Protein Engineering vol. 16 no. 7 pp. 489-496, 2003
© 2003 Oxford University Press
In silico mutations and molecular dynamics studies on a winged bean chymotrypsin inhibitor protein
Crystallography and Molecular Biology Division, Saha Institute of Nuclear Physics, 1/AF Bidhan Nagar, Kolkata 700 064, India
1 To whom correspondence should be addressed. e-mail: jiban{at}cmb2.saha.ernet.in
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Winged bean chymotrypsin inhibitor (WCI) has an intruding residue Asn14 that plays a crucial role in stabilizing the reactive site loop conformation. This residue is found to be conserved in the Kunitz (STI) family of serine protease inhibitors. To understand the contribution of this scaffolding residue on the stability of the reactive site loop, it was mutated in silico to Gly, Ala, Ser, Thr, Leu and Val and molecular dynamics (MD) simulations were carried out on the mutants. The results of MD simulations reveal the conformational variability and range of motions possible for the reactive site loop of different mutants. The N-terminus side of the scissile bond, which is close to a ß-barrel, is conformationally less variable, while the C-terminus side, which is relatively far from any such secondary structural element, is more variable and needs stability through hydrogen-bonding interactions. The simulated structures of WCI and the mutants were docked in the peptide-binding groove of the cognate enzyme chymotrypsin and the ability to form standard hydrogen-bonding interactions at P3, P1 and P2' residues were compared. The results of the MD simulations coupled with docking studies indicate that hydrophobic residues like Leu and Val at the 14th position are disruptive for the integrity of the reactive site loop, whereas a residue like Thr, which can stabilize the C-terminus side of the scissile bond, can be predicted at this position. However, the size and charge of the Asn residue made it most suitable for the best maintenance of the integrity of the reactive site loop, explaining its conserved nature in the family.
Keywords: docking studies/in silico mutation/molecular dynamics simulation/reactive site loop/winged bean chymotrypsin inhibitor
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The winged bean chymotrypsin inhibitor (WCI) protein belongs to the Kunitz (STI) family of serine protease inhibitors and inhibits
-chymotrypsin in a 1:2 molar ratio (Kortt, 1980
). WCI consists of a single polypeptide chain with 183 amino acid residues (Mr = 20.2 kDa). The three-dimensional structure of WCI indicates that the protein has a characteristic ß-trefoil fold (Figure 1) and the two reactive sites Gln63–Phe68 and Asn38–Leu43 (designated as ‘first’ and ‘second’ sites) are situated on surface loops (Dattagupta et al., 1999). The side chain of scaffolding residue Asn14 intrudes inside the reactive site loop Gln63–Phe68 and forms a number of hydrogen bonds with the main chain and side chain atoms of loop residues that impart stability to the reactive site loop conformation. The structural comparison of WCI with the other members of the Kunitz (STI) family indicates that this particular asparagine residue is conserved in the family and plays a similar role to a spacer residue (Meester et al., 1998). Interestingly, a conserved Asn residue is also observed in the Kazal family (Laskowski et al., 1980), where the residue’s side chain plays a strikingly similar role in stabilizing the reactive site loop conformation, e.g. in PSTI, the ND1 atom of Asn33 forms a hydrogen bond with the carbonyl oxygen of the P2 residue. Chymotrypsin inhibitor-2 (CI-2, PI-1 family) is another case where the reactive site loop is stabilized through strong electrostatic interactions mediated by two parallel arginine side chains, Arg65 and Arg67 (McPhalen and James, 1987). It has been observed that the replacement of conserved spacer residues, in particular those involved in non-covalent interactions between the loop and the scaffold, can result in considerable enhancement of loop mobility (Wagner et al., 1990). Moreover, a comparison of the binding constants of ovomucoid inhibitor or eglin c with those for octapeptides derived from their binding loops indicates a scaffolding contribution of approximately –33 kJ mol–1 (Bode and Huber, 1992). The above studies indicate that the canonical conformation of the reactive site loop of serine protease inhibitors is known to play a vital role in its inhibitory activity and it should be important to understand the role played by such scaffolding amino acids in the stability of the reactive site loop. With this in mind, we wanted to explore the role of Asn14 in WCI in somewhat more detail.
