PB2 Mutations D701N and S714R Promote Adaptation of an Influenza H5N1 Virus to a Mammalian Host

ABSTRACT

Mutation D701N in the PB2 protein is known to play a prominent role in the adaptation of avian influenza A viruses to mammalian hosts. In contrast, little is known about the nearby mutations S714I and S714R, which have been observed in some avian influenza viruses highly pathogenic for mammals. We have generated recombinant H5N1 viruses with PB2 displaying the avian signature 701D or the mammalian signature 701N and serine, isoleucine, and arginine at position 714 and compared them for polymerase activity and virus growth in avian and mammalian cells, as well as for pathogenicity in mice. Mutation D701N led to an increase in polymerase activity and replication efficiency in mammalian cells and in mouse pathogenicity, and this increase was significantly enhanced when mutation D701N was combined with mutation S714R. Stimulation by mutation S714I was less distinct. These observations indicate that PB2 mutation S714R, in combination with the mammalian signature at position 701, has the potential to promote the adaptation of an H5N1 virus to a mammalian host.

IMPORTANCE Influenza A/H5N1 viruses are avian pathogens that have pandemic potential, since they are spread over large parts of Asia, Africa, and Europe and are occasionally transmitted to humans. It is therefore of high scientific interest to understand the mechanisms that determine the host specificity and pathogenicity of these viruses. It is well known that the PB2 subunit of the viral polymerase is an important host range determinant and that PB2 mutation D701N plays an important role in virus adaptation to mammalian cells. In the present study, we show that mutation S714R is also involved in adaptation and that it cooperates with D701N in exposing a nuclear localization signal that mediates importin-α binding and entry of PB2 into the nucleus, where virus replication and transcription take place.

INTRODUCTION

Mammalian influenza viruses, including influenza viruses of humans, originate from a large reservoir of avian influenza viruses (AIV) that circulate in wild aquatic birds. If avian H5N1 viruses acquire the ability to be transmitted efficiently from human to human, an influenza pandemic may occur, as was the case in 1918, 1957, 1968, and 2009. Since the outbreak of highly pathogenic avian H5N1 influenza viruses (HPAIV) in China in 1996, these viruses have spread from Southeast Asia to Africa and Europe (1–4). Apart from now being endemic in wild aquatic birds and poultry, H5N1 viruses are occasionally transmitted to humans and other mammals such as cats and dogs (5–7). Until now, human infections with H5N1 viruses have been thought to be dead-end infections without or with only very limited human-to-human spread (8, 9). However, the ability of these viruses to cross species barriers is a matter of great concern. In fact, two independent studies with recombinant viruses in ferrets show that a few mutations in the hemagglutinin and polymerase genes are sufficient to allow airborne transmission of H5N1 HPAIV in mammalian hosts (10, 11).

H5N1 isolates obtained from humans and other mammals display mutations believed to be involved in adaptation to the new host. Many of these mutations are located in the PB2 subunit of the polymerase, with the most prevalent ones being E627K and D701N (for a review, see reference 12). Structural analysis of the C-terminal domain of the PB2 protein harboring both of these mutations provided information on some mechanisms underlying adaptation. Residue 627 is part of a largely positively charged surface. While the avian signature 627E alters this surface with its negative charge, mammal-specific 627K increases the positive charge (13). The alteration of the surface may be important for the interaction of PB2 with other viral proteins or with host factors (14, 15). A host factor that interacts with PB2 is importin-α, a component of the nuclear transport machinery. There is evidence that mutation E627K strengthens this interaction, resulting in enhanced polymerase activity in mammalian cells, and that this effect is independent of nuclear transport (16). Residue 701 also regulates the interaction of PB2 with importin-α but by a different mechanism. 701D, most commonly found in IAV, forms a salt bridge with 753R, masking a bipartite nuclear localization signal (NLS). When 701D is replaced with 701N, this salt bridge is disrupted, rendering the NLS easily accessible to importin-α (17). Mutation D701N therefore enhances nuclear entry of PB2 and polymerase activity in mammalian cells (18). Interestingly, adaptation by PB2 mutations E627K and D701N also involves a switch in the usage of different importin-α isoforms. While viruses with the avian signatures (627E and 701D) rely primarily on importin-α3, viruses with 627K and 701N depend on importin-α7 for enhanced replication in mammalian cells (19, 20).

