Susceptibility of Highly Pathogenic H5N1 Influenza Viruses to the Neuraminidase Inhibitor Oseltamivir Differs In Vitro and in a Mouse Model

Abstract

While the neuraminidase (NA) inhibitor oseltamivir is currently our first line of defense against a pandemic threat, there is little information about whether in vitro testing can predict the in vivo effectiveness of antiviral treatment. Using a panel of five H5N1 influenza viruses (H5 clades 1 and 2), we determined that four viruses were susceptible to the drug in vitro (mean 50% inhibitory concentration [IC₅₀], 0.1 to 4.9 nM), and A/Turkey/65-1242/06 virus was slightly less susceptible (mean IC₅₀, 10.8 nM). Two avian viruses showed significantly greater NA enzymatic activity (V_max) than the human viruses, and the five viruses varied in their affinity for the NA substrate MUNANA (K_m, 64 to 300 μM) and for oseltamivir carboxylate (K_i, 0.1 to 7.9 nM). The protection of mice provided by a standard oseltamivir regimen (20 mg/kg/day for 5 days) also varied among the viruses used. We observed (i) complete protection against the less virulent A/chicken/Jogjakarta/BBVET/IX/04 virus; (ii) moderate protection (60 to 80% survival) against three viruses, two of which are neurotropic; and (iii) no protection against A/Turkey/65-1242/06 virus, which induced high pulmonary expression of proinflammatory mediators (interleukin-1α [IL-1α], IL-6, alpha interferon, and monocyte chemotactic protein 1) and contained a minor subpopulation of drug-resistant clones (I117V and E119A NA mutations). We found no correlation between in vitro susceptibility and in vivo protection (Spearman rank correlation coefficient ρ = −0.1; P > 0.05). Therefore, the in vivo efficacy of oseltamivir against highly pathogenic H5N1 influenza viruses cannot be reliably predicted by susceptibility testing, and more prognostic ways to evaluate anti-influenza compounds must be developed. Multiple viral and host factors modulate the effectiveness of NA inhibitor regimens against such viruses and new, more consistently effective treatment options, including combination therapies, are needed.

Highly pathogenic avian H5N1 influenza viruses have spread intercontinentally and evolved into 10 phylogenetically distinct hemagglutinin (HA) clades; the most diverse, clade 2, comprises five subclades (33). Large outbreaks among poultry continue in far-ranging geographical areas, although human infections remain rare (411 confirmed cases since May 2003) (34). However, the pandemic potential of H5N1 influenza viruses should not be underestimated, and preparedness requires that appropriate prophylactic and therapeutic antiviral regimens be established. Importantly, human H5N1 infection differs markedly from human seasonal influenza (35). Viral pneumonia is considered a primary cause of death from H5N1 infection, but disseminated disease and multiorgan failure with renal and cardiac dysfunction, Reye's syndrome, and hemorrhage often occur (1, 4, 38). Infectious virus and viral RNA have been isolated from the upper and lower respiratory tract, brain, intestines, feces, blood, cerebrospinal fluid, and even from the placentas and fetuses of pregnant women (9, 35).

Antiviral drugs can play an important role in the initial response to pandemic influenza. One of the two classes of anti-influenza drugs, M2-ion channel blockers (amantadine and rimantadine), has limited usefulness, because clade 1 H5N1 viruses are frequently resistant (3, 22), although representatives of clade 2 are susceptible to adamantanes (15, 26). Most H5N1 isolates are susceptible in vitro to the second class of drugs, neuraminidase (NA) inhibitors (oseltamivir and zanamivir) (12). Natural genetic variations in NA were reported to affect the susceptibility of H5N1 viruses to oseltamivir in vitro (23), and some clade 2 viruses were found to be 15 to 30 times less susceptible to oseltamivir than clade 1 viruses, based on their 50% inhibitory concentrations (IC₅₀s) (18). Reduced susceptibility may be caused by NA antigenic mutation(s) and by the emergence of specific NA mutations under drug selection pressure (18, 23). NA mutations at positions 274 (H→Y) and 294 (N→S) are considered markers of the oseltamivir-resistant H5N1 phenotype (6, 17).

The NA enzyme inhibition assay measures the decrease in functional NA activity in the presence of the drug. This assay is considered the most reliable in vitro method of quantifying the susceptibility of seasonal influenza viruses to NA inhibitors, and it is well correlated with their susceptibility in animal models (29). However, it is unknown whether in vitro data can accurately predict the effectiveness of antiviral drugs against H5N1 viruses in vivo, since viral and host factors that modulate disease manifestations are incompletely understood (20). Experimental animal models are a logical approach to estimating drug effectiveness in vivo against lethal influenza virus infection. Studies in mice showed that more prolonged oseltamivir treatment is required to inhibit residual replication of a highly virulent representative of clade 1, A/Vietnam/1203/04 (H5N1) virus, than to inhibit a less virulent 1997 isolate (36). In a ferret model, the best antiviral effect against H5N1 virus was achieved by increasing the dose of oseltamivir and initiating treatment early (8). These observations show that the optimal dose and duration of an anti-H5N1 regimen may depend on virus virulence, although other viral factors can play a role. Some characteristics, such as the ability to spread systemically, tissue tropism (including neurotropism), virus fitness, the characteristics of individual virus proteins, and a preference for binding to α2,3- or α2,6-linked sialic acid receptors, clearly differ among H5N1 viruses and may affect the protection offered by antiviral therapy. It is also unknown whether the hypercytokinemia reported in human cases of H5N1 infection (7) represents an appropriate immune response or immune dysregulation that may alter the outcome of drug therapy.

