Molecular Mechanical Investigation of

The enzyme β-1,4-galactosyltransferase-1 catalyzes the transfer of a galactose residue from UDP-Gal to the C4-hydroxyl group of N-acetylglucosamine...
0 downloads 0 Views 3MB Size
Download PDF
11314

J. Phys. Chem. B 2009, 113, 11314–11319

Hybrid Quantum Mechanical/Molecular Mechanical Investigation of the β-1,4-Galactosyltransferase-I Mechanism Martin Krupicˇka and Igor Tvarosˇka* Contribution from the Institute of Chemistry, Centre for Glycomics, SloVak Academy of Sciences, Du´braVska cesta 9, 845 38 BratislaVa, SloVak Republic ReceiVed: May 20, 2009; ReVised Manuscript ReceiVed: July 3, 2009

The enzyme β-1,4-galactosyltransferase-1 (β4Gal-T1) catalyzes the transfer of a galactose residue from UDPGal to the C4-hydroxyl group of N-acetylglucosamine. The catalytic mechanism of β4Gal-T1 was investigated using the hybrid quantum mechanical/molecular mechanical (QM/MM) method, with the QM portion containing 253 atoms treated with density functional theory (DFT) at the BP/DZP and BP/TZ2P levels. The remaining parts of the β4Gal-T1 complex, 4527 atoms in all, were modeled using the AMBER molecular force field. A theoretical model of the Michaelis complex was built using the X-ray structure of β4Gal-T1 in a complex with the donor or acceptor substrate, respectively. The hybrid QM(DFT)/MM calculations identified an SN2type transition state for the nucleophilic attack of the O4a oxygen on the anomeric carbon C1 and the breaking of the C1-O1 glycosidic linkage. The activation barrier found for this process is 15 kcal/mol. In the transition state (TS) model, the sugar donor is partially cleaved from pyrophosphate, while nucleophilic oxygen O4a remains protonated with a low barrier hydrogen bond to the catalytic base D318. The structure of TS is characterized by the O4a-C1 and C1-O1 distances of 2.703 and 2.092 Å, respectively. When the obtained reaction sequence was used, the nature of the captured intermediate resembling the transition state structure (PDB/2FYD) was elucidated. This modeling QM/MM study has provided detailed insight into the mechanism of the Gal transfer catalyzed by β4Gal-T1 and has supplied further evidence for a concerted SN2-type displacement mechanism employed by inverting glycosyltransferases. 1. Introduction The glycosyltransferase (GT) superfamily comprises enzymes able to catalyze the formation of a glycosidic bond using activated saccharide donors. Donor substrates are most frequently nucleotide diphosphates, for example, UDP GlcNAc, UDP Glc, UDP Gal, and GDP Man, though nucleotide monophosphates and lipid phosphates are also utilized. Acceptor substrates used by glycosyltransferases are usually other carbohydrates but can also be other small molecules. Biosynthesized carbohydrate components of glycoproteins are involved in many physiological and pathological processes,1 and a lack of or mutations in glycosyltransferases are known to cause diseases in humans. Therefore, inhibitors of glycosyltransferases are promising targets with the potential to lead to therapeutic agents. Glycosyltransferases are classified into 90 distinct families based on amino acid similarities (http://www.cazy.org). Structural information from previously solved GT crystal structures reveals only two three-dimensional folds, termed GT-A and GT-B.2-4 Although another new fold, GT-C, has been recently predicted.5 An important class of glycosyltransferases is the galactosyltransferase family, which includes a subfamily of enzymes that use the same donor, UDP-Gal [uridine 5′-(R-Dgalactopyranosyl pyrophosphate)], to transfer the R-D-galactopyranose (Gal) residue to different acceptors. Over the last three decades, the structure and kinetic mechanism of the inverting β-1,4-galactosyltransferase-1 (β4Gal-T1; EC 2.4.1.38) that belongs to the GT7 family have been extensively investigated.6-25 This enzyme, in the absence of R-lactalbumin (LA), catalyzes * Corresponding author. E-mail: [email protected].

