Microdomains have been proposed to explain specificity in the myriad of possible cellular targets of cAMP. Local differences in cAMP levels can be generated by phosphodiesterases, which control the diffusion of cAMP. Here, we address the possibility that adenylyl cyclases, the source of cAMP, can be primary architects of such microdomains. Distinctly regulated adenylyl cyclases often contribute to total cAMP levels in endogenous cellular settings, making it virtually impossible to determine the contribution of a specific isoform. To investigate cAMP dynamics with high precision at the single-isoform level, we developed a targeted version of Epac2-camps, a cAMP sensor, in which the sensor was tagged to a catalytically inactive version of the Ca2+-stimulable adenylyl cyclase 8 (AC8). This sensor, and less stringently targeted versions of Epac2-camps, revealed opposite regulation of cAMP synthesis in response to Ca2+ in GH3B6 pituitary cells. Ca2+ release triggered by thyrotropin-releasing hormone stimulated the minor endogenous AC8 species. cAMP levels were decreased by inhibition of AC5 and AC6, and simultaneous activation of phosphodiesterases, in different compartments of the same cell. These findings demonstrate the existence of distinct adenylyl-cyclase-centered cAMP microdomains in live cells and open the door to their molecular micro-dissection.

Cyclic AMP impacts most aspects of the life of a cell. Obviously, such wide-ranging effects must be temporally and spatially constrained. Experimental evidence is accumulating for the existence of cAMP microdomains in which cAMP targets are discretely and selectively regulated in a manner that bears little relation to changes in gross cellular levels of cAMP (Bacskai et al., 1993; Hempel et al., 1996; Nikolaev et al., 2006; Rich et al., 2001; Willoughby et al., 2006; Zaccolo and Pozzan, 2002). Recent studies suggest that adenylyl cyclases (ACs), the originators of cAMP, can actually orchestrate their own microenvironment by recruiting a variety of signalling and scaffolding molecules (Bauman et al., 2006; Chou et al., 2004; Crossthwaite et al., 2006; Dupre et al., 2007; Piggott et al., 2008). Thus, the nine differently regulated isoforms of AC (Sunahara and Taussig, 2002) can also contribute significantly to the diversity of cellular cAMP microdomains.

All cells investigated endogenously express more than one AC isoform (Willoughby and Cooper, 2007). Global cAMP concentrations ([cAMP]) represent the spatiotemporal summation of their respective activities, and the contribution of individual ACs cannot be resolved by traditional methods such as measuring cAMP accumulation in cell populations (Evans et al., 1984). The properties of individual AC isoforms have been studied in cell systems where their activities can be made prominent by heterologous expression (Fagan et al., 1996; Fagan et al., 2000). Consequently, the contribution of less abundant AC isoforms in endogenous situations, as well as the outcome for spatially restricted cAMP signalling is unclear when different AC isoforms influence cAMP dynamics in the same cell.

Genetically-encoded, Förster resonance energy transfer (FRET)-based sensors for cAMP have been developed recently, which enable cAMP dynamics to be observed in single cells, thereby greatly enhancing the temporal resolution of cAMP signalling (DiPilato et al., 2004; Nikolaev et al., 2004a; Ponsioen et al., 2004). The majority of these sensors report global changes in [cAMP]i owing to their cytosolic expression (Willoughby and Cooper, 2008). Few attempts have been made to measure cAMP dynamics with greater spatial resolution. These approaches used versions of the sensors that were targeted to the plasma membrane, mitochondria or the nucleus (DiPilato et al., 2004). However, to address the role of individual AC isoforms, even more-specific targeting is required.

Amongst the mammalian AC family, AC1 and AC8 constitute the subgroup of Ca2+-stimulable enzymes (Cooper et al., 1995). The regulation of AC8 by Ca2+ entry has been studied in great detail; in non-excitable cells, AC8 is predominantly stimulated by Ca2+ ions entering the cells via capacitative calcium entry (CCE), which is triggered by the depletion of intracellular Ca2+ stores (Fagan et al., 1996; Parekh and Putney, 2005; Putney, 2001). However, with few exceptions (Martin et al., 2009; Watson et al., 2000), this characterization has been made in cell systems where the AC8 protein was heterologously expressed, so that little is known about the physiological regulation of AC8 in an endogenous context.

In the present study, we developed targeted versions of the Epac2-camps sensor (Nikolaev et al., 2004a) to investigate cAMP compartments in a system where AC8 is endogenously expressed. We chose GH3B6 pituitary-derived cells (Tashjian, 1979), which secrete prolactin (PRL) in response to thyrotropin-releasing hormone (TRH) and vasoactive intestinal peptide (VIP); two hypothalamic hormones that are believed to act by different signalling pathways (Bjoro et al., 1987; Guild and Drummond, 1984; Schlegel and Wollheim, 1984; Winiger and Schlegel, 1988). We used versions of Epac2-camps with increasing degrees of targeting precision – ranging from simple cytosolic (globalEpac2) to `generic' plasma membrane (pmEpac2) to specific for the AC8 isoform (Epac2AC8).

In response to physiological elevations of [Ca2+]i induced by TRH, Epac2-camps tagged to a catalytically inactive version of AC8 (Epac2AC8D416N) registered a dramatic increase in [cAMP] up to ∼15 μM, which we ascribe to the activity of endogenous AC8. By contrast, globalEpac2 and pmEpac2 detected a decrease in [cAMP] upon TRH-induced Ca2+-mobilization, which relies on the direct inhibition of AC5 and AC6 and the stimulation of phosphodiesterase 1 (PDE1) enzymes by Ca2+. However, Gs-protein-mediated stimulation of AC activity by VIP effectively increased [cAMP] as measured by globalEpac2 and pmEpac2, whereas it showed far less impact on [cAMP] detected by Epac2AC8D416N.

The locally disparate cAMP responses identified in this study provide the first direct evidence for AC-selective compartments within individual cells and establish the validity of exploring cAMP dynamics with sensor molecules that are targeted to restricted microdomains.

Development and characterization of Epac2AC8 cAMP sensors

The compartmentalization of ACs suggests that cAMP kinetics at the origin of its synthesis differ from those in the bulk cytosol. To enable investigation of the immediate AC environment, we fused the Epac2-camps sensor (Nikolaev et al., 2004a) to the N-terminus of the isoform AC8 (Epac2AC8; Fig. 1A). AC8 was chosen, because the literature indicates that this enzyme is positioned within cells to selectively respond to different stimuli within a strictly organized microenvironment.

When expressed in HEK293 cells, Epac2AC8 is localized to the plasma membrane (Fig. 1A). We next established that the addition of Epac2-camps to the N-terminus of AC8 had not compromised the activity of the enzyme. The responsiveness to Ca2+-calmodulin (CaM) of wild-type (wt) AC8 and Epac2AC8 was indistinguishable when measured in membranes prepared from transfected HEK293 cells. Both wtAC8 and Epac2AC8 were half-maximally activated by submicromolar free [Ca2+] (EC50 ± s.d., 416±22 nM and 336±20 nM, respectively; Fig. 1B). To allow Epac2AC8 to detect cAMP generated by other means than its own activity, we rendered the Epac2AC8 sensor incapable of producing cAMP by using a site-directed mutagenesis approach. This Epac2AC8D416N sensor was generated by replacing the invariant Asp416 in the catalytic region of AC8 with Asn (D416N, Fig. 1A). The corresponding substitution in the catalytic domain of AC5 (D475N) virtually eliminates AC activity (Tesmer et al., 1999). Indeed, Epac2AC8D416N was devoid of catalytic activity (Fig. 1B), although it was still delivered to the plasma membrane (Fig. 1A).

The expression levels of the three AC8 constructs were explored by western blotting using a polyclonal antibody, directed against the C-terminus of AC8. Apart from the empty vector control (pcDNA), three major signals were detected in each sample. The most prominent band represents the mature, glycosylated monomer (∼150 kDa, wtAC8 and ∼230 kDa, Epac2AC8 and Epac2AC8D416N; Fig. 1C). The two other species detected represent the non-glycosylated monomer (∼135 kDa, wtAC8 and ∼215 kDa, Epac2AC8 and Epac2AC8D416N; Fig. 1C) and a dimeric form of AC8 (>250 kDa for wtAC8, Epac2AC8 and Epac2AC8D416N; Fig. 1C). Dimeric forms of AC8 that withstand SDS-PAGE have been previously described (Pagano et al., 2009). The expression levels of the monomeric AC8 species were similar.

Fig. 1.

Development and characterization of Epac2AC8 cAMP sensors. (A) Localization of transiently expressed Epac2AC8 and Epac2AC8D416N in HEK293 cells. Expression was detected by recording the CFP emission. The plasma membrane was labelled using CellMask Deep Red plasma membrane stain. Scale bars: 10 μm. (B) AC activity was measured in crude membranes from HEK293 cells transiently expressing pcDNA, wtAC8, Epac2AC8 or Epac2AC8D416N in the presence of 1 μM calmodulin. Data, normalized to 0 mM Ca2+, are plotted as mean ± s.d., and are representative of three independent experiments. (C) Samples (∼20 μg protein) of the crude membrane preparations used in B were separated on a 7% SDS gel and transferred onto a nitrocellulose membrane. The membrane was probed with the anti-AC8 antibody (upper panel). To control for equal protein loading, the bottom half of the membrane was incubated with the anti-calnexin (Cnx) antibody (bottom panel). ##, dimeric AC8 species; #, glycosylated, monomeric AC8 species; *, non-glycosylated AC8 monomers.

Fig. 1.

Development and characterization of Epac2AC8 cAMP sensors. (A) Localization of transiently expressed Epac2AC8 and Epac2AC8D416N in HEK293 cells. Expression was detected by recording the CFP emission. The plasma membrane was labelled using CellMask Deep Red plasma membrane stain. Scale bars: 10 μm. (B) AC activity was measured in crude membranes from HEK293 cells transiently expressing pcDNA, wtAC8, Epac2AC8 or Epac2AC8D416N in the presence of 1 μM calmodulin. Data, normalized to 0 mM Ca2+, are plotted as mean ± s.d., and are representative of three independent experiments. (C) Samples (∼20 μg protein) of the crude membrane preparations used in B were separated on a 7% SDS gel and transferred onto a nitrocellulose membrane. The membrane was probed with the anti-AC8 antibody (upper panel). To control for equal protein loading, the bottom half of the membrane was incubated with the anti-calnexin (Cnx) antibody (bottom panel). ##, dimeric AC8 species; #, glycosylated, monomeric AC8 species; *, non-glycosylated AC8 monomers.

Targeting of Epac2AC8 and Epac2AC8D416N

One of the most striking features of AC8 in the intact cell is that it is stimulated by CCE (Parekh and Putney, 2005) but not by any other mode of Ca2+ entry (Fagan et al., 1996). This dependence on CCE seems to reflect a close apposition of AC8 and the CCE apparatus (Fagan et al., 1996; Smith et al., 2002). Consequently, as a test of correct functional targeting, we compared the sensitivity to CCE of the Epac2AC8 construct with wtAC8.

