Mechanism of Regulation of the Epac Family of cAMP-dependent RapGEFs*

Epac1 (cAMP-GEFI) and Epac2 (cAMP-GEFII) are closely related guanine nucleotide exchange factors (GEFs) for the small GTPase Rap1, which are directly regulated by cAMP. Here we show that both GEFs efficiently activate Rap2 as well. A third member of the family, Repac (GFR), which lacks the cAMP dependent regulatory sequences, is a constitutive activator of both Rap1 and Rap2. In contrast to Epac1, Epac2 contains a second cAMP binding domain at the N terminus, as does the Epac homologue from Caenorhabditis elegans. Affinity measurements show that this distal cAMP binding domain (the A-site) binds cAMP with much lower affinity than the cAMP binding domain proximal to the catalytic domain (the B-site), which is present in both Epac1 and Epac2. Deletion mutant analysis shows that the high affinity cAMP binding domains are sufficient to regulate the GEFs in vitro. Interestingly, isolated fragments containing the B-sites of either Epac1 or Epac2, but not the A-site from Epac2, inhibit the catalytic domains in trans. This inhibition is relieved by the addition of cAMP. In addition to the cAMP binding domains, both Epac1 and Epac2 have a DEP domain. Deletion of this domain does not affect regulation of Epac1 activity but affects membrane localization. From these results, we conclude that all three members of the Epac family regulate both Rap1 and Rap2. Furthermore, we conclude that the catalytic activity of Epac1 is constrained by a direct interaction between GEF and high affinity cAMP binding domains in the absence of cAMP. Epac1 becomes activated by a release of this inhibition when cAMP is bound.

Rap1 is a small GTPase closely related to Ras and implicated in the regulation of a variety of cellular processes including the control of platelet activation, T-cell anergy, B-cell activation, and neuronal differentiation (1,2). Very recently, Rap was shown to be involved in the control of cell adhesion (3), in particular the regulation of integrin activation by inside-out signaling (4). A large variety of extracellular stimuli, including growth factors, cytokines, and cell adhesion molecules, regulate Rap1 (5)(6)(7)(8)(9)(10). The intracellular pathways that activate Rap1 include common second messengers like cyclic adenosine monophosphate (cAMP), calcium, and diacylglycerol (DAG) 1 (11). A number of guanine nucleotide exchange factors (GEFs) have been identified that mediate the activation of Rap1. The first described Rap1GEF, C3G, is found in a complex with the protooncogene product c-Crk and may activate Rap1 as a consequence of complex formation and translocation induced by receptor tyrosine kinase signaling (12). CalDAG-GEFI has calcium binding EF-hands and a domain that resembles C1type DAG binding domains and may explain the activation of Rap1 by these two second messengers (13). Recently, another type of GEF for Rap1, called PDZ-GEF1 (14), nRapGEP (15), or RaGEF (16) was described. This GEF contains, in addition to the catalytic region, a Ras binding domain, which may interact directly with Ras and Rap1 in vitro, a PDZ domain that drives membrane association and a domain that is related to cAMP binding domains (RCBD) but does not bind cAMP. The most intriguing RapGEF, however, is Epac (exchange protein directly activated by cAMP), because this GEF represents a novel target for cAMP, independent from the classical target protein protein kinase A (PKA) (17,18).
Epac was identified in the data base as a genomic sequence with homology to cAMP-binding sites as well as GEFs for Ras-like proteins (17). Independently, the same protein, named cAMP-GEFI, was found in a differential display screen for novel brain-enriched genes related to signaling in the striatum (18). Epac contains a C-terminal catalytic region that activates Rap1 but not Ras, Ral, or R-ras. This region comprises the enzymatic GEF domain and the Ras exchange motif (REM), which is needed for stability of the GEF domain. The N-terminal part of the protein contains a DEP domain, of which the function is unclear, and a cAMP binding domain that is similar to the cAMP binding domains in the regulatory subunit of PKA. cAMP is required for the activation of full-length Epac in vitro, and deletion of the regulatory N-terminal part containing the cAMP binding domain results in the constitutive activation of Epac, indicating that it serves as an auto-inhibitory domain. In addition to Epac (from now on called Epac1), a second, closely related protein has been identified named Epac2 (or cAMP-GEFII (18)) as well as a related protein named Repac (for related to Epac) (or GRF (19)), which lacks the regulatory sequences present in Epac1 and Epac2.
