Whole-Genome Analysis Surveillance of Influenza A Virus Resistance to Polymerase Complex Inhibitors in Eastern Spain from 2016 to 2019

ABSTRACT

Molecular surveillance by whole-genome sequencing was used to monitor the susceptibility of circulating influenza A viruses to three polymerase complex inhibitors. A total of 12 resistance substitutions were found among 285 genomes analyzed, but none were associated with high levels of resistance. Natural resistance to these influenza A antivirals is currently uncommon.

INTRODUCTION

Antiviral compounds have been developed to target each of the three proteins that comprise the RNA-dependent RNA polymerase (RdRP) complex because of their highly conserved sequences. These antivirals disrupt viral mRNA synthesis and therefore affect the early stages of virus replication. One of these antivirals is the purine nucleoside analogue favipiravir, which is recognized as an alternate substrate by polymerase basic protein 1 (PB1), resulting in inaccurate viral RNA synthesis (1). Although being a broad-spectrum antiviral agent, favipiravir has already been licensed in Japan only for use against emerging influenza viruses resistant to other antivirals due to the risk of teratogenicity and embryotoxicity (2). Another polymerase inhibitor is pimodivir, which targets influenza A virus polymerase basic protein 2 (PB2), preventing the binding of host mRNA to PB2. However, due to structural differences in the PB2 cap-binding pocket, pimodivir is ineffective against influenza B viruses (3). The clinical development of pimodivir for influenza A virus was discontinued recently. The third polymerase inhibitor is baloxavir, which was licensed for the treatment of uncomplicated influenza in Japan and the United States in 2018. Baloxavir targets the polymerase acidic protein (PA) of influenza A and B viruses, inhibiting the cap-dependent endonuclease activity of the PA and, therefore, the cleavage of the host mRNA to create RNA fragments that will be used to prime viral transcription (4).

Studies evaluating mutations reducing antiviral susceptibility identified some amino acid substitutions that may be responsible for reduced susceptibility to these antivirals. A(H1N1)pdm09 viruses carrying the PB1 K229R substitution showed reduced favipiravir susceptibility in vitro (5), and the PA L666F substitution was commonly found after favipiravir administration but was linked to reduced polymerase activity (6). Reduced susceptibility of influenza A virus to pimodivir was associated with PB2 S324I/N/K/C/R, F325L, S337P, K376N/R, T378S, M431I, and N510T/K mutations (7–9). Finally, it has been demonstrated that the substitutions PA E23K/G, A36V, A37T, I38T/F/M/N/S/L/V, E119D/K, and E199G conferred reduced susceptibility to baloxavir, with most of these mutations emerging in treated patients (4, 10, 11; summarized at https://www.who.int/influenza/gisrs_laboratory/antiviral_susceptibility/Baloxavir_AAS_table.pdf).

In this study, we monitored the natural susceptibility of influenza A viruses to favipiravir, pimodivir, and baloxavir using next-generation sequencing (NGS) whole-genome analysis (WGA) among isolates circulating during three consecutive influenza seasons, from 2016 to 2019, identified in five tertiary-care hospitals from eastern Spain (Alicante, 1; Barcelona, 1; Castellón, 1; Valencia, 2) and in the context of influenza surveillance schemes approved by the corresponding institutional boards and performed in accordance with the 1964 Declaration of Helsinki and its later amendments.

Samples were retrieved from 285 influenza virus-positive untreated patients who were admitted after presenting with acute respiratory illness and were classified as mild or severe according to the criteria of admission to the intensive-care unit (ICU) or a fatal outcome. After RNA isolation from nasopharyngeal swabs and multisegment reverse transcription-PCR (RT-PCR) amplification (12), NGS libraries were generated using the Nextera XT kit (Illumina), and the complete genomes of 197 A(H3N2) and 88 A(H1N1)pdm09 viruses were sequenced in batches of 96 samples in a NextSeq sequencer (Illumina) with the 2 × 150 mid-output kit. An in-house automated bioinformatic pipeline implemented in R was used for demultiplexing, quality control, sequence pairing, assembly, the generation of consensus sequences, and variant analysis for each sample and segment of the viral genome, which is based on the use of prinseq-lite (13), FLASH (14), and snippy (https://github.com/tseemann/snippy). The publicly available INSaFLU platform was used to perform minority variant analysis (15). Subsequently, the deduced PB1, PB2, and PA protein sequences (see the accession numbers in Table 1) were analyzed for amino acid substitutions previously associated with reduced favipiravir, pimodivir, and baloxavir susceptibility, respectively (Tables 2 to 4).

