Modeling holo-ACP:DH and holo-ACP:KR complexes of modular polyketide synthases: a docking and molecular dynamics study
© Anand and Mohanty; licensee BioMed Central Ltd. 2012
Received: 18 November 2011
Accepted: 28 May 2012
Published: 28 May 2012
Modular polyketide synthases are multifunctional megasynthases which biosynthesize a variety of secondary metabolites using various combinations of dehydratase (DH), ketoreductase (KR) and enoyl-reductase (ER) domains. During the catalysis of various reductive steps these domains act on a substrate moiety which is covalently attached to the phosphopantetheine (P-pant) group of the holo-Acyl Carrier Protein (holo-ACP) domain, thus necessitating the formation of holo-ACP:DH and holo-ACP:KR complexes. Even though three dimensional structures are available for DH, KR and ACP domains, no structures are available for DH or KR domains in complex with ACP or substrate moieties. Since Ser of holo-ACP is covalently attached to a large phosphopantetheine group, obtaining complexes involving holo-ACP by standard protein-protein docking has been a difficult task.
We have modeled the holo-ACP:DH and holo-ACP:KR complexes for identifying specific residues on DH and KR domains which are involved in interaction with ACP, phosphopantetheine and substrate moiety. A novel combination of protein-protein and protein-ligand docking has been used to first model complexes involving apo-ACP and then dock the phosphopantetheine and substrate moieties using covalent connectivity between ACP, phosphopantetheine and substrate moiety as constraints. The holo-ACP:DH and holo-ACP:KR complexes obtained from docking have been further refined by restraint free explicit solvent MD simulations to incorporate effects of ligand and receptor flexibilities. The results from 50 ns MD simulations reveal that substrate enters into a deep tunnel in DH domain while in case of KR domain the substrate binds a shallow surface exposed cavity. Interestingly, in case of DH domain the predicted binding site overlapped with the binding site in the inhibitor bound crystal structure of FabZ, the DH domain from E.Coli FAS. In case of KR domain, the substrate binding site identified by our simulations was in proximity of the known stereo-specificity determining residues.
We have modeled the holo-ACP:DH and holo-ACP:KR complexes and identified the specific residues on DH and KR domains which are involved in interaction with ACP, phosphopantetheine and substrate moiety. Analysis of the conservation profile of binding pocket residues in homologous sequences of DH and KR domains indicated that, these results can also be extrapolated to reductive domains of other modular PKS clusters.
KeywordsMolecular dynamics Protein-ligand docking Protein-protein interaction Substrate binding site Evolutionary conservation Modular polyketide synthase Dehydratase domain Ketoreductase domain
Polyketides are secondary metabolites that constitute a major class of pharmaceutically important compounds. The biosynthesis is catalyzed by multi-functional enzymes using an assembly line mechanism. The various domains in these multi-functional enzymes are arranged as modules where each module catalyzes the addition of one extender unit. Each module is constituted by a set of domains responsible for adding and modifying the extender unit. The biosynthetic intermediate is covalently attached to Acyl Carrier Protein (ACP) domain by a thioester linkage and is transferred from one catalytic site to another. The condensation of extender unit with the elongating polyketide chain is brought about by ketosynthase (KS) and Acyltransferase (AT) domain. The beta-keto product formed by action of KS and AT domain can be further modified by a combination of ketoreductase (KR), dehydratase (DH) and enoyl-reductase (ER) domains to form a hydroxyl group (KR only), a double bond (KR and DH) or a single bond (DH, ER and KR) containing moiety . Although, earlier studies [2–6] have attempted to relate the sequences of DH and KR domains to their function, till recently, no crystal structures were available for these reductive domains from modular PKSs. The only structural information was available for Type II FAS DH [7–9] and Type II PKS KR [10–15] domains and it was presumed that the Type I PKS domains also utilized similar catalytic mechanisms as the type II enzymes. The recent availability of the crystal structures of these domains from Type I PKS [5, 16–20] has opened up an opportunity to utilize structural information to understand substrate recognition by these reductive domains and the mechanism of reactions catalyzed by them.
Ketoreductases reduce the beta carboxyl to a hydroxyl group and thereby introduce a chiral center in the polyketide products. Thus, they are responsible for determining the chirality of the polyketide product . Earlier studies have identified sequence motifs predictive of hydroxyl group stereochemistry brought about by KR domains. KRs that produce a hydroxyl group of “S” stereochemistry are called as A-type KRs and have a conserved tryptophan residue while B-type KRs containing a LDD motif produce a hydroxyl group of “R” stereochemistry . The KRs have been shown to maintain stereo-control in isolation also and thus, their role in determining alpha substituent chirality was proposed . This brings out the possibility of six types of KRs with characteristic motifs. As shown by Keatinge-Clay, A1 and B1 KRs catalyze reduction but no epimerization while A2 and B2 KRs catalyze epimerization as well as reduction .
