A close view of the configuration of glucose phosphorylation in the environment of GK active site. The structural model was constructed based on the snapshot at 2500 ps isolated from the 10-ns MD trajectory. W1 and W2 represent water molecules. ATP, Mg 2 + , glucose and important residues in GK are displayed in stick and labeled. Black dashed lines represent hydrogen bonds or salt bridges. Atomic numbering of ATP and glucose is displayed in the up-left panel. doi:10.1371/journal.pone.0006304.g002 

Contexts in source publication

Context 1

... X-ray crystal structure of human hexokinase type I [28,29]. Although human hexokinase type I is homologous to GK with sequence identity of , 55%, its three-dimensional (3D) structure is different, especially around the ATP binding pocket, as indicated by the structural superposition of the crystal structures of these two enzyme (Figure S1). The reliability of these GK catalytic models is therefore questionable. On the basis of the crystal structures of GK, now available, we built up a structural model for the GMAG complex (see Materials and Method section for detail). This model is more accurate than the previous models, and has also been validated by further MD and QM/MM calculations, as well as mutagenesis experiments and enzymatic assays (see discussion below). Structurally, GK is composed of a large domain and a small domain; the channel-shaped active site for phosphorylation is located in the deep cleft between these two domains, as shown in Figures 1A and 1B. The Mg 2 + ion and ATP (Mg 2 + -ATP) are predicted to bind to the left portion of the active site cleft, interacting with both domains [22–24]. The Mg 2 + -ATP binding site is formed by residues 78–83, 169, 225–228, 295–298, 331– 336, 410–413, and 415–417 (Figure 1C). Meanwhile, the glucose resides in a small pocket composed of residues 80–81, 151–154, 168–169, 204–207, 225–232, 254–259, 287 and 290 (Figure 1D). Our 3D model of GMAG complex reveals that GK provides a favorable microenvironment for the phosphorylation of glucose. Several important hydrogen bonds between ligands (ATP, Mg 2 + and glucose) and GK are observed in our model (Table S1), which are in agreement with a large number of mutagenesis analysis data [20,30–34]. The c -phosphate of ATP is close to the –O 6 H group of glucose with electrostatic interaction, indicating that the c phosphate is about to be transferred to glucose. To study the stability of the reaction environment, a 10-ns molecular dynamics (MD) simulation was performed on the GMAG complex. The time evolutions of the weighted Root-Mean Square Deviations (wRMSD) for the atoms of GK, ATP and glucose from their initial positions ( t = 0 ps) were monitored. The result indicates that all the reaction components are relatively stable except for the rotation of the –O 6 H group which causes an increase in wRMSD of glucose by 0.3 A ̊ at 1 ns (Figure S2). The H-bond network amongst GK, ATP and glucose also reflects the stability of the phosphorylation environment. These H-bonds were mostly maintained during the 10-ns MD simulation as indicated by their occupancies (Table S1). Under the restriction of GK, ATP, Mg 2 + and glucose form a highly advantageous configuration for the biochemical reaction (Figure 2). Mg 2 + ion in the binding site octahedrally coordinates with the oxygen atoms of two water molecules, O b 1 , O c 1 and O c 3 oxygen atoms of ATP and the O d 2 atom of Asp205. Nine H-bonds are formed between ATP and residues of GK, i.e. the N 1 atom of adenine with the –O c H group of Ser336, the O a 2 atom of ATP with the –NH groups of Thr82 and Asn83, the O a 1 atom of ATP with the –O c H group of Ser411, the O a 3 and O b 3 atoms of ATP with the –O c H Thr228, the O c 2 atom of ATP with the –NH group of Gly229, and the O c 1 and O c 2 atoms of ATP with the f + –N H 3 group of Lys169 (Figure 2 and Table S1). These H-bonds lead the ATP to adopt a conformation appropriate to coordinate with the Mg 2 + ion and to interact with glucose. On the other side, glucose forms seven H-bonds with the GK residues, viz. the –O 4 H group of glucose with the N d 1 atom of Asn204 and the O d 1 atom of Asp205, the –O 3 H group of glucose with the O e 1 and O e 2 atoms of Glu256 and the –NH group of Phe152, the –O 1 H group of glucose with the O e 1 atom of Glu290, the O 5 atom of glucose with f + the –N H 3 group of Lys169 (Figure 2 and Table S1). Most importantly, the –O 6 H group of glucose directly hydrogen bonds to the O c 2 atom of ATP, producing the requisite complex for glucose phosphorylation. Residues detected in the naturally occurring mutations (e.g. K169N, T228M, E256K, S336L, and S411F) from MODY families [20,30–32] are involved in the H- bond network for the binding of GK with glucose and ATP. Extensive work confirmed that S336L mutation decreased both the binding affinity of GK to ATP and the catalytic activity of the enzyme [33]. Several other site-directed mutants, including N204A, E256A and E290A, have been shown to alter GK’s enzymatic kinetics by decreasing the binding affinity of glucose and lowering V max in the catalytic process [31,34]. These mutagenesis and enzymatic results demonstrate that our 3D model of GMAG complex is reasonable. Among the conserved residues hydrogen bonding to ATP and glucose, Lys169 is of special interest because its cationic end, f + –N H 3 , bridges ATP and glucose together through H-bonds, placing the c -phosphorus (P c ) atom only 3.3 A ̊ far way from the –O 6 H group of glucose (Figure 2). This implies that Lys169 might play an important role in the phosphorylation of glucose. Additionally, it was reported that Lys169Asn (K169N) is one of the naturally occurring mutations in the GCK gene associated with familial mild fasting hyperglycemia [20]. The K169N mutant was also shown to experience a partial loss of glucose binding [30]. To address the biological function of Lys169, we constructed a 3D structural model for the GK K169A mutant (GK K169A ) in complex with ATP-Mg 2 + and glucose, and two additional MD simulations of the ATP-Mg 2 + -GK K169A and glucose-GK K169A complexes were conducted. Binding free energies of ATP and glucose with wild-type GK and GK K169A (Table S2) were then calculated using the MM-PBSA method encoded in AMBER (Version 8.0). Both ATP and glucose are able to bind tightly to the wild-type GK with calculated binding free energies of 2 18.67 6 4.59 and 2 44.06 6 3.94 kcal/mol, respectively. While neither ATP nor glucose can bind with GK K169A . The MD simulations and binding free energy calculations indicate that Lys169 may dominate the binding of both ATP and glucose, and hence the K169A mutant possibly loses the catalytic function for the phosphorylation of glucose. To verify this conclusion, a bioassay on the K169A mutant was performed