Theoretical studies of the function switch and mechanism of AceK as a highly active ATPase

Abstract

As a multi-function enzyme, AceK integrates kinase, phosphatase and ATPase activities in a single active site. In contrast to most kinases, AceK exhibits unusually high ATPase activity compared to its kinase and phosphatase activities. The reason that AceK possesses such a high ATPase activity and its multi-function regulation are still elusive. In this work, we have employed DFT methods to exploit the ATP hydrolysis mechanism of AceK and revealed a dissociative pathway with an activation energy of only 17.85 kcal mol⁻¹, which is highly favorable for ATPase activity. The high ATPase activity of AceK may play a role in producing ADP as a proton acceptor to fulfill its phosphatase function. Based on our calculation and structural analysis, binding with substrate ICDH causes a catalytically important residue, Asp477, to flip over and further suppress ATPase activity with a markedly increased activation energy of 21.68 kcal mol⁻¹, thus favoring kinase or phosphatase activity. Our work has shed new light on the function switch and ATP hydrolysis mechanism of AceK.

---

Introduction

Isocitrate dehydrogenase kinase/phosphatase (AceK) is a reversible enzyme in prokaryotes which regulates the action of isocitrate dehydrogenase (ICDH) at the branch point between the Krebs cycle and glyoxylate bypass in response to nutrient availability.^1,2 As the key metabolic and regulative enzyme for Escherichia coli, AceK uniquely integrates kinase, phosphatase and ATPase activities in the same active site.^3–5 AceK phosphorylates or dephosphorylates its substrate ICDH in its kinase or phosphatase reaction. Moreover, AceK possesses unusually high ATPase activity, which is about ten times higher than its kinase and phosphatase functions based on kinetic analyses.⁴ In contrast, the ATPase activity of typical kinases is only a very small fraction of their kinase activity. The molecular mechanism by which AceK is able to achieve such high ATPase activity and its physiological significance are poorly understood. Previous studies discovered that AceK's phosphatase activity is ATP/ADP-dependent.^3,4 A computational study by our group revealed that AceK uses ADP as the proton acceptor in the phosphatase reaction.⁶ These studies have led to a suggestion that hydrolysis of ATP might be a preliminary step for AceK to produce ADP to perform its phosphatase function. Furthermore, although great efforts have been made in the determination of the crystal structure and in studying the catalytic mechanisms of AceK, how AceK regulates its multiple functions remains elusive.^2,5–7

The multi-functional AceK is modulated by various regulators. For instance, AMP binding causes allosteric change in loop-β3αC (residues: 340–350), which moves downwards to the C-lobe and forms salt-bridges between Lys346 and Glu478. The movement of loop-β3αC leads to the formation of a close conformation which completely shields the active site. As a result, AceK's ATPase activity is completely abolished in the fully closed active site. The substrate ICDH is also capable of regulating AceK. In the AceK–ICDH complex structure (PDB code: 3LCB), the salt bridges in loop-β3αC are broken and expose AceK's active site to the solvent, so that Ser113 of ICDH is able to enter the active site to react. Intriguingly, the experimental evidence shows that the ATPase activity of AceK is partially inhibited when it binds to the substrate ICDH, while the kinase and phosphatase activities remain.⁸ A notable structural change caused by ICDH is found in the active site of AceK, when the residue Asp477 is squeezed to turn into the active pocket after binding with ICDH. As a negative charged residue, Asp477 may affect the ATP hydrolysis reaction, but the detailed mechanism remains unclear.

The crystal structure revealed that AceK shares some similarities with canonical ATPases in its catalytic core and ATP binding site. Similar to typical ATPases, AceK possesses an ATP binding motif and has an Mg²⁺ ion in its active site to stabilize the ATP molecule. A negative charged residue (Asp457) is positioned close to the phosphate end of the ATP binding site, which may act as the catalytic base or proton acceptor in hydrolysis reactions as in other ATPases. AceK also displays common structural characteristics of typical eukaryotic protein kinases (ePKs). For example, the kinase domain (KD) of AceK consists of two lobes (N-lobe and C-lobe). The “catalytic triad”, which is an essential structural feature of ePKs, is present in AceK at the same spatial position, consisting of residues Asp457, Asn462 and Asp475.⁵ Although the kinase mechanism of AceK has been investigated,⁷ how AceK hydrolyzes ATP with very high efficiency remains an open question. Detailed mechanistic studies of AceK as a highly active ATPase may also help us to understand AceK's other functions and provide new insights into its physiological role.

