Dias nummer 1

1
Preventing a future H5N1 flu pandemic by
vaccination
Anna Irene Vedel Sørensen, Thilde Warrer Jensen,
Lisbeth Elvira de Vries and Michael Back
Dalgaard Immunological Bioinformatics 2006, CBS,
DTU
Introduction H5N1 is an avian influenza. It was
detected in humans for the first time in 1997 in
Hong Kong. Since then the spread to humans has
been limited where 228 humans has been infected
of whom 130 has died (1). Evidence suggests that
the transmission so far has been bird-to-human
(2). Influenza binds to host cells via binding of
the viral surface protein hemagglutinin (HA) to
receptors containing a terminal sialic acid.
Human influenza preferentially binds to a-2,6
sialic acid-galactose-linkages whereas avian
influenza preferentially binds to a-2,3 sialic
acid-galactose-linkages (3). So far there has not
been an effective inter-human transmission which
potentially can lead to a pandemic (1, 4).To
obtain inter-human transmission, H5N1 must adapt
so it can bind to a-2,6 sialic acid-galactose-link
ages. In vitro studies has shown that only two
mutations in HA at residue 226 and 228 is needed
(2). Because of this knowledge we have designed a
vaccine directed against the mutated H5 to
prevent a future H5N1 pandemic. The vaccine is
intended as a recombinant protein vaccine to
obtain a B-cell response. Furthermore we have
designed a polytope vaccine against H5N1, where a
MHC class I and class II response is obtained.
This polytope vaccine is intended as a
DNA-vaccine or a peptide vaccine.
B Cell epitope We applied the following strategy
in the pursue of defining an appropriate
recombinant protein vaccine candidate eliciting a
humeral response which could neutralise an
invading mutated avian H5N1 virus. CPHmodels (6)
was used to obtain a protein structure from pdb
(8) for the H5N1 reference strain
(YP_308669.1_HA, 9) resulting in a nearly perfect
match with crystal structure found in pdb
(id1JSM, 8). Substitutions were introduced at aa
position Q226L and G228S inferring a HA protein
with affinity for the human a2,6-sialic
acid-galactose (2). Detecting areas of
conservation 66 human H5N1 isolates and
approximately 80 avian isolates were aligned
using ClustalW (5). We used BepiPred for the
prediction of possible linear B-cell epitopes
(6), and both DiscoTope (6) and CEP (8) for the
prediction of conformational B-cell epitopes.
Furthermore we also created a consensus model
combining the predictions from DiscoTope and CEP
(6, 8). PyMol (10) was used as protein
visualization tool. Further we investigated the
protein for MHC DR4 II binding using SYFPEITHI
(11), EasyGibbs (6) and ProPred (12) and found
some good binders e.g. CIGYHANNSTEQVDT.
Polytope vaccine We have designed a polytope
vaccine directed against H5N1 containing
three-five MHC class I epitopes and one class II
epitope. The epitopes were selected by the
principal wisdom of the crowd, using several
prediction servers (NetCTL, NetMHC, NetChop,
EasyGibbs (6), SYFPEITHI (11), and ProPred MHC
class-II binding peptide prediction server (12)).
A criteria for selecting the epitopes was that
they had to be conserved in all the H5N1 human
isolates (9) and in the H5N1 reference sequence
(YP_308669, 9, Figure 3). Finally, the polytopes
were optimized considering processing in the MHC
class I pathway (13, Figure 4). The selected
epitopes were found in the polymerase subunits
PB1 and PB2 and the nucleoprotein NP (Figure 3).
The polytope shown in figure 4.a contains
epitopes for MHC class I A1, A2, A3, B7, B44 and
class II DR4. In addition, the A2 epitope also
appears to be an epitope for A24 and the B7
epitope may also be an epitope for B8. Thus in
theory this polytope should cover 98,1-100 of
the worlds population (14). As shown in figure
4.a the polytope has strong internal proteasomal
cleavage sites in the A3 epitope. Furthermore a
new epitope is appearing in the DR4 epitope.
However, it was not possible to find an A3
epitope without internal cleavages in both the
native context and in the polytope. Therefore, we
have also optimized a smaller polytope containing
only the A2, A3, B7 and DR4 epitopes. This should
cover 83 - 88,5 and 5 of the worlds population
(14). As shown in figure 4.b this has a lower
probability of internal cleavage and has no new
epitopes. Thus we believe that this polytope
(figure 4.a) is the best candidate for a polytope
vaccine against H5N1.
PB2(344-353) pb1 (443-452) np
(462-471) A/Goose/Guangdong/1/96
(H5N1) KKEEEVLTGNLQTLKIRV...HSWIPKRNRSILNTS...QGRG
VFELSDEKATNP A/Human/Hong Kong/213/2003
(H5N1) KKEEEVLTGNLQTLKIRV...HSWIPKRNRSILNTS...QGRG
VFELSDEKATNP A/Human/Viet Nam/CL26/2004
(H5N1) KKEEEVLTGNLQTLKIRV...HSWIPKRNRSILNTS...QGRG
VFELSDEKATNP A/Viet Nam/1203/2004
(H5N1) KKEEEVLTGNLQTLKIRV...HSWIPKRNRSILNTS...QGRG
VFELSDEKATNP A/Human/Thailand/NK165/2005
(H5N1) KKEEEVLTGNLQTLKIRV...HSWIPKRNRSILNTS...QGRG
VFELSDEKATNP A2
B7
A3
Figure 1. Alignment with ClustalW (5) showing the
proposed introduction of mutations in aa position
Q226L and G228S (HA1/1-334) in the sialic acid
terminal receptor binding domain (RBD) of H5N1
avian influenza. This confers a switch from HA
found in avian species preferring a binding to
sialic acid in an a2,3-linkage to galactose to a
HA recognising sialic acid in a2,6-linkage which
is found in humans. The four top sequences are
obtained from human cases the bottom sequence is
the H5N1 reference strain. C
Figure 3. Multiple alignment of four human
isolates and the reference strain for H5N1 avian
influenza (9) in ClustalW (5) The number of the
first amino acid in the segments they are derived
from (PB2, PB1and NP respectively) are indicated
above the epitopes covering the MHC class I
supertypes A2, B7 and A3 respectively. Accession
numbersAF144300, AF144301, AF144303, AB212051,
AB212052, AB212055, DQ492896, DQ493422, DQ493160,
AY818126, AY818129, AY818138, DQ372598, DQ372597,
DQ372594.
Figure 4. Proteasomeatlases of two optimized
polytopes using polytope_cont3, Morten Nielsen,
CBS, DTU . A. Shows the proteasome cleavage (red
and black bars) of a polytope consisting of 6
epitopes (blue squeres) separated by linker
regions. The red squre illustrates a new A2
epitope arising in the optimized polytop. The
polytope contains 5 MHC class I epitopes and 1
MHC class II epitope (DR4). The MHC class I
epitopes cover superclass B7, A2, A3, B44 and
A1. The amino acid sequence of the polytope with
epitope sequences in upper case letters and
linker sequences i lower case letters
mysdIPKRNRSILarVLTGNLQTLyyenrGVFELSDEKntdakaFEDLRV
SSFfvSSDDFALIViLVGIDPFRLLQNSQVFSLiv. B. Shows the
proteasome cleavage of a polytope containing the
B7, A2, A3 and DR4 epitopes as shown in A. The
amino acid sequences mrmIPKRNRSILrarVLTGNLQTLarrr
GVFELSDEKrlvkLVGIDPFRLLQNSQVFSLiltp
Conclusion In the present study a potential human
recombinant protein vaccine candidate against the
highly virulent H5N1 avian influenza was designed
using the Hemagglutinin as a template. In the
recombinant HA1 subunit the amino acids 226 and
228 were replaced with leucine and serine which
in-vivo would enable the virus to bind to
receptors on human cells. Potential linear and
conformational B-cell epitopes exposed on the
surface of the protein were identified. Using
linear B-cell epitope prediction does not seem to
make much sense as these only comprise app. 10
of the total number of epitopes and the linear
prediction method does predict epitopes
internally in the protein which obviously not are
surface exposed. When we used two different
models for predicting conformational B-cell
epitopes we observed quite some differences in
the predictions. However surface areas exhibiting
consensus between the two models were identified
including the sialic acid terminal receptor
binding domain.Furthermore potential T-cell
epitopes for the MHC class II supertype DR4 were
identified. The proposed recombinant protein
vaccine needs trimming, cutting of the
trans-membrane region and confirmation that the
protein has folded and been glycosylated
correctly in our production organism e.g. yeast
or E. coli. In addition vaccination methods and
adjuvants must be considered. Two polytopes were
constructed selecting epitopes from the full
genome of influenza H5N1. All epitopes included
in the polytopes were conserved in all public
available sequences of human isolates of H5N1 and
in the reference sequence of H5N1. Unfortunately
it was not possible to avoid strong internal
cleavages in the epitope covering A3 thus
reducing the efficacy for this vaccination
polytope. The polytope is intended for use as a
DNA vector vaccine, but needs quite a lot of
optimization, and several other factors such as
promotors, adjuvants and vector type have to be
taken into consideration in the next step.
Figure 2. B-cell epitope predictions (red sticks)
and protein structures for the receptor
hemagglutinin (HA) from a H5N1 influenza virus.
The mature HA protein is composed of two subunits
HA1/chain a (green) and HA2/chain b (yellow) with
HA1 containing the sialic acid terminal receptor
binding domain (RBD) (blue). The very distal part
of the HA2 as to the RBD composes the
transmembrane structure. A. Linear B-cell
epitope prediction using BepiPred (6). B.
Conformational B-cell epitope prediction using
DiscoTope (6). C. Conformational B-cell epitope
prediction using CEP (7). D. Conformational
B-cell epitope prediction using consensus
prediction from DiscoTope (6) and CEP (7).