In the present study, we compared the in vitro NA inhibitor susceptibility and NA protein properties (enzymatic activity, affinity for substrate, and affinity for NA inhibitors) of five highly pathogenic H5N1 influenza viruses with the efficacy of oseltamivir treatment in a mouse model. Viruses of clade 1 and of the more diverse clade 2 were represented. Virus replication in the lungs and brain and production of proinflammatory cytokines were assessed, and virus clones were sequenced to identify minor subpopulations of variants and determine their effect on antiviral treatment. Here we demonstrate that the in vivo efficacy in mice of NA inhibitors against highly pathogenic H5N1 influenza viruses cannot be reliably predicted by susceptibility testing in vitro.

MATERIALS AND METHODS

Viruses, cells, and compounds.

Influenza A (H5N1) viruses were obtained through the World Health Organization network and were plaque purified in MDCK cells (American Type Culture Collection, Manassas, VA) before use. The NA inhibitors oseltamivir carboxylate ([3R,4R,5S]-4-acetamido-5-amino-3-[1-ethylpropoxy]-1-cyclohexene-1-carboxylic acid), the prodrug oseltamivir phosphate (oseltamivir) (ethyl[3R,4R,5S]-4-acetamido-5-amino-3-[1-ethylpropoxy]-1-cyclohexene-1-carboxylate), and zanamivir (4-guanidino-Neu5Ac2en) were provided by Hoffmann-La Roche, Ltd.

Infectivity of H5N1 viruses.

The 50% egg infectious dose (EID₅₀) was determined by injecting 100 μl of half-log serial dilutions of virus into the allantoic cavities of 10-day-old embryonated chicken eggs (24). After 48 h of incubation at 37°C, the hemagglutination activity was assayed. Infectivity in MDCK cells was determined by plaque assay and was expressed as the log₁₀PFU/ml. The dose of virus lethal to 50% of mice (MLD₅₀) was determined by inoculating groups of five mice with serial 10-fold dilutions of virus, followed by a 21-day observation period.

NA activity and kinetics.

A modified fluorometric assay was used to determine the NA activity of the viruses (21). We measured the NA enzyme kinetics at pH 6.5 with 33 mM MES [2-(N-morpholino)ethanesulfonic acid hydrate; Sigma-Aldrich], 4 mM CaCl₂, and the fluorogenic substrate MUNANA [2′-(4-methylumbelliferyl)-α-d-N-acetylneuraminic acid; Sigma-Aldrich] with a final substrate concentration of 0 to 3,333.3 μM. All H5N1 viruses were standardized to an equivalent dose of 10^7.5 PFU/ml. The reaction was conducted at 37°C in a total volume of 50 μl, and the fluorescence of released 4-methylumbelliferone was measured every 92 s for 45 min in a Fluoroskan II instrument (Labsystems) using excitation and emission wavelengths of 355 and 460 nm, respectively. To measure the inhibitory effect of oseltamivir carboxylate or zanamivir on NA activity, H5N1 viruses were preincubated for 30 min at 37°C in the presence of various concentrations of the drugs (0.05 to 500 nM). The kinetic parameters Michaelis constant (K_m), the maximum velocity (V_max) of substrate conversion, and the inhibitory constants (K_is) of the NAs were calculated by fitting the data to the appropriate Michaelis-Menten equations by using nonlinear regression in the commercially available GraphPad Prism 4 software.

Virus susceptibility to NA inhibitors in vitro.

H5N1 viruses were standardized to equivalent NA activity and incubated with NA inhibitors at concentrations of 0.00005 to 5 μM with MUNANA as a substrate. The concentration of NA inhibitor that reduced NA activity by 50% relative to a control mixture with no inhibitor (IC₅₀) was determined by plotting the dose-response curve of inhibition of NA activity as a function of the compound concentration. Values are the means of two to three independent determinations.

Oseltamivir efficacy in vivo.

Female 6-week-old BALB/c mice (weight, 18 to 20 g; Jackson Laboratories, Bar Harbor, ME) were used in all experiments. Mice were anesthetized by inhalation of isoflurane and inoculated intranasally with 50 μl of infectious virus. Treatment with oseltamivir phosphate (oseltamivir) (20 mg/kg/day by twice-daily oral gavage) was initiated 4 h before inoculation with 10 MLD₅₀ of H5N1 virus and continued for 5 days. Control (inoculated, untreated) mice received sterile phosphate-buffered saline (PBS) on the same schedule. Groups of 10 mice were observed for mortality daily for 21 days; animals that showed signs of severe disease and weight loss of >25% were sacrificed. Mean day-to-death values were statistically analyzed. Mice were weighed on days 0, 3, 6, 9, 12, 15, 18, and 21 after inoculation, and the loss or gain of weight was calculated for each mouse as a percentage of its weight before inoculation. Experiments with highly pathogenic H5N1 influenza viruses were conducted in an animal biosafety level 3+ containment facility approved by the U.S. Department of Agriculture. All studies were conducted under applicable laws and guidelines and after approval of the St. Jude Children's Research Hospital Animal Care and Use Committee.