|
To start with, Asn14 in WCI was mutated to Lys (N14K) and Asp (N14D) for X-ray structural studies. The crystal structure analysis of these mutants (N14K at 2.05 Å and N14D at 1.9 Å resolution) showed no major alteration in the reactive site loop conformation which was reflected in a minor difference in the dissociation constants (recombinant WCI 1.14x10–9 M, N14K 2.4x10–9 M, N14D 2.58x10–9 M) (Ravichandran et al., 2001
). In the case of N14K, the long side chain of Lys14, instead of destabilizing the loop conformation, has itself folded back, with an unusual rotamer forming hydrogen bonds with loop residues. In N14D, the hydrogen-bonding network at the loop region is maintained through a water molecule, the position of which corresponds to the side chain ND1 atom of Asn14. Therefore, the mutation of conserved Asn14 with these two polar residues indicates that the active site loop demands a hydrogen-bonding network for its stability and this need is satisfied even in mutants (N14K and N14D) either by the scaffolding residues or by the solvent molecules keeping the loop conformation intact. However, mutation of Asn14 with polar residues only may not be enough to understand its role in loop stability. So, we felt it necessary to replace this residue with a few other amino acids having different sizes and charges, which is likely to throw more light on the electrostatic nature of the loop in terms of its canonical conformation and the canonical interactions with chymotrypsin. The relationship between the protein structure and its dynamic nature, required to get the adoptability during protein–protein interaction, is a key question in biology, and is yet to be answered explicitly. Molecular dynamics (MD) simulation is an important tool to understand not only the effect of mutations on the structure, but more importantly, the dynamic nature of the protein molecule, as it can provide information about the structure at the atomic level on a reasonable time scale. Hence, in this work, we have replaced Asn14 with other residues in silico and MD simulations have been carried out on solvated native WCI as well as on its mutants. The residues having long/bulky side chains were excluded because we observed in N14K that the Lys side chain was not accommodated in the cavity and it folded back. Initially we started with N14G, where the side chain is fully truncated, followed by N14A, having just a methyl group as its side chain. Subsequently, mutations with hydrophobic residues like Val (N14V) and Leu (N14L) were also carried out and the results obtained from these mutants prompted us to replace the 14th residue with small, but not so polar, side chains like Thr (N14T) and Ser (N14S).
The resulting simulated structures of the mutants have been analysed in this paper, in terms of hydrogen-bonding interactions, dihedral angles and water orientation in the loop region. In each case, the snapshots obtained from the trajectory of the dynamics were docked at the active site of chymotrypsin to quantify the deviations in the canonical conformation in terms of the interaction energy and occurrence of conserved hydrogen bonds, as seen in the serine protease–protease inhibitor complexes.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Molecular dynamics simulation
The computation described in this paper utilized our X-ray coordinates of WCI (PDB code: 1EYL) as the starting models. The water molecules and sulphate ions were deleted from the X-ray structure. The MD simulations were performed on a SiliconGraphics O2 Workstation, using the DISCOVER-III module of Insight-II from Molecular Simulation Inc. (San Diego, CA, USA). Asn14 of WCI was mutated in silico to Gly, Ala, Ser, Thr, Val and Leu using the Insight-II Biopolymer module and the lowest energy rotamer for the mutated residues was chosen, whenever necessary. To start with, hydrogen atoms were generated at pH 7.0 and minimized, keeping all the heavy atoms fixed. Potentials were assigned using the CVFF force-field with a distance-independent dielectric constant of 4.0. No cross-terms were used in the energy expressions. For the treatment of non-bonded interactions, an atom-based method was used where non-bonded interactions were evaluated with a cut-off distance of 9.5 Å. Simulations in vacuo are known to suffer from serious artifacts, such as an excessive deviation from the native conformation. Moreover, in our present study, the understanding of the structural and dynamic properties of the solvent–protein interaction is an important task for the comprehension of protein functionality. Hence, the simulations were performed on the solvated protein molecules. Initially, a subset was defined in each case that included the following residues of the reactive site loop and neighbouring scaffold: Asp1–Gly16, Arg59–Val75. Then, the solvation of the defined subset was done, by creating a water shell of 10 Å capping using the SOAK utility of the DISCOVER-III module. The water molecules thus generated were minimized (steepest descent followed by conjugate gradient) and simulated for 30 ps, keeping the protein molecules fixed. Subsequently, the energy minimization followed by MD simulations were carried out on the solvated subset, fixing the rest of the protein molecule. The minimizations were performed using the steepest descent algorithm (down to a gradient of <100 kcal mol–1 Å–1) followed by conjugate gradient minimization (down to <10 kcal mol–1 Å–1 and then 0.001 kcal mol–1 Å–1) successively. This minimized assembly of the subset and solvent was used as a starting point for NVT (constant volume and temperature) MD to generate possible stable conformations. The simulations were carried out using the Verlet velocity algorithm (Swope and Anderson, 1982).