The DH and KR domains carry out their catalytic activity on the ACP bound substrate moiety. Even though, the crystal structures are available for DH and KR domains of modular PKS , no structural details are available for binding of the ACP domain to these reductive domains. The details about the interactions of ACP bound P-pant and substrate moieties with these domains are also not elucidated in details, as of now. In this study, we have made an attempt to model the holo-ACP:DH and holo-ACP:KR complexes and analyze P-pant as well as substrate binding sites on DH and KR domains to understand the mechanistic details of catalysis by these domains. In order to model holo-ACP:DH and holo-ACP:KR complexes, first apo-ACP:DH and apo-ACP:KR complexes have been modeled by protein-protein docking. This is followed by docking of P-pant and substrate moiety onto the apo-ACP:DH and apo-ACP:KR complexes. The necessary covalent bonds have been made between Ser of ACP, P-pant and substrate moiety to generate holo-ACP:DH and holo-ACP:KR complexes. In both the protein-protein as well as protein ligand docking, the solutions obtained have been filtered based on functional constraints arising from reactions catalyzed by these domains. Various recent studies have suggested that protein ligand docking followed by MD simulation studies to provide protein flexibility, can give useful information about critical residues involved in protein-ligand interactions [22–25]. The complexes obtained using docking studies by above-mentioned protocol have been further refined using MD simulations to incorporate effects of ligand and receptor flexibilities. Putative binding pocket residues of DH and KR domains involved in interactions with ACP, phosphopantetheine and substrate moiety have been identified from modeled apo-ACP:DH and apo-ACP:KR complexes and their conservation profile has also been analyzed.
Results and discussion
Modeling and analysis of substrate bound holo-ACP:DH complex
After obtaining the biologically meaningful structural model for apo-ACP:DH complex, the next step was to model the holo-ACP:DH complex by docking of P-pant moiety as described in the methods section. The total of 250 bound conformations of P-pant obtained from AutoDock were clustered using a RMSD cut off of 2 Å and this resulted in 113 clusters with different binding energy values. The central panel in Additional file 1: Figure S4 shows the binding energy values and number of conformations for these 113 clusters obtained from P-pant docking. Since, functionally meaningful P-pant conformation should form covalent bond between Ser of ACP and its thiol group should be close to the catalytic residues of DH domain, the cluster corresponding to the lowest energy and highest population might not always be biologically meaningful. Therefore, distance between Oγ atom of Ser 46 (ACP) and phosphate of P-pant as well as distance between –SH group of P-pant and Oδ atom of Asp 206 (DH) was computed for representative conformations from each of these 113 clusters. The representative conformations from each of the clusters having the values of above-mentioned distances less than 7 Å but more than 3.5 Å have been depicted in middle panel of Figure 4, while the corresponding values of distances as well as binding energy are shown in the bottom panel. The cases where the distance was less than 3.5 Å between Ser of ACP and phosphate group of P-pant showed a steric clash between them (Clusters 3 and 8 shown in lower panel of Figure 4). After removal of such solutions, the cluster 5 was chosen in terms of minimum binding energy as well as distance constraint required for formation of covalent bond. As can be seen from the middle and lower panels of Figure 4, all the solutions satisfying the functional constraints fall into the same deep cavity leading to the catalytic residues and binding energy of these solutions varies in the range of −3.19 to −5.24 kcal/mol. The Additional file 1: Figure S4 shows these two distances for all clusters (upper panel) and the relative orientation of the conformation of P-pant group from cluster 5 with respect to Ser 46 (ACP) and Asp 206 (DH) (lower panel). Therefore, docked conformation corresponding to cluster 5 was chosen for modeling of holo-ACP:DH complex by formation of required covalent bond between Ser 46 and P-pant moiety. The middle panel of Figure 4 shows in stick representation the covalently attached P-pant conformation (larger thickness and deep purple color) obtained after minimization of the complete complex containing DH, ACP and P-pant moiety from cluster 5.