Figure 1. Schematic diagram of the enzymatic reaction catalyzed by β4Gal-T1.

the transfer of the Gal residue from UDP-Gal to the O4 oxygen of N-acetylglucosamine (GlcNAc), thus creating a new β-1,4glycosidic linkage (Figure 1). The catalytic function of β4Gal-T1 requires the presence of the Mn2+ transition metal ion. The crystal structures of the catalytic domain of β4Gal-T1 have been solved both without any substrate and with bound substrates, and it appears that β4Gal-T1 adopts the GT-A fold.14 However, as in many glycosyltransferases, the crystal structure having both the donor and acceptor (or their analogues) bound to the enzyme was not solved. Based on the crystal structural data, a mechanistic scheme has been proposed for the β4Gal-T1 catalytic action. In the ordered sequential mechanism, Mn2+ binds first to the enzyme, followed by UDP-Gal, and then to the acceptor

10.1021/jp904716t CCC: $40.75  2009 American Chemical Society Published on Web 07/23/2009

β-1,4-Galactosyltransferase-I Mechanism

J. Phys. Chem. B, Vol. 113, No. 32, 2009 11315

GlcNAc. Experimental data support a direct displacement SN2type mechanism, with Asp318 serving as the catalytic base.25 While a large number of experimental studies on β4Gal-T1 structure, kinetics, and biochemistry are available in the literature, the problem has not been carefully investigated at a theoretical level, and many mechanistic aspects have not yet been elucidated. Therefore, we have carried out a computational hybrid QM/MM investigation26 of the catalytic mechanism of β4Gal-T1 as a continuation of our effort to explore the catalytic mechanism of glycosyltransferases.27-30 Recently, we employed a similar strategy to investigate the catalytic mechanism of the Golgi glycosyltransferases,31 N-acetylglucosaminyltransferase I (GnT-I), that led to a concerted SN2-type mechanism. In addition, an activation energy of 19 kcal/mol was estimated for the proposed “intermediate” transition state model. 2. Computational Details Enzyme-Substrate Initial Structure. The initial structure, consisting of residues 130-402 of the catalytic domain of the recombinant bovine Gal-T1 with its native substrates, was obtained through the alignment of two crystal structures: (i) β4Gal-T1 + UDP-Gal + Mn2+ (PDB code 1O0R14) and (ii) β4Gal-T1 + LA + GlcNAc (PDB code 1NQI10). One chain of the 1O0R structure was taken as the enzyme in the conformation closest to the Michaelis complex. As for the positioning of the remaining substrate, GlcNAc was obtained from 1NQI. Information about the position and conservation of water molecules in this complex was also obtained. Hydrogen atoms were added and all solvent molecules, except six waters present in the active site, were removed. QM(DFT)/MM Model. The hybrid QM/MM treatment was carried out by means of the Amsterdam Density Functional software, ADF 2006.01.32-34 The entire system, consisting of the donor UDP-Gal, acceptor GlcNAc, manganese divalent cation, 6 water molecules, and 272 amino acids (4527 atoms), was partitioned into two subsystems: the QM and the MM regions. Figure 2 illustrates the QM/MM division of the active site and the numbering for some relevant atoms. The QM subsystem was optimized using the Becke-Perdew (BP) functional.35,36 The atomic orbitals were taken from the ADF library and are described as a Slater-type double-ζ basis set with a single set of polarization functions (DZP). For final singlepoint energy calculations, a larger basis set was used, described as a Slater-type triple-ζ basis set with two sets of polarization functions (TZ2P). The same basis sets were used successfully in a study of the catalytic mechanism of N-acetylglucosaminyltransferase I.31 The MM subsystem was described by the AMBER95 all-atom force field.37 The QM subsystem, containing 253 atoms (Figure 2), is formed by the acceptor and donor substrates, the metal cofactor, and the side chains of 11 amino acids involved in the enzymatic reaction, binding the substrates, or bearing a charge near the reaction center. Thus, the QM part includes (i) the UDP-Gal donor, (ii) the GlcNAc acceptor, and (iii) the divalent metal cofactor Mn2+ coordinated with (iv) one water molecule W1, (v) aspartate D254, (vi) methionine M344, and (vii) histidine H347. The acceptor sugar interacts with (ix) tyrosine Y286, (x) tyrosine Y289, (xi) glutamate E317, and (xii) the catalytic base aspartate D318. Four amino acids are present in the vicinity of diphosphate, namely, (xiii) tryptophan W314, (xiv) arginine R349, and, from the opposite side, (xv) methionine M277 and (xvi) lysine K279. A bridge between the diphosphate and the acceptor sugar is formed by (xvii) water molecule W3 and