The major disadvantage in the MD simulations performed here is that the present computational system does not permit simulations for long enough. In order to partially overcome this disadvantage, high temperature MD simulations were carried out for WCI and all of its mutants. The kinetic energy for MD simulations was provided initially by a thermal bath at a constant temperature of 300 K and then the temperature was elevated up to 500 K. In each case, a total simulation run of 850 ps was performed with a 1 fs time step. An initial run of 250 ps at 300 K was followed by a 400 ps run at 500 K and again a 200 ps run at 300 K. After a 200 ps thermalization at 500 K, when the time dependence of the potential energy showed a stationary behaviour, a trajectory of 200 ps at 500 K followed by another 200 ps at room temperature were performed, collecting one conformation in every 0.2 ps. The last 100 ps dynamics run at 300 K was considered as the production run for the analysis.
Docking and analysis of hydrogen-bonding interactions
Twenty snapshots, with an interval of 5 ps, were collected from the last 100 ps production run and used for the analysis of the hydrogen-bonding pattern in the loop region as well as for the docking studies with chymotrypsin. Two different serine protease inhibitor complexes, STI–PPT (PBD code: 1AVW; Song and Suh, 1998) and OVO-CHY (PBD code: 1CHO; Fujinaga et al., 1987), were used as initial templates for the docking experiments. In the case of the STI–PPT template, the chymotrypsin molecule was first superposed on PPT and then the backbone of the (P3–P2') region of the reactive site loop of each snapshot was superposed with the corresponding part of STI to bring it to the peptide-binding groove of the enzyme. For OVO-CHY as the template, only the reactive site loop (P3–P2') of each snapshot was superposed on the corresponding part of OVO and the complexed chymotrypsin coordinate of OVO-CHY was used without any modification. The program MULTIDOCK (Jackson et al., 1998) was used next to refine the interface of the initial docked complex by rigid body minimization for better evaluation of the possible enzyme–inhibitor interactions.
Two different distance criteria were used to define the hydrogen-bonding pattern, one in the loop region of the inhibitors and the other in the docked enzyme–inhibitor interface. In the loop region, two O or N atoms were considered as hydrogen bonded if the donor–acceptor distance is <3.6 Å and the donor–hydrogen–acceptor angle is >120°. A protein O or N atom was considered to form a water-mediated interaction if a water O was <3.6 Å from each of them and if the angle subtended by these atoms at the water O atom was >90°. To analyse the hydrogen-bonding pattern, a 3.8 Å distance cut-off was used in the case of the enzyme–inhibitor interface. This is because a certain amount of structural changes occur during complex formation, but the program we used for docking, MULTIDOCK, performs only a rigid body motion with different side chain rotamers, without considering the flexibility in the backbone conformation of the protein during complexation. The hydrogen bond and the van der Waals contacts in the loop region and in the enzyme–inhibitor interface were monitored using the program CONTACT of the CCP4 suite (Collaborative Computational Project, 1994) and O (Jones et al., 1991).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
The objective of our MD simulation was to explore the effect of scaffold mutation on the reactive site loop conformation. Before proceeding, we felt it necessary to judge the reliability of our procedure, as done in other simulation exercises. Hence, we compared the simulated structure of WCI with its crystal structure. The reactive site loop of the simulated structure of WCI was superposed (backbone only) on that of the crystal structure, resulting in an r.m.s.d. value of 0.36 Å. Moreover, no major deviations were observed in the side chain orientation of the reactive site loop residues and Asn14 during simulation. The main chain r.m.s. fluctuations, calculated from the B-factor and obtained from dynamics, are also compared for residues around the reactive site loop of WCI (Figure 2). The agreement between the X-ray and the dynamics results are quite good, considering the involvement of the reactive site loop in crystallographic packing. MD simulation was also performed on a variant of WCI, where Asn14 was replaced by Lys (taking the lowest energy rotamer, staying inside the cavity). The simulated average structure of this mutant showed that the Lys14 side chain is folded in a similar manner (Figure 3) to that observed in the crystal structure of N14K (PDB code: 1FMZ).
