4A), but the two stimuli differed in their susceptibility to PKA inhibitors. Rp-8-CPT-cAMPS, an inhibitor of PKA that competes with cAMP by binding to the regulatory subunit of PKA, abolished the phosphorylation evoked by 8-Br-cAMP but only partially inhibited that evoked by PGE 2 (Fig. 4C). Conversely, H89, which inhibits PKA (and other kinases) by competing for the ATP-binding site, abolished the phosphorylation evoked by PGE 2 but caused lesser inhibition of the response to 8-Br-cAMP (Fig. 4C). Similar results were obtained when an antiserum to phospho-VASP was used to assess PKAmediated phosphorylation (Supplemental Fig. S2).
These results suggest that PKA activated by PGE 2 may be exposed to high local concentrations of cAMP, which might then effectively compete with the inhibitor Rp-8-CPT-cAMPS. Conversely, PKA activated by 8-Br-cAMP, which would probably be uniformly distributed within the cell, may be more accessible to ATP than PKA activated by PGE 2 , and so less susceptible to inhibition by H89.
H89 (10 mM) also attenuated the inhibition of histamineevoked Ca 21 signals by 8-Br-cAMP (DpIC 50 5 1.13 6 0.18 n 5 3) and PGE 2 (Fig. 5, A and B). In keeping with our analyses of protein phosphorylation (Fig. 4C), the inhibition of histamine-evoked Ca 21 signals by maximal concentrations of PGE 2 were less effectively inhibited by H89 than were the effects of maximal concentrations of 8-Br-cAMP (compare Fig. 5, A and B).
PKI inhibits PKA by competing with its peptide substrates. We could not achieve effective inhibition of PKA-mediated protein phosphorylation with myristoylated-PKI (10 mM, 20 minutes, data not shown). But using a baculovirus, we achieved expression of PKI in .90% of cells, and this caused 49% 6 6% (n 5 3) inhibition of the VASP phosphorylation evoked by PGE 2 (100 nM) (Supplemental Fig. S3). Expression of an inactive PKI (mut PKI) had no effect on PGE 2 -evoked protein phosphorylation (Supplemental Fig. S3). The effects of H89 and PKI on the inhibition of histamine-evoked  Ca 21 signals by PGE 2 were similar: Each substantially reduced the maximal inhibition without significantly affecting the IC 50 for PGE 2 (Fig. 5, A and C). The effects of H89 on inhibition of histamine-evoked Ca 21 signals by selective agonists of EP 2 (butaprost) and EP 4 (L902,688) receptors were similar to those observed with PGE 2 (Supplemental Fig. S4). We conclude that inhibition of histamine-evoked Ca 21 signals by PGE 2 is mediated by cAMP and requires PKA (Fig. 5D).
PGE 2 Does Not Inhibit Ca 21 Release Evoked by Direct Activation of IP 3 Rs. The rate at which [Ca 21 ] i recovered from the peak Ca 21 signal evoked by histamine was unaffected by PGE 2 (half-times for recovery were 19 6 1 and 17 6 1 seconds, after histamine alone or with PGE 2 , respectively; n 5 11) (Supplemental Fig. S5). This suggests that the attenuated Ca 21 signals do not result from PGE 2 stimulating Ca 21 extrusion from the cytosol.
We used flash-photolysis of ci-IP 3 to activate IP 3 R directly in Fluo-4-loaded ASMC. Single-cell analyses of ASMC established that most cells (99% 6 1%, from 12 fields) responded to histamine (1 mM) with an increase in [Ca 21 ] i , and that two successive challenges with histamine evoked indistinguishable Ca 21 signals (Fig. 6, A and B). PGE 2 reduced the peak amplitude of the Ca 21 signal evoked by a second histamine challenge by 28% 6 4% (n 5 65 cells), without significantly affecting the number of cells that responded (91% 6 8% and 83% 6 7% for control and PGE 2 -treated cells, respectively) ( Fig. 6, C and D). These results confirm that under the conditions used for uncaging ci-IP 3 , PGE 2 inhibits histamineevoked Ca 21 signals.
