Pleiotropic Effects of Influenza H1, H3, and B Baloxavir-Resistant Substitutions on Replication, Sensitivity to Baloxavir, and Interferon Expression

ABSTRACT

Baloxavir is an anti-influenza endonuclease inhibitor that targets the polymerase acidic (PA) protein of influenza A and B viruses. Our knowledge regarding the pleiotropic effects of baloxavir resistance-associated substitutions is limited. We generated recombinant A/California/04/09 (H1N1)-, A/Hong Kong/218849/2006 (H3N2)-, and B/Victoria/504/2000-like viruses that contained PA substitutions identified in baloxavir clinical trials and surveillance that could potentially be associated with baloxavir resistance. We characterized their susceptibility to baloxavir, impact on polymerase activity, viral growth, and ability to induce interferon (IFN) and IFN-stimulated genes expression in vitro. Four PA substitutions, H1N1 I38L/T, E199D, and B G199R, significantly reduced the sensitivity of the recombinant viruses to baloxavir (14.1-fold). We confirmed our findings by using the luciferase-based ribonucleoprotein minigenome assay and by using virus yield reduction assay in Calu-3 and normal human bronchial epithelial (NHBE) cells. We observed that I38L and E199D resulted in decreased viral replication of the H1N1 wild-type virus (1.4-fold) but the H1N1 I38T and B G199R substitutions did not significantly alter replication capacity in Calu-3 cells. In addition, H1N1 variants with PA I38L/T and E199D induced significantly higher levels of IFNB1 gene expression compared to the wild-type virus (4.2-fold). In contrast, the B variant, G199R, triggered the lowest levels of IFN genes in Calu-3 cells (1.6-fold). Because baloxavir is a novel anti-influenza therapeutic agent, identifying and characterizing substitutions associated with reduced sensitivity to baloxavir, as well as the impact of these substitutions on viral fitness, is paramount to the strategic implementation of this novel countermeasure.

INTRODUCTION

Influenza viruses can cause an acute, sometimes fatal, respiratory illness. Seasonal influenza outbreaks occur each year which result in upwards of 5 million cases of severe disease worldwide (1). The influenza A virus subtypes, H1N1 (H1) and H3N2 (H3), are the main cause of infection during outbreaks; however, influenza B virus (B) strains occasionally dominate during influenza seasons (2, 3). There have also been several influenza pandemics throughout history, with the most recent occurring in 2009 when a novel H1 emerged in late spring and quickly spread throughout the world, infecting millions and causing hundreds of thousands of deaths (4).

Antiviral drugs play an important role in the rapid response against influenza virus infection, especially when vaccination is not effective due to a poor match between the vaccine and the circulating influenza strain or poor immune responses among elderly patients. Until recently, influenza virus treatment has been restricted to a single class of therapeutics, the neuraminidase inhibitors (NAIs; oseltamivir, zanamivir, laninamivir, and peramivir) (5, 6). Unfortunately, the development of resistance through mutation(s) is a major hurdle in combatting influenza. M2 ion-channel inhibitors (amantadine and rimantadine) are no longer prescribed due to the near ubiquitous spread of amantadine resistance (7, 8). Moreover, the antiviral potency of the NAIs can be relatively weak, and an epidemic of NAI-resistant influenza strains was experienced as recently as 2009, when oseltamivir-resistant variants became widespread (8–10). Therefore, the availability of additional effective antiviral agents with novel mechanisms of action is needed for prevention and treatment of influenza virus infection, particularly in the event of widespread resistance to other classes of antivirals.

The influenza virus RNase polymerase (RNP) complex, which is comprised of three proteins, polymerase basic protein 1 (PB1), PB2, and polymerase acidic protein (PA), is an attractive target for influenza virus countermeasure development. The PB1 subunit is the viral RNA polymerase, the PB2 is responsible for the cap-snatching mechanism to initiate replication, and the PA protein contains an endonuclease function to cleave 5′ caps from host RNA (11). In 2018, baloxavir (prodrug: baloxavir marboxil, active form: baloxavir acid), an inhibitor of the PA cap-dependent endonuclease activity, was approved for use in the United States against influenza virus infection, was shown to be safe and effective at reducing duration of illness in influenza virus-infected individuals, and also had a significant impact on virus shedding (12–14). Additionally, several other compounds targeting influenza RNP have been investigated as potential influenza virus antivirals, including the PB1 inhibitor favipiravir, the PB2 inhibitor pimodivir, and the PA inhibitors RO-7 and AV5116 (15–18).

Baloxavir represents an important addition to the arsenal of interventions against influenza due to the novel mechanism of action. However, during clinical trials, resistant viruses emerged in baloxavir-treated subjects at a frequency ranging from ∼3 to 11% in adults to >23% in children (14, 19). The mechanism of influenza virus resistance to baloxavir is conferred mainly by PA substitutions at the conserved isoleucine (I) residue at position 38 (i.e., I38F/L/M/S/T) (14, 20–22). H1 PA E23K/R, A36V, and E119D, H3 PA E23K, A36V, A37T, and E119D/G, and B PA E23K, A36V, and E119G substitutions have also been shown to decrease influenza virus susceptibility to baloxavir (14, 21, 23). Interestingly, the impact of substitutions at PA residue 38 on baloxavir sensitivity and viral fitness differs between influenza virus subtypes. H1, H3, and B viruses with I38F/M/T exhibited 3- to 116-fold reductions in baloxavir susceptibility in vitro, with the greatest reduction seen in viruses containing I38T (21, 24–27). Whereas the PA I38F/M/T substitutions demonstrated negligible effects on polymerase activity of H1 and B strains, PA I38F/T significantly decreased polymerase activity of H3 viruses (26, 28). Further, PA I38M/S/T attenuated replication kinetics of H1 viruses both in vitro and in a ferret animal model, but PA I38L was reported to have a minimal effect on H1 viral replication in Madin-Darby canine kidney (MDCK) cells (20, 21, 25, 26, 29). However, H1, H3, and B viruses containing PA I38M/T were efficiently transmitted in ferrets (26). In addition, little to no reversion of PA I38M/F/T substitutions has been observed in H1, H3, and B viruses both in cell culture and in animal models (25, 26). Overall, our understanding of the impact of baloxavir resistance-associated mutations on influenza virus fitness is limited. The identification and close monitoring of novel viral pathways to resistance are urgently needed for the proper implementation of baloxavir in clinical settings.

