Research Article

Crystal Structure of a Complex Between the Catalytic and Regulatory (RIα) Subunits of PKA

See allHide authors and affiliations

Science  04 Feb 2005:
Vol. 307, Issue 5710, pp. 690-696
DOI: 10.1126/science.1104607

Abstract

The 2.0-angstrom structure of the cyclic adenosine monophosphate (cAMP)–dependent protein kinase (PKA) catalytic subunit bound to a deletion mutant of a regulatory subunit (RIα) defines a previously unidentified extended interface. The complex provides a molecular mechanism for inhibition of PKA and suggests how cAMP binding leads to activation. The interface defines the large lobe of the catalytic subunit as a stable scaffold where Tyr247 in the G helix and Trp196 in the phosphorylated activation loop serve as anchor points for binding RIα. These residues compete with cAMP for the phosphate binding cassette in RIα. In contrast to the catalytic subunit, RIα undergoes major conformational changes when the complex is compared with cAMP-bound RIα. The inhibitor sequence docks to the active site, whereas the linker, also disordered in free RIα, folds across the extended interface. The β barrel of cAMP binding domain A, which is the docking site for cAMP, remains largely intact in the complex, whereas the helical subdomain undergoes major reorganization.

The discovery of cAMP as a second messenger (1) led eventually to the identification of cAMP-dependent PKA (2) with regulatory subunits that are the major receptors for cAMP (3). PKA, ubiquitous in mammalian cells, regulates processes as diverse as growth, development, memory, metabolism, gene expression, and lipolysis. The PKA holoenzyme exists as an inactive complex of two cataytic (C) subunits and a regulatory (R) subunit dimer; binding of cAMP facilitates dissociation and activation of the C subunits.

The C subunit is comprised of a small and large lobe with the active site forming a cleft between the two lobes. This fold, first described for PKA, defines the protein kinase superfamily (4). The small lobe provides the binding site for adenosine 5′-triphosphate (ATP), and the large lobe provides catalytic residues and a docking surface for peptide/protein substrates. Opening and closing of the active site cleft is an essential part of catalysis (5). The activation loop in the large lobe contains a phosphorylation site, Thr197, essential for catalysis (6).

Like many signaling proteins, the R subunits are modular. They contain well-ordered domains as well as disordered regions. The dimerization domain at the N terminus is a well-organized four-helix bundle that also provides a docking surface for A kinase anchoring proteins (79). At the C terminus of each protomer are two tandem cAMP-binding domains (CBD-A and CBD-B), each representing a structural motif that has been conserved as a binding module for cAMP from bacteria to man (10). Each CBD is composed of a helical subdomain and an eight-stranded β barrel where cAMP binds. The essential feature of the β barrel is a conserved phosphate binding cassette (PBC) that anchors the cAMP and shields it from solvent (10). The CBDs are joined to the dimerization domain by a flexible linker (11), which includes a substratelike inhibitor sequence that docks to the active site cleft of the C subunit in the absence of cAMP (12).

Whereas structures of dimerization domains and the CBDs have been solved by nuclear magnetic resonance (7, 9) and crystallography (13, 14), respectively, the structure of an R:C complex has been elusive. Here, we report the crystal structure of a holoenzyme complex between an oxidation-resistant form of the C subunit and a deletion mutant of RIα, RIα(91–244), which contains the inhibitor sequence and the N-terminal cAMP binding domain (CBD-A). The complex was crystallized in the presence of a nonhydrolyzable analog of ATP, adenylyl-imidodiphosphate (AMP-PNP). The structure confirms the intrasteric mechanism of regulation, defines the intricate ordering of the linker region at the interface of the RIα and C subunits, reveals a global reorganization of the helical subdomain of RIα, and shows the unliganded molecular target that is seen by cAMP. The comparison of this structure with structures of RIα in its cAMP-bound conformation provides insight into the structural basis for cAMP-induced activation of PKA.

