Elucidation of the conformational free energy landscape in H.pylori LuxS and its implications to catalysis
© Bhattacharyya and Vishveshwara; licensee BioMed Central Ltd. 2010
Received: 25 March 2010
Accepted: 12 August 2010
Published: 12 August 2010
One of the major challenges in understanding enzyme catalysis is to identify the different conformations and their populations at detailed molecular level in response to ligand binding/environment. A detail description of the ligand induced conformational changes provides meaningful insights into the mechanism of action of enzymes and thus its function.
In this study, we have explored the ligand induced conformational changes in H.pylori LuxS and the associated mechanistic features. LuxS, a dimeric protein, produces the precursor (4,5-dihydroxy-2,3-pentanedione) for autoinducer-2 production which is a signalling molecule for bacterial quorum sensing. We have performed molecular dynamics simulations on H.pylori LuxS in its various ligand bound forms and analyzed the simulation trajectories using various techniques including the structure network analysis, free energy evaluation and water dynamics at the active site. The results bring out the mechanistic details such as co-operativity and asymmetry between the two subunits, subtle changes in the conformation as a response to the binding of active and inactive forms of ligands and the population distribution of different conformations in equilibrium. These investigations have enabled us to probe the free energy landscape and identify the corresponding conformations in terms of network parameters. In addition, we have also elucidated the variations in the dynamics of water co-ordination to the Zn2+ ion in LuxS and its relation to the rigidity at the active sites.
In this article, we provide details of a novel method for the identification of conformational changes in the different ligand bound states of the protein, evaluation of ligand-induced free energy changes and the biological relevance of our results in the context of LuxS structure-function. The methodology outlined here is highly generalized to illuminate the linkage between structure and function in any protein of known structure.
Enzymes perform their function by responding precisely to their cognate ligands, yet retaining their overall shape and structure. This phenomenon was beautifully formulated in classical models such as the "MWC" model of Monod  and the "KNF" model of Koshland . Such a phenomenological model, known as allostery, has played a major role in biochemistry in explaining the functioning of enzymes. At a quantitative level however, the effect of ligand binding on the enzyme can be subtle, multifold, ranging from minor local conformational change to major global changes or altering the conformational populations, which can affect the conformational free energy [3–5]. The pre-existence of an ensemble of pathways for allosteric communication has also been proposed . The concept of catalytic free energy landscape in the context of the function of enzymes has been discussed in literature [7–9]. A quantitative approach to elucidate such a landscape has been taken by Warshel's group [9–12]. The correlation of events such as protein conformational mobility and enzyme action through similar timescales has been an approach to elucidate the landscape [7, 8, 13]. Based on rigorous theoretical studies, it has been suggested that enzyme fidelity is determined by the proper pre-organization in the enzyme-substrate (ES) form  and catalysis is guided by the barrier to the chemical step in the ES complex and not affected by induced fit [10, 11] and these concepts have been extensively reviewed in . At the basic level, it is valuable to capture the conformational changes from a detailed molecular perspective. The structures of proteins bound to different ligands obtained from X-ray and NMR studies greatly enhance our understanding of the ligand-induced changes [14, 15]. The challenge has been to capture the subtle conformational changes and obtain the conformational populations in a precise manner that can help to construct a free energy landscape . Graph theoretical approaches have proved to be useful in capturing the side-chain interactions from a global perspective through the representation of protein structures in the form of protein structure networks (PSN) [16–18]. The importance of minor side-chain variations in the light of 'allostery' has also been established [4, 19]. Our previous studies on tRNA synthetases [20, 21] have shown that a combination of molecular dynamics simulations and protein structure network analysis effectively captures the allosteric paths of communication. It has also been suggested that the allosteric energy coupling in enzymes requires the consideration of the coupling with transition state energy . Furthermore, our recent studies have shown that the conformational changes described by the network of side-chain interactions and redistribution of conformational populations can be accurately mapped onto free energy changes .
In the present study, we have investigated the structure-function relationship in LuxS by performing MD simulations on the homodimeric protein LuxS from H.pylori and its different ligand bound structures, and by characterizing the ensemble of conformations through protein structure network analysis. We have also obtained the network of residues around the active sites, in the different ligand bound states of LuxS. Specifically we have identified ligand induced changes in conformation and free energy in LuxS and its different ligand bound forms using dynamically stable network parameters and essential dynamics approach in unison. Additionally we also demonstrate that the ligand-induced changes in the network of amino acid interactions in the enzyme can influence catalytic mechanisms.
LuxS is a bacterial protein, implicated in quorum sensing, which enables cross-talk across the bacterial 'societies' to bring about efficient synchronization of action as a function of cell density [25–27]. The specific signalling molecules involved are autoinducer 1 and 2  (abbreviated as AI-1 and AI-2 respectively). LuxS is involved in the biosynthetic pathway of AI-2, the universal signal for bacterial inter-species communication. In some organisms, quorum sensing by LuxS imparts a profound effect on pathogenicity by affecting toxin production/flagellar morphogenesis and hence motility and colonization. However, in some other species no direct pathogenic role has been assigned and LuxS is known to affect the metabolic processes . LuxS cleaves the thioether bond in S-Ribosylhomocysteine to produce homocysteine and DPD, the precursor molecule for AI-2 production in the S-adenosylmethionine (SAM) cycle . Such reactions are usually catalyzed by enzymes that require a redox-active cofactor. However, LuxS does not require such cofactors to catalyze this chemically difficult reaction . Previously we have probed into the structure-function relationships in LuxS (both crystallographic and modelled structures) from several bacterial species by employing the method of graph theory . In the present study, we have investigated ligand induced conformational changes in H.pylori LuxS, which provides important structural details and mechanistic insights into the catalytic cycle of this dimeric protein. It is an interesting system to target for development of broad-spectrum antibiotics as no LuxS homologues are present in the human. So its detailed investigation becomes exceedingly important.
