Oseltamivir-Resistant Influenza A Viruses Are Transmitted Efficiently among Guinea Pigs by Direct Contact but Not by Aerosol

Abstract

Influenza viruses resistant to the neuraminidase (NA) inhibitor oseltamivir arise under drug selection pressure both in vitro and in vivo. Several mutations in the active site of the viral NA are known to confer relative resistance to oseltamivir, and influenza viruses with certain oseltamivir resistance mutations have been shown to transmit efficiently among cocaged ferrets. However, it is not known whether NA mutations alter aerosol transmission of drug-resistant influenza virus. Here, we demonstrate that recombinant human influenza A/H3N2 viruses without and with oseltamivir resistance mutations (in which NA carries the mutation E119V or the double mutations E119V I222V) have similar in ovo growth kinetics and infectivity in guinea pigs. These viruses also transmit efficiently by the contact route among cocaged guinea pigs, as in the ferret model. However, in an aerosol transmission model, in which guinea pigs are caged separately, the oseltamivir-resistant viruses transmit poorly or not at all; in contrast, the oseltamivir-sensitive virus transmits efficiently even in the absence of direct contact. The present results suggest that oseltamivir resistance mutations reduce aerosol transmission of influenza virus, which could have implications for public health measures taken in the event of an influenza pandemic.

Influenza virus causes significant morbidity and mortality worldwide. The only measures currently available to control epidemic influenza are vaccination and antiviral medications, particularly the neuraminidase (NA) inhibitors oseltamivir and zanamivir. However, these interventions are not fail-safe. For example, during the influenza season of 2007 to 2008 in the United States, the proportion of deaths attributed to pneumonia and influenza was above the epidemic threshold for 19 consecutive weeks, and vaccine efficacy against circulating A/H3N2 and B viruses was compromised due to strain mismatch (4). Recent data collected by the World Health Organization demonstrate a sudden increase in oseltamivir-resistant influenza A/H1N1 viruses, predominantly in the northern hemisphere. A single amino acid change from histidine to tyrosine at position 274 in the subtype 1 NA (the H274Y mutation) confers relative resistance to oseltamivir; in the influenza season of 2007 to 2008, 15% of global H1N1 isolates were found to carry the H274Y mutation, which had been seen in fewer than 1% of isolates the prior season (32).

Because resistance to the M2 ion channel inhibitors amantadine and rimantidine is widespread among currently circulating strains, the NA inhibitors are, for all practical purposes, the only effective class of antiviral medications available to prevent and treat influenza disease (8). The NA inhibitors currently approved for clinical use consist of oseltamivir and zanamivir; oseltamivir is generally preferred because it is available in an oral preparation, whereas zanamivir requires a special inhaler for drug delivery. For epidemic and pandemic preparedness, it is critical to understand how mutations that increase resistance to oseltamivir affect the transmissibility of influenza viruses among humans.

In the subtype 2 influenza NA, a glutamic acid-to-valine substitution at position 119 (NA-E119V) has arisen in vivo in persons taking oseltamivir (2, 6, 16, 31; also Tamiflu [oseltamivir phosphate] package insert; Roche Laboratories Inc.). This mutation confers resistance to the drug on the order of 10- to 1,000-fold over the wild type, depending on the virus strain and the assay used to test for susceptibility (1, 2, 16, 23, 33, 34). In the ferret model of influenza transmission, an A/H3N2 virus with the NA-E119V mutation (A/Wuhan/359/95-like) has been shown to transmit efficiently by the contact mode (14, 33).

Here, we show that both a single point mutation (E119V) and a double mutation (E119V plus a change of isoleucine to valine at position 222 [NA-E119V+I222V]) in the NA of a recombinant human influenza virus [A/Panama/2007/1999(H3N2); hereafter, Pan/99] confer oseltamivir resistance. These resistant viruses also transmit efficiently by contact among cocaged guinea pigs, as has been shown for an A/H3N2 virus bearing the NA-E119V mutation in the ferret contact transmission model. However, in an aerosol transmission model, in which guinea pigs are housed in separate cages during the exposure period, the mutant viruses are deficient in transmission relative to the parental virus. Thus, aerosol (large or small respiratory droplet) transmission is compromised by oseltamivir resistance mutations at the active site of the subtype 2 NA, even as contact transmission remains highly efficient. These results suggest that mutations to the active site of NA and, in particular, such mutations that reduce sensitivity to oseltamivir affect the mode of transmission among mammals, which could have implications for public health measures taken in the event of a major influenza epidemic/pandemic.