We addressed this question in HEK293 cells, where CCE was triggered by the passive depletion of intracellular Ca2+ stores with thapsigargin, before adding Ca2+ to the extracellular medium (supplementary material Fig. S1A). Cyclic AMP accumulation was enhanced in cells expressing either wtAC8 or Epac2AC8 (Fig. 2A) in direct proportion to the extracellular Ca2+ concentration ([Ca2+]ex) applied. No obvious differences between wtAC8 and Epac2AC8 were observed at any [Ca2+]ex. HEK293 cells expressing empty vector (pcDNA) or Epac2AC8D416N, showed no response to CCE (Fig. 2A).

Fig. 2.

Targeting of Epac2AC8 and Epac2AC8D416N. (A) Cyclic AMP accumulation was measured in cell populations expressing pcDNA, wtAC8, Epac2AC8, or Epac2AC8D416N in response to increasing CCE. Data are plotted as mean ± s.d., and are representative of three independent experiments. (B) Localization of transiently expressed globalEpac2 and pmEpac2 in HEK293 cells. Expression was detected by recording the CFP emission. The plasma membrane was labelled using CellMask Deep Red plasma membrane stain. Scale bar: 10 μm. (Ci-Civ) CCE-induced changes in R/R0 [% max] in HEK cells transiently expressing Epac2-based cAMP sensors on a wtAC8 background. CCE was triggered by the addition of (i) 0.1 mM, (ii) 0.25 mM, (iii) 0.5 mM or (iv) 4.0 mM [Ca2+]ex after pre-treatment for 3 minutes with 200 nM thapsigargin in the presence of 100 μM EGTA. Maximal changes in R/R0 were achieved by stimulating AC activity with 10 μM FSK in the presence of 100 μM IBMX (MAX STIM). Data were plotted as mean ± s.e.m, (n>12). Inset in Ciii shows magnification of 1-2 minute region, visualizing the transient inhibitory effect of CCE on [cAMP] monitored with pmEpac2 at 0.5 mM [Ca2+]ex. (D) CCE dose-response curve of the CFP/YFP response (R/R0 [% max]) for globalEpac2, pmEpac2, Epac2AC8 and Epac2AC8D416N. Peak changes in R/R0 [% max] in response to CCE were plotted as a function of [Ca2+]ex. Data were sigmoid fitted (variable slope) and plotted as mean ± s.e.m. (n>12). (E) In vitro concentration response curves of Epac2 sensors. FRET responses (Δ CFP/YFP) elicited by known [cAMP] were measured fluorometrically. Data are plotted as mean ± s.d. (n=4), and sigmoid fitted (variable slope) using the GraphPadPrism software.

Fig. 2.

Targeting of Epac2AC8 and Epac2AC8D416N. (A) Cyclic AMP accumulation was measured in cell populations expressing pcDNA, wtAC8, Epac2AC8, or Epac2AC8D416N in response to increasing CCE. Data are plotted as mean ± s.d., and are representative of three independent experiments. (B) Localization of transiently expressed globalEpac2 and pmEpac2 in HEK293 cells. Expression was detected by recording the CFP emission. The plasma membrane was labelled using CellMask Deep Red plasma membrane stain. Scale bar: 10 μm. (Ci-Civ) CCE-induced changes in R/R0 [% max] in HEK cells transiently expressing Epac2-based cAMP sensors on a wtAC8 background. CCE was triggered by the addition of (i) 0.1 mM, (ii) 0.25 mM, (iii) 0.5 mM or (iv) 4.0 mM [Ca2+]ex after pre-treatment for 3 minutes with 200 nM thapsigargin in the presence of 100 μM EGTA. Maximal changes in R/R0 were achieved by stimulating AC activity with 10 μM FSK in the presence of 100 μM IBMX (MAX STIM). Data were plotted as mean ± s.e.m, (n>12). Inset in Ciii shows magnification of 1-2 minute region, visualizing the transient inhibitory effect of CCE on [cAMP] monitored with pmEpac2 at 0.5 mM [Ca2+]ex. (D) CCE dose-response curve of the CFP/YFP response (R/R0 [% max]) for globalEpac2, pmEpac2, Epac2AC8 and Epac2AC8D416N. Peak changes in R/R0 [% max] in response to CCE were plotted as a function of [Ca2+]ex. Data were sigmoid fitted (variable slope) and plotted as mean ± s.e.m. (n>12). (E) In vitro concentration response curves of Epac2 sensors. FRET responses (Δ CFP/YFP) elicited by known [cAMP] were measured fluorometrically. Data are plotted as mean ± s.d. (n=4), and sigmoid fitted (variable slope) using the GraphPadPrism software.

To determine whether the Epac2-modified AC8 constructs could be used to probe cAMP dynamics at the single-cell level, we compared the responsiveness of the Epac2 sensor in the AC8 constructs with less stringently targeted versions of Epac2: the non-modified, ancestral sensor Epac2-camps (globalEpac2), which is expressed uniformly throughout the cytosol (Fig. 2B) and a plasma-membrane-targeted version of Epac2 (pmEpac2). The pmEpac2 sensor was constructed by N-terminally modifying globalEpac2 with the `SH4' motif (GCINSKRKD) of Lyn kinase, which is post-translationally modified with myristate and palmitate (Resh, 2006; Zacharias et al., 2002). This modification led the pmEpac2 sensor to be localized to the plasma membrane (Fig. 2B).

All sensors were individually expressed on a wtAC8 background, which was necessary to allow CCE-stimulated cAMP production in cells expressing globalEpac2, pmEpac2 or Epac2AC8D416N. Changes in CFP/YFP (R/R0) upon triggering CCE (following store depletion with thapsigargin) with a range of [Ca2+]ex were recorded for each sensor, and expressed as percentages of the maximal change achieved by adding 10 μM forskolin in the presence of 100 μM isobutylmethylxanthine (IBMX; Fig. 2C). Expression of the data in this manner was important, because the normalized data indicate the impact of CCE on the respective local [cAMP], which varies depending on the different microenvironments. When Ca2+ was added to the external medium, a dose-dependent increase in R/R0 was observed for all sensors, whereby Epac2AC8 and Epac2AC8D416N were more responsive than globalEpac2 or pmEpac2 (Fig. 2Ci-iv). This difference was quantified by plotting the peak changes in R/R0 [% max] as a function of [Ca2+]ex (Fig. 2Cv). The EC50 values ± s.e.m. for [Ca2+]ex of Epac2AC8 and Epac2AC8D416N were similar (237±4 μM and 262±2 μM, respectively; Fig. 2Cv). The globalEpac2 sensor was about 2.5 times less responsive (EC50[Ca2+], 610±6 μM), and the pmEpac2 sensor was the least responsive (EC50[Ca2+], 837±5 μM; Fig. 2Cv). In fact the pmEpac2 sensor reported a transient inhibition of cAMP accumulation at low [Ca2+]ex before the stimulatory effect predominated (Fig. 2Ciii, inset). These findings might reflect the distances between the Epac2AC8 sensors that are close to the source of cAMP, versus the cytosolically placed globalEpac2. Although the pmEpac2 was expressed at the plasma membrane, it appeared to be in a different domain than AC8, and was possibly associated with the endogenous Ca2+-inhibitable AC6 of HEK293 cells.

Of course, the interpretation of responsiveness, which differs based on placement, relies on similar actual cAMP-sensitivities of the FRET responses amongst the sensors. These sensitivities were compared in vitro (Fig. 2D). The EC50[cAMP] values (± s.d.) for half maximal changes in CFP/YFP were 0.403±0.038 μM, 1.663±0.127 μM, 2.178±0.404 μM and 2.223±0.239 μM for globalEpac2, pmEpac2, Epac2AC8 and Epac2AC8D416N, respectively. These values conformed to recently published EC50 ranges for Epac-based cAMP sensors (Herget et al., 2008; Nikolaev et al., 2004b). The fact that globalEpac2 is about four times more sensitive than the Epac2AC8 sensors suggests that the observed differences in the responsiveness towards CCE-stimulated cAMP production (Fig. 2Cv) are even greater. Moreover, globalEpac2 is about five times more sensitive than pmEpac2 in vitro, which might also explain its apparently higher responsiveness towards CCE-stimulated cAMP production.

We also generated derivatives of our sensors that could not bind cAMP, to test for potential artefactual effects unrelated to cAMP binding. When the conserved Arg297 within the cyclic nucleotide-binding domain is replaced with Glu, Epac proteins do not bind cAMP (de Rooij et al., 1998; Mei et al., 2002). None of the R297E mutants showed changes in CFP/YFP (R/R0), even when AC activity was maximally stimulated (supplementary material Fig. S1B). We further addressed the possibility that our measurements had been compromised, by intermolecular FRET, which depends on the density of sensor molecules within a restricted space (Zacharias et al., 2002). For this purpose, the R0 values in transfected HEK293 cells expressing the R297E versions of the Epac2 sensors at different amounts were recorded. Within their typical expression ranges, the R0 values did not correlate with the expression level of the R297E sensors (supplementary material Fig. S1Ci-iv).

In summary, these results indicate that we had generated a functional and appropriately placed cAMP sensor (Epac2AC8D416N), which might allow us to monitor AC8 activity, in a cellular system where this enzyme is endogenously expressed.

Ca2+ reduces global [cAMP] in GH3B6 cells

Pituitary-derived GH3B6 cells are an ideal model to explore the issue of AC compartments. These cells secrete growth hormone and prolactin by both cAMP- and Ca2+-dependent mechanisms (Ooi et al., 2004). Hormones such as VIP elevate [cAMP] through Gs-linked mechanisms (Bjoro et al., 1987; Guild and Drummond, 1984). More controversially, TRH seems to act via both, Ca2+-dependent (Schlegel and Wollheim, 1984; Winiger and Schlegel, 1988) and cAMP-dependent mechanisms (Gautvik et al., 1983; Paulssen et al., 1992). It is uncertain whether TRH, which stimulates prolactin via the activation of Gq proteins (Aragay et al., 1992), can actually affect [cAMP] and if any such effect is direct or indirect. Part of the confusion arises from the fact that these cells express more than one AC isoform (Paulssen et al., 1994), each of which is subject to different modes of regulation. Thus, the complex regulation of cAMP signalling within GH3B6 cells provides an ideal situation to ask whether regulation of cAMP emanating from different sources might be resolved with discretely targeted probes.

We first determined the AC isoforms expressed in GH3B6 cells using an RT-PCR approach with rat sequence-specific primer pairs for the AC1, AC2, AC5, AC6 and AC8 (Fig. 3Ai). Most abundant (intensity relative to positive control) was the product for the Ca2+-inhibitable AC6. At lower levels, we also detected the Ca2+-insensitive AC2 and another Ca2+-inhibitable isoform, AC5 (Fig. 3Ai). Significant amounts of the Ca2+-CaM-stimulable AC8 were also detected, whereas no message for the other Ca2+-CaM-stimulable isoform, AC1, was found (Fig. 3Ai). To confirm the expression of AC8 at the protein level, we investigated its presence in membrane preparations. An immunoreactive band in the GH3B6 cell sample migrating at ∼150 kDa, corresponds to the apparent mass of the glycosylated, monomeric form of heterologously expressed AC8 (Fig. 3Aii).