Molecular Dynamics Simulations
Molecular dynamics simulations are performed at 300 K using AMBER9  with parm99 parameters  on five H.pylori LuxS systems, namely LuxS_apo (PDB: 1J6X), LuxS+SRH, LuxS+2SRH, LuxS+KRI, and LuxS+2KRI. The five simulations are carried out for 10 ns each in aqueous medium using TIP3P water model. The solvation box is 10Å from the farthest atom along any axis. The simulations are performed under NPT conditions. The van der Waals cutoff used is 10Å and the pressure and temperature relaxations are set to 0.5 ps-1. A time step of 2 fs is employed with the integration algorithm and the structures are stored every 1 ps.
Modeling of the ligand (SRH/KRI) bound structures from H.pylori
The crystal structure of H.pylori LuxS has been solved . However structures of H.pylori LuxS bound to its substrate S-Ribosylhomocysteine with intact ribose ring (SRH) or the 2-ketone intermediate (KRI) is not available. So we use the SRH and KRI bound LuxS structures from B.subtilis [PDB: 1JVI and 1YCL respectively] as a template to model the corresponding structures from H.pylori (LuxS+SRH-LuxS+2KRI). The orientation of the H.pylori LuxS which is suitable for complex formation with SRH/KRI is generated by superposing H.pylori LuxS on the corresponding ligand bound structures of B.subtilis LuxS (backbone rmsd is 0.742, all-atom rmsd is 0.644). The LuxS from H.pylori and B.subtilis has high sequence similarity (Score = 134, e-value = 2e-32) and identity of 46.6%. We further validated our docking by inspection of side-chain interactions involving the residues forming the active sites of LuxS in the two species. The communities (described in detail in the sub-section Network parameters associated with high connections) involving the histidines and cysteine forming the active site of LuxS has same/similar residues showing the similarities in the two ligand binding pockets [Additional file 1: Supplemental Figure SA1(a)]. We pictorially inspected the status of the ligand and the binding pocket after docking in H.pylori LuxS with reference to B.subtilis LuxS and an almost identical position is observed for the active site residues in the two species validating our docking exercise [Additional file 1: Supplemental Figure SA1(b)].
Construction of Protein Structure Network
where nij is number of distinct atom pairs between the side chains of amino acid residues i and j, which come within a distance of 4.5Å and Ni and Nj are the normalization factors for residues i and j. The pair of amino acid residues having interaction strength (Iij) greater than a user-defined cut-off (Imin) are connected by edges to give a protein structure network (PSN) graph for a given interaction strength Imin. Generally, Imins in the PSNs vary from 1% to 15%. The lower the Imin, the higher is the connectivity. An analysis of a large number of protein structures had shown that the optimal interaction strength in a protein structure is exhibited at the point the size of the largest noncovalently connected cluster undergoes a transition . This Imin which is around 2-4% for most of the protein structures (including the present system) is termed Icritical. The number of residue-residue interactions drastically reduces above this Imin. In the present study, an Imin of 2.5% is used for the evaluation of network parameters. Any pair of residues is considered to be connected if Imin is greater than 2.5%.
Dynamically Stable Hydrogen Bonds
The hydrogen bonds within the protein and between protein-ligand are generally identified in the structures obtained from X-ray crystallography. Although it is a reasonably accurate representation, some changes can be expected to take place in the solution environment. The average behaviour in the aqueous environment can be captured from the equilibrium dynamics simulations. Specifically, the hydrogen bonds (both mainchain-mainchain and mainchain-sidechain) are computed over the simulation trajectory and those present in > 50% of the snapshots are termed as dynamically stable ones and used for the analysis. The focus of this analysis here is mainly on the interactions between the ligands (SRH/KRI) and the protein.
Dynamic Cross-Correlation Maps
Dynamic Cross-correlation maps are used to identify the regions that move in or out of phase during the simulations. The construction and significance of such maps has been described in detail elsewhere. The cross-correlation coefficient varies from-1 (completely anti-correlated motion) to +1 (completely correlated motion).
In this study, the cross-correlation coefficients are computed by considering the backbone Cα atoms of amino acids in the protein using PTRAJ module in AMBER9 . The ensemble average structure is obtained from the snapshots from 0.5-10 ns.