MATERIALS AND METHODS

Cells and viruses.

Madin-Darby canine kidney (MDCK) cells were maintained in minimum essential medium supplemented with 10% fetal bovine serum, and 100 units/ml of penicillin-100 μg/ml of streptomycin (Pen/Strep). The recombinant influenza viruses used in these experiments (rPan/99 NA-wt, rPan/99 NA-E119V, and rPan/99 NA-E119V+I222V; wt is wild type) were rescued using a 12-plasmid reverse genetics system, as previously described (9, 28). Eight viral RNA expression plasmids were created by standard techniques by inserting the wild-type Pan/99 segment sequences into the pPol1 plasmid. To create the viral RNA expression plasmid pPol1 Pan/99 NA-E119V, the E119V point mutation was introduced into the pPol1 Pan/99 NA plasmid by a Quick Change II site-directed mutagenesis kit (Stratagene), as per the manufacturer's protocol. To create the pPol1 Pan/99 NA-E119V+I222V plasmid, the I222V point mutation was subsequently introduced into the pPol1 Pan/99 NA-E119V plasmid by the same method. To hinder reversion to the wild type, two nucleotides in each codon were changed (GAA to GTT for E119V and ATC to GTA for I222V). The recombinant viruses rPan/99 NA-wt, rPan/99 NA-E119V, and rPan/99 NA-E119V+I222V were rescued as previously described (9, 28), with minor modifications. Individual plaque clones were picked and suspended in 1 ml of phosphate-buffered saline (PBS) supplemented with 0.3% bovine serum albumin (BSA). Virus stocks were prepared by inoculation of 10-day-old embryonated chicken eggs with 100 μl of plaque suspension, followed by incubation at 37°C for 72 h. The sequence of the hemagglutinin (HA) and NA segments of all virus stocks were confirmed by sequencing of reverse transcription-PCR products.

NA complementation assay.

To evaluate plaque size in the presence and absence of exogenous sialidase (33), confluent MDCK monolayers in six-well plates were inoculated with 200 μl of each virus at a titer of 75 PFU/ml. After incubation for 1 h at 37°C, monolayers were overlaid with Dulbecco's modified Eagle's medium-F12 medium (Gibco) with 0.67% Oxoid agar and 1 μg/ml tosylsulfonyl phenylalanyl chloromethyl ketone-treated trypsin, either with or without 1 mU/ml of purified Vibrio cholerae NA (Roche). Plaques were visualized by immunostaining (21) with an anti-NP primary antibody and a horseradish peroxidase-conjugated secondary antibody.

Quantification of oseltamivir resistance.

Relative oseltamivir resistance was determined using an NA-Star influenza neuraminidase inhibitor resistance detection kit (Applied Biosystems) according to the manufacturer's protocol. Each virus was assayed in triplicate on two separate occasions, for a total of six trials per virus. Relative light units (RLUs) at each drug concentration are expressed as a percentage of the RLUs obtained in the absence of drug, with error bars representing one standard deviation (SD). The 50% inhibitory concentration (IC₅₀) was calculated by performing a nonlinear regression (in GraphPad Prism) for each trial; results are expressed as an average of all regressions ± SD.

In ovo growth kinetics.

In two separate experiments, 10-day-old embryonated chicken eggs were inoculated in triplicate with 500 PFU of virus stock. Eggs were incubated at 37°C for 24, 48, or 72 h in one experiment and for 12, 24, or 48 h in the other. At each time point, allantoic fluid from three eggs was harvested and frozen in aliquots at −80°C. Virus titers in allantoic fluid were determined by plaque assay of 10-fold serial dilutions of thawed aliquots on MDCK cells. Results of both experiments are combined, with data points representing the average of the logarithms of the virus titers (log₁₀ virus titer) in PFU/ml for all eggs at each time point ± SD.