We next examined the net effect of Ca2+ on this mixture of AC isoforms by measuring the in vitro Ca2+ response of AC activity in crude membrane preparations, in the presence and absence of CaM. A purely inhibitory response to Ca2+ was displayed by the AC activity in GH3B6 cell membranes, which was very similar to the responses observed in crude membranes prepared from HEK293 cells overexpressing Ca2+-inhibitable AC5 (Fig. 3B). Ca2+ concentrations of ∼5 μM inhibited FSK-stimulated AC activity by ∼60% and ∼50% in GH3B6 and HEK-AC5 cell membranes, respectively (Fig. 3B). The IC50 values for Ca2+ inhibition (± s.d.) were ∼130±52 nM for AC5 and ∼197±69 nM for GH3B6 ACs. The presence of 1 μM exogenous CaM failed to reveal any stimulation of AC activity (Fig. 3B), which suggested that the Ca2+-inhibitable isoforms AC5 and AC6 predominate in these cells. Such a CaM-independent, inhibitory effect of Ca2+ on AC activity in plasma membrane preparations from GH3B6 cells has been described previously (Boyajian and Cooper, 1990). This also agrees with the observations made in other cell systems predominantly expressing Ca2+-inhibitable ACs, such as pulmonary artery endothelial cells (Stevens et al., 1995).

As a prelude to considering the effects of cytosolic Ca2+ concentrations ([Ca2+]cyt) on [cAMP], we characterized the behaviour of [Ca2+]cyt upon hormonal stimulation with TRH. When examined in GH3B6 cells loaded with the Ca2+ indicator Fura-2, basal [Ca2+]cyt oscillated with great variation from cell to cell (Fig. 3C, inset). Upon treatment with TRH however, a biphasic Ca2+ signal was generated (Fig. 3C), comprising a sharp, initial burst in [Ca2+]cyt, which rapidly transited into a second oscillatory phase (Fig. 3C).

Fig. 3.

Ca2+ reduces global [cAMP] in GH3B6 cells. (Ai) GH3B6 cDNA was used as template for RT-PCR. Fragments for AC1, AC2, AC6, AC5 and AC8 were amplified using rat sequence-specific primer pairs, and resolved on a 0.75% agarose gel (top). Control reactions (bottom) were carried out using plasmid DNA containing the respective full-length AC cDNA as template. (Aii) Crude membrane preparations from HEK293 (0.5 μg), HEK-AC8 (0.5 μg) and GH3B6 cells (20 μg) were resolved on a 7% SDS gel, and transferred on a nitrocellulose membrane. The HEK293 and GH3B6 membrane samples were individually probed with the anti-AC8 antibody. (B) AC activity (100%=activity in Ca2+-free conditions) was measured in crude membranes prepared from GH3B6 cells, and from HEK293 cells expressing AC5. In vitro assays were carried out in the absence or presence of 1 μM CaM. Data were plotted as mean ± s.d., and are representative of at least three independent experiments. Data were fitted with the two site competition (no CaM, red), or the non-linear algorithm (+ CaM, green). (C) Changes in [Ca2+]cyt were monitored in single cells loaded with 2 μM Fura-2. GH3B6 cells were incubated in HBS buffer containing 1.8 mM Ca2+. TRH (100 nM) was added at 1 minute. Changes in 340/380 nm were shown as F/F0 from seven individual cells. Inset shows spontaneous Ca2+ oscillations monitored in non-treated GH3B6 cells (traces of four representative cells are shown). (D) Cyclic AMP accumulation was measured over a 3 minute period in GH3B6 cell populations under basal conditions (HBS + 1.8 mM Ca2+), and in response to 10 nM VIP or 200 nM TRH. Data are plotted as mean ± s.e.m. (n=5), **P<0.01 compared with basal level, n.s., no significant difference from basal (P>0.05, Dunnett's multiple comparison test; one-way ANOVA). (Ei) cAMP accumulation was measured in GH3B6 cell populations under basal conditions (HBS + 1.8 mM Ca2+), in response to 200 nM TRH, 1 μM FSK or 1 μM FSK + 200 nM TRH at different time points after stimulation. To facilitate temporal orientation, TRH-induced changes in [Ca2+]cyt were superimposed (blue trace), which had been monitored in GH3B6 cells loaded with 2 μM Fura-2. Data are plotted as mean ± s.e.m. (n=60). (Eii) Ei in the presence of 100 μM IBMX. Data in Ei and Eii were normalized to the cAMP accumulation at 20 s measured under basal conditions (Ei) or IBMX conditions (Eii). Data are plotted as mean ± s.e.m. (n=3). Statistical differences between corresponding FSK (green symbols) and FSK+TRH (red symbols) data were analysed using the unpaired t-test (*P<0.05).

Fig. 3.

Ca2+ reduces global [cAMP] in GH3B6 cells. (Ai) GH3B6 cDNA was used as template for RT-PCR. Fragments for AC1, AC2, AC6, AC5 and AC8 were amplified using rat sequence-specific primer pairs, and resolved on a 0.75% agarose gel (top). Control reactions (bottom) were carried out using plasmid DNA containing the respective full-length AC cDNA as template. (Aii) Crude membrane preparations from HEK293 (0.5 μg), HEK-AC8 (0.5 μg) and GH3B6 cells (20 μg) were resolved on a 7% SDS gel, and transferred on a nitrocellulose membrane. The HEK293 and GH3B6 membrane samples were individually probed with the anti-AC8 antibody. (B) AC activity (100%=activity in Ca2+-free conditions) was measured in crude membranes prepared from GH3B6 cells, and from HEK293 cells expressing AC5. In vitro assays were carried out in the absence or presence of 1 μM CaM. Data were plotted as mean ± s.d., and are representative of at least three independent experiments. Data were fitted with the two site competition (no CaM, red), or the non-linear algorithm (+ CaM, green). (C) Changes in [Ca2+]cyt were monitored in single cells loaded with 2 μM Fura-2. GH3B6 cells were incubated in HBS buffer containing 1.8 mM Ca2+. TRH (100 nM) was added at 1 minute. Changes in 340/380 nm were shown as F/F0 from seven individual cells. Inset shows spontaneous Ca2+ oscillations monitored in non-treated GH3B6 cells (traces of four representative cells are shown). (D) Cyclic AMP accumulation was measured over a 3 minute period in GH3B6 cell populations under basal conditions (HBS + 1.8 mM Ca2+), and in response to 10 nM VIP or 200 nM TRH. Data are plotted as mean ± s.e.m. (n=5), **P<0.01 compared with basal level, n.s., no significant difference from basal (P>0.05, Dunnett's multiple comparison test; one-way ANOVA). (Ei) cAMP accumulation was measured in GH3B6 cell populations under basal conditions (HBS + 1.8 mM Ca2+), in response to 200 nM TRH, 1 μM FSK or 1 μM FSK + 200 nM TRH at different time points after stimulation. To facilitate temporal orientation, TRH-induced changes in [Ca2+]cyt were superimposed (blue trace), which had been monitored in GH3B6 cells loaded with 2 μM Fura-2. Data are plotted as mean ± s.e.m. (n=60). (Eii) Ei in the presence of 100 μM IBMX. Data in Ei and Eii were normalized to the cAMP accumulation at 20 s measured under basal conditions (Ei) or IBMX conditions (Eii). Data are plotted as mean ± s.e.m. (n=3). Statistical differences between corresponding FSK (green symbols) and FSK+TRH (red symbols) data were analysed using the unpaired t-test (*P<0.05).

Under these conditions, TRH elicited no significant deviation from basal [cAMP] when cAMP accumulation was measured in populations of GH3B6 cells over a 3 minute period (Fig. 3D). By contrast, VIP, which is known to activate ACs in GH3B6 cells, robustly stimulated cAMP accumulation ∼10-fold (Fig. 3D). In previous studies, cells that express predominantly Ca2+-inhibitable isoforms of AC, namely NCB-20 and C-6 glioma cells, showed clear declines in cAMP accumulation in response to elevations of [Ca2+]cyt (Boyajian et al., 1991; Chiono et al., 1995; Fagan et al., 1998). Therefore, the time course of the Ca2+ signal elicited by TRH led us to analyse its effect on [cAMP] in greater detail. We compared the effect of TRH on cAMP accumulation in the presence and absence of FSK at different time points after stimulation (Fig. 3Ei), reflecting a progression along the course of the Ca2+-signal (superimposed in Fig. 3Ei,ii). Basal [cAMP] was again not affected by TRH; however, the cAMP accumulation stimulated by FSK was significantly reduced by TRH 3 minutes after stimulation (Fig. 3Ei). In the presence of IBMX, cAMP accumulation was generally increased by ∼75-100% compared with FSK alone (Fig. 3Eii). However, the inhibitory effect of TRH was still detected, which suggests that TRH reduces cAMP accumulation at least in part via direct Ca2+ inhibition of ACs (Fig. 3Eii).

The lack of any detectable Ca2+-stimulated AC activity could be explained by a low (relative to AC6 and AC5) and/or localized expression of AC8 activity in GH3B6 cells, which cannot be resolved using these conventional methods to measure total cAMP accumulation. To address whether spatiotemporal averaging was obscuring local increases in [cAMP], we monitored cAMP in `real time' using the Epac2-based sensors.

Epac2AC8D416N reveals Ca2+ release-stimulated cAMP signalling in GH3B6 cells

We transiently expressed globalEpac2, pmEpac2 and Epac2AC8D416N in GH3B6 cells. As with HEK293 cells (Fig. 1A, Fig. 2B), globalEpac2 was distributed throughout the cytosol, whereas pmEpac2 and Epac2AC8D416N were targeted to the plasma membrane (Fig. 4A). In these cells, changes in CFP/YFP (R/R0) upon stimulation with TRH and VIP were recorded, and expressed as a percentage of the maximal change elicited with 10 μM FSK in the presence of 100 μM IBMX.

Fig. 4.

Epac2AC8D416N reveals Ca2+-release-stimulated cAMP signalling in GH3B6 cells. (A) GH3B6 cells transiently expressing globalEpac2, pmEpac2, or Epac2AC8D416N. Expression was detected by recording CFP using the single-cell FRET imaging setup (see Materials and Methods). Scale bar: 10 μm. (B,C) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed globalEpac2 [n=23(B)/30(C)], pmEpac2 [n=26(B)/30(C)], and Epac2AC8D416N [n=30(B)/29(C)] in GH3B6 cells treated with either 10 nM VIP (B) or 100 nM TRH (C) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (D) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed pmEpac2 in GH3B6 cells treated with 10 μM MMX (red, n=26), 10 μM MMX + 100 nM TRH (blue, n=14) and 100 nM TRH (black, n=30) in the presence of 1.8 mM [Ca2+]ex. The arrow indicates the addition of substances. Data are plotted as mean ± s.e.m. (E) [Ca2+]cyt was monitored in GH3B6 cells loaded with 2 μM Fura-2. GH3B6 cells were treated with 100 nM TRH in the presence of 1.8 mM [Ca2+]ex (blue, n=60), in Ca2+-free HBS buffer containing 100 μM EGTA (grey, n=60), or after pre-treatment for 5 minutes with 5 μM U-73122 (red, n=60) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (F) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed Epac2AC8D416N in GH3B6 cells treated with 100 nM TRH in the presence of 1.8 mM [Ca2+]ex (blue, n=30), in Ca2+-free HBS buffer containing 100 μM EGTA (grey, n=26), after pre-treatment for 5 minutes with 5 μM U-73122 (red, n=17) in the presence of 1.8 mM [Ca2+]ex, or after a pre-treatment for 30 minutes with 1 μM chelerythrine chloride (CTC; green, n=21) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (G) The response of Epac2AC8D416N to 100 nM TRH in the presence of 1.8 mM [Ca2+]ex after pre-treatment with 5 μM U-73122 (red) for 5 minutes was subtracted from the response to TRH alone (blue) resulting in the dashed black trace. This represents the proportion of the Epac2AC8D416N response, which is caused by TRH-mediated Ca2+ release from intracellular stores.

