A simple electrostatic switch important in the activation of type I protein kinase A by cyclic AMP

Abstract

Cyclic AMP activates protein kinase A by binding to an inhibitory regulatory (R) subunit and releasing inhibition of the catalytic (C) subunit. Even though crystal structures of regulatory and catalytic subunits have been solved, the precise molecular mechanism by which cyclic AMP activates the kinase remains unknown. The dynamic properties of the cAMP binding domain in the absence of cAMP or C-subunit are also unknown. Here we report molecular-dynamics simulations and mutational studies of the RIalpha R-subunit that identify the C-helix as a highly dynamic switch which relays cAMP binding to the helical C-subunit binding regions. Furthermore, we identify an important salt bridge which links cAMP binding directly to the C-helix that is necessary for normal activation. Additional mutations show that a hydrophobic "hinge" region is not as critical for the cross-talk in PKA as it is in the homologous EPAC protein, illustrating how cAMP can control diverse functions using the evolutionarily conserved cAMP-binding domains.

Figure 1.

(Top) General domain organization of the regulatory subunits of protein kinase A. The A-domain is directly responsible for activating type Iα protein kinase A in a cAMP dependent manner. The cAMP-binding B-domain, which is not in this simulation but is present in the crystal structure, regulates access of cAMP to domain A but does not directly participate in the conformational changes that activate the kinase. (Middle) Region used in the molecular dynamics simulations presented here. Residues 91–112, which include the psuedosubstrate inhibitor are present in the crystallized construct but are not resolved in this structure. (Bottom) The construct in which we experimentally tested the mutations described. This region consists of the A-domain and pseudosubstrate inhibitor. Shown as ribbon diagrams are the crystal structures of the 91–244 domain when cAMP-bound (A) and C subunit-bound (B; bound C subunit not shown). Phosphate-binding cassette (PBC) (yellow); C-helix (white).

Figure 2.

Crystal structure of the RIα PBC, cAMP, and its interaction with the C-helix via Glu 200 and R241. Also shown are the three hydrophobic residues in the hinge region that we mutated to alanines. The cAMP is shown in yellow, while the B- and C-helices are white and the PBC is red. Dotted lines indicate hydrogen bonding.

Figure 3.

Plots of selected distances and dihedral angles during the simulations. Cyclic AMP-bound simulation is shown in gray dots; simulation without cAMP is shown as a solid black line; crystal structure distances are shown as straight dotted lines. (A) Salt bridge distance between OE1 of E143 and HH22 of R239. This salt bridge is not present in the crystal structure or apo simulation, but it forms in the simulation with cAMP. (B) Salt bridge distance between OE2 of E200 and HH21 of R241. This contact is present in the crystal structure but it breaks early with cAMP present and never reforms. Without cAMP, this contact breaks but reforms immediately when E200 flips toward the helical subdomain. (C) The side-chain torsion angle of E200 (CA,CB,CG,CD). With cAMP, it remains stable due to hydrogen bonding to the 2′-OH of the cAMP. Without cAMP, E200 flips toward the helical subdomain in order to form a salt bridge with R241. (D) The distance of the Cα atoms of E143 and R239. This represents the relative distances of the helices containing E143 and R239. The C helix containing R239 and R241 acts as a switch that moves toward the β-barrel without cAMP and moves toward the other helices when cAMP is present.

Figure 4.

End structures of the cAMP-bound and cAMP-free simulations. Shown in ribbons are the A-, B-, and C-helices. The white ribbons represent the end of the cAMP-free simulations, where the E200 flips away from the PBC and interacts more strongly with R241. In green is the end of the cAMP-bound simulations, where the R241/E200 contact is broken and the R239 interacts with E143 of the A-helix. While we do not expect the details of these simulations to be fully accurate, they demonstrate that the R241/E200 salt bridge is able to report on the occupancy of cAMP and thus allow a direct allosteric connection with the C-binding helices, via the C-helix. The orange arrows represent the movement of the C-helix. Figure courtesy of Nina Haste.

Figure 5.

Activation of wild-type and mutant RIα(91–244)/C holoenzymes. Holoenzymes were prepared as described in Materials and Methods. Holoenzymes (30 nM) were incubated with various concentrations of cAMP, and the activity was measured as described previously (Cook et al. 1981). Symbols used are wild type, □; I204A, ▪; R241A, Δ. The activation profiles for the L203A and Y229A mutants are very similar to those of wild type and are omitted for ease of viewing, but their activation constants are listed in Table 1.

Figure 6.

Thermal denaturation profiles of cAMP-free wild-type and mutant RIα (91–244). Samples (0.1 mg/mL) were scanned at 20°C/h from 20°–80°C using a 20-sec integration time. Ellipticity was monitored at 222 nm. Symbols used are WT, □; I204A, ▪; R241A,▵; L203A, •; Y229A, ▾.