Enzymatic Properties of the Neuraminidase of Seasonal H1N1 Influenza Viruses Provide Insights for the Emergence of Natural Resistance to Oseltamivir

Surveillance of the antiviral susceptibility of influenza viruses in Europe revealed the emergence of influenza A(H1N1) viruses naturally resistant to the anti-neuraminidase inhibitor oseltamivir (Tamiflu) [1]. Currently, resistant viruses are most prevalent in Europe (25%) but less prevalent in the Americas (16%) or the Western Pacific region (4%) [2]. In Europe, the prevalence varies between countries, with highest levels in Norway (66.5%) and France (46.6%) [3]. These frequencies are in sharp contrast with those observed for H1N1 viruses during previous seasons (0 to <1%) [4]–[8].

Resistance was linked to the H275Y mutation (H274Y in N2 numbering) of the N1 known to confer high level resistance to oseltamivir but not to the other antineuraminidase inhibitor, zanamivir (Relenza) [9][10][11][12]. Resistant H1N1 viruses with the H275Y change have been isolated from patients treated with oseltamivir and more frequently in children, especially in Japan, the country with the highest per capita usage of oseltamivir [10,13]. The current frequencies of resistant H1N1 viruses are not correlated with oseltamivir usage, which suggests that selective drug pressure has not been associated with continued transmission, although it may have been involved in their initial emergence. Clinical H1N1 isolates with the H275Y mutation were previously found to be generally less fit in terms of replication, infectivity for mice or ferrets, or transmission potential [14,15], although the mutation had a less pronounced and variable effect on virus fitness for laboratory strains such as WSN or PR8 viruses or for H5N1 viruses [9,[16][17][18]. To understand the molecular basis of the apparent fitness of the resistant H1N1 viruses that emerged during the 2007-2008, season we determined the enzymatic characteristics of their neuraminidase.
A selection of H1N1 viruses isolated by the National Influenza Center (Northern-France) from specimens received in the frame of routine surveillance through the GROG sentinel network between weeks 35/2007 and 03/2008 (Table 1) were studied. Using a standard neuraminidase inhibition assay, the IC50 values for oseltamivir ranged from 1.3 to 5.9 nM for sensitive viruses and were much higher (IC50, 624 to 942 nM) for resistant viruses (Table 1), as previously published [9,10,12,19]. All viruses were sensitive to zanamivir (IC50, 1.2 to 3.0 nM). All resistant viruses harbored the H275Y substitution in their N1.
Kinetic analyses of sialidase activities of the neuraminidase were performed using the MUNANA fluorogenic substrate in the absence or presence of neuraminidase inhibitors on whole virus suspensions as described [20]. The Michaelis-Menten constant (K m ), which reflects the affinity for the substrate, and the V m , which reflects the activity of the enzyme, were determined ( Table 1  reported for resistant H1N1 viruses (105 to 200 nM; [11,15,20]). Thus, the neuraminidase of H1N1 viruses from the 2007-2008 season exhibits an increased affinity for the substrate as well as for the two antineuraminidase inhibitors and a higher activity as compared to previously circulating viruses such as NC99 or SI06, except for isolate #0006/07, which behaved as an intermediate. As a result, the neuraminidase from recent resistant viruses that harbor the H275Y substitution has a slightly higher activity and affinity for the substrate than that from previously circulating sensitive viruses. These features may contribute to their overall fitness and ability to be transmitted, although the contribution from other genes cannot be excluded at present. When comparing the growth characteristics in vitro on MDCK SIAT-1 cells of the resistant viruses with that of sensitive viruses from the 2007-2008 season or from previous seasons, no significant differences in growth kinetics or final virus titers were observed ( Figure 1). These results indicated that, at least in vitro, the presence of the H275Y mutation did not significantly impair the fitness of the viruses, unlike what had been previously reported in the case of the A/Texas/36/ 91 virus on MDCK cells [15]. Whether the same holds true in vivo remains to be determined.  Figure 2). No specific amino acid changes that could be compensating for the presence of the H275Y substitution in the N1 were found in the H1 of resistant viruses. For instance, sensitive (#0497/07, #1149/08) and resistant (#0644/07 and #1170/08) viruses with the same HA and NA (except for the H275Y and G354D changes) amino acid sequences representing the consensus sequences of the recent H1N1 viruses had similar growth characteristics, similar V m values for their N1, and differed in their K m and K i values based solely on the two changes in the N1.