Here we have studied the regulation and function of the different Epac family members in more detail. First, we observe that all three members activate, in addition to Rap1, the close relative Rap2. Second, we identified an additional cAMPbinding site in Epac2, located N-terminal to the DEP domain. Third, mutant analysis revealed that the cAMP binding domains proximal to the catalytic domains in Epac1 and Epac2 (the B-sites) function as inhibitors of the GEF domains in the absence of cAMP. Finally, we show that the DEP domain is involved in membrane localization of Epac1 independent from cAMP signaling.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Constructs used for expression of GEFs and small GTPases in mammalian cells are cloned in the PMT2-SM-HA eukaryotic expression vector. Epac1 constructs are derived from human cDNA, and Epac2 constructs are derived from murine cDNA. HA-Epac1-DDEP contains Epac1 lacking amino acids 71-140, which span the DEP domain. For purification of glutathione S-transferase (GST) fusion constructs, all cDNAs were cloned in pGEX bacterial expression vectors. The catalytic domain of Epac1 contains amino acids 324 -881, the regulatory domain of Epac1 (Epac-RD) contains amino acids 2-329, the cAMP binding domain of Epac1 contains amino acids 149 -318, and Epac1-DDEP contains amino acids 149 -881. The catalytic domain of Epac2 contains amino acids 460 -993, the regulatory domain contains amino acids 1-463, the cAMP-binding site A contains amino acids 1-160, the B-site contains amino acids 280 -463, and Epac2-DDEP contains amino acids 280 -993. The GST fusion construct of Repac contains amino acids 2-580. The catalytic domain CalDAG-GEFI contains amino acids 3-422. The catalytically active PDZ-GEF1 construct contains amino acids 251-1001. The GST-PKA fusion construct used contained the R1a subunit of bovine PKA, lacking amino acids 1-91 (20). Protein production was induced in Bl-21 bacteria using 100 nM isopropyl-1-thio-␤-D-galactopyranoside for 20 h at room temperature. After protein production, bacteria were pelleted and lyzed in ice-cold phosphate-buffered saline containing 1% Triton X-100 and protease inhibitors. The lysate was sonicated three times for 10 s and centrifuged at 10,000 ϫ g to remove insoluble material. GST fusion proteins we purified from the cleared lysate by batchwise incubation with glutathione-agarose beads (Sigma), eluted from the beads in buffer containing 50 mM Tris, pH 7.5, 100 mM NaCl, 10% glycerol, and 10 mM glutathione, and dialyzed for 20 h in the same buffer without glutathione. When indicated, proteins were cleaved from the GST tag by incubation with thrombin and purified by gel filtration. Small GTPases used for in vitro experiments were described elsewhere (21).
In Vivo Activation of Rap-Cells were transfected with HA-tagged Rap1A or Rap2A and serum-starved for 20 h before the activation experiments. Cells were stimulated with forskolin (20 mM) and isobutylmethylxanthine (1 mM) for 10 min. The GTP-bound form of Rap1A was specifically isolated using GST-RalGDS as an activation-specific probe as described (5). Detection on Western blot was by 12CA5 monoclonal antibodies directed against the HA tag. In vivo labeling experiments for Rap2 were performed as described (22). Briefly, serumstarved cells were labeled with [ 32 P]orthophosphate for 5 h. Rap2A was precipitated using 12CA5 antibodies, and nucleotides were eluted and separated on polyethyleneimine-cellulose F thin layer chromatography (TLC) plates. Labeled nucleotides were visualized using a phosphoimager, and GTP/GDP ratios were calculated using the program ImageQuant.