Interface Dynamics Analysis
Protein-protein interactions generally take place through non-covalent interactions and this feature is also sensitive to environment such as the medium and the complex state of the protein. Graph theory based cluster analysis of the structures has been shown to elucidate the different types of interfaces in multimeric proteins such as lectins  and tryptophanyl-tRNA synthetase . In the present study of the dimeric protein LuxS, the interface clusters present in the snapshots from the simulations are identified (using the algorithm Depth First Search). The identified interface clusters are held together by non-covalent interactions. It has been previously shown that the clusters identified at the interaction strength Imin ≥ 6% play a significant role in stabilizing the interface . Hence we have used an Imin of 6% for the identification of interface clusters from the simulation snapshots.
Network parameters associated with high connections
In general network terminology, the parameters cliques/communities represent tightly connected regions of the network  and hubs represent highly connected points. In the context of PSN, these parameters are used to identify the rigid regions in the protein structures and to recognize the changes that take place due to the binding of ligands . In this study, the PSNs are constructed as described above for all the snapshots of the five MD simulated structures. We analyze these PSNs for k-clique communities and hubs on the basis of the definitions that follow.
A k-clique is defined as a set of k nodes (points represented by amino acids) in which each node is connected to all the other nodes. A community is defined as a union of smaller k-cliques that share node/s. According to mathematical literature, a k-clique community has been defined as the assemblage of k-cliques that can be percolated through a series of adjacent k-cliques. In the present study we have used a variation of the community definition and a k-clique community is one in which two k-cliques share k-1 or k-2 nodes. Community size is determined by the number of constituent cliques and is considered to be proportional to the compactness or packing density in the proteins.
k-Clique community finding algorithm
The community search approach employed by us is based on the algorithm proposed by Palla et al. We have used Cfinder to obtain the k-clique community from PSNs. In majority of the cases we obtain k = 3 cliques (with a few exception of k = 4 cliques) at the chosen Imin = 2.5%. k-clique communities with an overlap of k-1 nodes are obtained using Cfinder. The communities with k-2 node overlaps are obtained by manual inspection of the cliques and communities.
At a certain Imin different nodes have different number of connectivity (i.e. edges) with its neighbours. Hubs are defined as nodes connected by four or more edges to its neighbours in the PSN at the chosen value of Imin.
Dynamically Stable Cliques, Communities, and Hubs
The network parameters are considered to be dynamically stable if they are present in more than 50% of the simulation snapshots. The dynamically stable ones represent the major conformational states in the MD ensemble. The two-dimensional representation of these dynamically stable cliques and communities are drawn using Cfinder.
Essential Dynamics, Conformational Re-distribution, and Helmholtz Free Energy
Ideally the complete enumeration of the "catalytic free energy landscape" involves rigorous procedures taking into consideration the chemical barrier along the reaction coordinate [10–12]. Our present study however is limited to the investigations of the ligand induced variations in the protein conformations and the changes in the Helmholtz free energy of each system. We thus calculate discreet state information about definite points on the landscape. This involves the analysis of simulated trajectories in terms of essential dynamics and the method is briefly summarized below.
Where λ(i) is the ith eigenvalue and RCPF(n) is amount of motion associated with the subspace spanned by the first n eigenvalues. Sorting the eigenvalues in the descending order enables segregation of the conformational space in a low dimensional (essential subspace) subspace to which the major conformational fluctuations are restricted. It has been shown in previous studies that relevant conformational transitions are not detectable on other planes, and the corresponding eigenvectors are thus not suitable as conformational coordinates to describe the large conformational fluctuations [41, 42]. In our study, we have considered the two principal components with the highest eigenvalues to define the 'essential plane' for our analysis.
where Aref is the free energy of our reference cell, Aref = -RTln(pref/pTOT) and pTOT is the total number of conformations.
MD Simulations are performed on H.pylori LuxS (LuxS_apo) and its four ligand bound (S-ribosylhomocysteine in both the active (KRI) and inactive (SRH) forms bound to one or both the subunits) forms (LuxS+SRH - LuxS+2KRI) (10ns each) to obtain equilibrated structures which are analyzed as described in the subsequent sections.
Root Mean Square Deviation (RMSD) Profiles
Residue-wise RMSD (RRMSD) Profiles
In order to probe further into the regions contributing to the overall backbone RMSDs, the simulation averaged RRMSD values are evaluated for LuxS_apo-LuxS+2KRI [Additional file 1: Supplemental Figure SA2]. The RRMS deviations are maximum in LuxS_apo. The asymmetry between the two subunits is hinted at by Additional file 1: Supplemental Figure SA2 with residues from subunit A/B of a system fluctuating to different extents. Although such an analysis gives an idea of average deviation per residue over the trajectory for a particular system, it gives no clue to the correlated fluctuations of residues in the different systems in atomistic detail, an aspect, which is addressed in a later section.
Interactions at the Active Site
An important point to note is the asymmetric nature of the H-bonding pattern in the two subunits of LuxS in LuxS+2SRH and LuxS+2KRI [Figure 2(a-d)] inspite of the prediction about the homodimeric nature of the protein from its crystal structure (PDB_id:1J6X). The H-bonding pattern is highly symmetric as far as the homocysteine moiety of the ligands (SRH/KRI) is concerned. However, the asymmetry is marked in the nature of H-bonds between the ribosyl moiety (both in the open and closed forms) of the ligand and LuxS residues as highlighted in Figure 2(a-d). Particularly, in the 2-ketone intermediate bound form, LuxS+2KRI, the difference in the pattern of H-bonding around KRI in the two subunits is strikingly asymmetric. This asymmetry is further probed in our subsequent analysis of the simulation trajectories.
