Vous êtes sur la page 1sur 11

Journal of Biomolecular Structure &

Dynamics, ISSN 0739-1102


Volume 18, Issue Number 2, (2000)
©Adenine Press (2000)

Homology Model of Surface Antigen OmpC From A. Arockiasamy


Salmonella typhi and its Functional Implications and S. Krishnaswamy*
Bioinformatics Centre,
http://www.adeninepress.com Department of Genetic Engineering,
School of Biotechnology,
Abstract
Madurai Kamaraj University,
Homology based 3D structural model of the immunodominant major surface antigen OmpC Madurai-625 021, India
from Salmonella typhi, an obligatory human pathogen, was built to understand the possible
unique conformational features of its antigenic loops with respect to other immunologically
cross reacting porins. The homology model was built based on the known crystal structures
of the E. coli porins OmpF and PhoE. Structure based sequence alignment helped to define
the structurally conserved regions (SCRs). The SCR regions of OmpC were modelled using
the coordinates of corresponding regions from reference proteins. Surface exposed variable
regions were modelled based on the sequence similarity and loop search in PDB. Structural
refinement based on symmetry restrained energy minimization resulted in an agreeable
model for the trimer of OmpC. The resulting model was compared with other porin struc-
tures, having b-barrel fold with 16 transmembrane β-strands, and found that the variable
regions are unique in terms of sequence and structure. A ranking of the loops taking into
account the antigenic index, the sequence variability, the surface accessibility in the context
of the trimer, and the structural variability suggests that loop 4 (151-172), loop 5 (194-218)
and loop 6 (237-264) are the best ranked B-cell epitopes. The model provides possible expla-
nations for the functional and unique immunological properties associated with the surface
exposed regions and outlines the implications for structure based experimental design.

Introduction

Porins are pore forming proteins present as trimers on the outer membrane of Gram
negative bacteria which facilitate intake and disposal of nutrients and waste materi-
als of size <600 Da respectively. These water filled channels allow only hydrophilic
solutes to pass through the pore (1). Studies using monoclonal antibodies raised
against native porin trimers (2) as well as synthetic epitopes showed that porins are
highly cross-reactive and the crucial domains are conserved (3). Porins are shown to
be B-cell polyclonal activators (4) and also shown to induce cytokines in cultured
human monocytes (5). Also porins have attracted much attention, as they are poten-
tial cell surface antigens, which can display heterologus sequences on the bacterial
cell surface (6) towards the development of oral vaccines (7). Defence molecules like
lactoferrins and C1q, a component of complement system, were shown to bind to
enterobacterial porins (8,9). Salmonella porins were found to induce both humoral
and cell mediated immunity in animal models (10,11). A recent study demonstrated
that outer membrane porin could induce apoptosis in epithelial cell line SVC1 (12).
The surface exposed regions of these outer membrane proteins are believed to play a
crucial role in the survival of the bacterial pathogens inside the host system (13).
Results from these studies suggested us that investigations of variable surface
exposed regions of porins might help in understanding the diverse role of these pro-
teins in biology, especially in host interaction, pathogenesis and immunology. *Phone: +91-452-859141;
Fax: +91-452-859105;
E-mail: krishna@mrna.tn.nic.in
Crystal structures of few porins (14,15) from E. coli and other bacteria show a similar
three-dimensional architecture. Enterobacterial porins show a high degree of homology 261
262 in terms of sequence (16) and structure (17). General diffusion pores like OmpF and
PhoE have a 16 stranded β-barrel structure and specific porins like LamB and ScrY show
an 18-stranded barrel with surface exposed loops. High degree of sequence homology
Arockiasamy and Krishnaswamy among the porins of enterobacteria allowed a homology based model built for
Salmonella typhi OmpC (357 aa without the signal peptide and 39 kDa Mol.Wt.
/monomer) (18). Earlier we have predicted the sequential B-cell epitopic regions of S.
typhi OmpC using sequence analysis (19). Subsequently we have crystallised OmpC
(20) and the crystals diffract to low resolution. The homology based model will be used
in the structure determination of OmpC using Molecular Replacement (MR) method. In
the absence of crystal structure, the model is found useful to interpret the results obtained
from chemical modification and immunochemical studies, towards conformational epi-
tope mapping of S. typhi OmpC with Salmonella and enterobacterial porin specific mon-
oclonal antibodies. The functional implications of the model are discussed in this study.

