CTL epitopes for influenza A including the H5N1 bird flu; genome-, pathogen-, and HLA-wide screening

Vaccine 25 (2007) 2823–2831

CTL epitopes for influenza A including the H5N1 bird flu; genome-, pathogen-, and HLA-wide screening Mingjun Wang a , Kasper Lamberth b , Mikkel Harndahl b , Gustav Røder b , Anette Stryhn b , Mette V. Larsen c , Morten Nielsen c , Claus Lundegaard c , Sheila T. Tang c , Morten H. Dziegiel d , Jørgen Rosenkvist d , Anders E. Pedersen a , Søren Buus b,∗ , Mogens H. Claesson a , Ole Lund c a

c

Laboratory of Cellular Immunology, Department of Medical Anatomy, The Panum Institute, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen N, Denmark b Division of Experimental Immunology, Institute of Medical Microbiology and Immunology, The Panum Institute 18.3.12, University of Copenhagen, Blegdamsvej 3, 2200 Copenhagen N, Denmark Center for Biological Sequence Analysis, BioCentrum-DTU, Technical University of Denmark, Building 208, 2800 Lyngby, Denmark d H:S Blood Bank KI 2034, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen Ø, Denmark Received 6 July 2006; received in revised form 1 December 2006; accepted 12 December 2006 Available online 29 December 2006

Abstract The purpose of the present study is to perform a global screening for new immunogenic HLA class I (HLA-I) restricted cytotoxic T cell (CTL) epitopes of potential utility as candidates of influenza A-virus diagnostics and vaccines. We used predictions of antigen processing and presentation, the latter encompassing 12 different HLA class I supertypes with >99% population coverage, and searched for conserved epitopes from available influenza A viral protein sequences. Peptides corresponding to 167 predicted peptide–HLA-I interactions were synthesized, tested for peptide–HLA-I interactions in a biochemical assay and for influenza-specific, HLA-I-restricted CTL responses in an IFN-␥ ELISPOT assay. Eighty-nine peptides could be confirmed as HLA-I binders, and 13 could be confirmed as CTL targets. The 13 epitopes, are highly conserved among human influenza A pathogens, and all of these epitopes are present in the emerging bird flu isolates. Our study demonstrates that present technology enables a fast global screening for T cell immune epitopes of potential diagnostics and vaccine interest. This technology includes immuno-bioinformatics predictors with the capacity to perform fast genome-, pathogen-, and HLA-wide searches for immune targets. To exploit this new potential, a coordinated international effort to analyze the precious source of information represented by rare patients, such as the current victims of bird flu, would be essential. © 2007 Elsevier Ltd. All rights reserved. Keywords: Influenza A virus; Peptide; CTL epitope; Diagnostics; Vaccine

1. Introduction Influenza is a highly contagious, airborne respiratory tract infection associated with a significant disease burden. The annual “mild” influenza epidemics caused by antigenic drift of the virus affects 10–20% of the world’s population with up to 5 million cases of serious illness and 500,000 deaths (http://www.who.int/vaccine research/diseases/ari/en). At ∗

Corresponding author. Tel.: +45 3532 7885; fax: +45 3532 7696. E-mail address: [email protected] (S. Buus).

0264-410X/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.vaccine.2006.12.038

various intervals, new influenza subtypes emerge against which no immunity exists in the human population and these may cause global pandemics with an even higher disease toll. The current outbreak of a new influenza subtype A (H5N1), which can be directly, although at this time rarely, transmitted from birds to humans [1], is an example of a potential pandemic flu threat, which has currently reached phase 3 of 6 in the WHO delineation of a flu pandemic (http://www.who.int/ csr/disease/avian influenza/phase/en/index.html). It is highly pathogenic in birds, and when transmitted to humans it carries a mortality of over 50% [2]. So far, all the resulting

2824

M. Wang et al. / Vaccine 25 (2007) 2823–2831

human cases have been in close contact with infected flocks and there is no known example of a human-to-human transmission of the current H5N1 bird flu. However, it is the largest and most severe influenza epidemic ever registered among birds; since 2003, it has spread rapidly to poultry in many countries in Asia and most recently it seems to have established itself in Turkey. The size of this virus repertoire has caused concerns that re-assortment or mutations of influenza genes occurring in bird populations, or in infected humans, eventually will generate a virus that can be transmitted from person to person causing a highly contagious, and potentially devastating, pandemic [3]. The primary port of entry of the influenza virus is the mucosa of the respiratory tract. The adaptive immune system can provide immune protection against mucosal pathogens through secretory IgA and IgM immunoglobulins, which can effectively prevent the virus from infecting its target cells. Vaccination using inactivated influenza virus preparations remains the primary method of prevention. However, the virus attempts to escape neutralizing antibodies through constantly changing the composition of its surface antigens. This complicates the development of cross-protective immunity, i.e. the ability to cover several different isolates; rather, influenza vaccines must regularly be updated to match existing seasonal epidemic flu isolates. Current vaccine technology is likely to be too slow, and also of too low capacity, to produce enough vaccine against a new emerging flu isolate in time to prevent a true pandemic. Thus, the development of faster and more efficient vaccine technologies capable of delivering new, safe, efficacious and easy-managed protective influenza vaccines is of high priority. It is known that CD8+ T cell responses also play a major role in the control of primary influenza virus infection [4,5]. In mice, CTLs against conserved epitopes contribute to protective immunity against influenza viruses of various subtypes [6,7]. Identification of CTL epitopes, especially conserved epitopes shared by multiple viral strains, might therefore be a robust vaccine strategy against emerging influenza epidemics. One could argue that CTLs, being specific for short immunogenic peptides and being diversified by the highly polymorphic HLA system, are easier to target against conserved, and thereby potentially cross-protective, epitopes. Obviously, it is easier to find conservation in a short primary peptide sequence, a CTL target, than in a longer tertiary protein structure, such as a characteristic antibody target. Also, the HLA-restricted CTL immune system targets different epitopes in different individuals thereby reducing the risk of a population-wide virus escape through removal of highly conserved epitope targets. The flipside of trying to exploit the potential advantages of HLA-restricted CTL immunity is that one would have to identify several different CTL targets to encompass the diversity of the HLA system, and obtain sufficient coverage of the human population. This problem, however, appears somewhat alleviated by the recent discovery that HLA molecules largely can be grouped into 12 different “supertypes” of overlapping specificities [8].