Molecular architecture. The architecture of the RIα(91–244):C complex reveals an extended interface that covers nearly 3000 Å2 (Fig. 1, A and B, and movie S1). Although the C subunit assumes a fully closed conformation with Mn2AMP-PNP bound at the active site cleft, it does not undergo any other major conformational changes as a result of complex formation; instead, it serves as a stable scaffold for docking RIα. The docking surface on the C subunit is localized almost exclusively to the large lobe and is overlapping but distinct from the surface used for docking another PKA inhibitor, the heat stable protein kinase inhibitor (PKI) (Fig. 2D) (15). The binding surface extends from the inhibitor binding site at the active site cleft (site 1), across the G helix (site 2) and through to the activation loop (site 3). The electrostatic profile of this complex interface is shown in Fig. 1C.

Fig. 1.

Stereoview of the RIα(91–244):C:Mn2AMP-PNP complex. (A) The small lobe (residues 11 to 120) and large lobe (residues 121 to 350) of the C subunit are colored in gray and tan, respectively. The N and C termini (residues 11 to 350) are labeled NC and CC. The activation loop, the G helix (G), and the glycine-rich loop are colored in green. Two phosphorylated residues, Thr197C and Ser338C, are shown as red spheres. The N and C termini (residues 91 to 242) of RIα are labeled NR and CR. The inhibitory/linker region (IS) (residues 91 to 112) is colored in red, the PBC (P) in yellow, the B/C helix in gray, and the rest in cyan. AMP-PNP is shown in red. This figure was generated with Molscript (46) and Raster3D (47). (B) Looking down the C:RIα complex, rotated 90° around a horizontal axis from the view in (A). (C) Electrostatic surface potential of the complex (left) and with its interface opened up to view the surfaces of individual subunits (right). The linker segment complements site 3, the PBC complements site 2, and the inhibitor site complements site 1 of the C subunit. This figure was generated with GRASS (48). (D) Structural comparison of RIα in the cAMP-bound and C-bound conformations. The inhibitory/linker region (residues 91 to 112, in red) is disordered in the cAMP-bound conformation and ordered in the C-bound conformation. The PBC (residues 199 to 210, in yellow) is stretched away from the β barrel when the C subunit binds. The B and C helices become one extended helix (residues 226 to 242, in gray) in the complex. Tyr205R (yellow) at the tip of the PBC changes its orientation to interact with Tyr247C at the G helix of the C subunit. This figure was generated with Molscript (46) and Raster3D (47).

Fig. 2.

Surfaces of the C subunit that interact with RIα. (A) The complex interface on the large lobe with the F and G helices highlighted in gray. The activation segment (residues 184 to 208) is yellow. Asp208C and Arg280C are each indicated by a ball. Phe239C (49) is part of the hydrophobic docking site for PKI. Site 1 corresponds to the inhibitor docking site, site 2 to the G helix with Tyr247C, and site 3 to the activation loop containing Arg194C and Trp196C. This figure was generated with Insight II (50). (B) The inhibitor/linker segment of RIα(91–112). Structure alignment of inhibitor consensus sequences of PKI (inhibitor peptide IP20: Protein Data Bank accession code 1ATP) and RIα. Carbons of RIα are colored in gray. The consensus sequence and the amphipathetic helix of IP are shown as red bond and ribbon, respectively. Similar to IP20 (Arg-Arg-Asn-Ala-Ile), the inhibitory sequence (Arg94-Arg-Gly-Ala-Ile98) of RIα blocks the active site of the C subunit. The coordinates of these residues are nearly identical with an exception of Gly/Asn substitution at the P-1 site. The linker segment interacts not only with the C subunit but with the B/C helix. Specific interacting residues from C and RIα are indicated in black and red, respectively (arrows). As in the IP20 complex, the C subunit within the complex is in the closed conformation, with a root mean square deviation of 0.42 Å2 for equivalent Cα atoms. This figure was generated with Insight II (50). (C) The inhibitor region (residues 91 to 100) bound to the active site cleft (site 1). The R- and C-subunit residues are drawn with gray and tan carbons, respectively. Arg94R at the P-3 site interacts with Thr51C in the glycine-rich loop, Glu127C in the linker, Tyr330C in the C-terminal tail, and AMP-PNP. Arg95R at the P-2 site is tightly held by Glu170C in the catalytic loop and Glu230C in the F helix (Fig. 3). Farther down toward the linker, Ile98R and Ala100R make antiparallel β-sheet–like interactions with the P+1 loop. This figure was generated with Visual Molecular Dynamics (VMD) (51) and Molscript (46). (D) Surfaces of the C subunit used for recognition of PKI and RIα. This surface rendering of the C subunit shows the small lobe in white and the large lobe (residues 128 to 350) in tan. The inhibitors of C, PKI, and the R subunits use a common surface (magenta) on which a pseudosubstrate (PKI and RI) or a substrate (RII) docks to the active site cleft. To achieve high-affinity binding requires an additional peripheral recognition site (PRS). PKI uses the surface shown in blue, designated as PBS 1, whereas RIα uses the surface shown in red (PRS2). This figure was generated with VMD (51).

