Unraveling the potential of intrinsically disordered proteins as ... - IGIB

Unraveling the potential of intrinsically disordered...

http://www.rsc.org/delivery/_ArticleLinking/Article...

Molecular BioSystems Journals

Molecular BioSystems

Mol. BioSyst., 2009

Advance Articles

DOI: 10.1039/b905518p

DOI: 10.1039/b905518p

Paper

Unraveling the potential of intrinsically disordered proteins as drug targets: application to Mycobacterium tuberculosis Meenakshi Anurag and Debasis Dash* G.N.R. Knowledge Center of Genome Informatics, Institute of Genomics and Integrative Biology (IGIB), CSIR, New Delhi, 110007, India. E-mail: [email protected] Received 18th March 2009, Accepted 18th May 2009 First published on the web 15th July 2009 Many eukaryotic and prokaryotic proteins remain disordered under physiological conditions and often acquire a stable secondary structure on binding to their cellular targets. Though the process of binding is still under analysis, it has been found that the flexibility of proteins can add to their functionality. This motivated us to explore intrinsically disordered proteins (IDPs) as drug targets. In silico studies have been carried out on Mycobacterium tuberculosis, which, with emergence of hyper-virulent and drug resistant strains, XDRs and MDRs, is one of the most dreaded pathogens in the modern world. Our study reports 13 IDPs as potential drug targets, and three of them—FtsW (Rv2154c), GlmU (Rv1018c) and Obg (Rv2440c)—are chosen as key proteins and are described in detail. Future applications of this method can provide new insight into understanding the molecular mechanism of IDPs and their potential role as drug targets.

Introduction Tuberculosis (TB), which affects 8 million new individuals and results in the death of approximately 2 million every year, is caused by virulent strains of Mycobacterium tuberculosis. The drug therapy in use to date is a long process and an incomplete dose can cause a relapse of the disease. In addition, approximately 500 000 individuals are infected with the MDR (multi drug resistant) strain.1 The WHO report also suggests that India and China are the countries most epidemiologically burdened by TB. Drug treatment has been available for the past 50 years, yet TB continues to increase at a significant rate.2 Despite so many years of extensive study of the bacterium, the puzzle still remains unsolved. Since little has been attained through rational structure-based drug design, we suggest that intrinsically disordered essential proteins (IDEPs) be assessed as candidate drug targets. The availability of the M. tuberculosis H37Rv whole genome sequence3 and the assignment of probable functions in the form of comprehensive resources like Tuberculist (http://genolist.pasteur.fr/TubercuList/) and SysBorgTB (http://sysborgtb.osdd.net) provide a platform for the in silico study of the sequences. M. tuberculosis, despite being one of the most widely studied pathogens, has remained a puzzle and challenge for the world of science and medicine. The emergence of XDR and MDR strains has further worsened the situation. Our study is an attempt to emphasize the role of disordered regions in essential proteins for drug targeting. Intrinsically disordered proteins are proteins with a significant or near complete lack of folded structure and an extended conformation with high intramolecular flexibility and little secondary structure.4 The lack of a folded structure provides a functional advantage through entropic chains, which depend directly on the disordered state and are thus out of reach of globular proteins.5 In IDPs, the open structure provides for a disproportionately large binding surface and multiple contact points for a protein of the given size. Recently a few drug targets having an intrinsically disordered region (IDR) at their binding interface have been reported. These regions exhibit disordered-to-ordered transitions during the interaction with their substrates. One of the most studied cases is the p53–MDM2 interaction. p53 is a vital protein, which, when induced, results in apoptosis of tumor cells. There are several factors that regulate the activity of p53, one of them being the interaction with MDM2. p53 binds with MDM2 via an intrinsically disordered interface and undergoes a structural transition. A small molecule, Nutlin, has been found to mimic the binding of p53 to MDM2 and thus competes with p53 to bind the MDM2 pocket.6 Another example is of a nine-residue peptide which interacts with p53 and stabilizes it against denaturation.7 In this study we have adopted a systematic selection protocol to identify a set of crucial and targetable proteins in M. tuberculosis. We report thirteen IDPs that are essential for the survival of the organism and can be targeted to disrupt vital biological interactions and pathways.

