Functional Compensation of a Detrimental Amino Acid Substitution in a Cytotoxic-T-Lymphocyte Epitope of Influenza A Viruses by Comutations

Abstract

Influenza A viruses accumulate amino acid substitutions in cytotoxic-T-lymphocyte (CTL) epitopes, allowing these viruses to escape from CTL immunity. The arginine-to-glycine substitution at position 384 of the viral nucleoprotein is associated with escape from CTLs. Introduction of the R384G substitution in the nucleoprotein gene segment of influenza virus A/Hong Kong/2/68 by site-directed mutagenesis was detrimental to viral fitness. Introduction of one of the comutations associated with R384G, E375G, partially restored viral fitness and nucleoprotein functionality. We hypothesized that influenza A viruses need to overcome functional constraints to accumulate mutations in CTL epitopes and escape from CTLs.

Cytotoxic T lymphocytes (CTLs) play an important role in the control of viral infections, including those caused by influenza viruses (9, 18). Viruses exploit many strategies to evade recognition by virus-specific CTLs, including the accumulation of amino acid substitutions in or adjacent to CTL epitopes (7, 10). As a result, the epitope may no longer bind to its corresponding major histocompatibility complex class I molecule, be liberated from its native protein, and/or be recognized by the CTL specific for that epitope. This escape mechanism has been described predominantly for persistent virus infections. However, epitopes from the influenza virus nucleoprotein (NP) also exhibit amino acid variation associated with escape from recognition by CTLs (2, 15, 17). The rapid fixation of these mutations was explained by small selective advantages and population dynamics in a theoretical model, using the R384G mutation in the HLA-B*2705-restricted NP_383-391 epitope as an example (4). The R384G mutation resulted in the loss of the anchor residue and as a result ablated recognition by CTLs (15, 17). The loss of this epitope affected the human in vitro CTL response significantly (1). During this study, it appeared that the R384G substitution in the NP gene of influenza virus strain A/Hong Kong/2/68 (A/HK/68) prevented rescue of virus by reverse genetics, indicating that it was detrimental to viral fitness. Consultation of the influenza virus sequence database at http://www.flu.lanl.gov (8) revealed a number of comutations associated with the R384G substitution. We hypothesized that one or more of these comutations were functionally compensating for the detrimental effect of the R384G mutation. This hypothesis was tested with recombinant influenza viruses with the R384G mutation alone or in combination with one of the comutations.

Seven amino acid (aa) substitutions were associated with the R384G substitution (Table 1). However, some of these comutations (E18D, I197V, M239V, and S259L) were not exclusively associated with the presence of the R384G substitution, indicating that these additional amino acid substitutions may not be essential for viral fitness. The remaining three substitutions, R65K, D127E, and E375G, were tightly associated with the R384G substitution. Of these three, the substitution at position 375 was in closest proximity to the CTL epitope, and the effect of this comutation on NP functionality and fitness of viruses with and without the R384G mutation was studied. Also in human immunodeficiency virus type 1 CTL escape mutants, clustered mutations have been observed with a mutation in an HLA-B27-restricted CTL epitope in the gag protein (6)

TABLE 1.

Amino acid substitutions associated with the R384G substitution^a

Substitution	No. of virus strains with substitution/total (%)
	384R	384G
18E	46/46 (100)	2/6 (33)
65R	46/46 (100)	0/6 (0)
127D	46/46 (100)	0/6 (0)
197I	46/46 (100)	1/6 (17)
239M	46/46 (100)	2/6 (33)
259S	130/137 (94)	0/76 (0)
375E	136/137 (99.3)	0/76 (0)
18D	0/46 (0)	4/6 (67)
65K	0/46 (0)	6/6 (100)
127E	0/46 (0)	6/6 (100)
197V	0/46 (0)	5/6 (83)
239V	0/46 (0)	4/6 (67)
259L	8b/137 (6)	76/76 (100)
375G	1/137 (0.7)	76/76 (100)

^aFull-length NP amino acid sequences of influenza virus A/NL/18/94 and 51 influenza A (H3N2) viruses obtained from the influenza virus sequence database at LANL (8) and a 151-aa fragment (aa 240 to 391) of 162 Dutch influenza virus A (H3N2) strains (17) were analyzed. Seven amino acid substitutions were identified that were associated with the R384G substitution.

^bThese eight viruses with 259L circulated between 1968 and 1973.

For the generation of recombinant influenza viruses with R384 and G384, respectively, the NP gene segments of influenza viruses A/HK/68 and A/Netherlands/18/94 (A/NL/94) were amplified by reverse transcription-PCR as described previously (1) and inserted between the human polymerase I promoter and the hepatitis delta virus ribozyme sequence of plasmid pSP72-PhuThep (3). Site-directed mutagenesis of the respective NP genes was performed by PCR as previously described (16). SacI fragments (encoding aa 152 to 465) of the respective NP were exchanged with SacI fragments in the genomic NP construct of influenza A virus A/Puerto Rico/8/34 (A/PR/34) to obtain chimeric constructs with A/PR/34 sequences at the distal 5′ and 3′ends of the NP gene segments.

The plasmids pHMG-NP, pHMG-PB1, pHMG-PB2, and pHMG-PA, encoding the NP and the polymerase proteins PB1, PB2, and PA of influenza virus A/PR/8/34, respectively, were kindly provided by P. Palese (12). The bidirectional reverse genetics plasmids pHW181 through pHW188 for the transcription of viral gene segments of influenza virus A/WSN/33 were kindly provided by R. G. Webster (5). For the generation of viruses, the NP constructs of A/NL/94 or A/HK/68 were transfected into 293T cells together with pHMG-NP and all A/WSN/33 genomic constructs except the one encoding the NP gene segment, as described previously (1). Twenty-four hours after transfection, virus was passaged in Madin-Darby canine kidney (MDCK) cells and infectious virus titers were determined as described previously (14). The viruses were named after the NP construct that was used for their generation (Fig. 1). Only infectious virus was detected in the culture supernatant of 293T cells transfected with plasmids encoding the NP gene segments of the NL/94 wild type (WT) and NL/94 G384R mutant (data not shown). However, upon passage in MDCK cells, virus replication was observed with the NP gene segments of HK/68 WT, chimeric PR/34-HK/68 WT, NL/94, and PR/34-NL/94, the latter two with or without the G384R substitution. No virus was detected by using constructs containing NP sequences of HK/68 with the R384G substitution (HK/68 R384G or PR/34-HK68 R384G) (Fig. 2), confirming that this mutation is detrimental to viral fitness. For A/HK/68, the virus titers obtained after passage in MDCK cells were reproducibly higher with the chimeric constructs than with the full-length A/HK/68 gene segments, which may be related to more efficient packaging. For A/NL/94, it was the other way around. The reason for this observation is not clear.

FIG. 1.

NP constructs used in the present study. For the generation of genomic constructs, the NP gene segments of influenza viruses A/HK/68 (black) and A/NL/94 (gray) were amplified by PCR and cloned under control of the human polymerase I promoter. Amino acid substitutions at positions 375 and 384 were introduced by site-directed mutagenesis as indicated. In addition, chimeric NP gene segments were constructed by placing the SacI fragment encompassing the region encoding aa 152 to 465 of the NP derived from A/HK/68 and A/NL/94 in a background of A/PR/34 (hatched). WT indicates unmutated NP sequences of the respective NP genes.

FIG. 2.

Amino acid substitution R384G is detrimental to fitness of viruses containing the NP sequences of A/HK/68. Upon transfection of 293T cells and subsequent rescue in MDCK cells, infectious virus titers were determined for viruses made with chimeric NP gene segments (A) containing sequences obtained from A/PR/34 and A/HK/68 or A/NL/94, with or without reciprocal amino acid substitutions at position 384 as indicated, or full-length NP gene segments of A/HK/68 or A/NL/94 (B). The results of a representative experiment are shown. Amino acid substitution E375G restores fitness of viruses with A/HK/68-R384G NP sequences (C to F). Upon transfection of 293T cells and subsequent rescue in MDCK cells, infectious virus titers were determined for viruses made with chimeric NP gene segments (C and E) containing sequences obtained from A/PR/34 and A/HK/68 and full-length A/HK/68 NP sequences (D and F) without (WT) or with amino acid substitutions E375G and R384G in NP as indicated. The results of two independent experiments are shown. For influenza viruses A/HK/68 WT (•), A/HK/68-E375G (○), and A/HK/68-E375G/R384G (▾), grow curves were determined (G and H) after infection of MDCK cells at an MOI of 0.001 (G) or 0.01 (H). The data represent the average of two experiments.