Titration of virus in lungs and brain.

On days 3, 6, and 9 after inoculation with H5N1 influenza virus, three mice in each experimental and control group were sacrificed. The brains and lungs were removed, thoroughly rinsed with sterile PBS, homogenized, and suspended in 1 ml of cold PBS. The suspensions were cleared by centrifugation at 2,000 × g for 10 min, and a volume of 0.1 ml of the supernatants was injected into the allantoic cavities of 10-day-old embryonated chicken eggs to determine the EID₅₀. Virus titers were expressed as mean log₁₀ EID₅₀/ml ± the standard deviation (SD).

Lung cytokine/chemokine analysis.

The concentrations of interleukin-1α (IL-1α), IL-6, monocyte chemotactic protein 1 (MCP-1), gamma interferon (IFN-γ)-inducible protein (IP-10), tumor necrosis factor alpha (TNF-α), and IFN-α were determined in lung homogenates from groups of five experimental and control mice on day 3 after inoculation with H5N1 influenza virus, in PBS-inoculated mice treated with oseltamivir or PBS, and in clean mice. The concentration of cytokines and/or chemokines was determined by specific enzyme-linked immunosorbent assay (R&D Systems).

Virus sequence analysis.

Viral RNA was isolated directly from virus-containing cell culture fluid used for inoculation of BALB/c mice (virus inoculum) and from mouse lung homogenates on days 6 and 9 postinoculation (p.i.) by using the RNeasy Mini Kit (Qiagen). Samples were reverse transcribed and analyzed by PCR using primers specific for the HA (HA1 region) and NA gene segments, as described previously (10). A TOPO TA cloning kit for sequencing (Invitrogen) was used for clonal analysis of the virus population. Briefly, viral RNAs were extracted from plaque samples, and one-step reverse transcriptase PCR was performed. PCR products were purified with a QIAquick PCR purification kit (Qiagen), ligated to the pCR2.1-TOPO vector (Invitrogen), and used for the transformation of TOP10 competent cells (Invitrogen). Plasmid DNA was prepared by using the QIAprep spin miniprep kit. Sequencing was performed by the Hartwell Center for Bioinformatics and Biotechnology at St. Jude Children's Research Hospital. The DNA template was sequenced by using rhodamine or dRhodamine dye terminator cycle-sequencing ready-reaction kits with AmpliTaq DNA polymerase FS (Perkin-Elmer) and synthetic oligonucleotides. Samples were analyzed in a Perkin-Elmer Applied Biosystems DNA sequencer (model 373 or 377). DNA sequences were completed and edited by using the Lasergene sequence analysis software package (DNASTAR).

Statistical analysis.

The susceptibility of influenza A (H5N1) viruses to NA inhibitors, virus titers in mouse lungs and brains, and levels of cytokines/chemokines were compared between treatment and control groups by unpaired t test or analysis of variance (ANOVA). The probability of survival was estimated by the Kaplan-Meier method and compared between the treatment and control groups by using the log-rank test (30). A proportional hazard model was used to determine the death hazard ratio of the treatment and control groups (5). A probability value of 0.05 was prospectively chosen to indicate that the findings were not the result of chance alone. The Spearman rank correlation coefficient was used to compare IC₅₀s to the survival of mice (11).

RESULTS

Infectivity, NA inhibitor susceptibility, and NA enzyme kinetics of H5N1 influenza viruses.

The five selected H5N1 viruses replicated efficiently in embryonated chicken eggs and MDCK cells (data not shown) and had comparable infectivity titers in embryonated chicken eggs (9.5 to 10.3 log₁₀ EID₅₀/ml). Avian isolates grew to somewhat lower titers (8.0 to 9.6 log₁₀ PFU/ml) than human isolates (9.2 to 10.4 log₁₀ PFU/ml) in MDCK cells, but the difference was not statistically significant. All five viruses were lethal to BALB/c mice without prior adaptation. However, one virus, A/chicken/Jogjakarta/BBVET/IX/04 (H5N1), was significantly less pathogenic (mean MLD₅₀, 3,162 PFU) than other viruses tested (mean MLD₅₀, 79 to 1,000 PFU) (P < 0.001).

Susceptibility of the H5N1 viruses to NA inhibitors was determined by a fluorescence-based NA enzyme inhibition assay with MUNANA substrate (Table 1). Human isolate A/Thailand/1(Kan-1)/04 (H5N1) and avian isolate A/duck/Laos/25/06 (H5N1) were most susceptible to oseltamivir carboxylate (mean IC₅₀, 0.3 and 0.1 nM, respectively). Human isolate A/Turkey/65-1242/06 (H5N1) was the least susceptible (mean IC₅₀, 10.8 nM). All five viruses were susceptible to zanamivir, although the IC₅₀s of the human isolates were twice those of the avian isolates (Table 1).

TABLE 1.