In Vitro Activation of Small GTPases-Experiments were performed as described (21). Briefly, 100 nM purified GTPase (Rap1A or Rap2A) loaded with fluorescently labeled 2Ј,3Ј-bis(O)-N-methylantharanoloyl guanosine diphosphate (mantGDP) was incubated in the presence of excess unlabeled GDP with 50 nM purified GEF unless indicated differently. Release of mGDP was measured in real time as a decrease in fluorescence. To calculate reaction rates, single exponential functions were fit using the program Grafit3.0 (Eritacus). In all in vitro experiments, the Rap1A and Rap2A proteins were used.
Isothermal Titration Calorimetry-Binding of cAMP was investigated by isothermal titration calorimetry (ITC) (MicroCal Inc.). The isolated cAMP binding domains (cleaved from the GST tag) were thermostatted in the cell of the apparatus to 25°C, and cAMP was injected from a syringe in 40 steps up to a 2-4-fold molar excess. The cell contained 1.36 ml of protein solution, and typically, the nucleotide was added in steps of 6 l every 4 min. The data were analyzed using the manufacturer's software.
Cell Fractionation-Cells were scraped in mild lysis buffer (20 mM Hepes, pH 7.4, 5 mM EGTA, 1 mM sodium vanadate, 1 M leupeptin, 1 M aprotinin) and homogenized through a 23-gauche syringe. Intact cells and nuclear components were removed by two rounds of centrifugation at 6000 rpm for 1 min in an Eppendorf table centrifuge. Next, cytosolic and particulate fractions were separated by centrifugation at 50,000 ϫ g at 4°C for 90 min. The particulate fraction was dissolved in buffer containing 1% Triton X-100, 50 mM Hepes, pH 7.5, 50 mM NaCl, 5 mM EDTA, 1 mM sodium vanadate, 1 M leupeptin, and 1 M aprotinin. Distribution of Epac1 in both fractions was analyzed by SDSpolyacrylamide gel electrophoresis and immunoblotting using 12CA5 monoclonal antibody directed against the HA tag. Distribution of the endogenous epidermal growth factor receptor was analyzed using a monoclonal antibody (Transduction Laboratories), and the presence of p42 mitogen-activated protein kinase was investigated using a polyclonal antiserum described earlier (22).

RESULTS
Epac Family Members Activate Both Rap1 and Rap2-Currently, the Epac family of GEFs consists of three members, Epac1 and Epac2, which are regulated by cAMP and Repac (related to Epac), which lacks any apparent regulatory sequences ( Fig. 1a). Previously, it was shown that these GEFs activate Rap1 in vivo as well as in vitro but not the closely related GTPases Ras, R-ras, or Ral. We have extended these experiments and found that all three GEFs can directly activate Rap2A as well (Fig. 1b). Equal amounts (approximately 100 nM) of GST fusions of the catalytic domains ( Fig. 1a) were incubated with fluorescent mantGDP-loaded Rap1A or Rap2A (100 nM) in the presence of excess unlabeled GDP, and exchange of guanine nucleotides was followed in real time as a decrease in fluorescence. To compare the activity of the different GEFs toward Rap1A and Rap2A, single exponential curves were fit from which the exchange reaction rates were calculated. These rates were compared with the intrinsic exchange reaction rates of the GTPases measured in the same experiment. From these calculations a fold induction of guanine nucleotide exchange on Rap1A and Rap2A was derived, which is depicted in Fig. 1c. Epac1 activated Rap2A five times more efficiently than Rap1A, whereas Epac2 and Repac activated Rap2A 3-fold less efficiently (Fig. 1b). Activation of Rap2 is not a common feature of all Rap1GEFs, since C3G (21) and CalDAG-GEFI (Fig. 1B) did not exhibit catalytic activity toward Rap2A in vitro.
To validate the results obtained in vitro, we investigated whether the Epac family members also activate Rap1 and Rap2 in vivo. Cells were transfected with Epac cDNAs together with either Rap1A or Rap2A and stimulated with forskolin and isobutylmethylxanthine to raise the level of cAMP. Activation of Rap1A was measured using the previously described activation-specific probe assay, in which Rap1GTP is specifically precipitated with the Rap1 binding domain of RalGDS (5). As shown in Fig. 1c, all three GEFs activate Rap1A. The activation by Epac1 and Epac2 is enhanced by forskolin treatment.