Materials and Methods

Model building and energy minimization were carried out on a Silicon Graphics
Iris Crimson workstation using Biosym InsightII 95.0/Homology and Discover
modules (Biosym Technologies, San Diego). The energy calculations were done
initially with the consistent valence force field (CVFF) and later on with AMBER
force fields respectively (21). Electrostatic and solvent accessibility calculations
were done with DELPHI and SOLVATION modules of InsightII. Trimer generation
and final refinements were done using XPLOR (22). Molscript (23) was used to
generate the figures. WGCG package (24) was used for sequence analysis and JOY
(25) was used to represent the multiple sequence alignment with structural details.
Sequences used in this study were taken from EMBL and SwissProt and the refer-
ence structures were from PDB (1OMF, 1PHO).

Structure Based Sequence Alignment

Structural superposition of reference structures E. coli OmpF and PhoE identified the
structurally conserved regions (SCRs) among them. The multiple sequence align-
ment of S. typhi OmpC and E. coli OmpF & PhoE identified 9 homologous regions
with gaps inserted at SCRs between OmpF and PhoE. Final alignment was made with
the help of secondary structure information of OmpF and PhoE. Manual modification
of the multiple sequence alignment was done to make sure there was no inserted gaps
in the SCRs. Seven variable regions were found in the overall alignment.

Modelling of Structurally Conserved Regions

Coordinates for SCRs of OmpC were assigned from corresponding SCRs of OmpF
and PhoE. Though OmpF and PhoE share structural homology in these regions, to
assign the coordinates for OmpC, the respective SCRs of reference proteins were
selected based on the sequence identity. The residues that differed in OmpC with
respect to the reference structure were labeled as mutated residues for the model
refinement. In case the percentage identity for a SCR of OmpC was same with both
reference SCRs, selection was based on the origin of preceding and following
SCRs of that particular conserved region of OmpC. In the case of SCR, correspon-
ding to the β-strand 16 of OmpC, though it matches better with PhoE, coordinates
were assigned from OmpF since the structure for strand 1 was taken from OmpF
and strands 1 and 16 form a salt bridge giving a pseudo cyclic structure. Reference
structures were superimposed before transferring the coordinates to OmpC model,
to ensure the alignment of structural segments at the junctions.

Modeling of Variable Regions

The regions in between two SCRs were considered as structurally variable regions
(V1-V8) and they show significant difference in terms of sequence as noticed from
the multiple sequence alignment. The variable regions, except for loop 3, are sur- 263
face exposed (loops) in OmpF and PhoE. To model the variable regions in OmpC,
loop search was done using high-resolution structures (>2Å) in PDB. Best-matched
loops were selected based on the following criteria: a) Loops with similar topolo-
Homology Model of
gy and orientation b) Loops with low RMSD at the flanking regions. These vari- Surface Antigen OmpC
able segments were modelled using the backbone structure of the best-matched
loop from the PDB, which resulted in acceptable topology and orientation. Part of
the structure was taken from reference structures for some variable regions. Thus
all the variable regions could be modelled.

Refinement of the Initial Monomer Model

The refinement of the structure was essentially done in different levels to relieve
the short contacts and bad geometry. Side chains of mutated residues were assigned
from the rotamer library. Constrained energy minimization was done at the junc-
tion point between two segments if in case the two segments were taken from
OmpF and PhoE. Side chains of mutated residues in SCR and all atoms in loop
residues were relaxed using a few cycles of energy minimization. The resulting ini-
tial model was energy minimised initially using CVFF and then with AMBER force
fields with steepest descent and conjugate gradient minimisers.

Trimer Generation and Model Assessment

Since the porins are present as trimers in vivo, any conformational analysis of porin
will be more meaningful only if it is based on the trimer context. The S. typhi
OmpC trimer was generated using the symmetry information of E. coli OmpF.
Trimer model was energy minimised using XPLOR to relieve short contacts, which
might have been introduced while trimer generation. Stereochemistry of the model
was analysed using PROCHECK (26). The coordinates of the model have been
deposited in PDB (1IIV).

Electrostatic and Surface Accessibility Calculation

Electrostatic potential was calculated using the DELPHI program. The linear
Poisson-Boltzmann equation was solved, using Amber fractional charges for each
protein atom. Dielectric constant of 2 and 80 were used for inside and outside the
protein respectively. Program SOLVATION was used to calculate the surface
accessibility with the probe radius of 1.4Å. Surface accessibility of both monomers
and trimers were calculated and compared.

Prediction of Sequential Epitopes

Sequential epitopes of S. typhi OmpC predicted earlier, based on sequence align-


ment, has been modified using the 3-D model. The modified ranking strategy is
based on the antigenic index (AI) (27), ratio of variability in terms of sequence [Rv
(seq) = no of non-conserved residues in loop region based on multiple sequence
alignment /total number of residues in loop region] and the ratio of variability in
terms of structure [Rv (str) = no of non-conserved residues in loop region based on
the structural superposition /total number of residues in loop region] and surface
accessibility (SA) in the context of the monomer and trimer.

Results and Discussion

The homology based model was built to find the structurally unique regions of S.
typhi, which are likely to be exposed outside of the outer membrane and to outline
the features of its antigenic loops. Comparison of OmpC model with OmpF and
PhoE crystal structures shows overall structural conservation in the invariant,
membrane spanning, regions and significant differences in the loop regions.
264 Model Building and Validity of the Model

Porins from enterobacteria show high-level sequence similarity though there is less
Arockiasamy and Krishnaswamy homology observed between them and other bacteria (Table I). The level of
sequence similarity is reflected in the structure also (28). All non-specific porins
seem to have a 16-stranded β-barrel structure. Since S. typhi OmpC showed more
homology with E. coli OmpF and PhoE, these crystal structures were taken as ref-
erence to build the OmpC model. Though Rhodopseudomonas capsulatus and
Rhodopseudomonas blastica porin structures are known, they were not included in
reference structures due to poor sequence homology with OmpC.
Table I
The percentage identity (I) and similarity (S) of S. typhi OmpC, E. coli OmpF, E. coli PhoE,
R. capsulatus and R. blastica porin sequences are given.
E. coli E. coli R. capulatus R. blastica
OmpF PhoE porin porin
S. typhi I 63 64 22 22
OmpC S 77 78 48 51
E. coli I - 65 19 22
OmpF S - 80 48 49
E. coli I - - 18 23
PhoE S - - 49 48
R. capulatus I - - - 28
porin S - - - 55

E. coli OmpF and PhoE structures were compared and 8 structurally conserved
regions, 4 variable regions and a single residue insertion in variable region 3 (loop 3)
were found. These two structures were superimposed with RMSD of 0.76Å over
1244 backbone atom pairs. The SCR regions cover 92.8% of the total residues and
show 79% similarity in the sequence. SCRs of OmpF and PhoE comprise all mem-
brane spanning and periplasmic facing and part of some of the exposed loop regions.
Alignment of OmpC, OmpF and PhoE picked up 8 homologous regions, at the
sequence level, and helped locating the corresponding SCRs in OmpC. The SCRs
include identical positions and conservatively replaced positions of amino acids in
membrane spanning β-strands and part of other regions. The solvent accessible
regions of the β strands represent the membrane buried segments (Figure 1). The loop
regions are exposed outside of the cell. S. typhi OmpC has some of the largest loops.
Loops 6,7 and 8 show some structurally defined regions. Moreover the number of sta-
bilising side chain and main chain hydrogen bonds are more for the larger loops of S.
typhi OmpC. This probably reflects the minimisation of energy done in the absence
of interacting molecules such as receptors and probably reflects the stabilisation
required. The SCR regions were modelled first and then followed by variable regions.
11 regions on S. typhi OmpC comprising all transmembrane strands were assigned
coordinates from OmpF and PhoE. The variable regions V1, V2, V4, V5 and V6 were
built based on loop searches. Since there was no good match found for V3 using loop
search the structure was subsequently assigned from OmpF based on the sequence
match (71% similarity and 56% identity). V3 was split in to two segments before the
structure was assigned, as there was a gap due to insertion of Serine 121 of PhoE in
that SCR. Since part of the variable regions, at both ends of V7 and V8, of OmpC
showed good match with SCRs of OmpF and PhoE, coordinates were assigned from
PhoE on the basis of higher similarity and the core regions of these loops were
assigned from loop search. The residues (183-184) forming turn-5 were modelled as
SCR by taking the backbone from PhoE. Energy minimization with monomer and
generated trimer resulted in the final model, which was checked using PROCHECK
and the present model (Figure 2a, 2b) represents a valid structure (PDB code: 1IIV).

Model Analysis

S. typhi OmpC folds, as expected from the sequence homology, as E. coli OmpF
and PhoE in the transmembrane and periplasmic facing regions. The OmpC trimer
model was superimposed with OmpF and PhoE trimers at SCR regions with the 265
RMSD of 1.21 and 1.37Å respectively, over 2544 backbone atom pairs.
Superimposed monomers are seen in Figure 3. Variable regions of OmpC, com-
prising all surface exposed loops, are unique in terms of structure. Core regions of
Homology Model of
the membrane spanning b-strands of R. capsulatus and R. blastica could be rea- Surface Antigen OmpC
sonably superimposed with OmpC, OmpF and PhoE (with the RMSD ranging from
1.40 to 1.95Å over 492 backbone atom pairs in monomer level), in spite of the low
sequence homology. The variations in the surface exposed regions could be due to
evolutionary insertions and deletions which would have helped different bacteria to
adjust to their varying environmental conditions and the interaction with their
hosts, in case of animal and human pathogens. For example R. capsulatus and R.
blactica porins have simple barrel structure with short exposed loops unlike in
Gram-negative bacteria where loops are long and are more variable. Variable
regions V4 to V8 in OmpC have more number of amino acids than OmpF and PhoE
indicating that S. typhi OmpC has unique variable surface exposed regions.

Electrostatic potential

The electrostatic potential surface of E. coli OmpF, PhoE and S. typhi OmpC porins
show different spectrum of charge distribution at the cell surface (Figure 4a). The
difference in electrostatic potential at the exposed regions, which might involve in
antigen-antibody, host-parasite interaction, phage attachment and the solute prefer-
ence at the pore entrance, could play a key role in terms of their function, speci-
ficity and antigenic variation. These surface exposed regions can be engineered to
show different spectrum of charges at the cell surface, to understand the role of
charged amino acids in bringing about the host-bacterial interaction and their role
in infection and host immunity. Electrostatic properties of porins can also explain
and distinguish the role of variable regions in terms of physiological relevance
other than immunological importance.

Electrostatic properties of porins provide possible structure based explanation for Figure 2 a & b:The Molscript ribbon diagram of the
their biological specificity towards anions or cations as noted earlier (29,30). This monomer (Figure 2a: View along the membrane) and
specificity is conferred by the charge distribution at the pore entrance and several trimer (Figure 2b: View down the pore) of S. typhi
OmpC. The loops L1 to L8 are marked in the monomer.
mutational and modelling studies on loop 3 indicate the importance of the charged The loops are exposed outside of the cell and the short
amino acids at the pore constriction zone (31). The counterparts of S. typhi OmpC in turns, at the bottom of the β-barrel, face the periplasmic
E. coli and S. typhimurium have been shown to be cation selective. The electrostatic space. M1, M2 and M3 represent the monomers.
2a 2b
266 potential calculation shows that S. typhi OmpC could also be cation selective due to
enriched negative charges at the pore entrance (Figure 4b). While going down the
bottom of the pore, one could see only marginal differences between S. typhi OmpC
Arockiasamy and Krishnaswamy and the reference porins in the charge spectrum as observed in the earlier studies.

Surface Accessibility

Comparison of accessibility of the surface exposed variable regions of OmpC,


OmpF and PhoE resulted in interesting observation when they were analysed in
monomer and trimer levels. Though V1, V2 and V4 have more accessible area in
monomer level, there is a reasonable decrease in the context of the trimer (Table II)
due to the fact that these regions are involved in inter subunit interface. V2 of one
subunit comes in contact with the neighbouring subunit and there is a latching by
V2, which helps in trimer stability (32). There are not much variation in the amount
of surface exposed area among V5, V6, V7 and V8 since they are away from trimer
interface and are not involved in inter-subunit interaction. The results of compari-
son helped identify more exposed regions that could be of interest to design further
experiments and also to provide structure based explanation for the experimental
results got elsewhere as discussed below.

Model Implications

Porins have been shown to be potential surface antigens to elicit protective immune
response in animal models. The homology model provides a structural basis of the
antigenic variation, of exposed loops, which make the porins distinguishable in terms
of their varying antigenic properties. The three dimensional model also helps explain
properties of the conformational epitopes and uniqueness in terms of antigenecity.

The folding of buried or TM regions are similar in enterobacteria, while the exposed
loops show the difference. This could explain how monoclonal antibodies raised for
porins of Salmonella, which recognise the porin trimer on the cell surface (33), are
able to distinguish between E. coli and Salmonella porins. This shows the existence
of specific structural features in the overall exposed regions in the context of the
trimer. Since the native trimers were used for immunization and also some MAbs
recognise only the trimer and not the denatured monomers, it is obvious that the
exposed surface area display differential specificity, which results in cross reaction
and uniqueness. The MAbs have been shown to be useful in passive therapy against
Salmonella infection and also demonstrated to be useful in early diagnosis of
typhoid fever (34,35). In the absence of complex crystal structures of Porin-MAbs,
this model has been found useful to explain the experimental results obtained using
chemically modified derivatives of S. typhi OmpC and radio labelled antibodies to
map the conformational epitopes, specific to Salmonella porins and a common epi-
tope shared by enterobacterial porins, on the surface exposed region (36).

Synthetic peptides corresponding to V6 and V7 were shown to induce protective immu-


nity (37). The success of these could also lie in the better presentation afforded by these
loops which show greater stabilisation in terms of hydrogen bonding (Figure 1). The
possibility of synthetic V4, V5 and V8 inducing such response could be tried out since
V4 and V5 are better exposed than other variable regions. Combination of these regions
would be more useful in such studies where one can avoid preparing the whole protein
from pathogenic bacteria. Puente et. al., (6) have shown that S. typhi OmpC can be used
as a carrier for expressing heterologous epitopes on the cell surface. By altering the loops
V4 and V6, to study the expression of 18 amino acid long rotaviral epitope insert, it was
shown that expression in V6 (loop 6) was better than in V4 (loop 4). This could be
explained using the model, as V4 is better exposed than V6 in the structure since the
exposure of V6 is limited by the neighbouring loops (38). V4 of OmpC has 8 extra amino
acids than PhoE and OmpF and also more variable. The possibility of V5, V7 and V8
being useful in such a study is more valid in the structural point of view. This study indi-
267

Homology Model of
Surface Antigen OmpC

Figure 1: The final structure based sequence align-


ment of the S. typhi OmpC, E. coli OmpF and E. coli
PhoE are shown. The sequences are only of the
mature peptide. The SCRs are marked with the con-
tinuous line over the corresponding regions. The
residues forming beta strand are marked in blue,
alpha helix forming residues are in red and the 310
helix formers in maroon. The solvent inaccessible
residues are given in upper case. The solvent acces-
sible residues are given in lower case. The residues
forming hydrogen bond to the mainchain amide are
in bold and those forming hydrogen bond to main-
chain carbonyl are underlined. The loops (V1-V8)
are indicated above the corresponding regions.