Numerous theoretical and experimental studies have been performed to explore the hydrolysis mechanism of ATP in solution^9,10 and protein environments.^11–17 Two possible mechanisms, associative or dissociative,^12,14,15 are used by different ATPases. The essential difference between the two mechanisms is whether the cleavage of the P_γ–O_βγ bond of ATP takes place prior to attack by nucleophilic water or whether a stable P_γO₃⁻ group is formed in the course of reaction. A concerted mechanism for the ATP hydrolysis process has also been suggested, although explanation usually leans toward the dissociative or associative methods.¹⁷ Previous computational work found that the catalytic mechanism for AceK as a kinase follows a dissociative pathway. Whether it uses the same mechanism to fulfill its ATPase function remains to be investigated.

In this work, we investigated the AceK-catalyzed ATP hydrolysis mechanism via quantum mechanical calculations by using a density functional theory (DFT) method B3LYP.¹⁸ The dissociative mechanism was found to be the ATPase catalytic mechanism used by AceK with high activity, which produces ADP for the subsequent phosphatase function. We discovered that the flipping over of Asp477 changes the structure of the hydrogen bond network inside the active site and further suppresses the activity of ATPase. Our study of the ATPase mechanism provides a new understanding of the AceK function.

Models and methods

Models

The crystal structures of AceK (PDB code: 3LC6) and the AceK–ICDH complex (PDB code: 3LCB) have been resolved.^2,19,20 Both structures are in the form of a dimer. In the substrate-free AceK dimer structure, the second molecule (chain B) is bound with AMP, whereas AMP is not present in chain A. Instead of ATP being used in the crystallization, an ADP molecule is found in the active site of chain A, unlike chain B in which no nucleotide is bound. Since ADP is produced from ATPase hydrolysis, chain A is considered to be in a conformation with ATPase activity. In the AceK–ICDH complex the two chains are identical, and in each chain ICDH forms a tight homodimer with AceK and with an AMP molecule binding in the allosteric site, which inhibits the ATPase activity of AceK.⁵ Since chain A in the AceK structure is recognized as the active conformation with ATPase activity, we used it as the starting point for building a model to study the ATPase mechanism of AceK in this work. In our model, the ADP molecule in the active site of AceK was replaced by an ATP molecule. According to the experimental evidence that ICDH partially inhibits the ATPase activity of AceK⁸ and in order to investigate the structural basis of the substrate-mediated inhibition, we also investigate the AceK–ICDH complex structure in this work. We selected chain A from the dimer for model building. Before extracting the catalytic QM models, both crystal structures were subjected to molecular dynamics simulations. After adding hydrogen and missing heavy atoms, the complex was solvated in a rectangular box with a distance between the protein and the box boundary of 8.0 Å. We performed a total of 5000 steps of energy minimization to eliminate steric conflicts. The optimization was performed with 2500 steps of the steepest descent method combined with 2500 conjugate gradient steps. Following the minimization, both systems were heated from 0 K to 300 K by using a 100.0 kcal mol⁻¹ Ų constraint on the active site for 50 ps. Next, the heated structures were submitted to equilibration with the same constraint at 300 K for 10 ns. All of the MD simulations were carried out with the program Amber 9 using AMBER force field parameters^21,22 for the protein and TIP3P for water molecules. A 1 fs step size was used for the MD simulation and a total of 10⁷ conformers were sampled during a 10 ns simulation. After confirming that equilibrium had been reached, the final snapshots were used as the starting points for selecting the QM models.

