Metal-free cAMP-dependent protein kinase can catalyze phosphoryl transfer.

X-ray structures of several ternary product complexes of the catalytic subunit of cAMP-dependent protein kinase (PKAc) have been determined with no bound metal ions and with Na(+) or K(+) coordinated at two metal-binding sites. The metal-free PKAc and the enzyme with alkali metals were able to facilitate the phosphoryl transfer reaction. In all studied complexes, the ATP and the substrate peptide (SP20) were modified into the products ADP and the phosphorylated peptide. The products of the phosphotransfer reaction were also found when ATP-γS, a nonhydrolyzable ATP analogue, reacted with SP20 in the PKAc active site containing no metals. Single turnover enzyme kinetics measurements utilizing (32)P-labeled ATP confirmed the phosphotransferase activity of the enzyme in the absence of metal ions and in the presence of alkali metals. In addition, the structure of the apo-PKAc binary complex with SP20 suggests that the sequence of binding events may become ordered in a metal-free environment, with SP20 binding first to prime the enzyme for subsequent ATP binding. Comparison of these structures reveals conformational and hydrogen bonding changes that might be important for the mechanism of catalysis.

Signaling through the phosphorylation of proteins by protein kinase enzymes plays an important role in the regulation of a variety of physiological responses. ATP-dependent protein kinases are phosphotransferases that deliver the γ-phosphoryl group from the nucleotide to the hydroxyl of a serine, threonine, or tyrosine of a substrate protein to produce the phosphomonoester and ADP products. Among the diverse serine/threonine protein kinase family, cAMP-dependent protein kinase A (PKA) has served as an extensively studied model for investigations of the mechanism of all kinases.[1−3]In vivo, the inactive PKA holoenzyme consists of two catalytic (C) and two regulatory (R) subunits. An increase in cAMP concentration, due to β-adrenergic stimulation of cells, activates PKA by providing four cAMP molecules that bind to the R subunits, which in turn dissociate from the C subunits (referred to here as PKAc) that had been bound in an inactive holoenzyme state.[4]

Experiment and theory have demonstrated that the catalytic cycle of PKAc involves several structural and dynamic steps. apo-PKAc is in a dynamically uncommitted open conformation, showing very little conformational exchange on the microsecond to millisecond time scale.[5−7] Catalysis initiates when MgATP binds to produce the binary PKAc–MgATP complex, which causes PKAc to transition to an intermediate conformation that itself is partially rate-limiting at physiological Mg2+ concentrations (∼0.5 mM).[6−10] Substrate binding to PKAc–MgATP produces the ternary complex in which PKAc adopts the closed conformation necessary for chemical catalysis,[11] although the ternary Michaelis complex may cycle between the closed and intermediate conformations multiple times before phosphorylation actually take place.[7] Following the dissociation of the phosphorylated product from the active site, MgADP is released, which is concurrent with the enzyme’s return to the open conformation and is the second potentially rate-limiting step.[5,12]

Divalent metal cations are thought to play important roles throughout the catalytic cycle. Early studies pointed to a fundamental role for divalent metal cations in PKAc activity.[13,14] The PKAc active site possesses two metal-binding sites (M1 and M2; Figure 1). The occupancy of one site is sufficient for activity, whereas occupancy of the second site improves the affinity to both ATP and ADP. Indeed, in the absence of divalent cations in solution, no nucleotide binding or phopshotransferase activity has been detected.[15] Early reports suggested that at low physiological concentrations Mg2+ binds first to the M1 site, coordinated by β- and γ-phosphates of ATP.[16−18] The M2 site, coordinated by α- and γ-phosphates of ATP, becomes occupied at higher metal concentrations.[10,17,19] When both sites are occupied at high metal concentrations, ADP release becomes the rate-limiting process.[9]

Figure 1

A close-up view of the PKAc structure in cartoon representation (PDB ID 4IAC, Gerlits et al.[25]). The two Mg2+ cations at sites M1 and M2 are represented by magenta spheres, while a nucleotide is shown in stick representation colored by atom type. SP20 peptide is colored blue, and the substrate’s Ser21 is shown in stick representation colored by atom type.