Figure 2. Schematic representation of the QM subsystem of the active site for the QM(DFT)/MM calculations. The QM subsystem contains 253 atoms.

another (xviii) water molecule W2 that assists in binding the metal-coordinated water molecule W1 to the donor sugar. The MM region is composed of the remaining β4Gal-T1 atoms present in the crystallographic structure. Hydrogen link atoms were added to satisfy the valence of QM boundary atoms,38 always chosen as the R-carbons of amino acid residues. For our calculations, all QM and MM atoms were free to relax during the investigation of reaction mechanism. Prior to the potential energy surface calculations, a geometry optimization of the whole system was performed to acquire a refined location of the acceptor. Reaction Mechanism. The reaction that β4Gal-T1 catalyzes results in the formation of a new glycosidic linkage between the acceptor and donor, cleavage of the donor glycosidic linkage, and removal of a proton from the acceptor nucleophile. Based on our previous experience with modeling the catalytic mechanism of glycosyltransferases, the reaction mechanism was monitored by the reaction coordinate rC1-O4, defined as the distance between the anomeric carbon C1 of the donor and O4a oxygen of the acceptor hydroxyl group (Figure 2). The employed reaction coordinate represents the nucleophilic attack of the acceptor on the anomeric carbon of the donor UDP-Gal. The energy profile of a reaction path was determined by adiabatic mapping. The reaction coordinate rC1-O4 was varied by 0.2 Å increments, between 3.2 and 1.8 Å. All 13575 degrees of freedom of the system studied were optimized except the reaction coordinate. Once the energy profile was obtained, the structure of the energetic maximum was used to start the transition state (TS) search in ADF, using default methods for this type of computation. The refined TS structure was characterized by the frequency calculation on the QM part of the system. The frequency calculation procedure of the ADF program was found to be inefficient. Therefore, the calculations were performed at the DFT level of theory using the BP86

11316

J. Phys. Chem. B, Vol. 113, No. 32, 2009

Krupicˇka and Tvarosˇka TABLE 1: Relevant Geometric Parametersa and Selected Atomic Charges of Stationary Structures Obtained at the QM(DFT)/MM Level of Theory Using the BP/DZP Basis Set C1-O4a C1-O1 C1-O5 O4a-OB O4a-Ha OB-Ha C2-C1-H1 O5-C1-H1 O5-C1-C2 O-C1-O4a Q(C1) Q(O1) Q(O5) Q(O4a) Q(Ha) b

ES

TS

PC

3.34 1.498 1.385 2.449 1.087 1.397 111.895 106.220 113.664 149.404 0.488 –0.563 –0.445 –0.493 0.210

2.703 2.092 1.299 2.424 1.108 1.354 120.258 113.275 121.185 151.029 0.436 –0.675 –0.316 –0.490 0.194

1.507 3.398 1.395 2.680 1.733 1.012 110.307 111.953 116.213 148.853 0.464 –0.682 –0.443 –0.399 0.225

2FYDb 2.654 3.097 1.350 2.440

120.8 145.9

a Distances are given in angstroms; angles are given in degrees. Ref 25.

Figure 3. Active site models for the Michaelis complex (ES), transition-state (TS), and the products complex (PC) obtained using the QM(DFT)/MM method. For the sake of clarity, tryptophan W314 is not shown.

exchange-correlation functional35,36 and the LANL2DZ basis set, utilizing the GAUSSIAN03 package.39 Finally, the reaction path was traced from the TS back to reactants (ES) and forward to the product complex (PC), using a series of constrained optimizations in the ADF system. Molecular graphics images were produced using the UCSF Chimera package40 from the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco. 3. Results and Discussion The QM/MM results of a minimal energy profile suggest a direct displacement SN2-type mechanism. The reaction path led toward the Michaelis complex in one direction and to the product complex in the other. The search for further reaction channels was unsuccessful, and all attempts to locate another transition state or intermediate failed. The optimized structures of active site models in the ES, TS, and PC complexes are shown in Figure 3, and some relevant structural parameters of these structures are listed in Table 1. The Cartesian coordinates of the QM(DFT)/MM optimized stationary points are included in the Supporting Information. A geometrical representation of the TS model is given in Figure 4, and the proposed reaction mechanism based on these results is shown in Figure 5. Stationary Structures. As mentioned above, the starting model was generated upon inserting both substrates into the active site. The structure of the Michaelis complex (ES) model was obtained by optimization of this structure and from the transition state (TS) using the IRC path. The structure of ES was found to be very similar to the X-ray structures with a bound donor (PDB 1O0R) or acceptor (PDB 1NQI), respectively.10,14 The Mn2+ metal ion is coordinated by two oxygens from

