Epac (exchange protein directly activated by cAMP) is a critical cAMP receptor, which senses cAMP and couples the cAMP signal to the catalysis of guanine exchange in the Rap substrate. In the present paper, we review the NMR studies that we have undertaken on the CBD (cyclic-nucleotide-binding domain) of Epac1. Our NMR investigations have shown that cAMP controls distal autoinhibitory interactions through long-range modulations in dynamics. Such dynamically mediated allosteric effects contribute not only to the cAMP-dependent activation of Epac, but also to the selectivity of Epac for cAMP in contrast with cGMP. In addition, we have mapped the interaction networks that couple the cAMP-binding site to the sites involved in the autoinhibitory interactions, using a method based on the covariance analysis of NMR chemical shifts. We anticipate that this approach is generally applicable to dissect allosteric networks in signalling domains.

Introduction

cAMP is an ancient second messenger that controls a multitude of essential cellular processes by binding to downstream receptors. In eukaryotes, two ubiquitous cAMP receptors are PKA (protein kinase A) and the more recently discovered Epac (exchange protein directly activated by cAMP). The present mini-review will focus on the structural and dynamical basis for the cAMP-dependent activation of Epac. Epac is a multi-domain protein with five or six domains (Figure 1) partitioned into an N-terminal RR (regulatory region), containing at least one CBD (cyclic-nucleotide-binding domain), and a C-terminal CR (catalytic region), including the CDC25 homology domain that implements the guanine-exchange function of Epac [15]. The relative orientation of the RR and CR provides a basic, but effective, rationalization for the cAMP-dependence of the Epac catalytic efficiency [2,3].

Domain organization and activation model

Figure 1
Domain organization and activation model

(a) Schematic illustration of the domains of Epac1. The RR and CR are indicated and are separated by a vertical broken line between the CBD and REM (Ras-exchange motif) domains. CDC25HD, CDC25 homology domain; DEP, Dishevelled/Egl-10/pleckstrin; RA, Ras-association domain. (b) Thermodynamic cycle for the coupled activation and binding equilibria (Eq.). The inactive/closed states are indicated by (I), whereas the active/open states that allow for substrate Rap binding are indicated by (A).

Figure 1
Domain organization and activation model

(a) Schematic illustration of the domains of Epac1. The RR and CR are indicated and are separated by a vertical broken line between the CBD and REM (Ras-exchange motif) domains. CDC25HD, CDC25 homology domain; DEP, Dishevelled/Egl-10/pleckstrin; RA, Ras-association domain. (b) Thermodynamic cycle for the coupled activation and binding equilibria (Eq.). The inactive/closed states are indicated by (I), whereas the active/open states that allow for substrate Rap binding are indicated by (A).

In the absence of cAMP, the RR and CR adopt a predominantly closed topology, in which the RR sterically occludes access of the Rap1 substrate to the catalytic domain. This autoinhibited state of Epac is stabilized by a set of salt bridges, commonly referred to as the IL (ionic latch) and mediated by the N-terminal region of the CBD, which secures the RR in the vicinity of the CR [2]. A minor population of Epac molecules is also present in which the RR and CR adopt an open topology, whereby the Rap1 substrate can freely access the catalytic site. However, in the absence of cAMP, the effective concentration of Epac in the open topology is too low to result in any significant constitutive guanine exchange activity [6] (Figure 1b).

Upon cAMP binding, two events occur that result in a marked increase in the fractional population of the open Epac topology and in the consequent activation of Epac. First, cAMP causes a rotation in the C-terminal CBD helix. Secondly, cAMP weakens the IL region, stabilizing the open Epac topology and assisting the CBD hinge motion that rotates the RR away from the CR [15]. A key question addressed here is how does cAMP allosterically control distal autoinhibitory interactions, such as the IL?