QM setup

The cluster model was extracted from the last snapshot of the 10 ns MD simulation.^6,7 Since previous work found that the coordination sphere of Mg²⁺ has a profound impact on the catalytic reaction,^6,7 it is important to look into the coordination environment of Mg²⁺. After equilibration, Mg²⁺ is penta-coordinated by O_γ1, O_β1, O_α1 of ATP, Asn462 and Asp475. Thus we built the first model by adding one ligand water to complete the octahedral coordination of Mg²⁺ (Model_1-wat in Fig. 1). We noticed that in the crystal structure, the residue Asp475 is located at a distance of 3.46 Å from Mg²⁺. There is room between Asp475 and the Mg²⁺ ion to accommodate another ligand water. To make our investigation more comprehensive, we thus built a second model by inserting an extra ligand water between Asp475 and Mg²⁺ (Model_2-wat in Fig. S5†). In addition to Mg²⁺ and its ligands, the other residues included in our models are the methyl triphosphate tail of ATP and its first shell residues including Gly320, Met321, Val322, Met323, Lys336 and Lys461, potential proton acceptor Asp457 and a molecule of water as lytic water. As shown in the crystal structure, Asp475 forms a hydrogen bond with ATP; therefore Asp475 was protonated in our models. The same cluster models were also built from the equilibrium structure of the AceK–ICDH complex, which represents the partially inhibited ATPase conformation. Asp477 was included in these models given that ICDH binding squeezes Asp477 into AceK's active site in the crystal structure (Fig. 2). The carboxylate group of Asp477 was also protonated to form a hydrogen bond with ATP.

Fig. 1 Geometrical structures (left column) and schemes (right column) of Model_1-wat for AceK (a) and AceK–ICDH complex (b). Main-chain carbons are identified with asterisks.

Fig. 2 Superimposition of the active site and surrounding area of AceK (gray) and AceK–ICDH complex (orange). The residues corresponding to Model_1-wat are shown as yellow sticks and those from the AceK–ICDH complex structure are shown as green sticks. Rotation of Asp477 is obvious as shown from the structures.

All the QM models have a net charge of −1. The total numbers of atoms for Model_1-wat of AceK and AceK–ICDH complex models (one water ligand model) are 123 and 134, respectively, while those of Model_2-wat are 126 and 137.

Computational details

The DFT functional B3LYP^18,23,24 with 6-31G(d) basis sets was used to optimize the geometries of reactant, transition state and product. To mimic the polarization effects of the protein environment, a CPCM continuum model was employed with a dielectric constant of 4.0.^6,7 The zero-point energy (ZPE) effect was also estimated at the same level as in the geometry optimizations. More accurate energies were further evaluated by the B3LYP functional with larger basis sets (6-311+G(d) for Mg²⁺ and cc-pVTZ for the rest).^25,26 All the QM calculations were completed with the Gaussian09 program.²⁷ A dispersion correction was calculated using Grimme's D2 formula (implemented in XYZ-Viewer).

Results and discussions

ATPase mechanism of AceK

Based on the Model_1-wat and Model_2-wat of AceK, we investigated the ATP hydrolysis mechanism of AceK. Both pathways with associative and dissociative structural characteristics were explored. The calculated energetic data are listed in Table 1. Energetically, the dissociative mechanism of Model_1-wat is the favorable one, with an energy barrier of 17.85 kcal mol⁻¹, while the associative pathway of this model experiences a TS with an energy barrier of 25.97 kcal mol⁻¹.

Table 1 The relative energies in ATP hydrolysis process of the models built from AceK structure (kcal mol⁻¹)

	Model1-wat		Model2-wat
	Associative	Dissociative	Associative	Dissociative
Re	0	8.50	0	6.13
TS	25.97	17.85	34.69	21.28
Pr	−15.62	−9.44	−9.88	5.3

---

Associative pathway

The associative pathway of the ATP hydrolysis reaction by AceK was examined with Model_1-wat. The calculated activation energy is 25.97 kcal mol⁻¹. The optimized structures of reactant, transition state and product in the associative pathway are shown in Fig. 3. In the reactant, as in the crystal structure, the Mg²⁺ ion is coordinated by Asn462, Asp475, O_γ1, O_β1, O_α1 from ATP and a ligand water. The lytic water is located between γ-phosphoryl and Lys461 and bridges them with two hydrogen bonds (O_wat–O_γ2ATP = 2.74 Å, N_Lys461–O_wat = 2.78 Å). Lys461 forms two further hydrogen bonds with Asp457 and O_γ1 of ATP (N_Lys461–O_Asp457 = 2.69 Å, N_Lys461–O_γ1ATP = 2.94 Å). Asp475 interacts with the O_γ3 of ATP via a hydrogen bond (O_Asp475–O_γ3 = 2.54 Å), which stabilizes the ATP in the active pocket. Moreover, the ammonium group of Lys336 donates its proton to O_α2 of ATP via the hydrogen bond between them (O_α2–N_Lys336 = 2.63 Å). In the active site, the oxygen of lytic water is 3.22 Å away from the γ-phosphoryl group of ATP. These structural features of the reactant clearly show a tendency to follow the associative pathway.