During the FDA’s virology review of the clinical, virologic, and resistance data from clinical trials T0821, T0831, T0822, and CP40563, submitted to support the Original New Drug Application for baloxavir, several substitutions in the influenza RNP complex were identified (i.e., H1 PA A37S, I38L/T, E199D/K, P267S, K328E, T363I, A476S, and E677D; H3 PB2 R209K, PA F35L, V62I, T162I, E199D/K, Y321H, V432I, K492R, M595I, A618S, and G684R; and B PA I38V, G199R, K298R, T304A, V645A, and M682L). These amino acid substitutions were (i) located at amino acid positions associated with reduced susceptibility to baloxavir, (ii) treatment-emergent substitutions associated with virus rebound, or (iii) polymorphisms-associated with reduced susceptibility of baseline isolates. Because the observed substitutions may confer resistance to baloxavir, it was recommended that these substitutions be further evaluated for their impact on susceptibility of influenza virus to baloxavir (30). In this study, we further assessed the impact of the novel PB2 and PA substitutions on baloxavir acid (referred to as baloxavir throughout the text) antiviral activity by generating H1, H3, and B RNP complexes and recombinant viruses carrying the observed substitutions. We determined whether influenza variants carrying specific polymerase substitutions can be inhibited by favipiravir and/or the NAIs oseltamivir and zanamivir. We also examined whether the novel substitutions could affect influenza replicative ability and induction of interferons (IFNs) and IFN-stimulated genes (ISGs) in cultures of primary human respiratory epithelial cells.

RESULTS

Effect of polymerase substitutions on viral growth and polymerase activity.

We employed the eight plasmid reverse genetics system (31) to generate recombinant influenza viruses carrying specific PB2 and PA substitutions (Table 1). The H1, H3, and B mutants were rescued in A/California/04/09 (H1N1)-, A/Hong Kong/218849/2006 (H3N2)-, and B/Victoria/504/2000-virus backgrounds, respectively. To verify the presence of the desired substitutions, we sequenced the entire PB2 and PA genes of the virus variants (data not shown). Sequence analysis demonstrated that the recombinant H3 viruses with PB2 R209K, PA V162I, and PA E199D reverted to the H3 wild-type (WT) virus. All other variants contained the inserted substitutions, and no additional changes were detected.

TABLE 1.

Recombinant H1, H3, and B influenza viruses carrying substitutions selected for evaluation in the study and their growth characteristics

Type/subtype	Substitution	Rationale for evaluation	Virus background used in the study	Virus abbreviation	Virus yield (log10PFU/mL)a	Plaque size (mm)b
A/H1N1	WT	NA	A/California/04/2009	H1-WT	5.7 ± 0.2	0.2 ± 0.1*
	PA-I38T	Previously identified resistance-associated substitution (21, 23, 28, 32)		H1-I38T	7.1 ± 0.3°	0.6 ± 0.2*
	PA-A37S	Substitutions identified in NCBI database at amino acid positions where other resistance-associated substitutions have been identified		H1-A37S	7.4 ± 0.3°	0.5 ± 0.1*
	PA-I38L			H1-I38L	7.5 ± 0.5°	1.1 ± 0.4°
	PA-E199D			H1-E199D	6.9 ± 0.5°	0.7 ± 0.3*
	PA-E199K			H1-E199K	6.8 ± 0.3°	0.9 ± 0.4°
	PA-P267S	PA polymorphisms associated with elevated (≥90 percentile) baseline EC50 values of virus isolated from clinical specimens within type/subtype subsets within clinical trials T0821, T0831, T0822, and CP40563		H1-P267S	7.9 ± 0.7°	0.7 ± 0.3*
	PA-A476S			H1-A476S	7.9 ± 0.5°	1.2 ± 0.3°
	PA-K328E			H1-K328E	7.7 ± 0.1°	1.7 ± 0.3°
	PA-E677D			H1-E677D	7.8 ± 0.5°	1.2 ± 0.3°
	PA-T363I	Treatment-emergent substitution associated with virologic rebound in clinical trial CP40563		H1-T363I	7.2 ± 0.1°	1.2 ± 0.3°
A/H3N2	WT	NA	A/Hong Kong/218849/2006	H3-WT	6.0 ± 0.5	0.7 ± 0.2
	PA-F35L	PA polymorphisms associated with elevated (≥90 percentile) baseline EC50 values of virus isolated from clinical specimens within type/subtype subsets within clinical trials T0821, T0831, and T0822		H3-F35L	5.8 ± 0.2	0.8 ± 0.1
	PA-T162I			H3-T162I	NA	NA
	PA-Y321H			H3-Y321H	6.5 ± 0.4	1.0 ± 0.1°
	PA-V432I			H3-V432I	6.3 ± 0.4	0.5 ± 0.1*
	PA-M595I			H3-M595I	6.2 ± 0.3	0.8 ± 0.2
	PA-A618S			H3-A618S	6.2 ± 0.4	0.5 ± 0.2
	PA-G684R			H3-G684R	6.7 ± 0.2	0.9 ± 0.2
	PA-E199D	Substitutions identified in NCBI database at amino acid positions where other resistance-associated substitutions have been identified		H3-E199D	NA	NA
	PA-E199K			H3-E199K	6.4 ± 0.1	0.7 ± 0.1
	PA-V62I	Baseline substitutions associated with reduced virologic response as measured by the change in viral RNA from baseline and relative to consensus genotypes in clinical trials T0821 and T0831		H3-V62I	5.8 ± 0.6	0.6 ± 0.1
	PA-K492R			H3-K492R	6.3 ± 0.4	0.6 ± 0.2
	PB2-R209K	Treatment-emergent substitution associated with virologic rebound in clinical trial T0831		H3-R209K	NA	NA
B	WT	NA	B/Victoria/504/2000	B-WT	8.0 ± 0.5	1.6 ± 0.8
	PA-I38V	Substitution identified in NCBI database at amino acid positions where other resistance-associated substitutions have been identified		B-I38V	8.2 ± 0.3	1.5 ± 0.3
	PA-G199R	PA polymorphisms associated with elevated (≥90 percentile) baseline EC50 values of virus isolated from clinical specimens within type/subtype subsets within clinical trials T0821, T0831, and T0822		B-G199R	7.1 ± 0.4	0.9 ± 0.2°
	PA-K298R			B-K298R	7.3 ± 0.4	1.0 ± 0.2°
	PA-T304A			B-T304A	7.2 ± 0.3	1.3 ± 0.2
	PA-V645A			B-V645A	6.7 ± 0.1°	0.9 ± 0.2°
	PA-M682L	Baseline substitution associated with reduced virologic response as measured by the change in viral RNA from baseline and relative to consensus genotypes in clinical trials T0821 and T0831		B-M682L	7.6 ± 0.3	1.1 ± 0.2*

^aValues represent the mean log₁₀ PFU/mL ± standard deviations from three independent determinations. The number of PFU in MDCK-SIAT1 cells was measured by plaque assay after incubation at 37°C (H1 and H3) or 33°C (B) for 2 to 3 days with 10-fold serial dilutions of virus (71).