Results and discussion We have developed a new strategy for identifying essential IDPs involved in protein–protein interactions, which might then be studied as promising drug targets. The methodology undertaken (Fig. 1) led us to 13 IDPs, which might hold the key to target the pathogen. We have used two different approaches to predict disordered regions to increase the confidence of prediction. The study takes into account known protein–protein interactions, but interactions of the proposed drug targets should be explored further to elucidate novel pathways.

1 of 8

07/17/2009 02:29 PM



Fig. 1 Flowchart depicting the protocol adopted for proposing intrinsically disordered proteins as potential drug targets. Application of various filter criteria led to the identification of 13 potential drug targets.

Extent of disorderliness of the proteome IDPs occur across various classes of living organisms. Earlier studies reveal an average of 33.0% of eukaryotic, 4.2% of eubacteria and 2% of archeal proteins were composed of IDPs (i.e., disordered regions having continuous stretches of 30 amino acids).8 The M. tuberculosis proteome deviates from this trend, with 14.68% (586 proteins) of the proteome composed of IDPs as predicted by IUPred.9,10 This high incidence of IDPs could result from the pathogen trying to mimic the host machinery and successfully evading the defence mechanism. In order to ascertain the presence of IDPs in M. tuberculosis, we applied two different algorithms for disorder prediction —IUPred9,10 and DISpro,11 and selected only those proteins that qualified as IDPs in both the methods (Fig. 1). This approach ensured that the data set is likely to be free of any erroneous predictions. It led us to 222 proteins. This set marks its presence across various important functional classes (ESI, Fig. S1). The extent of disorderliness of these proteins was calculated as detailed in the Materials and methods section. The novelty of our approach is to highlight the importance of IDPs, which have been neglected by target-led identification approaches like rigid docking, as the disordered regions usually lack coordinates in a PDB structure. Studying IDPs might throw light on the less explored but nevertheless crucial targets. Most often IDPs participate in signaling, regulation and control events. As shown in Fig. 2, IDPs in M. tuberculosis were found to be well represented across various important functional classes including cell wall and cell processes, information pathways, intermediary metabolism pathways, and PE/PPE (proline and glutamate-rich protein family). The resources used are detailed under dataset and sources in the Materials and methods section.

Fig. 2 Percentage occurrence of proteins classified on the basis of functional classes, their representation in the genome, percentage representation of each class as IDPs, percentage of IDPs comprising the essential protein set. The high occurrence of IDPs in the PE/PPE class can be justified by the fact that these are low sequence complexity proteins and are rich in proline–glutamate and proline–proline–glutamate motifs (PE and PPE, respectively).3 Among cell wall and cell processes

2 of 8

07/17/2009 02:29 PM



proteins, IDPs are usually found in translocases and transmembrane proteins. Similarly among the information pathway-related proteins some of the ribosomal proteins, initiation factors, DNA/RNA polymerases, primase, ribonucleases, etc. have been found to have intrinsically disordered regions. Growth essentiality A proteome can be classified into essential and non-essential proteins. Essential proteins form the survivasome of the organism. We limited our study to these indispensable proteins that have disordered regions. Sassetti et al., 2003, conducted transposon site hybridization (TraSH) studies to identify proteins/genes that are required for the optimal growth of this pathogen, and identified 614 such proteins.12 We segregated such essential proteins from 222 IDPs, and labeled those as intrinsically disordered essential proteins (IDEPs). IDEP interacting partners Protein–protein interaction mesh and metabolic networks help in determining the hubs, bridges and direct interacting proteins in vital biological activities. IDPs frequently function by molecular recognition and they usually undergo a binding-induced folding transition upon binding to a suitable partner.13 They are reported to be interaction hubs in complex protein networks.14,15 It has been suggested that the assembly of protein complexes is enabled and to an extent promoted by protein disorder.14 Thus, studying protein–protein interactions is crucial in identifying hubs and choke points in networks. We limited our search by taking only those proteins for which there is experimental evidence of protein–protein interactions. This left us with a total of 21 proteins. In our study we found that a few of these proteins are immediate interacting partners that couple to well known targets. For instance, FtsW, PbpB, FtsK—all IDEPs, interact with FtsZ, a well studied target, during the divisome assembly (Fig. 3). Similarly RpoV, found to be responsible for the loss of virulence in M. bovis,16 interacts with RpoB, which has been associated largely with mutations leading to multi drug resistance.17