Figure 4. Geometrical representation of the substrates in the transition state (TS) model, obtained using the QM(DFT)/MM method.

diphosphate, Me344, Asp254, His347, and one water molecule, which is documented by their distances of 2.11, 2.28, 2.74, 2.25, 2.36, and 2.18 Å, respectively. The corresponding distances in the crystal structure are 2.14, 2.13, 2.85, 2.23, 2.28, and 2.33 Å, respectively. The GlcNAc residue is located near the hydrophobic pocket formed by four amino acid residues Trp314, Arg359, Phe360, and Ile361. Their interactions with the N-acetylamine group of GlcNAc help to orient the acceptor in the proper orientation. The GlcNAc oxygen O4a is positioned 3.34 Å from the anomeric carbon C1 of the donor, which is a suitable distance for nucleophilic attack. It is noteworthy that the distance of the O4a oxygen from the donor ring oxygen O5 is 3.69 Å. Therefore, electrostatic repulsive interactions between these two oxygens may facilitate the transition toward the planar orientation around donor carbon C1 as it has been suggested.25 A complex network of hydrogen bonds characterizes the ES and is depicted in Figure 2. A key low-barrier hydrogen bond41 (LBHB) engages the aspartate D318, oxygen OB, and the nucleophile oxygen O4a. This fairly strong interaction is characterized by short OB · · · O4a (2.45 Å) and Ha · · · OB (1.40 Å) distances and might be critical for catalysis because the D318 carboxylate has been proposed to function as a catalytic base.25 The OB · · · Ha-O4a hydrogen bond increases nucleophilicity of the acceptor oxygen O4a and facilitates the nucleophilic attack of the O4a oxygen on the anomeric carbon C1. In the ES complex, the sugar acceptor GlcNAc interacts with Y289 (H · · · O4a ) 1.59 Å), acting as a proton acceptor, and with the catalytic base D318 (Ha · · · OB ) 1.40 Å, HO3 · · · O2 ) 1.55 Å), acting as a proton donor. The

β-1,4-Galactosyltransferase-I Mechanism

J. Phys. Chem. B, Vol. 113, No. 32, 2009 11317

Figure 5. Schematic representation of the proposed mechanism for the Gal transfer catalyzed by β4Gal-T1.

sugar part of the donor Gal interacts with Y289, acting as an H-bond donor (HO2 · · · O289 ) 1.84 Å), D252 (HO3 · · · OB ) 1.69 Å). The phosphate group is stabilized by W314 (HNE · · · OP2 ) 1.70 Å), R349 (HNH1 · · · OP1 ) 1.60 Å), and most importantly, by the water molecule W3 (HW1 · · · OP2 ) 1.78 Å), which is also bound to the acceptor sugar (HW1 · · · O6 ) 1.78 Å) and is a H-bond acceptor to K279 (HNZ · · · OW1 ) 1.79 Å). A hybrid QM(DFT)/MM investigation of the β-1,2-GlcNAc transferase GnT-I31 supported a concerted catalytic mechanism with an SN2-type transition state. A careful examination of the potential energy surface of β4GalT-1 has revealed a similar concerted pathway, where the nucleophilic attack of the acceptor oxygen O4a and the breaking of the glycosidic C1-O1 bond occur simultaneously. In the TS structure model of β4GalT-1, the forming C1-O4a bond is 2.703 Å and the breaking glycosidic bond C1-O1 is 2.092 Å. The transition state can be characterized as the “early transition state”, based on the distance between C1-O4a and C1-O1.42 While the SN2-type TS is found for both examined transferases (GnT-I and β4GalT-1), relevant distances are slightly different. Both the forming C1-O2 bond of 1.912 Å and the breaking glycosidic bond C1-O1 of 2.542 Å characterize the TS model for GnT-I.42 However, the concerted character of the TS only involves formation of a new glycosidic bond and expulsion of UDP. These two processes are accompanied by a rearrangement of the donor ring, the anomeric center becoming essentially planar, which is documented by the sum of angles around the donor C1 carbon totaling 354.7°. The C1-O5 bond shortens from 1.385 to 1.299 Å, pointing to the delocalization of the O5 lone pairs to C1, and to a partial double bond character between the ring oxygen and anomeric carbon. The calculated activation energy for this SN2-type TS is 15 kcal/mol. A translation of donor C1 carbon toward the acceptor nucleophilic O4a is characteristic for the process from ES to TS. It is interesting that the breaking of the C1-O1 bond is accompanied by a rotation of the diphosphate group. Because a similar movement was also observed in the case of GnT-I, we assume that this might be a distinguishing characteristic for an SN2-type mechanism of inverting glycosyltransferases. It is likely that glycosyltransferases use this kind of movement to break glycosidic linkage, while substrates are held by the enzyme in their positions at their active sites. In the transition state, the hydrogen Ha is not fully transferred from the O4a oxygen. However,