In a recent study, a limited search for novel flu CTL epitopes was performed; only one epitope was found with a search restricted to only a part of the genome, one flu isolate and one HLA-I molecule [9]. Here, we have performed a genome-, pathogen- and HLA-wide search for new CTL epitopes directed against influenza A virus. The genome-wide aspect assures that all influenza virus proteins are considered, the pathogen-wide assures that maximum conservation is achieved, whereas the HLA-wide aspect assures maximum coverage of human populations. Others and we have developed immuno-bioinformatics methods to identify CTL epitopes. Here, we have used recently improved methods based upon combined HLA binding, TAP binding, and proteasome processing predictions [10–13]. Note, that these tools have been developed using large databases such as SYFPEITHI and Los Alamos database as biological benchmarks, i.e. they have been designed with optimal immune epitope identification in mind. The predicted peptides were synthesized and tested by biochemical methods for binding to the appropriate recombinant HLA-I protein, and by IFN-␥ ELISPOT analysis for CTL immune responses using PBMCs from healthy, elderly and HLA-typed Danish subjects, which are assumed to have been exposed to multiple influenza infections in their past. Here – even at the limited scale of this study – we suggest several HLA-I-restricted, influenza-specific 9mer epitopes, many of which are conserved among several different influenza virus isolates even including the H5N1 bird flu virus. Combining their HLArestriction specificity, they cover virtually the entire human population.

2. Materials and methods 2.1. Collection of blood samples Buffy coats of 500 ml whole blood from healthy Danish donors (age range: 35–65 years; donors have given written informed consent) were obtained from The Blood Bank at Rigshospitalet (Copenhagen, Denmark) and used within 24 h to isolate peripheral blood mononuclear cells (PBMC). The donors were selected, according to serological typing of their HLA-A and -B haplotype, to maximize coverage of the 12 HLA class I supertypes. A high-resolution sequence-based typing (SBT) of the HLA-A and HLA-B loci was subsequently established (Genome Diagnostics, Utrecht, Netherlands). Data for donors are shown in Table 1. 2.2. Isolation of PBMC Peripheral blood mononuclear cells (PBMC) were isolated from buffy coats by density gradient centrifugation using Lymphoprep (Nycomed Pharma AS, Oslo, Norway). The freshly isolated PBMC were cryopreserved for later use at 20 × 106 cells in 1 ml RPMI-1640 containing 20% FCS and 10% dimethyl sulfoxide (DMSO) at −140 ◦ C.

M. Wang et al. / Vaccine 25 (2007) 2823–2831

2825

Table 1 Donors used in the study Donor

Serologic typing

Sequence based typing

#

Sex

Age (years)

HLA-A

HLA-B

HLA-A

1 11 12 13 14 15 17 21 24 26 28 32 35 36

M M F M M M F M F M F M F F

36 47 37 60 58 44 35 54 36 58 62 55 43 50

A2,A3 A2,A3 A1,A24 A2 A1,A2 A2,A26 A2,A29 A1,A24 A2,A26 A1,A3 A3,A26 A2,A32 A29,A33 A2,A24

B35,B44 B7,B45 B8,B35 B39,B44 B8,B37 B7,B64 B44,B62 B7,B39 B38,B58 B8,B27 B7,B14 B5,B15 B17,B27 B35,B57

0201 0201 0101 0201 0101 0201 0201 0101 0206 0101 0301 0201 2902 0201

Immune reactivity (# of CTL epitopes)

HLA-B 0301 0301 2402 0201 0201 2601 2902 2402 2601 0301 2601 3201 3303 2402

3501 0702a 0801 3901 0801 0702 1501b 0702 3801 0801 0702 1501 2705 3508

4402 4501 3503 4402 3701 1401 4403 3901 5801 2705 1401 5101 5801 5701c

– 1 – – 1 1 3 3 – 3 – – 2 1

a Single underline indicates that the corresponding HLA molecule was a restricting element for one epitope, double underlines that it was restricting element for two epitopes. b HLA-B*1501 belongs to HLA-B62 supertype family. c HLA-B*5701 belongs to HLA-B58 supertype family.