Fig. 4.

Epac2AC8D416N reveals Ca2+-release-stimulated cAMP signalling in GH3B6 cells. (A) GH3B6 cells transiently expressing globalEpac2, pmEpac2, or Epac2AC8D416N. Expression was detected by recording CFP using the single-cell FRET imaging setup (see Materials and Methods). Scale bar: 10 μm. (B,C) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed globalEpac2 [n=23(B)/30(C)], pmEpac2 [n=26(B)/30(C)], and Epac2AC8D416N [n=30(B)/29(C)] in GH3B6 cells treated with either 10 nM VIP (B) or 100 nM TRH (C) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (D) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed pmEpac2 in GH3B6 cells treated with 10 μM MMX (red, n=26), 10 μM MMX + 100 nM TRH (blue, n=14) and 100 nM TRH (black, n=30) in the presence of 1.8 mM [Ca2+]ex. The arrow indicates the addition of substances. Data are plotted as mean ± s.e.m. (E) [Ca2+]cyt was monitored in GH3B6 cells loaded with 2 μM Fura-2. GH3B6 cells were treated with 100 nM TRH in the presence of 1.8 mM [Ca2+]ex (blue, n=60), in Ca2+-free HBS buffer containing 100 μM EGTA (grey, n=60), or after pre-treatment for 5 minutes with 5 μM U-73122 (red, n=60) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (F) Changes in the CFP/YFP response (R/R0 [% max]) of transiently expressed Epac2AC8D416N in GH3B6 cells treated with 100 nM TRH in the presence of 1.8 mM [Ca2+]ex (blue, n=30), in Ca2+-free HBS buffer containing 100 μM EGTA (grey, n=26), after pre-treatment for 5 minutes with 5 μM U-73122 (red, n=17) in the presence of 1.8 mM [Ca2+]ex, or after a pre-treatment for 30 minutes with 1 μM chelerythrine chloride (CTC; green, n=21) in the presence of 1.8 mM [Ca2+]ex. Data are plotted as mean ± s.e.m. (G) The response of Epac2AC8D416N to 100 nM TRH in the presence of 1.8 mM [Ca2+]ex after pre-treatment with 5 μM U-73122 (red) for 5 minutes was subtracted from the response to TRH alone (blue) resulting in the dashed black trace. This represents the proportion of the Epac2AC8D416N response, which is caused by TRH-mediated Ca2+ release from intracellular stores.

In accordance with the results described above (Fig. 3D), VIP globally increased [cAMP] in GH3B6 cells as reported by globalEpac2 and pmEpac2, which both responded to ∼85% of their maximal signals (Fig. 4B, grey and black traces, respectively). However, VIP seemed to be less effective in the vicinity of Epac2AC8D416N; this sensor responded to just 25% of its maximal signal (Fig. 4B, red trace).

Stimulation with TRH decreased cytosolic [cAMP] slightly as detected by the globalEpac2 sensor (Fig. 4C, grey trace). This effect of TRH was much more pronounced in the vicinity of the plasma membrane. The pmEpac2 sensor responded with a transient drop (up to 15%) in its signal upon stimulation with TRH (Fig. 4C, black trace), which agrees with the results from the cAMP accumulation measured in cell populations (Fig. 3E). In the presence of the PDE1-specific inhibitor 8-methoxymethyl-3-isobutyl-1-methylxanthine (MMX), basal cAMP levels were increased (Fig. 4D, red trace); simultaneous stimulation with TRH attenuated this increase (Fig. 4D, blue trace), whereas TRH alone continued to induce the drop in [cAMP] detected with pmEpac2 (Fig. 4D, black trace). This observation indicates the activation of PDE1 isoforms by TRH-induced Ca2+ signalling, in addition to an inhibition of AC5 or AC6. In striking contrast, an increase in [cAMP] was detected with Epac2AC8D416N, which responded to ∼65% of its maximal signal (Fig. 4C, red trace). Although both pmEpac2 and Epac2AC8D416N are localized to the plasma membrane, they report different cAMP dynamics, and reveal a lateral heterogeneity of the plasma membrane in GH3B6 cells. The original R/R0 average traces are shown in supplementary material Fig. S2 for both VIP and TRH. To visualize the variation of the respective maximal responses in R/R0, we have included a scatter-plot analysis (supplementary material Fig. S2Aii,Bii).

Since the cAMP signal detected with Epac2AC8D416N did not temporally superimpose with either component of the TRH-induced Ca2+ signal (compare with Fig. 4E), we wondered which, if any, part of the Ca2+ signal might have been responsible for the increase in [cAMP]. When GH3B6 cells were treated with TRH in Ca2+-free conditions, the initial Ca2+ release from internal stores remained almost unaffected, whereas the second phase, reflecting Ca2+ entry, was ablated owing to the lack of [Ca2+]ex (Fig. 4E, grey trace). Under these conditions, the effect of TRH on [cAMP] was unchanged (Fig. 4F, grey trace), which suggested that Ca2+ release, rather than entry, accounts for the stimulatory effect on AC activity. This conclusion was reinforced when we investigated the requirement of PLC for the effects of TRH on both Ca2+ and cAMP. Pre-treatment of GH3B6 cells with the PLC inhibitor U-73122 diminished TRH-induced Ca2+ release by ∼70% and almost completely prevented subsequent Ca2+ entry (Fig. 4E, red trace), as previously described (Smallridge et al., 1992). Following this treatment, the TRH-induced increase in cAMP was delayed by ∼30 seconds and attenuated by about 60% (Fig. 4F, red trace). We also considered that TRH can activate protein kinase C (PKC) (Kiley et al., 1991) and thereby might increase [cAMP] via stimulation of AC2 activity (Yoshimura and Cooper, 1993). Pre-treatment of GH3B6 cells with chelerythrine chloride (CTC) inhibits PKC by preventing its translocation to the plasma membrane (Herbert et al., 1990); however, the cAMP response was not affected by this treatment (Fig. 4E, green).

Fig. 5.

Ca2+-release-stimulated cAMP signalling in GH3B6 cells is compartmentalized. (A) In vitro parameters Fabs, EC50 and nH (Hill coefficient) for globalEpac2, pmEpac2 and Epac2AC8D416N (mean ± s.d.). The maximal, intracellular changes in R/R0 (Fmax) elicited with 10 μM FSK and 100 μM IBMX are given as mean ± s.e.m. (n⩾22). R/R0 values from single cells (see supplementary material Fig. S2 for R/R0 average traces) were converted into [cAMP] under (B) basal conditions (C) stimulation with 100 nM TRH (inset in C shows magnification of 0-1.5 minutes) and (D) stimulation with 10 nM VIP. (E) Cellular model of cAMP domains in GH3B6 cells. ER, endoplasmic reticulum; IP3, inositol-1,4,5-trisphosphate; PDE1/4, phosphodiesterase type 1/4; PLC, phospholipase C; TRH-R, Gq-coupled TRH receptor; VIP-R, Gs-coupled VIP receptor; +, stimulatory effect of Ca2+ ions on AC8 and Gs on global ACs; –, inhibitory effect of Ca2+ ions on AC5/6; red, cAMP emanating from global ACs; green, cAMP emanating from AC8.

Fig. 5.

Ca2+-release-stimulated cAMP signalling in GH3B6 cells is compartmentalized. (A) In vitro parameters Fabs, EC50 and nH (Hill coefficient) for globalEpac2, pmEpac2 and Epac2AC8D416N (mean ± s.d.). The maximal, intracellular changes in R/R0 (Fmax) elicited with 10 μM FSK and 100 μM IBMX are given as mean ± s.e.m. (n⩾22). R/R0 values from single cells (see supplementary material Fig. S2 for R/R0 average traces) were converted into [cAMP] under (B) basal conditions (C) stimulation with 100 nM TRH (inset in C shows magnification of 0-1.5 minutes) and (D) stimulation with 10 nM VIP. (E) Cellular model of cAMP domains in GH3B6 cells. ER, endoplasmic reticulum; IP3, inositol-1,4,5-trisphosphate; PDE1/4, phosphodiesterase type 1/4; PLC, phospholipase C; TRH-R, Gq-coupled TRH receptor; VIP-R, Gs-coupled VIP receptor; +, stimulatory effect of Ca2+ ions on AC8 and Gs on global ACs; –, inhibitory effect of Ca2+ ions on AC5/6; red, cAMP emanating from global ACs; green, cAMP emanating from AC8.

These findings support the idea that TRH increases [cAMP] via the PLC pathway, probably as a result of inositol-1,4,5-trisphosphate-mediated Ca2+ release. Nevertheless, the remaining cAMP increase observed when PLC was inhibited suggested at least one alternative signalling pathway for TRH, which can contribute to the stimulation of AC activity. A possible mechanism might involve the concomitant activation of TRH receptors that couple to Gs proteins, which has been suggested previously (Paulssen et al., 1992). We therefore aimed to visualize the cAMP signal that definitely relies on TRH-induced Ca2+ release by subtracting the cAMP signal trace recorded during inhibition of PLC from the original response to TRH. The resulting trace (Fig. 4G, dashed line) superimposed on the original TRH response during the first 30 seconds of cAMP increase. The peak change in R/R0, however, was reduced to ∼40% of the maximal signal, and the subsequent part of the signal became more transient (Fig. 4G).

These results indicate pronounced compartmentalization of cAMP signalling in GH3B6 cells. Whereas Gs-mediated activation of cyclases via VIP occurs as a global event, Ca2+ release elicited by TRH activates only a small and locally constrained subset of AC activity. We thus attempted to explore the dimensions and communication between these compartments by converting the changes in R/R0 into actual [cAMP].

Ca2+release-stimulated cAMP signalling in GH3B6 cells is compartmentalized

Cyclic AMP concentrations were calculated from the measured changes in R/R0 (supplementary material Fig. S2) according to the method published by Iancu and colleagues (Iancu et al., 2008). This analysis is based on parameters calculated from the in vitro calibration (Fabs, nH and EC50) as well as on the maximal change in R/R0 (Fmax) elicited in the intracellular environment (Fig. 5A).

