NORMAL GENOMIC

• Kelil A, Wang S, Brzezinski R, Fleury A
• CLUSS: clustering of protein sequences based on a new similarity measure.
• BMC Bioinformatics. 2007; 8: 286-286
• Display abstract

• BACKGROUND: The rapid burgeoning of available protein data makes the useof clustering within families of proteins increasingly important. Thechallenge is to identify subfamilies of evolutionarily related sequences.This identification reveals phylogenetic relationships, which provideprior knowledge to help researchers understand biological phenomena. Agood evolutionary model is essential to achieve a clustering that reflectsthe biological reality, and an accurate estimate of protein sequencesimilarity is crucial to the building of such a model. Most existingalgorithms estimate this similarity using techniques that are notnecessarily biologically plausible, especially for hard-to-align sequencessuch as proteins with different domain structures, which cause manydifficulties for the alignment-dependent algorithms. In this paper, wepropose a novel similarity measure based on matching amino acidsubsequences. This measure, named SMS for Substitution MatchingSimilarity, is especially designed for application to non-aligned proteinsequences. It allows us to develop a new alignment-free algorithm, namedCLUSS, for clustering protein families. To the best of our knowledge, thisis the first alignment-free algorithm for clustering protein sequences.Unlike other clustering algorithms, CLUSS is effective on both alignableand non-alignable protein families. In the rest of the paper, we use theterm "phylogenetic" in the sense of "relatedness of biological functions".RESULTS: To show the effectiveness of CLUSS, we performed an extensiveclustering on COG database. To demonstrate its ability to deal withhard-to-align sequences, we tested it on the GH2 family. In addition, wecarried out experimental comparisons of CLUSS with a variety of mainstreamalgorithms. These comparisons were made on hard-to-align and easy-to-alignprotein sequences. The results of these experiments show the superiorityof CLUSS in yielding clusters of proteins with similar functionalactivity. CONCLUSION: We have developed an effective method and tool forclustering protein sequences to meet the needs of biologists in terms ofphylogenetic analysis and prediction of biological functions. Compared toexisting clustering methods, CLUSS more accurately highlights thefunctional characteristics of the clustered families. It providesbiologists with a new and plausible instrument for the analysis of proteinsequences, especially those that cause problems for thealignment-dependent algorithms.

• Casbon JA, Saqi MA
• On single and multiple models of protein families for the detection ofremote sequence relationships.
• BMC Bioinformatics. 2006; 7: 48-48
• Display abstract

• BACKGROUND: The detection of relationships between a protein sequence ofunknown function and a sequence whose function has been characterisedenables the transfer of functional annotation. However in many cases theserelationships can not be identified easily from direct comparison of thetwo sequences. Methods which compare sequence profiles have been shown toimprove the detection of these remote sequence relationships. However, thebest method for building a profile of a known set of sequences has notbeen established. Here we examine how the type of profile built affectsits performance, both in detecting remote homologs and in the resultingalignment accuracy. In particular, we consider whether it is better tomodel a protein superfamily using a single structure-based alignment thatis representative of all known cases of the superfamily, or to usemultiple sequence-based profiles each representing an individual member ofthe superfamily. RESULTS: Using profile-profile methods for remote homologdetection we benchmark the performance of single structure-basedsuperfamily models and multiple domain models. On average, over allsuperfamilies, using a truncated receiver operator characteristic (ROC5)we find that multiple domain models outperform single superfamily models,except at low error rates where the two models behave in a similar way.However there is a wide range of performance depending on the superfamily.For 12% of all superfamilies the ROC5 value for superfamily models isgreater than 0.2 above the domain models and for 10% of superfamilies thedomain models show a similar improvement in performance over thesuperfamily models. CONCLUSION: Using a sensitive profile-profile methodwe have investigated the performance of single structure-based models andmultiple sequence models (domain models) in detecting remote superfamilymembers. We find that overall, multiple models perform better inrecognition although single structure-based models display betteralignment accuracy.