^bValues represent the mean plaque diameter (mm ± standard deviations) of 10 plaques as measured by use of a Finescale comparator. Fold changes compared to the respective WT virus are shown in parentheses. *, P < 0.05; °, P < 0.01 compared to the value for the respective WT virus by one-way analysis of variance (ANOVA). WT, wild-type; NA, not applicable.

We first examined the growth of the recombinant influenza variants in MDCK-SIAT1 cells (Table 1). All H1 variants grew to significantly higher titers and formed significantly larger plaques compared to the H1-WT virus (P < 0.05). In contrast, the recombinant H3 viruses had viral titers and plaque sizes similar to those of the respective H3-WT, except the H3-Y321H and H3-V432I variants, which formed significantly larger plaques than H3-WT (P < 0.05). We observed that B-V645A grew to significantly lower titers than B-WT and B variants, with G199R, K298R, V645A, and M682L substitutions forming significantly larger plaques (P < 0.05, Table 1). To evaluate the replicative ability of the rescued variants, we assayed their virus yield in comparison to that of the respective WT virus after multiple replication cycles in Calu-3 cells. As shown in Fig. 1, H1-A37S, H1-I38L, H1-E199D, H1-E199K, H1-P267S, H1-A476S, H1-E677D, H3-V432I, H3-M595I, H3-A618S, and B-G199R grew to significantly lower viral titers (1.2-fold) at 72 h postinfection (hpi) than the respective WT virus (P < 0.05). None of the H1, H3, or B substitutions resulted in significantly higher viral titers compared to the respective WT virus at 72 hpi.

FIG 1.

Replication of the recombinant (A and B) H1, (C and D) H3, and (E and F) B viruses in Calu-3 cells. Calu-3 cells were infected with virus at a multiplicity of infection (MOI) of 0.01 for 1 h before being washed and overlaid with fresh media and incubated at 37°C (H1 and H3) or 33°C (B virus). Every 24 hpi, for 72 h, aliquots of supernatant were collected and stored at −80°C until use. Samples were titrated by plaque assay using MDCK cells to determine viral yield. *, P < 0.05; °, P < 0.01 compared to the values for the respective WT virus by one-way ANOVA.

We next determined the effect of the PB2 and PA substitutions on viral RNP transcription activity by using a dual luciferase-based minigenome assay (Fig. 2). We observed that H1 PA I38L and E199D, H3 PA F35L, V62I, and E199D, and B PA M682L significantly increased polymerase activities of H1 (1.5-fold), H3 (1.3-fold), and B (1.4-fold) RNP complexes (P < 0.05). In contrast, a significant decrease in polymerase activity was observed in RNP complexes carrying H1 PA A37S, E199K, or K328E (1.7-fold), H3 PB2 R209K, PA T162I, or E199K (1.8-fold), and B PA G199R substitutions (2.2-fold, P < 0.05) (Fig. 2).

FIG 2.

Polymerase activity of RNP complexes of the recombinant (A) H1, (B) H3, and (C) B viruses. 293T cells were transfected with plasmids encoding the PB1, PB2, NP, PA (WT or mutated), and the reporter luciferases. The cells were incubated at 37°C (H1 and H3) or 33°C (B virus) for 24 h before being lysed and assayed using a dual luciferase-based reporter system. The data are expressed as percent activity compared to WT RNP and are from three independent experiments performed on different days. *, P < 0.05; °, P < 0.01 compared to the values for the respective WT virus by one-way ANOVA.

Effect of polymerase substitutions on susceptibility to baloxavir.

We next evaluated the susceptibility of the WT and mutant viruses to baloxavir by plaque reduction assay in MDCK-SIAT1 cells (Table 2). The H1-I38L, H1-I38T, and H1-E199D viruses showed significantly increased 50% effective concentration (EC₅₀) values of 11.8 ± 2.9 nM, 40.9 ± 6.5 nM, and 2.9 ± 0.2 nM (P < 0.05), respectively, compared to that of H1-WT (1.0 ± 0.7 nM). In contrast, all H3 viruses remained sensitive to baloxavir (EC₅₀ ≈ 0.3 ± 0.1 nM). For the B viruses, the B-G199R variant exhibited an increased EC₅₀ of 38.5 ± 4.2 nM (P < 0.01) compared to that of B-WT (18.9 ± 4.6 nM).

TABLE 2.