In contrast to the C subunit, RIα undergoes major conformational changes upon complex formation (Fig. 1D and movie S2). There are three general features that describe the binding of RIα to form the holoenzyme complex. First, the inhibitor sequence docks to the active site cleft. Second, the linker segment that connects the inhibitor peptide to CBD-A is ordered. Third, CBD-A docks onto the large lobe of the C subunit (Fig. 1C). The inhibitor peptide and linker region (Arg94R to Val112R) are disordered in the crystal structure of cAMP-bound RIα(91–379), whereas in the complex this segment binds as an extended chain along the surface of the active site cleft (Fig. 1, A to D). In addition to this transition from disorder to order, docking of RIα results in global reorganization of the helical subdomain of CBD-A (Fig. 1D). As predicted by mutagenesis and hydrogen/deuterium exchange (1618), this docking site involves the B helix; however, the details and magnitude of this conformational reorganization were not anticipated. The general conformation of the β barrel, which provides the binding site for cAMP in free RIα, remains largely unchanged when the C subunit binds (Fig. 1D).

Extended binding surface on the catalytic subunit.Figure 2A shows the extended surface on the C subunit that serves as a platform for docking RIα. RIα docks to at least three distinct subsites on the C subunit (Fig. 1C). Site 1 is where the basic inhibitor sequence docks to the acidic active site cleft, an interaction shared by all substrates and inhibitors of PKA. At site 2, the CBD-A, specifically the hydrophobic portion of the PBC containing Tyr205R, binds to an extensive hydrophobic surface that surrounds Tyr247C in the G helix. Finally, at site 3, the B/C helix in RIα docks to the activation loop of the C subunit. We consider site 1 first.

In this complex, AMP-PNP and RIα lock the C subunit into a fully closed conformation. The region of the linker corresponding to the inhibitor site is typically referred to as the P-3 through P+1 site (Arg94-Arg-Gly-Ala-Ile98), where the P site is filled by either Ser or Thr for substrates and Gly or Ala for pseudosubstrates. RIα and PKI are pseudosubstrates. As predicted from chemical modifications and from the structure of a 20-oligomer peptide containing the inhibitor consensus sequence of PKI (IP20) bound to the C subunit (15), this segment docks to the active site cleft in a manner similar to IP20 (Fig. 2B and fig. S3). Arg94R (P-3 site) and Arg95R (P-2 site) link the inhibitor sequence to several regions of the C subunit by interacting with Glu127C in the linker that joins the small and large lobes, Glu170C in the catalytic loop, and Glu230C in the F helix (Fig. 2C and movie S3). In this closed conformation, Tyr330C in the C-terminal tail interacts with the glycine-rich loop and hydrogen bonds to the 2′OH group of the ribose at the P-3 site. In addition, His87C in the C helix is hydrogen bonded to the phosphate on Thr197C in the activation loop. Ile98R binds to the P+1 loop. This part of the inhibitor sequence in RIα forms a complementary antiparallel β strand that links the glycine-rich loop and the N terminus of the C helix to the P+1 loop to form a β sheet (Fig. 2C). The detailed interactions are summarized in Fig. 3.

Fig. 3.

Specific interactions between the R and C subunits. The location of the each residue within the complex is listed alongside of each amino acid. Solvent molecules at the interface are shown as W. Ion pair, hydrogen-bond, and van der Waals interactions are notated as ↔, →, and Embedded Image, respectively.