• Davies AM, Rasia RM, Vila AJ, Sutton BJ, Fabiane SM
• Effect of pH on the active site of an Arg121Cys mutant of themetallo-beta-lactamase from Bacillus cereus: implications for the enzymemechanism.
• Biochemistry. 2005; 44: 4841-9
• Display abstract

• The zinc-dependent metallo-beta-lactamases are a group of bacterialenzymes that pose a threat to the future efficacy of present-dayantibiotics. Their mechanism is poorly understood, and there are noclinically useful inhibitors. While most members of the group contain twotightly bound zinc ions in their active sites, the Bacillus cereus enzymehas a much lower affinity for its second zinc (Zn2), thought to be due tothe presence of Arg121 immediately beneath the floor of the active site(cf. Cys/Ser/His121 in the bizinc enzymes). Crystal structures of theArg121Cys mutant of the B. cereus 569/H/9 enzyme were solved at pH 7.0,5.0, and 4.5, each in the presence of either 20 microM or 20 mM Zn(2+) togenerate the mono- and bizinc forms, respectively. Surprisingly, thestructure of the active site was unaffected by the mutation; a network ofordered water molecules replaced the interactions made by the arginineside chain, and the occupancy of Zn2 appeared minimally changed. As the pHwas lowered, Zn2 moved away from one of its ligands, Asp120, but was"tracked" by two others, Cys221 and His263. Furthermore, the hydroxide ion(and proposed nucleophile for beta-lactam hydrolysis) was bound to Zn1 atpH 5 and above but absent at pH 4.5. This provides experimental evidencefor an earlier proposed mechanism in which protonation of Asp120 and theZn1-bound hydroxide are the two events that lead to the loss of activityat low pH.

• Kann MG, Thiessen PA, Panchenko AR, Schaffer AA, Altschul SF, Bryant SH
• A structure-based method for protein sequence alignment.
• Bioinformatics. 2005; 21: 1451-6
• Display abstract

• MOTIVATION: With the continuing rapid growth of protein sequence data,protein sequence comparison methods have become the most widely used toolsof bioinformatics. Among these methods are those that useposition-specific scoring matrices (PSSMs) to describe protein families.PSSMs can capture information about conserved patterns within families,which can be used to increase the sensitivity of searches for relatedsequences. Certain types of structural information, however, are notgenerally captured by PSSM search methods. Here we introduce a program,Structure-based ALignment TOol (SALTO), that aligns protein querysequences to PSSMs using rules for placing and scoring gaps that areconsistent with the conserved regions of domain alignments from NCBI'sConserved Domain Database. RESULTS: In most cases, the alignment scoresobtained using the local alignment version follow an extreme valuedistribution. SALTO's performance in finding related sequences andproducing accurate alignments is similar to or better than that of IMPALA;one advantage of SALTO is that it imposes an explicit gapping model oneach protein family. AVAILABILITY: A stand-alone version of the programthat can generate global or local alignments is available by ftpdistribution (ftp://ftp.ncbi.nih.gov/pub/SALTO/), and has beenincorporated to Cn3D structure/alignment viewer. CONTACT:bryant@ncbi.nlm.nih.gov.

• Bhaduri A, Pugalenthi G, Sowdhamini R
• PASS2: an automated database of protein alignments organised as structuralsuperfamilies.
• BMC Bioinformatics. 2004; 5: 35-35
• Display abstract

• BACKGROUND: The functional selection and three-dimensional structuralconstraints of proteins in nature often relates to the retention ofsignificant sequence similarity between proteins of similar fold andfunction despite poor sequence identity. Organization of structure-basedsequence alignments for distantly related proteins, provides a map of theconserved and critical regions of the protein universe that is useful forthe analysis of folding principles, for the evolutionary unification ofprotein families and for maximizing the information return fromexperimental structure determination. The Protein Alignment organised asStructural Superfamily (PASS2) database represents continuously updated,structural alignments for evolutionary related, sequentially distantproteins. DESCRIPTION: An automated and updated version of PASS2 is, indirect correspondence with SCOP 1.63, consisting of sequences havingidentity below 40% among themselves. Protein domains have been groupedinto 628 multi-member superfamilies and 566 single member superfamilies.Structure-based sequence alignments for the superfamilies have beenobtained using COMPARER, while initial equivalencies have been derivedfrom a preliminary superposition using LSQMAN or STAMP 4.0. The finalsequence alignments have been annotated for structural features usingJOY4.0. The database is supplemented with sequence relatives belonging todifferent genomes, conserved spatially interacting and structural motifs,probabilistic hidden markov models of superfamilies based on thealignments and useful links to other databases. Probabilistic models andsensitive position specific profiles obtained from reliable superfamilyalignments aid annotation of remote homologues and are useful tools instructural and functional genomics. PASS2 presents the phylogeny of itsmembers both based on sequence and structural dissimilarities. Clusteringof members allows us to understand diversification of the family members.The search engine has been improved for simpler browsing of the database.CONCLUSIONS: The database resolves alignments among the structural domainsconsisting of evolutionarily diverged set of sequences. Availability ofreliable sequence alignments of distantly related proteins despite poorsequence identity and single-member superfamilies permit better samplingof structures in libraries for fold recognition of new sequences and forthe understanding of protein structure-function relationships ofindividual superfamilies. PASS2 is accessible athttp://www.ncbs.res.in/~faculty/mini/campass/pass2.html