and result is shown in Table 1. The kinetic assay indicates that K169A is indeed an inactivating mutation; its enzymatic activity is completely abolished. Far-UV CD spectra and fluorescence emission spectra results demonstrated that the K169A mutant had well-defined secondary structures; thereby we can exclude the possibility that this mutation may have caused GK to undergo misfolding (Figures S3, S4, S5). Thus, the MD prediction, regarding the importance of Lys169 to the binding of ATP and glucose with GK and to the catalytic activity of GK as well was validated. To further figure out the role of Lys169 in enzymatic catalysis, the reaction mechanism of glucose phosphorylation inside the active site of GK was investigated by using the QM/MM approach. The system for QM/MM simulation was constructed based on the snapshot at 2500 ps of the MD trajectory. The QM region was composed of Mg 2 + , water molecules coordinating with Mg 2 + , important groups of ATP, glucose, Lys169 and Asp205; the remainder of GK was included in the MM region (Figure S6). The partitioning scheme for QM and MM regions is described in the Materials and Methods section and Figure S6. We designate this structure as reagent system. For the reagent system, the initial coordination configuration of Mg 2 + with waters, ATP and glucose and the hydrogen bonding pattern amongst ATP, glucose, Lys169 and Asp205 have been described above (Figure 2). ONIOM, a QM/MM method encoded in Gaussion03 [35], was used for all the QM/MM calculations (Figure 3 and 4). The QM/MM optimized geometry of the reagent system showed that Mg 2 + octahedrally coordinates with the O b 1 and O c 1 atoms of ATP, the O atom of Asp205 and three oxygen atoms from three water molecules (Figure 3). The major difference of the optimized structure for the reaction center from the initial structure obtained by MD simulation is that one water molecule substituted the O c 3 atom to coordinate with the Mg 2 + ion. QM optimization resulted in important structural changes viz. one proton of the –N f H + 3 group of Lys169 transferred to the O c 3 f atom, and the resulting –N H 2 group formed a H-bond with the new water molecule coordinating with the Mg 2 + ion; on the other hand, the –O 6 H group of glucose adjusted its direction to point toward the O d 1 atom of Asp205, forming two strong H-bonds with the O d 1 and O d 2 atoms of Asp205. This structural reorganization made the –O 6 H group more propitious to attack the P c atom of ATP, implying that Lys169 might act as an acid catalyst and Asp205 as a base catalyst in the phosphorylation of glucose. Thus, we propose a mechanism for the glucose phosphorylation catalyzed by GK as Scheme 1 (Figure 5). Along this reaction path, the energies of the reagent (R), transition state (TS), and immediate product (P) were determined by two-dimensional QM/ MM potential energy surface by defining the distances of R (O _P ) and R (P _O ) as the reaction coordinates (Figure 3). In the optimized reagent, R (O b 3 _P c ) = 1.69 A ̊ and R (P c _O 6 ) = 3.39 A ̊ ; and in the optimized immediate product, the O b 3 _P c bond is broken and the P c _O 6 is formed, R (P c _O 6 ) = 1.71 A ̊. The calculated potential energy barrier of Scheme 1 is D E ? = 18.3 kcal/mol. The structure of the transition state (TS) in Scheme 1 was determined by adiabatic mapping at the QM/MM level (Figure 3). In the TS, R (O b 3 _P c ) = 2.1 A ̊ and R (P c _O 6 ) = 2.0 A ̊ , and the c c phosphate (–P O 3 ) is turned to be in a plane. This structure clearly illustrates that the P c _O 6 bond is partially formed and the O b 3 _P c bond is partially broken. The overall reaction is calculated to be ...

Context 2

... system showed that Mg 2 + octahedrally coordinates with the O b 1 and O c 1 atoms of ATP, the O atom of Asp205 and three oxygen atoms from three water molecules (Figure 3). The major difference of the optimized structure for the reaction center from the initial structure obtained by MD simulation is that one water molecule substituted the O c 3 atom to coordinate with the Mg 2 + ion. QM optimization resulted in important structural changes viz. one proton of the –N f H + 3 group of Lys169 transferred to the O c 3 f atom, and the resulting –N H 2 group formed a H-bond with the new water molecule coordinating with the Mg 2 + ion; on the other hand, the –O 6 H group of glucose adjusted its direction to point toward the O d 1 atom of Asp205, forming two strong H-bonds with the O d 1 and O d 2 atoms of Asp205. This structural reorganization made the –O 6 H group more propitious to attack the P c atom of ATP, implying that Lys169 might act as an acid catalyst and Asp205 as a base catalyst in the phosphorylation of glucose. Thus, we propose a mechanism for the glucose phosphorylation catalyzed by GK as Scheme 1 (Figure 5). Along this reaction path, the energies of the reagent (R), transition state (TS), and immediate product (P) were determined by two-dimensional QM/ MM potential energy surface by defining the distances of R (O _P ) and R (P _O ) as the reaction coordinates (Figure 3). In the optimized reagent, R (O b 3 _P c ) = 1.69 A ̊ and R (P c _O 6 ) = 3.39 A ̊ ; and in the optimized immediate product, the O b 3 _P c bond is broken and the P c _O 6 is formed, R (P c _O 6 ) = 1.71 A ̊. The calculated potential energy barrier of Scheme 1 is D E ? = 18.3 kcal/mol. The structure of the transition state (TS) in Scheme 1 was determined by adiabatic mapping at the QM/MM level (Figure 3). In the TS, R (O b 3 _P c ) = 2.1 A ̊ and R (P c _O 6 ) = 2.0 A ̊ , and the c c phosphate (–P O 3 ) is turned to be in a plane. This structure clearly illustrates that the P c _O 6 bond is partially formed and the O b 3 _P c bond is partially broken. The overall reaction is calculated to be exothermic by D E = 2 22.1 kcal/mol. To further address the catalytic role of Lys169, we investigated the reaction mechanism of glucose phosphorylation catalyzed by the GK K169A mutant employing the QM/MM method. Com- pared with the optimized structure of GK reagent system, the major difference for the ATP-GK K169A -glucose reaction center is that the –O 6 H group of glucose doesn’t form H-bond with either O d 1 or O d 2 atom of Asp205, but forms a strong H-bond with the O atom of ATP. This optimized structure implies an alternative path for the glucose phosphorylation as Scheme 2 (Figure 6): the proton of the –O 6 H group transfers to the O c 3 atom of ATP while the O 6 atom attacks the P c atom of ATP, and at the same time the O b 3 _P c bond is broken. According to this reaction path, we obtained the structures and energies of reagent (R), transition state (TS), and immediate product (P) by calculating the two- dimensional QM/MM potential energy surface; the distances of R (O b 3 _P c ) and R (P c _O 6 ) were defined