The sensors measured different cAMP concentrations under resting conditions. Global [cAMP] was calculated as ∼300 nM, whereas [cAMP] in the vicinity of pmEpac2 and Epac2AC8D416N was much greater (∼2.5 μM; Fig. 5B). The high [cAMP] reported by pmEpac2 and Epac2AC8D416N under resting conditions probably reflects their close proximity to the cAMP sources (ACs) at the plasma membrane. This would lead to the sensing of a higher [cAMP] compared with globalEpac2, if PDEs, which are certainly active in each compartment (supplementary material Fig. S3), were not as closely associated with the ACs. In fact, a high resting [cAMP] at the plasma membrane has previously been proposed to account for the maintenance of a hyperpolarization-activated inward current in GH3B6 cells. This proposal envisaged that the effect of cAMP on the hyperpolarization-activated inward current is due to the tonic activation of hyperpolarization-activated and cyclic nucleotide-gated channels (Kretschmannova et al., 2006). Given the high cAMP-sensitivity of HCN2 (K1/2 ∼0.5 μM) (Kaupp and Seifert, 2001), which is a predominant channel subunit in GH3B6 cells (Kretschmannova et al., 2006), a basal [cAMP] of ∼2-3 μM reported by pmEpac2 and Epac2AC8D416N would substantially activate the channel.

Stimulation with TRH transiently increased [cAMP] in the vicinity of Epac2AC8D416N to ∼15 μM, whereas global [cAMP] remained almost unchanged (Fig. 5C). Remarkably, the [cAMP] registered by pmEpac2 transiently declined from ∼ 2.5 to 2 μM (Fig. 5C). Although this decrease (∼20%) appeared to be quite moderate, the simultaneous increase observed with Epac2AC8D416N enhanced the spatial contrast of cAMP signalling. Based on the divergent behaviour of [cAMP] reported by the different sensors, we propose the following model for the organization of cAMP compartments in GH3B6 cells (Fig. 5E). TRH appeared to act on [cAMP] mostly via the induction of Ca2+ release from internal stores, which led us to propose the presence of Ca2+ release-stimulated cAMP-microdomains in GH3B6 cells. AC8 was assembled within these microdomains, whereas other subtypes were excluded. Moreover, the absence of any global effect of cAMP produced within Ca2+ release-stimulated cAMP-microdomains suggested that the total volume of these microdomains constituted just a minor fraction of the overall global volume, provided that free cAMP diffusion between compartments is allowed. The estimated volume of the Ca2+ release-stimulated cAMP-microdomains must then be less than 2% of the total global volume, because ∼15 μM cAMP, as measured with Epac2AC8D416N in response to TRH, did not impact the average global [cAMP] of ∼300 nM. The concomitant activation of PDE1 upon stimulation with TRH (Fig. 4D, Fig. 5E), could also contribute to the lack of a TRH-mediated increase in global [cAMP].

In contrast to TRH, VIP substantially increased global [cAMP] 20-fold to ∼5 μM, whereas [cAMP] as measured by Epac2AC8D416N only increased twofold to ∼4 μM (Fig. 5D, red and grey traces). Since this modest fold change in [cAMP] temporally followed, and did not exceed the sustained global [cAMP], it is likely to be caused by cAMP diffusion from the global compartment rather than by direct activation of AC activity within the microdomain. These observations could be explained either by a general Gs insensitivity of AC8 within Ca2+-release-stimulated cAMP microdomains, or by the exclusion of the VIP receptor and Gs proteins, as presented in the model (Fig. 5E). Both views were supported by the results obtained with pmEpac2. About 1 minute after stimulation with 10 nM VIP, [cAMP] in the vicinity of pmEpac2 exceeded 20 μM (Fig. 5D, black trace), which indicates an extreme proximity to the ACs activated by the VIP-Gs pathway. It is also possible that PDE4 is activated as a result of cAMP elevation (Willoughby et al., 2006) by VIP, potentially limiting diffusion of cAMP from its site of synthesis into the AC8 compartment and the global compartment, as indicated in the model (Fig. 5E).

We propose a high degree of organization for the cAMP signalling system in GH3B6 cells, although a detailed dissection of the architecture is beyond the scope of our current investigation. However, a considerable body of evidence suggests that the regulation of AC8 by CCE in HEK293 cells relies on its residence in lipid rafts (Fagan et al., 1996; Pagano et al., 2009; Smith et al., 2002), which are regions within the plasma membrane that are enriched in cholesterol and glycosphingolipids (Rajendran and Simons, 2005; Simons and Ikonen, 1997). We therefore wondered whether AC8 might also be targeted to rafts in GH3B6 cells. To address this issue, we asked whether the plasma membrane localization of pmEpac2 and Epac2AC8D416N in GH3B6 required the presence of cholesterol, by treating GH3B6 cells with sphingomyelinase, which specifically degrades sphingomyelins. This results in the assimilation of cholesterol and associated proteins from the plasma membrane to the ER, and disrupts the integrity of potential lipid rafts (Neufeld et al., 1996). Treatment of GH3B6 cells with sphingomyelinase led to a striking internalization of Epac2AC8D416N (Fig. 6A). In control conditions, the majority of both sensors occurred in the plasma membrane, as indicated by fluorescence intensity ratios PM/CYT (mean ± s.e.m.) of 19.4±2.5 and 13.7±1.1 for Epac2AC8D416N and pmEpac2, respectively (Fig. 6B). However, whereas treatment with sphingomyelinase did not significantly affect pmEpac2 (11.6±1.1; Fig. 6B), the PM/CYT ratio for Epac2AC8D416N was significantly lowered, by ∼50% (10.2±1.3; Fig. 6B). These findings suggest that indeed Epac2AC8D416N resides in rafts in GH3B6 cells. In addition, the difference in the extent of internalization indicates that pmEpac2 and Epac2AC8D416N do not reside in identical lipid environments, which goes some way to explaining their different reporting of cAMP.

Fig. 6.

Plasma membrane localization of Epac2AC8D416N requires cholesterol. (A) GH3B6 cells transiently expressing pmEpac2 (top) or Epac2AC8D416N (bottom). Expression of the sensors was detected by recording the YFP emission of the sensors (middle panels). The plasma membrane was labelled using wheatgerm agglutinin conjugated to Alexa Fluor 555 (WGA, left panels). Merged images (right panels) also show the DAPI-stained nuclei. Scale bars: 10 μm. A representative cell is shown in each panel for control cells and cells treated with sphingomyelinase (SMase). (B) Calculated ratio PM/CYT (see supplementary material Fig. S4) of fluorescence intensity associated with plasma membrane (PM) and the cytoplasm (CYT) for Epac2AC8D416N (control, n=40; SMase, n=35), and pmEpac2 (control, n=34; SMase, n=32). Statistical differences were analysed using the unpaired t-test (**P<0.01; n.s. indicates no significant difference).

Fig. 6.

Plasma membrane localization of Epac2AC8D416N requires cholesterol. (A) GH3B6 cells transiently expressing pmEpac2 (top) or Epac2AC8D416N (bottom). Expression of the sensors was detected by recording the YFP emission of the sensors (middle panels). The plasma membrane was labelled using wheatgerm agglutinin conjugated to Alexa Fluor 555 (WGA, left panels). Merged images (right panels) also show the DAPI-stained nuclei. Scale bars: 10 μm. A representative cell is shown in each panel for control cells and cells treated with sphingomyelinase (SMase). (B) Calculated ratio PM/CYT (see supplementary material Fig. S4) of fluorescence intensity associated with plasma membrane (PM) and the cytoplasm (CYT) for Epac2AC8D416N (control, n=40; SMase, n=35), and pmEpac2 (control, n=34; SMase, n=32). Statistical differences were analysed using the unpaired t-test (**P<0.01; n.s. indicates no significant difference).

We conclude that different compartments for cAMP can exist in a simple cell, and that different factors can impact the cAMP formed. For example, AC8 appears to be quite insensitive to VIP, however it can respond positively to TRH-mediated release of Ca2+. By contrast, a different subset of ACs (AC5 and AC6) in the same cells was highly responsive to VIP and inhibited by the same rise in Ca2+ that stimulates AC8 (Fig. 5E).

It must be recognized that in the intact pituitary tissue from which GH3B6 cells are derived (Ooi et al., 2004), distinct plasma membrane domains might be exposed to different surfaces and sources of VIP or TRH. The compartmentalization that we see in dissociated single cells would be heavily reinforced by the rigid architectural design in the pituitary tissue (Perez et al., 1995).

Our approach opens up the possibility of exploring the organization of cAMP dynamics even in the intact pituitary gland. In this regard, a major challenge would be delivery of the probes. Whether they have to be provided in a virus system, or whether transgenic strategies (e.g. Calebiro et al., 2009) would be required has to be investigated. Distinct from GH3B6 cells, which display a mixed somatolactotrope character (Tashjian, 1979), cells in the anterior pituitary are highly specialized. Five morphologically distinct endocrine cell types have been described, each producing specific hormones. Accordingly, they are classified as corticotropes, thyrotropes, gonadotropes, somatotropes and lactotropes (Ooi et al., 2004). This fact could be useful, because the cAMP sensors could be designed in such a way that their expression is controlled by the promoter for particular hormones, which would allow the study of cAMP signalling in distinct cell types.

In GH3B6 cells, the sophisticated organization of cAMP signalling might help to explain the physiological actions of hormones. Although cAMP generated upon stimulation with VIP can activate the secretion of PRL (Guild and Drummond, 1983; Onali et al., 1983), no such relationship was found for TRH. The main action of TRH on PRL secretion was instead attributed to elevations in [Ca2+]cyt, and the activation of PKC (Gershengorn, 1986). The increase in [cAMP] mediated by Ca2+ release observed in our system was independent of PKC activation, which led us to speculate about the downstream impact of Ca2+-release-regulated cAMP signalling on an event other than PRL secretion. Cyclic AMP in GH3B6 cells can regulate gene expression; phosphorylation of cAMP-dependent transcription factors, such as cAMP response element-binding protein by cAMP-dependent protein kinase has been reported (Coleman et al., 1996; Hayakawa et al., 2002; Lin et al., 1996). Likewise, TRH does impact gene expression, because it directly stimulates the promoter of the gene encoding PRL via activation of MAPK (Wang and Maurer, 1999). Interestingly, elevation of [cAMP]i can also activate MAPK by phosphorylation via cAMP-dependent protein kinase (Kievit et al., 2001). We can therefore envisage a role for TRH-mediated cAMP signalling in the regulation of PRL expression.

In this regard, compartmentalized signalling centred on AC8 might require AC8, PKA, CREB and PDEs, as well as phosphatases to be assembled in signalling complexes within the microdomain. Such a proposal is quite speculative at present because it is not known whether such complexes exist or how these domains of cAMP are maintained. However, evidence is accumulating that ACs interact with a variety of proteins, such as A-kinase-anchoring proteins, protein phosphatase 2A and the actin cytoskeleton (Bauman et al., 2006; Crossthwaite et al., 2006; Head et al., 2006; Piggott et al., 2008). Additionally, the assembly of signalling complexes is often accompanied by their residence in lipid rafts (Pike, 2003). Whether lipid rafts exist in GH3B6 cells is uncertain; however, the significance of the plasma membrane cholesterol content on GH3B6 cell physiology has been demonstrated (Lin et al., 2006; Zyzek et al., 1983). Consequently, the present finding that the expression of Epac2AC8D416N at the plasma membrane can be disrupted by sphingomyelinase in GH3B6 cells extends the evidence from studies in HEK293 cells that AC8 exists in raft domains and this could contribute to the organization of the AC8-centred cAMP microdomains (Fagan et al., 1996; Pagano et al., 2009; Smith et al., 2002).