The activation-specific probe assay is less suitable for measuring activation of Rap2A, due to the high basal level of Rap2GTP in cells (14). Therefore, we incubated the cells with 32 P-labeled orthophosphate followed by precipitation of Rap2A and separation of bound GDP and GTP. We observed that all three GEFs activate Rap2A in vivo (Fig. 1d). The activation of Rap2A by Epac1 and Epac2 is enhanced by forskolin treatment. From these results we conclude that all three members of the Epac family activate both Rap1 and Rap2.
Epac2 Has a Second, Low Affinity Binding Site for cAMP-In the completed sequence of the Caenorhabditis elegans genome, only one Epac-related gene could be found. This gene encodes a FIG. 1. Epac family members activate both Rap1 and Rap2. a, bacterially expressed GST fusion proteins containing the catalytic domains of Epac1 (E1), Epac2 (E2), Repac (Re), and CalDAG-GEFI (CD) were purified using glutathione-agarose beads, separated by SDS-polyacrylamide gel electrophoresis, and stained by Coomassie. b, purified catalytic domains were incubated at approximately 100 nM with purified Rap1A or Rap2A proteins (100 nM) loaded with mantGDP. A decrease in fluorescence was measured at intervals of 15 or 20 s. Data points shown represent the mean of 20 subsequent measurements. c, Rap1A was cotransfected in Cos-7 cells with the indicated full-length GEF constructs. Cells were stimulated with forskolin (20 M) and isobutylmethylxanthine (1 mM) for 10 min, and Rap1 activation was measured using GST-RalGDS-Ras binding domain as an activation-specific probe. The lower panel depicts a Western blot probed for total Rap1A present in the lysates. d, HA-Rap2A was cotransfected in Cos-7 cells with the indicated full-length GEF constructs. Cells were labeled with [ 32 P]orthophosphate, Rap2A was immunoprecipitated using 12CA5 monoclonal antibodies, nucleotides were eluted and separated by TLC, and GTP/GDP ratios were measured using a phosphoimager.
FIG. 2. Different cAMP binding sites in Epac1 and Epac2. a, alignment of different cAMP binding pockets (residues identical in more than 50% of the depicted domain are marked by dark boxes; light boxes indicate conserved residues) and schematic presentation and purification of GST fusion constructs containing the cAMP binding domains of Epac1 (B) and Epac2 (A and B) and the regulatory domain of Epac2 (RD) (Coomassie-protein that has a putative second cAMP binding domain at the N terminus apart from the reported cAMP binding domain proximal to the catalytic region. This domain is not present in Epac1, but in Epac2, a similar cAMP-binding site is present (Fig. 1a). As judged from primary sequences, this site is similar to the genuine cAMP-binding sites of Epac1, Epac2, and PKA but distinct from the RCBD domain in PDZ-GEF, a RapGEF that does not respond to cAMP (Fig. 2a). We named the Nterminal cAMP binding domain present in Epac2 and C. elegans Epac the A-site, and we named the cAMP binding domain proximal to the catalytic domains, which is present in Epac1, Epac2, and C. elegans Epac the B-site (Fig. 2a). To compare these different sites, we analyzed purified domains (Fig. 2a) for in vitro binding to cAMP by ITC. We found that the A-site of Epac2 binds cAMP with an apparent affinity of 87 M, whereas the B-site has an affinity of 1.2 M, which is comparable with the affinity of 4 M, observed for the cAMP binding domain of Epac1 (Fig. 2b). Apparently, the A-site has a much lower affinity for cAMP as compared with the B-site.
In the regulatory subunits of PKA two cAMP binding domains are present that cooperatively bind to cAMP, meaning that the binding of cAMP to one site influences the affinity of the second site for cAMP (20). To investigate whether sites A and B in Epac2 may also act cooperatively, we measured cAMP binding affinity to the complete regulatory region of Epac2. Best-fit analysis revealed two binding sites with K d values of 0.5 and 76 M (Fig. 2c), which are in the same range as the affinities of the isolated domains. As a final control, a mixture of the separately purified A and B cAMP-binding sites of Epac2, in which no cooperativity can occur, was analyzed in the same assay. This yielded exactly the same result as the titration of the complete regulatory domain of Epac2 (Fig. 2c, lower panel). The data from these measurements are summarized in Fig. 2d. We conclude that no cooperativity occurs in binding of cAMP to the regulatory domain of Epac2 and that also in the full-length Epac2 protein, the B-site has a much higher affinity for cAMP that the A-site.