NA inhibitor susceptibility and NA enzyme kinetics of H5N1 influenza viruses

H5N1 virus	HA clade	NA enzyme inhibition assay (mean IC50 [nM] ± SE)a		Kinetic NA enzyme parameters
				Mean Km (μM) ± SE (95% CI)b	Vmax ratioc	Ki (nM) ± SEd
		Oseltamivir carboxylate	Zanamivir			Oseltamivir carboxylate	Zanamivir
Human
A/Thailand/1(Kan-1)/04	1	0.3 ± 0.1	1.1 ± 0.1	113.9 ± 28.7 (49.0-178.8)	0.2	0.1 ± 0.1	0.5 ± 0.1
A/Turkey/65-1242/06	2.2	10.8 ± 1.5*	1.4 ± 0.1*	209.6 ± 19.8 (164.7-254.4)	0.1	7.9 ± 0.9*	1.0 ± 0.2*
Avian
A/chicken/Jogjakarta/BBVET/IX/04	2.1	4.9 ± 1.3*	0.7 ± 0.1*	301.1 ± 96.4 (83.1-519.1)	1.0	4.9 ± 0.3*	0.4 ± 0.1
A/whooper swan/Mongolia/244/05	2.2	1.5 ± 0.4*	0.8 ± 0.1*	63.6 ± 8.4 (44.5-82.6)	0.1	0.9 ± 0.2*	0.3 ± 0.1
A/duck/Laos/25/06	2.3	0.1 ± 0.1	0.7 ± 0.1*	193.3 ± 64.0 (48.7-338.0)	0.7	1.1 ± 0.1*	0.5 ± 0.1

^aThe results represent the means from three independent experiments. *, P < 0.01 compared to A/Thailand/1(Kan-1)/04 (H5N1) virus (one-way ANOVA).

^bThe results represent the means from one representative experiment done in triplicate. CI, confidence interval.

^cThat is, the ratio of the NA V_max of the H5N1 virus to that of A/Chicken/Jogjakarta/BBVET/IX/04 virus tested in parallel.

^d*, P < 0.01 compared to A/Thailand/1(Kan-1)/04 virus by one-way ANOVA.

To characterize the kinetic NA enzymatic parameters of the viruses, we determined their Michaelis-Menten constants (K_m), which reflect NA affinity for the MUNANA substrate; their inhibition constants (K_is) for both NA inhibitors (oseltamivir carboxylate and zanamivir); and their relative NA enzymatic activity (V_max), by using whole virus standardized to an equivalent virus dose of 10^7.5 PFU/ml. The K_m values did not differ significantly among the five H5N1 viruses (P = 0.12, Table 1). Interestingly, the NA glycoprotein of A/whooper swan/Mongolia/244/05 (H5N1) virus had a K_m value lower than those of the other strains, although the difference was not statistically significant. The viruses showed similar affinity for zanamivir (K_i, 0.5 to 1.0 nM), whereas they differed markedly in their affinity for oseltamivir carboxylate (K_i, 0.1 to 7.9 nM).

After determining the V_max values for each of the five viruses (Fig. 1), we calculated their V_max ratios in relation to that of A/chicken/Jogjakarta/BBVET/IX/04 NA glycoprotein, which had the highest V_max (Table 1). Two avian NAs showed significantly higher enzymatic activity than the human NAs of the same subtype, although the NA glycoprotein of avian A/whooper swan/Mongolia/244/05 virus had a V_max similar to that of human A/Turkey/65-1242/06 virus. Thus, the five viruses differed both in their NA enzyme kinetics and their susceptibility in vitro to oseltamivir carboxylate.

FIG. 1.

NA enzyme kinetics of five H5N1 influenza viruses. The velocity of substrate conversion by NA (V_o) is shown as a function of substrate concentration and expressed in the conversion of the MUNANA concentration (fluorescence units) per second. Fluorogenic MUNANA substrate was used at a final concentration of 0 to 3,333 μM. The viruses were standardized to an equivalent dose of 10^7.5 PFU/ml. Fluorescence was measured every 92 s for 45 min at 37°C, using excitation and emission wavelengths of 355 and 460 nm, respectively. The curves representing NA enzyme kinetics of the five H5N1 influenza viruses were drawn by approximation of Michaelis-Menten equation by using nonlinear regression.

Effect of oseltamivir on survival of mice lethally challenged with H5N1 influenza virus.

Mice were given 20 mg/kg/day of the prodrug oseltamivir phosphate (oseltamivir) for 5 days, starting 4 h before inoculation; this dose (adjusted for the interspecies difference in esterase activity [31]) is reported to produce a plasma concentration comparable to that of the recommended human oral dose (75 mg twice daily). Inoculated, untreated control mice lost weight rapidly; on day 6 p.i., weight loss ranged from 5.9% (A/duck/Laos/25/06 virus) to 22% (A/chicken/Jogjakarta/BBVET/IX/04 virus) of the initial weight (Table 2). The control animals continued to lose weight and died between days 8 and 13 p.i. (Fig. 2A). Oseltamivir protection against lethal challenge differed among the five H5N1 viruses. The greatest protection (100% survival; maximum weight change, −11.2%) was observed against A/chicken/Jogjakarta/BBVET/IX/04 virus. Oseltamivir protected 80% of animals inoculated with the human A/Thailand/1(Kan-1)/04 virus; the mice began regaining weight by day 10 p.i. and regained most of their weight by day 17 p.i. Oseltamivir treatment was less effective against A/whooper swan/Mongolia/244/05 and A/duck/Laos/25/06 influenza viruses (70 and 60% survival, respectively). Interestingly, weight loss occurred later during infection with A/duck/Laos/25/06 virus and was less pronounced than in other groups (Table 2), and death occurred on days 12 to 14 p.i. (Fig. 2A). The oseltamivir regimen provided partial or complete protection against lethal challenge with all viruses except A/Turkey/65-1242/06 virus (0% survival). We found no correlation between the survival of mice and the susceptibility of the viruses to NA inhibitors in vitro (IC₅₀s) (Spearman rank correlation coefficient ρ = −0.1; P > 0.05).

TABLE 2.

Effect of oseltamivir treatment on survival of BALB/c mice lethally challenged with H5N1 influenza viruses^a

H5N1 virus	Oseltamivir doseb (mg/kg/day)	No. of surviving mice/total no. of mice (%)	Mean day to death ± SEc	Hazard ratio (P)d	Mean wt change (% ± SE)e at:
					6 days p.i.	9 days p.i.	15 days p.i.
Human
A/Thailand/1(Kan-1)/04	20	8/10 (80)**	10.0 ± 0.0	0.09*	-10.7 ± 1.7**	-17.6 ± 1.6**	-2.1 ± 1.6
	0	0/10 (0)	10.4 ± 0.3	1	-15.4 ± 1.3	-25.4 ± 1.2	NA
A/Turkey/65-1242/06	20	0/10 (0)‡	8.0 ± 0.4‡	0.70‡	-24.4 ± 1.5‡	NA	NA
	0	0/10 (0)	8.0 ± 0.3‡	1	-20.8 ± 2.1	NA	NA
Avian
A/chicken/Jogjakarta/BBVET/IX/04	20	10/10 (100)**	>21.0**‡	0.00**	-8.0 ± 2.0**	-11.2 ± 2.3**	1.2 ± 2.5
	0	0/10 (0)	9.6 ± 0.2	1	-22.0 ± 1.7	-25.2 ± 2.5	NA
A/whooper swan/Mongolia/244/05	20	7/10 (70)*	9.7 ± 0.3	0.15*‡	-7.0 ± 2.0**	-18.2 ± 1.7	-10.6 ± 2.3‡
	0	0/10 (0)	9.3 ± 0.3†	1	-15.5 ± 1.3	-22.6f	NA
A/duck/Laos/25/06	20	6/10 (60)**	13.5 ± 0.3*‡	0.18*‡	-0.2 ± 1.5**‡	-5.0 ± 2.0**†	-7.6 ± 1.5
	0	0/9 (0)	10.9 ± 0.6	1	-5.9 ± 1.8†	-20.3 ± 2.3	NA

^a*, P < 0.01; or **, P < 0.001 compared to the PBS-treated virus-inoculated control group (unpaired Student t test). †, P < 0.05; or ‡, P < 0.01 compared to A/Thailand/1(Kan-1)/04 (H5N1) virus by one-way ANOVA. NA, not applicable (all mice in the group died).

^bOseltamivir or PBS was administered orally twice daily to 6-week-old BALB/c mice, starting 4 h before inoculation with 10 MLD₅₀ of H5N1 virus. Survival was observed for 21 days.

^cEstimated by the log-rank test.

^dThe hazard ratio for death (compared to the placebo group) was estimated by the proportional hazards model.

^eLoss or gain of weight was calculated for each mouse as a percentage of its weight on day 0.

^fResults obtained from one mouse.

FIG. 2.

Effect of oseltamivir treatment on survival and lung and brain virus titers of inoculated mice. (A) Survival of BALB/c mice given standard oseltamivir treatment and inoculated with 10 MLD₅₀ of the indicated H5N1 influenza viruses. Mice were given 20 mg/kg/day of oseltamivir by oral gavage twice daily for 5 days, beginning 4 h before virus exposure. Red lines, survival of treated mice; blue lines, survival of control (virus-inoculated, PBS-treated) mice. (B) Effect of standard oseltamivir treatment on lung and brain virus titers of mice inoculated with 10 MLD₅₀ of the indicated H5N1 influenza viruses. Each data point represents the mean virus titer (log₁₀ EID₅₀/ml) ± the SD in the lungs (solid lines) or brains (dashed lines) of three BALB/c mice. Red lines, virus titers of treated mice; blue lines, virus titers of control (virus-inoculated, PBS-treated) mice. The virus titers of treatment and control groups were compared by ANOVA. *, P < 0.05.

Effect of oseltamivir treatment on H5N1 virus replication in the lung and brain.