as the reaction coordinates. The result is shown in Figure 4. For the TS, R (O b 3 _P c ) = 2.1 A ̊ and R (P c _O 6 ) = 1.9 A ̊ . Remarkably, the calculated potential energy barrier of this reaction path is D E ? = 32.1 kcal/mol, which is , 14 kcal/mol higher than that of the reaction catalyzed by the wild-type GK. This result indicates that the phosphorylation of glucose could not occur without the assistance of Lys169. As mentioned above, GK is a key enzyme that phosphorylates glucose and triggers glucose utilization and metabolism. As an attractive drug target for discovering anti-diabetes drug, the importance of this enzyme has been appreciated. However, the unique catalytic mechanism of GK still remains unclear [17,18]. Several mutants of GK with the MODY phenotype have been identified recently; in particular, it was found that K169N is a naturally occurring mutation (K169N) in MODY [20]. In this study, we have investigated the phosphorylation mechanism of glucose catalyzed by GK, and revealed that Lys169 is a crucial residue for glucose phosphorylation. By using molecular modeling and simulation methods, we constructed a 3D structural model for the complex of GK with ATP, glucose and Mg 2 + (GMAG complex). This structural model is more reliable than previously published models [22–27,36] because it was constructed based on the crystal structures of human GK. In particular, this model provides a more realistic reaction environment for the phosphorylation of glucose. The validity of the structural model of GMAG was confirmed by the agreement between the available experimental mutagenesis and enzymatic data for GK and the important interactions of GK with ATP, glucose and Mg 2 + identified through computation [20,30–34]. Both ATP and glucose interact with the residues lining the binding pocket of GK through H-bonds (Figure 2). Among these residues, Lys169 plays a significant role, because it forms H-bonds with both ATP and glucose. This indicates that Lys169 contributes to the binding of GK with either ATP or glucose or both. Computational simulations demonstrated that both ATP and glucose are able to bind to the wild-type GK, and neither ATP nor glucose can bind to the GK K169A mutant (Table S2). Indeed, the experimental mutagenesis and enzymatic kinetic assay validated the computational prediction, i.e. the mutant of K169A could not bind to ATP and glucose and its enzymatic activity is completely lost (Table 1). The QM/MM calculations on the mechanisms of glucose phosphorylation catalyzed by both GK and GK K169A mutant further revealed the role of Lys169 in catalysis. During the phosphorylation process of glucose, in addition to enhancing the binding of ATP and glucose with GK, Lys169 also acts as a general acid catalyst, providing a proton to protonate the O c 3 atom of ATP; at the same time Asp205 acts as a general base catalyst, extracting the proton from the –O H group of glucose (Scheme 1). On the other hand, ATP and glucose bind to the GK K169A mutant with a different model than with the wild-type GK (Figures 3 and 4), whereby the phosphorylation of glucose adopts a different mechanism (Scheme 2). The QM/MM calculated activation energy of Scheme 2 is , 14 kcal/mol higher than that of Scheme 1. This suggests that the GK K169A mutant has significantly reduced catalytic activity than its wild-type form. The calculated result is in good agreement with our mutagenesis data (Table 1). Moreover, the QM/MM result also indicates that Asp205 of GK loses its function as a general base catalyst in the glucose phosphorylation when Lys169 is absent (Figure 4). The agreement between computation and experimental results validates the essential role of Lys169 in the phosphorylation of glucose. In summary, Lys169 plays at least three functional roles in the metabolism of glucose: (1) it enhances the binding of GK to both ATP and glucose; (2) it bridges ATP and glucose together; and (3) it acts as a general acid catalyst and directly participates in glucose phosphorylation. These results are important for understanding the catalytic mechanism of GK and the cause of the pathogenic mechanism of MODY due to the GK muta- tion[17,18]. Obviously, asparagine cannot act as a general acid catalyst because its side chain is incapable of providing a proton to ATP, as shown in Figure S7. Our study thus provides an explanation of why the catalytic activity of the mutant GK (K169N) in the MODY is lower than that of the wild-type GK. On the basis of crystal structures GK (the active human GK with glucose, PDB entry 1V4S) and two hexokinases (human brain hexokinase complexed with glucose and phosphate, PDB entry 1HKC, and monomeric hexokinase I complexed with ADP, PDB entry 1DGK) as templates, the whole structural model of GMAG complex (GK-Mg 2 + -ATP-glucose complex) was constructed by using a heuristic approach. The crystal structure of GK-glucose complex (1V4S) was used as a basal structure. The ATP conformation was generated according to the structure of ADP in the crystal structure of 1DGK, and the coordination configuration of Mg 2 + with ATP (Mg 2 + -ATP) was constructed based on the coordination state of Mg 2 + in the crystal structure of 1HKC. Then the Mg 2 + -ATP was ‘‘grafted’’ to the binding pocket of GK, and the primary position of Mg 2 + -ATP was decided through structural superimposition for the crystal structures of the three enzymes using the Homology module of Insight2005 (Accelrys, San Diego, CA). For the initial structure of the GMAG complex, residues within a radius of 6.5 A ̊ around the binding site were optimized by energy minimization using the AMBER force field implemented in the Sybyl6.8 (Tripos, St. Louis, MO) with following parameters: a distance-dependent dielectric function, non-bonded cutoff of 8 A ̊ , Amber charges for the protein, Gastieger-Hu ̈ckel charges for ATP and glucose. The structure was minimized by simplex method first, followed by Powell method to an energy gradient , 0.05 kcal/(mol ? A ̊ ). The structural model of G K169A MAG and G K169N MAG complex were simply constructed by replacing Lys169 with Ala and Asn respectively, followed by minimization using the same protocol as in the structural optimization for the model of GMAG. MD simulations were performed on the GMAG and G K169A MAG complexes. Before simulations, each complex was put into a suitably sized box, of which the minimal distance from the protein to the box wall was 15 A ̊ . Then the box was solvated with the SPC (Simple Point Charge) water model [37]. Each complex/ water system was submitted to energy minimization. Afterwards, counterions were added to the system to provide a neutral simulation system. The whole system was subsequently minimized again. The charges of the atoms of ATP and glucose were calculated by using the RESP ...