• References
• WHO http//www.who.int
• Stevens, J. et al., 2006, Structure and Receptor
  Specificity of the Hemagglutinin from an H5N1
  Influenza Virus, Science vol. 312 (404-10)
• Brown, E. G., 2000, Influenza virus genetics,
  Biomed Pharmacother, 54 (196-209)
• Center For Infectious Disease Research Policy,
  University of Minnesota http//www.cidrap.umn.edu
  
• ClustalW http//www.ebi.ac.uk/clustalw
• BepiPred, DiscoTope, CPHmode, EasyGibbs, NetCTL,
  NetChop, NetMHC http//www.cbs.dtu.dk/services
• Conformational Epitope onformational Server
  (CEP) http//202.41.70.748080/cgi-bin/cep.pl
• pdb http//www.rcsb.org/
• Influenza Virus Resource, NCBI
  http//www.ncbi.nih.gov/genomes/FLU/FLU.html
• Pymol http//pymol.sourceforge.net/
• SYFPEITHI http//www.syfpeithi.de
• ProPred MHC class-II binding peptide prediction
  server http//www.imtech.res.in/raghava/propred/i
  ndex.html
• polytope_cont3, Morten Nielsen, CBS, DTU
• Lund, O. et al., 2005, Immunological
  Bioinformatics, The MIT Press, Cambridge,
  Massachusetts, London, England