Our understanding of the role of metal in the catalytic cycle has been challenged recently by X-ray crystallography studies that revealed that M2 rather than M1 is occupied at low Mg2+ concentrations.[20] Computational studies also supported greater involvement of M2 in chemical catalysis than M1.[21−23] Further, although Mg2+ is considered to be the physiologically relevant metal based on its relatively high concentration in cells,[15,24] it has been demonstrated that all other divalent alkaline earth metals (Ca2+, Sr2+, and Ba2+) support the phosphoryl transfer in complexes with the 20-residue substrate analogue peptide SP20.[25] This unexpected result contradicted early kinetic measurements showing no PKAc activity in the presence of Ca2+ or Sr2+ using the Kemptide (LRRASLG) peptide substrate and no nucleotide binding when Ba2+ was present.[15]

The fact that ATP binding to PKAc is undetectable in the absence of divalent metals[15] leads us to propose that high substrate binding affinity can produce PKAc activity in a metal-free environment. SP20 has nanomolar binding affinity to PKAc.[26,27] Here, X-ray crystallography and single turnover enzyme activity measurements with [γ-32P]ATP were used to determine whether PKAc can bind ATP and transfer the γ-phosphoryl group to Ser21 of SP20 in the absence of free divalent metal ions. The results show that a stable PKAc–SP20 binary complex forms that can coordinate ATP to form the catalytically active ternary complex. The results reinforce the conclusions of several computational studies[21−23] that suggest that the primary role of the metal ions is to lower the transition state energy through electrostatic effects, rather than bond making or breaking during the chemical reaction. Importantly, the current study provides a major revision to our understanding of the role of the different components involved in the catalytic cycle of PKAc.

Materials and Methods

General Information

Pseudosubstrate peptide SP20 (TTYADFIASGRTGRRASIHD; residues 5–24 of the heat-stable PKAc inhibitor PKI, where positions 20 and 21 have been mutated to alanine and serine) was custom-synthesized by and purchased from Biomatik (Wilmington, Delaware, USA). ATP, as the disodium salt, and ATP-γS (in which sulfur replaces one of the terminal oxygen atoms), as a tetralithium salt, were purchased from Sigma-Aldrich (St. Louis, Missouri, USA). Protein purification supplies were purchased from GE Healthcare (Piscataway, New Jersey, USA). Crystallization reagents were purchased from Hampton Research (Aliso Viejo, California, USA).

Protein Expression and Purification

His6-tagged recombinant mouse PKAc was expressed in Escherichia coli using Luria-Bertani (LB) media at 16–24 °C for 16–20 h. The recombinant enzyme was purified by affinity chromatography using HisTrap fast-flow chromatography columns supplied by GE Healthcare. The enzyme was then buffer-exchanged with 50 mM 2-(N-morpholino)ethanesulfonic acid (MES)–NH4OH, 250 mM NH4Cl, 2 mM dithiothreitol (DTT) at pH 6.5 on a desalting column. For PKAc–Na2ADP–pSP20, the enzyme was buffer-exchanged with 50 mM MES–NaOH, 250 mM NaCl, 2 mM DTT, pH 6.5. Isoforms of PKAc, which differ by the number and positions of autophosphorylated residues, were not separated, without any obvious effect on crystallization of the ternary complexes.

Enzyme Activity Measurements

The activity of PKAc in the presence and absence of monovalent and divalent metals was measured in the single turnover regime utilizing 32P-labeled ATP (ATP*) and SP20 as substrates. The kinetics measurements were carried out in the presence of 40 μM ethylenediaminetetraacetic acid (EDTA) at 25 °C. The common reaction mix contained 50 mM 3-(N-morpholino)propanesulfonic acid (MOPS), pH 7.4, 10 μM SP20, 2 μM enzyme, 1 mM ATP, and ATP radiolabeled with 32P in the γ-position (ATP*, specific activity 500–1500 cpm/pmol) in a final volume of 20 μL. The various metal ions used were Mg2+, Ca2+, K+, and Na+ at final concentrations of 1 mM. The reaction was carried out as a time course experiment with 0, 60, and 300 min.

The reaction was initiated by adding 10 μL of the mix containing ATP, ATP*, and PKAc to 10 μL of the solution with metal (or no metal) and SP20. The γ-phosphoryl group of ATP* containing the radioactive marker was transferred to Ser21 of SP20 during the measurements. After each time-point, the reaction was quenched with 90 μL of 30% acetic acid. A 50 μL portion of the quenched reaction was then spotted on p81 phosphocellulose paper, washed 3 times for 5 min each with 5% phosphoric acid and once with acetone, air-dried, and counted on a liquid scintillation counter. The background counts were subtracted from the measured values. The difference in counts (total counts minus background counts) was divided by the activity of 32P ATP (cpm/pmol) to get the concentration of the phosphorylated pSP20 product.