|
|
As the specific hydrogen-bonding interactions of the serine protease inhibitor with the protease occur mainly through the backbone atoms of P3–P2' (Bode and Huber, 1992
), these regions of the mutants are superposed on that of WCI. No drastic changes in the loop conformation of the mutants were found and the r.m.s.d values were <1 Å, being the maximum for N14L (0.87 Å) and the minimum for N14T (0.45 Å) (Table I). The backbone superposition of the mutants on WCI (Figure 4) showed changes in the P1–P2' region, whereas the changes were minimum in the P4–P2 region.
|
|
In our previous structural studies on WCI, it was seen that the side chain amide group of Asn14 forms three major hydrogen bonds with reactive site loop residues: the Asn14 ND1 atom forms hydrogen bonds with the carbonyl oxygen atoms of P2 and P1', whereas Asn14 OD2 forms a hydrogen bond with the main chain N atom of the P4 residue along with the Asn14 N···P2'O interaction. In the reactive site loop of N14K, the side chain NZ atom of Lys14 was found to form hydrogen bonds with P4, P1' and P3'. In N14D also, Asp14 formed hydrogen bonds with P4, P2 and P1' residues, either through the side chain atoms or through a bridging water molecule. In both mutants, the main chain N atom of the 14th residue formed a strong hydrogen bond with the carbonyl oxygen atom of P2'. These observations directed us to the fact that, in all three cases described above, the hydrogen-bonding network in the loop region involves mainly the loop residues P4, P2, P1' and P2', and for them, no major alteration in the backbone conformation of the reactive site loop was observed. Analysis of the hydrogen-bonding pattern in the simulated structures of WCI reveals that two hydrogen bonds made by the side chain and main chain atoms of Asn14 with P1' and P2' are restored in all the snapshots. This observation, coupled with the backbone superposition, gives us an impression that the hydrogen bonds that include the residues P1' and P2' are important to restore the canonical conformation of the reactive site loop. With this in mind, we have analysed the hydrogen-bonding interactions in the loop region of the simulated structures of the mutants.
In N14G, where the side chain is truncated, and in N14A, having just a methyl group, hydrogen bonds between the side chain of the 14th residue and the loop residues are not possible. However, in both cases, the hydrogen bond involving the main chain N atom of the14th residue and the P2' carbonyl oxygen occurs with 100% frequency. Here, water molecule(s) play a role in the hydrogen-bonding interactions with the loop residues. In N14G, a water molecule Wat159 occupies a position that corresponds to the ND1 atom of Asn14. Using this bridging water molecule, the Ser OG atom of the P4 residue forms two water-mediated hydrogen bonds (Figure 5a) with the P2 carbonyl oxygen and the P1' OG, having 85 and 60% frequency of occurrence, respectively. Three water molecules are observed in the vicinity of the reactive site loop of N14A, which form a network of hydrogen bonds and connect the carbonyl oxygen atoms of P2, P1 and P1' (Figure 5b) with a frequency of occurrence ranging from 80 to 100%.