• Dal Peraro M, Vila AJ, Carloni P
• Substrate binding to mononuclear metallo-beta-lactamase from Bacilluscereus.
• Proteins. 2004; 54: 412-23
• Display abstract

• Structure and dynamics of substrate binding (cefotaxime) to the catalyticpocket of the mononuclear zinc-beta-lactamase from Bacillus cereus areinvestigated by molecular dynamics simulations. The calculations, whichare based on the hydrogen-bond pattern recently proposed by Dal Peraro etal. (J Biol Inorg Chem 2002; 7:704-712), are carried out for both the freeand the complexed enzyme. In the resting state, active site pattern andtemperature B-factors are in agreement with crystallographic data. In thecomplexed form, cefotaxime is accommodated into a stable orientation inthe catalytic pocket within the nanosecond timescale, interacting with theenzyme zinc-bound hydroxide and the surrounding loops. The beta-lactamring remains stable and very close to the hydroxide nucleophile agent,giving a stable representation of the productive enzyme-substrate complex.

• de Seny D et al.
• Mutational analysis of the two zinc-binding sites of the Bacillus cereus569/H/9 metallo-beta-lactamase.
• Biochem J. 2002; 363: 687-96
• Display abstract

• The metallo-beta-lactamase BcII from Bacillus cereus 569/H/9 possesses abinuclear zinc centre. The mono-zinc form of the enzyme displays anappreciably high activity, although full efficiency is observed for thedi-zinc enzyme. In an attempt to assign the involvement of the differentzinc ligands in the catalytic properties of BcII, individual substitutionsof selected amino acids were generated. With the exception ofHis(116)-->Ser (H116S), C221A and C221S, the mono- and di-zinc forms ofall the other mutants were poorly active. The activity of H116S decreasesby a factor of 10 when compared with the wild type. The catalyticefficiency of C221A and C221S was zinc-dependent. The mono-zinc forms ofthese mutants exhibited a low activity, whereas the catalytic efficiencyof their respective di-zinc forms was comparable with that of the wildtype. Surprisingly, the zinc contents of the mutants and the wild-typeBcII were similar. These data suggest that the affinity of thebeta-lactamase for the metal was not affected by the substitution of theligand. The pH-dependence of the H196S catalytic efficiency indicates thatthe zinc ions participate in the hydrolysis of the beta-lactam ring byacting as a Lewis acid. The zinc ions activate the catalytic watermolecule, but also polarize the carbonyl bond of the beta-lactam ring andstabilize the development of a negative charge on the carbonyl oxygen ofthe tetrahedral reaction intermediate. Our studies also demonstrate thatAsn(233) is not directly involved in the interaction with the substrates.

• Suarez D, Merz KM Jr
• Molecular dynamics simulations of the mononuclear zinc-beta-lactamase fromBacillus cereus.
• J Am Chem Soc. 2001; 123: 3759-70
• Display abstract

• Herein, we report molecular dynamics simulations of the mononuclear formof the Bacillus cereuszinc-beta-lactamase. We studied two differentconfigurations which differ in the presence of a zinc-bound hydroxide or azinc-bound water and in the protonation state of the essential His210residue. Contacts of the catalytically important residues (Asp90, His210,Cys168, etc.) with the zinc center are characterized by the MD analyses.The nature of the Zn-OH(2) --> His210 proton transfer pathway connectingthe two configurations was studied by means of QM calculations on clustermodels while the relative stability of the two configurations wasestimated from QM/MM calculations in the enzyme. From these results, atheoretical model for the kinetically active form of the B. cereusmetalloenzyme is proposed. Some mechanistic implications and the influenceof mutating the Cys168 residue are also discussed.

• Balaji S, Srinivasan N
• Use of a database of structural alignments and phylogenetic trees ininvestigating the relationship between sequence and structural variabilityamong homologous proteins.
• Protein Eng. 2001; 14: 219-26
• Display abstract