Fig. 3 Protein network information (STRING & literature) for the twenty one IDEPs. IDEPs are shown as red nodes.

Eliminating proteins with a human or microbiome counterpart Targeting a protein that has a counterpart in the host proteome can have adverse effects on the host and is thus unfit to be a potential drug target. Similarly there are a range of bacteria that reside within a (human) host, which have evolved with humans as symbiont. Such microbes primarily reside in human gut and oral cavity. If a target shares close similarity with these microbial proteins it will possibly have unfavorable side effects. Thus we eliminated proteins having similarity to the human proteome (FtsHRv3610c) or the microbiome (SecA1, IlvB1, SucB, RpsE, RpsC, RplD, RpsA), as targeting these proteins might result in inadmissible side effects. Thus the refined set of IDEPs that we propose as potential drug targets consists of thirteen essential proteins (Table 1). Table 1 Details regarding the function and percentage disorderliness of the thirteen proposed drug targets

3 of 8

RvID

Gene

Length

Function

IUPred (%)

DISPRO (%)

Rv0001

DnaA

507

Chromosomal replication initiator protein

9.27

12.03

07/17/2009 02:29 PM



RvID

Gene

Length

Function

IUPred (%)

DISPRO (%)

Rv1018c Rv1641

GlmU InfC

495 201

Probable UDP-N-acetylglucosamine pyrophosphorylase Probable initiation factor IF-3

11.11 22.89

6.26 14.93

Rv2154c

FtsW

524

Ftsw-like protein

15.84

11.64

Rv2163c Rv2343c

PbpB DnaG

679 639

Probable penicillin-binding membrane protein Probable DNA primase

10.90 6.89

8.25 5.32

Rv2440c

Obg

479

Probable GTP1/Obg-family GTP-binding protein

10.86

7.52

Rv2444c Rv2703

Rne SigA/RpoV

953 528

Possible ribonuclease E RNA polymerase sigma factor

45.86 41.48

16.47 21.02

Rv2916c

Ffh

525

Probable signal recognition particle protein

14.10

6.48

Rv3721c

DnaZX

578

DNA polymerase III (subunit gamma/tau)

20.07

14.36

Rv2748c Rv2839c

FtsK InfB

883 900

Possible cell division transmembrane protein Probable translation initiation factor if-2