the OB · · · Ha-O4a hydrogen bond certainly increases nucleophilic character of the acceptor oxygen O4a and contributes in stabilizing the transition state because it becomes stronger when passing from ES to TS. This is evidenced by a slight decrease of the OB · · · O4a (2.42 Å) and Ha · · · OB (1.35 Å) distances. A negative charge evolving on the pyrophosphate group is stabilized by neighboring amino acid groups possessing H-bond donors, namely, by W314 and R349. The nature of the transition state has been proven by a vibrational frequency calculation on the QM subsystem. There have been found 11 vibrations with the lowest lying imaginary frequencies that represent a rotation of the terminal methyl group of amino acids, which is an artifact caused by using only the capped QM part. The dominant components of the next lowest lying vibration, with an imaginary frequency of -194 cm-1, is the vibration representing the reaction coordinate translation of the donor C1 toward the acceptor O4a (the forming C1-O4a bond) with simultaneous rotation of the terminal UDP phosphate group away from the donor sugar moiety (breaking the C1-O1 bond). The product complex model (PC) shows complete transfer of galactose to the acceptor. The newly formed β-glycosidic linkage is indicated by the bonding distance of 1.507 Å. The pyrophosphate part of UDP is far away from galactose, the C1-O1 distance being 3.40 Å, representing a complete cleavage of this bond. The catalytic base is protonated and the distance O4a · · · OB increases to 2.68 Å. The calculated energy for the obtained product complex is 9 kcal/mol less stable than ES. It is likely that further relaxation of the complex (see discussion below) will lower the energy. The experimental work of Qasba et al.43 clearly illustrated that the inherent mobility of a protein loop is manifested in various steps constituting the catalytic cycle proposed for β4GalT1. In fact, two of the proposed five steps involve an intrinsic motion of one part of the enzyme’s structure (a loop of 345-365 residues). Modeling the formation of the Michaelis complex and the ascent along the SN2-type channel to the transition state was quite straightforward. However, to our surprise, the determination of the products complex structure was not as straightforward as we expected. A simple optimization of structures close to the TS model always led to the ES structure, and the structure of the products complex was obtained only when optimizations started from structures with the C1 · · · O4a