2.3. Bioinformatics search strategy for peptides derived from influenza A virus (H1N1) Conserved sequences in influenza A virus proteins were obtained from a multiple alignment. Protein sequences for each influenza A virus protein were translated from the NCBI RefSeq (http://www.ncbi.nlm.nih.gov/RefSeq/) entries NC 004518-25 corresponding to the eight segments of the influenza virus genome. Blast [14] was used to search NCBIs non-redundant database (NR) for homologues (using E < 0.05 as threshold). Multiple alignments based on the Blast alignments were constructed using the program view [14]. The sequences were clustered using 98% sequence identity as a threshold for similarity, and weighting each cluster equally, the probability for the different amino acids at each position in the alignment was calculated and used to calculate the probability of conservation of all 9mers in the consensus sequence derived from the multiple alignments. Binding to HLA molecules representing 12 different supertypes (HLAA1, -A2, -A3, -A24, -A26, -B7, -B8, -B27, -B39, -B44, -B58, -B62), prediction of proteasome cleavage and TAP binding was used to calculate a combined score [13]. Up to 17 of the top-ranking 9mer peptides with a conservation above 70% were selected for each supertype.

2.5. Biochemical peptide–HLA class I binding assay The biochemical assay for peptide–MHC-I binding was performed as previously described [15,16]. Briefly, denatured and purified recombinant HLA heavy chains were diluted into a renaturation buffer containing HLA heavy chain, ␤2 microglobulin, graded concentrations of the test peptide, and incubated at 18 ◦ C for 48 h allowing equilibrium to be reached. We have previously demonstrated that denatured HLA molecules can de novo fold efficiently, however, only in the presence of appropriate peptide [17]. The concentration of peptide–HLA complexes generated was measured in a quantitative enzyme-linked immunosorbent assay and plotted against the concentration of peptide offered [15]. Because the effective concentration of HLA (3–5 nM) used in these assays is below the equilibrium dissociation constant (KD ) of most high-affinity peptide–HLA interactions, the peptide concentration leading to half-saturation of the HLA is a reasonable approximation of the affinity of the interaction. An initial screening procedure was employed whereby a single high concentration (20,000 nM) of peptide was tested. If no complex formation was found, the peptide was assigned as a non-binder to the HLA molecule in question, conversely, if complex formation was found in the initial screening, a full titration of the peptide was performed to determine the affinity of binding.

2.4. Peptides The 9mer peptides were synthesized by standard 9fluorenylmethyloxycarbonyl (FMOC) chemistry, purified by reversed-phase high-performance liquid chromatography (at least 80%, usually >95% purity) and validated by mass spectrometry (Shafer-N, Copenhagen, Denmark). Peptides were distributed at 20 ␮g/vial and stored lyophilized at −20 ◦ C until use. Peptides were dissolved just before use.

2.6. IFN-γ ELISPOT assay PBMC were thawed, washed and resuspended to a concentration of 6–8 × 106 cells per ml in RPMI-1640 supplemented with 5% heat-inactivated AB serum (Valley Biomedical, Winchester, USA), 2 mM l-glutamine, 100 U/ml penicillin and 100 ␮g/ml streptomycin. Cells were cultured in 24-well

2826

M. Wang et al. / Vaccine 25 (2007) 2823–2831

plates (Nunc, Roskilde, Denmark). Individual peptides were added to a final concentration of 20 ␮g/ml per well and incubated for 10 days at 37 ◦ C, 5% CO2 in humidified air. RhIL-2 (Proleukin; Chiron, The Netherlands) 20 U/ml was added on day 1. Cells were harvested on day 10, washed twice in RPMI-1640 and resuspended in complete medium to a final concentration of 1 × 106 cells/ml. The IFN-␥ ELISPOT assay was performed as described previously [18] to quantify peptide-specific CTLs after in vitro expansion. Briefly, 96-well ELISPOT plates (Multiscreen, MAHAS4510; Millipore, Molsheim, France) were coated overnight at 4 ◦ C with 7.5 ␮g/ml anti-human IFN␥ (M-700A, Endogen; Pierce Biotechnology, Rockford, IL, USA). The plates were washed six times with sterile phosphate buffered saline (PBS) to remove unbound coating antibody. Cell suspensions were prepared and four to six replicate wells were seeded with 105 cells per well. Peptides were added in a final concentration of 10 ␮g/ml. As positive controls, cells were stimulated with 10 ␮g/ml phytohaemaglutinin (PHA) (Sigma-Aldrich, UK). Cells were incubated with complete medium alone as negative control. After incubation for 18–20 h at 37 ◦ C and 5% CO2 , plates were washed with PBS containing 0.05% Tween-20. Biotinylated antihuman IFN-␥ (M-701B, Endogen; Pierce Biotechnology, Rockford, IL, USA) was added at 0.75 ␮g/ml and plates were incubated for 2 h at room temperature. After incubation, plates were washed and incubated further for 1 h with streptavidin-conjugated peroxidase (Dako, Copenhagen, Denmark) diluted 1:500 in PBS/1% BSA. Next, the plates were washed as before and 200 ␮l of substrate [30 mg 4-chloro-1-naphthol (057h8927, Sigma) dissolved in 10 ml methanol, diluted in 50 ml triethanol-buffered H2 O (T1377, Sigma) at pH 7.5 and 25 ␮l H2 O2 30% (H1009, Sigma)] were added to the wells. Finally, plates were washed under running tap water and dried at room temperature. IFN-␥ spot-forming cells (SFC) were enumerated using an ELISPOT reader (KS ELISpot, Zeiss, Munich, Germany). 2.7. Intracellular IFN-γ FACS assay PBMC were thawed, washed and resuspended to a concentration of 1 × 107 cells per ml in Xvivo15 supplemented with 5% heat-inactivated AB serum (Valley Biomedical, Winchester, USA), 2 mM l-glutamine, 100 U/ml penicillin and 100 ␮g/ml streptomycin. Cells were stimulated over night with 5 ␮g/ml of the individual peptides in 24-well plates (Nunc, Roskilde, Denmark). The cells were washed and cultured for 9 days at 5 × 106 cells/ml at 37 ◦ C, 5% CO2 in humidified air. RhIL-2 (PepProTec; US) 50 U/ml was added on day 1. Dendritic cell (DC) was used as antigen-presenting cell (APC). PBMC were adhered for 2 h at 37 ◦ C in 24-well plates. Adherent cells were cultured for 9 days in Xvivo15 with 5% AB serum, 100 ng/ml GM-CSF, and 100 ng/ml IL-4. On day 7 the DCs were activated with a mixture of 10 ng/ml TNF-␣, 5 ng/ml IL-1␤, 20 ng/ml IL-6, and 1 ␮g/ml PGE2.

