Termination of cyclic adenosine monophosphate (cAMP) signaling via the extracellular Ca2+-sensing receptor (CaR) was visualized in single CaR-expressing human embryonic kidney (HEK) 293 cells using ratiometric fluorescence resonance energy transfer–dependent cAMP sensors based on protein kinase A and Epac. Stimulation of CaR rapidly reversed or prevented agonist-stimulated elevation of cAMP through a dual mechanism involving pertussis toxin–sensitive Gαi and the CaR-stimulated increase in intracellular [Ca2+]. In parallel measurements with fura-2, CaR activation elicited robust Ca2+ oscillations that increased in frequency in the presence of cAMP, eventually fusing into a sustained plateau. Considering the Ca2+ sensitivity of cAMP accumulation in these cells, lack of oscillations in [cAMP] during the initial phases of CaR stimulation was puzzling. Additional experiments showed that low-frequency, long-duration Ca2+ oscillations generated a dynamic staircase pattern in [cAMP], whereas higher frequency spiking had no effect. Our data suggest that the cAMP machinery in HEK cells acts as a low-pass filter disregarding the relatively rapid Ca2+ spiking stimulated by Ca2+-mobilizing agonists under physiological conditions.
cAMP controls crucial physiological cell functions such as cell growth, differentiation, transcriptional regulation, and apoptosis. Levels of intracellular cAMP principally reflect a balance between synthesis of the second messenger by adenylyl cyclases (ACs; activated through G protein–coupled receptors [GPCRs] coupled to Gαs), degradation by phosphodiesterases (PDEs), and the inhibitory action of pertussis toxin (PTX)–sensitive Gαi-coupled receptors that serve to limit cAMP formation by AC. The classical intracellular effector of the cAMP signal is protein kinase A (PKA), a holotetrameric complex that consists of two regulatory and two catalytic subunits that dissociate upon cAMP binding. More recently, a second class of intracellular cAMP targets has been identified: Epacs (exchange proteins activated by cAMP) are monomeric proteins that undergo significant conformational changes upon cAMP binding, allowing them to activate their target, Rap1 (Beavo and Brunton, 2002). Aberrations in cAMP signaling or inappropriate cAMP production can have serious pathological consequences (e.g., tumor formation and heart failure). Therefore, understanding how cAMP signals are terminated is just as important as understanding how they are generated in the first place.
In addition to inhibition through the classical PTX-sensitive Gαi, intracellular Ca2+ signaling pathways can exert powerful modulatory actions on cAMP accumulation (for review see Bruce et al., 2003). For example, some members of the extensive superfamily of PDEs (particularly those belonging to the PDE1 family) are activated by elevated intracellular Ca2+ (Houslay and Milligan, 1997; Goraya et al., 2004). In addition, specific isoforms of AC have been shown to respond to physiological changes in intracellular Ca2+ with either activation or inhibition of enzymatic activity (de Jesus Ferreira et al., 1998; Chabardes et al., 1999; Cooper, 2003). Ca2+ entering the cell via store-operated channels preferentially regulates cAMP production by AC and has been proposed to be much more effective than Ca2+ released from intracellular stores or influx via other types of Ca2+ channels (Cooper, 2003).
Conversely, the cAMP pathway can influence Ca2+ signaling at many levels. For example, PKA-dependent phosphorylation of intracellular release channels, such as the inositol 1,4,5-trisphosphate (InsP3) receptor, and Ca2+ extrusion mechanisms, such as the plasma membrane Ca2+ ATPase, can powerfully shape Ca2+ signals (for review see Bruce et al., 2003). Reciprocal modulation by cAMP and Ca2+ pathways will be expected to generate unique patterns of signaling molecules during concurrent activation of receptors linked to each of these signal transduction cascades.
Agonist-induced oscillations in intracellular ([Ca2+]i) are a well-described phenomenon. Based on the fact that intracellular Ca2+ can augment or inhibit cAMP accumulation, it is predicted that cAMP levels would fluctuate during oscillatory Ca2+ spiking in a cell type in which Ca2+-dependent ACs or PDEs were expressed (Cooper et al., 1995). Repetitive activation–inactivation cycles of the cAMP signaling pathway could conceivably encode information that differed from large static changes in the second messenger. Precisely this sort of regulation has been shown to be operative for Ca2+ in cells displaying oscillatory Ca2+ signaling events. The frequency, amplitude, and duration of Ca2+ spiking are known to differentially regulate key cellular functions such as gene transcription and the activation of plasma membrane ion channels (Thorn et al., 1993; Dolmetsch et al., 1998; Berridge et al., 2000).
In the present study we examined the interactions of the extracellular Ca2+-sensing receptor (CaR) with the cAMP signal transduction cascade. This widely expressed dimeric GPCR was first cloned more than a decade ago from bovine parathyroid cells (Brown et al., 1993). Since that time, expression of CaR has been demonstrated in several epithelial, glial, and neuronal cell types. CaR is able to sense small fluctuations in the extracellular [Ca2+] within the physiological range (Brown and MacLeod, 2001). Interestingly, this receptor is not strictly a calcium sensor, as it can be stimulated by a variety of physiologically relevant divalent and polycationic compounds. These compounds include Mg2+; amino acids; and polyamines, such as spermine and spermidine. CaR can also be activated through allosteric modulation by synthetic small molecule “calcimimetics,” such as NPS-R-467 (Conigrave et al., 2000; Hofer and Brown, 2003; Breitwieser et al., 2004; Nemeth, 2004).