11318

J. Phys. Chem. B, Vol. 113, No. 32, 2009

distance shortened to 2.2 Å. The question arises as to whether this reflects the possibility of another reaction step with a low energy barrier, for example, representing proton transfer to the catalytic base. Therefore, we have carried out constrained calculations for a set of structures with the C1 · · · O4a distance in the interval from 2.6 to 1.8 Å. The analysis revealed that the Ha proton remains bonded to the acceptor until the system achieves the TS structure, after which point the proton transfers and becomes bonded to the catalytic base D318 for the remainder of the process. This process is synchronized with an elongation of the O4a · · · OB distance from 2.42 Å in the transition state to 2.68 Å in the products complex. We emphasize again that all attempts to find another transition state or intermediate along the descent from the TS to P failed. Alternatively, the constrained minimization revealed the presence of an energetic plateau for the C1 · · · O4a distance between 2.6 to 2.4 Å. This may suggest a second explanation based on the catalytic cycle proposed for β4Gal-T1 by Qasba et al.,43 in which a release of the products is facilitated by conformational change of the loop covering the active site of β4Gal-T1 from the closed to the open conformation. We can only hypothesize that product formation and conformational change of the protein occurs simultaneously and that an adiabatic representation of the potential energy surface used is not well suited for this process. It is likely that β4Gal-T1 samples a series of thermally averaged equilibrium configurations with similar energy along the reaction coordinate, from the transition state to the products complex, and that the calculated points along the reaction coordinate represent only local minima. In such a case, the simple reaction coordinate represented by the C1 · · · O4a distance is not sufficient to properly describe the overall process. Because our primary goal of this paper was to investigate the catalytic mechanism and to determine the structure of the transition state reaction process from ES to TS, it is out of the scope of this work to further investigate this hypothesis. We are planning to employ a metadynamics method to address this question and investigate the whole catalytic cycle, including the loop motion, in the near future. The idea raised above is also supported by the captured intermediate 2FYD, in which the substrates are found in a geometry very similar to that of the transition state. In this case, however, relevant differences between the enzyme structures are observed. The most important seems to be a different orientation of ribose, especially the C6′ position and displacement of K279, which results in an inability to contribute to stabilizing the diphosphate anion. Water molecules in the active site probably also play an important role in keeping the substrates in appropriate positions and are likely involved in charge stabilization. Comparison of the β4Gal-T114 structure with the structure of 2FYD,25 where β4Gal-T1 is in complex with lactalbumin, glucose, and UDP-N-acetylgalactosamine, revealed that in the 2FYD structure a donor sugar is cleaved away from UDP, but a bond with the acceptor sugar is not formed. Thus, it was concluded25 that this structure represents an intermediate, close to the transition state of the studied reaction. We assume that interactions with lactalbumin alter the active site structure to a large extent, and therefore, the predicted TS structure and the 2FYD intermediate are of a different nature. The main differences are the following: (i) K279 is displaced by the N-acetyl moiety of the GalNAc donor, (ii) two water molecules W2 and W3 are missing, (iii) R349 is rotated outward, with a possible hydrogen bond forming with PEG4 (N · · · O distance of 2.752 Å), (iv) a water molecule is placed in the location that is occupied by the guanidine moiety of R349 in

Krupicˇka and Tvarosˇka the β4Gal-T1 structure, (v) a uridine moiety is positioned deeper in the enzyme, but the diphosphate part of the positioning is conserved, resulting in a different ribose orientation, and subsequently in the displacement of C6′ and O5′ atoms. A comparison of donor conformations revealed that the largest differences were found for the dihedral angle around C4′-C5′ bond. This angle is 13° and -122° in 2FYD and TS model, respectively, and also reflects a different orientation of ribose in both cases. In the crystal structure, the ribose residue is usually observed in the 3E conformation, whereas in the predicted TS structure model is in the 1E conformation. The dislocation of lysine K279 by the N-acetyl moiety of GalNAc disrupts the hydrogen bonding network important for the reaction process. Moreover, the lysine that stabilizes the negative charge of the diphosphate group is too far removed to interact effectively. In the case of R349, two different orientations are observed in the crystal structures.21,44 This suggests a larger flexibility of this group and that the actual orientation of the R349 residue may be affected by the enzyme microenvironment. The position of diphosphate is kept almost unchanged during the course of the reaction due to binding with the manganese cation. However, the observed rotation of the β-phosphate group appears to be associated with the movement of C6′ of ribose. Therefore, we assume that the donor sugar is probably presented in the form of an oxocarbenium ion, but is stabilized mainly by the diphosphate, which cannot move away and does not allow the reaction to proceed in 2FYD. A conformational change of β-phosphate during a cleavage of glycosidic linkage observed in 2FYD,25 GnT-I,31 and β4Gal-T1 might be characteristic for inverting glycosyltransferases. The above results pose the question of how literally the crystallographic data should be taken. It seems that for even a small change in a substrate, which is the case for 2FYD, the presence of the N-acetylamine group on O2 of the donor sugar cannot be considered as a slight perturbation as it influences the arrangement of the active site. As a result, not only is the energetic profile of the catalytic reaction altered, but the mechanism is modified, leading to a stable intermediate instead of expected products. The cause lies in the modification of the active site geometry, rather than solely on the reactivity of the substrate. The catalytic mechanism is illustrated in Figure 5 and can be described as a concerted SN2-type mechanism with synchronous bond breaking of C1-O1 and the evolving C1-O4a bond in the early transition state. This process is accompanied by the abstraction of the Ha proton from the nucleophile O4a oxygen by the catalytic base D318. The calculated barrier of 15 kcal/ mol is consistent with the experimentally observed barriers of 15-25 kcal/mol. The QM(DFT)/MM calculations indicate several ways in which the β4Gal-T1 environment is involved in the catalytic reaction. In the ES complex, the acceptor is placed in such a way that the nucleophile oxygen O4a is properly positioned for nucleophilic attack and for activation by the catalytic base D318. The O4a atom is at distance of 2.4 Å from the catalytic base OB oxygen, and 3.3 Å from the anomeric carbon C1 of the donor. The following interactions may be responsible for a stabilization of the transition state structure: (i) LBHB between O4a and OB(D318), (ii) hydrogen bond of R349 with the R-phosphate oxygen, (iii) hydrogen bond of W314 with β-phosphate, and (iv) interaction of K279 with water W3. The resulting products complex is stabilized by the hydrogen bond of K279 to the β-phosphate oxygen, which was previously bound to the water W3, thus stabilizing the diphosphate anion. Water W3 moves