Study of Interface Dynamics using Cluster Analysis
Asymmetry associated with dynamically stable interface residues between the two subunits of LuxS_apo-LuxS+2KRI
Difference in participation between residues from the two subunits (in %)
No. of residues for LuxS_apo
No. of residues for LuxS+SRH
No. of residues for LuxS+2SRH
No. of residues for LuxS+KRI
No. of residues for LuxS+2KRI
The trajectories obtained from the five simulations are used to construct protein structure networks for each snapshot, which are then analyzed to probe into the dynamical properties of the network. Cliques, communities and hubs form the rigid regions in a protein structure network (PSN), and the dynamically stable ones correspond to the major conformations in the MD ensemble. We have analysed all these parameters from the simulations of LuxS_apo-LuxS+2KRI as described below.
Cliques and Communities
The disjointed pattern in unliganded LuxS with several small cliques and communities is indicative of the flexibility of the native enzyme, which is proposed to facilitate the initial approach and binding of the ligand. On binding of KRI/SRH to one subunit of LuxS, there is an increase in the overall number of cliques. More complexity is introduced in the network as is indicated by the merging of cliques to give larger communities as compared to the unliganded protein. It is worth noting that there occurs a prominent re-organization in the cliques and community pattern in the vicinity of the active site to which SRH/KRI binds (the catalytic region in one of the subunits) in both the active and inactive form of the enzyme. Also it is interesting to note that the number of cliques and communities at the dimer interface undergoes a slight increment in LuxS+SRH and LuxS+KRI and the rigidity is partially communicated to the other subunit as well, thereby making the whole system partially rigid. The bi-liganded system (LuxS+2SRH and LuxS+2KRI) exhibits a dramatic change in the community configuration, mainly around the catalytic regions and their connecting interface, as compared to the native as well as the mono-liganded state with the emergence of giant communities encapsulating the active sites and the dimer interface. These giant communities increase the rigidity around the active sites and the dimer interface of LuxS. Thus the two subunits are now held in a highly rigid architecture and we propose that the formation of such highly modular structures involving both the subunits of LuxS gives rise to a particular highly populated conformation for LuxS+2SRH/LuxS+2KRI. However, it is worth mentioning that LuxS+2KRI is asymmetric in terms of community pattern around its active sites, one being more rigid and compact than the other [Additional file 1: Supplemental Figure SA6(b)]. This is in contrast to LuxS+2SRH where both the active sites are encapsulated by a single giant community with comparable rigidity around them [Additional file 1: Supplemental Figure SA5(c)].
Comparison between the number of cliques and clique forming residues in monoliganded and biliganded states of LuxS
Hubs are indicative of localized rigidity, whereas extended rigidity in the system is captured by cliques and communities. In the activated ligand bound LuxS states, the extended rigidity is increased at the expense of majorly peripheral localized rigidity with decrease in the number of hubs.
Water dynamics around Zn2+
Variations in the water coordination of Zn2+ in LuxS_apo-LuxS+2KRI
Conformational Mapping of Essential Dynamical Modes
The essential dynamics analysis (details in Methods section) captures the important dynamical modes of the simulated protein. The plot of RCPF [Additional file 1: Supplemental Figure SA9] indicates that the top two modes accounts for about 35-52% of the overall dynamics for LuxS_apo-LuxS+2KRI. The 'essential plane' constructed from the top two modes represents the major conformational degrees of freedom and the main conformational transitions of the system under study in its various states of ligation.
Our results for the five systems indicate that there is a marked re-distribution of protein conformational ensemble on binding of ligands (both SRH/KRI) to LuxS. For LuxS_apo, the population distribution profile has one global minima contributing to only about 30% of the total conformational ensemble and a very broad peak. On addition of single SRH/KRI to LuxS, the conformational space is spread out and composed of a large number of peaks in contrast to the unliganded LuxS. It is amazing to note that on binding of the second KRI/SRH, the whole system becomes highly rigid and stable as is also indicated by a narrow population distribution profile accounting for about 45-75% of the total ensemble of conformational states. This is indicative of the stability and compactness brought into the system on binding of the second ligand.
Mapping of 'Essential Conformational Space' with Network Parameters
It is striking to note that the results obtained from the conformational analysis in terms of network parameters such as cliques, communities and hubs are in good agreement with the population distribution profiles. The minimum number of cliques and communities are found in the unliganded LuxS implying maximum flexibility, which is also shown by its population distribution profile [Figure 7(a)]. In LuxS+2SRH and LuxS+2KRI, the number of cliques and communities are maximized with giant community/communities being formed around the active sites. This is also reflected in the narrow population distribution profiles for the bi-liganded forms of LuxS enzyme [Figure 7(c,e)]. However, LuxS+SRH and LuxS+KRI are interesting cases. Here we observe increased rigidity compared to LuxS_apo. Rigidity is induced mainly in the subunit bound to the ligand. Such a conformation appears as though it is an intermediate state and the system has the conformational features of both native and biliganded states [Figure 7]. Although, in general, rigid structures give rise to one major population, these results indicate partial rigidification can give multiple populations.