The cAMP-dependent control of Epac autoinhibitory interactions is assisted by dynamic changes

As a first attempt to understand how cAMP weakens the IL salt bridges mediated by the N-terminal helical bundle of Epac, we compared the apo and cAMP-bound forms of the essential CBD of Epac using NMR spectroscopy [79]. However, only minor chemical shift changes were observed between these two forms of Epac for the IL-spanning region, suggesting that cAMP does not cause any major structural change for the residues involved in the IL interactions [79]. This preliminary observation led us to hypothesize that cAMP may exert its control over the IL through dynamics as opposed to structural modulations [1016]. In order to test this hypothesis, we measured the 15N relaxation rates of the Epac1 CBD in the active cAMP effector-bound state as well as in two inactive states, the apo and the Rp-cAMPS (Rp isomer of adenosine 3′,5′-monophosphothioate) antagonist-bound states. The comparative analysis of the 15N relaxation rates in the active compared with the inactive states revealed that activation is associated with increased dynamics in the picosecond–nanosecond and microsecond–millisecond time scales for a significant portion of the IL-spanning region [8]. Considering that the NMR relaxation data was acquired for an isolated Epac CBD, the enhanced dynamics caused by cAMP in the IL region suggests that cAMP binding imposes an entropic penalty to the IL salt bridges, which in turn results in a weakening of the IL and of Epac autoinhibition [8,9]. We have also shown that this cAMP-dependent entropic control of the IL is a key determinant not only of the activation of Epac, but also of the selectivity of Epac for cAMP over cGMP [9] (Figure 2). Overall, the cAMP-dependent modulations in dynamics provide a first level of rationalization for IL control by cAMP [8]; however, they also open new questions on how the cAMP signals propagates from the PBC (phosphate-binding cassette), where cAMP docks, to the distal IL. To address these questions and obtain a better appreciation of the allosteric networks controlled by cAMP, we have proposed an NMR-based approach (Figure 3) that relies on the CHESCA (chemical shift covariance analysis) [17].

Model to rationalize the cAMP compared with cGMP selectivity in Epac1

Figure 2
Model to rationalize the cAMP compared with cGMP selectivity in Epac1

(a) In the absence of CBD, cAMP samples predominantly the anti conformation; however, the CBD of Epac preferentially selects the syn conformers of cAMP. (b) In contrast with cAMP, cGMP samples to a large extent the syn conformation in the absence of the Epac1 CBD. However, in the presence of the Epac1 CBD, there is a preferential selection for the anti conformation, which avoids the steric clash between the amino group of cGMP and the C-terminus of the PBC. Orange highlights indicate variations relative to (a). The cGMP-bound form of Epac does not introduce an IL entropic penalty as effective as that observed for cAMP [9].

Figure 2
Model to rationalize the cAMP compared with cGMP selectivity in Epac1

(a) In the absence of CBD, cAMP samples predominantly the anti conformation; however, the CBD of Epac preferentially selects the syn conformers of cAMP. (b) In contrast with cAMP, cGMP samples to a large extent the syn conformation in the absence of the Epac1 CBD. However, in the presence of the Epac1 CBD, there is a preferential selection for the anti conformation, which avoids the steric clash between the amino group of cGMP and the C-terminus of the PBC. Orange highlights indicate variations relative to (a). The cGMP-bound form of Epac does not introduce an IL entropic penalty as effective as that observed for cAMP [9].

Map of the CHESCA-based networks of Epac

Figure 3
Map of the CHESCA-based networks of Epac

(ac) Surface representation of the allosteric network. (d) Surface representation of the binding network. (b and c) Subsets of the allosteric network shown in (a). Selected regions subject to cAMP-dependent structural and dynamic changes are shown with a green and red ribbon respectively. Reproduced with permission from [17]: Selvaratnam, R., Chowdhury, S., VanSchouwen, B. and Melacini, G. (2011) Mapping allostery through the covariance analysis of NMR chemical shifts. Proc. Natl. Acad. Sci. U.S.A. 108, 6133–6138.

Figure 3
Map of the CHESCA-based networks of Epac

(ac) Surface representation of the allosteric network. (d) Surface representation of the binding network. (b and c) Subsets of the allosteric network shown in (a). Selected regions subject to cAMP-dependent structural and dynamic changes are shown with a green and red ribbon respectively. Reproduced with permission from [17]: Selvaratnam, R., Chowdhury, S., VanSchouwen, B. and Melacini, G. (2011) Mapping allostery through the covariance analysis of NMR chemical shifts. Proc. Natl. Acad. Sci. U.S.A. 108, 6133–6138.