It seems reasonable to anticipate that future studies exploiting differentially targeted, genetically encoded sensors for cAMP will expose other distinctly regulated cAMP micro-compartments in the same cell, which would be the first step in their dissection.

Materials

Protein molecular size standards and acrylamide:bisacrylamide 37.5:1 solution were from Bio-Rad (Hercules, CA). Restriction enzymes, DNA T4 ligase and calf intestinal phosphatase (CIP) were from New England Biolabs (Ipswich, MA). All radiochemicals were from Perkin Elmer Life Sciences (Beaconsfield, UK). The rabbit polyclonal anti-AC8 antibody was a kind gift from Jim Cali (Promega, Madison, WI). The rabbit polyclonal antibody anti-calnexin (ab22595) was from Abcam (Cambridge, UK). Thapsigargin was from Merck (Nottingham, UK). Oligonucleotides were from Sigma-Genosys (Haverhill, UK). All other chemicals, where not indicated, were from Sigma (Poole, UK).

Construction of pmEpac2, Epac2AC8, Epac2AC8D416N and the R297E mutants

A DNA fragment encoding globalEpac2 was generated by PCR using the Epac2-camps cDNA (Nikolaev et al., 2004a) as template. Primer sequences were: forward, 5′-GTACCGCCACCATGGTGAGCAAGGGCGAG-3′; and reverse, 5′-GTACCCTTGTACAGCTCGTC-3′. Subsequently, the PCR fragment was blunt-end cloned into pcDNA3-AC8(rat) (Cali et al., 1994) opened via a KpnI restriction site located within the 5′ non-translated region of the AC8 cDNA. The resulting construct (Epac2AC8) encodes a fusion protein between the Epac2-camps sensor C-terminally linked via 48 amino acids (GTGPCSGLLQSSSLGKRGSLPSGLRVSKNPYPQLPQPVSEPPSLRTPA) to the full-length AC8 protein. The D416N mutation of AC8, as well as the R297E mutations of Epac2, were introduced by site-directed mutagenesis according to the QuikChange protocol (Stratagene, La Jolla, CA) using the fusion high-fidelity polymerase kit (Finnzymes) according to manufacturers' instructions. The pmEpac2 sensor was generated as follows: the sequence encoding the `SH4' motif (MGCINSKRKD) of Lyn kinase followed by a NheI restriction site was introduced into pcDNA3 between BamHI and EcoRI using the annealed product of the oligonucleotides 5′-GATCCATGGGATGTATCAATAGCAAGCGCAAAGATGCTAGCAAAG-3′ and 5′-AATTCTTTGCTAGCATCTTTGCGCTTGCTATTGATACATCCCATG-3′. Second, a DNA fragment encoding EFYP was amplified by PCR and subcloned between NheI and EcoRI sites. Primer sequences were: forward, 5′-AAAGCTAGCATGGTGAGCAAGGGCGAG-3′ and reverse, 5′-AAAGGATCCCTTGTACAGCTCGTCCATGCCGA-3′. Finally, the remaining part of Epac2-camps (Epac2 cAMP-binding site and ECFP) was subcloned from the original plasmid using EcoRI and NotI.

RT-PCR experiments

Reverse transcription PCR reactions were performed by using SuperscriptII Reverse Transcriptase (Invitrogen, Paisley, UK) and 1-2 μg total cellular RNA as template isolated with the SV Total RNA Isolation system (Promega, Madison, WI). The PCR primers were designed to generate intron-spanning cDNA-fragments on the basis of published rat cDNA- and genomic sequences. Primer sequences and fragment sizes are as follows: AC1 (forward, 5′-CACTTGCTGGTCACGGC-3′ and reverse, 5′-GTCTCAATGTTATGAGTTTTC-3′; 835 bp), AC2 (forward, 5′-CATCATTGCCAGCATCCTC-3′ and reverse, 5′-GCTTCGCACATATCCAATCC-3′; 629 bp), AC5 (forward, 5′-CACCATCTACACGCTGCTG-3′ and reverse, 5′-GACCAACGAGATGGCCTC-3′; 642 bp), AC6 (forward, 5′-CGACATAGCTTCCGCCAG-3′ and reverse, 5′-CATCTCCACACAGCAGTGG-3′; 752 bp), and AC8 (forward, 5′-CAGAGGCGCAAATCGGAC-3′ and reverse, 5′-GCTCAGGCAGTCCTGACA-3′; 891 bp). Control PCR reactions were performed using 50 ng plasmid DNA containing the respective rat AC cDNA. PCR products were resolved on a 1% agarose gel containing Safe View nucleic acid stain (NBS Biologicals, Huntingdon, UK).

Cell culture and transfection

Human embryonic kidney cells (HEK293, European Collection of Cell Cultures, Porton Down, UK) were grown in minimum essential medium supplemented with 10% (v/v) fetal bovine serum and maintained at 37°C in a humidified atmosphere of 95% air and 5% CO2. GH3B6 cells (American Type Culture Collection, Rockville, MD) were grown in stock monolayer cultures in plastic culture flasks, as described (Tashjian, 1979). For transient transfections, HEK293 or GH3B6 cells were plated onto 90 mm dishes or 25 mm poly-L-lysine-coated coverslips, respectively, at 60% confluence ∼24 hours before transfection with 1-2 μg plasmid DNA using lipofectamine™ 2000 (Invitrogen) according to the manufacturer's instructions. Cells were used ∼48 hours after transfection. About 35% of transfected GH3B6 cells displayed detectable expression of the various probes (globalEpac2, ∼38%; pmEpac2, ∼34%; Epac2AC8D416N, ∼37%).

In vitro FRET measurements

48 hours after transfection, HEK293 cells were washed twice with ice-cold phosphate-buffered saline, scraped from the dishes, and centrifuged for 5 minutes at 200 g (RT). Cell pellets were resuspended in 5 mM Tris-HCl, 2 mM EDTA (pH 7.3) and lysed by passaging through a 21-gauge needle. Lysates from cells transfected with either Epac2AC8 or Epac2AC8D416N were then centrifuged for 30 minutes at 12,000 g (4°C). The insoluble pellet was resuspended in 5 mM Tris-HCl, 2 mM EDTA (pH 7.3). After the addition of an equal volume of a 2% CHAPS solution, tubes were briefly vortexed and incubated on ice for 30 minutes to solubilize membrane proteins. Initial lysates from cells transfected with globalEpac2, pmEpac2 or the solubilized membrane proteins from cells transfected with either Epac2AC8, or Epac2AC8D416N were subjected to ultracentrifugation (50,000 r.p.m., Beckmann SW55TI rotor, 20 minutes, 4°C). Subsequently, the fluorescence emission spectra of the supernatants (excitation at 436±8 nm, emission from 450-550) were measured in a Perkin-Elmer LS50B spectrofluorimeter, before and after adding increasing [cAMP]. [cAMP] was spectrofluorometrically determined at λ259nm.

Confocal imaging

Cells expressing Epac2-based cAMP sensors were plated onto glass-bottom dishes (WillcoWells, Amsterdam, The Netherlands) coated with poly-L-lysine. The cells were visualized on a Leica SP5 confocal microscope running LAS AF 1.6.0 software using a ×63 plan apochromat 1.4 NA oil-immersion objective (Leica Microsystems, Milton Keynes, UK).

To stain the plasma membrane of HEK293 cells, cells were incubated in 5 μg/ml Cell Mask™ Deep Red plasma membrane stain (Invitrogen) in Krebs buffer for 5 minutes at 37°C. During fluorescence imaging, cells were incubated at 37°C in HBS buffer containing 1.8 mM CaCl2. For visualization of the CFP fluorescence, a 458 nm laser line of an Argon laser was used with an emission bandwidth of 467-504 nm. For imaging using Cell Mask™ Deep Red plasma membrane stain, a 633 nm laser line of a HeNe laser was used with an emission bandwidth of 650-700 nm.

For internalization of pmEpac2 and Epac2A8D416N in GH3B6 cells (Fig. 6 and supplementary material Fig. S4), cells were plated as described above. 48 hours after transfection, cells were incubated with 200 mU sphingomyelinase (Sigma) in HBS buffer containing 1.8 mM Ca2+ for 90 minutes at 37°C. Control cells were incubated in HBS only. Subsequently, cells were incubated for 5 minutes at RT in PBS containing the plasma membrane marker wheatgerm agglutinin (1:200) conjugated to the Alexa Fluor 555 fluorophore, and fixed in 4% paraformaldehyde for 10 minutes at RT. YFP fluorescence was visualized upon excitation with a 514 nm laser line of an Argon laser using an emission bandwidth of 524-573 nm. For imaging of Alexa Fluor 555, a 543 nm laser line of a HeNe laser was used with an emission bandwidth of 586-692 nm.

Preparation of membrane proteins

Membrane proteins from transfected HEK293 and GH3B6 cells were prepared and immunoblotted as described previously (MacDougall et al., 2009; Wachten et al., 2006). Isolation of crude membranes, measurement of AC activity and CCE-induced cAMP accumulation in cell populations have been described elsewhere (Masada et al., 2009). Data points are presented as mean activities ± s.d. of triplicate determinations.

Single-cell Ca2+ measurements

These were performed as described previously (Willoughby and Cooper, 2006).

Single-cell FRET measurements to monitor cAMP

Fluorescence imaging of Epac2-based cAMP sensor-expressing HEK293 or GH3B6 cells was performed using an Andor Ixon+ camera and an Optosplit (505DC) to separate CFP (470 nm) and YFP (535 nm) emission images (Cairn Research). For dual emission-ratio imaging, cells were excited at 436 nm using a monochromator (Cairn Research) and 51017 filter set (Chroma, Rockingham, VT, USA) attached to a Nikon eclipse TE2000-S microscope (×40 objective). Emission images at 470 nm and 535 nm were collected at 0.1-0.2 Hz (200-500 millisecond integration times). Captured pictures were background subtracted, corrected for the spectral bleed-through of CFP into the YFP channel, and analysed with Metamorph imaging software (Molecular Devices). FRET data are plotted as changes in 470 nm (CFP) versus 535 nm (YFP) emission ratio (R) relative to the fluorescence ratio at 0 minutes (R0).

Calculation of [cAMP]

To calculate the [cAMP] detected by the Epac2-based sensors, we used the following relationship as described (Iancu et al., 2008):
\[\ [cAMP]=EC_{50}\left(\frac{F_{abs}(F+1)+F-F_{\mathrm{max}}}{F_{abs}(F_{\mathrm{max}}-F)-F+F_{\mathrm{max}}}\right)^{\frac{1}{n_{H}}},\ \]
(1)
where Fabs=maximal change in FRET measured in vitro using saturating [cAMP], Fmax=maximal change in CFP/YFP (R/R0) measured in single cells, and nH=Hill coefficient of the fitted in vitro calibration (Fig. 2D).