Figure 3 a & b: The structural superposition of the


monomers of S. typhi OmpC (red), E. coli OmpF
(green) and E. coli PhoE (blue). The figure was cre-
ated using InsightII of the BIOSYM software. (a)
View along the membrane, (b) View from the
periplasmic space.
268

Arockiasamy and Krishnaswamy

4a

4b

Figure 4 a & b: Comparison of the electrostatic


potentials of porin trimers S. typhi OmpC (lower left
panel), E. coli OmpF (upper left panel) and E. coli
PhoE (upper right panel). The potentials were calcu-
lated and displayed using DELPHI of BIOSYM soft-
ware. The negative electrostatic potential regions are
in red and the positive regions are in blue. The
region, from where the slice of the electrostatic
potential surface is taken, is shown in the lower right
panel with reference to the Cα trace of the S. typhi
OmpC trimer. The variations in electrostatic charge
distribution a) towards the top of the trimers and b)
near the membrane interface regions are shown.
cates the chances of S. typhi OmpC serving as a potential surface antigen to carry epi- 269
topes from other pathogens to develop diagnostic kits and its possible use in multivalent
vaccine design. Structural constraints like packing of the loops may play a key role in
accommodating the extra sequences of different length, in the variable regions.
Homology Model of
Moreover the loops are not exposed vertical to the plane of the membrane but are packed Surface Antigen OmpC
towards the axis passing through the pore, making some of the loops more accessible
than others (Table II). The capacity of these regions to accommodate foreign sequences,
with varying length, needs to be tested out.
Table II
The ranking of the antigenic loops on the basis antigenic index (AI), ratio of variability in terms of sequence [Rv (seq) = no of non-conserved
residues in loop region based on multiple sequence alignment (n2)/total number of residues in loop region] and ratio of variability in terms of
structure [Rv (str) = no of non-conserved residues in loop region based on multiple structural alignment (n2*) /total number of residues in loop
region] and surface accessibility (SA) per atom in the context of the monomer (SAM) and trimer(SAT).

Loop Antigenecity No. of AI/Rv No. of Surface accessibility AI/Rv (seq) Surface AI/Rv (seq) Rank
Index non-conserved (seq) non-conserved /Atom X accessibility X
residues residues (monomer) SAM/Rv (str) /Atom SAT (A2)
(seq) (str) SAM (A2) (trimer)
AI n2 n2* SAT (A2)
V1 (21-33) 1.29 3 0.30 11 5.11 1.30 3.57 0.91 VI
V2 (60-75) 0.93 9 0.52 14 4.85 2.21 2.16 0.98 V
V4 (151-172) 1.25 12 0.68 21 4.50 2.92 3.60 2.33 I
V5 (194-218) 0.96 15 0.57 21 4.07 1.95 4.21 2.02 II
V6 (237-264) 1.00 16 0.57 24 2.40 1.17 2.38 1.16 III
V7 (288-310) 0.82 12 0.43 18 3.16 1.06 3.18 1.07 IV
V8 (327-350) 0.70 10 0.29 13 4.13 0.65 4.04 0.63 VII

Since porins were found to be B-Cell mitogens, attention have been focused on
defined regions involved in this particular activity (39). Results of these reports
indicate that synthetic peptides corresponding to the exposed regions could be
more useful to delineate the residues involved in T-Cell independent B-cell induc-
tion and T-cell proliferation itself. Since defined fragments of E. coli OmpF
(153F:174V) and (157Y:174V) have been shown to activate macrophages (39), the
corresponding regions identified from the structural alignment in S. typhi OmpC
(145F:174Y) are likely to be involved in macrophage activation. This might
explain the involvement of OmpC in infection and pathogenesis since OmpC is
expressed throughout the infection period. Moreover the presence of antiporin anti-
bodies in human typhoid sera makes it a promising candidate antigen to be
explored. Lakshmi Mundkur (34) has shown that purified porin from S. typhi could
induce T-Cells in vitro conditions which again shows the importance of S. typhi
porin in cell mediated immunity. This has been supported by the results that
Salmonella porin could induce lymphocytes to secrete cytokines (5).