Mapping allosteric networks through the NMR CHESCA method

The CHESCA method is based on two main assumptions. First, it posits that modulation of dynamic processes in the fast or intermediate chemical-exchange regime results in subtle chemical shift changes [18,19], which are measurable with high accuracy and precision at high field [20]. Secondly, it assumes that linear chemical shift correlations between two given residues are indicative of their concerted response to a common set of perturbations. The first assumption implies that chemical shift variations may provide the key to map allosteric networks also in the case of dynamically driven (or assisted) allostery [17]. The second assumption implies that the chemical shift matrix, with residue-specific rows and perturbation-specific columns, is amenable to clustering analyses similar to those typically implemented in the context of gene network profiling [21]. For instance, hierarchical AC (agglomerative clustering) of the chemical shift correlation matrix indentifies two clusters of residues, which are represented as dendrograms and are associated with allosteric and cAMP-binding functions (Figure 3). The AC approach is then complemented by the SVD (singular value decomposition) of the covariance chemical shift matrix, which aims to assess the relative allosteric compared with binding contributions of each residue [17].

The CHESCA method was applied to the cAMP-binding domain of Epac1 using a perturbation set that included the apo form, the cAMP-bound state, the Rp-cAMPS antagonist-bound state as well as the Sp-cAMPS (Sp isomer of adenosine 3′,5′-monophosphothioate) and 2′-OMe-cAMP (2′-O-methyl-cAMP) agonist-bound states. Using these cAMP analogues, the CHESCA method resulted in an assessment of the residue-specific allosteric and binding contributions [17], which was consistent with independent measurements of Kd and kmax values for several Epac mutants [4,5]. In addition, the CHESCA results appeared to be in agreement also with the allosteric function of selected residues inferred by co-evolutionary analyses [22]. Overall, the available experimental data support the allosteric and binding clusters identified through CHESCA for the Epac1 CBD. Furthermore, the CHESCA clusters provided unanticipated insights into the allosteric and binding determinants of Epac.

Results from the application of CHESCA to Epac

The first striking feature of the allosteric CHESCA cluster in the Epac1 CBD (Figure 3a) is that it is, to a large extent, confined to the non-contiguous α-subdomain rather than to the contiguous β-subdomain. The latter appears therefore to play mainly a passive scaffolding role in the Epac allostery. Within the α-subdomain, two key allosteric subclusters stand out. The first subcluster (Figure 3c) is mostly composed of hydrophobic residues centred around Arg186, which, rather than being neutralized by a single salt bridge with a negatively charged group, forms multiple hydrogen bonds with several polar and non-charged oxygen atoms. The second subcluster (Figure 3b) includes two spines of hydrophobic residues at the interface between the so-called hinge–helix at the C-terminus of the CBD and a helix that belongs to the N-terminal helical bundle of the CBD [17]. Some of the residues in such hydrophobic spines were identified previously as being critical for the autoinhibition of Epac. For instance, Phe300 has been shown to impede the hinge–helix rotation in the absence of cAMP due to steric hindrance with the PBC [5]. However, CHESCA analysis reveals that several additional sites also play a role in the autoinhibition of Epac. In this respect, a residue of the CHESCA allosteric cluster, i.e. the β-branched Val307 is notable because, despite its low intrinsic propensity to adopt a helical conformation, it is part of the hinge–helix in the apo structure (Figure 3b). However, when cAMP binds Epac, the hinge–helix C-terminus, which includes Val307, becomes more disordered. These observations suggest that the autoinhibited state of Epac contains sites of molecular frustration (e.g. residues forced by the structural context into conformations that are different from those they would intrinsically prefer). Upon cAMP binding, such molecular frustration is eliminated or reduced. These observations suggest that the Epac1 CBD has evolved to exploit the cAMP-dependent release of molecular frustration in the apo state as an effective method to harness free energy from the protein structure and assist the conformational change associated with Epac activation.

CHESCA analysis reveals also a second cluster of approximately ten residues (Figure 3d), which appear to play a pivotal role in cAMP binding. Interestingly, the residues in this CHESCA cluster span not only the canonical PBC and BBR (base-binding region), which are known from structural studies to contact cAMP directly, but also several residues that are adjacent to the PBC and BBR, but do not interact directly with the cAMP ligand. For example, several residues in the loop between β-strands 2 and 3 as well as in β-strand 7 appear to contribute to binding, although they are not formally part of the PBC and BBR [17]. These results suggest that the cAMP-binding site should perhaps be redefined in more general terms to include also the regions that affect cyclic nucleotide binding, without interacting directly with the allosteric effector.