TABLE 1.

GISAID accession numbers of the 285 influenza A virus sequences analyzed

Subtype	GISAID accession no.
H3N2	EPI_ISL_345253
EPI_ISL_345255
EPI_ISL_345259
EPI_ISL_345262
EPI_ISL_345269
EPI_ISL_369232
EPI_ISL_369234
EPI_ISL_369237
EPI_ISL_369241
EPI_ISL_369256
EPI_ISL_377417–EPI_ISL_377525
EPI_ISL_377527
EPI_ISL_398155
EPI_ISL_507339–EPI_ISL_507414
H1N1pdm09	EPI_ISL_344884
EPI_ISL_345238–EPI_ISL_345251
EPI_ISL_369271
EPI_ISL_369273
EPI_ISL_372917
EPI_ISL_372919
EPI_ISL_372922
EPI_ISL_372925
EPI_ISL_372928–EPI_ISL_372937
EPI_ISL_372942–EPI_ISL_372955
EPI_ISL_376020–EPI_ISL_376033
EPI_ISL_376058–EPI_ISL_376081
EPI_ISL_414018
EPI_ISL_507313–EPI_ISL_507316

TABLE 2.

Amino acid substitutions implicated in reduced susceptibility to polymerase inhibitor antivirals retrieved from A(H3N2) sequences in five tertiary-care hospitals from eastern Spain from 2016 to 2019^a

Gene	No. of substitutions or mutation(s) (no. of isolates; mutation frequency [%])
	2016–2017		2017–2018		2018–2019		Total (n = 197)
	Mild(n = 30)	Severe (n = 24)	Mild (n = 9)	Severe (n = 18)	Mild (n = 103)	Severe (n = 13)
PB1	0	0	0	0	0	0	0
PB2	S337A (2; 1.4–2)	0	0	0	S337A (2; 1.3–1.5)	S337A (1; 1)	5
PA	I38V (1; 1.76)	A36E (1; 2.53), E119K (1; 14.19)	I38V (1; 14.63)	0	0	I38L (1; 3.22)	5
Total	3	2	1	0	2	2	10

^aNote that the fold changes (50% effective concentrations [EC₅₀s]) are as follows: unknown for A36E, 2 for I38V, 7 to 8 for I38L, 2 for E119K, and unknown for S337A (4, 10). Numbers of isolates and percent mutation frequencies within the intrahost viral population are shown in parentheses. Influenza virus-positive untreated patients with acute respiratory illness were classified as having severe illness according to the criteria of admission to the intensive-care unit (ICU) or fatal outcome.

TABLE 4.

Data on the 12 variant viruses with amino acid substitutions in the PB2 or PA gene