Animals.

Five- to six-week-old female Hartley strain guinea pigs weighing 350 to 400 g were obtained from Charles River Laboratories. Animals were allowed access to food and water ad libitum and kept on a 12-h light-dark cycle.

Infection of guinea pigs.

Before infection, all guinea pigs were anesthetized with a mixture of ketamine (30 mg/kg of body weight) and xylazine (5 mg/kg of body weight) administered intramuscularly. Preinfection blood samples were collected from all animals, and then half of the animals in each group were infected intranasally with rPan/99 NA-wt, rPan/99 NA-E119V, or rPan/99 NA-E119V+I222V. An inoculum of 1,000 PFU in 300 μl of PBS-BSA-Pen/Strep was instilled intranasally, 150 μl in each nostril, to the anesthetized guinea pigs, as previously described (17, 18).

GPID₅₀ experiments.

For each virus (rPan/99 NA-wt, rPan/99 NA-E119V, or rPan/99 NA-E119V+I222V), four groups of four guinea pigs each were inoculated intranasally with 1, 10, 100, or 1,000 PFU in 300 μl of PBS-BSA-Pen/Strep, as determined by plaque assay of a representative stock virus aliquot on MDCK cells (18). To confirm the virus dose given in the infection, titers of the same virus dilutions used to inoculate the guinea pigs were again determined by plaque assay on MDCK cells immediately after the inoculation. The infection status of each guinea pig was determined by performing plaque assays on nasal wash samples collected at 2 days postinoculation (dpi). The 50% guinea pig infectious dose (GPID₅₀) was calculated by the method of Reed and Muench (29).

Transmission experiments.

All transmission experiments were performed as previously described (17, 18). Infected guinea pigs were sequestered from naïve animals for the first 24 h after intranasal inoculation of virus. At 24 h postinoculation, four transmission pairs were set up, with each pair consisting of one virus-inoculated guinea pig and one naïve guinea pig. Experiments were carried out either under ambient conditions or in an environmentally controlled chamber (Caron model 6030) set at 20°C and 20% relative humidity (RH), which have been shown to be optimal conditions for influenza virus transmission (17); the conditions in which each experiment was conducted are noted with the results. Nasal wash samples were collected from both inoculated and exposed animals at 2, 4, 6, 8, and 10 dpi; for initial experiments, nasal washes of the mutant viruses were also collected at 12 dpi. In contact transmission experiments, the guinea pig pairs were placed together into a standard guinea pig cage. For aerosol transmission experiments, each animal was placed into a transmission cage, a standard rat cage with an open wire top, which was modified by having one plastic side wall replaced with a metal grille, as previously described (18). Exposed animals were handled before inoculated animals; investigators changed gloves between animals, and work surfaces were sanitized between animals.

Collection of guinea pig nasal wash samples.

Nasal washing was performed by instilling a total of 1 ml of PBS-BSA-Pen/Strep into the nostrils and allowing it to drain onto a sterile dish. Samples were collected in 1.5-ml tubes, centrifuged to pellet debris, and stored at −80°C before analysis by plaque assay.

Quantification of viral titers.

Nasal wash virus titers were determined by plaque assay of 10-fold serial dilutions (from 10° to 10⁻⁵ in PBS-BSA-Pen/Strep) on MDCK cells. Plaques were visualized by immunostaining (21) with Pan/99-infected guinea pig serum, followed by a horseradish peroxidase-conjugated secondary antibody.

RESULTS

Virus rescue and plaque phenotype.

Three recombinant Pan/99-based viruses—rPan/99 NA-wt, rPan/99 NA-E119V, and rPan/99 NA-E119V+I222V—were rescued by reverse genetics, and allantoic fluid stocks were grown up from plaque-purified transfectant virus clones. The NA and HA segments for all rescued viruses were sequenced and found to be identical to the wild-type virus, except for the desired point mutations in the NA segment.