In CaR-transfected human embryonic kidney (HEK) 293 cells (HEK CaR) and in several other cell models in which the receptor is expressed endogenously, CaR stimulation leads to robust oscillations in [Ca2+]i (Breitwieser and Gama, 2001; Young and Rozengurt, 2002; De Luisi and Hofer, 2003). In addition to this coupling to the phosphoinositide–InsP3–Ca2+ signaling pathway via Gαq/11, CaR has also been shown to inhibit cAMP production (as measured using standard biochemical techniques) through interactions with PTX-sensitive proteins, presumably Gαi (Chen et al., 1989; Chang et al., 1998). In the present study we used recently introduced fluorescence resonance energy transfer (FRET)–based fluorescent indicators for cAMP that allowed us to visualize the inhibitory actions of CaR activation on cAMP levels in single living HEK CaR cells in real time (Zaccolo and Pozzan, 2002; Ponsioen et al., 2004). Our data confirm that CaR interacts with PTX-sensitive Gαi to inhibit AC in cells stimulated with cAMP-generating agonists (prostaglandin E2 [PGE2], vasoactive intestinal peptide [VIP], and isoproterenol).
Pharmacological and Western blot evidence suggest that HEK 293 cells express principally Ca2+-insensitive isoforms of PDE (specifically, PDE4D3 and PDE4D5; Hoffmann et al., 1999; Rich et al., 2001b). These cells are also known to express multiple isoforms of AC (AC1, 3, 5, 6, 7, and 9 and soluble, bicarbonate-sensitive AC; Wayman et al., 1995; Ludwig and Seuwen, 2002; Geng et al., 2005), of which some are Ca2+ inhibitable (e.g., AC5 and AC6) and others are activated by Ca2+ or Ca2+/calmodulin (e.g., AC1; soluble, bicarbonate-sensitive AC; and possibly AC3). Because these enzyme subtypes may have antagonistic actions on cAMP accumulation, it is difficult to predict the effects of intracellular Ca2+ elevations on cAMP signaling. In this study we found that [Ca2+]i increases had a substantial inhibitory effect on cAMP levels, suggesting the prevailing action to be on Ca2+-inhibitable AC. We also found that in the presence of elevated cAMP, agonist-stimulated Ca2+ signaling is further augmented, providing additional inhibitory feedback on cAMP generation. These factors conspire to make CaR a particularly potent antagonist of cAMP generation.
Because CaR agonists reliably elicit robust oscillations in intracellular Ca2+ in HEK CaR cells, we reasoned that [cAMP] might also oscillate during CaR activation. However, although both the amplitude and frequency of Ca2+ oscillations, as measured by fura-2, were initially markedly enhanced by cAMP, these Ca2+ spikes were observed to eventually fuse into a sustained plateau of elevated [Ca2+]. At the same time, experiments using the PKA probe and a system of artificially generated Ca2+ pulses revealed that dissociation of PKA (the major intracellular target of the cAMP signal) is indifferent to patterns of rapid Ca2+ spiking, such as those elicited by an agonist in the absence of cAMP. In contrast, in the face of low-frequency, long-duration Ca2+ oscillations, a complex, descending staircase pattern of cAMP (as measured by PKA dissociation) was unmasked. These results show that the cAMP signaling machinery in HEK cells can discriminate between different patterns of Ca2+ oscillations and is, in principle, immune to the rapid spiking stimulated physiologically by Ca2+-mobilizing agonists.
Characterization of the cAMP probe in HEK 293 CaR cells
Several different types of fluorescent cAMP sensors have been described previously (Adams et al., 1991; Zaccolo et al., 2000; Zaccolo and Pozzan, 2002; DiPilato et al., 2004; Mongillo et al., 2004; Ponsioen et al., 2004; Landa et al., 2005). Here we used recently developed FRET sensors based on PKA (Zaccolo and Pozzan, 2002) and Epac (Ponsioen et al., 2004) to continuously monitor cAMP levels for extended periods of time in single HEK CaR cells. Fig. 1 A shows that stimulation of endogenous β-adrenoceptors using 100 nM isoproterenol produced rapid and reversible elevation in the FRET ratio. Repeated stimulation of prostanoid receptors (EP2/EP4) with 100 nM of the inflammatory mediator PGE2 yielded largely reproducible responses in the same cell. This agonist dose generally caused ratio responses that were smaller than the saturating activation elicited by a supramaximal dose (100 μM) of forskolin. This stimulation protocol was therefore used for much of the remainder of the study (typical of results from 32 cells in five experiments). Fig. 1 B shows pseudocolor images of the 480/535 nm FRET emission ratio of the PKA-based sensor, corresponding to the plot shown in Fig. 1 A. CFP- and YFP-labeled PKA subunits were distributed throughout the cytoplasm in transfected cells but were excluded from the nuclear compartment (not depicted). Significantly, because of the high sensitivity of these FRET-based techniques, we were able to detect ratio changes in response to low nanomolar concentrations of cAMP-generating agonists (e.g., 6.25 nM VIP or 10 nM isoproterenol; not depicted). One drawback, in fact, of the high sensitivity of these probes is that it is possible to saturate the indicators when GFP-tagged PKA subunits are dissociated to the maximal extent, even though intracellular cAMP continues to rise. This issue is discussed further on the following page. We also used a second type of cAMP probe based on Epac that exhibits a faster response time (Ponsioen et al., 2004). As illustrated in Fig. 1 C, reversible ratio changes after stimulation with 5 nM PGE2 were readily discriminated in HEK CaR cells using this sensor.