In the lung, only the replication of A/duck/Laos/25/06 virus was significantly reduced (P < 0.05) in treated mice at day 3 p.i. (Fig. 2B). Lung titers of treated mice lethally challenged with A/Thailand/1(Kan-1)/04 virus were 1.4 to 1.9 log₁₀ EID₅₀/ml lower than those in untreated challenged mice on days 6 and 9 p.i. (P < 0.05). In treated animals inoculated with A/chicken/Jogjakarta/BBVET/IX/04 and A/whooper swan/Mongolia/244/05 viruses, lung titers were significantly lower than those in controls on day 9 p.i. (P < 0.05). Notably, the virus load was not reduced in the lungs of treated mice inoculated with A/Turkey/65-1242/06 virus (Fig. 2B).

Oseltamivir indirectly controlled the spread of A/duck/Laos/25/06 virus to the brain early in the course of infection better than it controlled that of the other viruses (Fig. 2B). However, this virus was highly neurotropic at later stages of infection; an increase in brain virus titers was detected after discontinuation of therapy. The A/Thailand/1(Kan-1)/04 virus was detected at a low titer in the brains of both treated and untreated control animals. Replication of the A/chicken/Jogjakarta/BBVET/IX/04 virus in the brain was significantly inhibited (P < 0.05). Replication of A/whooper swan/Mongolia/244/05 virus in the brains of treated mice was not inhibited on days 3 and 6 p.i. but was significantly inhibited on day 9 p.i. (P < 0.05). Importantly, the brain titers of A/Turkey/65-1242/06 virus, like the lung titers, were not reduced by oseltamivir treatment on days 3, 6, or 9 p.i. (Fig. 2B). Overall, the extent to which antiviral therapy reduced virus replication in the internal organs differed among the five H5N1 viruses tested.

Effect of oseltamivir treatment on lung cytokine expression in inoculated mice.

In humans and in animal models, immune-mediated pathology plays a role in the disease caused by highly pathogenic H5N1 influenza viruses (20, 28). We therefore assessed the indirect effect of oseltamivir treatment on specific cytokine responses (IL-1α, IL-6, IL-10, TNF-α, IFN-α, and MCP-1) in the lungs. The concentration of IL-10 on day 3 p.i. was consistently below the threshold of detection (data not shown). Induction of the other five cytokines varied: TNF-α induction by three viruses was significantly inhibited by antiviral treatment (P < 0.001) (Fig. 3). IL-1α, IFN-α, and MCP-1 induction by two viruses was significantly inhibited by treatment, and IL-6 levels were comparable in the treatment and control groups (Fig. 3). Maximal cytokine inhibition (four of five cytokines studied) was observed in treated mice inoculated with A/duck/Laos/25/06 virus. Treatment inhibited two of five cytokines in mice inoculated with the A/Thailand/1(Kan-1)/04 and A/whooper swan/Mongolia/244/05 viruses. In treated mice inoculated with A/chicken/Jogjakarta/BBVET/IX/04 virus, only TNF-α expression was significantly reduced. Importantly, mice inoculated with A/Turkey/65-1242/06 virus showed marked inflammation and higher levels of IL-1α, IL-6, IFN-α, and MCP-1 than did mice inoculated with any of the other viruses. Overall, we observed no consistent inhibition of any specific cytokine/chemokine in challenged mice treated with oseltamivir.

FIG. 3.

Effect of oseltamivir on the pulmonary expression of cytokines (IL-1α, TNF-α, IL-6, and IFN-α) and the chemokine MCP-1. The cytokines and chemokine were assayed by enzyme-linked immunosorbent assay in lung homogenates obtained from BALB/c mice on day 3 p.i. with the indicated H5N1 influenza viruses. Bars indicate the mean concentration (ρg/ml) ± the SD from five mice. White bars represent mice treated with oseltamivir; solid bars represent virus-inoculated, PBS-treated mice. The dashed line indicates the level of cytokine/chemokine expression in mock-infected mice treated with oseltamivir or PBS. Abbreviations: A/TH/04, A/Thailand/1(Kan-1)/04; A/TK/06, A/Turkey/65-1242/06; A/CK/JOG/04, A/chicken/Jogjakarta/BBVET/IX/04; A/WS/MG/05, A/whooper swan/Mongolia/244/05; A/DK/Laos/06, A/duck/Laos/25/06. *, P < 0.05 compared to the results for the virus-inoculated untreated control group (unpaired Student t test). o, P < 0.01 compared to A/Thailand/1(Kan-1)/04 virus by one-way ANOVA.

Emergence of resistant variants during oseltamivir treatment.

To monitor the emergence of resistant variants during oseltamivir treatment, we determined the IC₅₀s and NA gene sequences of virus obtained from mouse lungs on day 6 or day 9 p.i. Four viruses showed no change in NA inhibitor susceptibility or NA amino acid sequence (Table 3). However, A/Turkey/65-1242/06 virus from one mouse treated with oseltamivir was moderately resistant to oseltamivir carboxylate and highly resistant to zanamivir in NA inhibition assays: the mean IC₅₀ increased by factors of 13 and 1,200, respectively. Sequence analysis of viral RNA obtained directly from this lung sample showed a mutation resulting in an amino acid change at position 119 (E→A) of the NA gene (Table 3). No NA amino acid changes were detected in virus from the lungs of the other two treated mice inoculated with A/Turkey/65-1242/06 virus.

TABLE 3.