The need for the 3D structure was pointed out in the study of mapping of antigenic
determinants of OmpC and OmpD from S. typhimurium (40,41) for accurate inter-
pretation of the MAb reactivity. We had earlier predicted the sequential epitopes of
S. typhi OmpC (19) based on the multiple sequence alignment, crystal structures of
E. coli OmpF and PhoE and partial modelling of two of the variable regions V1 and
V4. In the present study, variable regions have been taken along with 3 flanking
residues at both the ends in order to provide a presentation framework for the epi-
topes. These predicted regions were modified based on the surface accessibility of
the loop regions in the context of the monomer and the trimer from OmpC model.
For example V2 is better exposed in the monomer model. However it is less
exposed in the context of the trimer as it is involved in inter subunit locking. V1,
being the shortest loop, is not well exposed in the trimer and therefore the possi-
bility of being recognised by MAbs in the trimer context could be low (Table II).
V3 is not included in this analysis since it forms the pore constriction in the mid-
dle of the barrel and is not surface exposed. Though V6 forms the longest loop, the
end of the loop goes down towards pore unlike in E. coli OmpF.
270 The ranking of the possible B-cell epitopes (Table II) provides a rational strategy
for choosing between the epitopes. The chances of a particular stretch of a surface
exposed region being the best candidate for experimental applications is tested tak-
Arockiasamy and Krishnaswamy ing into account the antigenic index, variability in terms of sequence alignment, the
surface accessibility of the loops in the monomer context, variability based on the
structural alignment and the surface accessibility of the loop residues in the context
of the trimer. Any particular region, which ranks high based on these conditions, is
considered to be a possible candidate for experimental design and the ranking strat-
egy helps selecting one particular region over the other. The present study suggests
that the best epitopes are V4, V5, V6 and V7 and the less likely ones are V8, V2
and V1, based on the trimer model.

The superposition of three enterobacterial porins: S. typhi OmpC (model), E. coli


OmpF & PhoE and two other from non-pathogenic origins: R. capsulatus and R.
blastica porin, have helped to make structure based multiple sequence alignment.
This has led us to come up with sequence profiles from core regions of all 16 mem-
brane spanning strands, which fall in the SCRs, to predict transmembrane β-strand-
ed proteins from the sequence (42).

Acknowledgement

We are thankful to Profs. VR. Muthukkaruppan and K. Dharmalingam for encour-


agement and help. National Facility for High Resolution Graphics at
Bioinformatics centre, Madurai Kamaraj University, supported by DBT, Govt. of
India is gratefully acknowledged. A. A is a recipient of a fellowship (NET) from
CSIR. The research grant (SP/SO/D32/97) from DST, Govt. of India, is duly
acknowledged.