Role of CaR in terminating cAMP signaling
We next took advantage of the extraordinary sensitivity of these cAMP indicators to examine how activation of CaR influenced cAMP signaling. Fig. 2 A shows that the elevation of the emission ratio induced by 100 nM PGE2 was completely abolished by brief preexposure to NPS-R-467, a specific synthetic allosteric activator of CaR (n = 38 cells in six experiments; Nemeth, 2004). This calcimimetic by itself did not have any effect on cAMP levels, as measured by FRET. The data in Fig. 2 B further support this result. In the presence of the CaR agonist spermine, an extremely high dose (3 μM) of PGE2 produced only a very small transient increase in the ratio, followed by a small undershoot, showing that CaR is extremely effective in inhibiting cAMP production. The cAMP signal was partially recovered after spermine washout. Prestimulation of CaR with spermine or 3 mM Ca2+ was able to completely abolish the ratio increase when a lower dose (100 nM) of PGE2 was applied (Fig. 2 C; n = 16/19 cells in four experiments for spermine, and n = 15/15 cells in three experiments for 3 mM Ca2+). Thus, preexposure to a variety of CaR agonists effectively prevented the increase in cAMP production.
CaR stimulation was also able to reverse cAMP generation in cells previously stimulated with PGE2 or other agonists. Spermine addition at the peak of the PGE2-induced cAMP elevation caused a reversal of the ratio, as measured with this PKA-based probe (Fig. 2 D; n = 69 cells in eight experiments). However, we noted that in some experiments using this protocol, the inhibitory response to spermine addition occurred with some delay. We were concerned that we might be missing important dynamic information as the peak of the ratio change reported by this very sensitive probe approached saturation. We therefore performed additional experiments (Fig. 2 E) using a version of the cAMP probe (“R230K”) in which the RII regulatory subunit had been engineered with a reduced affinity for cAMP (Mongillo et al., 2004). The low-affinity reporter had an estimated EC50 value for cAMP-dependent dissociation of PKA of ∼31 μM (vs. submicromolar dissociation for the wild-type [WT] probe; Mongillo et al., 2004). Fig. 2 D also shows that the profile of cAMP reduction during CaR activation was a smooth monotonic function (typical of 13 cells in four experiments; control trace typical of 23 cells in five experiments).
In experiments similar to those in Fig. 2 D, we compared the acute action of the Ca2+-mobilizing agonist carbachol on the reversal of the 480/535 emission ratio during PGE2-stimulated cAMP production, as measured with the R230K low-affinity probe (Fig. 2 F; n = 29 cells in four experiments). The inhibitory action of carbachol was significantly less (only ∼35%) than that elicited by spermine (normalized rate of decline –0.006 ± 0.001 SD for carbachol vs. –0.017 ± 0.001 SD during spermine; P < 0.000001) and occurred with a slight delay. Carbachol was also unable to completely prevent the 480/535 emission ratio increase during PGE2 stimulation in experiments similar to those shown in Fig. 2 (A–C; and not depicted; n = 13 cells in two experiments). Because HEK 293 cells apparently do not express Gαi-coupled muscarinic M2 receptors (Stope et al., 2003), we thought that the increase in intracellular Ca2+ elicited by carbachol stimulation might account for the modest inhibitory action seen in these experiments.
Reciprocal modulation of cAMP and Ca2+ pathways
Parallel experiments were performed with fura-2–loaded HEK CaR cells to examine whether PGE2 had any effect on the pattern of intracellular Ca2+ signaling. CaR stimulation elicits robust and long-lasting oscillations in [Ca2+]i (Breitwieser and Gama, 2001; Young and Rozengurt, 2002; De Luisi and Hofer, 2003). As shown in Fig. 3 A, stimulation with spermine alone gave rise to Ca2+ oscillations with a frequency of 1.78 ± 0.03 per minute (n = 122 cells in four experiments). This frequency was dramatically increased after acute PGE2 addition (2.27 ± 0.06; P < 0.0001), until the spikes fused into an elevated plateau after ∼2 min (1.85 ± 0.08 min). A third exposure to spermine after PGE2 washout elicited oscillations that were faster than the control (2.41 ± 0.05; P < 0.0001, compared with the first stimulation), indicating that the effect of PGE2 was somewhat prolonged. The amplitude of the spikes was also significantly affected by PGE2, with the overall effect being an increased delivery of Ca2+ to the cytoplasm. As shown in Fig. 3 B, a 3-min preexposure to PGE2 (when [cAMP] is typically near its maximum; see Fig. 2 D) dramatically changed the profile of the spermine response. We observed a few initial rapid spikes, which eventually fused into a prolonged plateau after 2.2 ± 0.13 min (n = 118 cells in four experiments). A third challenge with spermine yielded a persistent effect on oscillation frequency compared with the first control stimulation (2.03 ± 0.03 spikes per minute in control vs. 2.66 ± 0.04 for third stimulation; P < 0.0001). Similar results were obtained when carbachol was used in place of spermine (not depicted; n = 90 cells in four experiments). Although the carbachol-stimulated Ca2+ oscillations under control conditions were typically much less robust than those observed during spermine stimulation, the pattern of initial increase in oscillation frequency that fused into a sustained plateau was still dependably observed. Fig. 3 C shows that the frequency of oscillations recorded during three sequential control stimulations with spermine were very consistent (n = 87 cells in four experiments; no statistical difference). The data on oscillation frequency are summarized in Fig. 3 D.