The next requirement was to dock the substrate moiety onto holo-ACP:DH complex such that the carboxyl group of the substrate will be at a position to form covalent bond with the S atom in P-pant and also the beta hydroxyl group of the substrate will be in close proximity of the catalytic residues (Asp 206 and His 44) in the DH domains. As described earlier in the methods section, the substrate for DH domain was docked as two separate fragments ( Additional file 1: Figure S1) and the required covalent bonds were made after selecting suitable solutions from these two AutoDock runs. The docking of the first fragment of the substrate ( Additional file 1: Figure S5) yielded 250 conformations belonging to 32 clusters (RMSD within cluster <2 Å) with different binding energy values. Additional file 1: Figure S5a shows the number of conformations, distance between Oδ atom of Asp 206 (DH) and beta hydroxyl of the substrate and also the distance between the carboxyl carbon of the substrate fragment and S atom of P-pant for each cluster. Inset to Additional file 1: Figure S5a shows the conformation of the substrate fragment 1 from the cluster 1 showing minimum values of these two distances also. Additional file 1: Figure S5b shows similar results from docking of the second fragment and distribution of distance between Cγ-Cδ in various docked clusters. The whole complex, hence obtained, was minimized after building the required bonds between ACP, P-pant and substrate fragments to obtain the final substrate bound holo-ACP:DH complex. Additional file 1: Figure S6a shows the final energy minimized structure for the substrate bound holo-ACP:DH complex. As can be seen, the substrate moiety as well as a portion of the P-pant group enters into a deep cavity which harbors the catalytic residues of the DH domain. In order to further validate the results from the current study, the bound conformation of the substrate obtained from the current study was compared with the bound conformation of the mechanistic inhibitor 3-decynoyl-N-acetylcysteamine which has been crystallized in complex with FabA (PDB ID 1MKA), the type II FAS DH enzyme from E. coli . Additional file 1: Figure S6b shows the superposition of 1MKA on the DH domain in substrate bound holo-ACP:DH complex obtained from the current study. As can be seen, in both cases the ligands bind into the narrow tunnel of DH leading to the catalytic residues ( Additional file 1: Figure S6b) and the bound conformation of the mechanism-based inhibitor 3-decynoyl-N-acetylcysteamine is also very similar to the conformation of the bound substrate in the ACP:DH complex. Additional file 1: Figure S6b shows the complete docked substrate backbone in green while the backbone atoms of the mechanism-based inhibitor have been depicted in blue color. The only difference in the conformation lies at the α-β positions because the inhibitor covalently binds to the catalytic His residue with cis conformation of the α-β bond, while the substrate for EryDH4 binds non-covalently in trans conformation of the corresponding α-β bond ( Additional file 1: Figure S6b). It shows that the binding site obtained by docking results is same as that observed by earlier experimental studies.
Contacts of DH domain with bound substrate and ACP in holo-ACP:DH complex
Contacts with substrate
Contacts with Phosphopantetheine
Contacts with ACP
Modeling and analysis of substrate bound holo-ACP:KR complex
The modeling of substrate bound holo-ACP:KR complex was carried out using a similar protocol (Figure 3) as that used for ACP:DH complex involving modeling of apo-ACP:KR complex by protein-protein docking, generation of holo-ACP:KR complex by docking of P-pant group and finally docking of the substrate group. However, the substrate moiety was docked in a single step because of its smaller size unlike the case of docking of substrate on DH domain.
In order to model the holo-ACP:KR complex, the P-pant moiety was docked on the final apo-ACP:KR complex obtained from protein-protein docking. Docking of P-pant moiety by AutoDock on the apo-ACP:KR complex resulted in 250 bound conformations belonging to 150 different clusters (RMSD cutoff of 2 Å). The lower left panel in Additional file 1: Figure S10 shows the number of conformations as well as binding energy values for each of these 150 clusters, while the top panel shows the distance between phosphate of P-pant and Ser 46 of ACP as well as the distance of the thiol group of P-pant from the KR bound NADPH for each of these clusters. Since, functionally meaningful P-pant conformation should form covalent bond between Ser 46 of ACP and its thiol group should be close to the catalytic residues of KR domain, the cluster corresponding to the lowest energy and highest population need not be biologically meaningful. Therefore, distance between Oγ atom of Ser 46 of ACP and phosphate of P-pant as well as distance between –SH group of P-pant and NADPH was used as functional constraints for filtering functionally meaningful bound conformation for P-pant group. The solutions with the values of above mentioned distances less than 8 Å but more than 3.5 Å have been depicted in middle panel of Figure 8 while the corresponding values of distances as well as binding energy are shown in the bottom panel. As can be seen from Figure 8, all these solutions fall into the same shallow cavity leading to the catalytic residues and binding energy of these solutions varies in the range of −1.75 to −6.13 kcal/mol. Even though cluster 8 was best solution in terms of minimum binding energy as well as distances, it had a steric clash between P-pant group and catalytic Tyrosine. Thus, the next best solution cluster 14 was chosen for modeling the holo-ACP:KR complex by formation of required covalent bond between Ser 46 and P-pant moiety. The middle panel of Figure 8 shows the stick representation (larger thickness and cyan color) of the P-pant conformation after minimization of the complete complex containing KR, ACP and P-pant moiety from cluster 14. Although, in the docked conformation of the P-pant moiety, the distance between NADPH and thiol group of P-pant was around 7 Å ( Additional file 1: Figure S10), after minimization the distance was reduced to 3.6 Å. In the holo-ACP:KR complex, the P-pant group was located in the shallow cavity which also contained NADPH binding site.