|
It was expected that the side chain OG atom of Thr and Ser in N14T and N14S would have a significant contribution in the hydrogen-bonding network with the loop residues. In N14T, the OG atom of Thr14 forms a hydrogen bond with carbonyl oxygen atoms of P1' and P2', whereas the main chain N atom of Thr14 forms a strong hydrogen bond with the P2' carbonyl oxygen (Figure 6a) with 100% frequency of occurrence. The hydroxyl group in N14S points in a different direction compared with that of N14T and can form two hydrogen bonds with the carbonyl oxygen atom of P2 and P1' through a bridging water molecule, Wat148 (Figure 6b), with 100% frequency of occurrence. But surprisingly, the main chain–main chain contact of the 14th residue with P2' is not observed in N14S.
|
Almost complete disruption of hydrogen-bonding interactions with the loop occurs in the case of N14L or N14V. Because of their hydrophobic nature, Leu and Val side chains are not capable of forming hydrogen bonds and also do not allow water molecules to come inside the reactive site loop, as was the case in N14G. The Leu14 N···P2'O bond is observed in N14L (Figure 7a) with 100% frequency of occurrence (average bond distance of 3.12 Å), whereas this bond (Val14 N···P2'O) is weaker in N14V (Figure 7b), having 70% frequency of occurrence and an average bond distance of 3.52 Å. The altered hydrogen-bonding pattern in mutants is likely to influence the resulting backbone conformation. Depending upon the nature of the mutated residue, the change in backbone conformation is different, but in all cases it is restricted around the scissile bond. These conformational changes presumably have some impact on enzyme–inhibitor association.
|
To quantify the change in the backbone conformation of the reactive site loop of WCI and its mutants, the dihedral angles around the scissile bond were calculated (Table II) and compared with the dihedral angle values for the canonical conformations of serine protease inhibitors (Bode and Huber, 1992
). According to Bode and Huber, the reactive site loop exhibits a quite characteristic conformation, mainly from P3 to P3', but in the case of the Kunitz (STI) family, the curvature of the reactive site loop is different from other families of serine protease inhibitors. This can be shown by measuring the distance between the C atoms of P4 and P3'. It is reported that the extension of the reactive site loop of the STI family (P4–P3') is among the smallest; 9 Å for STI, 9.2 Å for ETI and 8.8 Å for WCI compared with 11.3 Å for BPTI, 15 Å for MCTI of the squash family, 15.6 Å for PSTI of the Kazal family and 15.9 Å for eglin-C of the PI-1 family (Song and Suh, 1998). On that basis, we restricted our comparison of the dihedral angles mainly around P3–P2' with the canonical conformation of serine protease inhibitors defined by Bode and Huber (Bode and Huber, 1992). Our objective was to see whether the loop conformation falls within the range of canonical conformation along with their deviation from the conformation of native WCI. The above comparison (Table II) pointed to the fact that the 
/
angles of the P1 residue belong to the range of canonical conformation only in the case of native WCI and N14T; for other mutants, these values are out of range. Overall, deviation starts at P2 and maximum deviation is found in the P1'–P2' region. A significant deviation from canonical conformation is observed for N14L and N14V. In N14L, major deviations occur at the P1'–P2' bond along with the angle of P2 and angle of P1, whereas N14V experiences a deviation mainly in the P1–P1' region. For N14G, a substantial deviation is observed in the dihedral angles of the P1 residue, but in N14A, only a small deviation from canonical conformation is noticed at the angle of the P1' residue.