References and Footnotes

1. H. Nikaido, J. Biol.Chem. 269, 3905-3908 (1994).


2. V.R. Muthukkaruppan, K.S. Nandakumar and V. Palanivel, Immunol. Lett. 33, 201-206 (1992).
3. V. Simonet, M. Mallea, D. Fourel, J.M. Bolla and J.M. Pages, FEMS Microbiol. Lett. 136, 91-
97 (1996).
4. H.M. Vordermeier, H. Drexler and W.G. Bessler, Immun. Lett. 15, 121-126 (1987).
5. M. Cipollaro de L’ero, G. Donnarumma, A. Marcatili and F. Galdiero, Immunology 86, 612-
619 (1995).
6. J.L. Puente, D. Juarez, M. Bobadilla, C.F. Arias and E. Calva, Gene 156, 1-9 (1995).
7. R. Janssen and J. Tommassen, Int. Rev. Immunol. 11, 113-121 (1994).
8. J. Erdei, A. Forsgren and A.S. Naidu, Infect. Immun. 62, 1236-1240 (1994).
9. S. Alberti, G. Marques, S. Hernandez-Alles, X. Rubires, J.M. Tomas, F. Vivanco and V.J.
Benedi, Infect. Immun. 64, 4719-4725 (1996).
10. V. Udhyakumar and V.R. Muthukkaruppan, Infect. Immun. 55, 816-821 (1985).
11. A. Isibasi, V. Oritz, M. Vargas, J. Paniagua, C. Gonzalez, J. Moreno and J. Kumate, Infect.
Immun. 56, 2953-2959 (1988).
12. E. Buommino, F. Morelli, S. Metafora, F. Rossano, B. Perfetto, A. Baroni and M.A. Tufano,
Infect. Immun. 67, 4794-800 (1999).
13. T.F. Murphy, Trends. Microbiol. 2, 427-428 (1994).
14. T. Schirmer, J. Struct. Biol. 121, 101-109 (1998).
15. R. Dutzler, G. Rummel, S. Alberti, S. Hernandez-Alles, P. Phale, J. Rosenbusch, V. Benedi and
T. Schirmer, Structure 7, 425-434 (1999).
16. D. Jeanteur, J.H. Lakey and F. Pattus, Mol. Microbiol. 5, 2153-2164 (1991).
17. G.E. Schulz, Curr. Opin. Struct. Biol. 6, 485-490 (1996).
18. J.L. Puente, V. Alvarez-Scherer, G. Gosset and E. Calva, Gene 83, 197-206 (1989).
19. A. Arockiasamy and S. Krishnaswamy, J. Biosci. 20, 235-243 (1995).
20. A. Arockiasamy and S. Krishnaswamy, FEBS. Lett. 453, 380-382 (1999).
21. S.J. Weiner, P.A. Kollman, D.T. Nguyen and D.A. Case, J. Comp. Chem. 7, 230-252 (1986).
22. A.T. Brunger, X-PLOR (Version 3.1), Yale Univ. Press, New Haven (1993).
23. P.J. Kraulis, J. Appl. Cryst. 24, 946-950 (1991).
24. Wisconsin Package Version 9.0, Genetics Computer Group (GCG), Madison, Wisc.
25. K. Mizuguchi, C.M. Deane, T.L. Blundell, M.S. Johnson and J.P. Overington, Bioinformatics
14, 617-623 (1998).
26. R.A. Laskowski, M.W. MacArthur, D.S. Moss and J.M. Thornton, J. Appl. Cryst. 26, 283-291
(1993).
27. B.A. Jameson and H. Wolf, Comput. Appl. Biosci. 4, 181-186 (1988). 271
28. S.W. Cowan, T. Schirmer, G. Rummel, M. Steiert, R. Ghosh, R.A. Pauptit, J.N. Jansonius and
J.P. Rosenbusch, Nature 358, 727-733 (1992).
29. M.S. Weiss, U. Abele, J. Weckesser, W. Welte, E. Schiltz, G.E. Schulz, Science 254, 1627-1630 Homology Model of
(1991).
30. A. Karshikoff, V. Spassov, S.W. Cowan, R. Ladenstein and T. Schirmer, J. Mol. Biol. 240, 372- Surface Antigen OmpC
384 (1994).
31. K.L. Lou, N. Saint, A. Prilipov, G. Rummel, S.A. Benson, J.P. Rosenbusch and T. Schirmer, J.
Biol. Chem. 271, 20669-20675 (1996) .
32. P.S. Phale, A. Philippsen, T. Kiefhaber, R. Koebnik, V.P. Phale, T. Schirmer and J.P. Rosenbusch,
Biochemistry 37, 15663-15670 (1998).
33. K. Umesh Chandra Reddy, A. Arockiasamy and S. Krishnaswamy (unpublished).
34. Lakshmi Mundkur Ph.D thesis Madurai Kamaraj University, Madurai, India. (1995).
35. K.S. Nandakumar, V. Palanivel, V.R. Muthukkaruppan, Clin. Exp. Immunol. 94, 317-321
(1993).
36. A. Arockiasamy, G.S. Murthy and S. Krishnaswamy The Immunologist Supplement: 1 page 20
(1998).
37. Paniagua-Solis, J. Sanchez, V. Oritz-Navarrete, C.R. Gonzalez and A. Isibasi, FEMS Microbiol.
Lett. 141, 31-36 (1996).
38. S. Sujatha, A. Arockiasamy, S. Krishnaswamy and R. Usha (unpublished).
39. H.M. Vordermeier, P. Hoffmann, F.O. Gombert, G. Jung and W.G. Bessler, Infect. Immun. 58,
2719-2724 (1990).
40. S.P. Singh, S. Miller, Y.U. Williams, K.E. Rudd and H. Nikaido, Microbiology 142, 3201-3210
(1996).
41. S.P. Singh, S.R. Singh, Y.U. Williams, L. Jones and T. Abdullah, Infect. Immun. 63, 4600-4605
(1995).
42. T.V. Gnanasekaran, S. Peri, A. Arockiasamy and S. Krishnaswamy, Bioinformatics (in press)
2000.

Date Received: June 19, 2000

Communicated by the Editor Dino Moras

Vous aimerez peut-être aussi