9.51 32.11

13.02 28.11

Homologous proteins in various mycobacterial species We studied homologous proteins of the thirteen suggested drug targets to see how diverse these proteins are in context of their disorderliness. We did a comparative analysis on homologous proteins in seven different species—M. tuberculosis, Mycobacterium bovis, Mycobacterium avium, Mycobacterium leprae, Mycobacterium marinum, Mycobacterium ulcerans and Mycobacterium smegmatis. To find out the conservancy or divergence of the IDRs in homologous proteins across species, we calculated the difference in disordered index for each species with respect to the M. tuberculosis protein. d = (Um tb/Lm tb) − (Um s/Lm s) Where d is the difference in disordered index of M. tuberculosis (Um tb/Lm tb) and other mycobacterium species (Um s/Lm s). U is the length of the disordered region of a given protein and L is the total length of the corresponding protein. M. tuberculosis and M. bovis have highly conserved IDRs (ESI, Fig. S2). On the contrary there were remarkable differences between M. tuberculosis and M. smegmatis proteins. Interestingly, GlmU and InfB in M. smegmatis lack the disordered regions completely. Differential expression in the presence of four major anti-tuberculosis drugs To gain an insight into the effect of the known anti-tuberculosis drugs on these proteins, we extracted the differential expression data for the 13 proposed drug targets from target explorer.18 Data was drawn for isoniazid, ethambutol, pyrazinamide and rifampicin. In Fig. 4, the negative values (maroon) indicate down-regulation and positive values (cream) depict up-regulation of genes in presence of the corresponding drug. The data suggests that ethambutol fails to down-regulate any of the proteins in the selected set.

Fig. 4 Differential expression of the thirteen proposed drug targets in the presence of four vital anti-tuberculosis drugs— ethambutol, isoniazid, pyrazinamide and rifampicin.

Analysis of three potential drug targets The methodology that we have adopted here led us to thirteen essential proteins, which have been suggested to be important by other studies. Here we discuss three such proteins—FtsW, GlmU, Obg in details. Considering the protein disorder index between M. tuberculosis and M. smegmatis, IDR in Obg deviates the least whereas the deviation for FtsW and GlmU is considerably high, thus representing the extreme cases (ESI, Fig. S2). The sequence details of the IDRs of these proteins are shown in Table 2. Table 2 IDR details of the three key proteins shortlisted as drug targets—Rv ID, gene name, protein length (no. of amino acids), predicted disordered length (using IUPred and DISpro), location on the sequence and disordered amino acid sequences (as

4 of 8

07/17/2009 02:29 PM



predicted by IUPred)

Rv ID

Total IUPred DISpro Location Gene length IDR IDR IDR

Sequence (IUPred IDR )

Rv2154c FtsW 524 Rv1018c GlmU 495

83 55

61 31

C-terminal PRLEAFRDRKRANPQPAQTQPARKTPRTAPGQPARQMGLPPRPGSPRTADPPVRRSVHHGAG C-terminal DVPPGALAVSAGPQRNIENWVQRKRPGSPAAQASKRASEMACQQPTQPPDADQTP

Rv2440c Obg

52

36

C-terminal GEMTFDWEPQTPAGEPVAMSGRGTDPRLDSNKRVGAAERKAARSRRREHGDG

479

FtsW (Rv2154c) FtsW is a member of the SEDS (shape, elongation, divisome & sporulation) protein family. It is required to localize Pbp3 to the divisome and provides a link between FtsZ and the membrane.19 This was confirmed in mycobacterium and provided evidence for direct interactions between FtsW and FtsZ.20 It is proposed that in mycobacterium, the function of FtsW could have evolved to compensate for the absence of FtsA.21 Our methodology shortlisted FtsW as a potential drug target, with the disordered C-terminal region (Fig. 5). This region has been shown to posses the site of interaction with FtsZ.13

Fig. 5 FtsW, a potential drug target, has a disordered C-terminal region, which interacts with FtsZ thereby facilitating its interaction with penicillin-binding protein, another IDEPs. Fig. 5 also shows the role of PbpB (penicillin binding protein), another IDEP in the divisome assembly. In mycobacterium, the binding of FtsZ to the C-tail of FtsW is suggested to modulate its interactions with PBP3, which might lead to regulation of septal peptidoglycan biogenesis.13 With this supporting evidence, we propose that the dynamic and flexible nature of the IDR of FtsW can be treated as a potential drug target. GlmU (Rv1018c) GlmU is a bifunctional protein comprising an uridyl-transfer domain at the N-terminus and an acetyl-transfer domain towards the C-terminus. The functionality of both the domains is reported to be independent.22 Acetyl-transfer occurs prior to uridyl-transfer and is a vital step in peptidoglycan synthesis (Fig. 6). As per the IUPred predictions, the C-terminal region of the protein is a 55 amino acid long IDR (Table 2), which is also found to be conserved in M. bovis and M. tuberculosis CDC1551, but not in M. ulceran, M. smegmatis, etc. (NCBI BLAST). The crystal structure of Escherichia coli GlmU with both active sites containing the natural ligand (CoA, UD1) has been reported earlier (PDB: 1HV9).23 Recently, a partial structure for M. tuberculosis was solved (PDB: 2QKX). This structure, in accord with our analysis, lacks the coordinates for the IDR. Studies in M. smegmatis have shown that complete removal of GlmU or a reduction in its activity leads to non-viable cells, which have deformed shapes—stubby and rounded, and even lyse in some cases.24

