Recent H3N2 Influenza Virus Clinical Isolates Rapidly Acquire Hemagglutinin or Neuraminidase Mutations When Propagated for Antigenic Analyses

Abstract

Prior to serological testing, influenza viruses are typically propagated in eggs or cell culture. Recent human H3N2 strains bind to cells with low avidity. Here, we isolated nine primary H3N2 viral isolates from respiratory secretions of children. Upon propagation in vitro, five of these isolates acquired hemagglutinin or neuraminidase mutations that increased virus binding to cell surfaces. These mutations can potentially confound serological assays commonly used to identify antigenically novel influenza viruses.

TEXT

Influenza viruses continually acquire mutations in antigenic sites of hemagglutinin (HA) and neuraminidase (NA) glycoproteins, a process termed antigenic drift (1). H3N2 viruses caused a pandemic in 1968 and have continued to circulate in humans on a seasonal basis. Since 1968, H3N2 viruses have steadily acquired mutations in the exposed globular head of HA (2). Many HA mutations simultaneously alter the antigenicity and receptor binding properties of influenza viruses (3, 4). Recent H3N2 strains have dramatically reduced receptor binding avidity and altered receptor specificity (5, 6).

Twice a year, the World Health Organization (WHO) faces the challenging task of choosing seasonal influenza virus vaccine strains. Most neutralizing influenza virus antibodies are directed against the HA protein, and hemagglutination inhibition (HAI) assays are typically used to characterize the antigenic properties of circulating viral strains (7). HAI assays measure the amount of antisera required to prevent virus agglutination of red blood cells (RBCs) (8). HA is the sole attachment factor for the majority of influenza virus strains, and HA agglutinates RBCs by binding to sialic acid. HAI assays are intended to quantify relative amounts of HA antibodies; however, recent studies demonstrated that some H3N2 and H1N1 strains agglutinate RBCs through NA-sialic acid interactions (9–11).

It is widely known that influenza viruses can acquire HA and NA mutations when passaged in tissue culture or eggs; however, few studies have determined if these adaptive mutations influence antigenic assays. We completed a series of experiments to determine if HA and/or NA mutations commonly arise when contemporary clinical H3N2 isolates are propagated prior to antigenic testing. We used Madin-Darby canine kidney (MDCK) cells for these experiments, since they are routinely used by surveillance laboratories to isolate human viruses (12). