Homology Models of the Yersinia Pseudotuberculosisand Yersinia Pestis General Porins and ComparativeAnalysis of Their Functional and Antigenic Regions

http://www.jbsdonline.com

Abstract

The amino acid sequences of the Yersinia pseudotuberculosis porin (YPS) and Y. pestis porin(YPT) have recently deduced but their three-dimensional structures were not known. Thesesequences were analyzed using the servers 3D-PSSM and PredPort. The YPS and YPT porinswere shown to have a high degree of identity (above 50%) in primary and secondary struc-tures. The three-dimensional models of the Yersinia pseudotuberculosis porin (YPS) and Y.pestis porin (YPT) were obtained using the homology modeling approach, SWISS-MODELProtein Modeling Server and 3-D structure of PhoE porin from E. coli as template. Thesuperpositions of the Cα-atoms of the monomers of the Yersinia porins and PhoE poringave a root mean square deviations of 0.47 Å and 0.43 Å for YPS and YPT respectively.

Yersinia porins were found to be very similar in their three-dimentional structure to othernon-specific enterobacterial porins, having the same features of overall fold and disposi-tion of loop L3. The intrinsic structures of the monomer pores of YPS and YPT wereinvestigated and their conductances were predicted with the program HOLE. The goodcorrespondence between the theoretical and experimental magnitudes of YPS conductancewas found. The Yersinia porins were determined to be unusual in containing the substitu-tion, Glu replaced by Val, in a highly conserved pentapeptide (Pro-Glu-Phe-Gly-Gly-Asp),located in the loop L3 tip that disturbs the functionally important cluster of the acidicamino acids in the constriction site.

Comparative analysis of structural organization of YPS and E. coli OmpF porin in theregions involved in subunit association and pore lumen was performed. The YPS porinfunctional properties were predicted. The differences between these porins in polar interac-tions playing a signifiant role in stabilization of the porin trimers were found and discussedin term of the variations in trimer stability.

The Yersinia porins were shown to have the highest degree of the structural similarity. The dif-ferences between the porins were observed in their external loops. Their loops L6 and loopsL8 showed 71.4 and 52.9% of sequence identity, respectively. The arrangement of chargedresidues clustered in the channel external vestibule of these porins was found to be also dif-ferent sugesting the possible differences in their functional properties. The surface exposedregions of Yersinia porins involved in their potential sequential antigenic determinants werecompared. The structural basis of their cross reactivity and antigenic differences is discussed.

Introduction

Porins are transmembrane proteins located in the outer membrane of the Gram-neg-ative bacteria (1, 2). They play a significant role in maintaining the integrity of thecells and function as the dynamic interface between the bacterium and its sur-roundings. Porins form the channel, which are largely responsible for the overallpermeability of the membrane to nutrients and metabolites. The surface-exposed

Journal of Biomolecular Structure &Dynamics, ISSN 0739-1102Volume 23, Issue Number 2, (2005)©Adenine Press (2005)

G. N. Likhatskaya1,*

T. F. Solov’eva1

O. D. Novikova1

M. P. Issaeva1

K. V. Gusev1

I. B. Kryzhko2

E. V. Trifonov2

E. A. Nurminski2

1Pacific Institute of Bioorganic

Chemistry of Far East Branch of Russian

Academy of Sciences, Vladivostok

pr. 100 let Vladivostoku 159

690022, Russia 2Institute for Automation and Control

Processes of Far East Branch of Russian

Academy of Sciences

5 Radio Street

Vladivostok, 690041, Russia

1

*Fax: (4232)-314050Email: galin@piboc.dvo.ru

regions (external loops) of porins also serve as receptors for interaction withphages, colicins, and antibodies. During their interaction with the host, porins frompathogenic bacteria elicit immune response and have diverse biological activitieson several eukaryotic cell types (3). The role of porins in the pathogenesis ofGram-negative infections still needs to be clarified. Despite approximately 25years of extensive studies, bacterial porins are still at the center of researchers’attention as mechanisms underlying important properties of porin channel and bio-logical activities of porins both in vitro and in vivo are poorly understood.

Knowledge of the proteins structure is crucial to understanding their functions andbiological activities. Hence the most important progress in porin study arose fromthe elucidation of three-dimensional structures a number of trimeric porins by elec-tron diffraction and X-ray crystallography (4). However crystal structures areavailable for only a few porins, and therefore it is highly desirable to obtain theinformation about spatial protein structure from its primary sequence. At presentstructure prediction from amino acid sequence data is becoming a usual source ofprotein structural information. This approach was used to predict atomic structuresof the enterobacterial porins (5, 6).

Two representatives of Yersinia genus, Y. pseudotuberculosis and Y. pestis, havebeen studied extensively because of their ability to cause disease in both human andanimals (7). These organisms closely related at the genetic level (8). Moreover, ithas been suggested that Y. pestis is a recently emerged clone of Y. pseudotubercu-losis. Nevertheless, the symptoms of disease caused by the two Yersinia speciesare dramatically different, as are their mechanism of trasmission. Y. pseudotuber-culosis causes gastrointestinal disease, whereas Y. pestis is the causative agent ofplaque, an acute lethal disease.

Earlier the porins from Y. pseudotuberculosis were isolated and characterized con-cerning their immunochemical, functional, and structural properties (9, 10).Recently, the amino acid sequences of the porins from Y. pseudotuberculosis and Y.pestis were deduced (8, 11, 12).

In the work presented here the three-dimensional structures Y. pseudotuberculosisand Y. pestis porins, YPS and YPT, were predicted by homology modeling and thesurface exposed regions of these porins involved in the potential sequential anti-genic determinants were compared. The information regarding the structure-func-tion relationship in the Y. pseudotuberculosis porin was obtained. For this purposethe elaborated studies on the polar interactions stabilizing oligomeric structure ofthe Y. pseudotuberculosis porin and the organization of its pore region in compari-son with those of the E. coli OmpF porin were performed.

Materials and Methods

The tertiary structures of Y. pseudotuberculosis and Y. pestis CO92 porins were pre-dicted using the homology modeling approach. The primary structure of Y. pseudotu-berculosis porin (YPS) with Genbank accasion no AY855840 was the same as report-ed recently (11). The amino acid sequences of Y. pestis CO92 porin (YPT), Y. pseudo-tuberculosis IP 32953 porin (GBP), and Klebsiella pneumoniae osmoporin (Ompk36)were obtained from Swiss-Prot/TrEMBL databases of the ExPASy Molecular BiologyServer (www.expasy.org/sprot/) with the accession numbers Q8ZG94, Q66CG8, andQ48473. The X-ray structures of PhoE and OmpF porins from E. coli (4) were takenfrom the Brookhaven Protein Data Bank (PDB; www.rcsb.org/pdb/) (13) with thePDB codes 1PHO and 2OMF respectively. A multiple sequence alignment was doneusing the 3D-PSSM Server for protein fold recognition using 1D and 3D sequenceprofiles coupled with secondary structure and solvation potential information(www.sbg.bio.ic.ac.uk/~3dpssm/) (14) and the PredictProtein Server for sequenceanalysis (http://cubic.bioc.columbia.edu/predictprotein/).

2

Likhatskaya et al.

The alignment of Yersinia and E. coli porins was obtained with Clustal (matrix:BLOSUM62) using the molecular graphics package Molecular OperatingEnvironment (MOE; Chemical Computing Group; www.chemcomp.com/).Homology models of the YPS and YPT were generated using the program Swiss-PDBViewer 3.7 (SPDBV) (15) and Swiss-Model Server (www.expasy.org/spdbv/;www.expasy.org/swissmod/) (15-17).

The evaluation of structural parameters and the prediction quality of the modeledstructures were done using the programs WHAT_CHECK (18) and MOE. Thesuperposition of the coordinates of Cα-atoms of the theoretical models to porinswith known structures was carried out using the MOE Homology module. The sec-ondary structural elements and Ramachandran plots of porins were obtained usingthe program Molmol (19). Energy minimizations of the models of Yersinia porinmonomers were carried out on the Swiss-Model Server with the GROMOS96IFP43B1 parameters set (20) using steepest descent (200 cycles) and conjugate gra-dient (300 cycles) methods. The structures of YPS and YPT trimers were built fromthe monomers of Yersinia porins based on symmetry information for the biologicalunit of 2OMF porin using Magic Fit module of SPDBV 3.7 (15). Energy mini-mizations of the porin monomers and trimers were carried out by the program GRO-MACS 3.2.1 (www.gromacs.org/) (21) using steepest descent method and energyminimization options: force tolerance (emtol) equal 100.0 KJ/mol and initial step-size (emstep) equal 0.01. Pdb-files of Yersinia trimers, obtained with programSPDBV, and of 2OMF trimer were edited to make them suitable for GROMACS.

The programs SPDBV 3.7 (15), Molmol (19), and Rasmol (22) were used for visu-alization and analysis of the molecular structure of porins. The analysis of thedimensions of the pores running through porin monomers and prediction of con-ductance properties were carried out using program HOLE release 2.11 (23, 24)implemented on MVS-1000/16 of FEB RAS supercomputing center.

Results and Discussion

Enterobacterial porins involving Yersinia ones show a high degree of homology(above 50%) in term of sequence and structure. This allows the use of the homol-ogy modeling which is the best method for the generation of a structure proposalfor a protein with unknown 3-D structure, to translate Yersinia porins sequencesinto correct structure models of these porins.

The Description Porin Structure Models

Among all the porins with known three-dimentional structure Ompk36 porin fromKlebsiella pneumoniae, OmpF and PhoE porins from E. coli were found to be high-ly homologous to Yersinia pseudotuberculosis and Y. pestis ones in amino acidsequences. By comparing the amino acid sequences of Yersinia porins with those ofKlebsiella pneumoniae and E. coli porins, a set insertions and deletions for align-ment of the sequences has been deduced (Figure 1). The sequence identities betweenYersinia porins and Ompk36, PhoE and OmpF were 63%, 61%, and 56%, respec-tivly. However, PhoE porin but not Ompk36 porin was used as the template to avoidsome of the problems associated with modeling the loop L4. With this alignment asimilar foldings of Yersinia and E. coli porins have been predicted (Figure 1).

The three-dimensional models of Yersinia porins were generated by homology mod-eling method with PhoE porin structure as the template. The models would beexpected to be built with the good level of accuracy in the prediction of the positionof side chain atoms as well as main chain ones due to the high amino acid sequencehomology (56-61%) between Yersinia and E. coli porins. The energy values were -23002.141 KJ/mol for YPT and -23206.973 KJ/mol for YPS after energy minimiza-tions in vacuo of the Yersinia porin monomer models on the Swiss-Model Server

3

Homology Models of the Yersinia General Porins

and with the program GROMACS 3.2.1 (21) using steepest descent method. A com-parison of structures for obtained models and E. coli porins was made by superpo-sition of the coordinates of Cα-atoms of each structure to each other using the MOEHomology models module and the results are presented in Table I.

The difference in structural features between homology models of the Yersinia andE. coli porins was insignificant. The RMSD values computed between eachmodel and PhoE as a template were < 0.5 Å (Table I). The accuracy of the theo-retical models is high and then resolution of Yersinia porin models obtained cor-responds to template resolution.

In line with the predicted models Yersinia porins reveal common structural elementswith the other porins (4). The porin subunit monomer was determined to consist of16th β-strands with lengths ranging from 8 to 17 residues. The amino and carboxytermini of the porin sequences are linked by a salt bridge within the 16th β-strand,thus forming a pseudo-cyclic structure. The 16th β-strands form a completely anti-parallel β-barrel, enclosing a pore. All β-strands are connected to their neighbors byloops. The 8 loops at one end of the barrel (the smooth end) are short (2 to 6residues) whereas the 8 loops at the other end of the barrel (the rough end) are gen-erally longer (8 to 30 residues, with an average of 14 residues). The loop L2 pro-trudes from the barrel. The longest loop L3 folds into barrel leaving a gap in thewall between strands β4 and β7 that is than filled in by connecting loop L2* (aster-

Figure 1: Sequence alignment between theOmpk36 osmoporin from Klebsiella pneumoni-ae, the OmpF and PhoE porins from E. coli andthe porins from two strains of Y. pseudotubercu-losis (KS 3058, Russia, serotype I – YPS, and IP32953, France, serotype I – GBP), and Y. pestisCO-92 (YPT). β-Strands (β1- β16) and extra-cellular loops (L1-L8) are indicated as those ofOmpF. Asterisks denote different residues inYPS and YPT sequences, and grey colourmarks different residues of YPS and GBP.

4

Likhatskaya et al.

isks denote residues from an adjacent monomer) of a neighboring monomer. Theother external loops are tightly packed and extended towards the barrel axis.

Since porins are present as homotrimers in vivo, any conformational analysis ofporin will be more meaningful only if it based on the trimer context. The Yersiniaporin trimers were generated using the models of monomers and the symmetryinformation of E. coli OmpF porin trimer. Energy minimization of Yersinia trimerswere performed with the program GROMACS. The monomer subunit is found tobe non-uniform in height and its shorter sides are included in trimer interface.

The orientation of the porin trimers from Yersinia within the outer membrane wasinferred from the OmpF porin membrane topology, which was earlier establishedby comparative analysis of the atomic structure with the results of the investiga-tions on point mutation interference on phage and antibody binding (25). The topo-logical information deduced for OmpF, whereby the rough end of the barrel isexposed at the cell surface but the smooth end is located at the periplasmic side,was transferred to the Yersinia porins.

The superposition of the monomers of porins showed the most of the barrelresidues in Yersinia porins superimpose very well with their equivalent in OmpF,but there are marked differences in the loop regions. The most insertions and dele-tions are found in the extracellular loops. In Yersinia porins the loops L1 and L6are shorter by five and by three residues respectively and the loop L8 contains threeadditional residues as compared with the corresponding loops in OmpF. The essen-tial difference in the loop L6 packing between YPS porin and OmpF was found.The tip of the YPS loop L6 is shifted towards the loop L5 in distinction to OmpFloop L6 tip located in proximity with loop L8 (Figure 2).

As a result in YPS the loop L6 does not interact with the loop L8 whereas inOmpF the loop L6 seems to be linked with L8 (putative bonds Lys243/Gly325and Ile240/Leu324). The distinction in arrangement of the loops L6 might be ofadditional interest as these loops were suggested (26) to be involved in the extra-cellular domain which contributes in gating behavior of porins by voltage- andpH-dependent changing their conformation (27). From the above the spatialarrangement of the loops forming the mouth of external vestibule appears tovary between YPS and OmpF.

In order to analyze the channel structures of the YPS and YPTporin monomers their pore radius profiles were determined fromthe predicted structures with the spherical and capsule probesusing the program HOLE (23, 24) (Figure 3). From comparingthese pore radius profiles and OmpF one it is seen that theYersinia porins show the conservative pore geometry. The min-imum effective radiuses were estimated to be 3.63, 3.63, and3.83 Å for YPS, YPT, and OmpF, respectively. The YPS andYPT monomers pore conductance values in 1M KCl evaluatedfrom their 3-D structure were 680 and 670 pS, respectively.These values were higher than the conductance value of 590 pScalculated for OmpF monomer. The experimental conductancefor YPS trimers in 0.1M KCl (in black lipid bilayer) was foundto be 240 pS (about 80 pS for a single monomer) (28). Thus inthe case of YPS, the pore conductance values obtained experi-mentally and predicted seem to be in rather close agreement.

The secondary structure was calculated for the theoretical models of Yersiniaporins (Table II). In the case of YPS, the values predicted are in a good agreementwith the experimental data obtained from circular dichroism spectra by theProvencher method (10, 29).