In addition to the H275Y change, most, but not all, resistant viruses were characterized by the presence of a G354 as for the NC99 and SI06 viruses, whereas a D354 was found for sensitive viruses (Table 1). According to the three-dimen-sional structure of the N1 of an H5N1 avian influenza virus [21], residue 354 is located on the top external side of the neuraminidase tetramer at a distance from the catalytic site and subunit interfaces. It is therefore not likely to be compensating for the H275Y substitution. Indeed, as shown for isolate #0963/08 as compared to other resistant isolates, the presence of a D354 rather than a G354 does not have a major impact on the enzymatic characteristics of the N1 ( Table 1). Substitutions that distinguish the majority of H1N1 viruses from the 2007-2008 season from both NC99 and SI06 are H45N, K78E, E214G, R222Q, G249K, T287I, K329E, and D344N. Two of these positions are located in the stalk region (45 and 78), and three (222, 249, 344) in the vicinity of the catalytic site according to the threedimensional structure of the N1 [21]. Substitutions in the vicinity of the sub-  Overall, our results suggest that a specific combination of amino acids may have resulted in an increased affinity of the N1 of recent H1N1 viruses for its substrate and neuraminidase inhibitors. It will be of interest to determine more precisely which exact changes are involved through mutagenesis using the previously described transient N1 expression system for kinetic analyses of the neuraminidase activity [20].
Appropriate functional balance between the activities of the two influenza virus glycoproteins towards sialic acids, i.e., receptor binding (hemagglutinin) and sialidase activity (neuraminidase), is essential for virus fitness [23]. The H1 of viruses from the 2007-2008 season differ from both NC99 and SI06 by three substitutions (D35N, R188K, E273K), none of which are involved in direct interactions with the receptor and therefore not likely to result in changes of affinity of the H1 for the receptor. According to this hypothesis, which warrants further experiments, the increased affinity of the N1 of 2007-2008 viruses for its substrate would not have been compensated by an increased affinity of the H1 for the receptor. Therefore, viruses with a Y275 that have only a slightly higher affinity for the substrate as compared to H1N1 viruses that circulated previously may have a more appropriate balance of their hemagglutinin and neuraminidase activities than viruses with a H275 that have a 3-fold increased affinity of their neuraminidase for the substrate. As a result, as for influenza A viruses resistant to adamantanes [24][25][26][27][28], the Figure 2. Phylogenetic Relationships of the H1 and N1 Genes. The phylogenetic analysis was performed on the alignment of sequences from nucleotides 87 to 995 (H1) or 90 to 1286 (N1) (numbering from ATG). The dendrogram was constructed by genetic distance matrix and calculated with the DNADIST program using the Kimura-2 parameters model with transition-to-transversion ratio of 2.0 and neighbor-joining analysis in the PHYLIP package [32,33]. Sequences from the H1 or the N1 from A/NewCaledonia/20/99 were used as outgroup. Bootstrap values of 1,000 replicas are given as percentages at the nodes. Isolates from Northern-France from the 2006-2007 season are in italics. Viruses with a Y275 in the neuraminidase sequence are shown in bold and vaccine strains in capitals. Published sequences were issued from the influenza sequence database at Los Alamos National Laboratory [22]. doi:10.1371/journal.ppat.1000103.g002 recent resistant viruses would not be outcompeted upon circulation in the community. It should be emphasized, however, that the relative fitness and ability to be transmitted of the resistant versus sensitive viruses may be modulated by characteristics of other genes. This will require whole genome sequencing. The circulation of H1N1 viruses naturally resistant to oseltamivir underlines the fact that genetic variations may result in variations in sensitivity to oseltamivir in the absence of selective drug pressure, as shown for H5N1 viruses [20,29]. Genetic variations of the hemagglutinin and neuraminidase are mainly driven by the immune response, and adventitious properties that result in changes in fitness may be co-selected. Such a phenomenon could potentially take place for H5N1 viruses and also for H3N2 viruses. Genetic variations like these emphasize the need to carefully monitor the affinity of the neuraminidase for its substrate and antineuraminidase inhibitors in relation with the binding affinity of the hemagglutinin for its receptor for influenza viruses circulating in the population, as well as for avian influenza viruses with pandemic potential.

Sequence Accession Numbers
GenBank accession numbers are listed in Table 2 for the viruses included in this report.