• The database PALI (Phylogeny and ALIgnment of homologous proteinstructures) consists of families of protein domains of knownthree-dimensional (3D) structure. In a PALI family, every member has beenstructurally aligned with every other member (pairwise) and alsosimultaneous superposition (multiple) of all the members has beenperformed. The database also contains 3D structure-based andstructure-dependent sequence similarity-based phylogenetic dendrograms forall the families. The PALI release used in the present analysis comprises225 families derived largely from the HOMSTRAD and SCOP databases. Thequality of the multiple rigid-body structural alignments in PALI wascompared with that obtained from COMPARER, which encodes a procedure basedon properties and relationships. The alignments from the two proceduresagreed very well and variations are seen only in the low sequencesimilarity cases often in the loop regions. A validation of DirectPairwise Alignment (DPA) between two proteins is provided by comparing itwith Pairwise alignment extracted from Multiple Alignment of all themembers in the family (PMA). In general, DPA and PMA are found to varyrarely. The ready availability of pairwise alignments allows the analysisof variations in structural distances as a function of sequencesimilarities and number of topologically equivalent Calpha atoms. Thestructural distance metric used in the analysis combines root mean squaredeviation (r.m.s.d.) and number of equivalences, and is shown to varysimilarly to r.m.s.d. The correlation between sequence similarity andstructural similarity is poor in pairs with low sequence similarities. Acomparison of sequence and 3D structure-based phylogenies for all thefamilies suggests that only a few families have a radical difference inthe two kinds of dendrograms. The difference could occur when the sequencesimilarity among the homologues is low or when the structures aresubjected to evolutionary pressure for the retention of function. The PALIdatabase is expected to be useful in furthering our understanding of therelationship between sequences and structures of homologous proteins andtheir evolution.

• Concha NO et al.
• Crystal structure of the IMP-1 metallo beta-lactamase from Pseudomonasaeruginosa and its complex with a mercaptocarboxylate inhibitor: bindingdeterminants of a potent, broad-spectrum inhibitor.
• Biochemistry. 2000; 39: 4288-98
• Display abstract

• Metallo beta-lactamase enzymes confer antibiotic resistance to bacteria bycatalyzing the hydrolysis of beta-lactam antibiotics. This relatively newform of resistance is spreading unchallenged as there is a current lack ofpotent and selective inhibitors of metallo beta-lactamases. Reported hereare the crystal structures of the native IMP-1 metallo beta-lactamase fromPseudomonas aeruginosa and its complex with a mercaptocarboxylateinhibitor, 2-[5-(1-tetrazolylmethyl)thien-3-yl]-N-[2-(mercaptomethyl)-4-(phenylb utyrylglycine)]. The structures were determined by molecularreplacement, and refined to 3.1 A (native) and 2.0 A (complex) resolution.Binding of the inhibitor in the active site induces a conformationalchange that results in closing of the flap and transforms the active sitegroove into a tunnel-shaped cavity enclosing 83% of the solvent accessiblesurface area of the inhibitor. The inhibitor binds in the active sitethrough interactions with residues that are conserved among metallobeta-lactamases; the inhibitor's carboxylate group interacts with Lys161,and the main chain amide nitrogen of Asn167. In the "oxyanion hole", theamide carbonyl oxygen of the inhibitor interacts through a water moleculewith the side chain of Asn167, the inhibitor's thiolate bridges the twoZn(II) ions in the active site displacing the bridging water, and thephenylbutyryl side chain binds in a hydrophobic pocket (S1) at the base ofthe flap. The flap is displaced 2.9 A compared to the unbound structure,allowing Trp28 to interact edge-to-face with the inhibitor's thiophenering. The similarities between this inhibitor and the beta-lactamsubstrates suggest a mode of substrate binding and the role of theconserved residues in the active site. It appears that the metallobeta-lactamases bind their substrates by establishing a subset of bindinginteractions near the catalytic center with conserved characteristicchemical groups of the beta-lactam substrates. These interactions arecomplemented by additional nonspecific binding between the more variablegroups in the substrates and the flexible flap. This unique mode ofbinding of the mercaptocarboxylate inhibitor in the enzyme active siteprovides a binding model for metallo beta-lactamase inhibition withutility for future drug design.

• Wallqvist A, Fukunishi Y, Murphy LR, Fadel A, Levy RM
• Iterative sequence/secondary structure search for protein homologs:comparison with amino acid sequence alignments and application to foldrecognition in genome databases.
• Bioinformatics. 2000; 16: 988-1002
• Display abstract