Mutation D701N has been observed in the PB2 protein of about 7% of human H5N1 isolates, including strain A/Thailand/1(Kan-1)/04 (Kan-1 virus) (21). There is also evidence that this mutation contributes to the pathogenicity of H5N1 viruses in mice (19, 22). In the NLS region of PB2 where D701N is located, mutations have also been observed at position 714. Mutations S714R and D701N have been shown to promote the experimental adaptation of an H7N7 virus (SC35M) to mice (23). Mutation S714R has not been observed frequently in nature. The analysis of PB2 sequences of avian, human, and porcine viruses revealed that mutation S714R is present in only 7 (0.08%) of the 8,726 sequences analyzed. Moreover, it is displayed only in mammalian viruses of subtypes H1N1 and H3N2, including two human isolates of the 2009 swine origin pandemic virus. In contrast, avian H5N1 isolates obtained in 2002 that were highly pathogenic for mice displayed an amino acid change at the same site (S714I) without showing mammalian signatures at position 627 or 701 (24). This particular mutation is also present in a small number of avian H6N1 and porcine H1N1 isolates (6 [0.07%] of 8,726). Taken together, these observations suggested that mutations at position 714 of PB2 may alter the host range and pathogenicity of not only H7N7 but also H5N1 viruses.

We have therefore analyzed the effects of mutations D701N, S714I, and S714R in the PB2 protein on the polymerase activity, replication, and mouse pathogenicity of Kan-1 virus. The results show that PB2 mutation S714R, in combination with mutation D701N, promotes mammalian adaptation of an H5N1 influenza A virus.

MATERIALS AND METHODS

Plasmids for recombinant viruses and minigenome assays.

To generate recombinant Kan-1 viruses, we cloned all of the viral genes into bidirectional expression plasmid pHW2000 (25). We first extracted viral RNA with the QIAamp Viral RNA minikit (Qiagen, Hilden, Germany) according to the manufacturer's protocol and then subjected it to reverse transcription (RT)-PCR (One-Step RT-PCR kit; Qiagen) with gene-specific primers (26). For the minigenome assay, we subcloned the polymerase PB2, PB1, PA, and nucleoprotein (NP) genes from pHW2000 into pCAGGS. For this, we performed PCRs with cloning primers also used for pHW2000 cloning and followed the protocol described before (27). Plasmids with single-point and double mutations at residues 701 and 714 in the gene for PB2 were generated with the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) with the following primers: PB2 N701D Forward, ATT CTA GGC AAG GAG GAC AAA AGG TAT GGA CCA; PB2 N701D Reverse, TGG TCC ATA TCT TTT GTC CTC TCT GCC TAG AAT; PB2 S714I Forward, AGC ATC AAT GAA CTG ATC AAT CTT GCA AAA GGG; PB2 S714I Reverse, CCC TTT TGC AAG ATT GAT CAG TTC ATT GAT GCT; PB2 S714R Forward, AGC ATC AAT GAA CTG AGG AAT CTT GCA AAA GGG; PB2 S714R Reverse, CCC TTT TGC AAG ATT CCT CAG TTC ATT GAT GCT (the nucleotides changed for amino acid mutations are in bold). Reporter plasmids pPolI-NP-luc and MG3, encoding firefly luciferase flanked by the noncoding regions of influenza A viruses under the control of the human and chicken pol I promoters, respectively, where kindly provided by Thorsten Wolff, Berlin, Germany, and Martin Schwemmle, Freiburg, Germany. The Renilla luciferase expression plasmid pGL4.73 used for standardization was obtained from Promega (Mannheim, Germany).