5

Homology Models of the Yersinia General Porins

Figure 2: The packing of the loop L6 in the YPS andOmpF porins. The extra-cellular loops are shown theloops L6 and L8 are labeled. View monomer unit fromthe top. The model has been clipped.

Figure 3: Pore radius profiles for YPS, YPT and OmpF porin monomers.

The Channel Characterization and Prediction of Transport Activity of the YPS Porin

To explain the common properties and predict characteristic features of YPS porin,some functionally and structurally important regions in the porin were comparedwith those in the E. coli OmpF porin which is the best characreized porin in termsof structure-property relationship.

In the all porins with known structures the longest loop, L3, was found to fold intothe channel and form a constriction zone (“eyelet”) at about mid-height of the chan-nel (2). The unusual organization of the eyelet, with two half-rings of oppositecharge situated face to face in a restricted space, generates an intense electrostaticfield in the pore, controls the solute flux through the channel and governs poreactivity. In E. coli OmpF channel the constriction zone is delimited on one side bythe side chains of two acidic amino acids (Asp113 and Glu117) from L3, and onthe opposite side by a row of three arginines (Arg42, Arg82, Arg132) and Lys16from the barrel wall (4). The basic cluster is stabilized by the side chains of Asn64,Glu62, and Tyr102 (putative H-bonds: Arg42-Asn64, Arg82-Glu62, Arg132-Tyr102). The cluster of the positively charged residues Arg37, Arg77, Arg128,Lys16, is conserved in constriction zone of YPS (Figure 4). Although the residue,Glu57, alone contributes to the cluster stabilization, forming putative H-bonds withArg37 and Arg77. However the cluster formed by two acidic residues is disturbeddue to the substitution of glutamic acid at position 112 (Glu117 in OmpF) to valin,whereas the residue Asp108 (Asp113 in OmpF) remains intact. Interestingly, thereplacement Glu112Val is located within sequence of 111-115 residues (116-120,Pro-Glu-Phe-Gly-Gly-Asp, in OmpF), which forms a turn at the tip of L3 and ishighly conserved among enterobacterial porins (30). From the results of the com-parative analysis it may be suggested that in YPS the arrangement of the chargedresidues within the constriction zone is less rigid than in OmpF.

The loss of the negatively charged residue in the channel lumen would be expect-ed to effect on the channel properties of YPS porin. Consistent with this assump-tion was the observation that E. coli OmpF mutants Glu117Gln andAsp113Asn/Glu117Gln, which had virtually unchanged pore geometry, showedessentially reduced conductance and cation selectivity and increased pore size (31).The lack of the cation selectivity found experimentally for YPS porin might beresult of the substitution Glu112Val.

The porin functional properties are also affected by the position and the flexibilityof the loop L3 in the pore (32). These characteristics of the loop L3 were suggest-ed to determine by the hydrogen bond network and by the salt bridges between thetip and the root of L3 and the adjacent barrel wall (33). In OmpF a major player inhydrogen binding is the barrel wall residue Asp312 whose carboxyl group wouldbe in proximity of the peptidyl hydrogens of Glu117 and Phe118 at the L3 tip (4).This residue is conserved in YPS as Asp307. In spite of the replacement Asp/Valat the position 112 the hydrogen-bond network involving Asp307, Val112, andPhe113 (Asp312, Glu117and Phe118 in OmpF) is conserved in YPS, although theadditional hydrogen bond of the residue Val112 with Asp307 (in OmpF the inter-

6

Likhatskaya et al.

Figure 4: Structure of the channel constriction ofYPS porin. View from the extra-cellular side onto themonomer with strands depicted as broad ribbons. Themodel has been clipped to allow a full view of theconstriction.

action between the side chains of Asp312 and Glu117) and also H-bonds withTyr305 (Tyr310 in OmpF) and Tyr22 are lost. However the major portion of theH-bonds between the tip of L3 and the adjacent barrel wall, presenting in OmpF,are conserved in YPS. Hence the discussed above alterations in the amino acidresidues polar interactions are unlikely to effect markedly on the position and sta-bility of the loop L3 in the YPS channel.

In OmpF the putative salt-bridges between residues Asp126, Asp107, and Arg100,Arg140 play a key role in binding the root of the loopL3 with the barrel wall. Atthe root of the analogue loop L3 of YPS the residues Asp122 and Asp102 are pre-dicted to be also within salt-bridges distances from amino groups of Arg95 andArg136 at the barrel wall. However hydrogen bonding between the root L3residues Ser 121, Asp 122 (Ser 125, Asp126 in OmpF) and the loop L4 which wasshown to occur in OmpF is disrupt in YPS due to the replacements of the arginineresidues at the positions 163 and 164 (Arg 167 and Arg 168 in OmpF) at the rootof the loop L4 to Lys and Glu, respectively.

The discussed above differences between YPS and OmpF in the polar interactions ofthe loop L3 with the barrel wall are unlikely to affect markedly the position and sta-bility of this loop in the YPS channel in comparison with those of L3 in OmpF one.