Isolated B-sites Inhibit the Exchange Activity of Epac Catalytic Domains-To investigate the role of the different N-terminal domains in the regulation of Epac1 and Epac2 activity by cAMP, we made several deletion constructs (Fig. 3a). As shown in Fig. 3b, mutant Epac1 (Epac1-⌬DEP) and Epac2 (Epac2-⌬DEP) proteins containing, next to the catalytic domain, only the B-site cAMP binding domain respond to cAMP in vitro like full-length Epac1. The fact that Epac2-⌬DEP responded less strongly than Epac1-⌬DEP is most likely due to difficult purification of this construct (Fig. 3a, left panel), which resulted in less GEF being present in the reaction. As a negative control in these experiments we used AMP. This closely related small molecule did not activate Epac1-⌬DEP or Epac2-⌬DEP (data not shown), confirming the specificity of the cAMP binding domains for the cyclic nucleotide. Whereas Epac mutants that contain the B-site are regulated by cAMP, proteins that lack the B-sites as well as the other N-terminal domains are constitutively active (see Fig. 1b). This implies that the B-sites serve as auto-inhibitory domains.
Next we investigated whether a direct covalent linkage between the catalytic domain and the B-site is essential for this regulation or whether they can function as separate domains.
We therefore isolated the regulatory domains of both Epac1 and Epac2 and incubated them with the corresponding catalytic domains. As shown in Fig. 3c, both regulatory domains completely inhibit the catalytic activity of the corresponding GEF domains, showing that they can form a stable complex that prevents GEF activity. The addition of cAMP abolishes the inhibitory effect. To dissect the role of the two cAMP-binding sites in the regulatory domain of Epac2, purified domains of the A-and the B-site of Epac2 were incubated with the catalytic domain of Epac2. Only the B-site and not the A-site (even at high concentration) inhibits the catalytic domain of Epac2 (Fig.  3d). The use of cAMP binding domain constructs containing also the DEP domain did not alter the ability of the A-site or B-site to inhibit the catalytic activity (data not shown).
The Mechanism of Epac Regulation Is Conserved in a Subset of RapGEFs-To investigate whether the cAMP binding domain of Epac1 can regulate only the catalytic domain of Epac1, we incubated the regulatory domain of Epac1 with the catalytic domains of the other RapGEFs (Fig. 4a). As shown in Fig. 4b, Epac1-RD inhibited the catalytic activity of both Epac2 and Repac. Interestingly, also the catalytic activity of PDZ-GEF was inhibited. In contrast, the GEF activity of the catalytic domains of C3G and CalDAG-GEFI was not inhibited. From these results we conclude that the isolated regulatory domain of Epac1 can act as an inhibiting structure for a specific subset of RapGEFs. This indicates that this mechanism of regulation is conserved between Epac and PDZ-GEF. In PDZ-GEF, a structure related to cAMP binding domains (RCBD) is present that probably plays a similar role as the B-sites of Epacs in the regulation of GEF activity. Furthermore, this property is specific for certain cAMP binding domains only, because neither the A-site of Epac2 (Fig. 3c) nor a PKA construct containing both its cAMP binding domains (Fig. 4b) affected the activity of the catalytic domain of Epac2. Thus we conclude that a specific sequence or structure in the B-sites of Epacs enables these domains to form an inhibitory interaction with the catalytic domains of a subset of RapGEFs.