Context 3

... and several other hexokinases by using molecular modeling. Following this, the dynamic conformation change of the catalytic site of GK was investigated by using MD simulations. Furthermore, the modeling and simulation results were validated using mutagenesis and enzymatic kinetic assays. Encouraged by the compatible results from theoretical prediction and experiments, we explored the GK catalytic mechanism for the glucose phosphorylation based on a stable conformation of GMAG complex derived from the MD simulations by using quantum mechanics/molecular mechanics (QM/MM) method. The crystal structure of human GK had not been available until 2004 due to the high flexibility of the protein [21]. Moreover, it is easy for ATP to transfer its terminal phosphate to glucose in situ , making it difficult to gain accurate structural information of the catalytic environment of glucose inside GK, which plays a critical role in understanding the catalytic and regulation mechanisms of GK. Several models for GK catalytic environment have been published [22–27]. However, these models were built using homology modeling based on the X-ray crystal structure of human hexokinase type I [28,29]. Although human hexokinase type I is homologous to GK with sequence identity of , 55%, its three-dimensional (3D) structure is different, especially around the ATP binding pocket, as indicated by the structural superposition of the crystal structures of these two enzyme (Figure S1). The reliability of these GK catalytic models is therefore questionable. On the basis of the crystal structures of GK, now available, we built up a structural model for the GMAG complex (see Materials and Method section for detail). This model is more accurate than the previous models, and has also been validated by further MD and QM/MM calculations, as well as mutagenesis experiments and enzymatic assays (see discussion below). Structurally, GK is composed of a large domain and a small domain; the channel-shaped active site for phosphorylation is located in the deep cleft between these two domains, as shown in Figures 1A and 1B. The Mg 2 + ion and ATP (Mg 2 + -ATP) are predicted to bind to the left portion of the active site cleft, interacting with both domains [22–24]. The Mg 2 + -ATP binding site is formed by residues 78–83, 169, 225–228, 295–298, 331– 336, 410–413, and 415–417 (Figure 1C). Meanwhile, the glucose resides in a small pocket composed of residues 80–81, 151–154, 168–169, 204–207, 225–232, 254–259, 287 and 290 (Figure 1D). Our 3D model of GMAG complex reveals that GK provides a favorable microenvironment for the phosphorylation of glucose. Several important hydrogen bonds between ligands (ATP, Mg 2 + and glucose) and GK are observed in our model (Table S1), which are in agreement with a large number of mutagenesis analysis data [20,30–34]. The c -phosphate of ATP is close to the –O 6 H group of glucose with electrostatic interaction, indicating that the c phosphate is about to be transferred to glucose. To study the stability of the reaction environment, a 10-ns molecular dynamics (MD) simulation was performed on the GMAG complex. The time evolutions of the weighted Root-Mean Square Deviations (wRMSD) for the atoms of GK, ATP and glucose from their initial positions ( t = 0 ps) were monitored. The result indicates that all the reaction components are relatively stable except for the rotation of the –O 6 H group which causes an increase in wRMSD of glucose by 0.3 A ̊ at 1 ns (Figure S2). The H-bond network amongst GK, ATP and glucose also reflects the stability of the phosphorylation environment. These H-bonds were mostly maintained during the 10-ns MD simulation as indicated by their occupancies (Table S1). Under the restriction of GK, ATP, Mg 2 + and glucose form a highly advantageous configuration for the biochemical reaction (Figure 2). Mg 2 + ion in the binding site octahedrally coordinates with the oxygen atoms of two water molecules, O b 1 , O c 1 and O c 3 oxygen atoms of ATP and the O d 2 atom of Asp205. Nine H-bonds are formed between ATP and residues of GK, i.e. the N 1 atom of adenine with the –O c H group of Ser336, the O a 2 atom of ATP with the –NH groups of Thr82 and Asn83, the O a 1 atom of ATP with the –O c H group of Ser411, the O a 3 and O b 3 atoms of ATP with the –O c H Thr228, the O c 2 atom of ATP with the –NH group of Gly229, and the O c 1 and O c 2 atoms of ATP with the f + –N H 3 group of Lys169 (Figure 2 and Table S1). These H-bonds lead the ATP to adopt a conformation appropriate to coordinate with the Mg 2 + ion and to interact with glucose. On the other side, glucose forms seven H-bonds with the GK residues, viz. the –O 4 H group of glucose with the N d 1 atom of Asn204 and the O d 1 atom of Asp205, the –O 3 H group of glucose with the O e 1 and O e 2 atoms of Glu256 and the –NH group of Phe152, the –O 1 H group of glucose with the O e 1 atom of Glu290, the O 5 atom of glucose with f + the –N H 3 group of Lys169 (Figure 2 and Table S1). Most importantly, the –O 6 H group of glucose directly hydrogen bonds to the O c 2 atom of ATP, producing the requisite complex for glucose phosphorylation. Residues detected in the naturally occurring mutations (e.g. K169N, T228M, E256K, S336L, and S411F) from MODY families [20,30–32] are involved in the H- bond network for the binding of GK with glucose and ATP. Extensive work confirmed that S336L mutation decreased both the binding affinity of GK to ATP and the catalytic activity of the enzyme [33]. Several other site-directed mutants, including N204A, E256A and E290A, have been shown to alter GK’s enzymatic kinetics by decreasing the binding affinity of glucose and lowering V max in the catalytic process [31,34]. These mutagenesis and enzymatic results demonstrate that our 3D model of GMAG complex is reasonable. Among the conserved residues hydrogen bonding to ATP and glucose, Lys169 is of special interest because its cationic end, f + –N H 3 , bridges ATP and glucose together through H-bonds, placing the c -phosphorus (P c ) atom only 3.3 A ̊ far way from the –O 6 H group of glucose (Figure 2). This implies that Lys169 might play an important role in the phosphorylation of glucose. Additionally, it was reported that Lys169Asn (K169N) is one of the naturally occurring mutations in the GCK gene associated with familial mild fasting hyperglycemia [20]. The K169N mutant was also shown to experience a partial loss of glucose binding [30]. To address the biological function of Lys169, we constructed a 3D structural model for the GK K169A mutant (GK K169A ) in complex with ATP-Mg 2 + and glucose, and two additional MD simulations of the ATP-Mg 2 + -GK K169A and glucose-GK K169A complexes were conducted. Binding free energies of ATP and glucose with wild-type GK and GK K169A (Table S2) were then calculated using the MM-PBSA method encoded in AMBER (Version 8.0). Both ATP and glucose are able to bind tightly to the wild-type GK with calculated binding free energies of 2 18.67 6 4.59 and 2 44.06 6 3.94 kcal/mol, respectively. While neither ATP nor glucose can bind with GK K169A . The MD simulations and binding free energy calculations indicate that Lys169 may dominate the binding of both ATP and glucose, and hence the K169A mutant possibly loses the catalytic function for the phosphorylation of glucose. To verify this conclusion, a bioassay on the K169A mutant was performed and result is shown in Table 1. The kinetic assay indicates that K169A is indeed an inactivating mutation; its enzymatic activity is completely abolished. Far-UV CD spectra and fluorescence emission spectra results demonstrated that the K169A mutant had well-defined secondary structures; thereby we can exclude the possibility that this mutation may have caused GK to undergo misfolding (Figures S3, S4, S5). Thus, the MD prediction, regarding the importance of Lys169 to the binding of ATP and glucose with GK and to the catalytic activity of GK as well was validated. To further figure out the role of Lys169 in enzymatic catalysis, the reaction mechanism of glucose phosphorylation inside the active site of GK was investigated by using the QM/MM approach. The system for QM/MM simulation was constructed based on the snapshot at 2500 ps of the MD trajectory. The QM region was composed of Mg 2 + , water molecules coordinating with Mg 2 + , important groups of ATP, glucose, Lys169 and Asp205; the remainder of GK was included in the MM region (Figure S6). The partitioning scheme for QM and MM regions is described in the Materials and Methods section and Figure S6. We designate this structure as reagent system. For the reagent system, the initial coordination configuration of Mg 2 + with waters, ATP and glucose and the hydrogen bonding pattern amongst ATP, glucose, Lys169 and Asp205 have been described above (Figure 2). ONIOM, a QM/MM method encoded in Gaussion03 [35], was used for all the QM/MM calculations (Figure 3 and 4). The QM/MM optimized geometry of the reagent system showed that Mg 2 + octahedrally coordinates with the O b 1 and O c 1 atoms of ATP, the O atom of Asp205 and three oxygen atoms from three water molecules (Figure 3). The major difference of the optimized structure for the reaction center from the initial structure obtained by MD simulation is that one water molecule substituted the O c 3 atom to coordinate with the Mg 2 + ion. QM optimization resulted in important structural changes viz. one proton of the –N f H + 3 group of Lys169 transferred to the O c 3 f atom, and the resulting –N H 2 group formed a H-bond with the new water molecule coordinating with the Mg 2 + ion; on the other hand, the –O 6 H group of glucose adjusted its direction to point toward the O d 1 atom of Asp205, forming two strong H-bonds with the O d 1 and O d 2 atoms of Asp205. This structural reorganization made the –O 6 H group more propitious to attack the P c atom of ATP, implying that Lys169 might act as an acid catalyst and ...