Fig. 3 The optimized structures of Re, TS and Pr of Model_1-wat for the associative and dissociative pathways from QM calculations. Left column: Re, TS and Pr located along with the γ-phosphoryl group of ATP as the final proton acceptor (associative). Right column: Re, TS and Pr located along the Asp457 as proton acceptor (dissociative). Main-chain carbons are identified with asterisks.

As the reaction proceeds to the associative transition state, the lytic water approaches ATP with the O_wat–P_γATP distance decreasing to 2.22 Å, while the O_βγATP–P_γATP bond elongates from 1.71 Å to 1.78 Å. Meanwhile, a proton from the lytic water (H_wat) transfers to O_γ2 of ATP (H_wat–O_γ2 = 0.99 Å) (Fig. 3). The resulting hydroxide forms a stronger hydrogen bond with Lys461 compared to that in the reactant (O_wat–N_Lys461 = 2.53 Å), which causes a slight rotation of the ammonium group of Lys461 and weakens the hydrogen bond between Lys461 and O_γ1 of ATP. In the associative transition state, the O_βγATP–P_γATP bond is not fully broken, and a tetrahedral –OP_γO₃²⁻ is maintained in the reactant.

In the product, the O_βγATP–P_γATP bond breaks fully (4.98 Å) and an O_wat–P_γATP (1.65 Å) bond is formed. The O_γ2 accepts a proton from lytic water and forms a hydrogen bond with O_β1 (O_γ2–O_β1 = 2.62 Å). Moreover, the ammonium group of Lys336 retrieves its proton back from O_α2 of ATP and now forms two hydrogen bonds with O_α2 and O_βγ of ATP (O_α2–N_Lys336 = 2.65 Å, N_Lys336–O_βγ = 2.64 Å), respectively, which further stabilizes the product. The associative pathway is exergonic by 15.62 kcal mol⁻¹.

Dissociative pathway

The dissociative pathway was also investigated with Model_1-wat in this work. As illustrated in Fig. 3, a very important difference from the reactant of the associative pathway is that the γ-phosphoryl of ATP rotates away from the crystal structure and switches from O_β1–Mg²⁺ ligation to O_βγ–Mg²⁺ ligation. The dangling O_β1 is stabilized by a strong hydrogen bond from the ligand water (O_β1–O_{ligand–water} = 2.60 Å). This Lewis acid–base interaction obviously activates the O_βγ–P_γATP bond in the subsequent hydrolysis reaction. The calculated energy of the dissociative reactant is 8.50 kcal mol⁻¹ higher than that of the associative reactant. For the reactant state in the dissociative pathway, the lytic water forms several hydrogen bonds with the surrounding residues. In particular, it has a hydrogen bond (O_wat–O_Asp457 = 2.78 Å) with the carboxylate group of Asp457, which triggers the departure of a proton from the water. The hydrogen bond between Asp475 and O_γ3 of ATP (O_Asp475–O_γ3ATP = 2.55 Å) is similar to that in the associative pathway. Compared to that in the crystal structure, the amino group of the Lys461 side chain rotates slightly to bridge the ligand water of Mg²⁺ (N_Lys461–O_{ligand–water} = 2.98 Å) and O_γ1 of ATP (N_Lys461–O_γ1 = 2.66 Å). The oxygen of lytic water is 3.85 Å away from the γ-phosphoryl group of ATP. In short, the conformation of the reactant is primed for the lytic water to make a nucleophilic attack on the terminal phosphoryl of ATP and for the subsequent proton transfer.