• MOTIVATION: Sequence alignment techniques have been developed intoextremely powerful tools for identifying the folding families and functionof proteins in newly sequenced genomes. For a sufficiently low sequenceidentity it is necessary to incorporate additional structural informationto positively detect homologous proteins. We have carried out an extensiveanalysis of the effectiveness of incorporating secondary structureinformation directly into the alignments for fold recognition andidentification of distant protein homologs. A secondary structuresimilarity matrix based on a database of three-dimensionally alignedproteins was first constructed. An iterative application of dynamicprogramming was used which incorporates linear combinations of amino acidand secondary structure sequence similarity scores. Initially, onlyprimary sequence information is used. Subsequently contributions fromsecondary structure are phased in and new homologous proteins arepositively identified if their scores are consistent with thepredetermined error rate. RESULTS: We used the SCOP40 database, where onlyPDB sequences that have 40% homology or less are included, to calibratehomology detection by the combined amino acid and secondary structuresequence alignments. Combining predicted secondary structure with sequenceinformation results in a 8-15% increase in homology detection withinSCOP40 relative to the pairwise alignments using only amino acid sequencedata at an error rate of 0.01 errors per query; a 35% increase is observedwhen the actual secondary structure sequences are used. Incorporatingpredicted secondary structure information in the analysis of six smallgenomes yields an improvement in the homology detection of approximately20% over SSEARCH pairwise alignments, but no improvement in the totalnumber of homologs detected over PSI-BLAST, at an error rate of 0.01errors per query. However, because the pairwise alignments based oncombinations of amino acid and secondary structure similarity aredifferent from those produced by PSI-BLAST and the error rates can becalibrated, it is possible to combine the results of both searches. Anadditional 25% relative improvement in the number of genes identified atan error rate of 0.01 is observed when the data is pooled in this way.Similarly for the SCOP40 dataset, PSI-BLAST detected 15% of all possiblehomologs, whereas the pooled results increased the total number ofhomologs detected to 19%. These results are compared with recent reportsof homology detection using sequence profiling methods. AVAILABILITY:Secondary structure alignment homepage at http://lutece.rutgers.edu/ssasCONTACT: anders@rutchem.rutgers.edu; ronlevy@lutece.rutgers.eduSupplementary Information: Genome sequence/structure alignment results athttp://lutece.rutgers.edu/ss_fold_predictions.

• Bray JE, Todd AE, Pearl FM, Thornton JM, Orengo CA
• The CATH Dictionary of Homologous Superfamilies (DHS): a consensusapproach for identifying distant structural homologues.
• Protein Eng. 2000; 13: 153-65
• Display abstract

• A consensus approach has been developed for identifying distant structuralhomologues. This is based on the CATH Dictionary of HomologousSuperfamilies (DHS), a database of validated multiple structuralalignments annotated with consensus functional information forevolutionary protein superfamilies (URL: http://www.biochem.ucl.ac.uk/bsm/dhs). Multiple structural alignments have beengenerated for 362 well-populated superfamilies in the CATH structuraldomain database and annotated with secondary structure, physicochemicalproperties, functional sequence patterns and protein-ligand interactiondata. Consensus functional information for each superfamily includesdescriptions and keywords extracted from SWISS-PROT and the ENZYMEdatabase. The Dictionary provides a powerful resource to validate, examineand visualize key structural and functional features of each homologoussuperfamily. The value of the DHS, for assessing functional variabilityand identifying distant evolutionary relationships, is illustrated usingthe pyridoxal-5'-phosphate (PLP) binding aspartate aminotransferasesuperfamily. The DHS also provides a tool for examining sequence-structurerelationships for proteins within each fold group.

• Yang AS, Honig B
• An integrated approach to the analysis and modeling of protein sequencesand structures. III. A comparative study of sequence conservation inprotein structural families using multiple structural alignments.
• J Mol Biol. 2000; 301: 691-711
• Display abstract

• The information required to generate a protein structure is contained inits amino acid sequence, but how three-dimensional information is mappedonto a linear sequence is still incompletely understood. Multiplestructure alignments of similar protein structures have been used toinvestigate conserved sequence features but contradictory results havebeen obtained, due, in large part, to the absence of subjective criteriato be used in the construction of sequence profiles and in thequantitative comparison of alignment results. Here, we report a newprocedure for multiple structure alignment and use it to constructstructure-based sequence profiles for similar proteins. The definition of"similar" is based on the structural alignment procedure and on theprotein structural distance (PSD) described in paper I of this series,which offers an objective measure for protein structure relationships. Ourapproach is tested in two well-studied groups of proteins; serineproteases and Ig-like proteins. It is demonstrated that the quality of asequence profile generated by a multiple structure alignment is quitesensitive to the PSD used as a threshold for the inclusion of proteins inthe alignment. Specifically, if the proteins included in the aligned setare too distant in structure from one another, there will be a dilution ofinformation and patterns that are relevant to a subset of the proteins arelikely to be lost.In order to understand better how the samethree-dimensional information can be encoded in seemingly unrelatedsequences, structure-based sequence profiles are constructed for subsetsof proteins belonging to nine superfolds. We identify patterns ofrelatively conserved residues in each subset of proteins. It isdemonstrated that the most conserved residues are generally located in theregions where tertiary interactions occur and that are relativelyconserved in structure. Nevertheless, the conservation patterns arerelatively weak in all cases studied, indicating thatstructure-determining factors that do not require a particular sequentialarrangement of amino acids, such as secondary structure propensities andhydrophobic interactions, are important in encoding protein foldinformation. In general, we find that similar structures can fold withouthaving a set of highly conserved residue clusters or a well-conservedsequence profile; indeed, in some cases there is no apparent conservationpattern common to structures with the same fold. Thus, when a group ofproteins exhibits a common and well-defined sequence pattern, it is morelikely that these sequences have a close evolutionary relationship ratherthan the similarities having arisen from the structural requirements of agiven fold.