Cell culture.

MDCK II, A549, and HEK293 cells were grown and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 1% l-glutamine, and 1% penicillin-streptomycin. For infection, maintenance medium was changed to infection medium containing 0.2% bovine serum albumin (BSA; PAA, Cölbe, Germany) instead of FCS. Chicken LMH cells were grown in RPMI 1640 medium containing 10% chicken serum, 1% l-glutamine, and 1% penicillin-streptomycin. Quail QT6 cells were grown in Ham's F-12 medium containing 10% chicken serum, 1% l-glutamine, and 1% penicillin-streptomycin. Again, in infection medium, chicken serum was replaced with 0.2% BSA. All media and additives were obtained from Gibco/Life Technologies (Darmstadt, Germany) if not stated otherwise. Differentiated cultures of human tracheobronchial epithelial (HTBE) cells were prepared as described previously (28, 29). In brief, primary human tracheobronchial cells (Lonza, Basel, Switzerland) were expanded on plastic dishes and propagated on membrane supports (12-mm Transwell-Clear; Corning Inc., Corning, NY) at an air-liquid interface (ALI) in serum-free, hormone- and growth factor-supplemented growth medium. Fully differentiated 6-week-old cultures were used for the experiments.

Minigenome assays.

Minigenome assays were performed according to the Promega Dual-Luciferase Reporter Assay kit protocol. Briefly, cells were transfected with 50 ng each of pCAGGS-PB1, pCAGGS-PA, pCAGGS-NP, and pCAGGS-PB2; 200 ng of viral minigenome plasmid pPolI-NP-luc (HEK293) or MG3 (LMH and QT6); and Renilla luciferase for standardization. Minigenomes were under the control of the human or chicken RNA polymerase I promoter, respectively. In mock-transfected control cells, pCAGGS-PB2 was replaced with the empty pCAGGS plasmid. Cells were transfected with Lipofectamine 2000 (Invitrogen/Life Technologies, Darmstadt, Germany) according to the manufacturer's protocol, and transfection medium was exchanged for normal growth medium at 6 h posttransfection. After 24 h, cells were lysed and luciferase activity was measured in a Centro LB 960 luminometer (Berthold Technologies, Bad Wildbach, Germany).

Recombinant viruses.

Recombinant Kan-1 viruses were generated as described previously (25). In addition to the bidirectional plasmids encoding the eight viral genes, the viral genes for polymerase PB2, PB1, and PA, as well as the gene for NP, were added as pCAGGS expression plasmids to ensure efficient rescue. Briefly, HEK293 cells were transfected with 1 μg of each plasmid in the presence of Lipofectamine 2000 in Opti-MEM (serum-reduced, optimized Eagle's minimum essential medium) according to the manufacturer's protocol. At 6 h posttransfection, the transfection medium was exchanged with infection medium and cells were incubated at 37°C in 5% CO₂ for 48 h. After incubation, 1 ml of virus-containing supernatant was transferred to MDCK II cells. After 24 to 48 h, recombinant viruses were harvested and stored at −80°C and titers were determined by plaque assay. All experiments with recombinant H5N1 wild-type and mutant viruses were approved by the relevant German authorities, the Regierungspräsidium Giessen and the Behörde für Stadtentwicklung und Umwelt Hamburg, and conducted in biosafety level 3 facilities at the Institute of Virology of the University of Marburg and the Heinrich Pette Institute in Hamburg.

Growth kinetics.