In spite of sharing a common inhibitor site, IP20 and RIα use different surfaces of the C subunit to achieve high-affinity binding (Fig. 2D). Whereas the amphipathic helix of IP20 docks to a surface that lies N terminal to the inhibitor site, RIα docks to the surface that lies C terminal to the inhibitor site. As seen in Fig. 4A, binding of the P+1 residue (Ile98R) to the P+1 loop not only provides a local docking site but also appears to nucleate an extended hydrophobic interface between the C subunit (site 2) and CBD-A of RIα. A key element of this surface on the C subunit is the solvent exposed Tyr247C in the G helix, which is surrounded by a cluster of hydrophobic residues (Figs. 3 and 4A). Docking of Ile98R to the P+1 loop specifically links this hydrophobic cluster by means of Tyr247C to the hydrophobic surface surrounding Tyr205R in the PBC of RIα (Figs. 3 and 4, A and B). In its cAMP-bound state, cAMP is firmly anchored to the PBC through its interactions with two key residues, Arg209R and Glu200R (Fig. 4C). This complex of RIα and C shows that the C subunit also competes for the PBC by the interaction of Tyr247C of C with Tyr205R and Ile98R of RIα (Fig. 4B). The tip of the PBC in RIα is clearly stretched by these interactions, and this stretched PBC has lost its capacity to bind cAMP (Fig. 4C). Dislodging cAMP from the PBC is probably the key rate-limiting step required for holoenzyme formation.

Fig. 4.

Detailed features of site 2. (A) Extended hydrophobic interface created by binding of the P+1 residue (Ile98R) to the active site. On the left is the VDW surface of the C subunit and on the right is the complementing surface of RIα, both shown with the inhibitor/linker region and B/C helix of RIα. At the center are two surfaces together as they exist in the complex. Surfaces of RIα and C subunits are colored in yellow and gray, respectively. The inhibitor/linker region, shown as a ribbon, is color coded as follows: inhibitor sequence (residues 94 to 98), tan; linker segment (residues 99 to 105), green; residues 106 to 113, cyan. In cAMP-bound RIα, this entire segment is disordered. B and C helices are also in cyan. The surface of the C subunit is composed of P+1 loop and the hydrophobic face of the G helix. Two elements of the surface within the RIα are the hydrophobic residues from the helical tip of PBC and the X:NHelix to AHelix loop. The surface of the C subunit appears to be preformed, whereas the two elements forming the hydrophobic surface of RIα are brought together by conformational changes. (B) cAMP-binding site in the holoenzyme complex. Tyr205R at PBC forms a hydrogen bond (dotted line) with Tyr247C in the G helix. Surrounding this head-on hydrogen bonding between two tyrosines are clusters of hydrophobic residues from both subunits, as shown in (A). The side chain of the P+1 Ile nucleates this site. (C) The PBC (residues 199 to 210) of RIα in the cAMP-bound and C-bound conformations are aligned. The side chains, as in the C-bound conformation, are yellow, and the cAMP-bound conformations are gray. In addition to Tyr205R, Asp170R—which orients an important cAMP-binding residue, Arg209R—changes its rotomer position when C binds. All of the images were generated with VMD (51).

Site 3 on the docking surface of C is dominated by the activation loop containing Trp196C and Arg194C. They both interact with Glu105R at the linker segment of RIα (Fig. 5A). Earlier mutagenesis (19) and genetic (20) studies indicated that Trp196C would play an important role at the R:C interface. The phosphate of Thr197C links the C helix by means of His87C to the activation loop, and both of these segments of the C subunit interact directly with RIα. His87C also interacts with Gln84C, and Gln84C in turn binds to Ser99R in the linker region of RIα (Fig. 5A). Thus, like the glycine-rich loop, the N terminus of the C helix is an integral part of the RIα:C interface. The phosphate on Thr197C is also anchored to the catalytic loop at the active site through its interactions with Arg165C (Fig. 5B and fig. S2). Two regions of RIα, the C helix (Met234R) and the linker (Glu105R), interact with Trp196C and Arg194C in the activation loop (Figs. 3 and 5C).

Fig. 5.