Results very similar to those shown in Fig. 3 (A and B) were obtained when cAMP was elevated by a variety of means using 100 nM VIP (2.13 ± 0.05 oscillations per minute in control vs. 2.41 ± 0.08 with VIP; n = 51 cells; P < 0.001), 100 nM isoproterenol (2.22 ± 0.05 oscillations per minute in control vs. 2.63 ± 0.08 with isoproterenol; n = 63 cells; P < 0.0001), or 100 μM forskolin (2.34 ± 0.07 oscillations per minute in control vs. 2.70 ± 0.10 with forskolin after third stimulation; n = 60 cells; P < 0.003). Experiments with the membrane-permeable cAMP analogue Sp-8-Br-cAMPS suggest that cAMP elevation by itself and not other elements of the signal transduction cascade proximal to cAMP production were sufficient for the enhancement of Ca2+ signaling. 200 μM of the “potent” cAMP analogue Sp-8-Br-cAMPS caused the cAMP/FRET ratio to increase, albeit very slowly compared with native cAMP-generating agonists (not depicted; n = 31 cells in five experiments). Repeating the same maneuver in experiments with fura-2–loaded cells showed that acute addition of Sp-8-Br-cAMPS during Ca2+ spiking was able to reproduce the effect induced by PGE2 shown in Fig. 3 (1.94 ± 0.02 spikes per minute in control vs. 2.69 ± 0.03 in the presence of cAMP analogue; n = 135 cells in four experiments; P < 0.00001). Similar effects were achieved when cells were treated with Sp-8-Br-cAMPS before spermine stimulation (not depicted; 2.16 ± 0.04 spikes per minute in control vs. 2.73 ± 0.04 in the presence of cAMP analogue; n = 102 cells in four experiments; P < 0.000001).
We did not examine in detail the mechanisms accounting for the ability of cAMP to potentiate spermine or carbachol-stimulated Ca2+ signaling in HEK CaR cells. This phenomenon may derive from increased excitability of the InsP3 receptor because of PKA phosphorylation (for review see Bruce et al., 2003), as described previously in parotid acinar cells (Bruce et al., 2002). However, as shown in the subsequent sections, the conversion to a pattern of sustained Ca2+ delivery in the presence of cAMP has important implications for the regulatory feedback of Ca2+ on cAMP signaling.
Elevation of intracellular Ca2+ affects cAMP production
In the parathyroid gland and some other cell models, CaR is a pleiotropic receptor that can simultaneously increase [Ca2+]i and inhibit cAMP formation via Gαi. The data of Fig. 2 F provided some initial hints that intracellular Ca2+ may influence cAMP levels in HEK cells. We next attempted to dissect the relative contributions of Ca2+ signaling and putative Gαi-dependent actions on CaR-mediated inhibition of cAMP production in HEK 293 cells.
We first examined the actions of a large, single, “artificial” pulse of [Ca2+]i on cAMP formation in HEK 293 WT cells, which do not express CaR. Before the start of the experiment, WT cells were treated with the irreversible sarco/endoplasmic reticulum Ca2+-ATPase pump inhibitor thapsigargin to deplete intracellular Ca2+ stores. Cells were initially maintained in Ca2+-free solution. Under these conditions, it is expected that readmission of Ca2+ to the bath will provoke a large entry of Ca2+ into the cell because of influx through capacitative Ca2+ entry pathways (Berridge et al., 2000). Parallel experiments with fura-2 showed that addition of 5 mM Ca2+ resulted in a large increase in [Ca2+]i (Fig. 4 A, right), comparable in amplitude with the initial peak of the agonist-evoked Ca2+ spike (n = 110 cells in four experiments). In measurements using the PKA-FRET indicator under the same conditions, 5 mM Ca2+ administered during the peak of PGE2-stimulated cAMP production (100 nM PGE2) completely and reversibly decreased the ratio increase (Fig. 4 A; n = 28 cells in six experiments), although with a small delay.
We found similar results when we used a different experimental protocol to generate a persistent increase in [Ca2+]i in HEK 293 CaR cells. 10 μM ionomycin, a Ca2+ ionophore that releases internal Ca2+ stores and promotes Ca2+ entry from the extracellular space (as well as activates store-operated channels via store depletion), was able to completely abolish both PGE2- and forskolin-induced cAMP elevations, but only in the presence of external Ca2+ (Fig. 4 B, left; n = 28 cells in four experiments). As shown in Fig. 4 B (right), however, the responses to both PGE2 and forskolin were evident during ionomycin treatment in the absence of external Ca2+ (n = 28 cells in four experiments). Under these conditions, ionomycin is expected to yield only a transient (∼2-min duration) increase in intracellular Ca2+ because of store release. These experiments confirm that intracellular Ca2+ was indeed able to effectively modulate cAMP levels in HEK 293 cells and are in agreement with previous reports that Ca2+ entering the cell via store-operated channels preferentially regulates cAMP production by AC (Cooper, 2003). This explanation is also consistent with the data shown in Fig. 2 F, in which acute stimulation with the Ca2+-mobilizing agonist carbachol inhibited the cAMP signal, although this inhibition was diminished significantly compared with the potent actions of spermine.