Sensitivity of recombinant H1, H3, and B viruses to baloxavir

Virus	Baloxavir, nMa
	Plaque reduction assay (EC50)	Minigenome assay (IC50) b
H1-WT	1.0 ± 0.7	1.3 ± 0.3
H1-A37S	1.1 ± 0.3 (1.1)	1.3 ± 0.3 (1.0)
H1-I38L	11.8 ± 2.9 (11.5)°	8.4 ± 1.0 (6.6)°
H1-I38T	40.9 ± 6.5 (40.1)°	168.4 ± 33.7 (131.5)°
H1-E199D	2.9 ± 0.2 (2.9)*	3.0 ± 0.4 (2.3)°
H1-E199K	0.8 ± 0.1 (0.8)	1.9 ± 0.4 (1.5)
H1-P267S	0.9 ± 0.1 (0.9)	1.3 ± 0.3 (1.0)
H1-K328E	0.7 ± 0.4 (0.6)	0.9 ± 0.2 (0.7)
H1-T363I	1.4 ± 0.2 (1.4)	1.2 ± 0.5 (1.0)
H1-A476S	1.2 ± 0.3 (1.2)	1.3 ± 0.3 (1.1)
H1-E677D	1.2 ± 0.2 (1.2)	1.1 ± 0.2 (0.9)
H3-WT	0.3 ± 0.1	1.3 ± 0.1
H3-R209K	NA	1.8 ± 0.3 (1.4)
H3-F35L	0.3 ± 0.1 (1.0)	1.1 ± 0.2 (0.8)
H3-V62I	0.2 ± 0.1 (0.8)	1.4 ± 0.3 (1.1)
H3-T162I	NA	1.0 ± 0.2 (0.7)
H3-E199D	NA	2.0 ± 0.1 (1.6)°
H3-E199K	0.3 ± 0.1 (0.9)	1.3 ± 0.3 (1.0)
H3-Y321H	0.3 ± 0.1 (1.0)	1.8 ± 0.4 (1.3)
H3-V432I	0.2 ± 0.1 (0.6)	1.5 ± 0.3 (1.2)
H3-K492R	0.2 ± 0.1 (0.8)	1.4 ± 0.3 (1.0)
H3-M595I	0.2 ± 0.1 (0.8)	1.4 ± 0.3 (1.1)
H3-A618S	0.3 ± 0.1 (1.0)	1.0 ± 0.2 (0.8)
H3-G684R	0.3 ± 0.1 (1.1)	1.8 ± 0.3 (1.4)
B-WT	18.9 ± 4.6	16.0 ± 4.8
B-I38V	19.0 ± 3.4 (1.0)	18.0 ± 4.6 (1.1)
B-G199R	38.5 ± 4.2 (2.0)°	42.5 ± 12.2 (2.7)°
B-K298R	22.9 ± 4.6 (1.2)	18.4 ± 3.7 (1.2)
B-T304A	16.5 ± 3.5 (0.9)	12.8 ± 2.6 (0.8)
B-V645A	15.4 ± 3.9 (0.8)	17.5 ± 4.0 (1.1)
B-M682L	15.3 ± 4.0 (0.8)	18.4 ± 8.0 (1.2)

^aDecreased sensitivity to baloxavir is highlighted in bold. Fold changes compared to the respective WT virus are shown in parentheses. *, P < 0.05, °, P < 0.01 compared to the respective WT virus by one-way ANOVA. NA; not applicable.

^bThe concentration of baloxavir that caused a 50% reduction in RNP activity in minigenome assay compared to that in control cells without drug was defined as the IC₅₀.

We next used the luciferase-based minigenome assay, which eliminates variance caused by different replication kinetics of tested recombinant viruses, to confirm our data on baloxavir sensitivity measured by plaque reduction assay (Table 2). We observed that H1 RNP complexes with the PA I38L, I38T, or E199D substitutions exhibited significantly decreased sensitivity to baloxavir antiviral activity (2.3- to 131.5-fold decrease in 50% inhibitory concentration [IC₅₀] value, P < 0.01). The H3 RNP containing PA E199D exhibited a slightly increased IC₅₀ value (2.0 ± 0.1 nM) compared to that of the H3-WT RNP complex (1.3 ± 0.1 nM, P < 0.01). Furthermore, B PA G199R significantly decreased baloxavir sensitivity of the B RNP complex by 2.7-fold (P < 0.05, Table 2). Overall, the EC₅₀ values determined by plaque reduction assay in MDCK-SIAT1 cells were similar to the IC₅₀ values determined by the luciferase-based minigenome assay.

Since four recombinant viruses, H1-I38L, H1-I38T, H1-E199D, and B-G199R, demonstrated decreased sensitivity to baloxavir in plaque reduction and minigenome assays, we further assessed their impact on influenza virus sensitivity to baloxavir by virus yield reduction assay in Calu-3 and normal human bronchial epithelial (NHBE) cells (Fig. 3). Whereas two H1 variants, H1-I38L and H1-I38T, exhibited significantly increased resistance to baloxavir in Calu-3 and NHBE cells (13.7- and 115.2-fold, respectively, P < 0.05), the H1-E199D variant remained susceptible to baloxavir in Calu-3 cells (EC₅₀ = 0.6 ± 0.2 nM) (Fig. 3A). The B-G199R variant showed significant resistance to baloxavir in Calu-3 and NHBE cells (4.6-fold, P < 0.05, Fig. 3C and D). Taken together, the EC₅₀ values determined by virus yield reduction assay in NHBE cells (Fig. 3B and D) were in a good correlation with our findings observed in plaque reduction and minigenome assays (Table 2) and indicated that H1 PA I38L, I38T, E199D, and B PA G199R substitutions conferred resistance to baloxavir antiviral activity.

FIG 3.

Antiviral effect of baloxavir in (A and C) Calu-3 and (B and D) NHBE cells against recombinant H1 and B viruses as measured by virus yield reduction assay. Calu-3 and NHBE cells were pretreated with baloxavir for 2 h before being infected with influenza virus at an MOI of 0.1 PFU/cell. After 1 h, the cells were washed, overlaid with fresh media containing baloxavir, and then incubated at 37°C (H1 and H3) or 33°C (B virus). After 2 days, the supernatant was collected and stored at −80°C until use. Viral loads in samples were determined by plaque assay using MCDK cells. The concentration of virus that caused a 50% decrease in viral load was defined as the EC₅₀. Depicted results are the averages of two independent experiments. *, P < 0.05; °, P < 0.01 compared to the values for the respective WT virus by one-way ANOVA. EC₅₀, 50% effective concentration.

Effect of polymerase substitutions on IFN and ISG expression.

We next compared the ability of our recombinant H1, H3, and B viruses containing specific PA substitutions to induce IFN (i.e., IFNB1, IFNL1, and IFNL2/3) and ISG expression (i.e., IFIT1 and IFIT3) in Calu-3 cells (Fig. 4 to 6). Our results showed that the baloxavir-resistant variant, H1-I38T, induced significantly higher levels of IFNB1 and IFNL2/3 expression compared to H1-WT (5.0-fold, P < 0.01). Whereas the majority of the H1 variants triggered significantly decreased levels of IFNL2/3 and/or IFIT3 (4.5-fold, P < 0.01), the H1-E199D variant elicited increased levels of the IFNB1, IFNL2/3, IFIT1, and IFIT3 genes in Calu-3 cells (4.8-fold, P < 0.01, Fig. 4). We also observed the increased expression of the IFN genes after infection with the H3 recombinants carrying PA F35L, V432I, and M595I substitutions (3.2-fold, P < 0.01, Fig. 5). In contrast, H3-E199K induced significantly decreased expression of all IFN genes besides IFIT1 (2.6-fold, P < 0.01). The induction of all IFN and ISG expression by infection with the H3-A618S mutant was 3.2-fold stronger than that by infection with the respective H3-WT virus (P < 0.01, Fig. 5). In addition, infection of Calu-3 cells with all recombinant B viruses except B-I38V induced significantly lower expression levels of all studied IFN genes (1.7-fold, P < 0.05, Fig. 6).