|
For serine protease–protease inhibitor complexes, residues from different positions of the reactive site loop contribute to the diverse level of specificity in the global interaction process. In this type of complex, the residues from the N-terminal side of the scissile bond, P1 and P3, form an anti-parallel beta strand with the residues Ser214 and Gly216 of the enzyme, whereas the C-terminal side residue P2' forms another hydrogen bond with Phe41 of the enzyme. The P1 carbonyl carbon is close (
2.7 Å) to the catalytic Ser195 O
(Huber et al., 1974; Marquart et al., 1983; Bode and Huber, 1992) and the P1 carbonyl oxygen always projects into the oxy-anion hole (Robertus et al., 1978) of the enzyme where it can form two hydrogen bonds with Gly193 N and Ser195 N. With this information in mind, the simulated structure of WCI was docked in the peptide-binding groove of chymotrypsin (see Materials and methods) using MULTIDOCK and the ability to form standard hydrogen-bonding interactions at P3, P1 and P2' residues was compared. As expected, all these interactions were maintained here with 100% frequency of occurrence. Likewise, to understand the correlation between the changes in loop conformation with its canonical interactions, the standard hydrogen-bonding interactions at the enzyme inhibitor interface for the mutants are quantified in terms of the frequency of occurrence (in %) and tabulated in Table III. Analysis of the interactions at the interface point to the fact that, in all the mutants, except N14L and N14V, the hydrogen bonds made by the P1 residue with chymotrypsin are well restored, whereas the interactions made by the P3 and P2' residues with the enzyme have been affected to a different extent. In N14A, N14T and N14S, the distortion in P3 and P2' is lesser. But in the case of N14G, N14L and N14V, flipping at P3 is combined with the distortion in the P1'–P2' region resulting in a removal of those hydrogen bonds.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Owing to its crucial position, the role of a conserved Asn residue (Asn14 in WCI) was emphasized in the Kunitz family of serine protease inhibitors (Onesti et al., 1991
; Meester et al., 1998; Dattagupta et al., 1999). The OD2 atom of Asn14 forms a hydrogen bond with the main chain N atom of Ser62 (P4), whereas the ND1 atom acts as hydrogen bond donor and forms two hydrogen bonds with the carbonyl oxygens of Phe64 (P2) and Ser66 (P1). In the case of N14D, Asp14 can no longer act as a hydrogen bond donor; rather, the two carbonyl oxygens of P2 and P1' exert a repulsive force on the Asp14 side chain causing a 90° rotation of the same compared with that of Asn14 (Ravichandran et al., 2001). Here, a water molecule occupied the position, corresponding to the ND1 atom of Asn14, and maintained the hydrogen-bonding interactions with loop residues. These observations point to the fact that a hydrogen-bonding network, involving loop residues and the scaffolding residue, is necessary for maintaining the stability of the reactive site loop and any minor alteration of that may be reflected in the values of their binding constants with enzymes. In our present MD study, the manifestation of mutations at Asn14 is explored in more detail, since we have replaced Asn14 with residues of different sizes and electrostatic nature. The reactive site loop of the mutants, when superposed on the simulated structure of WCI, shows more deviation in the P1–P2' region compared with the P4–P2 region, though the hydrogen-bonding interaction of the 14th residue with the P4 residue is not observed for any of these mutants. It seems that the ß-strand ending at P4, which is a part of the ß-barrel, restricts the conformational freedom immediately after it, i.e. in the P4–P2 region (Figure 1). The C-terminal side of the scissile bond is far from any such secondary structural element and the change in the hydrogen-bonding pattern in this region is likely to affect the local conformation. Hence, all the hydrogen bonds observed in the reactive site loop region, may not be of equal importance, rather the hydrogen bonds that stabilize the C-terminal side of the scissile bond may play a more important role in maintaining its canonical conformation. The hydrogen-bonding pattern in the loop region of N14T and the smaller deviation of its loop supports the above fact. Here, the OG atom of Thr14 forms two hydrogen bonds with the P1' and P2' carbonyl oxygen and the Thr14 N···P2'O bond is well restored. This may be the reason for N14T maintaining the canonical conformation like that of WCI. On the other hand, in N14S, the hydroxyl group of Ser14 is oriented in such a manner that it cannot form any direct hydrogen bond with the loop residues, rather a water molecule is observed to bridge the P2 and P1' carbonyl oxygen with the Ser14 hydroxyl group. Moreover, the absence of the hydrogen bond, connecting P2'O with Ser14 N, results in a small conformational change in the C-terminal side of the scissile bond.
The loop deformation caused in N14L and N14V is a combination of the hydrogen bond disruption and/or steric repulsion. The size of the Leu side chain is comparable with that of Asn and therefore the terminal methyl groups of the Leu14 side chain come close to the carbonyl oxygen of P2 and P1'. These hydrophobic groups of Leu14 exert a repulsive force on Ser66 (P1'), resulting in a distortion of canonical conformation. The deformation at P1–P1' of N14V is not due to any such steric effect, as the size of the Val side chain is smaller compared with Leu and in this case the deviation is mainly due to complete disruption of hydrogen bonds around the P1–P1' region.