The loop L3 of YPS similarly the analogous OmpF loop seems to attach tightly tothe barrel wall (14-15 polar and hydrophobic putative bonds per 30 amino acidresidues of L3) that suggests the its rigid fixation in the channel. Hence the chan-nel property, such as the spontaneous gating activity, which is dependent on themotion or flexibility of the loop L3 (33), appears to be slightly expressed in YPSand to be closely related to that in OmpF. Indeed the mutant porins in which tightinteractions between L3 and barrel are lost showed notable increase of this typeactivity when compared with wild type porin (34).\

The functional activity of porins is greatly affected by modulators such aspolyamines and acidic pH (35, 36). The phenomenon of the pore activity modula-tion seems to be significant in bacterial physiology. In OmpF three residues, Asp113,Asp121, and Tyr294, are proposed to be putative sites of interaction between thechannel and polyamine-spermine (37) and the residue Arg132 is shown to be part ofpH sensor (38). In YPS the corresponding residues (Asp108, Asp116, Tyr289, andArg128) are conserved that supposes availability of regulation of the channel func-tion of this porin by above modulators. The substitution Glu112Val (Glu117 inOmpF) in YPS loop L3 could alter only slightly or not all the interaction ofpolyamine molecules with the porin and the pH response of the porin, as in OmpFthe negatively charged residues of the L3 loop other than Asp113 and Asp121(Asp108, Asp116 in YPS), do not contribute in the sensitivity of the channel topolyamine (35) and the residue Glu117 does not play a major role in pH sensing (38).Our experimental data on the inhibitory effect of acidic pH on YPS functional prop-erties are consist with predicted pH dependent regulation of the porin activity.

Predicted Stability and Integrity of the YPS Trimer

The porin trimers, unlike their monomeric counterparts, are extremely stable in thatthey withstand heat, detergents and exposure to proteolysis (39).

Some experimental data suggest the difference in trimer stability between YPS andOmpF. The YPS trimer was shown to dissociate partly in during isolation under stan-dard conditions (40). The OmpF trimer seems to be more stable under similar condi-tions. Furthermore, native OmpF trimer was dissociated at 75°C (41) while the disso-ciation temperature of the YPS porin trimer was about 67°C. This raises the questionof whether there is the structural basis of these differences. In this connection someregions of the YPS and OmpF structures contributing in trimer stability were compared.

7

Homology Models of the Yersinia General Porins

The stability and structural integrity of the trimers are determined by both intra- aswell as intersubunit interactions. The hydrophobic core around the symmetry axisthat exists within trimer contributes in the robustness of trimeric porins, as remov-ing specific monomer-monomer interactions by sites-directed mutagenesis resultsin a dramatic decrease in stability in the case of OmpF and PhoE (42). In YPS thehydrophobic interface between monomers of the trimeric structure is well con-served. In the monomer-monomer interface there is the structure element includ-ing N and C termini connected by the salt-bridge. It plays the important role in thetrimer stability as deletion or mutation of C-terminal residue, which is phenylala-nine, was shown to lower trimer stability (43), whereas deletion of the 16th β-strandabolish trimer formation (44). This structure motif is also conserved in YPS.

A central domain of OmpF involving the loop L3 and the eyelet region seems to bestrategic for stable trimerization of the porin, and the replacement Gly119(Gly119Asp) in this region reduces drastically trimer stability (45). The corre-sponding residue in YPS, Gly114, is conserved.

The rigid structure formed by the closely packed ionizable residues in constrictionzone of the pore, and also the loop L3 bonded tightly to the barrel wall are consid-ered to maintain structural integrity of the porin monomer and as a consequence thestability of the trimer as a whole. Indeed the E. coli mutants OmpFs where up to15 residues in the loop L3 are deleted still retain their functional activity (46)though they seem less stable (47). In the case of YPS, the conservation of the tightfixation of the loop L3 in the channel (as was discussed above), even with adecrease in rigidity of the arrangement of the ionizable residues in the channel eye-let due to the loss the negatively charged residue seems to result in the retention ofthe barrel rigidity and suggests the YPS and OmpF porins similarity in their barrelstability. However the loop L3 appears to play a less significant role in the struc-tural integrity of porin than the loop L2.

Undoubtly, the loop L2, which reaches between L2* and L4* of an adjacent monomer,thus filling the gap left by L3*, contributes strongly in high stability of porin trimersby polar and ionic intersubunit interactions (48). In OmpF the residue Glu71 on theloop L2 is integrated into an ionic network and forms salt bridges and hydrogen bondswith Arg100* and Arg132* on the channel wall in the adjacent subunit. The studieson the Glu71Ala, Glu71Gln, Arg132Ala, and Arg132Asp mutants showed that theloss of the salt bridge between Arg132 and Glu71 is responsible for a decrease in thethermal stability of the trimers (48). This structural motif involving the residue Glu66and residues Arg95*, Arg128* of the neighboring subunit is conserved in YPS.

The internal loop conformation structure may affect also the barrel stability andshape. In OmpF the residue Asp74 forms intra-loop H-bonds and is thought to con-tribute to the internal stabilization of the loop. The carboxylate of Asp74 interactswith the main chain amide and the side-chain hydroxyl of Ser70 at the L2 tip there-by tying them together. In YPS this interaction is lost as both of these residues,Asp69Gly and Ser65Ala, are substituted. In the Asp74Ala OmpF mutant in whichthis hydrogen bond is also lost the tip of the loop L2 becomes less kinked (48). Thesimilar alteration of the loop L2 conformation appears to occur also in YPS.However it seems unlikely that this has marked effect on the YPS stability.