Conclusions

The CHESCA method relies on the assumption that linear inter-residue chemical shift correlations reflect concerted responses to a common set of perturbations. As CHESCA only decodes information already embedded in the perturbation set, the CHESCA results depend critically on the choice of perturbations. When such perturbations include both agonists and antagonists, the CHESCA approach leads to a systematic dissection of binding compared with allosteric contributions at residue resolution. Furthermore, CHESCA identified allosteric networks that bridge sites subject to structural and/or dynamical changes, as sensed by minor chemical shift differences. We are now planning to extend the CHESCA applications to other CBDs, in order to identify which features of the allosteric and binding networks of Epac are conserved and which are Epac-specific. We are also planning to extend CHESCA to cases where perturbations are introduced by mutations, as opposed to cAMP analogues.

Signalling 2011: a Biochemical Society Centenary Celebration: A Biochemical Society Focused Meeting held at the University of Edinburgh, U.K., 8–10 June 2011. Organized and Edited by Nicholas Brindle (Leicester, U.K.), Simon Cook (The Babraham Institute, U.K.), Jeff McIlhinney (Oxford, U.K.), Simon Morley (University of Sussex, U.K.), Sandip Patel (University College London, U.K.), Susan Pyne (University of Strathclyde, U.K.), Colin Taylor (Cambridge, U.K.), Alan Wallace (AstraZeneca, U.K.) and Stephen Yarwood (Glasgow, U.K.).

Abbreviations

     
  • AC

    agglomerative clustering

  •  
  • BBR

    base-binding region

  •  
  • CBD

    cyclic-nucleotide-binding domain

  •  
  • CHESCA

    chemical shift covariance analysis

  •  
  • CR

    catalytic region

  •  
  • Epac

    exchange protein directly activated by cAMP

  •  
  • IL

    ionic latch

  •  
  • PBC

    phosphate-binding cassette

  •  
  • Rp-cAMPS

    Rp isomer of adenosine 3′,5′-monophosphothioate

  •  
  • RR

    regulatory region

Funding

This work was supported by the Canadian Institute of Health Research (CIHR) and the National Sciences and Engineering Research Council (NSERC). We are also indebted to the Heart and Stroke Foundation of Canada (HSFC) for a Maureen Andrew New Investigator to G.M.