The dissociative transition state structure is characterized by a planar P_γO₃⁻ moiety which is separated from ADP with the O_βγATP–P_γATP bond stretching from 1.81 Å to 2.23 Å. The O_wat–P_γATP distance decreases in the reactant from 3.85 Å to 2.47 Å. The H_wat–O_Asp457 distance also decreases to 1.76 Å compared with the initial value of 1.83 Å in the reactant, which may facilitate the proton transfer from the lytic water to Asp457 after formation of the O_wat–P_γATP bond. The transition state structure clearly indicates that the ATPase reaction follows a specific dissociative mechanism. The dissociative characters of the transition state can also be exemplified by Pauling's formula: D(n) = D(1) − 0.6 log n.²⁸ As described by Mildvan, D(1) stands for the standard P–O bond length which is assigned as 1.73 Å. D(n) is defined as the average value of the lengths of the cleaving and forming P–O bonds in the transition state, which is 2.35 Å in our calculated TS structure. Thus n in the equation, the fractional bond number, is calculated to be 0.09 which gives a TS with a high degree of dissociative characteristics at 91%. The local energy barrier for the dissociative pathway is calculated to be 9.35 kcal mol⁻¹. Counting from the lowest energy reactant (the associative one), the final energy barrier is 17.85 kcal mol⁻¹, which is 8.12 kcal mol⁻¹ more favorable than for the associative mechanism. This result also agrees with the experimental finding that AceK exhibits higher ATPase activity than its kinase and phosphatase functions.^4,6,7

In the optimized product, an O_wat–P_γATP bond is formed with a bond length of 1.70 Å. Meanwhile, the carboxylate group of Asp457 has accepted a proton from the lytic water (O_Asp457–H_wat = 0.99 Å). Compared to the reactant, some obvious changes could be found in the geometry of the product, especially for the ADP group. A proton from the ammonium group of Lys461 has transferred to O_β1 of ADP through the ligand water of Mg²⁺. An interesting observation about the product is that the β-phosphoryl group of ADP turns over, which results in the breaking of the hydrogen bond between O_β2 of ADP and Val322, Met323. A new hydrogen bond (O_β2–N_Lys336 = 2.71 Å) is formed between Lys336 and O_β2 of ADP. Compared to the reactant, the coordination shell of Mg²⁺ ion and the residues in the active site in the product are almost unchanged. As it is an exothermic reaction, the product is 17.94 kcal mol⁻¹ lower than the dissociative reactant.

Several clues could be gathered from the transition state structures to rationalize this energy barrier difference between the associative and dissociative pathways. First, in the dissociative transition state, the negatively charged residue Asp457 prepares the lytic water to attack the P_γO₃⁻ by forming a hydrogen bond with it (O_wat–O_Asp457 = 2.75 Å). In contrast, in the associative transition state, the positively charged residue Lys461 (O_wat–N_Lys461 = 2.53 Å) forms a hydrogen bond with the hydroxyl group after the lytic water transfers a proton to O_γ2 of ATP; the positive charge of Lys461 is obviously not feasible for a subsequent nucleophilic attack by the hydroxyl group on P_γO₃⁻. Furthermore, as described above, O_βγ–Mg²⁺ ligation in the dissociative transition state also activates the O_βγ–P_γATP bond by Lewis acid–base interaction. Moreover, as described in the previous study, the tetrahedral γ-phosphate group of ATP (–OP_γO₃²⁻) in the associative transition state is less readily attacked by a negatively charged OH_wat than the planar metaphosphate (P_γO₃⁻) in the dissociative transition state, which also leads to a higher energy barrier.²⁹ Additionally, the dissociative mechanism separates the breaking of the P_γ–O_βγ bond and cleavage of the lytic water into two steps, which also contribute to lowering the energy barrier of the reaction.