Fig. 6 Cellular process involved in peptidoglycan synthesis (a vital component of the mycobacterial cell wall). The N-terminal region of GlmU is responsible for uridyl-transfer and the C-terminal region for acetyl-transfer. The C-terminal region is the IDR and it follows the active site for acetyl CoA binding. Inhibiting GlmU activity can be crucial in distorting the protective mycobacterial cell wall. The biological function of acetyltransfer carried out by the protein is initiated by the trimerization of the protein. Inhibiting the trimer formation or competitive binding of a small compound to the acetyl CoA binding site will hamper the peptidoglycan synthesis. We suggest that the IDR of the protein aids in stabilization of the complex by acting as a clamp and creating a complementary pocket for binding of acetyl CoA. Structural analysis of the GlmU protein in the trimeric state possessing the IDR holds the key to understanding its interaction

5 of 8

07/17/2009 02:29 PM



with cellular targets and the effect of changing environmental conditions such as pH, temperature, etc. will help in understanding the molecular level interactions and further dissections of the complex reactions carried out by this remarkable enzyme. Obg (Rv2440c) Obg belongs to the family of GTP-binding proteins. In E. coli, Obg acts as a signalling protein that controls replication, translation and morphological development. Obg might reveal the existence of a new pathway of replication control and connections between replication, protein synthesis and cellular differentiation. In Bacillus subtilis the C-terminal region consists of a TGS domain,25 which is associated with stringent stress response and ppGpp biology in bacterial cells.26 This protein is speculated to check intracellular GTP levels and act as a switch to promote growth when bound to GTP but not GDP. Obg in B. subtilis is believed to act as a regulatory switch and have several targets, allowing it to carry out diverse functions.27 According to IUPred prediction, the C-terminus of Obg in M. tuberculosis has a 52 amino acid-long IDR (Table 2). ScanProsite (www.expasy.ch)28 predicts that residues 459–474 comprise the bipartite nuclear localization signal profile (score: 4), which mediates the transport of nuclear proteins into the nucleus and thus supervise selective accumulation.29 This IDR is not well conserved in mycobacterial species including ulcerian, avium, smegmatis, etc. We propose the role of IDR in aiding the protein to carry out an array of functions and targeting this protein could be crucial in curbing the pathogenic activity.