Emergence of drug-resistant variants during oseltamivir treatment of BALB/c mice infected with H5N1 influenza viruses

H5N1 virus	Before oseltamivir treatment (NA mutations)		After oseltamivir treatment
			NA mutations	NA enzyme inhibition assay (mean IC50 [nM] ± SE)d
	Dominant virus populationa	Individual clone(s) (no. clones/total no. sequenced)b	Dominant virus population (no. of infected mice/total no. of mice)c	Oseltamivir carboxylate	Zanamivir
Human
A/Thailand/1(Kan-1)/04	None	None	None (3/3)	0.2 ± 0.1	1.4 ± 0.1
A/Turkey/65-1242/06	None	E119A (2/21), I117V (6/21)	E119A (1/3)	136.9 ± 8.3	1,659.0 ± 441.9
Avian
A/chicken/Jogjakarta/BBVET/IX/04	None	None	None (3/3)	3.7 ± 1.5	0.8 ± 0.1
A/whooper swan/Mongolia/244/05	None	None	None (3/3)	1.2 ± 0.4	0.7 ± 0.1
A/duck/Laos/25/06	None	I222M (2/22)	None (3/3)	0.1 ± 0.1	0.6 ± 0.1

^aRNA was isolated directly from virus-containing allantoic fluid used as an inoculum.

^bTOPO TA cloning was performed by using PCR products obtained by amplification of the virus inoculum. Individual virus clones were analyzed by sequencing. Amino acid numbering is based on N2 NA (19).

^cRNA was isolated directly from lung homogenates of three mice on day 9 p.i.

^dValues are means from three independent experiments.

We sequenced virus samples from the brain of the mouse showing drug-resistant A/Turkey/65-1242/06 virus and confirmed the existence of the E119A NA mutation in the dominant virus population (data not shown). The NA and HA genes of the dominant populations of all five viruses had been sequenced before the experiments, but minor subpopulations of drug-resistant clones would not have been detected. Therefore, we sequenced the NA genes of 20 to 25 individual clones derived from each virus inoculum. We found an E119A NA amino acid change in 2 of 21 clones and an I117V change in 6 of 21 clones of the A/Turkey/65-1242/06 virus (Table 3). A minor population of clones (2/22) carrying an I222M NA mutation was detected in the A/duck/Laos/25/05 (H5N1) virus inoculum (Table 3). In the three other viruses, we observed no NA amino acid substitutions reported to be associated with resistance or decreased susceptibility to NA inhibitors.

DISCUSSION

We found that the in vitro oseltamivir susceptibility of five highly pathogenic H5N1 influenza viruses did not consistently predict the effectiveness of drug treatment in a mouse model. The five viruses differed in their susceptibility to oseltamivir in vitro: two viruses were highly susceptible to the drug (mean IC₅₀, 0.1 to 0.3 nM), two viruses had mean IC₅₀s comparable to those reported for susceptible H5N1 virus strains (12, 18) and human H1N1 (32) influenza viruses, but A/Turkey/65-1242/06 virus demonstrated slightly decreased susceptibility. Importantly, there was only a twofold difference in IC₅₀s between A/chicken/Jogjakarta/BBVET/IX/04 virus (100% of treated mice survived) and A/Turkey/65-1242/06 virus (no treated mice survived). Moreover, we did not observe a correlation between NA enzymatic activity and outcome of drug treatment in mice.

An oseltamivir regimen that approximated the recommended human regimen (31) afforded mice various levels of protection against lethal challenge with each of the five H5N1 viruses. Because the virulence of an H5N1 virus strain is a factor in the efficacy of standard therapy in experimental animal models (8, 36), we attribute the high efficacy of oseltamivir against the A/chicken/Jogjakarta/BBVET/IX/04 virus to the observed low virulence of the virus. At least three factors may explain the low efficacy of oseltamivir against A/Turkey/65-1242/06 virus. First, this virus had the lowest susceptibility to oseltamivir carboxylate in vitro. Although the plasma concentration of oseltamivir carboxylate achieved by this regimen is likely to have been much higher than the IC₅₀ against any of the viruses, it did not provide protection. Second, this virus induced high pulmonary expression of proinflammatory chemokines and cytokines; four of the five cytokines studied (IL-1α, IL-6, IFN-α, and MCP-1) were expressed at the highest levels observed. This finding may be partially attributed to the complete activation of macrophages and monocytes, which are the major source of MCP-1, IL-6, IL-1α, TNF-α, and IFN-α during inflammation (2, 16). Hypercytokinemia appears to contribute to the pathogenicity of H5N1 infection, although the mechanisms remain unclear (20, 35). The third potential factor is the minor subpopulation of drug-resistant variants in the virus inoculum. These variants were not detected by sequencing of the dominant virus population but only by subsequent analysis of individual virus clones.