• Carfi A, Duee E, Galleni M, Frere JM, Dideberg O
• 1.85 A resolution structure of the zinc (II) beta-lactamase from Bacilluscereus.
• Acta Crystallogr D Biol Crystallogr. 1998; 54: 313-23
• Display abstract

• Class B beta-lactamases are wide spectrum enzymes which require bivalentmetal ions for activity. The structure of the class B zinc-ion-dependentbeta-lactamase from Bacillus cereus (BCII) has been refined at 1.85 Aresolution using data collected on cryocooled crystals (100 K). The enzymefrom B. cereus has a molecular mass of 24 946 Da and is folded into abeta-sandwich structure with helices on the external faces. The activesite is located in a groove running between the two beta-sheets [Carfi etal. (1995). EMBO J. 14, 4914-4921]. The 100 K high-resolution BCIIstructure shows one fully and one partially occupied zinc sites. The zincion in the fully occupied site (the catalytic zinc) is coordinated bythree histidines and one water molecule. The second zinc ion is at 3.7 Afrom the first one and is coordinated by one histidine, one cysteine, oneaspartate and one unknown molecule (most likely a carbonate ion). In theB. cereus zinc beta-lactamase the affinity for the second metal-ion is lowat the pH of crystallization (Kd = 25 mM, 293 K; [Baldwin et al. (1978).Biochem. J. 175, 441-447] and the dissociation constant of the second zincion was thus apparently decreased at the cryogenic temperature. Inaddition, the structure of the apo enzyme was determined at 2.5 Aresolution. The removal of the zinc ion by chelating agents causes smallchanges in the active-site environment.

• Carfi A, Duee E, Paul-Soto R, Galleni M, Frere JM, Dideberg O
• X-ray structure of the ZnII beta-lactamase from Bacteroides fragilis in anorthorhombic crystal form.
• Acta Crystallogr D Biol Crystallogr. 1998; 54: 45-57
• Display abstract

• beta-Lactamases are extracellular or periplasmic bacterial enzymes whichconfer resistance to beta-lactam antibiotics. On the basis of theircatalytic mechanisms, they can be divided into two major groups:active-site serine enzymes (classes A, C and D) and the ZnII enzymes(class B). The first crystal structure of a class B enzyme, themetallo-beta-lactamase from Bacillus cereus, has been solved at 2.5 Aresolution [Carfi, Pares, Duee, Galleni, Duez, Frere & Dideberg (1995).EMBO J. 14, 4914-4921]. Recently, the crystal structure of themetallo-beta-lactamase from Bacteroides fragilis has been determined in atetragonal space group [Concha, Rasmussen, Bush & Herzberg (1996).Structure, 4, 823-836]. The structure of the metallo-beta-lactamase fromB. fragilis in an orthorhombic crystal form at 2.0 A resolution isreported here. The final crystallographic R is 0.196 for all the 32501observed reflections in the range 10-2.0 A. The refined model includes 458residues, 437 water molecules, four zinc and two sodium ions. Thesestructures are discussed with reference to Zn binding and activity. Acatalytic mechanism is proposed which is coherent withmetallo-beta-lactamases being active with either one Zn ion (as inAeromonas hydrophila) or two Zn ions (as in B. fragilis) bound to theprotein.