FIG 4.

H1-induced IFN and ISG expression in Calu-3 cells. Cells were infected with the recombinant H1 viruses (MOI of 7), and IFN and ISG expression levels were analyzed 24 hpi. Expression values for IFNB, IFNL1, IFNL2/3, and M1 genes were determined by comparison to the respective standard curve. For IFIT1 and IFIT3, values are fold change over mock-infected cells and normalized to the housekeeping gene, GAPDH. All values are normalized to the viral M1 gene and represent the mean and standard deviation from at least triplicate determinations. *, P < 0.05; °, P < 0.01 compared to the values for the H1-WT virus by one-way ANOVA. MOI, multiplicity of infection; IFN, interferon; ISG, interferon-stimulated gene.

FIG 5.

H3-induced IFN and ISG expression in Calu-3 cells. Cells were infected with the recombinant H3 viruses (MOI of 1), and IFN and ISG expression levels were analyzed 24 hpi. Expression values for IFNB, IFNL1, IFNL2/3, and M1 genes were determined by comparison to the respective standard curve. For IFIT1 and IFIT3, values are fold change over mock-infected cells and normalized to the housekeeping gene, GAPDH. All values are normalized to the viral M1 gene and represent the mean and standard deviation from at least triplicate determinations. *, P < 0.05; °, P < 0.01 compared to the values for the H3-WT virus by one-way ANOVA. MOI, multiplicity of infection; IFN, interferon; ISG, interferon-stimulated gene.

FIG 6.

B-induced IFN and ISG expression in Calu-3 cells. Cells were infected with the recombinant B viruses (MOI of 7) and IFN and ISG expression levels were analyzed 24 hpi. Expression values for IFNB, IFNL1, IFNL2/3, and M1 genes were determined by comparison to the respective standard curve. For IFIT1 and IFIT3, values are fold change over mock-infected cells and normalized to the housekeeping gene, GAPDH. All values are normalized to the viral M1 gene and represent the mean and standard deviation from at least triplicate determinations. *, P < 0.05; °, P < 0.01 compared to the values for the B-WT virus by one-way ANOVA. MOI, multiplicity of infection; IFN, interferon; ISG, interferon-stimulated gene.

We next plotted the polymerase activity values against the IFN and ISG expression and area under the curve (AUC) values to determine whether a statistically significant correlation could be observed (Fig. S1 to S3). No significant correlation was found between polymerase activity and IFN/ISG expression induced by the recombinant H1, H3, and B viruses (Fig. S2 to S3).

Stability of baloxavir-resistant substitutions and their effect on susceptibility to favipiravir and NAIs.

To evaluate the genetic stability of the baloxavir-resistant recombinant viruses in vitro, we serially passaged each virus three times in Calu-3 cells and then sequenced their PA genes. We observed that while H1-I38T was stable throughout the passages, the H1-I38L variant acquired the PA D394N substitution (designated H1-I38L-P3, Table S1) and the H1-E199D reverted to the H1-WT virus (designated H1-E199D-P3 [1], Table S1). To gain insight into the genetic stability of the H1-E199D mutant, we repeated the serial passaging of this virus, and no additional PA substitutions were observed (designated H1-E199D-P3 [2], Table S1). Our results indicated that the E199D substitution was stable during serial passaging 50% of the time. We also observed that no additional mutations were acquired after serial passaging of the baloxavir-resistant B variant, B-G199R, in Calu-3 cells (data not shown).

We next examined whether the serially passaged H1 variants altered their susceptibility to baloxavir compared to that of the parental virus variants by plaque reduction assay in MDCK-SIAT1 cells (Table S1). The H1 virus carrying double PA substitution, I38L and D394N, had an EC₅₀ value of 11.5 ± 1.0 nM, which was comparable to that of the parental H1-I38L counterpart (EC₅₀ = 11.8 ± 2.9 nM). The H1-E199D-P3 (1) and H1-E199D-P3 (2) variants demonstrated EC₅₀ values similar to those of the H1-WT and H1-E199D viruses, respectively (Table S1).

Last, we determined whether the PA substitutions that conferred resistance to baloxavir could affect viral susceptibility to favipiravir and/or NAIs oseltamivir and zanamivir. We first assessed the susceptibility of the H1-I38L, H1-I38T, H1-E199D, and B-G199R viruses to favipiravir by plaque reduction and luciferase-based minigenome assays. None of the baloxavir-resistant PA substitutions affected sensitivity to favipiravir compared to that of the respective WT virus (Table S2). Furthermore, all tested recombinant viruses remained sensitive to oseltamivir and zanamivir as seen by plaque reduction assay in MDCK-SIAT1 cells (Table S3).

DISCUSSION

The approval of baloxavir to treat uncomplicated influenza infection represents the next line of countermeasures against this pathogen. Since the effectiveness of baloxavir can be severely compromised by the emergence of drug-resistant viruses, it is paramount to identify substitutions that may lead to baloxavir resistance. In this study, we generated recombinant H1, H3, and B variants that contained substitutions in the PA protein identified in clinical trials and surveillance that could be associated with drug resistance. We characterized their susceptibility to baloxavir, impact on polymerase activity, viral growth, and ability to induce IFN and ISG expression. We found that two previously characterized substitutions, H1 PA E199D and B PA G199R, resulted in modestly decreased sensitivity to baloxavir treatment (∼2.5-fold). Interestingly, H1 E199D and B G199R did not confer detectable decrease in baloxavir susceptibility in recombinant A/WSN/33 (1.3-fold) and B/Maryland/1/59 (0.8-fold) backgrounds, respectively (30). The modest but measurable reduction observed in this study, in contrast to previous evaluation of these substitutions in different genetic backgrounds (30), highlights the potential effect of the virus background on determination of relative susceptibility of influenza virus to baloxavir. Additionally, we observed that the majority of the evaluated substitutions had a statistically significant impact on viral replication kinetics, RNP activity, and/or IFN and ISG expression in cultures of human airway epithelium.

