He whole allosteric network of the EPAC CBDIn order to further explore the allosteric network controlled by residue 305?10 of EPAC1 in the absence of cAMP, we implemented the chemical shift covariance analysis (CHESCA) method [26] using as basis set the Wt(apo), de312(apo), de310(apo) and the de305(apo) truncation mutants as well as E308A(apo), which also targets the 305?10 regions. Using these five apo EPAC1 samples, several linear inter-residue chemical shift correlations are observed (Fig. 5A, 5B), resulting in a residuecorrelation matrix (Fig. 5C) that reveals the presence of an extensive long-range network of interactions controlled by the 305?10 a6 region. Specifically, the agglomerative cluster analysis (Figure S2 in ZK-36374 site Supporting Information) of the correlation matrix (blue grid, Fig. 5C) indicates that perturbations on residues 305?310 propagate to all the known allosteric sites of the EPAC1 CBD, from the PBC and the b2-b3 loop to most of the N-terminal helical bundle (red highlights, Fig. 5C). Based on these observations, we conclude that the unwinding of residues 305?10 in a6 is coupled to the whole allosteric network controlled by cAMP (Fig. 5C).Destabilization of the hinge helix enhances the affinity for cAMPConsidering that the apo/active state binds cAMP more tightly than the apo/inactive state, the coupling between the C-terminal region of a6 revealed by the combined CHESPA and CHESCA methods, leads to the interesting Somatostatin-14 prediction that de305, the closest mimetic of the apo/active form in our current investigation of the hinge helix (Fig. 4B), should exhibit higher affinity for cAMP thanFigure 4. SVD analysis of the chemical shifts measured for the C-terminal truncation mutants de305, de310 and de312. a) This panel shows the PC1 vs. PC2 plot with 
three sets of loadings (diamonds) for each of 23977191 the C-terminal hinge helix deletion mutants: de312 (red), de310 (blue) and de305 (green). There are four loadings per mutant with each loading corresponding to a state referenced to Rp-cAMPS, as labelled in the figure. The smaller arrows correspond to the separation along PC1 between the Wt(apo) and the mutant(apo) state. The large arrows correspond to the separation along PC1 between the Wt(apo) and the cAMP-bound Wt(holo). b) The percentage ratio of the two separations measured in panel (a) (i.e. relative magnitude of the two arrows), provides a quantitative measure of the overall fractional shift toward activation caused by the mutation. doi:10.1371/journal.pone.0048707.gAuto-Inhibitory Hinge HelixFigure 5. Chemical shift covariance analysis (CHESCA) of the hinge helix mutants. a) and b) show representative inter-residue chemical shift correlation among the five apo states (318:Wt, 318:E308A, de312, de310, and de305) and `m’ defines the slope. c) The chemical shift correlation matrix. Residue pairs with absolute correlation coefficients 0.98 are marked with a dot. The blue grid represents the largest agglomerative cluster (Figure S2 in Supporting Information) [26], while regions highlighted in red correspond to key allosteric sites of the CBD other than the hinge helix. doi:10.1371/journal.pone.0048707.gthe Wt construct. This counter-intuitive prediction was experimentally confirmed by STD NMR measurements on both the de305 and the Wt construct (Fig. 6). As expected, Figure 6 clearly shows that the de305 mutant binds cAMP more tightly than Wt CBD with the full integral hinge helix. The ,8-fold decrease in KD observed in going from the.He whole allosteric network of the EPAC CBDIn order to further explore the allosteric network controlled by residue 305?10 of EPAC1 in the absence of cAMP, we implemented the chemical shift covariance analysis (CHESCA) method [26] using as basis set the Wt(apo), de312(apo), de310(apo) and the de305(apo) truncation mutants as well as E308A(apo), which also targets the 305?10 regions. Using these five apo EPAC1 samples, several linear inter-residue chemical shift correlations are observed (Fig. 5A, 5B), resulting in a residuecorrelation matrix (Fig. 5C) that reveals the presence of an extensive long-range network of 
interactions controlled by the 305?10 a6 region. Specifically, the agglomerative cluster analysis (Figure S2 in Supporting Information) of the correlation matrix (blue grid, Fig. 5C) indicates that perturbations on residues 305?310 propagate to all the known allosteric sites of the EPAC1 CBD, from the PBC and the b2-b3 loop to most of the N-terminal helical bundle (red highlights, Fig. 5C). Based on these observations, we conclude that the unwinding of residues 305?10 in a6 is coupled to the whole allosteric network controlled by cAMP (Fig. 5C).Destabilization of the hinge helix enhances the affinity for cAMPConsidering that the apo/active state binds cAMP more tightly than the apo/inactive state, the coupling between the C-terminal region of a6 revealed by the combined CHESPA and CHESCA methods, leads to the interesting prediction that de305, the closest mimetic of the apo/active form in our current investigation of the hinge helix (Fig. 4B), should exhibit higher affinity for cAMP thanFigure 4. SVD analysis of the chemical shifts measured for the C-terminal truncation mutants de305, de310 and de312. a) This panel shows the PC1 vs. PC2 plot with three sets of loadings (diamonds) for each of 23977191 the C-terminal hinge helix deletion mutants: de312 (red), de310 (blue) and de305 (green). There are four loadings per mutant with each loading corresponding to a state referenced to Rp-cAMPS, as labelled in the figure. The smaller arrows correspond to the separation along PC1 between the Wt(apo) and the mutant(apo) state. The large arrows correspond to the separation along PC1 between the Wt(apo) and the cAMP-bound Wt(holo). b) The percentage ratio of the two separations measured in panel (a) (i.e. relative magnitude of the two arrows), provides a quantitative measure of the overall fractional shift toward activation caused by the mutation. doi:10.1371/journal.pone.0048707.gAuto-Inhibitory Hinge HelixFigure 5. Chemical shift covariance analysis (CHESCA) of the hinge helix mutants. a) and b) show representative inter-residue chemical shift correlation among the five apo states (318:Wt, 318:E308A, de312, de310, and de305) and `m’ defines the slope. c) The chemical shift correlation matrix. Residue pairs with absolute correlation coefficients 0.98 are marked with a dot. The blue grid represents the largest agglomerative cluster (Figure S2 in Supporting Information) [26], while regions highlighted in red correspond to key allosteric sites of the CBD other than the hinge helix. doi:10.1371/journal.pone.0048707.gthe Wt construct. This counter-intuitive prediction was experimentally confirmed by STD NMR measurements on both the de305 and the Wt construct (Fig. 6). As expected, Figure 6 clearly shows that the de305 mutant binds cAMP more tightly than Wt CBD with the full integral hinge helix. The ,8-fold decrease in KD observed in going from the.