Crystallization and Data Collection

For crystallization trials, PKAc was concentrated to 8–12 mg/mL. The ternary and binary complexes with no metals, Na+ or K+, ATP (or ATP-γS), and SP20 were made before crystallization trials were set up. For the metal-free binary PKAc–SP20 complex, the enzyme solution was mixed with the peptide substrate, and (NH4)2EDTA was added to reach a final concentration of 20 mM. For the metal-free ternary complexes, the enzyme solution was combined with a premade solution of Na2ATP in 50 mM MES–NH4OH, 250 mM NH4Cl, 1 M (NH4)2EDTA, and 2 mM DTT at pH 6.5, and then peptide substrate was added. The molar ratio of PKAc/ATP(or ATP-γS)/SP20 was kept at 1:5:5. The final concentration of EDTA in the ternary complex solutions was 50 mM, and that of ATP was 5 mM. EDTA has about 105 higher affinity than ATP for Mg2+ and other divalent and trivalent metal ions, so it should chelate any trace amounts of metal present. For PKAc–Na2ADP–pSP20 and PKAc–K2ADP–pSP20, the enzyme solutions were mixed with Na2ATP and SP20. Crystals were grown in sitting drop microbridges or in 9-well glass plates using well solutions consisting of 100 mM MES–NH4OH, pH 6.5, 5 mM DTT, 15–20% PEG 4000, and 150 mM NaCl for PKAc–Na2ADP–pSP20 or 50 mM KCl for PKAc–K2ADP–pSP20 at 4–14 °C. PKAc–SP20 crystallized in 15–20% PEG3350.

Structure Determination and Refinement

X-ray crystallographic data were collected from frozen crystals at 100 K for all complexes. The data sets were collected on a Rigaku HomeFlux system, equipped with a MicroMax-007 HF generator and Osmic VariMax optics. The diffraction images were obtained using an RAXIS-IV++ image-plate detector. For the binary PKAc–SP20 complex, a 1.7 Å resolution diffraction data set was collected at the Advanced Photon Source on the beamline 19-ID. Diffraction data were collected, integrated, and scaled using HKL3000 software suite.[28] The structures were refined using SHELX-97[29] for PKAc–ADP–pSP20, PKAc–Na2ADP–pSP20, and PKAc–K2ADP–pSP20, and PHENIX[30] for PKAc–SP20, PKAc–ADP–psSP20. A summary of the crystallographic data and refinement is given in Supporting Table 1, Supporting Information. Similar to our previous observations,[20,25] all of the structures were of isoform 2 and contained three post-translationally phosphorylated residues, Ser139, Thr197, and Ser338.

The structure of the ternary complex of PKAc with 2Mg2+, ADP, and phosphorylated peptide pSP20 (PDB ID 4IAD)[25] was used as a starting model for structure determination. Structures were built and manipulated with program Coot,[31] whereas the figures were generated using the PyMol molecular graphics software (v.1.5.0.3; Schrödinger LLC). The absence of metal ions was confirmed by the lack of electron density at the M1 and M2 sites in the FO – FC difference electron density maps. The presence of metal ions (Na+ and K+) was established by ensuring that the strongest peaks in FO – FC maps with metals omitted and 2FO – FC maps with metals present corresponded to positions M1 and M2 and by checking that the coordination spheres of seven ligands were present around the metal ions. Differences in coordination geometries were also used to distinguish between metal ions. As an example, in the case of PKAc–Na2ADP–pSP20, we determined that the metal was sodium because the coordination around the metal ions differed from that found in our previously published product complex with magnesium, and because no magnesium was used during the sample preparation. Potassium has almost twice the number of electrons as sodium or magnesium; therefore, K+ was identified by a much larger electron density peak than that for the smaller sodium metal. The representative electron densities in the active sites of PKAc–ADP–psSP20, PKAc–Na2ADP–pSP20, and PKAc–K2ADP–pSP20 complexes are shown in Supporting Figures 1 and 2, Supporting Information, while coordination of metals in Na+ and K+ complexes are given in Supporting Table 2 and Supporting Figure 3, Supporting Information.