The E119V and E119V+I222V point mutations in the Pan/99 NA resulted in a small-plaque phenotype, relative to the recombinant parental virus (Fig. 1A). The mutant phenotype could be restored to that of the wild-type virus with the addition of exogenous V. cholerae NA to the agar overlay (Fig. 1B); no change was observed in the parental virus plaques in the presence of exogenous NA. These findings suggest that the small-plaque phenotype of the oseltamivir-resistant mutants results from impaired NA function, which can be complemented by an exogenously supplied sialidase.

FIG. 1.

Oseltamivir-sensitive and -resistant viruses demonstrate different plaque phenotypes in vitro but have similar growth kinetics in ovo. (A) After 3 days of incubation at 37°C under an agar overlay, plaques produced in MDCK cells by rPan/99 NA-E119V and rPan/99 NA-E119V+I222V are smaller than those produced by rPan/99 NA-wt. (B) When the agar overlay is supplemented with 1 mU/ml of V. cholerae NA, the parental virus plaques are phenotypically unchanged; however, the mutant virus plaques are enlarged to wild-type size under these conditions. (C) In ovo growth of oseltamivir-sensitive and -resistant viruses is not significantly different between 12 and 72 h postinoculation. Data points represent the average log₁₀ virus titer ± SD in allantoic fluid.

In ovo growth kinetics.

To compare viral growth in ovo, the allantoic cavities of 10-day-old embryonated chicken eggs were inoculated with 500 PFU of each virus and allowed to grow for 12, 24, 48, or 72 h. No significant difference was observed in the growth kinetics of the parental or mutant viruses at any time point (Fig. 1C).

IC₅₀ of oseltamivir.

To confirm that these putative oseltamivir resistance mutations do indeed confer a resistance phenotype in the Pan/99 background, the viruses were subjected to an NA Star chemiluminescent substrate cleavage assay. In this assay, the mutant virus NAs were significantly more resistant to oseltamivir inhibition than the wild-type NA (Fig. 2). The IC₅₀ of oseltamivir for rPan/99 NA-E119V was found to be 139-fold higher than that of the parental virus, and the IC₅₀ for the rPan/99 NA-E119V+I222V mutant was 224-fold higher.

FIG. 2.

Mutant viruses are less susceptible to inhibition by oseltamivir than rPan/99 NA-wt. (A) Log dose inhibition curves for the mutant viruses are shifted to the left of that of the parental virus, indicating greater oseltamivir resistance, as determined in a chemiluminescent substrate cleavage assay. Data points represent the average number of RLUs at each drug concentration, expressed as a percentage of the RLUs obtained in the absence of drug; error bars indicate SDs. (B) For the mutant viruses rPan/99 NA-E119V and rPan/99 NA-E119V+I222V, the IC₅₀s of oseltamivir are 139-fold and 224-fold (respectively) higher than the IC₅₀ for rPan/99 NA-wt. Results are expressed as average IC₅₀ (in nM) ± SD.

GPID₅₀.

To assess whether the oseltamivir resistance mutations alter virus infectivity, the GPID₅₀ for each virus was calculated (Table 1). The GPID₅₀s for rPan/99 NA-wt and rPan/99 NA-E119V were similar, at 66 and 70 PFU per guinea pig, respectively. The GPID₅₀ for rPan/99 NA-E119V+I222V was found to be approximately sevenfold higher, at 474 PFU per guinea pig. In the animals inoculated with the highest doses of each virus, the log₁₀ virus titers of the nasal washes collected at 2 dpi were not significantly different: 6.6 ± 0.6 for rPan/99 NA-wt, 5.6 ± 1.2 for rPan/99 NA-E119V, and 5.8 ± 1.3 for rPan/99 NA-E119V+I222V (data not shown).

TABLE 1.