The interaction of the loop L2 of one monomer with the loops L2*, L3*, L4*of theadjacent subunits appears to add also to the trimer stability. The hydrogen bondingbetween the loops L2 and L3* in OmpF (Glu71/Asp126) is conserved in YPS (puta-tive H-bond Glu66/Asp122). Whereas the interaction of the loop L2 with L4* is lostin YPS due to the substitution Asp/Gly at the position 69 of L2 (residue 71 in OmpF).

In the case of OmpF the loops L2 of the neighboring monomers are bound by hydro-gen bonds. The similar interactions between the same loops were modeled in YPS.

8

Likhatskaya et al.

In both of these porins the residues (Asn74 in YPS and Asn79, Gln76 in OmpF)involved in the hydrogen binding are located at the root parts of L2 facing the inter-monomeric region of the trimer at the three-fold trimeric axis. The side chains ofthese residues protrude into this space between monomers (Figure 5). However thereare differences between these porins in the arrangement of the hydrogen-bond net-works linking the loops L2 of adjacent monomers. In YPS the side chains of theresidues Asn74 participating in the loop L2 interaction are extended towards thetrimer axis and their amide groups are present in vicinity that results the hydrogenbonding between the corresponding atoms (OD1 and ND2) of Asn74 of the adjacentmonomers (Figure 5). It should be noted that these H-bonds are expected to be weak-er than traditional H-bonds in peptides as they are 3.12 Å in length between N and Oatoms whereas typically this distance is 2.8-3.0 Å (48). In OmpF the interactionbetween the side chain amides of Asn79 (Asn 74 in YPS) of the adjacent monomersis unlikely to occur as the side chains are extended along the barrels walls, making theincreased distance between the corresponding atoms (OD1 and ND2) of Asn79 of theadjacent monomers (Figure 5). However the OmpF residues Asn79 are hydrogenbonded to the residues Gln76* of the loops L2 of the adjacent monomers (putative H-bonds between the Asn79 main chains and Gln76* side chain amides) (Figure 5).

In addition the intra-loop hydrogen bond seems to exist between the main chains ofthese residues. The ensembles of H-bonds involved in the loop L2 interaction inYPS and OmpF seem to be essentially different in stability. The differencesbetween the YPS and OmpF trimers in this structural element located at theentrance to hydrophobic pocket at the trimer threefold axis so as in the other onesdiscussed above would contribute to the distinct trimer stability of these porins.

The other structural element, which appears to add to the trimer stability is in inter-monomer region at the trimer bottom end. The residues Asn9 and Tyr336 (Tyr338in OmpF) from the strands of the adjacent monomers linked with by hydrogen bondsare involved in this structure. This structural motif is conserved in both porins.

Compared in details of the structural organization the E. coli OmpF and YPSregions involved in subunit association and pore lumen are shown to be highly con-served ones. The revealed structural differences between these porins determineespecial features making each porin to be unique. The predicted functional prop-erties of Y. pseudotuberculosis porin are suggested to examine in experiments.

Comparison of Yersinia Porins Structures

The highest degree of sequence identity (92%) was found for YPS and YPT. TheY. pestis porin seems to be closely similar to Y. pseudotuberculosis porin in theregions involved in subunit association and pore lumen. The differences betweenthese porins were observed at comparison of their external loops. In YPS and YPTthe corresponding external loops posses the same length and with the exception of

9

Homology Models of the Yersinia General Porins

Figure 5: H-binding between the loops L2 from adja-cent monomers in Y. pseudotuberculosis (YPS) andOmpF porins. For clarity, only the backbones of theloops L2 and putative hydrogen bond network con-necting the amino acid residues of the loops L2 fromthe adjacent monomers in trimer are shown. Dashedlines indicate possible H-bonds. Porin monomers arelabeled M1, M2, and M3.

the loops L6 and L8 show high identity in sequence that suggests a high similarityin their conformation. The loops L6 and notably the loops L8 in these porins aredifferent in sequence showing 71.4 and 52.9% of identity respectively.

The differences between porins found mainly in their external loops result in the dif-ferences in the distributions of ionizable residues in the vestibule on the extracellu-lar side of the channel (Figure 6). In YPT the aspartic acid residues 279 and 317 atthe loops L7 and L8 replace the uncharged residues (Asn 279 and Leu317) in YPSand introduce the additional negative charges. Additionally, two basic residues,Lys315 and Arg323, of YPS at the loop L8 are lost in YPT due to their substitutionsto acidic and neutral residues (Glu315 and Gly323). As a consequence the vestibuleof YPT seems to be more negative than that of YPS even though the loss of twoacidic residues (Glu320 and Asp328 of YPS at L8) occurs in YPT. This might resultin the differences in the channel transport activity between these porins as thearrangement of charged residues clustered in the external vestibule plays a role inpreselecting the type of charged molecules or ions to be transported more favorablyacross the membrane pore (2). Thus, in spite of the highest degree of the structuralsimilarity of Yersinia porins the differences in their functional properties seem tooccur. This might be expected. Actually, because of the importance of porins in theinteraction of bacteria with the environment, wide differences in habitat between Y.pseudotuberculosis and Y. pestis should be reflected in their channel properties.