The structures have been deposited to the Protein Data Bank and were assigned the following PDB codes: 4IB3 for PKAc–ADP–pSP20 (ternary complex of PKAc with products ADP and phosphorylated SP20); 4O21 for PKAc–ADP–psSP20 (ternary complex of PKAc with products ADP and thiophosphorylated SP20); 4IB0 for PKAc–Na2ADP–pSP20 (ternary complex of PKAc with two sodium cations and products ADP and phosphorylated SP20); 4IB1 for PKAc–K2ADP–pSP20 (ternary complex of PKAc with two potassium cations and products ADP and phosphorylated SP20); and 4O22 for PKAc–SP20.

Results

Enzyme Activity

In the presence of Mg2+ and Ca2+, the phosphotransfer reaction was already at equilibrium after 60 min, and the amount of the product remained unchanged after 300 min (Figure 2). The PKAc activity was significantly lower when no metal or monovalent metals were present in the reaction mixture, producing ∼2% and 5–10% of the phosphorylated product after 60 and 300 min, respectively. The enzyme’s activity is substantially impaired, but not abolished, without divalent metal ions relative to its performance with Mg2+ and Ca2+. Interestingly, metal-free PKAc was as active as that with Na+ or K+, indicating that the alkali metal ions do not play a role in the phosphoryl transfer.

Figure 2

Amounts of pSP20 synthesized by PKAc at two time points of 60 (green bars) and 300 min (blue bars) in the presence of no metal, Na+, K+, Mg2+, and Ca2+.

Ternary Complexes with ADP, pSP20, and No Metal

The active site of PKAc–ADP–pSP20 is shown in Figure 3a and is free of bound metal. Contrary to previous studies, the phosphotransfer reaction has occurred, and the products, ADP and pSP20, are trapped (Figure 3b). The positions of the products are stabilized by a network of potential hydrogen bonds and water-mediated interactions. The PKAc–ADP–psSP20 structure is very similar to PKAc–ADP–pSP20, with the only structural difference confined to the glycine-loop region (Supporting Figure 4, Supporting Information). This flap-like hairpin structure spans residues 49–57, contains three glycine residues, and acts as a lid for the catalytic site cavity. The flap is slightly more compact in PKAc–ADP–psSP20 due to a 1.5 Å shift of residues 52–54 away from the alternate conformation of thiophosphate side chain of psSer21SP20. As a consequence, a hydrogen bond does not form between the thiophosphate of psSer21SP20 and the side chain OH of Ser53 in the glycine-rich loop in PKAc–ADP–psSP20, unlike in PKAc–ADP–pSP20 and the divalent metal-bound ternary product complexes reported previously.[25] Instead, a new hydrogen bond forms between the amide of Gly52 and the carbonyl of Gly55 of PKAc–ADP–psSP20. This 2.8–3.2 Å long hydrogen bond is found in all product complexes containing divalent metals, indicating that, in addition to being flexible as a whole, the flap is also internally flexible.

Figure 3

(a) A close-up view of the enzyme active site in PKAc–ADP–pSP20 product complex showing ADP, phosphorylated pSer21SP20, and the enzyme’s residues important for catalysis. Possible hydrogen bonds are represented as red dashed lines. (b) Electron density for the active site components ADP, Asp166, Lys168, and pSer21SP20 contoured at the 1.5σ level (the contour level is 1.0σ for ADP to show the phosphates). (c) Superposition of the active sites from PKAc–ADP–pSP20 and PKAc–Mg2ADP–pSP20. The distance between the phosphorus atoms in pSer21SP20 residues is shown as black dashed arrow, indicating that the product phosphate moved over 1 Å in the metal-free structure toward Asp166 and Lys168 relative to its position in the magnesium structure. Distances are in Å.

The β-phosphate of ADP is quite mobile in both metal-free ternary complexes, having B factors of the order of 70–100 Å2, whereas the pSer21SP20 phosphate and psSer21SP20 thiophosphate groups are less mobile, having the B factors of the order of 40–50 Å2. The pSer21SP20’s phosphate group is drawn 1.3 Å closer to the active site residues relative to its position in PKAc–Mg2ADP–pSP20,[25] possibly due to the lack of metal ions. This shift is accomplished without a rotation around the Oγ–Cβ bond of the phosphorylated serine (Figure 3c). The pSer21SP20 side chain position creates hydrogen bonds with Asp166 and Lys168 in the metal-free PKAc–ADP–pSP20 structure that are absent from the metal–product complexes. Similar hydrogen bonds form with the sulfur atom of psSer21SP20 in one of the alternate conformations.