We have also extracted the specific snapshots from the trajectories giving rise to a particular peak (LuxS_apo-LuxS+2KRI have given rise to 2, 4, 2, 4, and 1 major peak/s as highlighted in Additional file 1: Supplemental Figure SA10) in the population distribution profile. We have further analyzed these peaks in terms of cliques and communities [Additional file 2: Supplemental Table SB8]. These parameters beautifully capture the intricate side-chain variations and thus allow us to compare the conformations within and across simulations in molecular detail, in contrast to the conventional analysis techniques. It is evident that conformations under one peak are distinctly different from that under other peaks within a system. Further, the conformational similarity decreases between peaks from different systems, as measured by the number of common cliques.
Ligand Induced Changes in the Free Energy Landscape
Using the population distribution profile, Helmholtz free energy is calculated from equation 3[43, 44] and the free energy contour maps are plotted [Figure 7(a-e)] for all the five simulations. It is clear that the conformational space accessible to LuxS+SRH and LuxS+KRI (monoliganded forms of LuxS) is much higher with a large number of energetically close conformations than their corresponding biliganded counterparts. This may indicate that the mono-liganded system (LuxS+SRH and LuxS+KRI) is not very stable and thus not capable of catalytic function with efficacy. On the other hand, the rigid and highly modular bi-liganded state may be necessary for the functionality of the enzyme. Each system has an ensemble of conformations in equilibrium with a major conformation/s being dominant. The lowest Helmholtz free energy values for the minima in the free energy landscape for each of the five systems (LuxS_apo-LuxS+2KRI) are 10.439, 13.517, 11.843, 12.894, and 12.850 KJ/mol respectively.
In the Introduction section we have discussed about the various facets and challenges involved in understanding the role of protein conformational changes in enzyme catalysis. It has also been pointed out that allostery can be guided by enthalpic factors and/or entropic factors [3, 4]. In order to calculate these contributions from the simulations, a rigorous assessment of the associated conformational changes is a necessity. The simulation trajectory analyses such as the root mean square deviation (RMSD), residue wise RMSD and pairwise correlations between residues provide some information on the conformational fluctuations. However, minor variations at the side-chain level and perturbations due to collective influence of a set of residues are largely elusive from these analyses. Our recent studies on tryptophanyl-tRNA synthetase have shown that the ligand induced changes in the free energy landscape can be efficiently evaluated using the concepts of essential dynamics and network parameters in unison . The essential dynamics analysis enables us to represent the population distribution profile for the protein in different ligand bound states. The network parameters on the other hand enable identification of the amino acid networks that act in coherence with the shift in conformational equilibrium. Thus our method beautifully captures the intricate changes at the molecular level that is associated with the re-distribution of protein conformational states in response to ligand binding. We have applied our methodology to the bacterial quorum sensing protein LuxS from H.pylori to probe into the structure and function of this dimeric protein from a molecular perspective. We emphasize the fact that major conformational changes at the backbone level need not necessarily take place during ligand binding to a protein. Rewiring of side chain interactions are often sufficient to bring about functionally relevant changes in the rigidity/flexibility of the protein structure. Our investigation of the five systems (LuxS_apo-LuxS+2KRI) of LuxS and its complexes unravels the relationship between its structure and function. Major reorganisations are observed in the structure of LuxS in its differentially liganded forms as is evident from the analysis of dynamically stable network parameters [Figures 4, 5 and 6, Additional file 2: Supplemental Table SB8]. Below, we have addressed issues like asymmetry and co-operativity between the two subunits, the accessible conformational space, free energy changes and the dynamics of water coordinated to Zn2+ at the active site in both the active and inactive ligand bound states. Ideally longer simulations are required to obtain the full dynamics involving all possible states. However, certain conformational changes, especially at the side chain interaction level, do take place in a timescale of 10 ns simulations (on multiple complexes of LuxS) which provide sufficient data to obtain relative free energies of different accessed states. The rigorous information on free energy changes from existing techniques (like MM-PBSA and Umbrella sampling) is more expensive and here we propose an alternative and computationally less expensive method to obtain information on free energy changes subsequent to ligand binding. Based on these investigations, we have finally probed the mechanistic aspects of catalysis.
Distribution of cliques in the core (subunit A/B) and the dimer interface for LuxS_apo-LuxS+2KRI
LuxS_apo (k = 3)
LuxS+SRH (k = 3)
LuxS+KRI (k = 3)
LuxS+2KRI (k = 3)
(k = 3)
(k = 4)
Subunit A: 14
Subunit A: 21
Subunit A: 15
Subunit A: No cliques
Subunit A: 8
Subunit A: 19
Subunit B: 14
Subunit B: 11
Subunit B: 13
Subunit B: 1
Subunit B: 17
The present study has enabled us to track the ligand induced changes in the protein conformation, redistribution of the populations and associated free energy changes effectively through structure network representation. The Helmholtz free energy contour maps for the five systems portray the ligand induced changes in the free energy landscape for LuxS. Upon ligand binding, the proteins may undergo conformational changes and/or changes in the distribution of their population. Here we are able to obtain changes in the atomic description of conformations and their populations in equilibrium as a function of the bound ligand. Such an investigation provides a tool to capture the subtle conformational changes in proteins in response to ligand binding in a detailed and quantitative manner.