Contacts of KR domain with bound substrate and ACP in holo-ACP:KR complex
Contacts with substrate
Contacts with Phosphopantetheine
Contacts with ACP
In this study, the binding of DH and KR domains to their cognate apo-ACP domains has been modeled using protein-protein docking methods. The energetically favorable solutions obtained from docking studies have been further filtered to satisfy functional constraints like proximity of catalytic Ser of ACP to active sites of DH and KR domains. The apo-ACP:DH and apo-ACP:KR complexes generated by this approach have been utilized to model holo-ACP:DH and holo-ACP:KR complexes by docking of P-pant moiety and substrate fragments. The protein-ligand docking for modeling holo-ACP complexes has also been carried out based on functional constraints of covalent linkage between Ser of ACP, P-pant and substrate moiety. Since, the protein-protein and protein-ligand docking do not incorporate protein flexibility, the holo-ACP complexes obtained from docking studies have been subjected to unrestrained molecular dynamics simulations in explicit solvent environment for a period of 20 ns. Interestingly, the holo-ACP:DH and holo-ACP:KR complexes were found to be stable even during unrestrained MD simulations.
This study gives an insight into how ACP binds to DH and KR domains and also how the substrate moieties are swung into the active sites by P-pant arm of ACP. It was found that substrate enters into a deep tunnel in case of DH domain while in case of KR domain the substrate binds in a shallow cavity exposed to the surface. Based on these modeling studies, the specific residues on DH and KR domains involved in interaction with ACP, P-pant and substrate moiety have been identified. The substrate binding site on DH domain identified from the current modeling study has been validated by comparison with the crystallographically determined bound conformation of the mechanism based inhibitor of FabA protein. Similarly, in case of KR domain, the substrate is positioned in proximity to the known sequence motifs responsible for epimerization as well as stereo-chemical control, as explained earlier. Even though, the current modeling study has been carried out for DH and KR domains from erythromycin synthase, analysis of conservation profile of binding pocket residues indicate that the results might be extrapolated to DH and KR domains of other PKS clusters also.
The DH and KR domains carry out their catalytic activity on a polyketide intermediate which is covalently attached to the P-pant moiety of the holo-ACP. Therefore, the protocol for in silico modeling of substrate bound apo-ACP:DH and apo- ACP:KR complexes involved several steps which are illustrated in the flowchart depicted in Figure 3. First, apo-ACP:DH and apo-ACP:KR complexes were generated by protein-protein docking approach using FTDOCK . Subsequently, the respective acylated P-pant moieties, the substrates for the DH and KR domains respectively, were docked to the apo-ACP:DH and apo-ACP:KR complexes by protein-ligand docking approach using AutoDock . However, the acylated P-pant groups which are cognate substrates for the DH and KR domains used in this study are large chemical moieties with too many freely rotatable bonds. Therefore, instead of docking them as a single ligand moiety, protein-ligand docking was carried out in several steps by docking the P-pant group and fragments of substrates separately as shown in Additional file 1: Figure S1. The various solutions obtained from protein-ligand docking were filtered using functional constraints like proximity of docked ligand to the catalytic residues of DH and KR domains, covalent bonding distance between phosphate of P-pant and Ser of ACP, covalent bonding distance between thiol group of P-pant and substrate fragments etc. Details of various steps involved in modeling substrate bound holo-ACP:DH and holo-ACP:KR complexes are described below.
Docking of homology models of cognate ACPs onto crystal structures of DH and KR domains
The cognate ACP for the crystal structure of DH domain of erythromycin synthase is the ACP from module 4 while ACP from module 1 of erythromycin is the cognate ACP for the crystal structure of the KR domain. Since, no crystal or NMR structures were available for these two ACPs, NMR structure (PDB ID 2JU2) of the ACP from module 2 of erythromycin PKS was used as template for modeling the structures of these cognate ACPs for DH and KR domains. Homology modeling of these ACPs was carried out using Modeller version 9v8 .
Docking of the structural models of apo ACP domains without P-pant onto the crystal structures of DH and KR domain was carried out using the protein-protein docking software FTDOCK version 2 . During the protein-protein docking, the larger DH or KR domain was treated as receptor and kept static, while the smaller ACP domain was kept mobile. The DH domain was projected onto 176 × 176 × 176 grid with a grid step of 0.696 Å and a global surface thickness of 1.3 Å. The grid size for KR domain was 192 × 192 × 192 with a grid step of 0.703 Å. Both for DH-ACP and KR-ACP docking studies, the rotation angle step was 12°, a total of 9,240 rotations were evaluated in total and three best complexes were selected from each rotational step. FTDOCK runs were carried out with electrostatic potential turned on. Finally, a total of 10000 top scoring complexes were given by FTDOCK as possible solutions. These 10000 complexes were further re-ranked using RPScore  module of FTDOCK package which uses a scoring function similar to residue based statistical pair potentials. Few hundred high scoring complexes based on positive RPScore values were further filtered based on functional or biological constraints like proximity of catalytic sites of DH or KR domains to the P-pant attachment site on the ACP domain.