Substitutions at position 38 of the influenza PA protein are a major pathway to reduced sensitivity to baloxavir antiviral activity (14, 21, 23, 28, 32). Namely, H1 PA I38L and I38T reduced the influenza sensitivity to baloxavir by ∼8.0 and 116.0-fold, respectively, in MDCK and MDCK-SIAT1 cells (21, 25, 33). Our data are in a good correlation with these previous findings. In addition, the A/Victoria/361/2011 (H3N2) virus carrying the PA E199G substitution was shown to be 4.5-fold less susceptible to baloxavir inhibition in MDCK cells (21). Although we were unable to rescue the A/Hong Kong/218849/2006 (H3N2) virus with PA E199D, we observed that H3 RNP with E199D exhibited reduced baloxavir sensitivity in minigenome assay, indicating that further assessment of the impact of the H3 PA substitutions at residue 199 on drug sensitivity should be considered using different H3N2 virus backgrounds. Importantly, we demonstrated that H1 PA E199D and B PA G199R were able to confer baloxavir resistance in MDCK-SIAT1, 293T, and NHBE cells. Interestingly, in contrast to Calu-3, the NHBE cells were one of the most sensitive cell cultures capable of distinguishing drug-resistant variants compared to their respective wild-type counterparts. Therefore, differentiated cultures of human airway epithelium can be employed more frequently in the evaluation of baloxavir resistance-associated substitutions in the future. In addition, since a good correlation was observed between results obtained by minigenome assay and those from plaque reduction and virus yield reduction assays, the luciferase-based minigenome assay should be considered for assessment of potential baloxavir-resistant substitutions, especially when rapid testing is required and rescuing of the viruses is not feasible.

Monitoring of emergence of drug-resistant mutations is an integral part of influenza virological surveillance worldwide. To determine if the PB2 or PA substitutions evaluated in this study might also occur in circulating influenza viruses, we analyzed human influenza sequences deposited to the Influenza Research Database from 2009 to 2021 (www.fludb.org). As shown in Table S4, all H1 PA substitutions were observed at frequencies of ≤0.7%. The H3 PA V62I, V432I, and K492R and B PA K298R and M682L substitutions were present at frequencies of >1.0% in several years (Tables S5 and S6). Notably, the baloxavir-resistant substitutions (i.e., H1 PA I38L, I38T, E199D, and B PA G199R) were present at frequencies of <1.0% in selected years of the respective surveyed populations, indicating that their spontaneous emergence is an extremely rare event.

To gain insight into the effect of the studied PA substitutions on the H1, H3, and B PA proteins (i.e., PA, PA-X, PA-N155, and PA-N182), we analyzed the changes in the charge, polarity, and hydrophobicity that resulted from the studied substitutions (Tables S7 and S8). The most drastic changes in H1 PA were conferred by I38T, which increased the PA charge and resulted in a hydrophobic to hydrophilic change, whereas H1 PA I38L and E199D did not confer changes in any of the properties studied (Table S7). Moreover, as seen in Table S8, the B PA G199R substitution had the greatest impact on charge, polarity, and hydrophobicity among all B PA substitutions and induced a change from neutral to positive charge, nonpolar to polar, and hydrophobic to hydrophilic. Molecular mapping showed that the PA residue 38 contacts the V shape of the baloxavir tail-group and the I38T substitution leads to a change in rotomer resulting in the reduced binding of the PA protein with the compound (21). One could speculate that, in contrast to I38L, the I38T substitution may result in the reduction of the PA binding with baloxavir not only due to the confirmational change but also due to the changes in charge, polarity, and/or hydrophobicity. Further, the 199 residue links the N- and C-terminal domains of the PA protein and is located outside the baloxavir-binding domain (21). Although the mechanism on how substitutions at position 199 affect binding with baloxavir is unclear, concomitant changes in the PA properties might decrease the interaction with the drug molecule. Overall, our findings may shed light on the mechanism by which PA substitutions confer resistance to baloxavir and may help to guide the identification of other possible substitutions that could potentially be associated with drug resistance.

Because the emergence of a drug-resistant variant with comparable or increased fitness as its wild-type counterpart will reduce the effectiveness of clinical use of an antiviral drug, it is important to understand the impact of resistance-conferring substitutions on viral fitness. We observed that whereas two baloxavir-resistant variants, H1-I38T and B-G199R, did not diminish viral replication ability, H1-I38L and H1-E199D exhibited reduced replication kinetics in Calu-3 cells. In contrast, the latter two viruses showed increased polymerase activity and increased viral growth in MDCK-SIAT1 cells. In addition, serial passaging of the H1 virus with E199D resulted in reversion to the WT virus 50% of the time in Calu-3, indicating that the fitness deficit hinders the ability of this variant to successfully replicate and spread in human airway epithelium. Our results seen in Calu-3 cell cultures are in line with the previous studies on evaluation of the impact of I38T on viral growth of the A/Quebec/144147/2009 (H1N1) and A/California/04/2009 (H1N1) viruses in MDCK and MDCK-SIAT1 cells (24, 29). Controversially, a few reports showed that the PA I38T substitution may result in significant fitness deficit in MDCK and MDCK-SIAT1 cells when A/California/04/09 (H1N1)-, A/WSN/33 (H1N1)-, and A/Illinois/08/2018 (H1N1)-like virus backgrounds are used (21, 25, 32). Furthermore, our data observed in MDCK-SIAT1 cells corroborate with those reported by Chesnokov et al., where the PA I38L substitution was shown not to diminish the viral growth of the A/Illinois/37/2018 (H1N1) virus (25). Taken together, the impact of the baloxavir resistance-associated substitutions on viral fitness may depend on the virus type/subtype and may differ dramatically dependent on which cell line is used. Therefore, the appropriate cell culture that most closely resembles human airway epithelium should be considered for assessment of viral fitness of baloxavir-resistant variants and results from different studies should be compared only if the data are obtained in the same in vitro system.

