Phosphoryl transfer by protein kinase A is captured in a crystal lattice

Abstract

The catalytic (C) subunit of cAMP-dependent protein kinase (PKA) is a serine/threonine kinase responsible for most of the effects of cAMP signaling, and PKA serves as a prototype for the entire kinase family. Despite multiple studies of PKA, the steps involved in phosphoryl transfer, the roles of the catalytically essential magnesium ions, and the processes that govern the rate-limiting step of ADP release are unresolved. Here we identified conditions that yielded slow phosphoryl transfer of the γ-phosphate from the generally nonhydrolyzable analog of ATP, adenosine-5'-(β,γ-imido)triphosphate (AMP-PNP), onto a substrate peptide within protein crystals. By trapping both products in the crystal lattice, we now have a complete resolution profile of all the catalytic steps. One crystal structure refined to 1.55 Å resolution shows two states of the protein with 55% displaying intact AMP-PNP and an unphosphorylated substrate and 45% displaying transfer of the γ-phosphate of AMP-PNP onto the substrate peptide yielding AMP-PN and a phosphorylated substrate. Another structure refined to 2.15 Å resolution displays complete phosphoryl transfer to the substrate. These structures, in addition to trapping both products in the crystal lattice, implicate one magnesium ion, previously termed Mg2, as the more stably bound ion. Following phosphoryl transfer, Mg2 recruits a water molecule to retain an octahedral coordination geometry suggesting the strong binding character of this magnesium ion, and Mg2 remains in the active site following complete phosphoryl transfer while Mg1 is expelled. Loss of Mg1 may thus be an important part of the rate-limiting step of ADP release.

Figure 1

Crystal structure of PKA displays partial phosphoryl transfer of AMP-PNP onto a substrate peptide. Panel A) The overall 1.55 Å PKA structure, 4HPU, is displayed in cartoon representation with the small lobe (12–126) colored white, large lobe (127–350) colored olive, SP20 colored red, P-site serine and ANP-PNP shown in stick representation and colored by element, and magnesium ions shown as spheres and colored blue. Panel B) The 2F_o-F_c electron density map contoured to 1 σ is displayed showing partial phosphoryl transfer and partial intact AMP-PNP. The structure is modeled with 45% phosphoryl transfer, 55% reactants. Panel C) The same density as in B with the free phosphate from 1RDQ displayed in gray to illustrate that the electron density corresponds to phosphoryl transfer and not hydrolyzed, free phosphate.

Figure 2

Partial phosphoryl transfer is further verified with F_o-F_c maps with different components omitted. The F_o-F_c electron density maps contoured to 3 σ are shown when AMP-PNP and no phosphoryl transfer are modeled Panel A; when complete phosphoryl transfer to the P-site serine and ADP are modeled Panel B; and when no phosphoryl transfer and ADP are modeled Panel C. When either P-site phosphorylation (A) or the γ-phosphate of AMP-PNP (B) are excluded from the model, the electron density map clearly shows positive electron density that corresponds to each phosphate group. This finding is also confirmed when both P-site Ser phosphorylation and the γ-phosphate of AMP-PNP are excluded from the model (C).

Figure 3

The 2.15 Å crystal structure displays complete phosphoryl transfer of AMP-PNP onto the substrate peptide. Panel A) The 2F_o-F_c electron density map at 1 σ is displayed for 4HPT showing electron density for complete phosphoryl transfer. Panel B) The hydrolyzed phosphate from 1RDQ is aligned with this structure and colored gray showing that the electron density corresponds to phosphoryl transfer and not hydrolyzed phosphate. Panel C) The F_o-F_c electron density map generated from a model with ADP and no phosphoryl transfer modeled in the structure is displayed at 3 σ showing positive electron density for phosphoryl transfer with no density for the γ-phosphate of AMP-PNP.

Figure 4

Mg1 is expelled from the active site following complete phosphoryl transfer. The 2F_o-F_c electron density map at 2 σ is displayed for the magnesium ions in the 1.55 Å, partial phosphoryl transfer structure, 4HPU, Panel A and in the 2.15 Å, complete phosphoryl transfer structure, 4HPT, Panel B. There is much less electron density for Mg1 than typically seen for magnesium ions in the 2.15 Å complete phosphoryl transfer structure which indicates partial, if not complete, loss of this magnesium ion.

Figure 5

Coordination of the magnesium ions before and after phosphoryl transfer. Panel A) A stereo view of the magnesium ions in the 1.55 Å partial phosphoryl transfer structure is displayed. AMP-PNP is colored by element, AMP-PN (without γ-phosphate) is colored black, water molecules are colored gray, and magnesium ions are colored blue. The distances between Asp184 and Mg1 are shown in Angstroms. Panel B) A stereo view of the magnesium ions in the 2.15 Å complete phosphoryl transfer structure is shown. The AMP-PN from the structure is colored by element, water molecules are colored gray, magnesium ions are colored blue, and AMP-PNP from the 1.55 Å partial phosphoryl transfer structure is aligned and shown in black. Mg2 recruits an additional water molecule that fulfills the vacancy left after transfer of the γ-phosphate highlighted in yellow. Panel C) Coloring is the same as in Panel A. The F_o-F_c electron density map at 3 σ is shown in blue for the area near the new water molecule recruited by Mg2 following phosphoryl transfer in the 1.55 Å partial phosphoryl transfer structure. Thus Mg2 immediately recruits this new water molecule which was included in the final model.

Figure 6

Changes in the Gly-rich loop with substrates bound and following phosphoryl transfer. Panel A) The Gly-rich loop is slightly raised in the structures reported here, full transfer (4HPT) depicted in olive and partial transfer (4HPU) depicted in gray, compared to the fully closed structure typically observed with ATP and IP20 (depicted in red from PDB: 3FJQ). The raised loop may result from the presence of a serine rather than alanine residue at the P-site as evident from previous C-subunit structures containing a substrate bound at the active site such as SP20 (depicted in blue from PDB: 4DG0) and RIIβ(108–268) (depicted in black from PDB: 3IDB). Panel B) 4DG0 and 3FJQ are depicted in the same colors as Panel A with ATP from 3FJQ colored by element and AMP-PNP from 4DG0 colored black. This alignment highlights that the raised Gly-rich loop may be the result of substrate binding alone and not phosphoryl transfer since 4DG0 which was bound to SP20 but exhibited no phosphoryl transfer also has a raised Gly-rich loop.

Figure 7

Catalytic cycle of PKA. A similar cycle is also predicted for CDK2.