Specific interactions associated with site 3. (A) Hydrogen-bonding network at site 3. Carbons from RIα are gray; carbons from the C subunit are tan. Specific interactions are also summarized in Fig. 3. (B) Interactions of the activation loop. Phosphorylation of Thr197C creates parts of the RIα:C interface. Each plays a critical role. Whereas Leu198C, Thr201C, and Gly200C contribute to recognition of the peptide, Trp196C and Arg194C provide a critical phosphorylation-dependent docking site for RIα. Asterisk is the reactive cystine. This figure was generated with Insight II (50). (C) Interactions between the activation loop and the B/C helix. The interface of Trp196C/Arg194C with the B/C helix of RIα is shown. Specific interactions between Trp196C/Arg194C of the C subunit and Glu105R/Met234R of RIα are highlighted. (D) Glu143R in the A helix binds to Lys213C of the C subunit (Fig. 3) as predicted by mutation study (16). The C helix and the A helix of RIα interact closely, thus helping to stabilize the inhibitory conformation. All of the images except (B) were generated with VMD (51).

Mutagenesis predicted that Lys213C in the C subunit formed an ion pair with Glu143R in RIα (19). The structure confirms this interaction and, in addition, shows that Lys213C brings together the A and C helices, which are far apart in cAMP-bound RIα (Fig. 5D). Lys213C is located in a highly conserved segment that links the activation loop with the F helix; we refer to this segment as the APE-F linker (Fig. 6A), where Glu208C in the APE motif provides the conserved C-terminal anchor for the activation loop. Whereas Lys213C interacts directly with RIα, two other nearby residues, Tyr215C and Lys217C, reach up and buttress the activation loop by hydrogen bonding to the backbone carbonyls of Thr197C and Ser191C, respectively. A peptide corresponding to this region was also predicted to be part of the RIα:C interface based on its protection from deuterium exchange (17, 18).

Fig. 6.

(A) The APE-F linker motif anchors the activation loop. The activation segment and the APE-F linker motif are rendered as dark gray and light gray ribbons, respectively. (B) Conserved features of the CBD in RIα. CBD-A is highly conserved where two charged residues, Arg209R and Glu200R, anchor the cAMP. Hydrophobic residues, shown with the van der Waals surfaces, surround the cAMP and shield the adenine ring. The B and C helices (cyan) interface with the PBC (red) through a hydrophobic hinge at the B helix and through hydrophobic capping of the adenine ring by Trp260R. In our construct, residues 248 to 260 (gold) are missing. This figure was adapted from figure 7C of (21). (C) Comparison of RIα in its cAMP-bound (residues 113 to 242) (14) and C subunit–bound (residues 91 to 242) conformations. The C subunit–bound conformation is on the left and cAMP-bound conformation is on right. The inhibitory/linker region and X:N and A helices are in yellow. PBC is in red and the B and C helices are in cyan. (D) Structure alignment of the two conformations of RIα. Specific conformational changes associated with the B and C helices (residues 226 to 242), the X:N and A helices (residues 113 to 150), and the β barrel (residues 151 to 225) are shown. The C-bound conformation is in cyan and the cAMP-bound conformation is in gray with PBC colored red in both conformations. The β-barrel subdomains minus the PBC (residues 151 to 200 and 210 to 225) of the two conformations align with a Cα root mean square deviation of 1.05 Å2, whereas the helical subdomains do not align. All of the figures were generated with Insight II (50).

Interaction sites on RIa. Based on a comparison of six cAMP-bound structures of CBDs, two elements of the CBD are predicted to be essential for signaling (21). The highly conserved PBC, embedded in the β-barrel subdomain, is the anchoring site for cAMP, as described earlier. The second element is the B/C helix (Fig. 6B). This element, which serves as a docking surface for interacting with other proteins or domains, is tethered to the PBC by a hydrophobic “hinge” between the B helix and the tip of the PBC (13, 22). This RIα:C complex is the first time that any CBD has been seen in both a cAMP-bound state and in a cAMP-free state bound to another protein. Removing cAMP from this deletion mutant of RIα is not sufficient to induce a major conformational change (23). It is, rather, binding of the C subunit that induces major conformational changes (Fig. 6C).