The ability of CaR to block cAMP production is PTX sensitive
We used the specific blocker of Gαi, PTX, to determine whether CaR expressed in HEK cells could also inhibit cAMP production through this pathway. Previous attempts to characterize this aspect of CaR signaling in HEK CaR cells using conventional radioimmune assays for cAMP have been only partially successful (Chang et al., 1998). During these experiments we eliminated the Ca2+ signaling component by pretreating the cells with thapsigargin in Ca2+-free solutions. Control experiments (Fig. 5 A) demonstrate that store-depleted HEK CaR cells maintained in Ca2+-free solutions were still able to repeatedly respond to PGE2 and forskolin (n = 37/39 cells in six experiments). Somewhat surprisingly, this response was entirely abolished by spermine (Fig. 5 B; n = 28/29 cells in four experiments). This implies that activation of CaR is capable of regulating cAMP signaling through a pathway entirely independent of Ca2+. However, the response to PGE2 was recovered in the presence of spermine when Gαi, the link between CaR and AC inactivation, was inhibited by PTX (n = 19 cells in five experiments; Fig. 5 C).
The preceding results demonstrate that CaR exerts a dual action on cAMP signaling, working through intracellular Ca2+ and PTX-sensitive pathways. Based on the strong oscillatory pattern of the Ca2+ signal (Fig. 3, A–C), we wondered if conditions existed under which cAMP would also display a complex oscillatory or descending “staircase” profile. Such cAMP oscillations might arise from repetitive coupling of CaR to AC via Gαi, or as a secondary consequence of intracellular Ca2+ spiking.
Fig. 6 A shows that in the absence of Gαi-mediated inhibition (PTX pretreatment), spermine still caused a smooth monotonic reduction in the FRET ratio (compare with control in Fig. 2 E), and this occurred with a substantial delay (n = 20 cells in four experiments). We expected to see fluctuations or stepwise decrements in cAMP because parallel experiments in fura-2–loaded cells revealed initial persistence of Ca2+ oscillations (similar to those observed in Fig. 3; not depicted) after PTX treatment, and the data of Fig. 5 showed cAMP accumulation to be very sensitive to intracellular Ca2+. We considered the possibility that the PKA-based sensors we were using had not been fast enough to resolve rapid fluctuations in cAMP because a relatively complex binding reaction involving four cAMP molecules and the dissociation of four PKA subunits is required to see a change in FRET. However, a similar profile was observed in experiments using the fast monomeric Epac-based sensor described on the previous page. As shown in Fig. 6 B, spermine addition during stimulation of cells with a low dose (5 nM) of PGE2 caused a smooth reversible decline in the FRET ratio. Fig. 6 C illustrates the powerful inhibitory effect of the PTX-sensitive component of CaR stimulation on cAMP signaling. After PTX pretreatment, the action of spermine-induced Ca2+ signals can be examined in isolation. A smooth, slow decline in the FRET ratio that occurred with a notable delay was observed upon spermine addition. Concurrent measurements of the 340:380 nm fura-2 excitation ratio and 480/535 nm FRET emission ratio in fura-2–loaded HEK CaR cells expressing the Epac sensor confirmed that the presence of Epac did not alter Ca2+ oscillations (Fig. 6 D). However, we noted significant “bleed through” of the fura-2 signal into the FRET channels (see Materials and methods for details), giving rise to apparent oscillations in the FRET ratio that were never observed in cells not loaded with fura-2 (n = 9 cells in five experiments).
In numerous experiments using the protocols shown in Fig. 2 (A–F; on hundreds of individual cells) using both low-affinity WT PKA-based indicators and varying doses and combinations of receptor agonists, we observed only a regular, uniform decline in the ratio after CaR activation. The question remains, why are fluctuations in the FRET ratio, driven by intracellular Ca2+ spikes, not evident in these experiments? The following data show that the frequency and duration of Ca2+ spiking are critical determinants of whether the inhibitory actions of intracellular Ca2+ are translated to the cAMP signaling pathway.
We generated repetitive artificial Ca2+ spikes in HEK WT cells using a pattern that mimicked spermine-stimulated Ca2+ oscillations in HEK CaR cells. WT cells were pretreated with thapsigargin in Ca2+-free solutions, and intermittent pulses of Ca2+ (1 mM) were applied during PGE2-stimulated cAMP accumulation. As shown in Fig. 7 A, with a frequency of one Ca2+ spike (lasting 1 min) every 2 min, no effect could be discerned on the FRET ratio (n = 8 cells in three experiments). Fig. 7 B shows that the low-affinity PKA-based indicator should, in principle, be able to respond to alterations in [cAMP] on the minute time scale. In these experiments, cells were permeabilized with digitonin in an intracellular-like buffer. Cells were then alternately superfused for 1-min pulses with solutions containing 30 or 50 μM cAMP (n = 9 cells in four experiments). These seemingly high concentrations of cAMP were used because in situ calibration of the low-affinity probe revealed this to be the range of [cAMP] levels achieved during stimulation with 500 nM PGE2 (not depicted; n = 12 cells in five experiments). In contrast to the results of Fig. 7 A, when the frequency and duration of the Ca2+ pulses were altered (one 3-min-long spike every 6 min), distinctive stepwise reductions in Ca2+ were evident (Fig. 7 C; n = 22 cells in five experiments). It was consistently observed that >1 min was required after admission of Ca2+ in the superfusion before the FRET ratio began to decline (during the first Ca2+ pulse, the ratio started to decrease only after 1.63 ± 0.60 min, for the second Ca2+ pulse, after 1.40 ± 0.66 min, and for the third pulse, after 1.22 ± 0.70 min; n = 22 cells in five experiments), explaining why the pattern of rapid Ca2+ spiking in Fig. 7 A failed to elicit any effect. These results indicate that in this model system the cAMP machinery is immune to rapid Ca2+ spiking but does respond to large static increases in Ca2+ (as sensed by PKA and Epac), which sets cAMP at a new, reduced level.