The IFN response is a complex signaling pathway that staunches viral infection until more potent arms of the immune response are recruited. During influenza replication, viral pathogen-associated molecular patterns are recognized by cellular pattern recognition receptors, which results in downstream signaling events leading to the production of IFNs, which are secreted from the infected cells and trigger an antiviral state through upregulation of ISG expression (34, 35). Several studies showed that administration of type I and III IFNs inhibits influenza infection both in vitro and in vivo (36–44). In addition, type I and III IFN receptor knockout mice infected with H1N1 influenza virus developed higher tissue vial loads and more severe disease (45–48). The ISGs, such as IFITM, MxA, BST-2, and viperin, were also shown to restrict influenza replication at different stages of the virus replication cycle in vitro (49–54), and ISG knockout mice developed enhanced influenza virus infection (55–58). However, influenza viruses have evolved several strategies to suppress the IFN response to ensure their efficient replication and spread using nonstructural 1 (NS1), PB2, PB1-F2, and PA-X proteins as IFN and ISG antagonists (51, 59–64). Here, we observed that H1-I38T and B-G199R inhibited expression of the studied ISGs and IFNs in Calu-3, respectively, which may have provided an advantage for their growth. In contrast, the increased ISG expression triggered by the H1-E199D variant could be one of the factors leading to its reduced replication ability in Calu-3. Our findings may point to the important role of the influenza PA protein that it can play in antagonism against the host IFN response during infection (65, 66). Overall, future studies are warranted to assess the impact of baloxavir-resistant PA substitutions on induction of the innate response.

In conclusion, our findings add to the growing body of data describing influenza virus PA substitutions that confer resistance to baloxavir and their impact on viral fitness, polymerase activity, and innate immune response. To our knowledge, this is the first report that describes the ability of the H1 PA E199D and B PA G199R substitutions to confer modest resistance to baloxavir. Further assessment of the impact of these substitutions on drug antiviral activity is needed in animal models. Because baloxavir is a novel anti-influenza virus therapeutic agent, identifying and characterizing substitutions associated with reduced sensitivity to baloxavir, as well as the impact of these substitutions on viral fitness, is paramount to the strategic implementation of this countermeasure. Timely monitoring of all viral genetic determinants that mediate reduced susceptibility to baloxavir will guide clinical use of this novel anti-influenza drug.

MATERIALS AND METHODS

Cells.

MDCK, human embryonic kidney cell line (293T), and the human lung epithelial cell line (Calu-3) were obtained from the American Tissue Culture Collection (Manassas, VA, USA) and were maintained as described previously (67). MDCK cells transfected with cDNA encoding human 2,6-sialyltransferase (MDCK-SIAT1 cells) were kindly provided by Mikhail N. Matrosovich and were maintained as described previously (68). Primary NHBE cells were obtained from Lonza (Walkersville, MD, USA) and were grown on membrane supports (6.5-mm Transwell, Corning Inc., Corning, NY, USA) at the air-liquid interface in serum-free and hormone- and growth factor-supplemented medium as described previously (69). Once confluent, the cells were allowed to fully differentiate for at least 3 weeks before experiments.

Compounds.

Baloxavir acid (baloxavir) and favipiravir were obtained from MedChemExpress (Monmouth Junction, NJ, USA). Oseltamivir carboxylate (oseltamivir) ([3R,4R,5S]-4-acetamido-5-amino-3-[1-ethylpropoxy]-1-cyclohexene-1-carboxylic acid) was provided by Roche Diagnostics GmbH (Mannheim, Germany). Zanamivir (4-guanidino-Neu5Ac2en) was obtained from Sigma-Aldrich (St. Louis, MO, USA).

Viruses.

Recombinant influenza A/California/04/09 (H1N1)-, A/Hong Kong/218849/2006 (H3N2)-, and B/Victoria/504/2000-like influenza viruses were generated by DNA transfection of human embryonic kidney 293T cells (31). Point mutations to induce amino acid substitutions in the H3 PB2 (R209K), H1 PA (A37S, I38L/T, E199D/K, P267S, K328E, T363I, A476S, and E677D), H3 PA (F35L, V62I, T162I, E199D/K, Y321H, V432I, K492R, M595I, A618S, and G694R), and B PA (I38V, G199R, K298R, T304A, V645A, and M682L) genes of the respective WT viruses were introduced using a Quick-Change site-directed mutagenesis kit (Table 1, Stratagene, La Jolla, CA, USA). Virus stocks were prepared by incubation in 10-day-old embryonated chicken eggs for 48 h at 37°C or 33°C for the H1 and H3 or B viruses, respectively. The entire PB2 and PA genes of the H1, H3, and B viruses were sequenced to verify the presence of desired mutations and absence of additional mutations. All experimental work was performed in a biosafety level-2 laboratory approved for use of these strains by the U.S. Department of Agriculture and the U.S. Centers for Disease Control and Prevention.

Virus sequence analysis.

Viral RNAs were isolated from cell culture supernatants after transfection or after 3 passages in Calu-3 cells. Samples were then reverse transcribed and analyzed by PCR using universal primers specific for PB2 or PA gene segments as described previously (70). Sequencing was performed by the Research Central Facility for Biotechnology Resources at the U.S. Food and Drug Administration, Silver Spring, MD, USA. DNA sequences were completed and edited using the DNASTAR Laser gene sequence analysis software package (Madison, WI, USA).

Stability and infectivity of recombinant viruses.

The genetic stability of the viruses was monitored by sequencing of the PA genes after three passages in Calu-3cells at an MOI of ∼1 PFU/mL. The infectivity of recombinant viruses was determined by plaque assay (71). Briefly, confluent cultures of MDCK-SIAT1 cells were incubated for 1 h with 10-fold serial dilutions of each virus at 37°C or 33°C for H1 and H3 or B viruses, respectively. The cells were then washed and overlaid with minimal essential medium containing 0.3% bovine serum albumin (BSA), 0.45% Bacto agar, and 1 μg/mL l-(tosylamido-2-phenyl)ethylchloromethylketone (TPCK)-treated trypsin. After 2 to 3 days of incubation at 37°C (H1 and H3) or 33°C (B), the cells were stained with 0.1% crystal violet in 10% formaldehyde solution and the number of PFU/mL and plaque size of any 10 plaques were determined using a Finescale magnifying comparator.

Baloxavir cytotoxicity.