DC and T cell cultures were harvested on day 9. DC were pulsed with or without peptide for 1 h and added to the T cells. The cells were incubated for 4 h with Brefeldin A at 37 ◦ C. Subsequently, staining for CD8, CD69, and intracellular IFN␥ (Pharmingen) was carried out. The cells were analyzed on a FACScalibur. 2.8. Statistics Wilcoxon rank-sum test was used to analyze the quantitative differences between the experimental wells and control in Elispot assays. A P-value below 0.05 was considered significant.

3. Results 3.1. Prediction of HLA-I binding, proteasome cleavage site, and TAP binding An example of the proteins used as input for combined predictions of HLA-I binding, TAP binding and proteasome cleavage site, here exemplified by the proteins from the influenza A/Puerto Rico/8/34/Mount Sinai (H1N1) virus isolate, is shown in Table 2. Note that the search for conserved CTL epitopes skewed the selection towards the polymerases and the nucleoprotein, whereas it strongly deselected for the classical antibody targets, the Hemagglutinin and the Neuraminidase. For each of the 12 HLA-I supertypes, we selected up to 17 peptides with the top combined scores and with a conservation filter that assured they are present in more than 70% of the clusters of influenza sequences (see Section 2 for details, see Table 3 for an example). Peptides corresponding to a total of 167 peptide–HLA-I interactions were selected and synthesized. 3.2. Biochemical validation of HLA-I binding To determine whether the peptides indeed were binders to the relevant HLA-I proteins, they were tested in a biochemical binding assay (see Section 2). This included peptides predicted to bind to HLA-A1 (8 peptides), -A2 (14 peptides), -A3 (15 peptides), -A24 (15 peptides), -B7 (17 Table 2 The influenza A viral proteins for peptide prediction Entry

Protein

No. of predicted epitopes

NP NP NP NP NP NP NP NP

Polymerase (PB2) Polymerase (PB1) Polymerase (PA) Hemagglutinin (HA) Nucleoprotein (NP) Neuraminidase (NA) Matrix protein (M2) Matrix protein (M1)

37 57 39 1 17 7 1 8

775529.1 775530.1 775531.1 775532.1 775533.1 775534.1 775535.1 775536.1

Entry, database name for protein; Protein, common name for protein.

M. Wang et al. / Vaccine 25 (2007) 2823–2831

2827

Table 3 Characteristics of the epitopes Entry

Protein

Peptide

Sequence

HLA

KD

Pa

Pc

TAP

Combi

Cons 9

Responding donor

NP NP NP NP NP NP NP NP NP NP NP NP NP

Polymerase PB1 Polymerase PB1 Polymerase PB1 Polymerase PB1 Polymerase PB1 Polymerase PB1 Polymerase PB1 Polymerase PA Nucleoprotein NP Nucleoprotein NP Nucleoprotein NP Nucleoprotein NP Matrix protein M1

PB141–49 PB1591–599 PB1166–174 PB1349–357 PB1347–355 PB1566–574 PB1540–548 PA601–609 NP383–391 NP44–52 NP199–207 NP225–233 M1173–181

DTVNRTHQY VSDGGPNLY FLKDVMESM ARLGKGYMF KMARLGKGY TQIQTRRSF GPATAQMAL SVKEKDMTK SRYWAIRTR CTELKLSDY RGINDRNFW ILKGKFQTA IRHENRMVL

A26 A1 A2 B27 B62 B62 B7 B8 B27 A1 B58 B8 B39

6 6 51 246 178 88 6 NB 38 7 42 664 13

11.34 0.63 0.47 0.43 0.41 0.46 0.67 7.82 0.38 0.56 0.45 9.77 7.08

1.00 1.00 0.97 0.97 0.88 0.54 1.00 1.00 0.98 0.97 0.08 0.97 1.00

2.70 2.90 0.09 2.94 3.31 2.52 0.54 0.71 2.06 2.75 1.03 −0.35 1.26

2.06 3.74 1.25 2.03 1.47 1.46 2.09 1.28 1.72 3.35 1.65 1.42 1.29

0.98 0.90 0.77 0.98 0.97 0.92 0.99 0.78 0.87 0.81 0.99 0.99 0.85

Donor 15 Donor 21 Donor 17 Donor 26 and 35 Donor 17 Donor 17 Donor 11 Donor 26 Donor 26 and 35 Donor 21 Donor 36 Donor 14 Donor 21