Alternative coordination sphere of Mg²⁺

As mentioned in the Models and methods section, a water molecule could replace Asp475 to ligate with Mg²⁺ in the active site, which is represented by our Model_2-wat. To gain a more complete understanding, we also investigated the hydrolysis mechanism of AceK based on this model. The geometries of reactants, transition states and products are depicted in Fig. S1† and key structural parameters are listed in Table S1.† For Model_2-wat, the dissociative pathway is still the energetically favorable one, accounting for 93% of the dissociative characteristics as calculated by Pauling's formula. This finding is also consistent with Model_1-wat. Compared to that of Model_1-wat, the energy barrier for the dissociative Model_2-wat is raised by 3.43 kcal mol⁻¹, with a value of 21.28 kcal mol⁻¹. Observing the geometry of the reactant of Model_2-wat in Fig. S1,† we can see that the second ligand water replacing Asp475 forms a longer hydrogen bond with γ-phosphoryl of ATP, which enables the phosphoryl to rotate slightly to form another hydrogen bond with the lytic water. This hydrogen bond is not present in the reactant of Model_1-wat and must be broken before the lytic water can attack the phosphoryl. Thus, the local barrier becomes higher. Furthermore, estimation of the binding energy of the second water indicates that its ligation is energetically unfavorable.³⁰ Taken together, the one-ligand–water model (Model_1-wat) represents the more probable reactant state with more favorable activation energy.

Previous studies on the ATPase reaction found that the hydrolysis process usually involves more than one water molecule in the proton transfer progress.^12–17 Besides, the unique single Mg²⁺ ion character for AceK has been verified to facilitate kinase activity whereas it inhibits phosphatase activity compared to the double Mg²⁺ ion in the catalytic core.^6,7 Given these results, the water wire and Mg²⁺ ion effects on the ATPase reaction were also taken into consideration in this paper. In addition, in contrast to the kinase activity, the high ATPase activity for AceK was also analyzed. Detailed information about this is presented in ESI.†

Catalytic roles of the critical active-site residues

In the active site of AceK, several residues, such as Asp457, Asp475, Lys336 and Lys461 are identified as playing very important roles in AceK's kinase and phosphatase activities. Our calculations suggest that these residues are also critical for the ATPase function. As in the kinase reaction pathway, Asp457 acts as the proton acceptor in the ATP hydrolysis process. The negatively charged carboxylate of Asp457 facilitates proton transfer from either Ser113 or lytic water, thus completing the kinase and ATPase reaction.⁷ The positively charged Lys336 forms a hydrogen bond with α-phosphoryl of ATP and possibly a second hydrogen bond with β-phosphoryl, which stabilizes ATP or ADP at different stages of the hydrolysis reaction. Furthermore, Lys461 facilitates the departure of γ-phosphoryl via a hydrogen bond between them. Experiment has found that a D475A mutation completely abolishes AceK's ATPase activity,² indicating that Asp475 plays a critical role in the ATP hydrolysis reaction. Based on the structure and our QM computations, although Asp475 is not directly involved in the bond breaking, it stabilizes Mg²⁺ as a potent ligand in the active site, which further binds with ATP as the substrate of the hydrolysis reaction. Mutation of Asp475 to Ala would release Mg²⁺ and cause the collapse of the active site. Furthermore, based on our calculations, Asp475 is also involved in the hydrogen bonding interaction inside the active site. Replacement of Asp475 ligation by water ligation induces a delicate change in the hydrogen bond network and slows down the ATP hydrolysis reaction by raising the active barrier by 3.43 kcal mol⁻¹.

The partially inhibited ATPase activity with substrate ICDH

With regard to AceK's ATPase activity, there is evidence that substrate ICDH binding is capable of partially inhibiting the activity.⁸ Comparing the structure of AceK with that in the AceK–ICDH complex, the most noticeable difference is that on binding with ICDH the side chain of AceK Asp477 moves into the active site. A previous computational study has identified that with Asp477 located inside the active site, the kinase activity of AceK would be suppressed.⁷ This observation strongly suggests that Asp477 might also restrain AceK's ATPase activity when it turns into the active site. To confirm that the Asp477 rotation is the reason that ICDH-binding suppresses AceK's ATPase activity, we performed a DFT calculation on the cluster model built from the AceK–ICDH complex structure. Starting from the model, we located all the stationary points along the reaction path. The geometries of reactants, transition states and products are depicted in Fig. S4† and key structural parameters are listed in Table S4.† The dissociative mechanism is identified as the dominant pathway with an energy barrier of 21.68 kcal mol⁻¹ (28.82 kcal mol⁻¹ for the associative pathway). This energy barrier is 3.83 kcal mol⁻¹ higher than that calculated for the model built from AceK (Model_1-wat), which is in good agreement with the experiment showing that ICDH-binding restrains the ATPase activity of AceK but does not fully abolish it.