MDCK II, A549, and LMH cells were seeded to 90% confluence the day before infection. Directly before infection, the cells were washed twice with phosphate-buffered saline (PBS). Cells were inoculated with 25 PFU diluted in PBS and incubated for 1 h at 37°C in 5% CO₂. After incubation, cells were washed twice with PBS and 5 ml of infection medium was added. Samples of 100 μl were taken at the time points indicated and stored at −80°C until titration (0-h samples were obtained directly after the addition of infection medium). Plaques were visualized either by crystal violet staining or immunohistochemical staining of NP with antiserum raised against A/FPV/Rostock/34 (H7N1) (Institute of Virology, University of Marburg) and a horseradish peroxidase (HRP)-conjugated secondary antibody (Dako, Hamburg, Germany). Briefly, MDCK cells were infected with 10-fold serial dilutions of harvested samples for 1 h. Overlay medium containing 1.25% Avicel (FMC BioPolymers, Ladyburn Works, Girvan, Scotland, United Kingdom) (30) in Eagle's minimal essential medium was then added to the cells, and plaque formation was allowed for 24 h at 37°C in 5% CO₂. After incubation, cells were fixed with 4% paraformaldehyde (Merck, Darmstadt, Germany) solution, permeabilized with 0.3% Triton X-100 (Sigma-Aldrich, Munich, Germany) in PBS, and subsequently incubated with antiserum (1:2,000), an HRP-conjugated secondary antibody (1:3,000), and True Blue HRP substrate (KPL, Gaithersburg, MD).

Replication of viruses in HTBE cultures was studied as described previously (29). In brief, replicate cultures were washed five times with 0.2 ml of PBS and inoculated with 200 PFU in 0.2 ml of growth medium. Inoculates were removed 1 h later, and the cultures were incubated at 37°C under ALI conditions. At 24, 48, 72, and 96 h postinfection (p.i.), apical sides of the cultures were incubated with 0.2-ml aliquots of DMEM for 30 min at 37°C. The apical washes were harvested, stored at −80°C, and analyzed by plaque titration and immunohistochemical staining of NP as described above.

Animal experiments.

Animal experiments were performed according to the guidelines of the German animal protection law. All animal protocols were approved by the relevant German authority, the Behörde für Stadtentwicklung und Umwelt Hamburg. Four- to 6-week-old BALB/c mice were anesthetized with ketamine-xylazine (100 and 10 mg/kg, respectively) and inoculated intranasally with 50 μl of virus diluted in PBS. Animals were infected with 10¹ PFU and observed for 14 days for weight loss. For determination of organ viral titers, three mice in each group were euthanized on day 3 p.i. Virus titers in lung and brain homogenates were determined by plaque assay.

Analysis of PB2 sequences.

To analyze PB2 sequences for the occurrence of mutations S714R and S714I among IAV, full-length sequences of avian, human, and porcine IAV were downloaded from the NCBI database (http://www.ncbi.nlm.nih.gov/genomes/FLU/FLU.html) and aligned with BioEdit software (v 7.0.5.3; http://www.mbio.ncsu.edu/bioedit/bioedit.html).

RESULTS

Effects of mutations D701N, S714I, and S714R on the activity of the viral polymerase in mammalian cells.

Pathogenicity often depends on the efficiency of the viral polymerase in the respective host. To understand whether mutations D701N, S714R, and S714I in the PB2 subunit of the viral polymerase contribute to an increase in replication in mammals by enhancing the activity of the viral polymerase, we generated PB2 mutants of Kan-1 virus in which the avian signature 701D and the mammalian signature 701N were combined with 714S, 714I, and 714R. The PB2 panel used for analysis is shown in Table 1. It includes six constructs, among which PB2 DS, PB2 DI, PB2 NS, and PB2 NR have the amino acid constellations observed with most avian H5N1 viruses, some H5N1 isolates with high mouse pathogenicity (24), Kan-1 virus, and SC35M virus, respectively.

TABLE 1.

Amino acids at positions 701 and 714 of the PB2 constructs

PB2 construct	Amino acid at position:
	701	714
DS (avian-like)	D	S
DI (HK2001-like)	D	I
DR	D	R
NS (wild-type Kan-1)	N	S
NI	N	I
NR (SC35 M-like)	N	R

In HEK293 cells, comparison of PB2 NS with PB2 DS shows that mutation D701N leads to a 4.5-fold increase in polymerase activity (Fig. 1A). Interestingly, both the PB2 DR and PB2 DI mutants also show a significant increase that, in the case of mutation S714R, is comparable to that caused by mutation D701N. The combination of the mammalian signature at position 701 with 714I and 714R has additive effects; polymerase activity with PB2 NI and PB2 NR was 7- and 30-fold higher, respectively, than with PB2 DS.