775530.1 775530.1 775530.1 775530.1 775530.1 775530.1 775530.1 775531.1 775533.1 775533.1 775533.1 775533.1 775536.1

KD , measured affinity in nM; Pa: Predicted HLA binding propensity (for A26, B8 and B39 supertypes bindings have been predicted using matrix methods where outputs above 7 are predicted binders. For the other supertypes neural networks have been used and numbers above 0.42 correspond to binding affinities predicted to be better than 500nM); Pc, predicted proteasomal cleavage propensity; TAP (transporter associated with antigen processing), predicted TAP binding propensity; Combi, combined score; Cons 9, fraction of clusters (with more than 98% sequence identity) that contain the 9mer; NB, non-binder (>5000 nM). The prediction servers are available at http://www.cbs.dtu.dk/services/.

peptides), -B8 (12 peptides), -B27 (15 peptides), -B44 (15 peptides), -B58 (14 peptides), -B62 (15 peptides), A26 (13 peptides), or B39 (14 peptides). Consistent with previous classifications, the binding affinity (KD ) of the 167 9mer predicted peptides studied can be divided into groups of high affinity binders (n = 34; KD ≤ 50 nM), intermediate affinity binders (n = 55; 50 nM < KD ≤ 500 nM), low affinity binders (n = 23; 500 nM < KD ≤ 5000 nM) and peptides with low or no affinity for MHC-I molecules (n = 55, KD > 5000 nM) (see Table 4).

personal communication Lars Nielsen, State Serum Institute, Denmark). PBMCs were tested in an IFN-␥ ELISPOT assay. Positive peptides were tested at least twice in the same HLA matched donors to confirm responsiveness. Negative peptides were tested in PBMCs from 2 to 3 HLA matched donors, which responded to at least one of the positive peptides confirming that these donors had previously been exposed to influenza A virus. According to this criterion thirteen peptides belonging to 9 supertypes (A1, A2, A26, B27, B39, B58, B62, B7 and B8) were immunogenic suggesting that the peptides are true CTL epitopes. The ELISPOT data are given in Table 5, whereas peptide and donor specific overviews are given in Table 3 and 1, respectively. The IFN-␥ ELISPOT assay used above does not distinguish between CD4+ and CD8+ T cell mediated secretion of IFN-␥. To ascertain that CD8+ T cells indeed were involved in these responses, donor T cells were pulsed with the relevant peptides and expanded for 9 days. At that time, they

3.3. Immunogenicity of the predicted peptides All the peptides were also tested for their ability to stimulate influenza A-specific, HLA-matched T cells from healthy Danish subjects aged 35–65, i.e. assumed to have previously been exposed to natural influenza A virus (the annual seasonal influenza incidence in Denmark is between 5 and 10%, Table 4 Measured affinity of the predicted peptide binders HLA supertype

KD ≤ 50

50 < KD ≤ 500

A1 A2 A3 A24 A26 B7 B8 B27 B39 B44 B58 B62

5 5

4 1 1 1 1

3 5 4 3 2 6 1 5 2 2 10 12

Total

34

55

9 1 6

500 < KD ≤ 5000 1 5

1 3 2 5 3 1 2 23

KD > 5000

Non-binders

Total

2 3 1 2 1 1 1

3 1 3 8 1 7 2 5 8 1

8 14 15 15 13 17 12 15 14 15 14 15

16

39

167

5

KD , the equilibrium dissociation constant; a measurement of the affinity of peptides binding to the relevant HLA molecules in nM. The lower the value, the stronger the binding.

2828

M. Wang et al. / Vaccine 25 (2007) 2823–2831

Table 5 Elispot analysis of peptide-specific donor responses Peptide

HLA-restriction

Experiment 1 +Peptide

PB1591–599 NP44–52 PB1166–174 PB141–49 PB1540–548 NP225–233 PA601–609 PB1349–357 NP383–391 M1173–181 NP199–207 PB1566–574 PB1347–355

HLA-A1 HLA-A1 HLA-A2 HLA-A26 HLA-B7 HLA-B8 HLA-B8 HLA-B27 HLA-B27 HLA-B39 HLA-B58 HLA-B62 HLA-B62

18 34 74 40 7 9 23 10 39 14 28 15 77

± ± ± ± ± ± ± ± ± ± ± ± ±

2 5 10 3 2 4 6 6 6 3 5 5 20

Experiment 2 −Peptide 3 4 11 20 2 1 1 1 1 3 1 2 3

± ± ± ± ± ± ± ± ± ± ± ± ±

3 1 6 7 1 1 1 1 1 1 1 2 2

+Peptide 12 13 140 38 13 19 119 14 40 84 15 21 91

± ± ± ± ± ± ± ± ± ± ± ± ±

4 4 36 5 2 7 8 4 6 11 6 2 8

−Peptide 1 0 20 24 6 2 2 1 2 3 2 2 10

± ± ± ± ± ± ± ± ± ± ± ± ±

1 0 7 3 1 2 1 1 1 1 2 0 3

Data represent average SFC/105 ± S.D. Two independent experiments for each peptide/epitope are included.

were stimulated with peptide-pulsed dendritic cells and analyzed by flowcytometry for intracellular IFN-␥ release as well as for phenotypic markers such as CD69 (activation marker) and CD8 (CTL marker). Fig. 1 clearly demonstrates the presence of peptide-specific CD8+ T cells. Further demonstrating the involvement of CD8+ T cells, we successfully blocked ELISPOT responses using the pan-specific HLA-I antibody, W6/32 (data not shown).