Our previous work has shown that resistant H5N1 virus variants can become dominant under drug selection pressure and, assuming adequate fitness (37), can contribute significantly to therapeutic failure. NA mutation at residue I117V was reported previously in an H5N1 isolate with reduced susceptibility to oseltamivir carboxylate (12); this mutation may affect interactions with residue R118, one of the three arginines in the NA catalytic site that bind the carboxylate of the substrate sialic acid (25). An E119A substitution is associated with zanamivir resistance in viruses of the N1 NA subtype; however, our experiments showed that the E119A mutation can be maintained in the NA of clade 2 virus during oseltamivir therapy. Therefore, the carboxylate groups of the two NA inhibitors appear to interact similarly with the conserved framework residues of viral NA (25). The likelihood that the preexisting minor population of E119A may have contributed to the lack of standard oseltamivir treatment against of A/Turkey/65-1242/06 virus in this experiment is, however, low, since this mutation became more prevalent only in one of three mice, while none of the mice were protected from death by the 20-mg/kg/day dose of oseltamivir.

We explored the effectiveness of higher oseltamivir dosages and found that 50 mg/kg/day for 8 days resulted in 80% survival of BALB/c mice lethally infected with A/Turkey/65-1242/06 virus (results not shown). The mechanistic basis for the impact of pathogenicity on dose requirement is unknown, but from the data generated here, it is tempting to speculate that hypercytokinemia may be induced by certain H5N1 influenza strains leading to increased overall pathogenicity due to immune mediated tissue damage on top of direct virus induced cell death. Higher doses of antivirals may be needed early to more effectively block virus replication and prevent triggering dysregulation of the immune response. Once triggered, the immune-mediated tissue damage is possibly not very sensitive to the presence of antiviral agents. Thus, the NA inhibitor oseltamivir alone can promote survival in mice, although it is unlikely to significantly reduce the effect of host-mediated determinants of pathogenicity.

The oseltamivir regimen that was used provided moderate protection (60 to 80% survival) against three H5N1 influenza viruses. Although A/duck/Laos/25/06 virus was highly susceptible in vitro (mean IC₅₀, 0.1 nM), only 60% of mice survived lethal virus challenge. We did not observe high levels of cytokines in these mice, but the virus was detected in their brains after cessation of therapy, and brain virus titers were increasing on days 6 and 9 p.i. in untreated control animals, demonstrating that the virus is neurotropic. The primary antiviral mechanism of oseltamivir is direct inhibition of viral NA enzyme activity in the extracellular fluid of the respiratory tract (29). Therefore, the neurotropic features of A/duck/Laos/25/06 may have been the deciding factor in the outcome of oseltamivir treatment. Little is known about the bioavailability of oseltamivir carboxylate in the brain or digestive tract. The latter factor may be even more crucial in human H5N1 infections, which are characterized by extrapulmonary viral replication, hypercytokinemia, and multiorgan failure (35). A study in the rat model suggested that oseltamivir carboxylate has limited ability to cross the blood-brain barrier (27). It is also unknown how the absorption and pharmacokinetics of oseltamivir are altered by diarrhea and multiorgan failure, which are common features of H5N1 infection.

The human isolate A/Thailand/1(Kan-1)/04 was highly susceptible to oseltamivir carboxylate in vitro, but only 80% of treated mice survived. The observed high levels of IL-6, IFN-α, and MCP-1, which are not controlled by antiviral treatment, may have contributed to incomplete protection. The moderate level of protection against avian isolate A/whooper swan/Mongolia/244/05 (70% survival) was associated with poor control of virus replication in the lungs and elevated expression of IL-1α and TNF-α, again suggesting that inflammation may modulate the efficacy of NA inhibitor therapy.

Like all results obtained in experimental animal models, our findings cannot necessarily be generalized to humans. The H5N1 viruses are likely to have growth kinetics, cytokine induction, and tropism characteristics that differ between humans and mouse models. Another factor that may affect NA inhibitor susceptibility in humans is potential mistmatch between host receptors and the HA receptor binding properties of the virus. Binding of the HA proteins of avian viruses (which have specificity for 2,3-linked sialyl receptors) to receptors in the human respiratory tract (predominantly 2,6-linked sialyl receptors) may be weak, reducing the importance of NA functions. Moreover, the receptor and binding specificities for systemic organs and tissues may also be a factor in avian H5N1 influenza viruses that replicate beyond the respiratory tract.

Therapy with oseltamivir, an anti-influenza drug that specifically targets the viral NA protein, has effectively controlled H5N1 infection in animal models (8, 36). Our findings that are limited to prophylactic administration of oseltamivir in mice suggest that the optimal drug regimen for infection with highly pathogenic H5N1 influenza viruses depends on multiple factors. Therefore, antiviral treatment can be optimized only through improved understanding of the biology of the H5N1 virus and the factors that affect its pathogenicity. Studies are needed to evaluate new modalities of oseltamivir treatment, as well as the potential benefit of adjunctive treatments such as immunomodulatory therapy, IFN, or ribavirin. We previously demonstrated that oseltamivir is more effective against H5N1 viruses in mice when combined with amantadine (13). Oseltamivir and ribavirin showed principally additive efficacy against both clade 1 and clade 2 H5N1 influenza viruses, although marginal synergy or marginal antagonism was occasionally observed at certain drug concentrations (14). Combination therapy consisting of specific anti-influenza drugs and inhibitors of inflammation, such as celecoxib and mesalazine, was recently reported to be a promising approach to control of H5N1 infection in mice (39). These observations highlight the need for additional antiviral agents that can replace or be used in combination with oseltamivir for the appropriate management of H5N1 influenza virus infections.