Materials and methods Dataset and sources Protein sequences and annotations for M. tuberculosis H37Rv were retrieved from Tuberculist (http://genolist.pasteur.fr /TubercuList). The proteome dataset of H37Rv strain comprises of 3999 proteins. Functional classification of the protein was studied using data from target explorer (http://saclab.tamu.edu/target_explorer.html),18 Tuberculist—TB Gene (http://tuberculist.epfl.ch) and WebTB portal (http://www.webTB.org). All the data were downloaded from target explorer using an automated downloading script. This data provided us with additional knowledge about most of the proteins, including their ranking as drug targets, differential expression data, essentiality, homology, drug response, functional classes and annotations, etc., in the form of a consolidated and comprehensive platform. Protein disorder prediction and parameters Two different disorder prediction algorithms—IUPred 9,10 and DISpro were used for predicting the IDRs. First of all, the proteome dataset was scanned for IDRs using the standalone version of IUPred. It predicts the IDRs of proteins on the basis of their estimated energy content. To crosscheck the results the same dataset was passed through the other software, DISpro, which takes into account evolutionary information profiles, predicted secondary structure and solvent accessibility. The results obtained from the two different algorithms were parsed using in-house scripts with a threshold of the predicted disordered region to be 30 consecutive amino acids. The extent of disorderliness was calculated as the percentage of protein predicted as IDRs. A similar exercise was carried out for results from both the algorithms. Filter criteria Essential intrinsically disordered proteins. The database of essential genes (DEG) 30,31 was used to identify the essential genes in M. tuberculosis which in turn have been derived from the transposon site hybridization (TraSH) study of the bacteria conducted to find genes essential for growth of the bacteria using high density mutagenesis.12 A stand-alone version of BLAST was used to identify the essential genes. The aim of these filtering criteria was to find unstructured proteins which are part of the minimal set of proteins essential for the optimal growth of bacteria. Pathway and interaction studies. Proteins interact in complicated ways because of their vast complex structure. The study of protein–protein interactions is a major potential source to single out new potential drug targets. Protein interactions add an insight into the indispensability of a protein to the organism. STRING 32–36—a database for protein–protein interactions and STITCH37—a database for protein-chemical interactions were used to study the interaction network on the basis of various factors, primarily from experimental evidence. Absence of homologs in the human proteome and human gut and oral microbiome. The identification of proteins that are non-homologous to human proteins but essential for the survival of the pathogen represents a promising means for the identification of novel drug targets.38 The full human proteome was downloaded from NCBI (http://www.ncbi.nlm.nih.gov/genome/guide/human/) and BLAST39 (blastall 2.2.18) was used to filter out the proteins from the IDEPs dataset which had homology (identity > 45%) with human protein. Human gut and oral flora is constituted of microbes that are considered to influence the physiology, nutrition, immunity and development of the host and their hampered metabolism may have adverse effects. The completed proteome of eight gut flora (http://www.metagenome.jp/microbes/data.html, http://www.metagenome.jp/microbes/genomes.html) and 27 oral flora (http://www.homd.org) were downloaded, cumulatively run through cd-hit40 (similarity = 60% and word size = 4) and similar sequences were removed from the list of potential targets.

Conclusion Earlier, signifying the role of IDPs, Cheng et al.6 suggested that protein interactions leading from disorder-to-order transitions can provide a new dimension to drug target discovery. Further experimentation is required to elucidate the precise role of the identified proteins and to validate them as drug targets. This study provides a foundation and rationale for targeting IDPs in M. tuberculosis. Detailed analysis of overlapping or isolated but unique pathways can help in understanding the chokepoints. We firmly believe that further investigations will help unravel the regulatory and metabolic significance of the proteins and shun the pathogen's survivasome.

6 of 8

07/17/2009 02:29 PM



Acknowledgements The authors wish to acknowledge Dr M. Madan Babu for his valuable suggestions for structuring the manuscript and sincerely thank Dr Anshu, G. P. Singh, Vikram Chots and Namit Bharija for their input during discussion. Funding support from the Council of Scientific and Industrial Research (CSIR) through the In Silico Biology Task Force project is duly acknowledged.