The key hydrogen bonding interactions for the reactant and transition states of the AceK–ICDH model are shown in Fig. 4. We observe that the P_γO₃⁻ group of ATP forms four hydrogen bonds with Lys461, Met321, Asp475 and Asp477. When the reaction proceeds to the transition state, with the rotation of the P_γO₃⁻ group, the hydrogen bond between P_γO₃⁻ and Met321 breaks, which costs extra energy. However, in the model built from AceK in which Asp477 is absent, only three hydrogen bonds are formed between the P_γO₃⁻ group and surrounding residues and no hydrogen bonds will be lost in the transition state. In other words, Asp477 in the active site complements the restraining of the P_γO₃⁻ group, which further inhibits AceK's ATPase activity.

Fig. 4 The partial structures of the one-ligand water models in AceK and AceK–ICDH complex. (a) The partial structures of reactant (Re) and transition states (TS) in AceK model. (b) The partial structures of reactant and transition states in AceK–ICDH complex model.

Based on our structural analyses and QM calculations, Asp477 is verified to play an inhibiting role in the ATPase activity. Binding with substrate ICDH causes Asp477 to flip over and further inhibits AceK ATPase activity by influencing the hydrogen bond change in the ATPase reaction, thus favoring kinase or phosphatase activities accordingly.

Regulation mechanism of AceK

As a multi-function enzyme, AceK uniquely possesses kinase, phosphatase and ATPase activities in the same active site, which necessitates a delicate balance among these functions. As described above, the essential regulator AMP allosterically regulates the movement of loop-β3αC. In the presence of AMP, loop-β3αC takes a closed conformation which shields the ATP molecule from the active site, thus fully abolishing its ATPase activity. Through complexing with ICDH, which partially breaks the salt bridges in the loop-β3αC, its kinase and ATPase activities are still suppressed, which ensures that AceK can play its role as a phosphatase. When AMP dissociates from AceK's allosteric site, due to the opening of loop-β3αC, the active site is accessible to the ATP molecule. Depending on whether it is complexing with ICDH or not, AceK switches to exhibit kinase or ATPase activity. As a kinase, with Asp477 now located in the active site, AceK's ATPase function is partially inhibited. When Asp477 leaves the active site upon disassociation of ICDH, AceK converts to an ATPase only to produce ADP, which is utilized in the phosphatase reaction. High ATPase activity is necessary because AceK's ATPase function is an integral part of its phosphatase activity.

Conclusions

We have investigated the catalytic mechanism of the AceK ATPase function, which features unusually high ATPase activity, as well as its multi-function switch using a QM method. Our study reveals that the highly dissociative characteristics and the specific spatial constraints in AceK enable the enzyme's dissociative ATPase reaction mechanism and a single water pathway. The calculated low energy barrier (17.85 kcal mol⁻¹) indicates that as a multi-functional enzyme, the strong ATPase activity is essential to provide ADP for its phosphatase reaction. Analysis of the structural change along our calculated reaction pathway suggests that the carboxylate of Asp457 acts as the final base to accept a proton from the lytic water during the ATP hydrolysis process through the dissociative mechanism.

The unique structural characteristics and the allosteric effect of AMP as well as the combination with substrate ICDH enable AceK's delicate functional switch. Besides, DFT calculations for the ATPase reaction in the AceK–ICDH complex model also explain the fact that ATPase activity is partially inhibited in the presence of substrate ICDH through the conformational change in Asp477. Furthermore, our investigations of the ATPase activity of AceK also provide a new understanding of the regulating biological function of AceK.

Acknowledgements

We thank S. Marothy (Stockholm University) for providing XYZ-Viewer to create all the figures of the molecule models. This work was supported by grants from the National Natural Science Foundation of China (No. 21131003, 21573020, 21503018 and 21571019) and Natural Science and Engineering Research Council of Canada.