β-1,4-Galactosyltransferase-I Mechanism away from the sugar acceptor and is bound preferentially to the diphosphate anion. 4. Conclusion In this paper a theoretical investigation of the catalytic mechanism of glycosyltransferase β4Gal-T1 has been carried out at the QM(DFT)/MM level using the BP functional. The obtained results suggest that the catalytic mechanism of the galactose transfer by the inverting β-1,4-galactosyltransferase proceeds along a reaction path representing the concerted mechanism with the nucleophilic attack of the acceptor oxygen O4a and the breaking of the glycosidic C1-O1 bond occurring simultaneously. The SN2-type transition state is characterized by the calculated reaction barrier of 15 kcal/mol at the QM(BP/ TZ2P)/MM(Amber95) level of theory and can be described as the “early transition state”. No evidence has been found for the presence of another transition state or intermediate. The results are in agreement with experimental data achieved by Qasba et al.43 and shed some light on the presence of the intermediate in the 2FYD crystal structure, where β4Gal-T1 is in complex with lactalbumin, glucose, and UDP-N-acetylgalactosamine.25 Acknowledgment. We thank Pradman Qasba of the Structural Glycobiology Section at the National Cancer Institute, Frederick, MD, U.S.A., for suggesting this project. This work was supported by grants from the EC under Contract MRTNCT-2006-035866, and the Slovak Research and Development Agency under Contract APVV-0607-07. This work has also benefited from the Centers of Excellence program of the Slovak Academy of Sciences (COMCHEM, Contract No. II/1/2007). Molecular graphics images were produced using the UCSF Chimera package from the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by NIH P41 RR-01081). Supporting Information Available: Calculated geometries of ES, TS, and PC; complete ref 39. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Glycoproteins and Disease; Montreuil, J., Vliegenthart, J. F. G., Schachter, H., Eds.; Elsevier Science B. V.: Amsterdam, 1996; Vol. 30. (2) Wilson, I. B. H.; Breton, C.; Imberty, A.; Tvaroska, I. In Glycoscience Chemistry and Chemical Biology; Fraser-Reid, B. O., Tatsuta, K., Thiem, J., Cote, G. L., Flitsch, S., Ito, Y., Kondo, H., Nishimura, S.-I., Yu, B., Eds.; Springer-Verlag: Berlin, Heilderberg, 2008; p 2267. (3) Lairson, L. L.; Henrissat, B.; Davies, G. J.; Withers, S. G. Annu. ReV. Biochem. 2008, 77, 521. (4) Bourne, Y.; Henrissat, B. Curr. Opin. Struct. Biol. 2001, 11, 593. (5) Liu, J.; Mushegian, A. Protein Sci. 2003, 12, 1418. (6) Masibay, A. S.; Qasba, P. K. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 5733. (7) Masibay, A. S.; Balaji, P. V.; Boeggeman, E. E.; Qasba, P. K. J. Biol. Chem. 1993, 268, 9908.