• Fabiane SM et al.
• Crystal structure of the zinc-dependent beta-lactamase from Bacilluscereus at 1.9 A resolution: binuclear active site with features of amononuclear enzyme.
• Biochemistry. 1998; 37: 12404-11
• Display abstract

• The structure of the zinc-dependent beta-lactamase II from Bacillus cereushas been determined at 1.9 A resolution in a crystal form with twomolecules in the asymmetric unit and 400 waters (space group P3121; Rcryst= 20.8%). The active site contains two zinc ions: Zn1 is tightlycoordinated by His86, His88, and His149, while Zn2 is loosely coordinatedby Asp90, Cys168, and His210. A water molecule (W1) lies between the twozinc ions but is significantly closer to Zn1 and at a distance of only 1.9A is effectively a hydroxide moiety and a potential, preactivatednucleophile. In fact, Asp90 bridges W1 to Zn2, and its location is thusdistinct from that of the bridging water molecules in the binuclear zincpeptidases or other binuclear zinc hydrolases. Modeling of penicillin,cephalosporin, and carbapenem binding shows that all are readilyaccommodated within the shallow active site cleft of the enzyme, and theZn1-bound hydroxide is ideally located for nucleophilic attack at thebeta-lactam carbonyl. This enzyme also functions with only one zinc ionpresent. The Zn1-Zn2 distances differ in the two independent molecules inthe crystal (3.9 and 4.4 A), yet the Zn1-W1 distances are both 1.9 A,arguing against involvement of Zn2 in W1 activation. The role of Zn2 isunclear, but the B. cereus enzyme may be an evolutionary intermediatebetween the mono- and bizinc metallo-beta-lactamases. The broadspecificity of this enzyme, together with the increasing prevalence ofzinc-dependent metallo-beta-lactamases, poses a real clinical threat, andthis structure provides a basis for understanding its mechanism anddesigning inhibitors.

• Sonnhammer EL, Eddy SR, Durbin R
• Pfam: a comprehensive database of protein domain families based on seedalignments.
• Proteins. 1997; 28: 405-20
• Display abstract

• Databases of multiple sequence alignments are a valuable aid to proteinsequence classification and analysis. One of the main challenges whenconstructing such a database is to simultaneously satisfy the conflictingdemands of completeness on the one hand and quality of alignment anddomain definitions on the other. The latter properties are best dealt withby manual approaches, whereas completeness in practice is only amenable toautomatic methods. Herein we present a database based on hidden Markovmodel profiles (HMMs), which combines high quality and completeness. Ourdatabase, Pfam, consists of parts A and B. Pfam-A is curated and containswell-characterized protein domain families with high quality alignments,which are maintained by using manually checked seed alignments and HMMs tofind and align all members. Pfam-B contains sequence families that weregenerated automatically by applying the Domainer algorithm to cluster andalign the remaining protein sequences after removal of Pfam-A domains. Byusing Pfam, a large number of previously unannotated proteins from theCaenorhabditis elegans genome project were classified. We have alsoidentified many novel family memberships in known proteins, including newkazal, Fibronectin type III, and response regulator receiver domains.Pfam-A families have permanent accession numbers and form a library ofHMMs available for searching and automatic annotation of new proteinsequences.

• Watanabe K, Hata Y, Kizaki H, Katsube Y, Suzuki Y
• The refined crystal structure of Bacillus cereus oligo-1,6-glucosidase at2.0 A resolution: structural characterization of proline-substitutionsites for protein thermostabilization.
• J Mol Biol. 1997; 269: 142-53
• Display abstract

• The crystal structure of oligo-1,6-glucosidase (dextrin6-alpha-glucanohydrolase, EC 3.2.1.10) from Bacillus cereus ATCC7064 hasbeen refined to 2.0 A resolution with an R-factor of 19.6% for 43,328reflections. The final model contains 4646 protein atoms and 221 orderedwater molecules with respective root-mean-square deviations of 0.015 A forbond lengths and of 3.166 degrees for bond angles from the ideal values.The structure consists of three domains: the N-terminal domain (residues 1to 104 and 175 to 480), the subdomain (residues 105 to 174) and theC-terminal domain (residues 481 to 558). The N-terminal domain takes a(beta/alpha)8-barrel structure with additions of an alpha-helix (Nalpha6') between the sixth strand Nbeta6 and the sixth helix N alpha6, analpha-helix (N alpha7') between the seventh strand Nbeta7 and the seventhhelix N alpha7 and three alpha-helices (N alpha8', N alpha8" and Nalpha8'" between the eighth strand Nbeta8 and the eighth helix N alpha8.The subdomain is composed of an alpha-helix, a three-stranded antiparallelbeta-sheet, and long intervening loops. The C-terminal domain has abeta-barrel structure of eight antiparallel beta-strands folded in doubleGreek key motifs, which is distorted in the sixth strand Cbeta6. Threecatalytic residues, Asp199, Glu255 and Asp329, are located at the bottomof a deep cleft formed by the subdomain and a cluster of the twoadditional alpha-helices N alpha8' and N alpha8" in the(beta/alpha)8-barrel. The refined structure reveals the locations of 21proline-substitution sites that are expected to be critical to proteinthermostabilization from a sequence comparison among three Bacillusoligo-1,6-glucosidases with different thermostability. These sites lie inloops, beta-turns and alpha-helices, in order of frequency, except forCys515 in the fourth beta-strand Cbeta4 of the C-terminal domain. Theresidues in beta-turns (Lys121, Glu208, Pro257, Glu290, Pro443, Lys457 andGlu487) are all found at their second positions, and those inalpha-helices (Asn109, Glu175, Thr261 and Ile403) are present at their N1positions of the first helical turns. Those residues in both secondarystructures adopt phi and phi values favorable for proline substitution.Residues preceding the 21 sites are mostly conserved upon prolineoccurrence at these 21 sites in more thermostable Bacillusoligo-1,6-glucosidases. These structural features with respect to the 21sites indicate that the sites in beta-turns and alpha-helices have moreessential prerequisites for proline substitution to thermostabilize theprotein than those in loops. This well supports the previous finding thatthe replacement at the appropriate positions in beta-turns oralpha-helices is the most effective for protein thermostabilization byproline substitution.