Curve fitting and statistical analysis

Sigmoidal dose-response curves, linear regressions, and kinetic parameters were calculated using GraphPad Prism Version 4 (GraphPad Software, La Jolla, CA). Data points of representative experiments are shown as mean ± s.d. of triplicate determinations. Data from three or more experiments are expressed as mean ± s.e.m., as stated. Correlation analysis was carried out applying the two-tailed Pearson algorithm (supplementary material Fig. S1Ci-iv).

[Ca2+]cyt measurements in cell populations

[Ca2+]cyt of Fura-2-loaded HEK293 cells was detected as described previously (Chiono et al., 1995).

This work was supported by The Wellcome Trust (RG 31760). D.M.F.C. is a Royal Society Wolfson Research Fellow. We thank Michelle Halls and Debbie Willoughby for careful revision of the manuscript. Deposited in PMC for release after 6 months.

Aragay, A. M., Katz, A. and Simon, M. I. (
1992
). The G alpha q and G alpha 11 proteins couple the thyrotropin-releasing hormone receptor to phospholipase C in GH3 rat pituitary cells.
J. Biol. Chem.
267
,
24983
-24988.
Bacskai, B. J., Hochner, B., Mahaut-Smith, M., Adams, S. R., Kaang, B. K., Kandel, E. R. and Tsien, R. Y. (
1993
). Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons.
Science
260
,
222
-226.
Bauman, A. L., Soughayer, J., Nguyen, B. T., Willoughby, D., Carnegie, G. K., Wong, W., Hoshi, N., Langeberg, L. K., Cooper, D. M. F., Dessauer, C. W. et al. (
2006
). Dynamic regulation of cAMP synthesis through anchored PKA-adenylyl cyclase V/VI complexes.
Mol. Cell
23
,
925
-931.
Bjoro, T., Ostberg, B. C., Sand, O., Gordeladze, J., Iversen, J. G., Torjesen, P. A., Gautvik, K. M. and Haug, E. (
1987
). Vasoactive intestinal peptide and peptide with N-terminal histidine and C-terminal isoleucine increase prolactin secretion in cultured rat pituitary cells (GH4C1) via a cAMP-dependent mechanism which involves transient elevation of intracellular Ca2+.
Mol. Cell Endocrinol.
49
,
119
-128.
Boyajian, C. L. and Cooper, D. M. F. (
1990
). Potent and cooperative feedback inhibition of adenylate cyclase activity by calcium in pituitary-derived GH3 cells.
Cell Calcium
11
,
299
-307.
Boyajian, C. L., Garritsen, A. and Cooper, D. M. F. (
1991
). Bradykinin stimulates Ca2+ mobilization in NCB-20 cells leading to direct inhibition of adenylylcyclase. A novel mechanism for inhibition of cAMP production.
J. Biol. Chem.
266
,
4995
-5003.
Calebiro, D., Nikolaev, V. O., Gagliani, M. C., de Filippis, T., Dees, C., Tacchetti, C., Persani, L. and Lohse, M. J. (
2009
). Persistent cAMP-signals triggered by internalized G-protein-coupled receptors.
PLoS Biol.
7
,
e1000172
.
Cali, J. J., Zwaagstra, J. C., Mons, N., Cooper, D. M. F. and Krupinski, J. (
1994
). Type VIII adenylyl cyclase. A Ca2+/calmodulin-stimulated enzyme expressed in discrete regions of rat brain.
J. Biol. Chem.
269
,
12190
-12195.
Chiono, M., Mahey, R., Tate, G. and Cooper, D. M. F. (
1995
). Capacitative Ca2+ entry exclusively inhibits cAMP synthesis in C6-2B glioma cells. Evidence that physiologically evoked Ca2+ entry regulates Ca2+-inhibitable adenylyl cyclase in non-excitable cells.
J. Biol. Chem.
270
,
1149
-1155.
Chou, J. L., Huang, C. L., Lai, H. L., Hung, A. C., Chien, C. L., Kao, Y. Y. and Chern, Y. (
2004
). Regulation of type VI adenylyl cyclase by Snapin, a SNAP25-binding protein.
J. Biol. Chem.
279
,
46271
-46279.
Coleman, D. T., Chen, X., Sassaroli, M. and Bancroft, C. (
1996
). Pituitary adenylate cyclase-activating polypeptide regulates prolactin promoter activity via a protein kinase A-mediated pathway that is independent of the transcriptional pathway employed by thyrotropin-releasing hormone.
Endocrinology
137
,
1276
-1285.
Cooper, D. M. F., Mons, N. and Karpen, J. W. (
1995
). Adenylyl cyclases and the interaction between calcium and cAMP signalling.
Nature
374
,
421
-424.
Crossthwaite, A. J., Ciruela, A., Rayner, T. F. and Cooper, D. M. F. (
2006
). A direct interaction between the N terminus of adenylyl cyclase AC8 and the catalytic subunit of protein phosphatase 2A.
Mol. Pharmacol.
69
,
608
-617.
de Rooij, J., Zwartkruis, F. J. T., Verheijen, M. H. G., Cool, R. H., Nijman, S. M. B., Wittinghofer, A. and Bos, J. L. (
1998
). Epac is a Rap1 guanine-nucleotide-exchange factor directly activated by cyclic AMP.
Nature
396
,
474
-477.
DiPilato, L. M., Cheng, X. and Zhang, J. (
2004
). Fluorescent indicators of cAMP and Epac activation reveal differential dynamics of cAMP signaling within discrete subcellular compartments.
Proc. Natl. Acad. Sci. USA
101
,
16513
-16518.
Dupre, D. J., Baragli, A., Rebois, R. V., Ethier, N. and Hebert, T. E. (
2007
). Signalling complexes associated with adenylyl cyclase II are assembled during their biosynthesis.
Cell. Signal.
19
,
481
-489.
Evans, T., Smith, M. M., Tanner, L. I. and Harden, T. K. (
1984
). Muscarinic cholinergic receptors of two cell lines that regulate cyclic AMP metabolism by different molecular mechanisms.
Mol. Pharmacol.
26
,
395
-404.
Fagan, K. A., Mahey, R. and Cooper, D. M. F. (
1996
). Functional co-localization of transfected Ca2+-stimulable adenylyl cyclases with capacitative Ca2+ entry sites.
J. Biol. Chem.
271
,
12438
-12444.
Fagan, K. A., Mons, N. and Cooper, D. M. F. (
1998
). Dependence of the Ca2+-inhibitable adenylyl cyclase of C6-2B glioma cells on capacitative Ca2+ entry.
J. Biol. Chem.
273
,
9297
-9305.
Fagan, K. A., Smith, K. E. and Cooper, D. M. F. (
2000
). Regulation of the Ca2+-inhibitable adenylyl cyclase type VI by capacitative Ca2+ entry requires localization in cholesterol-rich domains.
J. Biol. Chem.
275
,
26530
-26537.
Gautvik, K. M., Gordeladze, J. O., Jahnsen, T., Haug, E., Hansson, V. and Lystad, E. (
1983
). Thyroliberin receptor binding and adenylyl cyclase activation in cultured prolactin-producing rat pituitary tumor cells (GH cells).
J. Biol. Chem.
258
,
10304
-10311.
Gershengorn, M. C. (
1986
). Mechanism of thyrotropin releasing hormone stimulation of pituitary hormone secretion.
Annu. Rev. Physiol.
48
,
515
-526.
Guild, S. and Drummond, A. H. (
1983
). Adenosine 3′,5′-cyclic monophosphate-dependent release of prolactin from GH3 pituitary tumour cells. A quantitative analysis.
Biochem. J.
216
,
551
-557.
Guild, S. and Drummond, A. H. (
1984
). Vasoactive-intestinal-polypeptide-stimulated adenosine 3′,5′-cyclic monophosphate accumulation in GH3 pituitary tumour cells. Reversal of desensitization by forskolin.
Biochem. J.
221
,
789
-796.
Hayakawa, J., Ohmichi, M., Tasaka, K., Kanda, Y., Adachi, K., Nishio, Y., Hisamoto, K., Mabuchi, S., Hinuma, S. and Murata, Y. (
2002
). Regulation of the PRL promoter by Akt through cAMP response element binding protein.
Endocrinology
143
,
13
-22.
Head, B. P., Patel, H. H., Roth, D. M., Murray, F., Swaney, J. S., Niesman, I. R., Farquhar, M. G. and Insel, P. A. (
2006
). Microtubules and actin microfilaments regulate lipid raft/caveolae localization of adenylyl cyclase signaling components.
J. Biol. Chem.
281
,
26391
-26399.
Hempel, C. M., Vincent, P., Adams, S. R., Tsien, R. Y. and Selverston, A. I. (
1996
). Spatio-temporal dynamics of cyclic AMP signals in an intact neural circuit.
Nature
384
,
166
-169.
Herbert, J. M., Augereau, J. M., Gleye, J. and Maffrand, J. P. (
1990
). Chelerythrine is a potent and specific inhibitor of protein kinase C.
Biochem. Biophys. Res. Comm.
172
,
993
-999.
Herget, S., Lohse, M. J. and Nikolaev, V. O. (
2008
). Real-time monitoring of phosphodiesterase inhibition in intact cells.
Cell. Signal.
20
,
1423
-1431.
Iancu, R. V., Ramamurthy, G., Warrier, S., Nikolaev, V. O., Lohse, M. J., Jones, S. W. and Harvey, R. D. (
2008
). Cytoplasmic cAMP concentrations in intact cardiac myocytes.
Am. J. Physiol. Cell Physiol.
295
,
C414
-C422.
Kaupp, U. B. and Seifert, R. (
2001
). Molecular diversity of pacemaker ion channels.
Annu. Rev. Physiol.
63
,
235
-257.
Kievit, P., Lauten, J. D. and Maurer, R. A. (
2001
). Analysis of the role of the mitogen-activated protein kinase in mediating cyclic-adenosine 3′,5′-monophosphate effects on prolactin promoter activity.
Mol. Endocrinol.
15
,
614
-624.
Kiley, S. C., Parker, P. J., Fabbro, D. and Jaken, S. (
1991
). Differential regulation of protein kinase C isozymes by thyrotropin-releasing hormone in GH4C1 cells.
J. Biol. Chem.
266
,
23761
-23768.
Kretschmannova, K., Gonzalez-Iglesias, A. E., Tomic, M. and Stojilkovic, S. S. (
2006
). Dependence of hyperpolarisation-activated cyclic nucleotide-gated channel activity on basal cyclic adenosine monophosphate production in spontaneously firing GH3 cells.
J. Neuroendocrinol.
18
,
484
-493.
Lin, J.-H., Jang, Y.-C., Wen, D.-C. and Wang, F.-F. (
1996
). Synergistic activation of cAMP and calcium on cAMP-response-element-mediated gene expression in GH3 pituitary tumor cells.