The DEP Domain Localizes Epac to the Membrane Fraction-In addition to the cAMP binding domains, both Epac1 and Epac2 have a DEP domain. Such a domain was previously recognized in Disheveled, Egl-10, and pleckstrin (hence the name DEP domain) (23), and in the case of Disheveled, it was found to be involved in Frizzled-induced membrane localization (24). In Fig. 3b we showed that the DEP domain is not required for regulation of Epac1 by cAMP. We therefore tested the possibility that the DEP domain is involved in membrane localization of Epac1. Cos-7 cells expressing Epac1 or Epac1-⌬DEP were separated in particulate and cytosolic fractions, and the distribution of Epac1 in these fractions was determined. As shown in Fig. 5a, full-length Epac1 is only observed in the particulate fraction, whereas Epac1-⌬DEP is to a large extent present in the cytosolic fraction. As a control for this fractionation experiment, the distribution of the epidermal growth factor receptor, a transmembrane protein, and p42 mitogen-activated protein kinase, an exclusively cytosolic protein, in the same Epac-transfected fractions, was analyzed (Fig.  5a, lower panel). Both control proteins were detected almost exclusively in the expected fractions, proving a clear separation of membrane and cytosolic fraction. This result indicates that the DEP domain is involved in the localization of Epac1 to membrane structures. Importantly, stimulation of these cells with forskolin to induce cAMP did not have any effect on the distribution of Epac1 over the two fractions, indicating that the intracellular localization of Epac1 by the DEP domain is not dependent on cAMP. Next we investigated whether the DEP domain is required for in vivo regulation of Epac1 by cAMP. As shown in Fig. 5b, in NIH3T3-A14 cells, Epac1-⌬DEP strongly activates Rap1A in response to forskolin. From these results we conclude that the DEP domain is not directly involved in the cAMP-induced regulation of Epac1. DISCUSSION We have extended the characterization of the Epac family of cAMP-dependent Rap1GEFs. We found that all three members, Epac1, Epac2, and Repac, can activate the closely related GTPase Rap2A. Furthermore, we clarified the mechanism by which Epac1 and Epac2 are regulated by cAMP. The DEP domain targets Epac1 to membrane structures independent of cAMP signaling. The high affinity cAMP binding domains in Epac1 and Epac2 (B-sites) inhibit the catalytic activity of the GEF domains in the absence of cAMP. If cAMP levels rise, the auto-inhibition is relieved. The RapGEFs identified thus far can be placed in four different families. Whereas C3G and CalDAG-GEFI show little or no exchange activity toward Rap2 (21), the Epac family members and PDZ-GEF1 can efficiently activate Rap2 as well. This means that, next to Rap1, Rap2 proteins are also targets for PKA-independent cAMP-signaling routes. Presumably, the difference in GTPase specificity can be explained by the presence of specific sequence properties in the catalytic domains of Epacs and PDZ-GEF, which are absent from CalDAG-GEF and C3G. From sequence alignments, however, we were not able to identify such sequences, indicating that subtle differences may be sufficient.
Rap1 and Rap2 share 70% homology and differ in one residue in the effector domain. Like Rap1, Rap2 proteins have a threonine at the 61 position, at which a glutamine is present in Ras. As a consequence, the Rap proteins have a relatively low intrinsic GTPase activity (25). Our finding that Rap2 shares some but not all GEFs with Rap1 indicates that Rap2 is regulated to a certain extent by the same signals as Rap1. However, the basal level of Rap2GTP is much higher than that of Rap1GTP in a number of cell lines that have currently been tested (14). 2 This may be explained by the fact that Rap1GAP, a ubiquitously expressed GAP for Rap1, has a 40-fold lower activity toward Rap2 (26). Indeed, overexpression of Rap1GAP leads to complete inactivation of Rap1, even in the presence of overexpressed, active Rap1GEFs, whereas the basal and GEFinduced GTP-levels of Rap2 are only marginally reduced. 2 Epac1 and Epac2 are both regulated by cAMP. However, some clear differences exist between the regulatory domains of these GEFs. Epac1 has a single cAMP binding domain, and Epac2, as well as C. elegans Epac, has a second domain that is homologous to described cAMP binding domains. The affinity of this second domain, which we called the A-site, is much lower (87 M) than that of the cAMP binding domains proximal to the GEF domains of Epac1 and Epac2 (the B-sites, 4 and 1.2 M, respectively). The affinity of the PKA holoenzyme for cAMP was determined at approximately 0.8 M in cells (27). This is in the same range as the affinity of the B-sites of Epac1 and Epac2 for cAMP. Furthermore, in hepatocytes, which have relatively high levels of cAMP, it was calculated that the intracellular concentration of cAMP ranges from 0.3 to maximally 36 M (27), 3 which makes it likely that the B-sites and not the A-site of Epac2 are regulated by cAMP in cells. Perhaps the A-site interacts with a different cAMP-like small molecule or an unrelated compound, which together with cAMP, is responsible for full activation of Epac2. At present the function remains elusive.