The infectivities of rPan/99 NA-wt, rPan/99 NA-E119V, and rPan/99 NA-E119V+I222V as measured by the GPID₅₀

Virus	Dose (PFU/guinea pig)	No. of infected animals	No. of uninfected animals	% Infected	GPID50 (PFU/guinea pig)a
rPan/99 NA-wt	2,100	4	0	100	66
	210	3	1	80
	21	1	3	20
	2	0	4	0
rPan/99 NA-E119V	1,500	4	0	100	70
	150	3	1	75
	15	0	4	0
	2	0	4	0
rPan/99 NA-E119V+I222V	2,200	4	0	100	474
	220	1	3	25
	22	0	4	0
	2	0	4	0

^aThe infectivities of rPan/99 NA-wt and rPan/99 NA-E119V are similar; rPan/99 NA-E119V+I222V is slightly less infective, with a GPID₅₀ sevenfold higher.

Transmissibility among guinea pigs.

Transmission experiments were performed in both noncontact (aerosol) and contact models. The aerosol transmission experiments were conducted twice each: series 1, under ambient conditions during November and December 2007; and series 2, in an environmental chamber set at 20°C and 20% RH in February and March 2008 (Fig. 3). Contact transmission experiments were performed once, in the environmental chamber set at 20°C and 20% RH, in April 2008 (Fig. 4).

FIG. 3.

Aerosol transmission of rPan/99 NA-wt is more efficient than that of rPan/99 NA-E119V or rPan/99 NA-E119V+I222V. The parental virus was transmitted by aerosol among seven of eight guinea pig pairs, while the single mutant was transmitted among two of eight pairs, and the double mutant was transmitted among zero of eight pairs. Data represent two experiments (four guinea pig pairs per experiment) per virus. Series 1 was performed under ambient conditions, and series 2 was performed in an environmental chamber at 20°C and 20% RH. The time course of log₁₀ nasal wash virus titers collected from inoculated animals is represented by dotted lines, while the time course of those collected from exposed animals is represented by solid lines. Each inoculated-exposed guinea pig pair is represented by a unique symbol (square, triangle, diamond, or circle).

FIG. 4.

Contact transmission of rPan/99 NA-E119V and rPan/99 NA-E119V+I222V is highly efficient. Each mutant virus transmitted among four of four cocaged guinea pig pairs at 20°C and 20% RH. The time course of log₁₀ nasal wash virus titers collected from inoculated animals is represented by dotted lines, while the time course of those collected from exposed animals is represented by solid lines. Each inoculated-exposed guinea pig pair is represented by a unique symbol (square, triangle, diamond, or circle).

Aerosol transmission of oseltamivir-sensitive and -resistant viruses.

Aerosol transmission of rPan/99 NA-wt was highly efficient, with 88% (seven of eight) of naïve guinea pigs becoming productively infected when placed in a cage next to an inoculated guinea pig (Fig. 3). All seven guinea pigs with positive nasal wash specimens seroconverted, as assessed by a hemagglutination inhibition (HI) assay of postinfection serum tested against rPan/99 virus (data not shown).

In contrast, aerosol transmission of the single mutant rPan/99 NA-E119V was relatively inefficient, with only 25% (two of eight) of naïve guinea pigs becoming infected, as assessed by virus growth from nasal wash specimens. Both transmission events occurred during series 1 (ambient conditions). One of the two guinea pigs infected by aerosol demonstrated borderline seroconversion, with an HI titer of 1:40 after infection; however, this animal had positive nasal wash specimens on three days (8, 10, and 12 dpi), with a peak nasal wash titer of 10⁶ PFU/ml on day 10. The other guinea pig with positive nasal wash specimens seroconverted at a titer of 1:80. A third guinea pig in series 2 (20°C and 20% RH), from which live virus was never obtained by nasal wash, also seroconverted at a titer of 1:80 (data not shown).

Aerosol transmission of rPan/99 NA-E119V+I222V did not occur in either series 1 or 2 (zero of eight animals), despite productive infection in the inoculated guinea pigs. In concordance with the nasal wash data, no exposed guinea pigs seroconverted, as determined by HI assay (data not shown).