Modeling of holo-ACP:DH and holo-ACP:KR complexes
In order to model complexes of holo-ACP with DH and KR domains, it was necessary to dock the phosphopantetheine (P-pant) group on the final structures of apo-ACP:DH and apo-ACP:KR complexes obtained by functional constraints based filtering of the high scoring solutions given by FTDOCK. However, in these complexes, ACP blocks the entrance to the active site pockets of DH and KR domains. Therefore, during the docking of phosphopantetheine group, ACP was removed from these complexes and docking was carried out on DH and KR domains alone. The docking of phosphopantetheine group onto the DH and KR complexes was carried out by using the program AutoDock version 4 . The coordinates for the phosphopantetheine (P-pant) moiety were obtained from the NMR structure of the holo-ACP from Spinach (PDB ID 2FVA) . The structure 2FVA contained a stearic acid bound phosphopantetheine attached to the catalytic Ser of ACP. The hydrogens were added to the ligand and Gasteiger charges  were assigned to various atoms of the ligand. During docking, all bonds other than peptide bonds were kept rotatable. The docking grid consisting of 126 × 126 × 126 points with a grid spacing of 0.375 Å was centered on the catalytic site of DH (His 44) or KR (Tyr 1813) domain and covered a large portion of DH or KR complex. Docking was carried out using the Lamarckian genetic algorithm (LGA) as conformational search method . The docking parameters were set to 27,000 generations, 25,00,000 energy evaluations, and 250 docking runs and default values were used for the other parameters. The final set of 250 receptor bound conformations for P-pant obtained from docking studies on DH and KR complexes were clustered using a cluster radius of 2 Å RMSD.
The various different P-pant bound DH and KR domains obtained from the AutoDock analysis were transformed onto the apo-ACP:DH and apo-ACP:KR complexes. This resulted in ACP:DH and ACP:KR complexes in which P-pant group was bound in different conformations and orientations. In the holo-ACP:DH and holo-ACP:KR complexes, the phosphate of the P-pant group should be covalently bonded to Ser of ACP and the terminal -SH group of P-pant should be proximal to the catalytic residues of DH or KR domains. Out of the 250 solutions obtained by docking, P-pant conformations which had the minimum distance between the S atom of P-pant and Cα of respective catalytic residues in DH/KR domains as well as between phosphate of P-pant and Ser of ACP were selected using an in-house Perl script. The P-pant bound ACP:DH and ACP:KR complexes which satisfied the above mentioned functional constraints were used to model holo-ACP:DH and holo-ACP:KR complexes by forming the required covalent bonds between the Ser of ACP and the phosphate of P-pant. The holo-ACP:DH and holo-ACP:KR complexes obtained by this approach were further energy minimized using CVFF forcefield and Insight II package.
Modeling of substrate bound holo-ACP:DH and holo-ACP:KR complexes
In order to model holo-ACP:DH and holo-ACP:KR complexes bound to their native substrates, the cognate substrate moieties for the DH and KR domains were docked onto the holo-ACP:DH and holo-ACP:KR complexes obtained by P-pant docking. Complexes were energy minimized after formation of covalent bonds, as mentioned earlier. The substrate for KR domain was a diketide while the substrate for DH domain was (2R,3R,4R,6R,7S,8S,9R)-3,7,9-trihydroxy-5-oxo-2,4,6,8 tetramethylundecanoate . The structural models for these substrate moieties were built using Biopolymer module of Insight II. The longer substrate for DH domain which consisted of 10 carbon atoms was modeled starting from the coordinates of the 18 carbon stearic acid from 2FVA and making relevant substitutions of hydrogen atoms by hydroxyl or methyl groups as per desired stereochemistry. The structures of cognate substrates generated by Insight II were also energy minimized in isolation prior to their docking on the holo-ACP:DH and holo-ACP:KR complexes.
In case of the substrates for DH domain, the number of freely rotatable bonds was larger. Therefore, the 10 carbon atom substrate was docked as two separate fragments one after another. Additional file 1: Figure S1 shows the fragments of DH substrate which were docked in two different steps. As in the case of P-pant docking, the ACP domain was removed from the ACP:DH or ACP:KR complex for facilitating access of the substrate to the catalytic pocket of DH or KR domain. However, the P-pant group was bound to the DH and KR domain and it was kept rigid during the docking of the substrate moieties for DH and KR domains. The docking was carried on using AutoDock 4 and all other parameters were same as in the case of P-pant docking. The various bound conformations of substrates or substrate fragments obtained from AutoDock 4 were analyzed in terms of their proximity to the catalytic residues of DH or KR domains and distance between the terminal carboxyl of the substrate to the –SH group of the P-pant with which the substrates should form covalent bonds. The docked conformations of the substrates which satisfied these functional constraints were selected for further analysis. After selecting the final conformation of the substrates, the P-pant and substrate fragment bound DH or KR domains were transformed onto the apo-ACP:DH or apo-ACP:KR complexes by superposition of the DH or KR domain only. This was done in order to get coordinates of ACP domain with respect to substrate and P-pant bound DH or KR domain. The required bonds between –SH group of P-pant and carboxyl group of substrate as well as bonds between two different substrate fragments of DH domains were created. Thus, the complete structure of the substrate bound holo-ACP:DH or holo-ACP:KR domain obtained by this approach was energy minimized first by CVFF forcefield of Insight II and subsequently by AMBER package .