Ternary Complexes with ADP, pSP20, and Excess Na+ and K+

Metal was observed in both the M1 and M2 sites of the PKAc–Na2ADP–pSP20 and PKAc–K2ADP–pSP20 structures with the phosphotransfer products despite the fact that ionic radii of Na+ (1.0 Å) and K+ (1.38 Å) are significantly larger than that of Mg2+ (0.7 Å). The active sites of the Na+ and K+ structures are very similar to PKAc–Mg2ADP–pSP20, including the position and orientation of the ADP and pSer21SP20 phosphate groups.[25] Further, the pSer21SP20 phosphoryl groups are both oriented away from Asp166 and Lys168, which maintains a 2.5–2.6 Å long hydrogen bond to the OH of Ser53 in the glycine-rich loop (Supporting Figure 5, Supporting Information), which compares well with the 2.8 Å distance in PKAc–Mg2ADP–pSP20. Ser53’s side chain (Cβ–Oγ bond) is rotated by almost 180° from its conformation in PKAc–Mg2ADP–pSP20, which allows it to face Phe54 and to make a hydrogen bond with the imidazole of His23SP20.

The octahedral geometry around the magnesium ions in PKAc–Mg2ADP–pSP20 does not form in PKAc–Na2ADP–pSP20 or PKAc–K2ADP–pSP20. Na+ and K+ have seven ligands at M1 and M2, consistent with preferences of alkali metals for coordination numbers above six. Here, the seventh ligands are water molecules. The metal ion coordination spheres, metal–ligand distances, and metal hydration in the alkali metal complexes differ from the magnesium product complex (Supporting Table 2 and Supporting Figures 3 and 5, Supporting Information). As was observed in the Ca2+, Sr2+, and Ba2+ complexes, the terminal phosphoryl groups on ADP and pSer21SP20 are slightly (∼0.5 Å) father away from each other compared with their locations in PKAc–Mg2ADP–pSP20. These different active site geometries may be explained by the larger Na+ and K+ ionic radii, as well as the higher coordination numbers.

Binary Complex with SP20 and No Metal

The active site of the PKAc–SP20 complex is shown in Figure 4a. The intact peptide substrate is bound to the enzyme, and the unmodified OH group of Ser21SP20 can be seen (Figure 4b). No density that can be attributed to metal ions is present in either the M1 or M2 sites. The side chain of Ser21SP20 is rotated toward PKAc, forming hydrogen bonds with Asp166 and Lys168. In contrast, the Cβ–Oγ bond is rotated by ∼140° toward the bulk solvent in the metal-free and metal-bound product complexes (Figure 5). The absence of a nucleotide causes PKAc to adopt a conformation that is intermediate between the apo-form open conformation and the closed conformation of the ternary complexes. There is a significant shift of the glycine-rich loop (4 Å for Ser53) upward toward the α-helix in the small lobe of PKAc, displacing residues 76 through 84 from the positions adopted in PKAc–ADP–pSP20 and PKAc–Mg2ADP–pSP20 (Supporting Figure 5, Supporting Information).

Figure 4

(a) A close-up view of the enzyme active site in PKAc–SP20 binary complex showing intact Ser21SP20 and the enzyme’s residues important for catalysis. Possible hydrogen bonds are represented as red dashed lines. (b) Electron density for the active site residues, Ser21SP20, and water molecules contoured at the 2.0σ level (the contour level is 1.5σ for H2O). Distances are in Å.

Figure 5

Superposition of PKAc–SP20 and PKAc–ADP–pSP20 showing the conformation of the Cβ–Oγ bond of the Ser21SP20 side chain before and after the phosphotransfer reaction. Hydrogen bonding formed by Ser21SP20 and pSer21SP20 with active-site residues Asp166 and Lsy168 is shown as dashed lines. Distances are in Å.