In the six aerosol transmission experiments, virus titers collected by nasal wash of the inoculated guinea pigs between 2 and 8 dpi were not significantly different, with a mean area under the curve (AUC) of 25.9 ± 3.6 for the eight animals inoculated with rPan/99 NA-wt, a mean AUC of 24.6 ± 2.3 for those inoculated with rPan/99 NA-E119V (P = 0.41 versus the wild-type virus, by two-tailed t test), and a mean AUC of 26.5 ± 5.1 for animals inoculated with rPan/99 NA-E119V+I222V (P = 0.77 versus the wild-type virus, by two-tailed t test). Mean log virus titers for the three viruses at all time points were not significantly different, although there was a nonsignificant trend toward lower titers at 2 dpi for rPan/99 NA-E119V (mean log PFU/ml of 4.9 ± 1.3) than for rPan/99 NA-wt (mean log PFU/ml of 6.0 ± 1.0) and rPan/99 NA-E119V+I222V (mean log PFU/ml of 6.2 ± 0.3) at 2 dpi. These data indicate that the deficiency in transmission of the mutant viruses was not the result of inadequate infection of or viral growth in the donor guinea pigs.

Contact transmission of oseltamivir-resistant viruses.

To assess the ability of the mutant viruses to transmit by direct contact, one series of contact experiments was performed. Because the parental virus transmitted with high efficiency under noncontact conditions and because we routinely see more efficient transmission in the contact model than in the aerosol model, only rPan/99 NA-E119V and rPan/99 NA-E119V+I222V were assessed in a contact model (Fig. 4). Similar to results in the aerosol transmission experiments, peak nasal wash titers in the inoculated animals were 10⁶ to 10⁷ PFU/ml at 4 dpi. In contrast to the aerosol transmission experiments, however, the mutant viruses transmitted efficiently when the inoculated and exposed guinea pigs were cocaged from day 1 to day 8 postinfection. Both viruses exhibited 100% contact transmission (four of four animals per virus) by plaque assay of nasal wash specimens and by seroconversion, as assessed by HI assay (data not shown).

DISCUSSION

The NA of the influenza A and B viruses is an integral membrane-bound enzyme with receptor-destroying activity; it cleaves the terminal sialic acid moieties to which the viral HA binds, allowing progeny virus to release efficiently from the host cell (26). NA inhibitors interfere with this enzymatic activity, resulting in inefficient virus release; clinically, oseltamivir has been shown to reduce duration of viral shedding and severity of symptoms and to impede infection when used as prophylaxis (11, 12, 24).

In the influenza A and B virus NAs, the active site of the enzyme comprises catalytic residues, which directly contact the sialic acid substrate, and framework residues, which stabilize the active site of the NA but do not make contact with the substrate (3, 5, 7, 10). When the crystal structure of oseltamivir bound to a subtype 9 influenza NA was resolved, the C₃ amino group of oseltamivir was found to form a strong charge-charge type hydrogen bond interaction with the framework residue glutamic acid at position 119 (E119) (15), suggesting that this residue plays a direct role in the positioning of the oseltamivir molecule at the NA active site. Indeed, under the selective pressure of oseltamivir treatment, influenza A/H3N2 viruses with mutations at this residue have been isolated in vivo (2, 6, 16; also Tamiflu package insert; Roche Laboratories Inc.), and resistant viruses have been rescued by reverse genetics (33, 34).

Due to persistent influenza infection in the context of an immunodeficient host, a Canadian child with severe combined immunodeficiency disease was treated with oseltamivir for approximately three months in 2005. During this time, she shed viruses with various NA mutations, including E119V, and, later in the oseltamivir course, a virus with both the E119V and I222V substitutions. In the case report, the authors note that the E119V and I222V mutations together had a synergistic effect on oseltamivir resistance. The NA-E119V mutation increased the oseltamivir IC₅₀ by 131-fold over that of the wild-type NA, while the addition of the I222V mutation to the preexisting E119V substitution more than doubled its oseltamivir resistance, increasing the IC₅₀ by 293-fold over the wild type (2). The values obtained here were very much in accord with these data, with an IC₅₀ for rPan/99 NA-E119V found to be 139-fold higher and an IC₅₀ for rPan/99 NA-E119V+I222V 224-fold higher than that of the parental strain.