Cell. Signal.
8
,
111
-115.
Lin, M. W., Wu, A. Z., Ting, W. H., Li, C. L., Cheng, K. S. and Wu, S. N. (
2006
). Changes in membrane cholesterol of pituitary tumor (GH3) cells regulate the activity of large-conductance Ca2+-activated K+ channels.
Chin. J. Physiol.
49
,
1
-13.
MacDougall, D. A., Wachten, S., Ciruela, A., Sinz, A. and Cooper, D. M. F. (
2009
). Separate elements within a single IQ-like motif in adenylyl cyclase type 8 impart Ca2+/CaM binding and autoinhibition.
J. Biol. Chem.
23
,
15573
-15588.
Martin, A. C. L., Willoughby, D., Ciruela, A., Ayling, L.-J., Pagano, M., Wachten, S., Tengholm, A. and Cooper, D. M. F. (
2009
). Capacitative Ca2+ entry via Orai1 and stromal interacting molecule 1 (STIM1) regulates adenylyl cyclase type 8.
Mol. Pharmacol.
75
,
830
-842.
Masada, N., Ciruela, A., Macdougall, D. A. and Cooper, D. M. F. (
2009
). Distinct mechanisms of regulation by Ca2+/calmodulin of type 1 and 8 adenylyl cyclases support their different physiological roles.
J. Biol. Chem.
284
,
4451
-4463.
Mei, F. C., Qiao, J., Tsygankova, O. M., Meinkoth, J. L., Quilliam, L. A. and Cheng, X. (
2002
). Differential signaling of cyclic AMP: opposing effects of exchange protein directly activated by cyclic AMP and cAMP-dependent protein kinase on protein kinase B activation.
J. Biol. Chem.
277
,
11497
-11504.
Neufeld, E. B., Cooney, A. M., Pitha, J., Dawidowicz, E. A., Dwyer, N. K., Pentchev, P. G. and Blanchette-Mackie, E. J. (
1996
). Intracellular trafficking of cholesterol monitored with a cyclodextrin.
J. Biol. Chem.
271
,
21604
-21613.
Nikolaev, V. O., Bunemann, M., Hein, L., Hannawacker, A. and Lohse, M. J. (
2004a
). Novel single chain cAMP sensors for receptor-induced signal propagation.
J. Biol. Chem.
279
,
37215
-37218.
Nikolaev, V. O., Gambaryan, S., Engelhardt, S., Walter, U. and Lohse, M. J. (
2004b
). Real-time monitoring of live cell's PDE2 activity: Hormone-stimulated cAMP hydrolysis is faster than hormone-stimulated cAMP synthesis.
J. Biol. Chem.
3
,
1716
-1719.
Nikolaev, V. O., Bunemann, M., Schmitteckert, E., Lohse, M. J. and Engelhardt, S. (
2006
). Cyclic AMP imaging in adult cardiac myocytes reveals far-reaching beta1-adrenergic but locally confined beta2-adrenergic receptor-mediated signaling.
Circ. Res.
99
,
1084
-1091.
Onali, P., Eva, C., Olianas, M. C., Schwartz, J. P. and Costa, E. (
1983
). In GH3 pituitary cells, acetylcholine and vasoactive intestinal peptide antagonistically modulate adenylate cyclase, cyclic AMP content, and prolactin secretion.
Mol. Pharmacol.
24
,
189
-194.
Ooi, G. T., Tawadros, N. and Escalona, R. M. (
2004
). Pituitary cell lines and their endocrine applications.
Mol. Cell. Endocrinol.
228
,
1
-21.
Pagano, M., Clynes, M. A., Masada, N., Ciruela, A., Ayling, L.-J., Wachten, S. and Cooper, D. M. F. (
2009
). Insights into the residence in lipid rafts of adenylyl cyclase AC8 and its regulation by capacitative calcium entry.
Am. J. Physiol. Cell Physiol.
296
,
C607
-C619.
Parekh, A. B. and Putney, J. W., Jr (
2005
). Store-operated calcium channels.
Physiol. Rev.
85
,
757
-810.
Paulssen, R. H., Paulssen, E. J., Gautvik, K. M. and Gordeladze, J. O. (
1992
). The thyroliberin receptor interacts directly with a stimulatory guanine-nucleotide-binding protein in the activation of adenylyl cyclase in GH3 rat pituitary tumour cells. Evidence obtained by the use of antisense RNA inhibition and immunoblocking of the stimulatory guanine-nucleotide-binding protein.
Eur. J. Biochem.
204
,
413
-418.
Paulssen, R. H., Johansen, P. W., Gordeladze, J. O., Nymoen, O., Paulssen, E. J. and Gautvik, K. M. (
1994
). Cell-specific expression and function of adenylyl cyclases in rat pituitary tumour cell lines.
Eur. J. Biochem.
222
,
97
-103.
Perez, F. M., Rose, J. C. and Schwartz, J. (
1995
). Anterior pituitary cells: getting to know their neighbors.
Mol. Cell Endocrinol.
111
,
C1
-C6.
Piggott, L. A., Bauman, A. L., Scott, J. D. and Dessauer, C. W. (
2008
). The A-kinase anchoring protein Yotiao binds and regulates adenylyl cyclase in brain.
Proc. Natl. Acad. Sci. USA
105
,
13835
-13840.
Pike, L. J. (
2003
). Lipid rafts: bringing order to chaos.
J. Lipid. Res.
44
,
655
-667.
Ponsioen, B., Zhao, J., Riedl, J., Zwartkruis, F., van der Krogt, G., Zaccolo, M., Moolenaar, W. H., Bos, J. L. and Jalink, K. (
2004
). Detecting cAMP-induced Epac activation by fluorescence resonance energy transfer: Epac as a novel cAMP indicator.
EMBO Rep.
5
,
1176
-1180.
Putney, J. W., Jr (
2001
). The pharmacology of capacitative calcium entry.
Mol. Interv.
1
,
84
-94.
Rajendran, L. and Simons, K. (
2005
). Lipid rafts and membrane dynamics.
J. Cell Sci.
118
,
1099
-1102.
Resh, M. D. (
2006
). Trafficking and signaling by fatty-acylated and prenylated proteins.
Nat. Chem. Biol.
2
,
584
-590.
Rich, T. C., Fagan, K. A., Tse, T. E., Schaack, J., Cooper, D. M. F. and Karpen, J. W. (
2001
). A uniform extracellular stimulus triggers distinct cAMP signals in different compartments of a simple cell.
Proc. Natl. Acad. Sci. USA
.
98
,
13049
-13054.
Schlegel, W. and Wollheim, C. B. (
1984
). Thyrotropin-releasing hormone increases cytosolic free Ca2+ in clonal pituitary cells (GH3 cells): direct evidence for the mobilization of cellular calcium.
J. Cell Biol.
99
,
83
-87.
Simons, K. and Ikonen, E. (
1997
). Functional rafts in cell membranes.
Nature
387
,
569
-572.
Smallridge, R. C., Kiang, J. G., Gist, I. D., Fein, H. G. and Galloway, R. J. (
1992
). U-73122, an aminosteroid phospholipase C antagonist, noncompetitively inhibits thyrotropin-releasing hormone effects in GH3 rat pituitary cells.
Endocrinology
131
,
1883
-1888.
Smith, K. E., Gu, C., Fagan, K. A., Hu, B. and Cooper, D. M. F. (
2002
). Residence of adenylyl cyclase type 8 in caveolae is necessary but not sufficient for regulation by capacitative Ca2+ entry.
J. Biol. Chem.
277
,
6025
-6031.
Stevens, T., Nakahashi, Y., Cornfield, D. N., McMurtry, I. F., Cooper, D. M. F. and Rodman, D. M. (
1995
). Ca2+-inhibitable adenylyl cyclase modulates pulmonary artery endothelial cell cAMP content and barrier function.
Proc. Natl. Acad. Sci. USA
92
,
2696
-2700.
Sunahara, R. K. and Taussig, R. (
2002
). Isoforms of mammalian adenylyl cyclase: multiplicities of signaling.
Mol. Interv.
2
,
168
-184.
Tashjian, A. H., Jr (
1979
). Clonal strains of hormone-producing pituitary cells.
Methods Enzymol.
58
,
527
-535.
Tesmer, J. J. G., Sunahara, R. K., Johnson, R. A., Gosselin, G., Gilman, A. G. and Sprang, S. R. (
1999
). Two-metal-ion catalysis in adenylyl cyclase.
Science
.
285
,
756
-760.
Wachten, S., Schlenstedt, J., Gauss, R. and Baumann, A. (
2006
). Molecular identification and functional characterization of an adenylyl cyclase from the honeybee.
J. Neurochem.
96
,
1580
-1590.
Wang, Y.-H. and Maurer, R. A. (
1999
). A role for the mitogen-activated protein kinase in mediating the ability of thyrotropin-releasing hormone to stimulate the prolactin promoter.
Mol. Endocrinol.
13
,
1094
-1104.
Watson, E. L., Jacobson, K. L., Singh, J. C., Idzerda, R., Ott, S. M., DiJulio, D. H., Wong, S. T. and Storm, D. R. (
2000
). The type 8 adenylyl cyclase is critical for Ca2+ stimulation of cAMP accumulation in mouse parotid acini.
J. Biol. Chem.
275
,
14691
-14699.
Willoughby, D. and Cooper, D. M. F. (
2006
). Ca2+ stimulation of adenylyl cyclase generates dynamic oscillations in cyclic AMP.
J. Cell Sci.
119
,
828
-836.
Willoughby, D. and Cooper, D. M. F. (
2007
). Organization and Ca2+ regulation of adenylyl cyclases in cAMP microdomains.
Physiol. Rev.
87
,
965
-1010.
Willoughby, D. and Cooper, D. M. F. (
2008
). Live-cell imaging of cAMP dynamics.
Nat. Meth.
5
,
29
-36.
Willoughby, D., Wong, W., Schaack, J., Scott, J. D. and Cooper, D. M. F. (
2006
). An anchored PKA and PDE4 complex regulates subplasmalemmal cAMP dynamics.
EMBO J.
25
,
2051
-2061.
Winiger, B. P. and Schlegel, W. (
1988
). Rapid transient elevations of cytosolic calcium triggered by thyrotropin releasing hormone in individual cells of the pituitary line GH3B6.
Biochem. J.
255
,
161
-167.
Yoshimura, M. and Cooper, D. M. F. (
1993
). Type-specific stimulation of adenylylcyclase by protein kinase C.
J. Biol. Chem.
268
,
4604
-4607.
Zaccolo, M. and Pozzan, T. (
2002
). Discrete microdomains with high concentration of cAMP in stimulated rat neonatal cardiac myocytes.
Science
295
,
1711
-1715.
Zacharias, D. A., Violin, J. D., Newton, A. C. and Tsien, R. Y. (
2002
). Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells.
Science
296
,
913
-916.
Zyzek, E., Desmazes, J. P., Gourdji, D., Dufy, B. and Georgescauld, D. (
1983
). Modifications of membrane cholesterol content affects electrical properties and prolactin release of cultured pituitary cells.
Experientia
39
,
56
-58.

Supplementary information