Ium. d) The hydrophobic “spine”, a network of residues involving the hydrophobic contacts between the hinge 
helix and adjacent helices (a4 and a5). doi:10.1371/journal.pone.0048707.gdestabilize the helix or induce unwinding shift the apo/inactive vs. apo/active pre-equilibrium toward the latter state, i.e. an active state without cAMP (Fig. 1C). To test this hypothesis, we designed three successive deletion LED 209 web mutations of the 149?18 EPAC1 construct [10,21], which spans the CBD and which from here on forth will be referred to as the Wt-EPAC. Specifically, these mutants are C-terminally 25033180 truncated at positions 305, 310, and 312 (called de305, de310, and de312, respectively from here on forth) and act as perturbations that destabilize the hinge helix of apoEPAC, mimicking the cAMP-induced unwinding (Fig. 1B). In order to explore how the mutations affect the inactive vs. active conformational equilibrium of apo-EPAC, we employed the previously proposed chemical shift projection analysis (CHESPA), which provides residue-specific fractional shift towards activation for each mutant (Fig. 2A) [27]. In addition, the allosteric role of the hinge helix was further probed by the chemical shift covariance analysis (CHESCA), for which mutations were utilized as source of perturbations, unlike in previous CHESCA applications where cAMP and analogs were used to perturb the allosteric system [26]. Our results confirm the hypothesis that the C-terminal residues of the hinge helix (i.e. residues 305?10) are a pivotal 194423-15-9 chemical information determinant of EPAC auto-inhibition, showing that the hinge helix is extensively coupled to the other conserved allosteric elements ofthe CBD, even in the absence of cAMP. These results also lead to the counter-intuitive prediction that deletion of this C-terminal region causes an enhancement in cAMP-affinity, due to an increase in the apo/active relative population. This unexpected prediction was corroborated by the measurement of cAMPbinding isotherms through saturation transfer difference (STD) NMR experiments and the relevance of these results for the substrate-dependent sensitization to cAMP is also discussed [34,35].Materials and Methods Sample preparationThe deletion mutations de312, de310, and de305 were generated by inserting a stop codon at position 313, 311, and 306, respectively, by PCR in the 23727046 Wt construct (EPAC1149?18) and confirmed by DNA sequencing. Wt and all mutant constructs including E308A were purified and labelled according to published methods [26].NMR MeasurementsSpectra were acquired with a Bruker Avance 700-MHz spectrometer equipped with a 5 mm TCI cryoprobe at 306 K. Gradient and sensitivity enhanced [1H-15N] heteronuclear singleAuto-Inhibitory Hinge Helixbetween the Wt(apo) and the Wt(holo) was calculated as the magnitude of the activation vector B in Figure 2A. The chemical shift (ppm) of the 15N was downscaled by a factor of 0.2, as indicated in Figure 2A. The cos H and fractional activation X 
were calculated as:I Icos HA .BDA DDB DI I??I IXA .B DB DI??Chemical Shift Covariance Analysis (CHESCA)The inter-residue correlation matrix was generated according to published protocols [26]. However, in contrast to previous applications of CHESCA, the perturbation set was composed of select mutations that destabilize the C-terminal end of the hinge helix. Such mutations were analyzed in the apo state, where the extended hinge helix is stable. Thus, the perturbation set used here for CHESCA consisted of: Wt(apo), de305(apo),.Ium. d) The hydrophobic “spine”, a network of residues involving the hydrophobic contacts between the hinge helix and adjacent helices (a4 and a5). doi:10.1371/journal.pone.0048707.gdestabilize the helix or induce unwinding shift the apo/inactive vs. apo/active pre-equilibrium toward the latter state, i.e. an active state without cAMP (Fig. 1C). To test this hypothesis, we designed three successive deletion mutations of the 149?18 EPAC1 construct [10,21], which spans the CBD and which from here on forth will be referred to as the Wt-EPAC. Specifically, these mutants are C-terminally 25033180 truncated at positions 305, 310, and 312 (called de305, de310, and de312, respectively from here on forth) and act as perturbations that destabilize the hinge helix of apoEPAC, mimicking the cAMP-induced unwinding (Fig. 1B). In order to explore how the mutations affect the inactive vs. active conformational equilibrium of apo-EPAC, we employed the previously proposed chemical shift projection analysis (CHESPA), which provides residue-specific fractional shift towards activation for each mutant (Fig. 2A) [27]. In addition, the allosteric role of the hinge helix was further probed by the chemical shift covariance analysis (CHESCA), for which mutations were utilized as source of perturbations, unlike in previous CHESCA applications where cAMP and analogs were used to perturb the allosteric system [26]. Our results confirm the hypothesis that the C-terminal residues of the hinge helix (i.e. residues 305?10) are a pivotal determinant of EPAC auto-inhibition, showing that the hinge helix is extensively coupled to the other conserved allosteric elements ofthe CBD, even in the absence of cAMP. These results also lead to the counter-intuitive prediction that deletion of this C-terminal region causes an enhancement in cAMP-affinity, due to an increase in the apo/active relative population. This unexpected prediction was corroborated by the measurement of cAMPbinding isotherms through saturation transfer difference (STD) NMR experiments and the relevance of these results for the substrate-dependent sensitization to cAMP is also discussed [34,35].Materials and Methods Sample preparationThe deletion mutations de312, de310, and de305 were generated by inserting a stop codon at position 313, 311, and 306, respectively, by PCR in the 23727046 Wt construct (EPAC1149?18) and confirmed by DNA sequencing. Wt and all mutant constructs including E308A were purified and labelled according to published methods [26].NMR MeasurementsSpectra were acquired with a Bruker Avance 700-MHz spectrometer equipped with a 5 mm TCI cryoprobe at 306 K. Gradient and sensitivity enhanced [1H-15N] heteronuclear singleAuto-Inhibitory Hinge Helixbetween the Wt(apo) and the Wt(holo) was calculated as the magnitude of the activation vector B in Figure 2A. The chemical shift (ppm) of the 15N was downscaled by a factor of 0.2, as indicated in Figure 2A. The cos H and fractional activation X were calculated as:I Icos HA .BDA DDB DI I??I IXA .B DB DI??Chemical Shift Covariance Analysis (CHESCA)The inter-residue correlation matrix was generated according to published protocols [26]. However, in contrast to previous applications of CHESCA, the perturbation set was composed of select mutations that destabilize the C-terminal end of the hinge helix. Such mutations were analyzed in the apo state, where the extended hinge helix is stable. Thus, the perturbation set used here for CHESCA consisted of: Wt(apo), de305(apo),.