In vivo, cells are bombarded with a variety of extracellular first messengers that may have complementary or antagonistic actions. These stimuli must be integrated dependably at the level of intracellular second messenger cascades if the appropriate biological endpoint is to be achieved. In addition, although there are many specific cell-surface receptors, there are relatively few second messengers. An emerging concept in the field of signal transduction is the existence of signaling networks that exploit complex spatiotemporal patterns of second messenger molecules, allowing encoding of an expanded repertoire of messages using limited numbers of messengers (Zaccolo and Pozzan, 2003).
In this study we examined the interactions of the widely expressed GPCR, CaR, with other receptors linked to cAMP generation. Our data show that in HEK cells, CaR is exceptionally competent in terminating cAMP signaling initiated by the inflammatory mediator PGE2 (Figs. 2 and 6) as well as cAMP signals generated by isoproterenol and VIP (not depicted). The reasons for this potent action are threefold. First is the somewhat predictable inhibition through the classical PTX-sensitive Gαi pathway that we demonstrated here to be very effective by itself in HEK CaR cells (Figs. 5 and 6). Although it was known from prior biochemical studies that CaR is able to interact with Gαi in several cell types, the use of sensitive FRET-based imaging techniques allowed us to examine this interaction with unprecedented spatiotemporal detail in single living cells. The second reason CaR stimulation is so efficient in preventing or reversing cAMP signaling is that it activates intracellular Ca2+ signaling cascades. Our data demonstrate that cAMP accumulation was sensitive to [Ca2+]i in HEK cells (Fig. 4), as has been described previously for other cell models. Under control conditions, these Ca2+ signals are manifested as oscillations, which persist with very little desensitization as long as CaR agonists are present. However, in the presence of cAMP, this oscillatory pattern of Ca2+ signaling was ultimately converted to a larger, sustained Ca2+ elevation that was shown here to be much more effective in modulating cAMP levels (Figs. 3 and 4). This enhancement of the Ca2+ signal by cAMP constitutes the third mechanism that serves to reinforce the prompt termination of the cAMP elevation in HEK CaR cells.
CaR and other receptors that are able to interact through both Gαi and Ca2+ may be useful targets for pharmacological control of cAMP signaling in several tissues. Of course, it must be kept in mind that the action of Ca2+ on cAMP signaling will depend strongly on the particular repertoire of PDE and AC isoforms expressed in a given cell type and whether they are activated, unaffected, or inhibited by Ca2+ (Cooper et al., 1995; for review see Bruce et al., 2003). Landa et al. (2005) recently provided evidence for rapid Ca2+-dependent augmentation of cAMP signaling using Epac-based cAMP sensors in insulin-secreting MIN6 β cells, a cell type that expresses a very different panel of ACs and PDEs. de Jesus Ferreira et al. (1998) described coexpression of Ca2+-inhibitable (type 6) AC and CaR in the cortical thick ascending limb of the kidney and further showed potent antagonism of cAMP signaling during CaR stimulation by physiological levels of extracellular [Ca2+]. CaR is also known to be concentrated in neuronal synapses, where predictions from modelling and direct measurements have shown significant depletions in extracellular [Ca2+] during neuronal activity because of entry though voltage-operated Ca2+ channels (Egelman and Montague, 1999; Cohen and Fields, 2004). Under these conditions, it might be predicted that lowering extrasynaptic [Ca2+] would relieve repression of cAMP signaling through CaR's interaction with Gαi (keeping in mind, however, that Ca2+-activated ACs expressed in neuronal cells might mitigate this effect). This type of regulation could have profound implications for learning and memory. We recently provided functional evidence that physiological fluctuations in extracellular [Ca2+] can affect the secretory activity of the intact gastric mucosa through a PTX-sensitive interaction of CaR with the cAMP signaling pathway (Caroppo et al., 2004). We termed this effect the “third messenger” action of Ca2+ because CaR activation resulted from dynamic extracellular [Ca2+] fluctuations secondary to stimulation of the tissue with a Ca2+-mobilizing cholinergic agonist. Interestingly, as has been reported in several other cell systems, direct CaR stimulation did not produce an intracellular Ca2+ signal in gastric cells (Hofer et al., 2004). These findings highlight the complexity of signaling networks in vivo, where the effects of [Ca2+] changes in both intracellular and extracellular compartments must be taken into account in tissues where CaR or other Ca2+ sensors are expressed.