In both N14G and N14A, the main chain contact of the 14th residue with the P2' carbonyl oxygen is present and hydrogen bonds involving the loop residues are mediated through water molecules. The difference in the loop conformation between them, however, lies in the nature of their hydrogen-bonding pattern. Two water-mediated hydrogen bonds are present in the loop region of N14G, and among them the hydrogen bond that connects P4 and P1' with 60% frequency of occurrence is the only contributor in the loop stability. A squeezing of the N14G loop, where the C distance between P3–P2' is 10.24 Å compared with 11.30 Å of WCI, may be a result of the above fact.
On the other hand, in N14A, the methyl group of Ala14, being not too big to exert any repulsive force on the loop like Leu or Val nor too small to leave a vacant space inside the loop like Gly, can partially play the role of a spacer. Moreover, the flexible C-terminal part of the scissile bond is getting stabilized through a network of hydrogen bonds involving three water molecules and the carbonyl oxygen atoms of P2, P1 and P1'. These two facts are possibly acting in tandem to stabilize the reactive site loop of N14A, whereas this is not the case in N14G, having only water molecule, Wat159, near the loop.
Analysis of the docking results of WCI and the mutants indicate that the deformations in the loop region are manifested in the canonical interactions with the cognate enzyme (Table III). All the important canonical hydrogen-bonding/hydrophobic interactions are identified in each of the snapshots of WCI, when docked with chymotrypsin. But the complete retention of the standard hydrogen-bonding pattern at the enzyme–inhibitor interface is not observed for any of these mutants. The interaction energy of the P1 residue with the enzyme, which is energetically most important in chymotrypsin-like protease inhibitor and determines its specificity, is compared for different mutants. This value, obtained from MULTIDOCK, is highest for WCI (9.2 kcal mol–1) and is comparable with N14T (8 kcal mol–1) and N14A (8.2 kcal mol–1). N14G (4.9 kcal mol–1) and N14S (5.8 kcal mol–1) show intermediate values and the values are a minimum for N14L (2.4 kcal mol–1) and N14V (2 kcal mol–1).
An examination of Figure 4 shows that the deformations in the loop region are least for N14T and N14A. N14T maintains the canonical interactions with almost 100% frequency of occurrence. N14S and N14G show the intermediate deformation, whereas the deformation is highest for N14L and N14V. In N14G, the three hydrogen bonds made by the P1 residue with the enzyme, are found with 100% frequency of occurrence and the P3···S3 interaction occurs with only 67% frequency. The absence of P2'···S2' interaction here may be an effect of the squeezing in the reactive site loop. The worst interaction with chymotrypsin has been observed for N14L and N14V. In N14L, a deviation from canonical conformation is found in the P1'–P2' region and a flipping of the P3 carbonyl oxygen (1.5 Å away from that of WCI), causes a complete disruption of the P3···S3 bond (distance 4.5 Å). Though the P2'···S2' interaction occurs here with 70% frequency, the distance of this hydrogen bond is 3.8 Å. In N14V, almost all the interactions occur with lesser frequency. The P1 carbonyl oxygen of this mutant does not even point towards the oxy-anion hole and is not in a position to form bonds with Gly193 N. Flipping of the P3 carbonyl oxygen is also seen here causing only a 15% occurrence of the P3···S3 interaction. So, it seems that the presence of these two hydrophobic residues in the 14th position causes some deformations in the reactive site loop region and the cumulative effect of these deformations results in a major disruption of enzyme–inhibitor interactions.