Isoleucine-222 begins a highly conserved sequence among influenza A and B NAs, the sequence ILRTQES at residues 222 to 228. It is located in the substrate binding cleft (23) and, with tryptophan-178, forms a hydrophobic pocket for the methyl group of the C₄ acetamide of oseltamivir (15). Substitutions of valine or threonine for this isoleucine (I222V/T) have been generated in vitro after passage of influenza A viruses, in both the N1 and N2 backgrounds, in the presence of oseltamivir (Tamiflu package insert; Roche Laboratories Inc.) and have also been isolated from clinical specimens of influenza A/H1N1 and B viruses (23). However, to our knowledge, no substitutions at I222 of the subtype 2 NA have been reported in a human isolate, other than its appearance in association with the E119V mutation in the child with severe combined immunodeficiency disease (2). The superimposition of the I222V mutation on the more common E119V mutation in the setting of extended oseltamivir treatment is of interest; the selection of the second mutation could occur because it increases oseltamivir resistance, improves viral fitness, or both. In intranasally inoculated guinea pigs, the course of infections, as measured by individual nasal wash titers on 2, 4, 6, and 8 dpi and by the AUC of all nasal wash titers between days 2 and 8 pi, are not significantly different; however, the double mutant rPan/99 NA-E119V+I222V and the wild-type rPan/99 NA-wt demonstrate a trend toward day 2 titers and AUCs higher than rPan/99 NA-E119V values. It is tempting to speculate that the double mutant virus might have a slight in vivo fitness advantage in addition to having higher in vitro oseltamivir resistance; regardless, in these experiments, the I222V mutation did not enhance the transmissibility of the NA-E119V virus by the aerosol route.

We hope that the use of the guinea pig as a new model for influenza virus transmission will facilitate vital research into the determinants of virus spread among mammalian hosts as a means toward understanding human epidemics and pandemics. While mice are commonly used to study influenza virus infection and pathogenesis, they transmit the virus poorly or not at all (18, 30). Ferrets are a well-characterized mammalian model for influenza virus transmission; unlike mice, they are highly susceptible to infection with primary human isolates, and infected ferrets can transmit unadapted influenza virus to uninfected ferrets. However, several practical considerations, such as the ferret's size, cost, availability, and difficulty in handling, have limited the use of this model (20). Like humans and ferrets, guinea pigs can be productively infected by human (i.e., unadapted) influenza virus isolates, have a primarily upper respiratory tract infection, and transmit virus to other guinea pigs in both contact and noncontact situations (18). However, because guinea pigs are relatively small, inexpensive, and easy to handle, guinea pig transmission experiments, even when performed in multiple replicates, require less space, money, and specialized animal husbandry than similar experiments in ferrets.

In ferrets, an A/H3N2 virus with the NA-E119V mutation has been shown to transmit efficiently by contact, while in the same background an arginine-to-lysine mutation at position 292 (NA-R292K) abolished transmission. An A/H1N1 virus with the NA-H274Y mutation has also demonstrated efficient transmission among cocaged ferrets (13, 14, 33). In the present experiments, we show that contact transmission of a recombinant A/H3N2 virus with the oseltamivir resistance mutation NA-E119V is highly efficient in the guinea pig model; these data replicate what has been previously observed in the ferret model. However, in addition, we show that transmission occurs much less frequently, if at all, among guinea pigs in an aerosol transmission model. While the parental rPan/99 NA-wt transmitted efficiently in these experiments (88%), the single point mutation E119V reduced aerosol transmission to 25%, as determined by detection of virus in nasal wash specimens; the addition of the second framework residue mutation I222V to the E119V substitution completely abrogated aerosol transmission in these experiments.