We undertook this study with the initial expectation that cAMP, as reported by cAMP-dependent conformational changes of two major cellular targets of the cAMP signal, PKA and Epac, might undergo “oscillations” or other dynamic fluctuations during concurrent stimulation of CaR and other cAMP-generating receptors, based on the robust oscillatory Ca2+ response during CaR activation in HEK CaR cells. Gorbunova and Spitzer (2002) reported slow, spontaneous cAMP transients lasting several minutes in embryonic Xenopus laevis spinal neurons that followed bursts of spontaneous Ca2+ oscillations. These investigators identified optimal patterns of Ca2+ spiking during the preceding cluster of Ca2+ oscillations that gave rise to these intermittent cAMP elevations. We examined the cAMP signal over a much shorter time frame and only during contemporaneous stimulation of CaR and cAMP pathways. In our studies we only observed a smooth monotonic decline in the FRET ratio under these conditions. We cannot exclude the possibility that complex spatiotemporal cAMP dynamics occur in localized domains (Tasken and Aandahl, 2004) and are therefore undetectable by the PKA- or Epac-based sensors. Rich et al. (2001a) provided evidence for localized cAMP signals under the plasma membrane of HEK cells that differed temporally from those of the bulk cytoplasm, as measured by currents through transfected cyclic nucleotide gated cation channels. However, our data clearly show that the relatively fast frequencies of Ca2+ oscillations that are typical of stimulation with native Ca2+-mobilizing agonists cannot exert effects on PKA subunit dissociation (Fig. 7 A) or Epac conformational changes (Fig. 6, B and C). This important result means that, in principle, high-frequency Ca2+ signaling can take place without affecting the activity of two of the major targets of the intracellular cAMP signal, PKA and Epac.
Our data revealed a complex twist to this story, namely, that during concurrent stimulation of cAMP and Ca2+ pathways, cAMP initially causes the Ca2+ oscillation frequency to increase. However, these spikes eventually fuse into a persistent Ca2+ elevation, a pattern that is effective in inhibiting cAMP accumulation. This may help explain why the effects of Ca2+-mobilizing agonists (spermine and carbachol) on the FRET ratio are only evident in Fig. 2 F and Fig. 6 (A and C) after some delay. Initially, rapid Ca2+ spiking would not be effective in turning off cAMP production, but after several minutes, during the elevated plateau phase of the signal, Ca2+ would be able to exert its inhibitory action on cAMP.
Application of artificial low-frequency, long-duration signals yielded a complex staircase profile of cAMP decline (Fig. 7 C). This mode of low-frequency, long-duration Ca2+ spiking has not been observed in this cell type during a single stimulation with a native agonist; however, this profile would be generated during a brief (∼3-min duration) stimulation with a Ca2+-mobilizing agonist in the presence of cAMP. Because the molecular targets of cAMP (e.g., type I and II regulatory subunits of PKA and Epac) have a wide range of affinities for cAMP, it is possible that a large uniform Ca2+ pulse of moderate duration could serve to “reset” [cAMP] to a lower level that favors activation of a higher affinity cAMP-binding target. This may be one more example of how complex spatial and temporal interactions between different signaling systems can be used to create unique messenger profiles that elicit diverse biological endpoints.
Materials And Methods
HEK 293 cells stably expressing the human CaR and WT HEK 293 were grown in high-glucose DME + glutamax (Life Technologies) supplemented with 10% FBS containing 10 U/ml penicillin and 10 mg/ml streptomycin. Cells were maintained in a humidified incubator under 5% CO2 and 95% O2 at 37°C.
FRET-based measurement of cAMP in single cells
Intracellular cAMP was imaged in single cells expressing a genetically encoded FRET-based indicator described previously (Zaccolo and Pozzan, 2002; Zaccolo, 2004). In brief, the probe was generated by fusing the regulatory and the catalytic subunits of PKA to CFP and YFP, respectively. When [cAMP] is low, PKA is in its holomeric form and the two fluorophores, CFP and YFP, are separated physically by a few nanometers. The excitation energy of the CFP (“donor,” 440 nm) can pass via Förster energy transfer to the acceptor molecule, YFP, which in turn emits at its own wavelength (535 nm). This gives rise to maximal FRET. When intracellular cAMP rises, the two PKA subunits dissociate and FRET is abolished. FRET can be estimated as the ratio of donor (480 nm) to acceptor emission (535 nm) intensities when cells are excited at the donor excitation wavelength (440 nm). Changes in the fluorescence emission ratio (480/535 nm) are directly correlated to changes in catalytic and regulatory subunit association and thus to cAMP levels (Mongillo et al., 2004).
A second type of sensor for cAMP relying on the cAMP-dependent conformational change of the monomeric cAMP-binding protein Epac (monitored as a change in FRET between CFP- and YFP-labeled domains) was also used. CFP-Epac(∂DEP)-YFP, a catalytically inactive version of the cAMP effector protein Epac with a cytosolic localization (Ponsioen et al., 2004), was a gift from K. Jalink and colleagues (the Netherlands Cancer Institute, Amsterdam, Netherlands). The 480/535 nm emission ratio (excitation at 440 nm) of this fast cAMP sensor was monitored as for the PKA-based probes.