References

1. D. C. LaPorte and D. Koshland Jr, Nature, 1982, 305, 286–290.
2. J. Zheng and Z. Jia, Nature, 2010, 465, 961–965.
3. C. S. Stueland, T. P. Ikeda and D. C. LaPorte, J. Biol. Chem., 1989, 264, 13775–13779.
4. S. P. Miller, E. J. Karschnia, T. P. Ikeda and D. C. LaPorte, J. Biol. Chem., 1996, 271, 19124–19128.
5. J. Zheng, S. P. Yates and Z. Jia, Philos. Trans. R. Soc. London, Ser. B, 2012, 367, 2656–2668.
6. S. Wang, Q. Shen, G. Chen, J. Zheng, H. Tan and Z. Jia, Chem. Commun., 2014, 50, 14117–14120.
7. Q. Li, J. Zheng, H. Tan, X. Li, G. Chen and Z. Jia, PLoS One, 2013, 8, e72048.
8. C. S. Stueland, K. R. Eck, K. T. Stieglbauer and D. LaPorte, J. Biol. Chem., 1987, 262, 16095–16099.
9. C. B. Harrison and K. Schulten, J. Chem. Theory Comput., 2012, 8, 2328–2335.
10. N. V. Plotnikov, B. R. Prasad, S. Chakrabarty, Z. T. Chu and A. Warshel, J. Phys. Chem. B, 2013, 117, 12807–12819.
11. B. L. Grigorenko, A. V. Rogov, I. A. Topol, S. K. Burt, H. M. Martinez and A. V. Nemukhin, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 7057–7061.
12. B. L. Grigorenko, I. A. Kaliman and A. V. Nemukhin, J. Mol. Graphics Modell., 2011, 31, 1–4.
13. S. Hayashi, H. Ueno, A. R. Shaikh, M. Umemura, M. Kamiya, Y. Ito, M. Ikeguchi, Y. Komoriya, R. Iino and H. Noji, J. Am. Chem. Soc., 2012, 134, 8447–8454.
14. H. Freedman, T. Laino and A. Curioni, J. Chem. Theory Comput., 2012, 8, 3373–3383.
15. J. Akola and R. Jones, J. Phys. Chem. B, 2006, 110, 8121–8129.
16. M. McCullagh, M. G. Saunders and G. A. Voth, J. Am. Chem. Soc., 2014, 136, 13053–13058.
17. M. J. McGrath, I.-F. W. Kuo, S. Hayashi and S. Takada, J. Am. Chem. Soc., 2013, 135, 8908–8919.
18. A. D. Becke, J. Chem. Phys, 1993, 98, 1372–1377.
19. J. Zheng, A. X. Ji and Z. Jia, Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun., 2009, 65, 1153–1156.
20. J. Zheng, D. C. Lee and Z. Jia, Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun., 2009, 65, 536–539.
21. D. A. Case, T. Darden, T. E. Cheatham III, C. Simmerling, J. Wang, R. E. Duke, R. Luo, K. M. Merz, D. A. Pearlman and M. Crowley, AMBER9, University of California, San Francisco, 2006, p. 45.
22. Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P. Cieplak, R. Luo and T. Lee, J. Comput. Chem., 2003, 24, 1999–2012.
23. C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785.
24. A. D. Becke, J. Chem. Phys, 1993, 98, 5648–5652.
25. R. Krishnan, J. S. Binkley, R. Seeger and J. A. Pople, J. Chem. Phys, 1980, 72, 650–654.
26. R. A. Kendall, T. H. Dunning Jr and R. J. Harrison, J. Chem. Phys, 1992, 96, 6796–6806.
27. M. Frisch, G. Trucks, H. B. Schlegel, G. Scuseria, M. Robb, J. Cheeseman, G. Scalmani, V. Barone, B. Mennucci and G. Petersson, Gaussian 09, Revision D.01, Inc., Wallingford, CT, 2009, p. 200.
28. A. S. Mildvan, Proteins: Struct., Funct., Bioinf., 1997, 29, 401–416.
29. F. A. Kiani and S. Fischer, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, E2947–E2956.
30. M. R. Blomberg, T. Borowski, F. Himo, R.-Z. Liao and P. E. Siegbahn, Chem. Rev., 2014, 114, 3601–3658.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11873a

---