J. Phys. Chem. B, Vol. 113, No. 32, 2009 11319 (8) Boeggeman, E. E.; Balajai, P. V.; Qasba, P. K. Glycoconjugate J. 1995, 12, 865. (9) Qasba, P. K.; Ramakrishnan, B.; Shah, P. S. Glycobiology 1999, 9, 1129. (10) Ramakrishnan, B.; Qasba, P. K. J. Mol. Biol. 2001, 310, 205. (11) Ramakrishnan, B.; Shah, P. S.; Qasba, P. K. J. Biol. Chem. 2001, 276, 37665. (12) Boeggeman, E.; Qasba, P. K. Glycobiology 2002, 12, 395. (13) Qasba, P. K.; Ramakrishnan, B.; Balaji, P. V. Abstracts of Papers, 223rd National Meeting of the American Chemical Society, Orlando, FL, United States, April 7-11, 2002; American Chemical Society: Washington, DC, 2002, COMP. (14) Ramakrishnan, B.; Balaji, P. V.; Qasba, P. K. J. Mol. Biol. 2002, 318, 491. (15) Ramakrishnan, B.; Boeggeman, E.; Qasba, P. K. Biochem. Biophys. Res. Commun. 2002, 291, 1113. (16) Ramakrishnan, B.; Qasba, P. K. J. Biol. Chem. 2002, 277, 20833. (17) Boeggeman, E. E.; Ramakrishnan, B.; Qasba, P. K. Protein Exp. Purif. 2003, 30, 219. (18) Gunasekaran, K.; Ma, B.; Ramakrishnan, B.; Qasba, P. K.; Nussinov, R. Biochemistry 2003, 42, 3674. (19) Ramakrishnan, B.; Qasba, P. K. J. Biomol. Struct. Dyn. 2003, 21, 1. (20) Ramasamy, V.; Ramakrishnan, B.; Boeggeman, E.; Qasba, P. K. J. Mol. Biol. 2003, 331, 1065. (21) Ramakrishnan, B.; Boeggeman, E.; Qasba, P. K. Biochemistry 2004, 43, 12513. (22) Ramakrishnan, B.; Boeggeman, E.; Ramasamy, V.; Qasba, P. K. Curr. Opin. Struct. Biol. 2004, 14, 593. (23) Ramakrishnan, B.; Boeggeman, E.; Qasba, P. K. Biochemistry 2005, 44, 3202. (24) Ramasamy, V.; Ramakrishnan, B.; Boeggeman, E.; Ratner, D. M.; Seeberger, P. H.; Qasba, P. K. J. Mol. Biol. 2005, 353, 53. (25) Ramakrishnan, B.; Ramasamy, V.; Qasba, P. K. J. Mol. Biol. 2006, 357, 1619. (26) Senn, H. M.; Thiel, W. Angew. Chem., Int. Ed. 2009, 48, 1198. (27) Tvaroska, I.; Andre, I.; Carver, J. P. J. Am. Chem. Soc. 2000, 122, 8762. (28) Tvaroska, I.; Andre, I.; Carver, J. P. Glycobiology 2003, 13, 1. (29) Andre, I.; Tvaroska, I.; Carver, J. P. Carbohydr. Res. 2003, 338, 865. (30) Tvaroska, I. Carbohydr. Res. 2004, 339, 1007. (31) Kozmon, S.; Tvaroska, I. J. Am. Chem. Soc. 2006, 128, 16921. (32) SCM Amsterdam, 2006. (33) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca, C.; Guerra, S. J.; Gisbergen, A. v.; Snijders, J. G.; Ziegler, T. J. Comput. Chem. 2001, 22, 931. (34) Fonseca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, 391. (35) Becke, A. D. Phys. ReV. A 1988, 38, 3098. (36) Perdew, J. P. Phys. ReV. B 1986, 33, 8822. (37) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179. (38) Swart, M. Int. J. Quantum Chem. 2003, 91, 177. (39) Frisch, M. J. Gaussian03; Gaussian, Inc.: Wallingford CT, 2003. (40) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J. Comput. Chem. 2004, 25, 1605. (41) Schiøtt, B.; Iversen, B. B.; Madsen, G. K. H.; Larsen, F. K.; Bruice, T. C. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 12799. (42) Tvaroska, I. Trends Glycosci. Glycotechnol. 2005, 17, 177. (43) Qasba, P. K.; Ramakrishnan, B.; Boeggeman, E. Trends Biochem. Sci. 2005, 30, 53. (44) Ramakrishnan, B.; Ramasamy, V.; Qasba, P. K. Glycobiology 2004, 14, 1075.

JP904716T