Similar to other bacterial porins, YPS has been shown to bepotential surface antigen and to elicit protective immuneresponse in animal model (50, 51). However the location of spe-cific immunodominant epitopes in YPS molecule was not stud-ied. In the porins some linear B- and T- cell epitopes were foundwithin the loops (52). The Yersina porin loops L1-L5 seem toharbour the common epitopes, which are likely candidates forcross-reactivity between these porins. The amino-acid polypep-tide sequences included in these loops might be involved in pro-tective determinants. In Salmonella typhi porin the syntheticpeptides corresponding to the loops (L6 and L7) were shown toinduce protective immunity (53). The potentially immunoactivepeptide fragments of the porin loops might be synthesized inorder to create a synthetic vaccine against Yersinia infections.

The significant structural differences between Yersinia porins in the L6 and L8loops make YPS and YPT distinguishable in terms of their varying antigenic prop-erties. Synthetic peptides corresponding to L6 and L8 loops could be the possiblecandidates to develop assays for differentiation of these Yersinia species. The sur-face-exposed putative epitopes of Yersinia porins might be potentially useful anti-genic regions for diagnostic assays and vaccine development.

In summary, Yersinia porins are very similar in structure to other non-specific porins,having the same features of overall fold and disposition of loop L3. Their pore sizesare similar to that of other non specific enterobacterial porins. The main differencebetween Yersinia and other non-specific porins is the distribution of charges at the poreconstriction zone. There is the substitution, Glu replaced by Val, in a highly conservedpentapeptide (Pro-Glu-Phe-Gly-Gly-Asp), located in the loop L3 tip, that disturbs thefunctionally important cluster of the acidic amino acids in the constriction site. Somedifferences in polar interactions helping in trimer stability between the YPS and OmpFporins seem to be found suggesting the differences in their trimeric stability. The struc-tural features of Yersinia porins suggest their distinctive properties yet to be studied.

The differences between porins from Yersinia were observed mainly in their externalloops. A great divergence between these proteins was found in the loop L8, suggest-ing a possible role of this loop in formation of specific cell surface-exposed epitopes.

10

Likhatskaya et al.

Figure 6: Distribution of the ionizable residues facing the channel vestibulein the Y. pseudotuberculosis (YPS) and Y. pestis (YPT) porins. It is shown theacidic and basic residues that are not identical in YPS and YPT. Side view ofmonomeric units tilted (about 30°) about vertical axis towards the viewer.

Acknowledgements

The program of Presidium of RAS “Molecular and cell biology” and FEBRASgrant 05-III-A-05-069 supported this work.

References and Footnotes

11

1.2.3.4.

5.6.

7.8.9.

10.

11.

12.

13.

14.15.16.

17.18.19.20.

21.22.23.24.25.

26.27.28.

29.30.31.

32.

33.

34.35.36.37.38.39.40.

41.42.43.44.