Introduction of the E375G mutation in the HK/68-R384G or PR/34-HK/68 NP-R384G gene restored the possibility to rescue virus by reverse genetics (Fig. 2C and D). The titers that were obtained with the E375G/R384G double mutants were not as high as in the original virus, indicating that multiple comutations may be required to fully restore viral fitness. The E375G mutation by itself did not influence the fitness of influenza virus A/HK/68 to a great extent. This was also confirmed by assessing the multistep growth kinetics of these viruses after infection of MDCK cells by using an equivalent multiplicity of infection (MOI) of 0.001 and 0.01 50% tissue culture infectious dose (TCID₅₀) per cell (Fig. 2G and H). The addition of the G375E substitution to the G384R substitution in the NP of A/NL/94 did not affect the fitness of these viruses to a great extent (Fig. 2B). However, the introduction of the G375E substitution alone was detrimental to viral fitness. This confirms that the 375E/384G combination affects viral fitness, as was already observed with the A/HK/68 NP constructs.

An NP transcomplementation assay was performed as described previously (17) to correlate differences in viral fitness with functionality of the NP variants. The NP-encoding sequences of influenza virus A/HK/68 variants were cloned into a modified version of the eukaryotic expression plasmid pcDNA3.

These plasmids were transfected into 293T cells with plasmids pHMG-PB1, pHMG-PB2 and pHMG-PA and RF419, from which the green fluorescent protein (GFP) gene flanked by the influenza A virus noncoding region of the NS gene segment is transcribed in a negative orientation. Transcription of the GFP mini-replicon was assessed by GFP expression (Fig. 3). The plasmid from which the HK/68-R384G NP was expressed consistently resulted in fewer GFP-positive cells than after transfection with WT HK/68 NP or HK/68-E375G NP. Introduction of the comutation E375G in the HK/68-R384G gene increased the GFP expression in transfected cells, although the percentage of GFP-positive cells was not as high as that observed with the WT HK/68 NP. Thus, the negative effect of R384G on viral fitness correlated with reduced functionality of the RNP complex, which was partially overcome by the E375G comutation. Since the NP has multiple functions and interactions with viral and host cell proteins (13), its conformational integrity and functionality need to be retained for viral fitness. A similar finding was described recently for simian immunodeficiency virus escape from CTL recognition at a structurally constrained epitope in the gag protein (11).

FIG. 3.

Functional analysis of A/HK/68 NP with or without amino acid substitution at positions 375 and 384. Functionality of the NP was tested in a transcomplementation assay. 293T cells were transfected with plasmids pHMG-PB1, pHMG-PB2, pHMG-PA, and RF419 and a plasmid encoding the NP of A/HK/68 with or without substitutions at positions 375 and 384 as indicated. Cells transfected with plasmid pcDNA3 served as a negative control, while cells cotransfected with pHMG-NP were used as positive control. The results are shown as the percentage of cells showing GFP expression, as measured by flow cytometry. The results of three independent experiments are shown. With the plasmid pEGFP-N1, typically transfection efficiencies of >95% were obtained.

Collectively, the data show that an amino acid substitution in a CTL epitope, which allows the virus to escape from the recognition by virus-specific CTLs, is not tolerated because of functional constraints. Only when the R384G substitution occurs in the presence of other functionally compensating amino acid substitutions is this otherwise detrimental amino acid substitution tolerated. We hypothesize that influenza viruses can exert sufficient flexibility to fix these mutations in the NP, although it should be realized that these events are relatively rare in comparison with the mutation rate in other proteins, like hemagglutinin. It also may explain that in some instances even immunodominant CTL epitopes, like the HLA-A*0201-restricted M1_58-66 epitope from the matrix protein, are fully conserved. For escape from specific CTLs, additional substitutions may be required to overcome functional constraints imposed by the amino acid sequence of this epitope.