|Home | About | Journals | Submit | Contact Us | Français|
Viruses can exploit a variety of strategies to evade immune surveillance by cytotoxic T lymphocytes (CTL), including the acquisition of mutations in CTL epitopes. Also for influenza A viruses a number of amino acid substitutions in the nucleoprotein (NP) have been associated with escape from CTL. However, other previously identified influenza A virus CTL epitopes are highly conserved, including the immunodominant HLA-A*0201-restricted epitope from the matrix protein, M158-66. We hypothesized that functional constraints were responsible for the conserved nature of influenza A virus CTL epitopes, limiting escape from CTL. To assess the impact of amino acid substitutions in conserved epitopes on viral fitness and recognition by specific CTL, we performed a mutational analysis of CTL epitopes. Both alanine replacements and more conservative substitutions were introduced at various positions of different influenza A virus CTL epitopes. Alanine replacements for each of the nine amino acids of the M158-66 epitope were tolerated to various extents, except for the anchor residue at the second position. Substitution of anchor residues in other influenza A virus CTL epitopes also affected viral fitness. Viable mutant viruses were used in CTL recognition experiments. The results are discussed in the light of the possibility of influenza viruses to escape from specific CTL. It was speculated that functional constraints limit variation in certain epitopes, especially at anchor residues, explaining the conserved nature of these epitopes.
Cytotoxic T lymphocytes (CTL) play an important role in the control of viral infections (10). To evade these CTL responses, viruses can exploit a variety of mechanisms to prevent recognition by specific CTL, including the accumulation of amino acid substitutions in or adjacent to CTL epitopes (28, 39). This strategy has been shown predominantly by certain RNA viruses, such as human immunodeficiency virus (HIV) (7, 21, 22, 35, 44, 47, 48), hepatitis C virus (9, 60), and lymphocytic choriomeningitis virus (36, 46), which are known for their high mutation rate. Also for influenza A viruses, a number of amino acid substitutions in the nucleoprotein (NP) have been associated with escape from human CTL. One of them, the R-to-G substitution at position 384 (R384G), which is at the anchor residues of the HLA-B*0801-restricted NP380-388 and HLA-B*2705-restricted NP383-391 epitopes, resulted in the loss of these epitopes (51, 58). This substitution reduced the in vitro virus-specific CTL response in HLA-B*2705-positive individuals significantly (2). Although the R384G substitution was tolerated only in the presence of one or more functionally compensating comutations (50, 52), it was fixed rapidly. This was explained by small selective advantages and population dynamics in a theoretical model (19). A third variable epitope in the influenza A virus NP is the HLA-B*3501-restricted epitope NP418-426, which displayed considerable variability in T-cell contact residues, affecting recognition by specific CTL (4, 5). Thus, in contrast to the two epitopes described above, the NP418-426 epitope retained its anchor residues for binding to HLA-B*3501. Other previously identified epitopes in influenza A virus proteins are highly conserved, such as the immunodominant HLA-A*0201-restricted epitope from the matrix protein, M158-66. It is likely that selective pressure by CTL against this epitope is high, considering the immunodominant nature of the epitope (3, 55) and the high prevalence of HLA-A*0201 in the human population (34). Yet, the amino acid sequence of this nine-mer epitope is conserved, even between different subtypes of human influenza A virus. We hypothesize that functional constraints are responsible for the inability of the virus to accumulate amino acid substitutions in this and other conserved epitopes, limiting immune escape from virus-specific CTL responses. To test this hypothesis, we performed a mutational analysis of various epitopes and tested the effect of selected amino acid substitutions on viral fitness and immune recognition by CTL. For this purpose, we employed a plasmid-driven rescue system for the generation of recombinant influenza viruses. For the epitope M158-66 (GILGFVFTL), we performed alanine replacements for each of the nine amino acid positions. In addition, various other amino acid substitutions were introduced in this and four other epitopes, namely the HLA-A*0101-restricted epitopes PB1591-599 and NP44-52, the HLA-B*2705-restricted epitope NP174-184, and the HLA-B*3501-restricted epitope NP418-426. Single mutations at anchor residues could result in the loss of the epitopes, which would constitute the most economical way for the virus to escape from immune surveillance by specific CTL. Therefore, we focused on the mutational analysis of anchor residues of the respective epitopes. The data obtained in the present study on viral fitness and recognition of influenza viruses with mutations in CTL epitopes are discussed in the light of natural evolution of CTL epitopes.
For the generation of recombinant influenza viruses, a bidirectional reverse genetics system based on influenza virus A/Netherlands/178/95 (A/NL/178/95; H3N2), was used. The viral gene segments were amplified by reverse transcription-PCR using segment-specific primers, purified by electrophoresis in agarose gels according to standard methods, and cloned into a modified pHW2000 vector as previously described (11, 24). Subsequently, site-directed mutagenesis was performed (QuikChange site-directed mutagenesis kit; Stratagene, La Jolla, CA) to substitute single amino acids in several influenza virus CTL epitopes, as listed in Table Table1.1. Sequence analysis was performed for all recombinant plasmids, using a Big Dye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) and an ABI PRISM 3100 genetic analyzer (Applied Biosystems), according to the instructions of the manufacturer. All PCR primer sequences and plasmid maps are available on request.
The recombinant bidirectional plasmids were transfected into 293T cells, using the calcium phosphate precipitation method as described previously (11). After 48 h, culture supernatants were harvested and used for subsequent infection of confluent Madin-Darby canine kidney (MDCK) cells (2). After 3 days, culture supernatants were harvested, cleared by low-speed centrifugation, aliquoted, and stored at −80°C until use. The recombinant viruses were designated “A/NL/95-” followed by their specific substitutions. In order to confirm the introduction of the mutations and to exclude the introduction of second site mutations, the nucleotide sequences of the corresponding full-length genes were assessed. Infectious virus titers were determined as previously described (49). Multistep growth kinetics were determined for all recombinant influenza viruses after infection of MDCK cells, using an equivalent multiplicity of infection (MOI) of 0.001 50% tissue culture infectious dose (TCID50) per cell, which was used as a measure of viral fitness. Viral fitness was considered reduced when a statistically significant difference was observed compared to plasmid-derived wild-type virus.
Generation of CD8+-T-cell clones directed against the HLA-A*0201-restricted epitope M158-66, the HLA-B*3501-restricted epitope NP418-426, and the HLA-A*0101-restricted epitopes NP44-52 and PB1591-599 was described previously (5, 58).
B-lymphoblastoid cell lines, established as described previously (53), and three C1R cell lines, kindly provided by P. Romero (HLA-A*0201-transfected C1R cell line), M. Takiguchi (HLA-B*3501-transfected C1R cell line), and P. Cresswell (HLA-A*0101-transfected C1R cell line), were used as target cells. Peptide labeling was performed by incubating 106 cells/ml overnight with 5 μM peptide in RPMI 1640 medium (Cambrex, East Rutherford, NJ) supplemented with 10% fetal calf serum (FCS) and antibiotics (R10F). Peptides were manufactured, high-performance liquid chromatography purified (immunograde, >85% purity), and analyzed by mass spectrometry (Eurogentec, Seraing, Belgium). For infection with the recombinant influenza viruses, 106 target cells were infected at an MOI of 3 in a volume of 1 ml. After incubation for 1 h at 37°C, the cells were resuspended in R10F and incubated for 16 to 18 h.
The CD8+-T-cell clones were adjusted to a concentration of 106 cells/ml in R10F supplemented with Golgistop (monensin; Pharmingen, Alphen a/d Rijn, The Netherlands). Sixty thousand effector cells were incubated with 3 × 105 stimulator cells, which were infected, pulsed with peptides, or left untreated, for 6 h at 37°C in U-bottom plates. Subsequently intracellular gamma interferon (IFN-γ) staining was performed as described previously (6). In brief, the cells were washed with phosphate-buffered saline (PBS) containing 2% FCS (P2F) and Golgistop, stained with monoclonal antibody (MAb) directed to CD3 (Becton Dickinson, Alphen a/d Rijn, The Netherlands) and CD8 (Dako, Glostrup, Denmark), fixed and permeabilized with Cytofix and Cytoperm (Pharmingen), and stained with a MAb specific for IFN-γ (Pharmingen). At least 5 × 103 gated CD3+ CD8+ events were acquired using a FACSCalibur (Becton Dickinson) flow cytometer. The data were analyzed using the software program Cell Quest Pro (Becton Dickinson).
Enzyme-linked immunospot (ELISPOT) assays were performed as described previously (3). In brief, 96-well Silent Screen plates (Nalge Nunc International, Rochester, NY) were coated with 2.5 μg/ml of anti-IFN-γ MAb 1-DIK (Mabtech, Stockholm, Sweden) and blocked with RPMI 1640 medium supplemented with 10% human AB serum (Sanquin Bloodbank, Rotterdam, The Netherlands), antibiotics, and 20 μM β-mercaptoethanol (R10H). Three thousand cells of CD8+-T-cell clones were incubated with 3 × 104 target cells, which were infected, pulsed with peptides, or left untreated, for 4 h. Next, the plates were washed with PBS-0.05% Tween 20 (Sigma Chemical Co., St. Louis, MO), and secreted IFN-γ was detected using biotinylated anti-IFN-γ MAb 7-B6-1 (Mabtech; dilution of 1:5,000). Subsequently, streptavidin labeled with alkaline phosphatase was added, which was visualized with the phosphatase substrate BCIP/NBT (5-bromo-4-chloro-3-indolylphosphate/nitroblue tetrazolium) (Kirkegaard & Perry Laboratories, Gaithersburg, MD). Numbers of spots were counted using an automated image analysis ELISPOT reader (Aelvis; Sanquin Bloodbank, Amsterdam, The Netherlands).
Chromium release assays were performed as described previously (1). In brief, 7.5 × 105 cells per target were labeled with 75 μCi Na2[51Cr]O4 and incubated with CD8+-T-cell clone at effector/target (E:T) ratios of 10, 5, 2.5, and 1. Target cells were also incubated with 10% Triton X-100 or R10F to determine the maximum and spontaneous release, respectively. After 4 h of incubation, the supernatants were harvested (Skatron Instruments, Sterling, VA) and radioactivity was measured by gamma counting. The percentage of specific lysis was calculated by the following formula: [(experimental release − spontaneous release)/(maximum release − spontaneous release)] × 100%. The chromium release assays were performed in quadruplicate, and the data are presented as the average.
The ratio of synonymous and nonsynonymous nucleotide substitutions was calculated using a synonymous/nonsynonymous analysis program (SNAP) (27, 38, 40) at www.hiv.lanl.gov. The NP nucleotide sequences of influenza viruses A/England/878/69 (AY210221) and A/New York/12/2003 (CY000124) obtained from the Influenza Sequence Database (www.flu.lanl.gov) (32) were used in comparison for analysis by SNAP. These viruses were selected since the NP genes belonged to the same lineage of influenza A (H3N2) viruses (30).
In order to obtain an impression of the selective pressure mediated by virus-specific CTL, we performed a synonymous/nonsynonymous analysis with the NP nucleotide sequences of a pair of influenza A (H3N2) viruses consisting of influenza virus A/England/878/69, isolated shortly after the introduction of H3N2 virus in the human population, and a more recent strain, A/New York/12/2003. The NP gene was selected for this type of analysis since 14 of the known epitopes are located within this protein. The synonymous/nonsynonymous (ds/dn) ratio for the sequence encoding the 14 epitopes was 8.67, whereas the ds/dn ratio for the rest of the protein was 19.73, which is suggestive for selective pressure on the CTL epitopes. However, BLAST search of up to 450 influenza A H3N2 virus sequences indicated that all known CTL epitopes, including those located within other viral proteins, retained their anchor residues, with the exception of the HLA-B*0801- and HLA-B*2705-restricted epitopes NP380-388 and NP383-391 (see Discussion).
Since the HLA-A*0201-restricted epitope M158-66 (GILGFVFTL) is highly conserved, we selected this epitope to examine the effect on viral fitness of alanine replacements at each of the nine positions of the epitope. Mutant viruses could be rescued with alanine replacements at all positions within the M158-66 epitope, except for the second position (Fig. (Fig.1A).1A). The alanine replacement at position 59 of the matrix protein, which is the anchor residue of the M158-66 epitope, was detrimental to viral fitness. Although viruses with alanine replacements at the other eight positions were rescued, the virus replication kinetics of these mutants was affected compared to that of wild-type virus (Fig. (Fig.1B).1B). Especially, mutant viruses A/NL/95-M1 F62A and -M1 F64A yielded >100-fold less progeny virus than the wild-type virus at 12 h postinfection. At 48 h postinfection, the differences from wild-type virus were still at least 50-fold. In addition, the L60A substitution caused a reduction of 75-fold in virus production compared to wild-type virus from 24 h onwards. Since the dramatic effect of the alanine replacement at the anchor residue was of special interest, we decided to study the effect of more conservative substitutions at position 59. We also replaced the isoleucine at this position with a leucine and a valine (M1 I59L and I59V) and found that in contrast to M1 I59A these replacements were tolerated by the virus to a certain extent, since these mutant viruses were readily rescued (Fig. (Fig.1C).1C). We also performed multistep growth curves with these mutant viruses. Six and 12 h postinfection, these mutant viruses yielded 100-fold and 30-fold less progeny virus than wild-type virus, respectively. From 24 h postinfection onwards, these differences were no longer statistically significant (Fig. (Fig.1D1D).
Because of the impact of the alanine replacement at the anchor residue of the M158-66 epitope, we decided to study the effect on anchor residue substitutions of the HLA-B*3501-restricted epitope NP418-426 (LPFEKSTVM). This epitope displays a high degree of variability but retained its anchor residues for binding to HLA-B*3501 (4). Replacement of the proline at position 419 or the methionine at position 426 with an alanine at these positions was detrimental to viral fitness (Fig. (Fig.2A).2A). A more conservative substitution at position 419 (NP P419G) also prevented rescue of viable virus. The conservative NP M426I substitution was not detrimental to viral fitness, and only 6-h postinfection virus replication kinetics were significantly impaired (38-fold) compared to those of wild-type virus (Fig. 2A and B).
In addition, conservative amino acid substitutions at the anchor residues of the HLA-A*0101-restricted epitopes PB1591-599 (VSDGGPNLY) and NP44-52 (CTELKLSDY) and the HLA-B*2705-restricted epitope NP174-184 (RRSGAAGAAVK) were introduced. The D593N substitution in PB1 was detrimental to viral fitness. The E46Q and R175K substitutions in the viral NP were not detrimental to viral fitness (Fig. (Fig.3A),3A), although mutant virus A/NL/95-NP E46Q yielded up to 133-fold less progeny virus within the first 24 h postinfection than wild-type virus. For mutant virus A/NL/95-NP R175K, significantly lower virus titers were observed than for wild-type virus from 48 h postinfection onwards (Fig. (Fig.3B3B).
The recognition of HLA-A*0201-positive cells infected with the various M158-66 mutant viruses by specific CTL was determined by intracellular IFN-γ staining, ELISPOT assay, and classical chromium release assays. As shown in Fig. Fig.4,4, the M158-66-specific CTL clone recognized C1R-A2 cells infected with wild-type virus and mutant A/NL/95 virus with the M1 G58A, L60A, or L66A substitution, but not cells infected with A/NL/95 mutant viruses with the M1 G61A, F62A, V63A, or T65A substitution or noninfected cells, as determined by intracellular IFN-γ staining and flow cytometry. These observations were confirmed by ELISPOT (Fig. 4K and L) and by chromium release assays (Fig. 4M and N). A control CTL clone specific for the NP418-426 epitope recognized target cells infected with all mutant A/NL/95 viruses similarly, indicating that the infection of the cells and the processing and presentation of immunogenic peptides were comparable for all viruses (Fig. 4L and N). The recognition of A/NL/95-M1 I59A and F64A could not be tested, since these mutant viruses could not be propagated to sufficiently high titers. C1R-A2 cells infected with A/NL/95 mutant viruses with the more conservative amino acid substitutions M1 I59V and I59L were fully recognized in all three assays (Fig. (Fig.55).