References 1 WHO, Global tuberculosis control—surveillance, planning, financing, 2008, http://www.who.int/tb/publications/global_report /2008/en. 2 A. C. Murillo, H. Y. Li, T. Alber, E. N. Baker, J. M. Berger, L. T. Cherney, M. M. Cherney, Y. S. Cho, D. Eisenberg, C. R. Garen, C. W. Goulding, L. W. Hung, T. R. Ioerger, W. R. Jacobs, M. N. James, C. Kim, I. Krieger, J. S. Lott, R. Sankaranarayanan, B. W. Segelke, T. C. Terwilliger, F. Wang, S. Wang and J. C. Sacchettini, Infect. Disord. Drug Targets, 2007, 7, 127–139 [Links]. 3 S. T. Cole, R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris, S. V. Gordon, K. Eiglmeier, S. Gas, C. E. Barry III, F. Tekaia, K. Badcock, D. Basham, D. Brown, T. Chillingworth, R. Connor, R. Davies, K. Devlin, T. Feltwell, S. Gentles, N. Hamlin, S. Holroyd, T. Hornsby, K. Jagels, A. Krogh, J. McLean, S. Moule, L. Murphy, K. Oliver, J. Osborne, M. A. Quail, M. A. Rajandream, J. Rogers, S. Rutter, K. Seeger, J. Skelton, R. Squares, S. Squares, J. E. Sulston, K. Taylor, S. Whitehead and B. G. Barrell, Nature, 1998, 393, 537–544 [Links]. 4 P. Tompa, Trends Biochem. Sci., 2002, 27, 527–533 [Links]. 5 A. K. Dunker, C. J. Brown, J. D. Lawson, L. M. Iakoucheva and Z. Obradovic, Biochemistry, 2002, 41, 6573–6582 [Links]. 6 Y. Cheng, T. LeGall, C. J. Oldfield, J. P. Mueller, Y. Y. Van, P. Romero, M. S. Cortese, V. N. Uversky and A. K. Dunker, Trends Biotechnol., 2006, 24, 435–442 [Links]. 7 N. Issaeva, A. Friedler, P. Bozko, K. G. Wiman, A. R. Fersht and G. Selivanova, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 13303–13307 [Links]. 8 J. J. Ward, J. S. Sodhi, L. J. McGuffin, B. F. Buxton and D. T. Jones, J. Mol. Biol., 2003, 337, 635–645. 9 Z. Dosztanyi, V. Csizmok, P. Tompa and I. Simon, J. Mol. Biol., 2005, 347, 827–839 [Links]. 10 Z. Dosztanyi, V. Csizmok, P. Tompa and I. Simon, Bioinformatics, 2005, 21, 3433–3434 [Links]. 11 J. Cheng, M. Sweredoski and P. Baldi, Data Mining and Knowledge Discovery, 2005, 11, 213–222. 12 C. M. Sassetti, D. H. Boyd and E. J. Rubin, Mol. Microbiol., 2003, 48, 77–84 [Links]. 13 H. Hegyi, E. Schad and P. Tompa, BMC Struct. Biol., 2007, 7, 65 [Links]. 14 C. Haynes, C. J. Oldfield, F. Ji, N. Klitgord, M. E. Cusick, P. Radivojac, V. N. Uversky, M. Vidal and L. M. Iakoucheva, PLoS Comput. Biol., 2006, 2, e100. 15 G. P. Singh, M. Ganapathi and D. Dash, Proteins, 2007, 66, 761–765. 16 A. J. Steyn, D. M. Collins, M. K. Hondalus, W. R. Jacobs Jr, R. P. Kawakami and B. R. Bloom, Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 3147–3152 [Links]. 17 C. Mani, N. Selvakumar, S. Narayanan and P. R. Narayanan, J. Clin. Microbiol., 2001, 39, 2987–2990 [Links]. 18 S. Hasan, S. Daugelat, P. S. Rao and M. Schreiber, PLoS Comput. Biol., 2006, 2, e61. 19 P. Datta, A. Dasgupta, A. K. Singh, P. Mukherjee, M. Kundu and J. Basu, Mol. Microbiol., 2006, 62, 1655–1673 [Links]. 