FIG 1.

Polymerase activities of PB2 mutants. Shown are the activities in HEK293 cells (A) and LMH cells (B) of polymerase complexes reconstituted by PB1, PA, NP, and PB2 mutations at residues 701 and 714. Polymerase activity was determined by dual-luciferase assay (Promega) in at least three independent experiments. Values relative to PB2 DS activity are indicated. Double asterisks indicate P values of <0.005 (Student's t test).

To study whether the increase in polymerase activity is dependent on the host cell, we also transfected cells from chickens and quails, the two poultry species where the mutation S714I was originally observed (24). Mutation D701N, as well as mutations S714I or S714R, led to only moderately enhanced polymerase activity in chicken cells that, in most instances, did not reach the level observed in mammalian cells (Fig. 1B). The results obtained with quail cells were similar to those obtained with chicken cells (data not shown).

These findings indicate that adaptive mutations in the PB2 protein of H5N1 viruses can occur at residues 701 and 714. Compared to avian cells, the effect of the mutations on the activity of H5N1 virus polymerase is significantly higher in mammalian cells, indicating that these mutations enhance replication and transcription specifically in a mammalian host. The combination of mutations at both sites promotes adaptation in an additive manner, and this effect is most distinct when mutation D701N is combined with S714R.

Effects of mutations D701N, S714I, and S714R on viral growth.

In order to address the question of whether an increase in polymerase activity translates to an increase in replication efficiency, we generated six recombinant variants of Kan-1 virus. They contained the PB2 mutations shown in Table 1 and were designated accordingly.

We then analyzed virus replication in human A549 cells (Fig. 2A). Comparison of the growth curves obtained with the PB2 DS and PB2 NS viruses indicated that mutation D701N resulted in an at least 10-fold increase in the replication rate. Comparison of the PB2 DS virus with the PB2 DI and PB2 DR viruses showed that mutations at position 714 had little effect in combination with the avian signature at position 701. However, when position 701 showed the mammalian signature, both mutations at position 714 further increased replication efficiency, as indicated by comparison of the PB2 NS virus with the PB2 NI and PB2 NR viruses. Similar results were obtained when virus replication in MDCK cells was analyzed. In contrast, mutations D701N and S714I or S714R alone or in combination had no effect on virus replication in avian LMH cells (Fig. 2B).

FIG 2.

Replication of PB2 mutant viruses. Human A549 cells (A) or avian LMH cells (B) were infected with recombinant PB2 mutant viruses (multiplicity of infection, 10⁻⁵). Virus release into cell culture supernatant was titrated by plaque assay at the time points indicated. Each curve is the average of three independent experiments. The asterisk indicates a P value of <0.05 (Student's t test).

We also tested the effects of mutations D701N, S714I, and S714R on virus replication in differentiated HTBE cell cultures, which model the upper respiratory tract in mammals more accurately than permanent laboratory cell lines. As shown in Fig. 3, the PB2 NS virus replicated very efficiently in HTBE cells, whereas in three of four cultures infected with the PB2 DS virus, titers were significantly lower. Mutations at position 714 had little effect in combination with the avian signature at position 701, as is the case in A549 cells.

FIG 3.

Replication of PB2 mutant viruses in HTBE cells. Differentiated HTBE cells were infected with 10³ PFU, and virus was collected at the time points indicated. For each virus, four cultures (numbered 1 to 4) were sampled independently and analyzed by plaque assay. An asterisk(s) indicates a P values of <0.05 (*) or >0.01 (**) (Student's t test).

Taken together, these observations indicate that mutation D701N in the PB2 subunit significantly enhances H5N1 growth in mammalian cells. Combination with mutation S714I or S714R results in a further increase in virus replication.