ASMC loaded with ci-IP 3 responded to UV flashes with rapid increases in Fluo-4 fluorescence (F/F 0 , see Materials and Methods). The amplitudes of these signals were less than those evoked by a maximal concentration of histamine (Fig. 6, E and F), confirming that responses to photolysis of ci-IP 3 were not saturated. Although cells responded similarly to successive histamine challenges (Fig. 6, A and B), the response to a second photolysis of ci-IP 3 was smaller than the first (Fig. 6G), presumably because each stimulus depleted a fraction of the ci-IP 3 . We therefore used two methods to assess the effects of PGE 2 on the Ca 21 signals evoked by photolysis of ci-IP 3 . Cells were either stimulated twice with a UV stimulus, and the amplitude of the second response (with or without PGE 2 ) was compared with the first response for each cell (R2/R1) (Fig. 6, G-I), or cells were stimulated once with UV flashes alone or in the presence of PGE 2 (Fig. 6J). Both analyses concur in demonstrating that PGE 2 has no significant effect on the Ca 21 signals evoked by IP 3 (Fig. 6, G-J). The results with ci-IP 3 therefore demonstrate that PGE 2 does not affect the Inhibition of Ca 21 Signals by cAMP Signaling Junctions interactions of IP 3 with IP 3 R. Furthermore, because the peak IP 3 -evoked Ca 21 signals were unaffected by PGE 2 under conditions where it attenuates responses to histamine (Fig.  6, I and J), the results provide additional evidence that PGE 2 does not stimulate Ca 21 removal from the cytosol.
The product of ci-IP 3 photolysis is an active but modified form of IP 3 (D-2,3-O-isopropylidene-myo-inositol 1,4,5-trisphosphate) (Dakin and Li, 2007) that is not a substrate for IP 3 3-kinase and may differ from IP 3 in its rate of dephosphorylation. Our results do not therefore exclude the possibility that PGE 2 may accelerate degradation of IP 3 . These results suggest that PGE 2 attenuates histamine-evoked Ca 21 signals by inhibiting IP 3 formation, stimulating IP 3 degradation, and/or disrupting IP 3 delivery to IP 3 Rs.
PGE 2 Attenuates Histamine-Evoked Accumulation of IP 3 . Using an assay that reports PLC activity (stimulation after blocking inositol monophosphate degradation by Li 1 ), histamine (1 mM, 30 minutes) stimulated a small accumulation of 3 H-inositol phosphates in ASMC. Although the response was modestly attenuated by PGE 2 (10 mM), the effect was not statistically significant (Fig. 7A). Using an IP 3 R-based bioassay that detects only (1,4,5)IP 3 , histamine stimulated IP 3 accumulation, and the response was attenuated by PGE 2 , although the latter again failed to achieve statistical significance (Fig. 7B). We also attempted to measure histamine-evoked IP 3 formation in single cells using a fluorescence resonance energy transfer (FRET)-based IP 3 sensor (Gulyas et al., 2015), but the signals were too small to resolve reliably any inhibitory effect of PGE 2 . Available genetically encoded IP 3 sensors are known to have limited dynamic range and limited capacity to resolve small changes in intracellular IP 3 concentration (Miyamoto and Mikoshiba, 2017).