Subtype and season (yrs)	Sample name	Collection date (day/mo/yr)	Patient outcomea	Substitution (gene)	GISAID accession no.
H3N2
2016–2017	A/Catalonia/173509208/2017	22/01/2017	Mild	S337A (PB2)	EPI_ISL_507397
A/Catalonia/770004109_170611195/2017	25/01/2017	Mild	S337A (PB2)	EPI_ISL_507405
A/Catalonia/770004107_170605075/2017	11/01/2017	Mild	I38V (PA)	EPI_ISL_507403
A/Valencia/07_1120_20170208	08/02/2017	Severe	A36E (PA)	EPI_ISL_507357
A/Catalonia/770004112_173508442/2017	08/01/2017	Severe	E119K (PA)	EPI_ISL_507407
2017–2018	A/Catalonia/10701/2018	16/02/2018	Mild	I38V (PA)	EPI_ISL_507375
2018–2019	A/Valencia/10_1542_20190201	01/02/2019	Mild	S337A (PB2)	EPI_ISL_377502
A/Castellon/02_1550_20190218	18/02/2019	Mild	S337A (PB2)	EPI_ISL_377448
A/Castellon/02_1450_20190208	08/02/2019	Severe	S337A (PB2)	EPI_ISL_377444
A/Valencia/10_1349_20190119	19/01/2019	Severe	I38L (PA)	EPI_ISL_377495
H1N1pdm09
2017–2018	A/Catalonia/2242523NS.2/2017	09/01/2018	Severe	E119K (PA)	EPI_ISL_507316
2018–2019	A/Valencia/07_1643_20190215	15/02/2019	Mild	N510T (PB2)	EPI_ISL_376029

^aInfluenza virus-positive untreated patients with acute respiratory illness were classified as having severe illness according to the criteria of admission to the ICU or fatal outcome.

We did not find substitutions associated with reduced favipiravir susceptibility (PB1 K229R or PA L666F) in any of the 285 isolates analyzed. In five cases, we observed PB2 S337A substitutions as minority variants in A(H3N2) viruses. Four corresponded to mild cases from the 2016–2017 and 2018–2019 seasons, and one was from a severe case in the 2018–2019 season. Furthermore, PB2 N510T was detected as a minority variant in one A(H1N1)pdm09 isolate from a mild case in the 2018–2019 season. The PB2 substitutions S337P and N510T reduce susceptibility to pimodivir in vitro (8, 9), while the effect of S337A remains unknown. Finally, 6 amino acid substitutions in PA were found as minority variants. Five of these mutations were detected in A(H3N2) viruses, 2 from mild cases and 3 from severe cases. Substitutions in PA at position 38 were found in isolates from the three seasons: two mild cases (I38V) and one severe case (I38L). Some substitutions at this position, such as I38T/F/M/N/S, are associated with a significant decrease of baloxavir susceptibility (>10-fold reduction) and presumably emerge after exposure to the antiviral (10, 11), while I38L reduces baloxavir susceptibility approximately 8-fold in A(H1N1)pdm09 but only 2-fold in A(H3N2) viruses and appeared spontaneously in an untreated patient (4, 11). Finally, I38V seems a less relevant substitution because it reduces baloxavir susceptibility only 1- to 2-fold (4, 10). The two other PA mutations found, A36E and E119K, were detected independently as minority variants in two A(H3N2) isolates from severe cases from the 2016–2017 season. Moreover, E119K was also found in an A(H1N1)pdm09 isolate from a severe case in the 2017–2018 season. The PA A36V and E119K substitutions modestly reduce baloxavir susceptibility in vitro, 4-fold and 6-fold for A36V in A(H1N1) and A(H3N2) viruses, respectively (10), or 2-fold for an E119E/K mixed population of A(H3N2) viruses (4), while the effect of the PA A36E substitution is still unknown.

In summary, amino acid substitutions previously described in the literature altering favipiravir, pimodivir, and baloxavir susceptibility were screened by WGA in almost 300 influenza A viruses from untreated inpatients during three influenza seasons. Overall, the frequency of amino acid substitutions associated with reduced susceptibility to these three antivirals was low: 12 variant viruses with a single substitution each among all 285 influenza A virus genomes analyzed, with 10 (5 in PB2 and 5 in PA) detected in A(H3N2) and 2 (1 in PB2 and 1 in PA) detected in A(H1N1)pdm09 viruses. One can speculate that, even in untreated patients, emergence of resistance associated with PA substitutions may occur at higher rates in A(H3N2) than in A(H1N1)pdm09 viruses, as has been proposed for baloxavir-treated patients (16). None of the substitutions found have been associated with high levels of resistance for any of the three antivirals: PB1 K229R (30-fold reduction in favipiravir susceptibility) (5), PB2 M431I (57-fold reduction in pimodivir susceptibility) (7), or PA I38T/F/M/N/S (27- to 57-, 11- to 20-, 14-, 24- to 10-, and 6- to 12-fold reductions in susceptibility to baloxavir, respectively) (10, 11). In turn, most of the substitutions found may have little effect on antiviral susceptibility (<3-fold reduction), and all of them appeared as intrahost minority variants with frequencies below 15% of the viral population (Tables 2 and 3). Furthermore, substitutions occurred at similarly low frequencies in viruses from inpatients with mild (0.033 substitutions/patient) or severe (0.068 substitutions/patient) outcomes, who had not been treated with any polymerase complex inhibitor.