Of four mutants tested, mutant virus A/NL/95-NP M426I was the only virus with an amino acid substitution in the NP418-426 epitope that proved viable and with which CTL recognition was studied. HLA-B*3501- and -A*0201-positive cells infected with wild-type virus were recognized by M158-66-specific CTL and NP418-426-specific CTL by intracellular IFN-γ staining and ELISPOT and chromium release assays (Fig. 6B, E, and G). However, cells infected with A/NL/95-NP M426I were recognized by M158-66-specific CTL, but not by NP418-426-specific CTL (Fig. 6C, F, and H). These results were confirmed by showing that the functional avidity of the NP418-426-specific CTL decreased more than 100-fold by the NP M426I substitution, using serial dilutions of wild-type and mutant peptide in ELISPOT assays (data not shown).
The recognition of influenza virus A/NL/95-PB1 D593N was not tested, since this mutant virus could not be rescued. Although influenza virus A/NL/95-NP E46Q could not be propagated to sufficiently high titers, the recognition of HLA-A*0101-positive cells infected at a low MOI (0.02) was examined by ELISPOT. It was found that the substitution at the anchor residue abrogated recognition by NP44-52-specific CTL (data not shown). Influenza virus A/NL/95-NP R175K was not tested, since no specific CTL clone was available for this epitope.
In the present paper, the effect of amino acid substitutions in CTL epitopes on viral fitness and T-cell recognition was evaluated. It was concluded that functional constraints imposed on CTL epitopes limit escape from virus-specific CTL without loss of viral fitness.
The synonymous/nonsynonymous analysis revealed that in the 90 amino acids that constitute the 14 known epitopes located in the NP, relatively more nonsynonymous mutations occurred between 1969 and 2003 than in the rest of the protein. The hypervariable epitope NP418-426 had a major impact on the lower ds/dn ratio, and 5 out of the 14 partially overlapping epitopes were fully conserved. Some points in this analysis should be taken into consideration. First, since commonly old prototypic strains like A/Puerto Rico/8/34 have been used for the identification of influenza virus CTL epitopes, there is a bias towards the identification of conserved epitopes (12-14, 18, 25, 32, 33, 54, 57, 59). Recent work in our laboratory indicates that a significant number of epitopes are not conserved (Berkhoff et al., unpublished data). Second, the conserved epitopes and the variable epitopes, including the NP418-426 epitope, have in common that they all retained their anchor residues for binding to their corresponding HLA molecules. The only exception to this is an amino acid substitution at position 384 of the NP. The R384G substitution, which is at the anchor residues of the HLA-B*0801- and -B*2705-restricted epitopes NP380-388 and NP383-391, resulted in the loss of their epitopes and abrogated recognition of virus-infected cells by specific CTL (51, 58). However, introduction of a glycine at position 384 of the NP of influenza virus A/Hong Kong/2/68 was detrimental to viral fitness, and several comutations associated with the R384G substitution in epidemic influenza virus strains were required to functionally compensate for the detrimental effect of the R384G substitution (50, 52). Similar findings have been observed for CTL escape mutants of HIV and simian immunodeficiency virus (SIV), which also accumulated extraepitopic comutations in the gag protein for restoration of viral fitness in the presence of mutations in CTL epitopes (17, 26, 42). Apparently, RNA viruses display sufficient flexibility to escape from CTL and retain viral fitness. For HIV and SIV, the selective pressure is mediated by CTL during the chronic infection of individual hosts, while for influenza viruses this takes place by CTL immunity at the population level (19). It is of special interest that also for HIV, CTL escape mutants can be identified at the population level (45), although transmission rates of this virus are much lower than those for influenza viruses. Thus, influenza virus CTL epitopes are either conserved, display variation at non-anchor residues, or lose their anchor residues at the cost of viral fitness, which is functionally compensated for by the accumulation of comutations. To assess the impact of amino acid substitutions in conserved epitopes on viral fitness and recognition by specific CTL, we conducted a mutational analysis of the epitope M158-66 (GILGFVFTL). This epitope is immunodominant and recognized by a large portion of individuals in the population, but is highly conserved. Replacement of the anchor residue at position 2 of the epitope (M1 I59A) was detrimental to viral