We assessed the responses of ASMC to other stimuli (ATP, bradykinin, carbachol, phenylephrine, and thrombin) that might be expected to evoked Ca 21 signals through receptors that stimulate Gq (results not shown). Only thrombin reproducibly evoked substantial increases in [Ca 21 ] i . Thrombin is a protease that cleaves the type 1 protease-activated receptor (PAR1) to unmask an N-terminal ligand. Thrombin and the PAR1 peptide itself evoked concentration-dependent increases in [Ca 21 ] i in ASMC (Fig. 7C). In parallel analyses, PGE 2 (10 mM, 5 minutes) attenuated the Ca 21 signals evoked by histamine without affecting those evoked by PAR1 peptide (Fig. 7D). Although the maximal increase in [Ca 21 ] i evoked by PAR1 peptide was larger than that evoked by histamine, with concentrations of histamine and the PAR1 peptide that evoked comparable increases in [Ca 21 ] i , only the response to histamine was inhibited by PGE 2 (Fig. 7E). After heterologous expression of human muscarinic M3 acetylcholine receptors in ASMC, carbachol evoked a concentration-dependent (pEC 50 5 7.72 6 0.04, n 5 3) increase in [Ca 21 ] i , with a maximal increase (272 6 31 nM, n 5 3) comparable to that evoked by histamine (204 6 13 nM, Fig. 2A). However, the responses to carbachol were unaffected by PGE 2 (Fig. 7F). These results, demonstrating that PGE 2 selectively inhibits the Ca 21 signals evoked by histamine, suggest that the inhibition probably does not arise downstream of PLC.  Methods) for cells stimulated with a UV flash alone or with PGE 2 (10 mM added 5 minutes before flash). *P , 0.05, relative to the response to histamine alone (D, paired Student's t test) or to the UV flash alone (F, unpaired Student's t test).

Inhibition of Ca 21 Signals by cAMP Signaling Junctions
Histamine-Evoked Ca 21 Signals Are Inhibited by Local cAMP Signals. Inhibitors of AC (SQ/DDA) attenuated PGE 2 -evoked cAMP formation (by 79% 6 2%, n 5 4) (Fig. 8A) and protein phosphorylation (Fig. 4B). However, SQ/DDA had no effect on the inhibition of histamine-evoked Ca 21 signals evoked by PGE 2 or butaprost (Fig. 8, B and C). Although cAMP mediates the inhibition of Ca 21 signals by PGE 2 , the response to a maximal concentration of PGE 2 might survive substantial inhibition of AC because it stimulates formation of more cAMP than needed to maximally inhibit Ca 21 signals (Fig. 1B). However, the same argument cannot account for the lack of effect of SQ/DDA on responses to submaximal concentrations of PGE 2 . How might a submaximal response to PGE 2 be unaffected by substantial inhibition of cAMP formation and PKA activity ( Fig. 4B; Fig. 8, A and B)?
A possible explanation is that SQ 22356 and DDA, related inhibitors that bind to the ATP-binding site of AC (Brand et al., 2013), selectively inhibit subtypes of AC distinct from those that mediate the effects of PGE 2 . Available antibodies do not allow quantitative assessment of the expression of AC

542
subtypes, but QPCR analysis shows that human ASMC express similar amounts (∼30%) of AC3, AC7, AC9, some AC6 (∼10%), and detectable AC4 (∼2%) (Supplemental Fig. S6). AC9 probably does not mediate the effects of PGE 2 on Ca 21 signals because AC9 is insensitive to forskolin and NKH 477 (Seifert et al., 2012), which mimic the effects of PGE 2 on Ca 21 signals (Fig. 1, C-E; Fig. 2D). Among the remaining ACs expressed in ASMC, SQ22536 and DDA probably have some selectivity for AC6 over AC3 and AC7 despite some inconsistent reports (Pierre et al., 2009;Seifert et al., 2012). From analyses of individual AC isoforms, maximally effective concentrations of SQ 22356 (and other P-site inhibitors) inhibit catalytic activity by only ∼80% [Brand et al. (2013), but see Onda et al. (2001)]. This is similar to the ∼80% inhibition of PGE 2 -evoked cAMP accumulation by SQ/DDA in ASMC (Fig. 8A), suggesting that the incomplete inhibition observed in ASMC probably does not reflect the unperturbed activity of SQ/DDA-insensitive ACs. Furthermore, the effects histamine-evoked Ca 2+ signals. Activation of EP 2 or EP 4 receptors with butaprost or L902,688, respectively, inhibits histamine-evoked Ca 2+ signals, but the latter evokes formation of less cAMP. We propose, from our results with SQ/DDA, that cAMP is locally delivered to PKA within each signaling junction at a concentration more than sufficient to cause a maximal effect. The concentration-dependent inhibition of Ca 2+ signals by prostanoids is proposed to result from recruitment of all-or-nothing cAMP signaling junctions, rather than from graded increases in activity within individual junctions. EP 2 receptors deliver more cAMP than EP 4 receptors and are therefore more resistant to inhibition of AC by SQ/DDA. Hence inhibition of Ca 2+ signals by EP 4 receptors is partially inhibited by SQ/DDA, whereas the response to EP 2 receptors is insensitive (right). *P , 0.05, paired Student's t-test relative to control.