References

References
1
Kraemer
A.
Rehmann
H.R.
Cool
R.H.
Theiss
C.
de Rooij
J.
Bos
J.L.
Wittinghofer
A.
Dynamic interaction of cAMP with the rap guanine-nucleotide exchange factor Epac1
J. Mol. Biol.
2001
, vol. 
306
 (pg. 
1167
-
1177
)
2
Rehmann
H.
Das
J.
Knipscheer
P.
Wittinghofer
A.
Bos
J.L.
Structure of the cyclic-AMP-responsive exchange factor Epac2 in its auto-inhibited state
Nature
2006
, vol. 
439
 (pg. 
625
-
628
)
3
Rehmann
H.
Arias-Palomo
E.
Hadders
M.A.
Schwede
F.
Llorca
O.
Bos
J.L.
Structure of Epac2 in complex with a cyclic AMP analogue and RAP1B
Nature
2008
, vol. 
455
 (pg. 
124
-
127
)
4
Rehmann
H.
Schwede
F.
Doskeland
S.O.
Wittinghofer
A.
Bos
J.L.
Ligand-mediated activation of the cAMP-responsive guanine nucleotide exchange factor epac
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
38548
-
38556
)
5
Rehmann
H.
Prakash
B.
Wolf
E.
Rueppel
A.
de Rooij
J.
Bos
J.L.
Wittinghofer
A.
Structure and regulation of the cAMP-binding domains of Epac2
Nat. Struct. Biol.
2003
, vol. 
10
 (pg. 
26
-
32
)
6
Hilser
V.J.
Garcia-Moreno
E.B.
Oas
T.G.
Kapp
G.
Whitten
S.T.
A statistical thermodynamic model of the protein ensemble
Chem. Rev.
2006
, vol. 
106
 (pg. 
1545
-
1558
)
7
Mazhab-Jafari
M.T.
Das
R.
Fotheringham
S.A.
SilDas
S.
Chowdhury
S.
Melacini
G.
Understanding cAMP-dependent allostery by NMR spectroscopy: comparative analysis of the EPAC1 cAMP-binding domain in its apo and cAMP-bound states
J. Am. Chem. Soc.
2007
, vol. 
129
 (pg. 
14482
-
14492
)
8
Das
R.
Mazhab-Jafari
M.T.
Chowdhury
S.
SilDas
S.
Selvaratnam
R.
Melacini
G.
Entropy-driven cAMP-dependent allosteric control of inhibitory interactions in exchange proteins directly activated by cAMP
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
19691
-
19703
)
9
Das
R.
Chowdhury
S.
Mazhab-Jafari
M.T.
Sildas
S.
Selvaratnam
R.
Melacini
G.
Dynamically driven ligand selectivity in cyclic nucleotide binding domains
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
23682
-
23696
)
10
Smock
R.G.
Gierasch
L.M.
Sending signals dynamically
Science
2009
, vol. 
324
 (pg. 
198
-
203
)
11
Tsai
C.J.
del Sol
A.
Nussinov
R.
Allostery: absence of a change in shape does not imply that allostery is not at play
J. Mol. Biol.
2008
, vol. 
378
 (pg. 
1
-
11
)
12
Kuriyan
J.
Eisenberg
D.
The origin of protein interactions and allostery in colocalization
Nature
2007
, vol. 
450
 (pg. 
983
-
990
)
13
Goodey
N.M.
Benkovic
S.J.
Allosteric regulation and catalysis emerge via a common route
Nat. Chem. Biol.
2008
, vol. 
4
 (pg. 
474
-
482
)
14
Changeux
J.P.
Edelstein
S.J.
Allosteric mechanisms of signal transduction
Science
2005
, vol. 
308
 (pg. 
1424
-
1428
)
15
Kern
D.
Zuiderweg
E.R.
The role of dynamics in allosteric regulation
Curr. Opin. Struct. Biol.
2003
, vol. 
13
 (pg. 
748
-
757
)
16
Boehr
D.D.
Nussinov
R.
Wright
P.E.
The role of dynamic conformational ensembles in biomolecular recognition
Nat. Chem. Biol.
2009
, vol. 
5
 (pg. 
789
-
796
)
17
Selvaratnam
R.
Chowdhury
S.
VanSchouwen
B.
Melacini
G.
Mapping allostery through the covariance analysis of NMR chemical shifts
Proc. Natl. Acad. Sci. U.S.A.
2011
, vol. 
108
 (pg. 
6133
-
6138
)
18
Daily
M.D.
Gray
J.J.
Local motions in a benchmark of allosteric proteins
Proteins
2007
, vol. 
67
 (pg. 
385
-
399
)
19
Zhuravleva
A.
Korzhnev
D.M.
Nolde
S.B.
Kay
L.E.
Arseniev
A.S.
Billeter
M.
Orekhov
V.Y.
Propagation of dynamic changes in barnase upon binding of barstar: an NMR and computational study
J. Mol. Biol.
2007
, vol. 
367
 (pg. 
1079
-
1092
)
20
Skrynnikov
N.R.
Dahlquist
F.W.
Kay
L.E.
Reconstructing NMR spectra of “invisible” excited protein states using HSQC and HMQC experiments
J. Am. Chem. Soc.
2002
, vol. 
124
 (pg. 
12352
-
12360
)
21
Eisen
M.B.
Spellman
P.T.
Brown
P.O.
Botstein
D.
Cluster analysis and display of genome-wide expression patterns
Proc. Natl. Acad. Sci. U.S.A.
1998
, vol. 
95
 (pg. 
14863
-
14868
)
22
Kannan
N.
Wu
J.
Anand
G.S.
Yooseph
S.
Neuwald
A.F.
Venter
J.C.
Taylor
S.S.
Evolution of allostery in the cyclic nucleotide binding module
Genome Biol.
2007
, vol. 
8
 pg. 
R264