4. Discussion One of the major drawbacks of a peptide-based CTL vaccine strategy is that the restricting HLA genes are extremely polymorphic resulting in a vast diversity of peptide-binding HLA specificities and a low population coverage for any given peptide–HLA specificity. To increase population coverage, one might include defined epitopes for each HLA-I

Fig. 1. Peptide specific CD8+ T cell response. PBMCs were stimulated with peptide for 9 days. Presence of specific CD8+ T cells were analyzed by intracellular IFN-␥ staining and flow cytometry after peptide stimulation. The CD8+ T cells were gated and CD69 and IFN-␥ staining of these are displayed in the panel. The top row represented HLA-A2 restricted PB1166–174 peptide specific responses and the bottom row represented HLA-B57 restricted NP199–207 specific responses. In parentheses are shown the percentage of IFN-␥+ CD8+ T cells of unstimulated cultures in left columns and of the stimulated cultures in the right column.

M. Wang et al. / Vaccine 25 (2007) 2823–2831

allele, however, this would lead to a vaccine comprising hundreds of peptides. One way to reduce this complexity is to group HLA-I molecules into so-called HLA-I supertypes; a classification that refers to a group of HLA alleles with largely overlapping peptide binding specificities [19–21]. Ideally this means that a peptide, which binds to one allele within a supertype, has a high probability of binding to other allelic members of the same supertype. For HLA-I molecules, Sette and Sidney [20] defined nine supertypes (A1, A2, A3, A24, B7, B27, B44, B58, B62) and suggested that >99.6% of persons in all ethnic groups surveyed possessed at least one allele within at least one of the nine supertypes. Thus, 9 epitopes representing each of these 9 MHC-I supertypes would lead to virtually complete population coverage. Recently, Lund et al. [8] defined 3 more MHC class I supertypes (A26, B8 and B39). The present study includes all 12 HLA-I supertypes. The majority (85%) of the immunogenic peptides reported here had intermediate to high HLA-I binding affinity; this is consistent with previous reports that the development of immune reactivity towards peptides with high binding affinity for MHC-I in general is better than that of peptides with intermediate or low binding affinities [22]. However, not all peptides were found to be immunogenic in this study. In fact, only 11 of 89 of the intermediate to high HLA-I affinity binding influenza peptides could be demonstrated to be immunogenic in our donor cohort. This is not unexpected since several other requirements besides HLA-I binding must be met to assure CTL immunity; requirements like peptide generation and TAP translocation, which were not validated here. Also, we do not know whether our donors have been exposed to influenza isolates representing the exact peptides selected here. It follows that we could have expected to identify even more epitopes among our peptides if we had examined an even larger donor cohort. One peptide, PA601–609 , appeared to present an anomaly since it had no measured binding affinity to its predicted HLA-I restriction element,

2829

HLA-B*0801, yet, it induced a strong IFN-␥ ELISPOT response in PBMCs from Donor 26 (HLA-A*0101, -A*0301, -B*0801, -B*2705). One possible explanation could be that the low affinity binding (KD = 1221 nM) of peptide PA601–609 to HLA-A*0301 might be sufficient to support a HLA-A*0301-restricted CTL response. Among these thirteen epitopes, two HLA-A1-restricted peptides PB1591–599 , NP44–52 [23], and the HLA-B27 restricted peptide NP383–391 [24] have been reported previously. In addition, PB1349–357 is known to be restricted by mouse H-2Dk [25]. Since all the epitopes selected here are conserved and therefore present in many subtypes of influenza A virus, our epitopes might include most influenza A subtypes including the emerging H5N1 bird flu. This was confirmed upon closer inspection of several sequenced influenza A H5N1 virus strains. Thus all 13 of the peptides identified here were also found in H5N1 isolates (Table 6). The immunogenic peptides identified in the present study bind to 9 of the 12 HLA-I supertypes so far identified thereby giving close to 100% population coverage. These epitopes could be used diagnostically to detect influenza specific CTL responses in patients and after vaccination. Even though a CTL based influenza vaccine might not prevent infection, it might still protect against the disease [7,26]. In the recent outbreak of bird flu, more than half of those infected with the bird flu virus have died. Most cases have occurred in children and young adults (under 40 years old). Why does the H5N1 virus attack the young and spare the elder? Since the majority of conserved epitopes are also conserved in H5N1 subtypes, one possibility is that the elder people have been exposed to and developed protective immunity to a greater variety of influenza A viruses than the younger ones. Given the development of an efficient CTL epitope delivery technology, these epitopes might eventually become vaccines in their own right either using individual epitopes, or epitopes fused as polytopes. Following a screening like the