Discussion

The phosphotransferase activity of PKAc is highly dependent on divalent metal ions. It has been previously shown that PKAc binds divalent metals weakly in the absence of a nucleotide, with KD > 1 mM.[13] Similarly, PKAc does not bind ATP without divalent metals.[15] For a short peptide substrate Kemptide, which has moderate binding affinity, PKAc shows activity with only a few types of divalent metal ions.[15] In contrast, PKAc is capable of transferring the γ-phosphoryl group from ATP to the SP20 peptide substrate, which binds tightly to PKAc, when any alkaline earth metal is present.[25] Further, the results presented here conclusively demonstrate that phosphotransfer can take place without divalent metals (Figure 2), although the enzyme activity is greatly diminished. Interestingly, alkali metal ions have no effect on the kinase activity with SP20, because the amount of pSP20 was measured to be similar in the reactions with no metal, Na+, and K+, raising the possibility that these cations may not be present when the phosphotransfer takes place. The only other kinase known to function without divalent metal ions is the pseudokinase CASK, whose activity is actually inhibited by divalent cations in the order of Mg2+ > Mn2+ > Ca2+.[32] It is important to note that previous studies have investigated removal of key functionalities from the PKAc active site, albeit by mutagenesis, and also failed to completely abolish the enzyme’s activity. The single mutation variants Lys72Ala, Lys168Ala, and Asp166Ala showed greatly diminished phosphotransferase activity, being below 1% that of the wild-type protein,[33,34] much like the present results. These three residues form hydrogen bonds with the phosphate groups of ATP and substrate,[16,17,20,24,25] and their primary roles in the reaction are to facilitate the phosphoryl transfer.[1] Substitutions of the other active site residue, Ser53, had no effect on the steady-state phosphorylation of the substrate Kemptide, indicating that it does not participate in the reaction.[35]

In the current study, the single-turnover enzyme activity measurements quantitatively determined the amounts of pSP20 that were phosphorylated by metal-free PKAc, although SP20 phosphorylation was not detected in steady-state kinetics experiments. While the measurements did not provide a reaction rate constant and utilized reaction incubation times much longer than steady-state techniques,[19] the results demonstrate phosphorylation of a substrate in the absence of divalent cations. In contrast, no kinase activity was detected via steady-state kinetics measurements when PKAc was reacted with Kemptide in the metal-free environment or in the presence of Na+ or K+. The ability of PKAc to phosphorylate SP20, but not Kemptide, in metal-free conditions may be related to the extremely high binding affinity (∼100 nM) of SP20, whose sequence is based on the heat-stable protein kinase inhibitor (PKI).[26,27] The productive binding of a substrate and MgATP to the apo-active site of PKAc has been considered formally random,[19] but almost simultaneous,[7] with the preferred initial binding of MgATP.[36] Based on the present results, we propose that the sequence of binding events may become ordered in a metal-free environment, with SP20 binding first to prime the enzyme for subsequent ATP binding. The PKAc–SP20 binary complex is stable and has an intermediate, partially closed, conformation (Supporting Figure 6, Supporting Information), suggesting that such a sequence of events is possible. Previously, the X-ray structure of the similar binary complex containing phosphorylated SP20 has been reported.[37] In addition, studies on the pH dependence of the kinetic mechanism demonstrated that pSP20 bound prior to MgADP in the reaction in the direction of MgADP phosphorylation at pH 6.5,[38] which supports our hypothesis. An equimolar mixture of ATP and Mg2+ contains predominantly the MgATP2– complex, with low but appreciable amounts of ATP4– and HATP3–.[39] HATP3–, in which one of the γ-phosphoryl oxygens is protonated, has the ionization constant (pKa) of 6.5;[40] therefore, it is expected that 50% of ATP will be in the form of HATP3– in the studies reported here. The PKAc active site has a net negative charge, which would repel the bare ATP4– more strongly than it would the singly protonated HATP3–. Consequently, it is reasonable to propose that the monoprotonated nucleotide species HATP3– is the reactive molecule that binds to the PKAc–SP20 binary complex to generate a catalytically viable ternary complex that subsequently reacts to produce the product complex PKAc–ADP–pSP20 observed in our crystal structure.