Context 4

... in GK. The main goal of this study is to investigate the catalytic process of human GK by using computational methods in conjunction with enzymatic assay. Such a study, however, needs an accurate structural model for the whole catalytic environment of GK, i.e. the three-dimensional (3D) structure of the complex of GK with ATP, glucose and Mg 2 + (GK-Mg 2 + -ATP-glucose complex, designated as GMAG complex hereinafter). Accordingly, a 3D structural model of GMAG complex was constructed based on the crystal structures of GK and several other hexokinases by using molecular modeling. Following this, the dynamic conformation change of the catalytic site of GK was investigated by using MD simulations. Furthermore, the modeling and simulation results were validated using mutagenesis and enzymatic kinetic assays. Encouraged by the compatible results from theoretical prediction and experiments, we explored the GK catalytic mechanism for the glucose phosphorylation based on a stable conformation of GMAG complex derived from the MD simulations by using quantum mechanics/molecular mechanics (QM/MM) method. The crystal structure of human GK had not been available until 2004 due to the high flexibility of the protein [21]. Moreover, it is easy for ATP to transfer its terminal phosphate to glucose in situ , making it difficult to gain accurate structural information of the catalytic environment of glucose inside GK, which plays a critical role in understanding the catalytic and regulation mechanisms of GK. Several models for GK catalytic environment have been published [22–27]. However, these models were built using homology modeling based on the X-ray crystal structure of human hexokinase type I [28,29]. Although human hexokinase type I is homologous to GK with sequence identity of , 55%, its three-dimensional (3D) structure is different, especially around the ATP binding pocket, as indicated by the structural superposition of the crystal structures of these two enzyme (Figure S1). The reliability of these GK catalytic models is therefore questionable. On the basis of the crystal structures of GK, now available, we built up a structural model for the GMAG complex (see Materials and Method section for detail). This model is more accurate than the previous models, and has also been validated by further MD and QM/MM calculations, as well as mutagenesis experiments and enzymatic assays (see discussion below). Structurally, GK is composed of a large domain and a small domain; the channel-shaped active site for phosphorylation is located in the deep cleft between these two domains, as shown in Figures 1A and 1B. The Mg 2 + ion and ATP (Mg 2 + -ATP) are predicted to bind to the left portion of the active site cleft, interacting with both domains [22–24]. The Mg 2 + -ATP binding site is formed by residues 78–83, 169, 225–228, 295–298, 331– 336, 410–413, and 415–417 (Figure 1C). Meanwhile, the glucose resides in a small pocket composed of residues 80–81, 151–154, 168–169, 204–207, 225–232, 254–259, 287 and 290 (Figure 1D). Our 3D model of GMAG complex reveals that GK provides a favorable microenvironment for the phosphorylation of glucose. Several important hydrogen bonds between ligands (ATP, Mg 2 + and glucose) and GK are observed in our model (Table S1), which are in agreement with a large number of mutagenesis analysis data [20,30–34]. The c -phosphate of ATP is close to the –O 6 H group of glucose with electrostatic interaction, indicating that the c phosphate is about to be transferred to glucose. To study the stability of the reaction environment, a 10-ns molecular dynamics (MD) simulation was performed on the GMAG complex. The time evolutions of the weighted Root-Mean Square Deviations (wRMSD) for the atoms of GK, ATP and glucose from their initial positions ( t = 0 ps) were monitored. The result indicates that all the reaction components are relatively stable except for the rotation of the –O 6 H group which causes an increase in wRMSD of glucose by 0.3 A ̊ at 1 ns (Figure S2). The H-bond network amongst GK, ATP and glucose also reflects the stability of the phosphorylation environment. These H-bonds were mostly maintained during the 10-ns MD simulation as indicated by their occupancies (Table S1). Under the restriction of GK, ATP, Mg 2 + and glucose form a highly advantageous configuration for the biochemical reaction (Figure 2). Mg 2 + ion in the binding site octahedrally coordinates with the oxygen atoms of two water molecules, O b 1 , O c 1 and O c 3 oxygen atoms of ATP and the O d 2 atom of Asp205. Nine H-bonds are formed between ATP and residues of GK, i.e. the N 1 atom of adenine with the –O c H group of Ser336, the O a 2 atom of ATP with the –NH groups of Thr82 and Asn83, the O a 1 atom of ATP with the –O c H group of Ser411, the O a 3 and O b 3 atoms of ATP with the –O c H Thr228, the O c 2 atom of ATP with the –NH group of Gly229, and the O c 1 and O c 2 atoms of ATP with the f + –N H 3 group of Lys169 (Figure 2 and Table S1). These H-bonds lead the ATP to adopt a conformation appropriate to coordinate with the Mg 2 + ion and to interact with glucose. On the other side, glucose forms seven H-bonds with the GK residues, viz. the –O 4 H group of glucose with the N d 1 atom of Asn204 and the O d 1 atom of Asp205, the –O 3 H group of glucose with the O e 1 and O e 2 atoms of Glu256 and the –NH group of Phe152, the –O 1 H group of glucose with the O e 1 atom of Glu290, the O 5 atom of glucose with f + the –N H 3 group of Lys169 (Figure 2 and Table S1). Most importantly, the –O 6 H group of glucose directly hydrogen bonds to the O c 2 atom of ATP, producing the requisite complex for glucose phosphorylation. Residues detected in the naturally occurring mutations (e.g. K169N, T228M, E256K, S336L, and S411F) from MODY families [20,30–32] are involved in the H- bond network for the binding of GK with glucose and ATP. Extensive work confirmed that S336L mutation decreased both the binding affinity of GK to ATP and the catalytic activity of the enzyme [33]. Several other site-directed mutants, including N204A, E256A and E290A, have been shown to alter GK’s enzymatic kinetics by decreasing the binding affinity of glucose and lowering V max in the catalytic process [31,34]. These mutagenesis and enzymatic results demonstrate that our 3D model of GMAG complex is reasonable. Among the conserved residues hydrogen bonding to ATP and glucose, Lys169 is of special interest because its cationic end, f + –N H 3 , bridges ATP and glucose together through H-bonds, placing the c -phosphorus (P c ) atom only 3.3 A ̊ far way from the –O 6 H group of glucose (Figure 2). This implies that Lys169 might play an important role in the phosphorylation of glucose. Additionally, it was reported that Lys169Asn (K169N) is one of the naturally occurring mutations in the GCK gene associated with familial mild fasting hyperglycemia [20]. The K169N mutant was also shown to experience a partial loss of glucose binding [30]. To address the biological function of Lys169, we constructed a 3D structural model for the GK K169A mutant (GK K169A ) in complex with ATP-Mg 2 + and glucose, and two additional MD simulations of the ATP-Mg 2 + -GK K169A and glucose-GK K169A complexes were conducted. Binding free energies of ATP and glucose with wild-type GK and GK K169A (Table S2) were then calculated using the MM-PBSA method encoded in AMBER (Version 8.0). Both ATP and glucose are able to bind tightly to the wild-type GK with calculated binding free energies of 2 18.67 6 4.59 and 2 44.06 6 3.94 kcal/mol, respectively. While neither ATP nor glucose can bind with GK K169A . The MD simulations and binding free energy calculations indicate that Lys169 may dominate the binding of both ATP and glucose, and hence the K169A mutant possibly loses the catalytic function for the phosphorylation of glucose. To verify this conclusion, a bioassay on the K169A mutant was performed and result is shown in Table 1. The kinetic assay indicates that K169A is indeed an inactivating mutation; its enzymatic activity is completely abolished. Far-UV CD spectra and fluorescence emission spectra results demonstrated that the K169A mutant had well-defined secondary structures; thereby we can exclude the possibility that this mutation may have caused GK to undergo misfolding (Figures S3, S4, S5). Thus, the MD prediction, regarding the importance of Lys169 to the binding of ATP and glucose with GK and to the catalytic activity of GK as well was validated. To further figure out the role of Lys169 in enzymatic catalysis, the reaction mechanism of glucose phosphorylation inside the active site of GK was investigated by using the QM/MM approach. The system for QM/MM simulation was constructed based on the snapshot at 2500 ps of the MD trajectory. The QM region was composed of Mg 2 + , water molecules coordinating with Mg 2 + , important groups of ATP, glucose, Lys169 and Asp205; the remainder of GK was included in the MM region (Figure S6). The partitioning scheme for QM and MM regions is described in the Materials and Methods section and Figure S6. We designate this structure as reagent system. For the reagent system, the initial coordination configuration of Mg 2 + with waters, ATP and glucose and the hydrogen bonding pattern amongst ATP, glucose, Lys169 and Asp205 have been described above (Figure 2). ONIOM, a QM/MM method encoded in Gaussion03 [35], was used for all the QM/MM calculations (Figure 3 and 4). The QM/MM optimized geometry of the reagent system showed that Mg 2 + octahedrally coordinates with the O b 1 and O c 1 atoms of ATP, the O atom of Asp205 and three oxygen atoms from three water molecules (Figure 3). The major difference of the optimized structure for the reaction center from the initial structure obtained by MD simulation is that one water molecule substituted the O c 3 atom to coordinate with the Mg 2 + ion. QM optimization resulted in important structural changes ...

Context 5

... (Figure 2). Mg 2 + ion in the binding site octahedrally coordinates with the oxygen atoms of two water molecules, O b 1 , O c 1 and O c 3 oxygen atoms of ATP and the O d 2 atom of Asp205. Nine H-bonds are formed between ATP and residues of GK, i.e. the N 1 atom of adenine with the –O c H group of Ser336, the O a 2 atom of ATP with the –NH groups of Thr82 and Asn83, the O a 1 atom of ATP with the –O c H group of Ser411, the O a 3 and O b 3 atoms of ATP with the –O c H Thr228, the O c 2 atom of ATP with the –NH group of Gly229, and the O c 1 and O c 2 atoms of ATP with the f + –N H 3 group of Lys169 (Figure 2 and Table S1). These H-bonds lead the ATP to adopt a conformation appropriate to coordinate with the Mg 2 + ion and to interact with glucose. On the other side, glucose forms seven H-bonds with the GK residues, viz. the –O 4 H group of glucose with the N d 1 atom of Asn204 and the O d 1 atom of Asp205, the –O 3 H group of glucose with the O e 1 and O e 2 atoms of Glu256 and the –NH group of Phe152, the –O 1 H group of glucose with the O e 1 atom of Glu290, the O 5 atom of glucose with f + the –N H 3 group of Lys169 (Figure 2 and Table S1). Most importantly, the –O 6 H group of glucose directly hydrogen bonds to the O c 2 atom of ATP, producing the requisite complex for glucose phosphorylation. Residues detected in the naturally occurring mutations (e.g. K169N, T228M, E256K, S336L, and S411F) from MODY families [20,30–32] are involved in the H- bond network for the binding of GK with glucose and ATP. Extensive work confirmed that S336L mutation decreased both the binding affinity of GK to ATP and the catalytic activity of the enzyme [33]. Several other site-directed mutants, including N204A, E256A and E290A, have been shown to alter GK’s enzymatic kinetics by decreasing the binding affinity of glucose and lowering V max in the catalytic process [31,34]. These mutagenesis and enzymatic results demonstrate that our 3D model of GMAG complex is reasonable. Among the conserved residues hydrogen bonding to ATP and glucose, Lys169 is of special interest because its cationic end, f + –N H 3 , bridges ATP and glucose together through H-bonds, placing the c -phosphorus (P c ) atom only 3.3 A ̊ far way from the –O 6 H group of glucose (Figure 2). This implies that Lys169 might play an important role in the phosphorylation of glucose. Additionally, it was reported that Lys169Asn (K169N) is one of the naturally occurring mutations in the GCK gene associated with familial mild fasting hyperglycemia [20]. The K169N mutant was also shown to experience a partial loss of glucose binding [30]. To address the biological function of Lys169, we constructed a 3D structural model for the GK K169A mutant (GK K169A ) in complex with ATP-Mg 2 + and glucose, and two additional MD simulations of the ATP-Mg 2 + -GK K169A and glucose-GK K169A complexes were conducted. Binding free energies of ATP and glucose with wild-type GK and GK K169A (Table S2) were then calculated using the MM-PBSA method encoded in AMBER (Version 8.0). Both ATP and glucose are able to bind tightly to the wild-type GK with calculated binding free energies of 2 18.67 6 4.59 and 2 44.06 6 3.94 kcal/mol, respectively. While neither ATP nor glucose can bind with GK K169A . The MD simulations and binding free energy