Inhibition of Ca 21 Signals by cAMP Signaling Junctions of PGE 2 on protein phosphorylation in ASMC are inhibited by SQ/DDA (Fig. 4B), again suggesting that the ACs activated by PGE 2 are inhibited. We conclude that the lack of effect of SQ/DDA on PGE 2 -mediated inhibition of histamine-evoked Ca 21 signals is probably not the result of ineffective inhibition of an SQ/DDA-resistant subtype of AC.
To account for the results with SQ/DDA, we suggest that cAMP is delivered locally to PKA at concentrations more than sufficient to fully inhibit Ca 21 signals. The concentrationdependent effects of PGE 2 might then result from recruitment of these "hyperactive" cAMP signaling junctions, rather than from increased activity within individual junctions (Fig. 8F). This interpretation is consistent with analyses of the effects of SQ/DDA on the inhibition of Ca 21 signals by selective activation of EP 4 receptors. Although activation of EP 2 and EP 4 receptors causes similar maximal inhibition of histamineevoked Ca 21 signals, EP 4 receptors cause less stimulation of AC (Pantazaka et al., 2013). This suggests that EP 4 receptors may less effectively saturate the cAMP signaling junctions. Whereas inhibition of AC with SQ/DDA had no effect on the inhibition of histamine-evoked Ca 21 signals by PGE 2 or butaprost (to selectively activate EP 2 receptors), the sensitivity to L902,688, a selective agonist of EP 4 receptors, was modestly reduced by SQ/DDA (DpIC 50 5 0.32 6 0.10, n 5 5) (Fig. 8, B-E). This observation supports our suggestion that the subtype(s) of AC that link prostanoid receptors to inhibition of Ca 21 signals are sensitive to SQ/DDA. Furthermore, these results are consistent with the scheme shown in Fig. 8F, where we suggest that cAMP is locally delivered within "hyperactive" signaling junctions at concentrations more than sufficient to maximally activate the PKA that inhibits Ca 21 signals.
We considered whether AKAPs, which are widely implicated in assembling PKA with its regulators and effectors (Smith et al., 2017), might contribute to organization of the cAMP signaling through PKA that leads to inhibition of histamine-evoked Ca 21 signals. A membrane-permeant peptide that disrupts association of AKAPs with PKA (st-Ht31) but not its inactive analog (st-Ht31P), significantly attenuated the protein phosphorylation evoked by PGE 2 , but neither peptide affected the concentration-dependent inhibition of histamine-evoked Ca 21 signals by PGE 2 (Supplemental Fig.  S7). These results suggest that AKAPs are probably not important components of the signaling pathway from PGE 2 to inhibition of Ca 21 signals.

Discussion
In human ASMC, the IP 3 -mediated Ca 21 signals evoked by activation of H 1 histamine receptors are attenuated by PGE 2 . Several lines of evidence show that this inhibition is mediated by cAMP. The concentration-effect relationships for regulation of AC and Ca 21 signals by PGE 2 are consistent with cAMP lying upstream of Ca 21 in the signaling pathway (Fig. 1B), direct activation of AC or membrane-permeant analogs of cAMP mimic PGE 2 , and maximal concentrations of these drugs are not additive (Figs. 1 and 2). Our conclusion that cAMP mediates the inhibition of Ca 21 signals in human ASMC is consistent with evidence that many receptors, via stimulation of AC, attenuate Ca 21 signaling in smooth muscle, including VSM (Morgado et al., 2012). Inhibition of histamineevoked Ca 21 signals by PGE 2 does not require activation of EPACs (Fig. 2, D and E). The inhibition is not mediated by accumulation of cGMP after inhibition of PDEs since neither cGMP nor inhibition of PDEs effectively mimicked PGE 2 (Fig.  3). Inhibition of histamine-evoked Ca 21 signals by PGE 2 or 8-Br-cAMP was attenuated by inhibition of PKA using H89, PKI, or Rp-8-CPT-cAMPS (Fig. 4). We conclude that inhibition of histamine-evoked Ca 21 signals by PGE 2 is (at least largely) mediated by PKA (Fig. 5D).