Appreciation of RIα(91–244) in its cAMP-bound state compared with its conformation in the complex can be best achieved by focusing on three regions (Figs. 1D and 6C). We discussed many specific residues earlier, and here we consider these global changes in RIα. The charged and hydrophilic linker region, completely disordered in cAMP-bound RIα, now becomes ordered at the interface (fig. SA). The B and C helices snap into a single fully extended helix where residues that were previously fully exposed to solvent are now embedded at the RIα:C interface. Finally, the remainder of the helical subdomain that includes the X:N helix and the A helix together with the interlinking loop rearranges to fill the space that was vacated by the movement of the C helix (Figs. 1D and 6C).

As described earlier, docking of the inhibitor site to the active site cleft results in ordering of the remaining linker. This region, highly conserved in RIα, contributes many charged residues (Figs. 1C and 2C). By interacting not only with C but also with the CBD, they function almost like a zipper to fasten the CBD-A to the large lobe of the C subunit. Most of the conformational changes in CBD-A are in the helical subdomain, with the exception of the tip of the PBC where cAMP docks. This segment is extended slightly by its interactions with Tyr247C in C and Ile98R in RIα (Figs. 1D, 4C, and 6C). The rest of the β barrel remains unchanged. In contrast, the helical subdomain of CBD-A undergoes major conformational changes as it docks to the surface of the C subunit. Most notably, the B and C helices become a single extended helix (Figs. 1D and 6D). Trp196C in the activation loop of the C subunit helps to anchor this extended helix through its interactions with Met234R in the C helix of RIα (Figs. 3 and 5C). In its cAMP-bound conformation, this entire surface of the B/C helix of the CBD-A is exposed to solvent (Fig. 1D). This surface, which includes Arg226R, Arg230R, Arg231R, Met234R, and Lys240R, now becomes embedded at the interface created by the large lobe of the C subunit (Fig. 5C).

Mechanism of inhibition and activation. Understanding how a protein kinase is inhibited is as important as understanding how it functions as a catalyst. Whereas appreciation of the kinase as a catalyst has focused primarily on the dynamic small lobe and the opening and closing of the active site cleft (5), this structure of the RIα:C complex draws our attention to the large lobe and reveals the importance of the C subunit as a scaffold. PKA has at least seven known physiological inhibitors, four functionally nonredundant R-subunit isoforms (24), and three PKI isoforms (25); all bind to C with subnanomolar affinities. Although all share a common substrate-like inhibitor site that binds to the active site cleft, the peripheral sites that allow each to bind with high affinity are distinct (Fig. 2, B to D, and fig. S1). PKI uses a rather small surface for docking an amphipathic helix that is N terminal to the inhibitor site (15), whereas RIα uses a very large surface C terminal to the inhibitor site and a mechanism that involves major conformational changes in the helical region of RIα (Fig. 2D). Preliminary evidence, based on mutagenesis and hydrogen-deuterium exchange, suggests that RIIβ may use yet another variation (26). Collectively, these sites span a substantial surface of the large lobe and define the large lobe as a major scaffold. In spite of the extended surface, the docking sites shown in Fig. 2A are all preformed; no conformational changes are induced in the C subunit, other than the closing of the active site cleft, as a consequence of these inhibitors binding. PKI, like RIα, requires two Mg2+ ions and ATP to form a stable complex (27), and the C subunit must assume a fully closed conformation. Formation of holoenzyme with RII subunits does not require ATP.

In addition to the inhibitor site, both the G helix and activation loop of the large lobe are essential for binding to RIα. In particular, Tyr247C in the G helix and Trp196C in the activation loop are important key residues that anchor RIα (Figs. 4, A and B, and 5C). Because both are highly conserved as aromatic residues in the protein kinase family, protein recognition will likely be a common feature shared by these residues in many protein kinases (28). The importance of the G helix as a hydrophobic docking site was demonstrated previously in the crystal structure of cdk2 bound to kinase-associated phosphatase (KAP), the phosphatase that removes the activation loop phosphate from Thr160 (29). In that complex, the phosphate on Thr160 of cdk2 binds directly to the catalytic site of KAP. Unlike the E, F, and H helices, which form the highly stable core of the large lobe and are extremely resistant to deuterium exchange (30), the G helix is more solvent accessible. Its backbone amides are only shielded from solvent when inhibitors are bound. This protection enabled us to identify this element as part of the RIα:C interface (18). We predict that the G helix will be a dynamic and solvent-accessible sensor for protein docking in many protein kinases.