Acquisition of PB2 mutation E627K in HTBE cells.

Interestingly, virus titers observed 72 h after PB2 DS infection were 3 to 4 log higher in HTBE culture no. 4 than in the other cultures (Fig. 3). The high replication rate suggested that the virus might have mutated. The PB2 genes of the virus present in each culture over a period of 96 h were therefore subjected to sequence analysis. The results showed that the PB2 DS virus had acquired the well-known adaptive mutation E627K in the gene for PB2. The mutant virus was already detected 36 h after infection and rapidly outgrew the original virus within the next 12 h. Sequence analysis revealed the E627K mutation neither in the inoculum nor in the virus from the other three replicate HTBE cultures, indicating that the mutant was either present below the detection limit or emerged during replication in HTBE cells.

Effects of mutations D701N, S714I, and S714R on virus pathogenicity in mice.

Since mutations D701N and S714I/R enhance replication rates in mammalian cells, it was of interest to study their effects on pathogenicity. Mice were chosen as an animal model.

As shown in Fig. 4A, infection with viruses containing the mammalian signature at position 701 led to a pronounced loss of body weight from day 3 on, with the PB2 NR virus having the strongest impact. Surprisingly, when infected with the viruses showing the avian signature at position 701 and 714I or 714R (PB2 DI and PB2 DR viruses), mice did not lose weight, in contrast to the animals infected with the avian-like PB2 DS virus.

FIG 4.

Infection of BALB/c mice with PB2 mutant viruses. Thirteen mice were infected with 10¹ PFU i.n. (or treated with PBS as a control) and monitored for weight loss (A) and survival (B) for 14 days. At day 3 p.i., three mice in each group were sacrificed for determination of virus titers in tissue samples (C). Viral titers in organ homogenates were determined by plaque assay. Standard deviations represent the titers of three mouse organs. An asterisks indicates a P value of <0.05 (Student's t test).

As depicted in Fig. 4B, survival of mice infected with each virus correlated with the loss of weight. After infection with virus showing the avian signature at position 701 and 714S (PB2 DS) 50% of the mice died, whereas all of the mice survived when isoleucine or arginine was introduced at position 714 (PB2 DI and PB2 DR viruses). In contrast, infection with viruses showing the mammalian signature at position 701 killed most (PB2 NS and PB2 NI viruses) or all (PB2 NR virus) of the animals.

Lung and brain viral titers also correlated with weight loss and death (Fig. 4C). The PB2 DS virus replicated in the lung to moderate titers that were significantly reduced when 714S was replaced with I or R. None of the viruses spread to the brain. In contrast, when animals were infected with viruses displaying mammalian signature N at position 701, relatively high lung titers and, in the cases of the PB2 NS and PB2 NR viruses, dissemination into the brain was also observed in one of three animals.

These observations, taken together, indicate that introduction of the mammalian signature N at position 701 significantly enhances the pathogenicity of H5N1 viruses for mice. When combined with 701N, mutation S714R further enhances pathogenicity, whereas mutation S714I has no effect. When combined with the avian signature at position 701, as in the PB2 DI and PB2 DR viruses, these mutations even decrease pathogenicity. These results are explicable by the fact that the wild-type Kan-1 virus carries 701N.

DISCUSSION

PB2 mutation D701N plays an important role in the adaptation of H5N1 viruses to mammalian hosts, as shown previously with isolates obtained from ducks (22). Our studies with the Kan-1 strain of human origin (21) strengthen this concept. They show that mutation D701N leads to enhanced polymerase activity and higher rates of virus replication in MDCK and A549 cells. This increased virus replication, however, was most distinct in differentiated HTBE cell cultures which show many features of the epithelium lining human airways (28, 31). This observation supports the view that at least some steps leading to human adaptation of H5N1 viruses may occur in the respiratory tract, from which most human isolates have been obtained (21, 32). The increase in polymerase activity and replication rate exerted by mutation D701N was paralleled by enhanced pathogenicity of the virus for mice. Viruses with the mammalian signature at position 701 grew to significantly higher titers in the lung and, unlike virus with the avian signature, spread to the brain. Viral dissemination to the brain may result from the high virus load in the lung, or it may be due to other factors such as a differential usage of importin-α isoforms by PB2. The latter concept is supported by the observation that mutation D701N is responsible for a specificity switch from importin-α3 to importin-α7, which may account for differences in tropism (19).