Table 6 Epitopes conserved in H5N1 isolates H5N1 isolates from humans and birds in various countries in Asia

PB141–49 PB1591–599 PB1166–174 PB1349–357 PB1347–355 PB1566–574 PB1540–548 PA601–609 NP383–391 NP44–52 NP199–207 NP225–233 M1173–181

DTVNRTHQY VSDGGPNLY FLKDVMESM ARLGKGYMF KMARLGKGY TQIQTRRSF GPATAQMAL SVKEKDMTK SRYWAIRTR CTELKLSDY RGINDRNFW ILKGKFQTA IRHENRMVL

HK483

HK486

HK212

HK213

Thai1

Thai676

VN1203

VN014

CKJ

GG

CKIn

+ + + + + + + − + − + + +

+ + − + + + + − + − + + +

+ + + + + + + + + + + + +

+ + + − + + + + + + + + +

+ − + + + − + + + + + + +

+ + + + + + + + + + + + +

+ + + + + + + + + + + + +

+ + + + + + + + + + − + +

+ + + + + − + + + + + + +

+ − + + + + + + + + + + +

+ + + + + + + + + + + + +

HK483, A/Hong Kong/483/97 (H5N1); HK486, A/Hong Kong/486/97 (H5N1); HK212, A/HK/212/03 (H5N1); HK213, A/HK/213/03 (H5N1); Thai1, A/Thailand/1(KAN-1)/2004 (H5N1); Thai676, A/Thailand/676/2005 (H5N1); VN1203, A/Viet Nam/1203/2004 (H5N1); VN014, A/Viet Nam/BL-014/05 (H5N1); CKJ, A/chicken/Jilin/9/2004 (H5N1); GG, A/Goose/Guangdong/1/96 (H5N1); CKIn, A/Ck/Indonesia/PA/2003 (H5N1). +, conserved in the H5N1 isolate; −, not conserved in the H5N1 isolate.

2830

M. Wang et al. / Vaccine 25 (2007) 2823–2831

one described here, and before using the resulting peptide information in a vaccine formulation, one should verify that the peptide-specific CTLs can recognize virus infected cells. In terms of formulating an epitope-based vaccine, there are several possibilities including DNA vaccines, virus-derived vectors, etc., however, more research and development is needed before the full vaccine potential of epitope information can be realized. Animal studies have demonstrated that epitope-based vaccines in some cases can confer immune protection [20,27]. One could hope that a vaccine including all the epitopes identified in the present study could reduce the disease burden of the seasonal influenza, as well as of the unpredictable pandemic variants of the influenza virus. Finally, a note on the use of immuno-bioinformatics and epitope information in the battle against infectious diseases: immuno-bioinformatics is maturing rapidly and has now reached the stage where genome-, pathogen- and HLA-wide scanning for immunogenic epitopes are possible at a scale and speed that makes it possible to exploit genome information as fast as it can be generated. During the recent SARS epidemic, the SARS genome was identified in a matter of weeks, and a complete CTL epitope scanning – just barely possible at that time – was completed a few months later [16]. A faster and more complete scanning – involving more peptide and more patients – is today entirely possible and only limited by the modest resources needed. In this context, the patients are an important source of information unlocking the epitopes of the natural CTL responses; and they are especially precious when they are scarce such as during the recent SARS epidemic and during the present bird flu threat. Assuring a fast and complete gathering of such information for the benefit of the development of diagnostics and eventually of vaccines is a task that should be internationally guided and coordinated. In summary, we have identified ten novel influenza Aderived CTL epitopes, and confirmed three previously known ones. These epitopes appear to be conserved in different isolates of the highly pathogenic H5N1 influenza virus. These results have important implications for the rational design of CTL epitope-based influenza A virus diagnostics and vaccines applicable to all ethnic groups.

Acknowledgements This work was supported by NIAID Contracts no. HHSN266200400083C, HHSN266200400025C and EU 6FP 503231. We thank Ms. Trine Devantier for her excellent technical assistance.

References [1] Chan PK. Outbreak of avian influenza A (H5N1) virus infection in Hong Kong in 1997. Clin Infect Dis 2002;(34(Suppl. 2)):S58–64. [2] Beigel JH, Farrar J, Han AM, Hayden FG, Hyer R, de Jong MD, et al. Avian influenza A (H5N1) infection in humans. N Engl J Med 2005;353(13):1374–85.