MD simulations on substrate bound holo-ACP:DH and holo-ACP:KR complexes
The structural models of substrate bound apo-ACP:DH and apo-ACP:KR complexes obtained from protein-protein docking and protein-ligand docking contained an unusual side chain residue where catalytic Ser of ACP was covalently bonded to acylated P-pant moiety. Antechamber module of AMBER 9  was used to assign charges and other molecular mechanics forcefield parameters to this chemically modified Ser moiety ( Additional file 1: Figure S2a and Figure S2b). Coordinate and topology files for these substrate bound holo-ACP:DH and holo-ACP:KR complexes were generated using xleap module of AMBER 9. The force field used was ff03  for amino acids and TIP3P water model was used to solvate the protein. The water box extended 14 Å beyond the coordinates of the outer most atom of the protein-protein complexes along all three axes. The solvated structures were minimized using steepest descent minimization with a convergence criterion of 0.001 kcal/mole/Å as RMS gradient of potential energy. After minimization, MD simulations were carried out on the solvated protein-protein complexes using a time step of 1 fs and SHAKE  was used for constraining any bonds involving heavy atoms and hydrogen. The temperature of the system was raised to 300 K using NVT ensemble over a period of 200 ps in order to distribute the kinetic energy added into the system due to heating to 300 K among all degrees of freedom. Temperature coupling was performed using Langevin dynamics with a collision frequency 3 ps-1. The reference pressure was set to 1 atm using NPT ensemble over a further period of 200 ps. The equilibration was carried on till the density (~1 g/cm3) as well as temperature (~300 K) became stable. The pressure was scaled using isotropic position scaling with a pressure relaxation time of 1 ps. The production simulation was carried out for 50 ns for both the substrate bound holo-ACP:DH as well as holo-ACP:KR complexes using NPT ensemble. The non-bonded cutoff of 8 Å was used for van der Waals interactions. Particle Mesh Ewald (PME) summation  was used to compute long-range electrostatic interactions. The cutoff of 8 Å is used to define the range within which the direct sum computation of electrostatics will occur and beyond the cutoff reciprocal sum calculation of electrostatics is performed.
During the simulation, the coordinates were saved at an interval of 1 ps. Various analyses on the MD trajectories were performed using ptraj module of AMBER 9 as well as other in-house Perl scripts.
Calculation of the persistence of contacts during MD simulations
A set of 1000 structures were extracted from 50 ns trajectory with conformations taken at an interval of 50 ps. The residues contacting the substrate and P-pant moiety were calculated for each of these structures. The contacting residues included those where any atom of the residue lied within a cutoff distance of 6 Å from any atom of P-pant and substrate moieties. For each of the contacting residues, the percentage of structures in which the corresponding contact was present was reported.
Authors thank Director, NII for encouragement and support. SA thanks CSIR, India for award of senior research fellowship. The work has been supported by grants to NII from Department of Biotechnology (DBT), India. DM also acknowledges financial support from DBT, India under BTIS project and National Bioscience Career Development award.
- Fischbach MA, Walsh CT: Assembly-line enzymology for polyketide and nonribosomal peptide antibiotics: logic, machinery, and mechanisms. Chem Rev 2006, 106(8):3468–3496. 10.1021/cr0503097View ArticlePubMedGoogle Scholar
- Ansari MZ, Yadav G, Gokhale RS, Mohanty D: NRPS-PKS: a knowledge-based resource for analysis of NRPS/PKS megasynthases. Nucleic Acids Res 2004, 32: 405–413. 10.1093/nar/gkh359View ArticleGoogle Scholar
- Dillon SC, Bateman A: The hotdog fold: wrapping up a superfamily of thioesterases and dehydratases. BMC Bioinformatics 2004, 5: 109. 10.1186/1471-2105-5-109PubMed CentralView ArticlePubMedGoogle Scholar