PB2 mutations D701N and E627K are quite common in avian influenza viruses adapted to mammals, but in general, they do not occur in combination. This concept is supported by the observation made here that the Kan-1 mutant with the avian signature 701D spontaneously acquired the adaptive mutation 627K. Similar findings have been obtained before (33, 34). Thus, it appears that both mutations can replace each other, although they presumably differ significantly in mechanistic terms (12, 35–37).

Since the amino acid exchanges S714R and S714I are relatively rare (0.08 and 0.07% of the full-length PB2 sequences deposited in the NCBI database, respectively) and since little is known about their contributions to host range, it was of particular interest to study the effects of these mutations on polymerase activity, virus growth, and pathogenicity. When introduced into PB2 with the avian signature D at position 701, mutations S714I and S714R led to moderately increased polymerase activity in mammalian but also in avian cells. However, our data also show that this increase in polymerase activity does not translate into enhanced replication efficiency and mouse pathogenicity, supporting previous observations that there is no strict linear correlation between these parameters (23). It therefore remains an open question whether substitution S714I, as observed in early H5N1 isolates from chicken and quail with a high pathogenicity for mice (24), contributed to the increased host range of these viruses.

In contrast to the results obtained with viruses displaying the avian signature 701D, mutations at position 714 enhanced polymerase activity and virus growth in mammalian cells as well as mouse pathogenicity, when position 701 had the mammalian signature N. The stimulatory effect was less distinct with 714I, but it was particularly strong with mutation S714R, which led to a 30-fold increase in polymerase activity and 100% lethality for mice.

On the basis of the data presented here and the known structure of the C-terminal domain of PB2 (17), we propose a model that explains how the mutations at positions 701 and 714 cooperate to facilitate NLS exposure. The NLS is located in the flexible region at the extreme C-terminal end of PB2 (amino acids 737 to 759), which is tethered to the main body of the domain by a salt bridge between 753R and 701D (Fig. 5A). NLS unfolding, which allows importin-α binding in mammalian cells, is triggered when the salt bridge is disrupted by mutation D701N, and it is driven further by the repulsive forces between 714R and 737R (Fig. 5B). Unfolding is less efficient when 714S is replaced with an uncharged amino acid such as isoleucine (Fig. 5A). This concept is supported by the findings made here that PB2 NR has higher polymerase activity than PB2 NI and that the PB2 NR virus is more pathogenic than the PB2 NI virus. Taken together, these observations support the concept that mutation S714R promotes the binding of PB2 to importin-α (38), as is the case with mutation D701N (18), and they indicate that mutation S714R has adaptive potential when it coemerges with mutation D701N in H5N1 viruses.

FIG 5.

Interaction of amino acids 701 and 714 with the NLS in the C-terminal domain of PB2. The flexible NLS peptide consisting of amino acids 737 to 759 (purple) is folded back to the globular region containing the mammal-specific mutations at positions 701 and 714 (A). 701D forms a salt bridge with 753R that must be disrupted during NLS unfolding (B). Mutation D701N facilitates unfolding by destroying this salt bridge. Mutation S714R facilitates unfolding owing to repulsive forces between 714R and 737R, which flips around and away from 714R in the unfolded conformation. There is no repulsion when 714R is replaced with 714I. These molecular models were prepared on the basis of the crystal structures 2GDQ (A) and 2GMO (B) (17) with the PyMol Molecular Graphics System (v. 1.6; Schrödinger, LLC).