• Concha NO, Rasmussen BA, Bush K, Herzberg O
• Crystal structure of the wide-spectrum binuclear zinc beta-lactamase fromBacteroides fragilis.
• Structure. 1996; 4: 823-36
• Display abstract

• BACKGROUND: The metallo-beta-lactamase from Bacteroides fragilishydrolyzes a wide range of beta-lactam antibiotics, and is not clinicallysusceptible to any known beta-lactamase inhibitors. B. fragilis isassociated with post-surgery hospital infections, and there has been arecent report of plasmid-mediated dissemination of the enzyme. Effectiveinhibitors are therefore urgently needed. Knowledge of thethree-dimensional structure will aid in the drug design effort. RESULTS:The crystal structure of the enzyme has been determined by usingmultiwavelength anomalous diffraction at the zinc absorption edge andrefined to 1.85 A resolution. The structure is a four-layeralpha/beta/beta/alpha molecule. The active site, found at the edge of thebeta sandwich contains a binuclear zinc center with several novelfeatures. One zinc is tetrahedrally coordinated, the other has a trigonalbipyramidal coordination; a water/hydroxide molecule serves as a ligandfor both metals. The residues that coordinate the two zincs are invariantin all metallo-beta-lactamases that have been sequenced, except for twoconservative replacements. Despite the existence of the pattern forbinuclear zinc binding, the reported structure of the Bacillus cereusenzyme contains only a single zinc. CONCLUSIONS: Structural analysisindicates that affinity for the penta-coordinated zinc can be modulated byneighboring residues, perhaps explaining the absence of the second zinc inthe B. cereus structure. Models of bound substrates suggest that theactive-site channel can accommodate a wide variety of beta-lactams. Wepropose that the zinc cluster prepares an hydroxide, probably thehydroxide that ligates both zincs, for nucleophilic attack on the carbonylcarbon atom of the beta-lactam. The resulting negatively chargedtetrahedral intermediate implicated in catalysis is stabilized by anoxyanion hole formed by the side chain of the invariant Asn 193 and thetetrahedral zinc.

• Parker JM, Hodges RS
• HomologyPlot: searching for homology to a family of proteins using adatabase of unique conserved patterns.
• J Comput Aided Mol Des. 1994; 8: 193-210
• Display abstract

• A new database of conserved amino acid residues is derived from themultiple sequence alignment of over 84 families of protein sequences thathave been reported in the literature. This database contains sequences ofconserved hydrophobic core patterns which are probably important forstructure and function, since they are conserved for most sequences inthat family. This database differs from other single-motif or signaturedatabases reported previously, since it contains multiple patterns foreach family. The new database is used to align a new sequence with theconserved regions of a family. This is analogous to reports in theliterature where multiple sequence alignments are used to improve asequence alignment. A program called HomologyPlot (suitable for IBM orcompatible computers) uses this database to find homology of a newsequence to a family of protein sequences. There are several advantages tousing multiple patterns. First, the program correctly identifies a newsequence as a member of a known family. Second, the search of the entiredatabase is rapid and requires less than one minute. This is similar toperforming a multiple sequence alignment of a new sequence to all of theknown protein family sequences. Third, the alignment of a new sequence tofamily members is reliable and can reproduce the alignment of conservedregions already described in the literature. The speed and efficiency ofthis method is enhanced, since there is no need to score for insertions ordeletions as is done in the more commonly used sequence alignment methods.In this method only the patterns are aligned. HomologyPlot also providesgeneral information on each family, as well as a listing of patterns in afamily.