TABLE 3.

Amino acid substitutions implicated in reduced susceptibility to polymerase inhibitor antivirals retrieved from A(H1N1)pdm09 sequences in five tertiary-care hospitals from eastern Spain from 2016 to 2019^a

Gene	No. of substitutions or mutation(s) (no. of isolates; mutation frequency [%])
	2016–2017		2017–2018		2018–2019		Total (n = 88)
	Mild (n = 0)	Severe (n = 0)	Mild (n = 0)	Severe (n = 6)	Mild (n = 70)	Severe (n = 12)
PB1	0	0	0	0	0	0	0
PB2	0	0	0	0	N510T (1; 1)	0	1
PA	0	0	0	E119K (1; 2.80)	0	0	1
Total	0	0	0	1	1	0	2

^aNote that the fold changes (EC₅₀s) are as follows: 2 for E119K and >10 for N510T (4, 9). Numbers of isolates and percent mutation frequencies within the intrahost viral population are shown. Influenza virus-positive untreated patients with acute respiratory illness were classified as having severe illness according to the criteria of admission to the ICU or fatal outcome.

Substitutions at the conserved position 38 of PA are most common after baloxavir administration (17, 18), which may be associated with increased virus titers, prolonged shedding, an initial delay in symptom alleviation, or, infrequently, symptom rebound (19). Children (<12 years old) show higher rates of PA position 38 substitutions after baloxavir administration (17), and human-to-human transmission of a PA I38T virus variant from baloxavir-treated patients has been suggested (17, 20). Fitness assays of PA I38 variants are not conclusive due to experimental conditions and the viral genetic background (18, 21). Finally, both in vitro and in vivo competition experiments show that I38T or I38M variants are genetically stable, although wild-type viruses can outcompete the I38 variant viruses (21). Possible human-to-human transmission was also suspected for a PA E23K virus variant recovered from an untreated child (22). PA E23K reduces baloxavir susceptibility 7- to 9-fold in A(H1N1)pdm09 (22) and 6-fold in A(H3N2) (10) viruses, although the replication of this variant was significantly reduced in A(H1N1)pdm09 virus compared to the wild‐type virus in vitro (22).

In the current study, no PA I38T/F/M/N/S or E23K substitutions were found, but PA I38L/V substitutions were found as naturally occurring variations, as previously described (11). Although these variants, PA I38L [8-fold in A(H1N1)pdm09 (4) and 2-fold in A(H3N2) (11) viruses] and PA I38V (1- to 2-fold) (4, 10), showed low levels of reduced susceptibility to baloxavir, they showed replicative fitness similar to that of their respective A(H1N1) and A(H3N2) wild‐type counterparts in vitro (11). Other PA substitutions, such as E23K/G, A37T, and E199G, have also been associated with A(H1N1)pdm09 viral rebound and should also be monitored irrespective of their effect on antiviral susceptibility (23).

Collectively, our results and published data suggest that the emergence of antiviral resistance to influenza A virus polymerase complex inhibitors is not common in the absence of the antiviral compound and support the importance of expanding the molecular surveillance of influenza viruses to WGA through NGS to monitor for mutations associated with drug resistance or increased pathogenicity that could spread in the population, including amino acid substitutions such as the ones found in this study, which may be present as intrahost minority variants (24).