- Starcevic A, Zucko J, Simunkovic J, Long PF, Cullum J, Hranueli D: ClustScan: an integrated program package for the semi-automatic annotation of modular biosynthetic gene clusters and in silico prediction of novel chemical structures. Nucleic Acids Res 2008, 36(21):6882–6892. 10.1093/nar/gkn685PubMed CentralView ArticlePubMedGoogle Scholar
- Keatinge-Clay AT: A tylosin ketoreductase reveals how chirality is determined in polyketides. Chem Biol 2007, 14(8):898–908. 10.1016/j.chembiol.2007.07.009View ArticlePubMedGoogle Scholar
- Caffrey P: Conserved amino acid residues correlating with ketoreductase stereospecificity in modular polyketide synthases. Chembiochem 2003, 4(7):654–657. 10.1002/cbic.200300581View ArticlePubMedGoogle Scholar
- Kimber MS, Martin F, Lu Y, Houston S, Vedadi M, Dharamsi A, Fiebig KM, Schmid M, Rock CO: The structure of (3R)-hydroxyacyl-acyl carrier protein dehydratase (FabZ) from Pseudomonas aeruginosa. J Biol Chem 2004, 279(50):52593–52602. 10.1074/jbc.M408105200View ArticlePubMedGoogle Scholar
- Kostrewa D, Winkler FK, Folkers G, Scapozza L, Perozzo R: The crystal structure of PfFabZ, the unique beta-hydroxyacyl-ACP dehydratase involved in fatty acid biosynthesis of plasmodium falciparum. Protein Sci 2005, 14(6):1570–1580.PubMed CentralView ArticlePubMedGoogle Scholar
- Leesong M, Henderson BS, Gillig JR, Schwab JM, Smith JL: Structure of a dehydratase-isomerase from the bacterial pathway for biosynthesis of unsaturated fatty acids: two catalytic activities in one active site. Structure 1996, 4(3):253–264. 10.1016/S0969-2126(96)00030-5View ArticlePubMedGoogle Scholar
- Korman TP, Tan YH, Wong J, Luo R, Tsai SC: Inhibition kinetics and emodin cocrystal structure of a type II polyketide ketoreductase. Biochemistry 2008, 47(7):1837–1847. 10.1021/bi7016427PubMed CentralView ArticlePubMedGoogle Scholar
- Korman TP, Hill JA, Vu TN, Tsai SC: Structural analysis of actinorhodin polyketide ketoreductase: cofactor binding and substrate specificity. Biochemistry 2004, 43(46):14529–14538. 10.1021/bi048133aPubMed CentralView ArticlePubMedGoogle Scholar
- Hadfield AT, Limpkin C, Teartasin W, Simpson TJ, Crosby J, Crump MP: The crystal structure of the actIII actinorhodin polyketide reductase: proposed mechanism for ACP and polyketide binding. Structure 2004, 12(10):1865–1875. 10.1016/j.str.2004.08.002View ArticlePubMedGoogle Scholar
- Keatinge-Clay AT, Maltby DA, Medzihradszky KF, Khosla C, Stroud RM: An antibiotic factory caught in action. Nat Struct Mol Biol 2004, 11(9):888–893. 10.1038/nsmb808View ArticlePubMedGoogle Scholar
- Javidpour P, Korman TP, Shakya G, Tsai SC: Structural and biochemical analyses of regio- and stereospecificities observed in a type II polyketide ketoreductase. Biochemistry 2011, 50(21):4638–4649. 10.1021/bi200335fPubMed CentralView ArticlePubMedGoogle Scholar
- Javidpour P, Das A, Khosla C, Tsai SC: Structural and Biochemical Studies of the hedamycin type ii polyketide ketoreductase (HedKR): molecular basis of stereo- and regiospecificities. Biochemistry 2011, 50(34):7426–7439. 10.1021/bi2006866PubMed CentralView ArticlePubMedGoogle Scholar
- Zheng J, Keatinge-Clay AT: Structural and functional analysis of C2-type ketoreductases from modular polyketide synthases. J Mol Biol 2011, 410(1):105–117.View ArticlePubMedGoogle Scholar
- Zheng J, Taylor CA, Piasecki SK, Keatinge-Clay AT: Structural and functional analysis of a-type ketoreductases from the amphotericin modular polyketide synthase. Structure 2010, 18(8):913–922.View ArticlePubMedGoogle Scholar
- Akey DL, Razelun JR, Tehranisa J, Sherman DH, Gerwick WH, Smith JL: Crystal structures of dehydratase domains from the curacin polyketide biosynthetic pathway. Structure 2010, 18(1):94–105.PubMed CentralView ArticlePubMedGoogle Scholar
- Keatinge-Clay A: Crystal structure of the erythromycin polyketide synthase dehydratase. J Mol Biol 2008, 384(4):941–953. 10.1016/j.jmb.2008.09.084PubMed CentralView ArticlePubMedGoogle Scholar
- Keatinge-Clay AT, Stroud RM: The structure of a ketoreductase determines the organization of the beta-carbon processing enzymes of modular polyketide synthases. Structure 2006, 14(4):737–748. 10.1016/j.str.2006.01.009View ArticlePubMedGoogle Scholar
- Maier T, Leibundgut M, Ban N: The crystal structure of a mammalian fatty acid synthase. Science 2008, 321(5894):1315–1322. 10.1126/science.1161269View ArticlePubMedGoogle Scholar