fitness, whereas alanine replacements at the other eight positions did not prevent rescue of recombinant influenza virus and were tolerated to various extents. The M158-66 epitope is located in the fourth N-terminal α-helix of the M1 protein. Mutations in this region may disturb the functional and structural integrity of the protein, as has been described for mutations in the M1 “helix six” domain (8, 31). The reduced virus titers obtained with a number of these mutant M1 viruses correlated with the number of productively infected cells, as measured by immunofluorescence assay using an NP-specific monoclonal antibody 6 h postinfection of MDCK cells, suggesting that the virus replication cycle was affected at an early pretranscriptional stage (data not shown). Conservative amino acid substitutions at position 2 of the M158-66 epitope (M1 I59L and I59V) were less critical, although the kinetics of viral replication was somewhat affected. More importantly, the A/NL/95-M1 I59L and I59V mutant viruses were fully recognized by M158-66 specific CTL, which makes it unlikely that these variants would ever emerge in the human population. Although some of the other alanine replacements resulted in the partial loss of recognition by M158-66-specific CTL, their impaired replication kinetics is not in favor of the emergence of these mutants. We speculate that there must be a trade-off between viral fitness and immune recognition of which we have little insight at present. The T-cell recognition patterns that were observed here with mutant virus-infected cells were in agreement with those observed with mutant M158-66-peptides in previous studies (1, 20, 41). Although the use of T-cell clones may not reflect the situation in vivo, the analysis of anchor residues boils down to recognition of the epitope or not, which is not different between clonal and polyclonal T-cell populations. In the analysis of T-cell receptor contact residues, as done for the M158-66 epitope, the situation is more complicated. However, the M158-66-specific CTL response is oligoclonal in nature and dominated by T cells carrying the T-cell receptor with Vβ 17 chains (29, 37). Fitness costs also limit variation in the highly immunodominant Gag p11C, C-M CTL epitope of SIV and escape from specific CTL (43). Therefore, this phenomenon may be more universal and apply to more RNA viruses, which are under selective pressure mediated by CTL. It even may contribute to shaping of the T-cell repertoire and have an influence on the hierarchy of epitope dominance.
Next, we wished to evaluate the conservative anchor residues of the otherwise hypervariable epitope NP418-426 (LPFEKSTVM). The relatively conservative NP P419A and P419G substitutions at position 2 of the epitope were both detrimental to viral fitness, indicating that the proline at this position is essential. Amino acid substitutions at position 9 of the epitope, the second anchor residue, yielded interesting results. First, the NP M426A substitution was detrimental to viral fitness. Second, with the conservative NP M426I substitution, the HLA-B*3501 binding motif was retained (16, 23, 56) and viral fitness was not affected to a great extent. Of special interest, HLA-B*3501-positive cells infected with influenza virus A/NL/95-NP M426I were poorly recognized by NP418-426-specific T-cell clones. Since the NP M426I mutant epitope retained its capacity to bind to HLA-B*3501, it may have undergone conformational changes in T-cell receptor contact residues, preventing recognition by CTL, as has been described previously for another HLA-B*3501-restricted epitope (15). Conservative amino acid substitutions at the anchor residues of the epitopes PB1591-599 (VSDGGPNLY), NP44-52 (CTELKLSDY), and NP174-184 (RRSGAAGAAVK) also affected viral fitness. The PB1 D593N substitution in particular was detrimental to viral fitness. Although the conservative NP E46Q substitution resulted in the loss of the anchor residue and would allow the virus to escape from specific CTL, the loss of viral fitness may limit the emergence of this variant in the human population.
Based on the data presented here, we speculate that influenza A viruses display a limited degree of variation in CTL epitopes despite selective pressure on these epitopes mediated by CTL. Functional constraints imposed on influenza virus CTL epitopes may limit efficient escape from CTL and could constitute the Achilles heel of these viruses, limiting the impact of epidemic and pandemic outbreaks of influenza on severe morbidity and mortality.
This work was supported by European Union grant QLRT-2001-01034 (Novaflu).
The authors thank T. M. Bestebroer and M. I. J. Spronken for excellent technical assistance.