calculations indicate that Lys169 may dominate the binding of both ATP and glucose, and hence the K169A mutant possibly loses the catalytic function for the phosphorylation of glucose. To verify this conclusion, a bioassay on the K169A mutant was performed and result is shown in Table 1. The kinetic assay indicates that K169A is indeed an inactivating mutation; its enzymatic activity is completely abolished. Far-UV CD spectra and fluorescence emission spectra results demonstrated that the K169A mutant had well-defined secondary structures; thereby we can exclude the possibility that this mutation may have caused GK to undergo misfolding (Figures S3, S4, S5). Thus, the MD prediction, regarding the importance of Lys169 to the binding of ATP and glucose with GK and to the catalytic activity of GK as well was validated. To further figure out the role of Lys169 in enzymatic catalysis, the reaction mechanism of glucose phosphorylation inside the active site of GK was investigated by using the QM/MM approach. The system for QM/MM simulation was constructed based on the snapshot at 2500 ps of the MD trajectory. The QM region was composed of Mg 2 + , water molecules coordinating with Mg 2 + , important groups of ATP, glucose, Lys169 and Asp205; the remainder of GK was included in the MM region (Figure S6). The partitioning scheme for QM and MM regions is described in the Materials and Methods section and Figure S6. We designate this structure as reagent system. For the reagent system, the initial coordination configuration of Mg 2 + with waters, ATP and glucose and the hydrogen bonding pattern amongst ATP, glucose, Lys169 and Asp205 have been described above (Figure 2). ONIOM, a QM/MM method encoded in Gaussion03 [35], was used for all the QM/MM calculations (Figure 3 and 4). The QM/MM optimized geometry of the reagent system showed that Mg 2 + octahedrally coordinates with the O b 1 and O c 1 atoms of ATP, the O atom of Asp205 and three oxygen atoms from three water molecules (Figure 3). The major difference of the optimized structure for the reaction center from the initial structure obtained by MD simulation is that one water molecule substituted the O c 3 atom to coordinate with the Mg 2 + ion. QM optimization resulted in important structural changes viz. one proton of the –N f H + 3 group of Lys169 transferred to the O c 3 f atom, and the resulting –N H 2 group formed a H-bond with the new water molecule coordinating with the Mg 2 + ion; on the other hand, the –O 6 H group of glucose adjusted its direction to point toward the O d 1 atom of Asp205, forming two strong H-bonds with the O d 1 and O d 2 atoms of Asp205. This structural reorganization made the –O 6 H group more propitious to attack the P c atom of ATP, implying that Lys169 might act as an acid catalyst and Asp205 as a base catalyst in the phosphorylation of glucose. Thus, we propose a mechanism for the glucose phosphorylation catalyzed by GK as Scheme 1 (Figure 5). Along this reaction path, the energies of the reagent (R), transition state (TS), and immediate product (P) were determined by two-dimensional QM/ MM potential energy surface by defining the distances of R (O _P ) and R (P _O ) as the reaction coordinates (Figure 3). In the optimized reagent, R (O b 3 _P c ) = 1.69 A ̊ and R (P c _O 6 ) = 3.39 A ̊ ; and in the optimized immediate product, the O b 3 _P c bond is broken and the P c _O 6 is formed, R (P c _O 6 ) = 1.71 A ̊. The calculated potential energy barrier of Scheme 1 is D E ? = 18.3 kcal/mol. The structure of the transition state (TS) in Scheme 1 was determined by adiabatic mapping at the QM/MM level (Figure 3). In the TS, R (O b 3 _P c ) = 2.1 A ̊ and R (P c _O 6 ) = 2.0 A ̊ , and the c c phosphate (–P O 3 ) is turned to be in a plane. This structure clearly illustrates that the P c _O 6 bond is partially formed and the O b 3 _P c bond is partially broken. The overall reaction is calculated to be exothermic by D E = 2 22.1 kcal/mol. To further address the catalytic role of Lys169, we investigated the reaction mechanism of glucose phosphorylation catalyzed by the GK K169A mutant employing the QM/MM method. Com- pared with the optimized structure of GK reagent system, the major difference for the ATP-GK K169A -glucose reaction center is that the –O 6 H group of glucose doesn’t form H-bond with either O d 1 or O d 2 atom of Asp205, but forms a strong H-bond with the O atom of ATP. This optimized structure implies an alternative path for the glucose phosphorylation as Scheme 2 (Figure 6): the proton of the –O 6 H group transfers to the O c 3 atom of ATP while the O 6 atom attacks the P c atom of ATP, and at the same time the O b 3 _P c bond is broken. According to this reaction path, we obtained the structures and energies of reagent (R), transition state (TS), and immediate product (P) by calculating the two- dimensional QM/MM potential energy surface; the distances of R (O b 3 _P c ) and R (P c _O 6 ) were defined as the reaction coordinates. The result is shown in Figure 4. For the TS, R (O b 3 _P c ) = 2.1 A ̊ and R (P c _O 6 ) = 1.9 A ̊ . Remarkably, the calculated potential energy barrier of this reaction path is D E ? = 32.1 kcal/mol, which is , 14 kcal/mol higher than that of the reaction catalyzed by the wild-type GK. This result indicates that the phosphorylation of glucose could not occur without the assistance of Lys169. As mentioned above, GK is a key enzyme that phosphorylates glucose and triggers glucose utilization and metabolism. As an attractive drug target for discovering anti-diabetes drug, the importance of this enzyme has been appreciated. However, the unique catalytic mechanism of GK still remains unclear [17,18]. Several mutants of GK with the MODY phenotype have been identified recently; in particular, it was found that K169N is a naturally occurring mutation (K169N) in MODY [20]. In this study, we have investigated the phosphorylation mechanism of glucose catalyzed by GK, and revealed that Lys169 is a crucial residue for glucose phosphorylation. By using molecular modeling and simulation methods, we constructed a 3D structural model for the complex of GK with ATP, glucose and Mg 2 + (GMAG complex). This structural model is more reliable than previously published models [22–27,36] because it was constructed based on the crystal structures of human GK. In particular, this model provides a more realistic reaction environment for the phosphorylation of glucose. The validity of the structural model of GMAG was confirmed by the agreement between the available experimental mutagenesis and enzymatic data for GK and the important interactions of GK with ...