[3] Horimoto T, Kawaoka Y. Influenza: lessons from past pandemics, warnings from current incidents. Nat Rev Microbiol 2005;3(8):591–600. [4] Bender BS, Croghan T, Zhang L, Small Jr PA. Transgenic mice lacking class I major histocompatibility complex-restricted T cells have delayed viral clearance and increased mortality after influenza virus challenge. J Exp Med 1992;175(4):1143–5. [5] McMichael AJ, Gotch FM, Noble GR, Beare PA. Cytotoxic T-cell immunity to influenza. N Engl J Med 1983;309(1):13–7. [6] Kuwano K, Scott M, Young JF, Ennis FA. HA2 subunit of influenza A H1 and H2 subtype viruses induces a protective cross-reactive cytotoxic T lymphocyte response. J Immunol 1988;140(4):1264–8. [7] Ulmer JB, Donnelly JJ, Parker SE, Rhodes GH, Felgner PL, Dwarki VJ, et al. Heterologous protection against influenza by injection of DNA encoding a viral protein. Science 1993;259(5102):1745–9. [8] Lund O, Nielsen M, Kesmir C, Petersen AG, Lundegaard C, Worning P, et al. Definition of supertypes for HLA molecules using clustering of specificity matrices. Immunogenetics 2004;55(12):797–810. [9] Toebes M, Coccoris M, Bins A, Rodenko B, Gomez R, Nieuwkoop NJ, et al. Design and use of conditional MHC class I ligands. Nat Med 2006;12(2):246–51. [10] Buus S, Lauemoller SL, Worning P, Kesmir C, Frimurer T, Corbet S, et al. Sensitive quantitative predictions of peptide–MHC binding by a ’Query by Committee’ artificial neural network approach. Tissue Antigens 2003;62(5):378–84. [11] Nielsen M, Lundegaard C, Worning P, Lauemoller SL, Lamberth K, Buus S, et al. Reliable prediction of T-cell epitopes using neural networks with novel sequence representations. Protein Sci 2003;12(5):1007–17. [12] Nielsen M, Lundegaard C, Worning P, Hvid CS, Lamberth K, Buus S, et al. Improved prediction of MHC class I and class II epitopes using a novel Gibbs sampling approach. Bioinformatics 2004;20(9):1388–97. [13] Larsen MV, Lundegaard C, Lamberth K, Buus S, Brunak S, Lund O, et al. An integrative approach to CTL epitope prediction: a combined algorithm integrating MHC class I binding, TAP transport efficiency, and proteasomal cleavage predictions. Eur J Immunol 2005;35(8):2295–303. [14] Schaffer AA, Aravind L, Madden TL, Shavirin S, Spouge JL, Wolf YI, et al. Improving the accuracy of PSI-BLAST protein database searches with composition-based statistics and other refinements. Nucleic Acids Res 2001;29(14):2994–3005. [15] Sylvester-Hvid C, Kristensen N, Blicher T, Ferre H, Lauemoller SL, Wolf XA, et al. Establishment of a quantitative ELISA capable of determining peptide-MHC class I interaction. Tissue Antigens 2002;59(4):251–8. [16] Sylvester-Hvid C, Nielsen M, Lamberth K, Roder G, Justesen S, Lundegaard C, et al. SARS CTL vaccine candidates; HLA supertype-, genome-wide scanning and biochemical validation. Tissue Antigens 2004;63(5):395–400. [17] Østergaard Pedersen L, Nissen MH, Hansen NJ, Nielsen LL, Lauenmoller SL, Blicher T, et al. Efficient assembly of recombinant major histocompatibility complex class I molecules with preformed disulfide bonds. Eur J Immunol 2001;31(10):2986–96. [18] Svane IM, Pedersen AE, Johnsen HE, Nielsen D, Kamby C, Gaarsdal E, et al. Vaccination with p53-peptide-pulsed dendritic cells, of patients with advanced breast cancer: report from a phase I study. Cancer Immunol Immunother 2004;53(7):633–41. [19] Sidney J, del Guercio MF, Southwood S, Engelhard VH, Appella E, Rammensee HG, et al. Several HLA alleles share overlapping peptide specificities. J Immunol 1995;154(1):247–59. [20] Sette A, Sidney J. Nine major HLA class I supertypes account for the vast preponderance of HLA-A and -B polymorphism. Immunogenetics 1999;50(3–4):201–12. [21] del Guercio MF, Sidney J, Hermanson G, Perez C, Grey HM, Kubo RT, et al. Binding of a peptide antigen to multiple HLA alleles allows definition of an A2-like supertype. J Immunol 1995;154(2):685–93. [22] Sette A, Vitiello A, Reherman B, Fowler P, Nayersina R, Kast WM, et al. The relationship between class I binding affinity and immunogenic-

M. Wang et al. / Vaccine 25 (2007) 2823–2831 ity of potential cytotoxic T cell epitopes. J Immunol 1994;153(12): 5586–92. [23] DiBrino M, Tsuchida T, Turner RV, Parker KC, Coligan JE, Biddison WE. HLA-A1 and HLA-A3 T cell epitopes derived from influenza virus proteins predicted from peptide binding motifs. J Immunol 1993;151(11):5930–5. [24] Huet S, Nixon DF, Rothbard JB, Townsend A, Ellis SA, McMichael AJ. Structural homologies between two HLA B27-restricted peptides suggest residues important for interaction with HLA B27. Int Immunol 1990;2(4):311–6.

2831

[25] Tourdot S, Herath S, Gould KG. Characterization of a new H-2D(k)restricted epitope prominent in primary influenza A virus infection. J Gen Virol 2001;(82(Pt 7)):1749–55. [26] Taylor PM, Askonas BA. Influenza nucleoprotein-specific cytotoxic T-cell clones are protective in vivo. Immunology 1986;58(3): 417–20. [27] Rodriguez F, An LL, Harkins S, Zhang J, Yokoyama M, Widera G, et al. DNA immunization with minigenes: low frequency of memory cytotoxic T lymphocytes and inefficient antiviral protection are rectified by ubiquitination. J Virol 1998;72(6):5174–81.