The importance of a single phosphate in the activation loop was appreciated immediately when the first set of crystal structures was solved (4, 3134). In its fully phosphorylated and activated state, p-Thr197C is linked to the active site through Arg165C (Fig. 5B) (35). In inactive kinases, such as cdk2 (31), the loop is not phosphorylated and is typically disordered. The essential phosphorylation site on the activation loop is always coupled with an Arg-Asp (RD) motif (Arg165-Asp166 in the C subunit), in which the equivalent to Arg165C is anchored to the phosphate with Asp166C positioned at the active site in the catalytic loop, where Asp166C functions as a catalytic base and proton sink (3638). Biochemical studies with mutant proteins have demonstrated the importance of the activation loop phosphorylation for catalysis (6, 39, 40). The RIα:C structure highlights another important consequence of phosphorylating the activation loop: creating a binding surface for RIα. The importance of Trp196C for R-subunit inhibition was first recognized by a genetic screen designed to identify mutants that could not be regulated (20), and we confirmed experimentally that Trp196C was an essential feature of the R:C interface (19). However, this is only true when the C subunit is phosphorylated on Thr197C; dephosphorylated C subunit cannot bind RIα (41). As described earlier, Trp196C is packed against residues in the B/C helix of RIα(91–244), and, similar to Tyr205R, these residues are fully exposed to solvent when cAMP is bound to Δ(1–91)RIα (Figs. 5C and 6B).

The most notable revelation that emerges from the RIα:C complex is the remarkable conformational malleability of RIα. Clearly, the extended linker region needed to become ordered upon binding to the C subunit, and it was not possible to predict the molecular features of this docking in advance of the structure. However, the reorganization of the helical subdomain was not anticipated and the magnitude of the conformational changes is substantial (Figs. 1D and 6C). Although the β barrel that provides the docking site for the phosphate of cAMP remains largely intact, the helical subdomain is completely reorganized. Understanding the dynamics and the pathway for this change is another challenge for the future. We predict that in the absence of cAMP, the helical subdomain should be quite dynamic. The molecule is much less stable, and unfolding is no longer cooperative when cAMP is removed (42, 43). A stable conformation is created only in the presence of cAMP or C subunit. This mechanism of stabilizing very different conformations by either a small molecule ligand or another protein has not been observed previously.

Ligand-induced activation of the RIα holoenzyme is a highly cooperative and ordered allosteric process with cAMP binding first to CBD-B and then to CBD-A (44). Binding of cAMP to CBD-A then leads to release of the C subunit. In the complex examined here, CBD-B is missing. This domain is tightly anchored by multiple hydrophobic interactions to the segment that follows the C helix (residues 245 to 259). Trp260R at the beginning of the A helix of CBD-B provides the hydrophobic cap for the adenine ring of cAMP bound to CBD-A (Fig. 6B). Thus, when cAMP is released from the CBD-A, it removes a major hydrophobic anchor between the two domains, as predicted by Berman et al. (21). Because CBD-B is tightly anchored to residues 245 to 259 (14), it will move along with the C helix of CBD-A when the C subunit is bound. The conformational changes in the helical subdomain of CBD-A in RIα, as revealed in this complex, thus predict that there will be a major change in the orientation of CBD-B relative to CBD-A when RIα is bound to the C subunit. A notable conformational change was observed from the neutron scattering studies of the RIα holoenzyme compared to the RIα homodimer, and this data can now be modeled more precisely with the availability of this RIα:C complex (45). Another challenge now will be to confirm the spatial and temporal features of these global changes in conformation in the larger protein.

Supporting Online Material

www.sciencemag.org/cgi/content/full/307/5710/690/DC1

Materials and Methods

Figs. S1 to S3

Table S1

Movies S1 to S3

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article