In summary, we have offered a novel generalized method to visualize the major conformational space associated with a protein and its various ligand bound states and simultaneously explore the molecular detail of the associated conformational transitions in terms of the dynamically stable network parameters and Helmholtz free energy changes. We have applied the method to H.pylori LuxS and its different ligand bound states to investigate the structure-function relationship. The study uncovers some of the structural events that lead to an efficient functional outcome (i.e, the catalytic action) in LuxS. Here we are able to capture the ligand induced conformational changes which includes the side-chain re-orientations and re-distribution of conformational populations. We are also able to identify the salient features resulting from the binding of the active (KRI) and the inactive (SRH) forms of S-ribosylhomocysteine to H.pylori LuxS. Thus our analysis of LuxS has illuminated upon the catalytic mechanism of action of the enzyme. Our analyses also provided a generalized tool to capture the subtle ligand induced conformational changes in any protein of known structure.
Protein Structure Network
S-Ribosylhomocysteine (ribose form)
catalytic 2-ketone intermediate
Dynamic Cross-correlation map
We acknowledge the grants from Microsoft Research, India. We are also grateful to the Department of Biotechnology (DBT), Govt. of India and the Supercomputer Education and Research Centre, Indian Institute of Science, Bangalore for computational facilities.
- Monod J, Wyman J, Changeux JP: On the nature of allosteric transitions: a plausible model. J Mol Biol 1965, 12: 88–118. 10.1016/S0022-2836(65)80285-6View ArticlePubMedGoogle Scholar
- Koshland DE, Nemethy G, Filmer D: Comparison of Experimental Binding Data and Theoretical Models in Proteins Containing Subunits. Biochemistry 1966, 5(1):365–385. 10.1021/bi00865a047View ArticlePubMedGoogle Scholar
- Tsai C, del Sol A, Nussinov R: Allostery: Absence of a change in shape does not imply that allostery is not at play. J Mol Biol 2008, 378: 1–11. 10.1016/j.jmb.2008.02.034PubMed CentralView ArticlePubMedGoogle Scholar
- Tsai C, del Sol A, Nussinov R: Protein allostery, signal transmission and dynamics: a classification scheme of allosteric mechanisms. Mol BioSyst 2009, 5(3):207–216. 10.1039/b819720bPubMed CentralView ArticlePubMedGoogle Scholar
- Gunasekaran K, Ma B, Nussinov R: Is allostery an intrinsic property of all dynamic proteins? Proteins: Structure, Function, and Bioinformatics 2004, 57: 433–443. 10.1002/prot.20232View ArticleGoogle Scholar
- Del Sol A, Tsai C, Ma B, Nussinov R: The origin of allosteric functional modulation: multiple pre-existing pathways. Structure 2009, 17: 1042–1050. 10.1016/j.str.2009.06.008PubMed CentralView ArticlePubMedGoogle Scholar
- Benkovic SJ, Hammes GG, Hammes-Schiffer S: Free-Energy Landscape of Enzyme Catalysis. Biochemistry 2008, 47(11):3317–3321. 10.1021/bi800049zView ArticlePubMedGoogle Scholar
- Goodey NM, Benkovic SJ: Allosteric regulation and catalysis emerge via a common route. Nat Chem Biol 2008, 4: 474–482. 10.1038/nchembio.98View ArticlePubMedGoogle Scholar
- Kamerlin SCL, Warshel A: At the dawn of the 21st century: Is dynamics the missing link for understanding enzyme catalysis? Proteins. 2010, 78(6):1339–1375.PubMed CentralPubMedGoogle Scholar
- Pisliakov AV, Cao J, Kamerlin SCL, Warshel A: Enzyme millisecond conformational dynamics do not catalyze the chemical step. Proc Natl Acad Sci U.S.A 2009, 106(41):17359–17364. 10.1073/pnas.0909150106PubMed CentralView ArticlePubMedGoogle Scholar
- Roca M, Messer B, Hilvert D, Warshel A: On the relationship between folding and chemical landscapes in enzyme catalysis. Proc Natl Acad Sci U.S.A 2008, 105(37):13877–13882. 10.1073/pnas.0803405105PubMed CentralView ArticlePubMedGoogle Scholar
- Xiang Y, Goodman MF, Beard WA, Wilson SH, Warshel A: Exploring the role of large conformational changes in the fidelity of DNA polymerase beta. Proteins: Structure, Function, and Bioinformatics 2008, 70(1):231–247. 10.1002/prot.21668View ArticleGoogle Scholar
- Henzler-Wildman KA, Thai V, Lei M, Ott M, Wolf-Watz M, Fenn T, Pozharski E, Wilson MA, Petsko GA, Karplus M, Hubner CG, Kern D: Intrinsic motions along an enzymatic reaction trajectory. Nature 2007, 450(7171):838–844. 10.1038/nature06410View ArticlePubMedGoogle Scholar