The mechanism by which cAMP regulates the activity of Epacs was studied in vitro using deletion mutants. We observed that the B-sites alone are sufficient to provide cAMPdependent activation. Importantly, isolated B-sites can inhibit the activity of the isolated GEF domains, indicating that the regulation is mediated by an interaction between these domains, which is strong enough to survive separation of the two domains in different protein constructs. This interaction leads to the inactivation of the GEF domain in the absence of cAMP. The binding of cAMP either abolishes the interaction or changes it in such a way that it no longer prevents GEF activity.
The ability of the B-sites to inhibit the catalytic activity as a separate domain prompted us to investigate whether an overexpressed B-site, mutated in its cAMP binding pocket, could function as an interfering mutant in cAMP-induced, Epacmediated Rap1 activation. Although we were able to find mutants that inhibit the catalytic domains even in the presence of cAMP, these mutant domains could not inhibit cAMP-induced activation of Rap1, which was mediated by Epac1-⌬DEP (containing the B-site) in vitro or full-length Epac1 in vivo. This indicates that a separate B-site cannot compete with the intrinsic B-site. Possibly, steric hindrance by the cAMP-bound intrinsic B-site prevents the interaction of the ectopic, mutated B-site with the catalytic domain.
Interestingly, Epac1-RD was able also to inhibit the catalytic domains of Epac2, Repac, and PDZ-GEF. This indicates that the mechanism by which the B-site interacts with the catalytic domain is rather conserved. It is obvious that the presence of both domains in one protein facilitates this mode of regulation, but it could be hypothesized that originally, in early evolution, the two domains were expressed as separate proteins. This question is particularly interesting with respect to Repac, which lacks any intrinsic regulatory domain. It is well possible that a separate regulatory domain, which has not yet been identified, regulates this GEF. Alternatively, Repac is a constitutively active GEF, which is responsible for basal levels of Rap1GTP and Rap2GTP.
As shown by the in vitro experiments with mutants lacking the DEP domain, this domain is not involved in the regulation of GEF activity. Instead it is involved in the membrane localization of Epac1, presumably by binding to a membrane-associated protein, a function that is also assigned to the DEP domain of Disheveled (24). So far, however, no proteins interacting with a DEP domain have been identified. Alternatively, DEP domains may interact with specific lipid molecules. Since Epac is completely membrane-associated both in the absence 2 J. de Rooij, unpublished observations. 3 S. O. Doskeland, personal communication.
FIG. 5. CAMP-independent membrane targeting of Epac1 by the DEP domain. a, Cos-7 cells were transfected with HA-Epac1 or Epac1 lacking the DEP domain (HA-Epac1-⌬DEP). Cells were lysed in a syringe, and particulate fractions (P) and cytosolic fractions (C) were separated by centrifugation. Distribution of Epac in these fractions was analyzed by Western blotting using 12CA5 monoclonal antibodies directed against the HA tag. Lower panels show distribution of the epidermal growth factor receptor (EGF Rec.) as a transmembrane protein and mitogen-activated protein kinase (Mapk) as cytosolic protein for as a control for the fractionation. wt, wild type. b, Rap1A was cotransfected with increasing amounts of HA-Epac1-⌬DEP (30 ng, 100 ng, 300 ng, and 1 g), and Rap1 activation was measured using activation-specific probes and 12CA5 monoclonal antibodies on a Western blot. and presence of cAMP, we conclude that cAMP does not regulate translocation. This is in contrast to many other GEFs, for instance for Ral, Ras, and members of the Arf family, where membrane translocation may be the most important mechanism of activation.