PKA can enhance Ca 21 removal from the cytosol by stimulating Ca 21 pumps (Tada and Toyofuku, 1998) or the Na 1 /Ca 21 exchanger (Karashima et al., 2007). However, accelerated removal of cytosolic Ca 21 does not mediate inhibition of histamine-evoked Ca 21 signals by PGE 2 in human ASMC (Supplemental Fig. S5). Nor would this mechanism be consistent with the lack of effect of PGE 2 on the Ca 21 signals evoked by stimulation of endogenous PAR1 or heterologously expressed M3 muscarinic receptors (Fig. 7, D-F). Cyclic AMP has been proposed to inhibit IP 3 -evoked Ca 21 release (Bai and Sanderson, 2006), but PKA (IP 3 R1 and IP 3 R2) and cAMP (IP 3 R1-3) more often potentiate responses to IP 3 (Taylor, 2017). However, under conditions where PGE 2 inhibited histamine-evoked Ca 21 signals, it had no effect on the sensitivity of IP 3 Rs to IP 3 (Fig. 6). Steps linking receptors to PLC can also be inhibited by cAMP (see references in Yang et al. (1999)). Although two different assays suggested that PGE 2 attenuated histamine-evoked PLC activity in human ASMC, neither analysis demonstrated a statistically significant effect (Fig. 7, A and B). However, the lack of effect of PGE 2 on the Ca 21 signals evoked by PAR1 and muscarinic M3 receptors (Fig. 7, D-F) suggests that the inhibition of histamine-evoked Ca 21 signals by cAMP/PKA is probably the result of uncoupling of H 1 histamine receptors from G q/11 . PKA has been reported to phosphorylate H 1 histamine receptors (Kawakami et al., 2003;Horio et al., 2004), but the functional consequences have not been thoroughly examined (Miyoshi et al., 2006). We conclude that in human ASMC, PGE 2 , through EP 2 and EP 4 receptors (Pantazaka et al., 2013), stimulates AC, leading to formation of cAMP and uncoupling of histamine from stimulation of PLC, most probably by PKA-mediated phosphorylation of H 1 receptors.
Cyclic AMP can be locally delivered to intracellular targets (Zaccolo, 2011;Cooper and Tabbasum, 2014). AKAPs play prominent roles in targeting cAMP through PKA to specific cellular responses (Smith et al., 2017), but our results suggest that AKAPs probably do not contribute to inhibition of histamine-evoked Ca 21 signals by PGE 2 (Supplemental Fig. S7). Our results do, however, reveal an additional complexity in the pathways linking PGE 2 to inhibition of histamine-evoked Ca 21 signals. Although cAMP mediates this inhibition, the concentration-dependent effects of PGE 2 were insensitive to substantial inhibition of AC (Fig. 8). These results and analyses of the effects of selective activation of EP 2 and EP 4 receptors lead to the scheme shown in Fig. 8F. We suggest that communication between EP receptors and the PKA that inhibits histamine-evoked IP 3 formation is mediated by delivery of cAMP within signaling junctions. Activation of a junction allows local delivery of a supersaturating concentration of cAMP to PKA, allowing each junction to function as a robust on-off switch. We suggest that the concentration-dependent effects of PGE 2 arise from recruitment of these junctions and not from graded activity within individual junctions. Such digital signaling from receptors to intracellular targets via hyperactive junctions (Fig. 8F) allows robust and reliable communication, and may be a general feature of signaling by diffusible intracellular messengers (Tovey et al., 2008).