Conclusions
A number of MD simulations and docking studies are presented in this paper, and here we summarize the significant observations. Along with the role of Asn14, this study also helped us to understand the conformational variability and range of motions possible for the reactive site loop of different mutants. During MD simulation, we observed that the loop has an in-built stability, which is further supplemented by the scaffolding residues. The superpositioning data demonstrate that the N-terminal side of the scissile bond, which is close to ß-barrel, may not require hydrogen bonds for its stabilization, while the C-terminal side is far from any such secondary structural element and needs stability through hydrogen-bonding interactions. Along with Gly, the choice of hydrophobic residues like Val and Leu are ruled out at the 14th position of the inhibitor, as they can dramatically affect the binding site conformation and hence the canonical interactions. But a residue like Thr can be predicted here as it can stabilize the C-terminal part of the scissile bond through hydrogen bonds. Hence, it is clear that, despite its in-built stability, the loop needs a spacer of suitable size and charge for the best maintenance of its canonical conformation and, from that point of view, Asn is the most suitable one to complement the environment inside the loop.
|
Acknowledgements |
|---|
We thank Mr Satyabrata Das for computational help and many useful discussions. This work is supported by a grant (BT/PRO0139/R&D/15/11/96) from the Department of Biotechnology, Government of India.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Bode,W. and Huber,R. (1992) Eur. J. Biochem., 204, 433–451.[Web of Science][Medline]
2 Collaborative Computational Project, Number 4 (1994) Acta Crystallogr. D, 50, 760–763.[CrossRef][Medline]
Dattagupta,J.K., Podder,A., Chakrabarti,C., Sen,U., Mukhopadhyay,D., Dutta,S.K. and Singh,M. (1999) Proteins: Struct. Funct. Genet., 35, 321–331.[CrossRef][Web of Science][Medline]
Fujinaga,M., Sielecki,A.R., Read,R.J., Ardelt,W., Laskowski,M.,Jr and James,M.N.G. (1987) J. Mol. Biol., 195, 397.[CrossRef][Web of Science][Medline]
Huber,R., Kukla,D., Bode,W., Schwager,P., Bartels,K., Deisenhofer,J. and Steingemann,W. (1974) J. Mol. Biol., 89, 73–101.[CrossRef][Web of Science][Medline]
Jackson,R.M., Gabb,H.A. and Sternberg,M.J.E. (1998) J. Mol. Biol., 276, 265–285.[CrossRef][Web of Science][Medline]
Jones,T.A., Zou,J.Y., Cowan,S.W. and Kjeldgaard,M. (1991) Acta Crystallogr. A, 47, 110–119.
Kortt,A.A. (1980) Biochim. Biophys Acta, 624, 237–248.[Medline]
Laskowski,M.,Jr, Kato,I., Kohr,W.C., March,C.J. and Bogard,W.C. (1980) Protides Biol. Fluids, 28, 123–128.
Marquart,M., Walter,J., Deisenhofer,J., Bode,W. and Huber,R. (1983) Acta Crystallogr. B, 39, 480–490.[CrossRef]
McPhalen,C.A. and James,M.N.G. (1987) Biochemistry, 26, 261–269.[CrossRef][Medline]
De Meester,P., Brick,P., Lloyd,L.F., Blow,M.D. and Onesti,S. (1998) Acta Crystallogr. D, 54, 589–597.[Medline]
Onesti,S., Brick,P. and Blow,D.M. (1991) J. Mol. Biol., 217, 153–176.[CrossRef][Web of Science][Medline]
Ravichandran,S., Dasgupta,J., Chakrabarti,C., Ghosh,S., Singh,M. and Dattagupta,J.K. (2001) Protein Eng., 14, 349–357.
Robertus,J.D., Alden,R.A., Birktoft,J.J., Kraut,J., Powers,J.C. and Wileox,P.E. (1978) Biochemistry, 11, 2439–2449.
Song,H.K. and Suh,S.W. (1998) J. Mol. Biol., 275, 347–363.[CrossRef][Web of Science][Medline]
Swope,W.C. and Anderson,H.C. (1982) J. Chem. Phys., 76, 637–649.[CrossRef]
Wagner,G., Hyberts,S.G., Heinz,D.W. and Grütter,M.G. (1990) DNA Protein Complexes and Protein, Vol. 2. Adeninc Press, Guilderl, NY, pp. 93–101.
Received October 17, 2002; revised June 6, 2003; accepted June 19, 2003.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||