Previous work in this laboratory has shown that ambient conditions in winter are favorable for aerosol transmission of influenza virus while noncontact transmission becomes less efficient in the warmer, more humid months (unpublished observations). Subsequent studies under controlled environmental conditions demonstrated that noncontact transmission is abolished by high temperature (30°C) and high humidity (80% RH) and is inefficient at moderate humidity (50% RH) (17, 19). In series 1 of the aerosol transmission experiments, the ambient humidity during the rPan/99 NA-wt trial was unusually high for November, reaching 55% RH during the exposure period. The ambient humidity may explain why the parental strain transmitted better in series 2, which was controlled at a temperature and humidity more conducive to influenza virus transmission by aerosol. In contrast, in the series 1 rPan/99 NA-E119V transmission experiment, the ambient humidity remained less than 27% RH for the entire exposure period, and two of four naive animals became infected. A second set of animals was exposed to rPan/99 NA-E119V in the environmental chamber; no live virus was isolated from nasal wash specimens in this experiment, but one animal did seroconvert at an HI titer of 1:80, suggesting that a subclinical virus exposure may have occurred.

These results suggest that influenza viruses with mutations at the NA active site are less transmissible than the parental strain under certain conditions and that oseltamivir-resistant viruses may have more restricted modes of transmission than wild-type strains. This mode restriction could arise in either the donor host or the recipient host. In the former, a deficient NA hampers efficient mutant virus release from the infected respiratory epithelium, and thus the amount of virus exhaled in airborne droplets is not sufficient for a full infectious dose to be delivered to the naïve guinea pig in the next cage. In the latter, the recipient animal requires a higher inoculum of the mutant virus to be productively infected, and thus the donor guinea pig must be in very close proximity to deliver a full infectious dose. Because the influenza virus NA has been implicated in virus entry into human tracheobronchial epithelial cells (22) as well as in virus release, we compared the infectivity of the oseltamivir-sensitive and -resistant viruses. Although we found the GPID₅₀ of rPan/99 NA-E119V+I222V to be sevenfold higher than that of the parental virus or rPan/99 NA-E119V, the method of determining the GPID₅₀ is crude, and these values may be within the error of measurement. In fact, in a prior study with nonrecombinant Pan/99 virus, the GPID₅₀ was found to be approximately 5 PFU/guinea pig (18), which is one log lower than the GPID₅₀ determined here for rPan/99 NA-wt. This discrepancy could result from small differences between the wild-type and recombinant viruses; more probably, it results, at least in part, from the error inherent in the method of determining an ID_50.

The oseltamivir resistance mutations in the NA active site likely compromise effective virus release rather than compromising virus infectivity. The sialidase activity is important in preventing progeny viruses from adhering to each other and to the cell surface from which they have budded; it also breaks down the mucins in respiratory secretions to allow effective penetration of virus to and from the epithelial border (22, 25, 27). The nasal wash procedure may dislodge NA-defective virus from epithelial surfaces and wash out virus-rich mucus, which would result in nasal wash virus titers equivalent to the titer of the parental virus. However, the air turbulence created in the guinea pig nasopharynx by breathing and vocalizations may not be forceful enough to disrupt virus agglutination and overcome the deficient release of the mutant viruses. To test this hypothesis, we attempted to boost aerosol transmission of rPan/99 NA-E119V+I222V—which was transmitted only by direct contact—by treating infected guinea pigs with intranasal V. cholerae NA; however, transmission was not observed with or without sialidase treatment, so we are unable to prove that reduced NA activity and deficient virus release are the cause of the observed lack of aerosol transmission.

In epidemic or pandemic influenza, NA inhibitor use, both for treatment and for prophylaxis, is likely to increase dramatically, which would, in turn, increase the opportunities for a pandemic strain to be selected for NA inhibitor resistance. Simple measures undertaken by the general population, such as good hand hygiene, could slow the spread of a resistant virus that transmits efficiently by contact but not by aerosol. Transmission prevention in health care settings—through contact, droplet, and airborne isolation precautions—requires the marshalling of diverse resources; the mode of transmission of NA inhibitor-resistant viruses has drastic implications in the management of infected patients in clinics, hospitals, and nursing homes. Thus, an understanding of how drug resistance mutations affect the frequency and mode of transmission is critical to preparing public health measures aimed at controlling epidemic and pandemic influenza.