- Pandit J, Forman MD, Fennell KF, Dillman KS, Menniti FS: Mechanism for the allosteric regulation of phosphodiesterase 2A deduced from the X-ray structure of a near full-length construct. Proc Natl Acad Sci U.S.A 2009, 106(43):18225–18230. 10.1073/pnas.0907635106PubMed CentralView ArticlePubMedGoogle Scholar
- Popovych N, Tzeng S-R, Tonelli M, Ebright RH, Kalodimos CG: Structural basis for cAMP-mediated allosteric control of the catabolite activator protein. Proc Natl Acad Sci U.S.A 2009, 106(17):6927–6932. 10.1073/pnas.0900595106PubMed CentralView ArticlePubMedGoogle Scholar
- Brinda KV, Vishveshwara S: A Network Representation of Protein Structures: Implications for Protein Stability. Biophys J 2005, 89: 4159–4170. 10.1529/biophysj.105.064485PubMed CentralView ArticlePubMedGoogle Scholar
- Brinda KV, Mitra N, Surolia A, Vishveshwara S: Determinants of quaternary association in legume lectins. Protein Science 2004., 13: 10.1110/ps.04651004Google Scholar
- Brinda KV, Vishveshwara S: Oligomeric protein structure networks: Insights into protein-protein interactions. BMC Bioinformatics 2005., 6: 10.1186/1471-2105-6-296Google Scholar
- Tsai C-J, del Sol A, Nussinov R: Allostery: Absence of a Change in Shape Does Not Imply that Allostery Is Not at Play. J Mol Biol 2008, 378: 1–11. 10.1016/j.jmb.2008.02.034PubMed CentralView ArticlePubMedGoogle Scholar
- Ghosh A, Vishveshwara S: A study of communication pathways in methionyl-tRNA synthetase by molecular dynamics simulations and structure network analysis. Proc Natl Acad Sci U.S.A 2007, 104: 15711–15716. 10.1073/pnas.0704459104PubMed CentralView ArticlePubMedGoogle Scholar
- Ghosh A, Vishveshwara S: Variations in Clique and Community Patterns in Protein Structures during Allosteric Communication: Investigation of Dynamically Equilibrated Structures of Methionyl tRNA Synthetase Complexes. Biochemistry 2008, 47: 11398–11407. 10.1021/bi8007559View ArticlePubMedGoogle Scholar
- Xiang Y, Oelschlaeger P, Florian J, Goodman MF, Warshel A: Simulating the Effect of DNA Polymerase Mutations on Transition-State Energetics and Fidelity: Evaluating Amino Acid Group Contribution and Allosteric Coupling for Ionized Residues in Human Pol. Biochemistry 2006, 45(23):7036–7048. 10.1021/bi060147oView ArticlePubMedGoogle Scholar
- Bhattacharyya M, Ghosh A, Hansia P, Vishveshwara S: Allostery and conformational free energy changes in human tryptophanyl-tRNA synthetase from essential dynamics and structure networks. Proteins 2010, 78(3):506–517. 10.1002/prot.22573View ArticlePubMedGoogle Scholar
- Rader BA, Campagna SR, Semmelhack MF, Bassler BL, Guillemin K: The quorum sensing molecule autoinducer 2 regulates motility and flageller morphogenesis in helicobacter pylori. J. Bacteriol. 2007, 189: 6109–6117. 10.1128/JB.00246-07PubMed CentralView ArticlePubMedGoogle Scholar
- Vendeville A, Winzer K, Heurlier K, Tang CM, Hardie KR: Making 'sense' of metabolism: autoinducer-2, LuxS and pathogenic bacteria. Nature reviews 2005, 3: 383–396. 10.1038/nrmicro1146PubMedGoogle Scholar
- De Keersmaecker SCJ, Sonck K, Vanderleyden J: Let LuxS speak up in AI-2 signaling. TRENDS in Microbiol 2006, 14: 114–119. 10.1016/j.tim.2006.01.003View ArticleGoogle Scholar
- Xavier KB, Bassler BL: LuxS quorum sensing: more than just a numbers game. Current Opinion in Microbiology 2003, 6: 191–197. 10.1016/S1369-5274(03)00028-6View ArticlePubMedGoogle Scholar
- Pei D, Zhu J: Mechanism of action of S-ribosylhomocysteinase (LuxS). Current Opinion in Chemical Biology 2004, 8: 492–497. 10.1016/j.cbpa.2004.08.003View ArticlePubMedGoogle Scholar
- Bhattacharyya M, Vishveshwara S: Functional correlation of bacterial LuxS with their quaternary associations: interface analysis of the structure networks. BMC Struct.Biol. 2009, 9: 8. 10.1186/1472-6807-9-8PubMed CentralView ArticlePubMedGoogle Scholar
- Case DA, Darden TA, Cheatham TE III, Simmerling CL, Wang J, Duke RE, Luo R, Merz KM, Pearlman DA, Crowley M, Walker RC, Zhang W, Wang B, Hayik S, Roitberg A, Seabra G, Wong KF, Paesani F, Wu X, Brozell S, Tsui V, Gohlke H, Yang L, Tan C, Mongan J, Hornak V, Cui G, Beroza P, Mathews DH, Schafmeister C, Ross WS, Kollman PA: AMBER9. University of California, San Fransisco; 2006.Google Scholar