Cook et al.[19] previously showed that KCl increases the KM of MgATP and the Ki of MgADP at physiological magnesium concentrations, when one Mg2+ ion binds to the PKAc active site. The same phenomenon was observed for Na2SO4, which eliminates competitive inhibition of PKAc by chloride ions. These observations can be explained by the alkali metal binding to the second metal site at concentrations of ∼100 mM, thereby increasing the counterion charge to 3 and resulting in the diminished affinity of ATP and ADP for the active site. The product structures PKAc–Na2ADP–pSP20 and PKAc–K2ADP–pSP20 demonstrate that alkali metals can bind to PKAc even though alkali metals have extremely low ATP-binding constants (∼100 mM).[41] However, the single turnover kinetics data indicate that the rate of SP20 phosphorylation is about the same in the presence of Na+ or K+ as it is in the absence of any metals. We propose that Na+ and K+ may not come to the PKAc active site bound to ATP but instead diffuse opportunistically into the PKAc–ADP–pSP20 complex after the reaction takes place.

Recent molecular dynamics simulations of the Michaelis complexes reported that the OH group of a substrate peptide’s serine adopts a substrate-dependent conformation. In Kemptide, the OH was rotated toward Asp166, whereas in SP20, it is pointed away from Asp166 into the bulk solvent.[42] In PKAc–SP20, the side chain of Ser21SP20 is rotated toward Asp166 and Lys168, forming hydrogen bond interactions with the aspartate’s carboxylic and the lysine’s amino groups. The same “inward” conformation of the hydroxyl was found in earlier PKAc complexes that mimicked the transition state and product structures.[34,43] Intriguingly, in PKAc–ADP–pSP20 and in all product complexes with metals[25] the hydroxyl moiety of pSP20 adopts a different “outward” conformation, being rotated out by ∼140° relative to its position in PKAc–SP20 (Figure 5). Conformational flexibility of the substrate’s OH may be an important factor in the mechanism of phosphoryl transfer.

Enzymes are remarkable catalysts that can dramatically lower the energy barrier that a chemical reaction has to overcome in several ways including lowering the reaction entropy by bringing together and correctly positioning reactants (which improves the chance of collision between the reactants relative to solution) and lowering the activation energy of the reaction by distorting reactants and stabilizing the transition state through the creation of a favorable environment.[44,45] Each part of an enzyme, whether it be the catalytic residues, cofactors, or moving ensembles of residues (loops or domains), can be involved, but the reaction pathway is not dramatically altered relative to solution chemistry. In the case of PKAc, computational studies suggest that metals stabilize the phosphotransfer reaction’s transition state having the pentacoordinated γ-phosphorus.[21−23,46,47] Our observations of structures of PKAc and its activity when no free metal is present indicate that divalent cations do not constrain the active site geometrically to promote the reaction and are not engaged in the chemical transformation. Instead, they provide a favorable electrostatic environment for the reaction thereby lowering the energy of the transition state.

It follows that the absence of PKAc bound metals dramatically increases the reaction energy barrier for the tightly bound SP20, but the reaction still proceeds because the enzyme promotes a favorable reduction in reaction entropy through the binding of reactants. Baseline catalytic activity has also been reported for other enzymes (which may be reduced by orders of magnitude) in the absence of seemingly critical functionalities. For example, NADH oxidase from Thermus thermophilus showed detectable activity, albeit reduced by about 2 orders of magnitude, in the absence of flavin adenine mononucleotide, which is an essential electron acceptor.[48] Similarly, the metalloenzyme cytidine 3′,5′-cyclic monophosphate phospho-diesterase can catalyze the hydrolysis reaction with no divalent metal cofactor present.[49] For these other enzymes, it is likely that a less optimal pathway has been found to facilitate the reaction. However, the results presented here for PKAc suggest that we should not discount the general significance of the simple reduction in reaction entropy provided by enzymes when they bind reactants, that is, the “entropy trap” effect remains an important aspect of enzyme catalysis.

Conclusions

We have provided new evidence that PKAc is active in the presence of other divalent metal cations, rather than just with the previously assumed physiological Mg2+ cofactor. This raises the important possibility that Ca2+ may also be a physiological cofactor. Further, we have demonstrated that PKAc is active in the absence of metals, which is only the second example of such a kinase. This result has profound implications for our understanding of the role that metals play in PKAc catalysis, expanding on our previous work showing activity in the presence of all alkali earth metals.[25] Excitingly, the metal-free catalytic cycle appears to involve an ordered binding sequence of substrate followed by ATP, in contrast to the random, nearly simultaneous binding that takes place with divalent metals. This raises the interesting possibility that the metal-free complex may be a suitable model system for resolving dynamic states, and the transitions between them associated with ATP binding.