H. Nikaido. Microbiol. Mol. Biol. Rev. 67, 593-656 (2003).B. K. Jap and P. J. Walian. Physiol. Rev. 76, 1073-1088 (1996).M. Galdiero, M. Vitiello, and S. Galdiero. FEMS Microbiol. Letters 226, 57-64 (2003).S. W. Cowan, T. Schrimer, G. Rummel, M. Steiert, R. Ghosh, R. A. Pauptit, J. N. Jansonius,and J. P. Rosenbusch. Nature 358, 727-733 (1992).A. Arockiasamy, S. Krishnaswamy. J. Biomolec. Struct. Dyn. 18, 261-271 (2000).S. Alberti, F. Rodriquez-Quinones, T. Schirmer, G. Rummel, J. M. Tomas, and J. P.Rosenbusch. Infect. Immun. 63, 903-910 91995).R. R. Brubaker. Curr. Top. Microbiol. Immunol. 57, 111-158 (1972).J. Parkhill, B. W. Wren, N. R. Thomson, R. W. Titball, et al. Nature 413, 523-527 (2001).G. N. Likhatskaya, O. D. Novikova, T. F. Solovjeva, and Yu. S. Ovodov. Biol. Membrany(Russian) 2, 1218-1224 (1985).O. D. Novikova, G. M. Frolova, T. I. Vakorina, Z. A. Tarankova, V. P. Glasunov, T. F.Solovjeva, and Yu. S. Ovodov, Bioorg. Khim. (Russian) 15, 763-772 (1989).M. P. Isaeva, O. D. Novikova, T. F. Solov’eva, V. A. Rasskasov, and S. Kh. Degtyarev. Adv.Exp. Med. Biol. 529, 257-260 (2003).P. S. G. Chains, E. Carniel, F. W. Larimer, J. Lamerdin, P. O. Stoutland, W. M. Regala, A.M. Georgescu, L. M. Vergez, M. L. Land, V. L. Motin, R. R. Brubacker, J. Fowler, J.Hinnebusch, M. Marceau, C. Medigue, M. Simonet, V. Chenal-Francisque, B. Souza, D.Dacheux, J. M. Elliott, A. Derbise, L. J. Hauser, and E. Garcia. Proc. Natl. Acad. Sci. USA103, 13826-13831 (2004).H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H. Weissig, I. N. Shindyalov,and P. E. Bourne. Nucleic Acids Research. 28, 235-242 (2000).L. A. Kelley, R. M. MacCallum, M. J. E. Sternberg. J. Mol. Biol. 299, 499-520 (2000).N. Guex and M. C. Peitsch. Electrophoresis 18, 2714-2723 (1997).T. Schwede, J. Kopp, N. Guex, and M. C. Peitsch. Nucleic Acids Research 31, 3381-3385 (2003).M. C. Peitsch. Bio/Technology 13, 658-660 (1995).R. W. W. Hooft, G. Vriend, C. Sander, E. E. Abola. Nature 381, 272 (1996).R. Koradi, M. Billeter, K. Wethrich. J. Mol. Graphics 14, 51-55 (1996).W. F. van Gunsteren, S. R. Billeter, A. A. Eising, P. H. Hünenberger, P. Krüger, A. E. Mark, W.R. P. Scott, I. G. Tironi. Biomolecular Simulation: The GROMOS96 Manual and User Guide.VdF: Hochschulverlag AG an der ETH Zürich and BIOMOS b.v., Zürich, Groningen, (1996).E. Lindahl, B. Hess, D. van der Spoel. J. Mol. Mod. 7, 306-317 (2001).R. A. Sayle, E. J. Milner. White Trends in Biochemical Sciences 20, 374-376 (1995).O. S. Smart, J. M. Goodfellow, B. A. Wallace. Biophys. J. 65, 2455-2460 (1993).O. S. Smart, J. Breed, G. R. Smith, M. S.P. Sansom. Biophys. J. 72, 1109-1126 (1997).J. Tommassen. In Membrane Biogenesis. Heidelberg, Germany: Springer-Verlag, 351-373 (1988).D. P. Tieleman and H. J. C. Berendsen. Biophysical J. 74, 2786-2801 (1998).D. J. Muller and A. Engel. J. Mol. Biol. 285, 92-94 (1999).O. D. Novikova, G. N. Likhatskaya, G. M. Frolova, O. P. Vostrikova, V. A. Khomenko, N. F.Timchenko, T. F. Solov’eva, and Yu. S. Ovodov, Biol. Membrany (Russian). 7, 453-461 (1990).C. W. Provencher and J. Glöcker. Biochemistry 20, 33-37 (1981).D. Jeanteur, J. H. Lakey, and F. Pattus. Mol. Microbiol. 5, 2153-2164 (1991).P. S. Phale, A. Philippsen, C. Widmer, V. P. Phale, J. P. Rosenbusch, and T. Schirmer.Biochemistry 40, 6319-6325 (2001).P. S. Phale, T. Schirmer, A. Prilipov, K.-L. Lou, A. Hardmeyer, and J. P. Rosenbusch. Proc.Natl. Acad. Sci. USA 94, 6741-6745 (1997).A. Karshikoff, V. Spassov, S. W. Cowan, R. Ladenstein, and T. Schirmer. J. Mol. Biol. 240,372-384 (1994).N. Liu and A. H. Delcour. Protein Engineering 9, 797-802 (1998).H. Samartzidou and A. H. Delcour. J. Bacteriol. 181, 791-798 (1999).J. C. Todt, W. J. Rocque, and E. J. McGroarty. Biochemistry 31, 10471-10478 (1992).R. Iyer, Z. Wu, P. M. Woster, and A. H. Delcour. J. Mol. Biol. 297, 933-945 (2000).N. Liu and A. H. Delcour. FEBS Lett. 434, 160-164 (1998).B. Lugtenberg and L. Van Alphen. Biochim. Biophys. Acta 737, 51-115 (1983).O. P. Vostrikova, O. D. Novikova, N. Yu. Kim, G. N. Likhatskaya, and T. F. Solov’eva. Adv.Exp. Med. Biol. 529, 261-263 (2003).D. Fourel, S. Mizushima, A. Bernadac, and J.-M. Pages. J. Bacteriol. 175, 2754-2757 (1993).P. Van Gelder and J. Tommassen. J. Bacteriol. 178, 5320-5322 (1996).M. Struyve, M. Moons, and C. Tomassen. J. Molec. Biol. 218, 141-148 (1991).D. Bosch, M. Scholten, and C. Tomassen. J. Molec. Genet. 216, 144-148 (1989).

12

Likhatskaya et al.

45.46.47.48.

49.50.

51.

52.

D. Fourel, S. Mizushima, A. Bernadac, and J.-M. Pages. J. Bacteriol. 175, 2754-2757 (1993).S. A. Benson, J. L. L Occi, and B. A. Sampson. J. Molec. Biol. 203, 960-971 (1988).W. J. Rocque and E. J. McGroarty. Biochemistry 29, 5344-5351 (1990).R. S. Phale, A. Phillpsen, T. Keifhaber, R. Koebnik, V. P. Phale, T. Schirmer, and J. P.Rosenbusch. Biochemistry 37, 15663-15670 (1998).I. L. Karle. J. Mol. Struct. 474, 103-112 (1999).O. Yu. Portnyagina, O. D. Novikova, O. P. Vostrikova, and T. F. Solovjeva. Byul. Eksp. Biol.Med. (Russian) 128, 437-440 (1999).I. V. Darmov, I. V. Marakulin, I. P. Pogorelsky, O. D. Novikova, O. Yu. Portnyagina, and T.F. Solovjeva. Byul. Eksp. Biol. Med. (Russian) 127, 221-223 (1999).K. M. Williams, E. C. Bigley, and R. B. Raybourne. Infect. Immun. 68, 2535-2545 (2000).J. Paniagua-Solis, V. Sanchez, V. Oritz-Navarrete, C. R. Gonzalez, and A. Isibasi. FEMSMicrobiol. Lett. 141, 31-36 (1996).

Date Received: January 30, 2005

Communicated by the Editor Ramaswamy H Sarma