20 P. Datta, A. Dasgupta, S. Bhakta and J. Basu, J. Biol. Chem., 2002, 277, 24983–24987 [Links]. 21 F. Gueiros-Filho, Cell Division, in Bacillus: Cellular and Molecular Biology, ed. P. Graumann, Horizon Scientific Press/Caister Academic Press, Norwich, UK, 2007. 22 A. M. Gehring, W. J. Lees, D. J. Mindiola, C. T. Walsh and E. D. Brown, Biochemistry, 1996, 35, 579–585 [Links]. 23 L. R. Olsen and S. L. Roderick, Biochemistry, 2001, 40, 1913–1921 [Links]. 24 W. Zhang, V. C. Jones, M. S. Scherman, S. Mahapatra, D. Crick, S. Bhamidi, Y. Xin, M. R. McNeil and Y. Ma, Int. J. Biochem. Cell Biol., 2008, 40, 2560–2571 [Links]. 25 J. Buglino, V. Shen, P. Hakimian and C. D. Lima, Structure, 2002, 10, 1581–1592 [Links]. 26 Y. I. Wolf, L. Aravind, N. V. Grishin and E. V. Koonin, Genome Res., 1999, 9, 689–710 [Links]. 27 J. M. Scott and W. G. Haldenwang, J. Bacteriol., 1999, 181, 4653–4660 [Links]. 28 E. Gasteiger, A. Gattiker, C. Hoogland, I. Ivanyi, R. D. Appel and A. Bairoch, Nucleic Acids Res., 2003, 31, 3784–3788 [Links]. 29 M. Cokol, R. Nair and B. Rost, EMBO Rep., 2000, 1, 411–415 [Links]. 30 R. Zhang, H. Y. Ou and C. T. Zhang, Nucleic Acids Res., 2004, 32, D271–D272 [Links]. 31 R. Zhang and Y. Lin, Nucleic Acids Res., 2009, 37, D455–D458 [Links]. 32 L. J. Jensen, M. Kuhn, M. Stark, S. Chaffron, C. Creevey, J. Muller, T. Doerks, P. Julien, A. Roth, M. Simonovic, P. Bork and M. C. von, Nucleic Acids Res., 2009, 37, D412–D416 [Links]. 33 B. Snel, G. Lehmann, P. Bork and M. A. Huynen, Nucleic Acids Res., 2000, 28, 3442–3444 [Links]. 34 M. C. von Huynen, D. Jaeggi, S. Schmidt, P. Bork and B. Snel, Nucleic Acids Res., 2003, 31, 258–261 [Links]. 35 M. C. von, L. J. Jensen, B. Snel, S. D. Hooper, M. Krupp, M. Foglierini, N. Jouffre, M. A. Huynen and P. Bork, Nucleic Acids Res., 2005, 33, D433–D437 [Links]. 36 M. C. von, L. J. Jensen, M. Kuhn, S. Chaffron, T. Doerks, B. Kruger, B. Snel and P. Bork, Nucleic Acids Res., 2007, 35, D358–D362 [Links]. 37 M. Kuhn, M. C. von, M. Campillos, L. J. Jensen and P. Bork, Nucleic Acids Res., 2008, 36, D684–D688 [Links]. 38 K. R. Sakharkar, M. K. Sakharkar and V. T. K. Chow, Biocomputational Strategies for Microbial Drug Target Identification, in New Antibiotic Targets, ed. W. S. Champney, Humana Press, Totowa, New Jersey, 2008, pp. 1–9. 39 S. F. Altschul, W. Gish, W. Miller, E. W. Myers and D. J. Lipman, J. Mol. Biol., 1990, 215, 403–410 [Links]. 40 W. Li and A. Godzik, Bioinformatics, 2006, 22, 1658–1659 [Links].

7 of 8

07/17/2009 02:29 PM



Footnotes This article is part of a Molecular BioSystems themed issue on Computational and Systems Biology. Electronic supplementary information (ESI) available: Supplementary figures, Fig. S1 and S2. Links to view 3D visualisations of structures using FirstGlance. See DOI: 10.1039/b905518p

This journal is © The Royal Society of Chemistry 2009

8 of 8

07/17/2009 02:29 PM