HEK 293 CaR and HEK 293 WT cells were plated onto sterilized glass coverslips resting in the bottom of 60-mm plastic culture dishes and transiently transfected with cAMP probes using effectene transfection reagent (QIAGEN). Imaging experiments were performed on subconfluent cells 2 d after the transfection using a ratio imaging setup running Metafluor software (Universal Imaging Corp.). Coverslips with probe-transfected cells were mounted in an open-topped perfusion chamber (Series 20; Warner Instrument Corp.) and placed on the heated stage of an inverted microscope (TE200; Nikon) equipped with a CFP/YFP FRET filter set (XF88-2/E; Omega Optical). Cells were excited at 440 nm for 80 ms through a 40× (NA 1.4) oil immersion objective (455-nm DRLP). Emitted light was captured alternately at 480 and 535 nm using a microprocessor-controlled filter wheel placed in the emission light path. Pairs of fluorescence images at the two wavelengths were captured by a charge-coupled device camera (ORCA ER; Hamamatsu) every 5 s and converted to a ratio image by the Metafluor software.
Measurement of [Ca2+]i
Parallel measurements of cytoplasmic-free [Ca2+] were performed using fura-2–AM–loaded HEK 293 cells as described previously (De Luisi and Hofer, 2003). Using the imaging system described for cAMP, cells were excited alternately at 340 and 380 nm for 80 ms through a 40× (NA 1.4) oil immersion objective. The excitation wavelengths were generated using a computer-controlled filter wheel (Sutter Instrument Co.) placed in the path of a 100-W mercury light source. Pairs of fluorescence images (emission collected >510 nm) were captured by the charge-coupled device camera every 4 s and converted to a ratio image by the Metafluor software.
Concurrent measurements of fura-2 and CFP-YFP FRET
HEK CaR cells expressing the Epac sensor were loaded with fura-2 and sequential fura-2, and CFP/YFP ratio pairs were collected, keeping only the 455-nm dichroic mirror in place (Landa et al., 2005). Significant contamination of the FRET ratio with the fura-2 signal was noted using this configuration.
Experiments on digitonin-permeabilized HEK cells expressing R230K PKA-based sensor
As described previously (Hofer and Machen, 1993), cells were rinsed briefly in a high K+ solution (125 mM KCI, 25 mM NaCI, 0.1 mM MgCl2, and 10 mM Hepes, pH 7.20) and then exposed for 2–3 min to an intracellular buffer (the same solution with free [Ca2+] clamped to 170 nM using Ca2+/EGTA buffers and supplemented with 1 mM Na2ATP) also containing 5 μg/ml digitonin at 37°C. After plasma membrane permeabilization, cells were continuously superfused with intracellular buffer (without digitonin, but containing varying [cAMP]). An in situ calibration established that 500 nM PGE2 caused cAMP levels to increase to ∼30–50 μM, as measured by the low-affinity R230K sensor.
Solutions and materials
Unless otherwise stated, all chemicals were purchased from Sigma-Aldrich. Experiments were performed with a Ringer's solution containing 121 mM NaCl, 2.4 mM K2HPO4, 0.4 mM KH2PO4, 1.2 mM CaCl2, 1.2 mM MgCl2, 10 mM glucose, and 10 mM Hepes/NaOH, pH 7.40. Bradykinin, ionomycin, and Sp-8-Br-cAMPS (adenosine 3′, 5′-cyclic monophosphorthioate, 8-Bromo, Sp-Isomer, and Na+ salt) were obtained from Calbiochem-Novabiochem; fura-2–AM and BAPTA-AM were obtained from Invitrogen. NPS-R-467 was obtained from E. Nemeth (NPS Pharmaceuticals, Salt Lake City, UT). When DMSO or ethanol was used as a solvent, the final solvent concentration never exceeded 0.01 or 0.1%, respectively.
At the end of each experiment using the cAMP probes, cells were superfused with a supramaximal dose (50–100 μM) of forskolin, a reliable activator of AC. Because cells were occasionally fluorescent but nevertheless nonresponsive to agonists (possibly because of improper folding or targeting of the probe), only the cells that responded to the forskolin with a robust ratio increase were statistically averaged. Statistical significance was determined using a paired t test. Data from several cells in a particular experimental run were averaged, and experiment averages were used to calculate the mean ± SEM. Traces shown are typical of at least three similar experiments unless otherwise noted.
We are very grateful to Dr. Kees Jalink for the kind gift of the Epac sensor. We also thank Jessica Roy for expert technical assistance, Professor Edward M. Brown for helpful comments on our data, and Dr. Ed Nemeth for graciously providing us with NPS-R-467.
These studies were supported by grants from the Medical Research Service of the Veteran's Administration and a pilot/feasibility award from the Brigham and Women's Surgical Research Group (both to A.M. Hofer). A. Gerbino was supported by a doctoral fellowship awarded jointly from the University of Bari and the European Community (FSE). Work in the laboratory of M. Zaccolo is supported by Telethon Italy (TCP00089), the European Union (QLK3-CT-2002-02149), the Italian Cystic Fibrosis Research Foundation, and the Fondazione Compagnia di San Paolo.
A. Gerbino's present address is Dipartimento di Fisiologia Generale ed Ambientale, Universita' di Bari, 70126 Bari, Italy.
Abbreviations used in this paper: AC, adenylyl cyclase; [Ca2+]i, intracellular calcium concentration; CaR, extracellular Ca2+-sensing receptor; Epac, exchange protein activated by cAMP; FRET, fluorescence resonance energy transfer; GPCR, G protein–coupled receptor; HEK, human embryonic kidney; InsP3, inositol 1,4,5-trisphosphate; PDE, phosphodiesterase; PGE2, prostaglandin E2; PKA, protein kinase A; PTX, pertussis toxin; VIP, vasoactive intestinal peptide; WT, wild-type.