- Gajendrarao P, Krishnamoorthy N, Sakkiah S, Lazar P, Lee KW: Molecular modeling study on orphan human protein CYP4A22 for identification of potential ligand binding site. J Mol Graph Model 2010, 28(6):524–532. 10.1016/j.jmgm.2009.11.010View ArticlePubMedGoogle Scholar
- Bouaziz-Terrachet S, Toumi-Maouche A, Maouche B, Tairi-Kellou S: Modeling the binding modes of stilbene analogs to cyclooxygenase-2: a molecular docking study. J Mol Model 2010, 16(12):1919–1929. 10.1007/s00894-010-0679-7View ArticlePubMedGoogle Scholar
- Khare G, Gupta V, Gupta RK, Gupta R, Bhat R, Tyagi AK: Dissecting the role of critical residues and substrate preference of a fatty acyl-coa synthetase (FadD13) of Mycobacterium tuberculosis. PLoS One 2009, 4(12):e8387. 10.1371/journal.pone.0008387PubMed CentralView ArticlePubMedGoogle Scholar
- Zhao P, Liao QH, Ren CF, Wei J: Identification of ligand binding site on RXRgamma using molecular docking and dynamics methods. J Mol Model 2011, 17(6):1259–1265.View ArticlePubMedGoogle Scholar
- Gabb HA, Jackson RM, Sternberg MJ: Modelling protein docking using shape complementarity, electrostatics and biochemical information. J Mol Biol 1997, 272(1):106–120. 10.1006/jmbi.1997.1203View ArticlePubMedGoogle Scholar
- Moont G, Gabb HA, Sternberg MJ: Use of pair potentials across protein interfaces in screening predicted docked complexes. Proteins 1999, 35(3):364–373. 10.1002/(SICI)1097-0134(19990515)35:3<364::AID-PROT11>3.0.CO;2-4View ArticlePubMedGoogle Scholar
- Parris KD, Lin L, Tam A, Mathew R, Hixon J, Stahl M, Fritz CC, Seehra J, Somers WS: Crystal structures of substrate binding to Bacillus subtilis holo-(acyl carrier protein) synthase reveal a novel trimeric arrangement of molecules resulting in three active sites. Structure 2000, 8(8):883–895. 10.1016/S0969-2126(00)00178-7View ArticlePubMedGoogle Scholar
- Anand S, Prasad MV, Yadav G, Kumar N, Shehara J, Ansari MZ, Mohanty D: SBSPKS: structure based sequence analysis of polyketide synthases. Nucleic Acids Res 2010, 38: W487-W496. 10.1093/nar/gkq340PubMed CentralView ArticlePubMedGoogle Scholar
- Morris GM, Goodsell DS, Halliday RS, Huey R, Hart W, Belew RK, Olson AJ: Automated docking using a lamarckian genetic algorithm and an empirical binding free energy function. JComputChem 1998, 19: 1639–1662.Google Scholar
- Alekseyev VY, Liu CW, Cane DE, Puglisi JD, Khosla C: Solution structure and proposed domain domain recognition interface of an acyl carrier protein domain from a modular polyketide synthase. Protein Sci 2007, 16(10):2093–2107. 10.1110/ps.073011407PubMed CentralView ArticlePubMedGoogle Scholar
- Sali A, Blundell TL: Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol 1993, 234(3):779–815. 10.1006/jmbi.1993.1626View ArticlePubMedGoogle Scholar
- Zornetzer GA, Fox BG, Markley JL: Solution structures of spinach acyl carrier protein with decanoate and stearate. Biochemistry 2006, 45(16):5217–5227. 10.1021/bi052062dPubMed CentralView ArticlePubMedGoogle Scholar
- Gasteiger J, Marsili M: Iterative partial equalization of orbital electronegativity - a rapid access to atomic charges. Tetrahedron 1980, 36(22):3219–3228. 10.1016/0040-4020(80)80168-2View ArticleGoogle Scholar
- Case DA, Cheatham TE, Darden T, Gohlke H, Luo R, Merz KM, Onufriev A, Simmerling C, Wang B, Woods RJ: The amber biomolecular simulation programs. J Comput Chem 2005, 26(16):1668–1688. 10.1002/jcc.20290PubMed CentralView ArticlePubMedGoogle Scholar
- Case DA, Darden TA, Cheatham TE, Simmerling CL, Wang J, Duke RE, Luo R, Merz KM, Pearlman DA, Crowley M, et al.: AMBER 9. University of California, San Francisco; 2006.Google Scholar
- Duan Y, Wu C, Chowdhury S, Lee MC, Xiong G, Zhang W, Yang R, Cieplak P, Luo R, Lee T, et al.: A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. J Comput Chem 2003, 24(16):1999–2012. 10.1002/jcc.10349View ArticlePubMedGoogle Scholar
- Ryckaert JP, Ciccotti G, Berendsen HJC: Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J Comput Phys 1977, 23(3):327–341. 10.1016/0021-9991(77)90098-5View ArticleGoogle Scholar
- Darden TA, Pedersen LG: Molecular modeling: an experimental tool. Environ Health Perspect 1993, 101(5):410–412. 10.1289/ehp.93101410PubMed CentralView ArticlePubMedGoogle Scholar