The baloxavir cell cytotoxicity (CC₅₀, the drug concentration that caused a 50% decrease in cell viability compared to untreated control wells) was previously determined to be 34.1 ± 1.9, 10.1 ± 2.1 μM, and 7.8 ± 0.9 μM at 24, 48, and 72 h, respectively, in MDCK-SIAT1 cells and 3.0 ± 1.3 μM in MDCK cells (20, 27). To assess baloxavir cytotoxicity in 293, Calu-3, and NHBE cells, cells were treated with baloxavir ranged from 0.001 to 1,000 nM diluted in the appropriate infection medium and concurrent with each antiviral assay described below. Cytotoxicity was measured via CellTiter-Blue cell viability assay (Promega, Madison, WI, USA), and CC₅₀ values were found to be >1,000 nM in all cell lines tested.

Viral replication kinetics.

To determine the multistep growth curves for each virus, Calu-3 cells were infected with H1, H3, or B viruses at an MOI of 0.01 PFU/cell. After incubation for 1 h at 37°C or 33°C for H1 and H3 or B viruses, respectively, the cells were washed and then overlaid with medium containing 0.3% BSA and 1 μg/mL TPCK-treated trypsin. The supernatants were collected at the indicated time points and kept at −80°C until titration.

Plaque reduction assay.

Drug sensitivity of the recombinant influenza viruses was determined by plaque reduction assay, as described previously (71). Briefly, confluent MDCK-SIAT1 cells were inoculated with 50 to 100 PFU of influenza virus for 1 h and then overlaid with minimal essential medium containing 0.3% BSA, 0.45% Bacto agar, 1 μg/mL of TPCK-treated trypsin, and baloxavir (ranging from 0.05 nM to 100 nM), favipiravir (ranging from 0.05 to 1,000 μM), or NAIs (ranging from 0.001 nM to 100 μM). After 3 days of incubation at 37°C or 33°C for H1 and H3 or B viruses, respectively, the cells were stained with 0.1% crystal violet in 10% formaldehyde solution and the plaques were counted to determine the PFU/mL. The drug concentration that caused a 50% decrease in the PFU titer compared to that in control wells without drug was defined as the EC₅₀. The results of two independent experiments were averaged.

Virus yield reduction assay.

The virus yield reduction assay was performed as described previously (44). Briefly, 24-well plates containing confluent Calu-3 or NHBE cells were pretreated with baloxavir ranging from 0.05 nM to 100 nM for 2 h and then infected with influenza virus at an MOI of 0.1 PFU/cell for 1 h. After incubation, the cells were washed, and Calu-3 cells were overlaid with drug-containing media. NHBE cells were cultured under air-liquid interface conditions. The supernatants were collected after 2 days and virus yields were determined as the number of PFU/mL in MDCK cells. The drug concentration that caused a 50% decrease in the PFU titer in comparison to control wells without drug was defined as the EC₅₀. The results of two independent experiments were averaged.

Minigenome assay for polymerase activity.

293T cells were pretreated with different concentrations of baloxavir (ranging from 0 nM to 1,000 nM) or favipiravir (ranging from 0 μM to 1,000 μM) for 1 h before being transfected with a mix of PB1, PB2, PA (WT or mutated), and NP plasmids in quantities of 1, 1, 1, and 2 μg, respectively, a luciferase reporter plasmid (enhanced green fluorescent protein [EGFP] open reading frame in pHW72-EGFP substituted with a firefly luciferase gene), and pGL4.75[hRluc/CMV] vector, which expresses Renilla luciferase (Promega) and was used as an internal control for transfection efficiency. After 24 h, cell extracts were harvested and lysed, and luciferase levels were assayed with a dual luciferase-based assay system (Promega). The effect of each substitution on RNP activity was assessed by comparing the luciferase levels of cells transfected with mutant RNP to those transfected with the respective WT virus. The concentration of baloxavir or favipiravir that caused a 50% reduction in RNP activity in minigenome assay compared to that in control cells without drug was defined as the IC₅₀. Experiments were performed at least in triplicate.

qPCR of IFN, ISG, and viral matrix (M1) genes.

Quantification of changes in gene expression was carried out by quantitative real-time PCR (qPCR) analyses of individual IFNs (IFNB1, IFNL1, and IFNL2/3), ISGs (IFIT1 and IFIT3), and influenza M1 gene. Calu-3 cells were infected with virus (H1 and B: MOI of 7, H3: MOI of 1), and total cellular RNA was harvested 24 h later using the RNeasy minikit (Qiagen, Germantown, MD, USA). The collected RNA samples were then treated with DNase, and then 1 μg of each purified RNA sample was reverse transcribed to cDNA with Quantiscript reverse transcriptase (Qiagen). The cDNAs were mixed with RT² SYBR green qPCR Mastermix (Qiagen), and qPCR analyses were performed using the ViiA 7 instrument (Applied Biosystems, Waltham, MA). IFNB1, IFNL1, IFNL2/3, and influenza M1 gene copy numbers were assayed using TaqMan gene expression assay primer/probe sets and master mix (Life Technologies, Carlsbad, CA, USA) and ViiA 7 software v.1.2.2 (Applied Biosystems). The values were determined by comparison to respective standard curves for each gene. Changes in ISGs expression levels were expressed as the mean-fold increase relative to the untreated control gene expression levels after normalization to the housekeeping gene, GAPDH. Values for each gene were normalized to M1 gene expression for each virus and then plotted against the polymerase activity of the virus. Statistical analysis of the qPCR results was performed using Prism 9.0 (GraphPad Software, La Jolla, CA, USA). Values are the means of three independent determinations.

NA enzyme inhibition assay.

H1-WT and B-WT viruses were standardized to equivalent NA activity and incubated for 30 min at 37°C or 33°C for H1-WT or B-WT, respectively, with NAIs at concentrations of 0.0001 to 5 μM and then with MUNANA (Sigma-Aldrich) as a substrate. The reaction was then terminated by adding 14 mM NaOH, and fluorescence was quantified in a Synergy 2 multimode microplate reader. The IC₅₀ concentration of each NAI was determined by plotting the dose-response curve of inhibition of NA activity as a function of the compound concentration. Values represent the mean of at least two independent experiments, each consisting of two or three replicates.

Statistical analysis.

Virus yield, plaque size and number, IC₅₀ and EC₅₀ values, levels of IFNs and ISGs expression, and polymerase activities of RNP complexes of the WT and mutant H1, H3, and B viruses were compared by analysis of variance (ANOVA). The degree of association between polymerase activities and levels of IFN/ISG expression and AUC values was determined by the Spearman correlation test. Probability values of ≤0.05 indicated statistically significant differences.

Data availability.

Sequences have been deposited in GenBank under accession numbers OM585550 to OM585577.