- Cheatham TE III, Cieplak P, Kollman PA: A modified version of the Cornell et al. force field with improved sugar pucker phases and helical repeat. J Biomol Struct Dyn 1999, 16: 845–862.View ArticlePubMedGoogle Scholar
- Lewis HA, Furlong EB, Laubert B, Eroshkina GA, Batiyenko Y, Adams JM, Bergseid MG, Marsh CD, Peat TS, Sanderson WE: A structural genomics approach to the study of quorum sensing: crystal structures of three LuxS orthologs. Structure 2001, 9: 527–537. 10.1016/S0969-2126(01)00613-XView ArticlePubMedGoogle Scholar
- Ruzheinikov SN, Das SK, Sedelnikova SE, Hartley A, Foster SJ, Horsburgh MJ, Cox AG, McCleod CW, Mekhalfia A, Blackburn GM: The 1.2 A Structure of a Novel Quorum-Sensing Protein, Bacillus subtilis LuxS. J Mol Biol 2001, 313: 111–122. 10.1006/jmbi.2001.5027View ArticlePubMedGoogle Scholar
- Rajan R, Zhu J, Hu X, Pei D, Bell CE: Crystal structure of S-ribosylhomocysteinase (LuxS) in complex with a catalytic 2-ketone intermediate. Biochemistry 2005, 44: 3745–3753. 10.1021/bi0477384View ArticlePubMedGoogle Scholar
- Kannan N, Vishveshwara S: Identification of side-chain clusters in protein structures by a graph spectral method. J Mol Biol 1999, 292: 441–464. 10.1006/jmbi.1999.3058View ArticlePubMedGoogle Scholar
- Brinda KV, Surolia A, Vishveshwara S: Insights into the quaternary association of proteins through structure graphs: A case study of lectins. Biochem J 2005, 391: 1–15. 10.1042/BJ20050434PubMed CentralView ArticlePubMedGoogle Scholar
- Case DA, Cheatham TE III, Darden T, Gohlke H, Luo R, Merz KM Jr, 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
- Hansia P, Ghosh A, Vishveshwara S: Ligand dependent intra and inter subunit communication in human tryptophanyl tRNA synthetase as deduced from the dynamics of structure networks. Mol Biosyst 2009, 5: 1860–1872. 10.1039/b903807hView ArticlePubMedGoogle Scholar
- Palla G, Derenyi I, Farkas I, Vicsek T: Uncovering the overlapping community structure of complex networks in nature and society. Nature 2005, 435: 814–818. 10.1038/nature03607View ArticlePubMedGoogle Scholar
- Adamcsek B, Palla G, Farkas IJ, Derenyi I, Vicsek T: CFinder: locating cliques and overlapping modules in biological networks. Bioinformatics 2006, 22: 1021–1023. 10.1093/bioinformatics/btl039View ArticlePubMedGoogle Scholar
- Amadei A, Linssen ABM, Berendsen HJC: Essential dynamics of proteins. Proteins: Structure, Function, and Genetics 1993, 17: 412–425. 10.1002/prot.340170408View ArticleGoogle Scholar
- van Aalten DMF, Findlay JBC, Amadei A, Berendsen HJC: Essential dynamics of the cellular retinol-binding protein evidence for ligand-induced conformational changes. Protein. Eng 1995, 8(11):1129–1135. 10.1093/protein/8.11.1129View ArticlePubMedGoogle Scholar
- Daidone I, Amadei A, Di Nola A: Thermodynamic and kinetic characterization of a beta-hairpin peptide in solution: An extended phase space sampling by molecular dynamics simulations in explicit water. Proteins: Structure, Function, and Bioinformatics 2005, 59: 510–518. 10.1002/prot.20427View ArticleGoogle Scholar
- Patel S, Sasidhar YU: A shorter peptide model from staphylococcal nuclease for the folding-unfolding equilibrium of a [beta]-hairpin shows that unfolded state has significant contribution from compact conformational states. J Struct Biol 2008, 164: 60–74. 10.1016/j.jsb.2008.06.003View ArticlePubMedGoogle Scholar
- Zhu J, Hu X, Wavreille A, Park J, Pei P: S-Ribosylhomocysteinase (LuxS) Is a Mononuclear Iron Protein. Biochemistry 2003, 42: 4717–4726. 10.1021/bi034289jView ArticlePubMedGoogle Scholar
- Zhu J, Hu X, Dizin E, Pei D: Catalytic Mechanism of S-Ribosylhomocysteinase (LuxS): Direct Observation of Ketone Intermediates by 13C NMR Spectroscopy. J Am Chem Soc 2003, 125(44):13379–13381. 10.1021/ja0369663View ArticlePubMedGoogle Scholar
- Zhu J, Patel R, Pei D: Catalytic mechanism of S-ribosylhomocysteinase (LuxS): stereochemical course and kinetic isotope effect of proton transfer reactions. Biochemistry 2004, 43: 10166–10172. 10.1021/bi0491088View ArticlePubMedGoogle Scholar
- Gopishetty B, Zhu J, Rajan R, Sobczak AJ, Wnuk SF, Bell CE, Pei D: Probing the Catalytic Mechanism of S-Ribosylhomocysteinase (LuxS) with Catalytic Intermediates and Substrate Analogues. J Am Chem Soc 2008, 131(3):1243–1250. 10.1021/ja808206wView ArticleGoogle Scholar