Context 6

... the recent findings of mutagenesis experiments and related kinetic studies. MD simulations further revealed that Lys169 may play an essential role in both ligand binding and GK catalytic process. The computational prediction was verified by additional experimental mutagenesis and kinetic analysis. Based on these results, we propose an atomistic catalytic mechanism of GK for glucose phosphorylation using QM/MM calculations. Findings from this work provide a better understanding of the enzymatic mechanism of GK, and also a potential explanation of the pathogenic mechanism of MODY caused by the mutation in GK. The main goal of this study is to investigate the catalytic process of human GK by using computational methods in conjunction with enzymatic assay. Such a study, however, needs an accurate structural model for the whole catalytic environment of GK, i.e. the three-dimensional (3D) structure of the complex of GK with ATP, glucose and Mg 2 + (GK-Mg 2 + -ATP-glucose complex, designated as GMAG complex hereinafter). Accordingly, a 3D structural model of GMAG complex was constructed based on the crystal structures of GK and several other hexokinases by using molecular modeling. Following this, the dynamic conformation change of the catalytic site of GK was investigated by using MD simulations. Furthermore, the modeling and simulation results were validated using mutagenesis and enzymatic kinetic assays. Encouraged by the compatible results from theoretical prediction and experiments, we explored the GK catalytic mechanism for the glucose phosphorylation based on a stable conformation of GMAG complex derived from the MD simulations by using quantum mechanics/molecular mechanics (QM/MM) method. The crystal structure of human GK had not been available until 2004 due to the high flexibility of the protein [21]. Moreover, it is easy for ATP to transfer its terminal phosphate to glucose in situ , making it difficult to gain accurate structural information of the catalytic environment of glucose inside GK, which plays a critical role in understanding the catalytic and regulation mechanisms of GK. Several models for GK catalytic environment have been published [22–27]. However, these models were built using homology modeling based on the X-ray crystal structure of human hexokinase type I [28,29]. Although human hexokinase type I is homologous to GK with sequence identity of , 55%, its three-dimensional (3D) structure is different, especially around the ATP binding pocket, as indicated by the structural superposition of the crystal structures of these two enzyme (Figure S1). The reliability of these GK catalytic models is therefore questionable. On the basis of the crystal structures of GK, now available, we built up a structural model for the GMAG complex (see Materials and Method section for detail). This model is more accurate than the previous models, and has also been validated by further MD and QM/MM calculations, as well as mutagenesis experiments and enzymatic assays (see discussion below). Structurally, GK is composed of a large domain and a small domain; the channel-shaped active site for phosphorylation is located in the deep cleft between these two domains, as shown in Figures 1A and 1B. The Mg 2 + ion and ATP (Mg 2 + -ATP) are predicted to bind to the left portion of the active site cleft, interacting with both domains [22–24]. The Mg 2 + -ATP binding site is formed by residues 78–83, 169, 225–228, 295–298, 331– 336, 410–413, and 415–417 (Figure 1C). Meanwhile, the glucose resides in a small pocket composed of residues 80–81, 151–154, 168–169, 204–207, 225–232, 254–259, 287 and 290 (Figure 1D). Our 3D model of GMAG complex reveals that GK provides a favorable microenvironment for the phosphorylation of glucose. Several important hydrogen bonds between ligands (ATP, Mg 2 + and glucose) and GK are observed in our model (Table S1), which are in agreement with a large number of mutagenesis analysis data [20,30–34]. The c -phosphate of ATP is close to the –O 6 H group of glucose with electrostatic interaction, indicating that the c phosphate is about to be transferred to glucose. To study the stability of the reaction environment, a 10-ns molecular dynamics (MD) simulation was performed on the GMAG complex. The time evolutions of the weighted Root-Mean Square Deviations (wRMSD) for the atoms of GK, ATP and glucose from their initial positions ( t = 0 ps) were monitored. The result indicates that all the reaction components are relatively stable except for the rotation of the –O 6 H group which causes an increase in wRMSD of glucose by 0.3 A ̊ at 1 ns (Figure S2). The H-bond network amongst GK, ATP and glucose also reflects the stability of the phosphorylation environment. These H-bonds were mostly maintained during the 10-ns MD simulation as indicated by their occupancies (Table S1). Under the restriction of GK, ATP, Mg 2 + and glucose form a highly advantageous configuration for the biochemical reaction (Figure 2). Mg 2 + ion in the binding site octahedrally coordinates with the oxygen atoms of two water molecules, O b 1 , O c 1 and O c 3 oxygen atoms of ATP and the O d 2 atom of Asp205. Nine H-bonds are formed between ATP and residues of GK, i.e. the N 1 atom of adenine with the –O c H group of Ser336, the O a 2 atom of ATP with the –NH groups of Thr82 and Asn83, the O a 1 atom of ATP with the –O c H group of Ser411, the O a 3 and O b 3 atoms of ATP with the –O c H Thr228, the O c 2 atom of ATP with the –NH group of Gly229, and the O c 1 and O c 2 atoms of ATP with the f + –N H 3 group of Lys169 (Figure 2 and Table S1). These H-bonds lead the ATP to adopt a conformation appropriate to coordinate with the Mg 2 + ion and to interact with glucose. On the other side, glucose forms seven H-bonds with the GK residues, viz. the –O 4 H group of glucose with the N d 1 atom of Asn204 and the O d 1 atom of Asp205, the –O 3 H group of glucose with the O e 1 and O e 2 atoms of Glu256 and the –NH group of Phe152, the –O 1 H group of glucose with the O e 1 atom of Glu290, the O 5 atom of glucose with f + the –N H 3 group of Lys169 (Figure 2 and Table S1). Most importantly, the –O 6 H group of glucose directly hydrogen bonds to the O c 2 atom of ATP, producing the requisite complex for glucose phosphorylation. Residues detected in the naturally occurring mutations (e.g. K169N, T228M, E256K, S336L, and S411F) from MODY families [20,30–32] are involved in the H- bond network for the binding of GK with glucose and ATP. Extensive work confirmed that S336L mutation decreased both the binding affinity of GK to ATP and the catalytic activity of the enzyme [33]. Several other site-directed mutants, including N204A, E256A and E290A, have been shown to alter GK’s enzymatic kinetics by decreasing the binding affinity of glucose and lowering V max in the catalytic process [31,34]. These mutagenesis and enzymatic results demonstrate that our 3D model of GMAG complex is reasonable. Among the conserved residues hydrogen bonding to ATP and glucose, Lys169 is of special interest because its cationic end, f + –N H 3 , bridges ATP and glucose together through H-bonds, placing the c -phosphorus (P c ) atom only 3.3 A ̊ far way from the –O 6 H group of glucose (Figure 2). This implies that Lys169 might play an important role in the phosphorylation of glucose. Additionally, it was reported that Lys169Asn (K169N) is one of the naturally occurring mutations in the GCK gene associated with familial mild fasting hyperglycemia [20]. The K169N mutant was also shown to experience a partial loss of glucose binding [30]. To address the biological function of Lys169, we constructed a 3D structural model for the GK K169A mutant (GK K169A ) in complex with ATP-Mg 2 + and glucose, and two additional MD simulations of the ATP-Mg 2 + -GK K169A and glucose-GK K169A complexes were conducted. Binding free energies of ATP and glucose with wild-type GK and GK K169A (Table S2) were then calculated using the MM-PBSA method encoded in AMBER (Version 8.0). Both ATP and glucose are able to bind tightly to the wild-type GK with calculated binding free energies of 2 18.67 6 4.59 and 2 44.06 6 3.94 kcal/mol, respectively. While neither ATP nor glucose can bind with GK K169A . The MD simulations and binding free energy calculations indicate that Lys169 may dominate the binding of both ATP and glucose, and hence the K169A mutant possibly loses the catalytic function for the phosphorylation of glucose. To verify this conclusion, a bioassay on the K169A mutant was performed and result is shown in Table 1. The kinetic assay indicates that K169A is indeed an inactivating mutation; its enzymatic activity is completely abolished. Far-UV CD spectra and fluorescence emission spectra results demonstrated that the K169A mutant had well-defined secondary structures; thereby we can exclude the possibility that this mutation may have caused GK to undergo misfolding (Figures S3, S4, S5). Thus, the MD prediction, regarding the importance of Lys169 to the binding of ATP and glucose with GK and to the catalytic activity of GK as well was validated. To further figure out the role of Lys169 in enzymatic catalysis, the reaction mechanism of glucose phosphorylation inside the active site of GK was investigated by using the QM/MM approach. The system for QM/MM simulation was constructed based on the snapshot at 2500 ps of the MD trajectory. The QM region was composed of Mg 2 + , water molecules coordinating with Mg 2 + , important groups of ATP, glucose, Lys169 and Asp205; the remainder of GK was included in the MM region (Figure S6). The partitioning scheme for QM and MM regions is described in the Materials and Methods section and Figure S6. We designate this structure as reagent system. For the reagent system, the initial coordination configuration of Mg 2 + with waters, ATP and glucose and the hydrogen bonding pattern amongst ATP, glucose, Lys169 and Asp205 have been described above ...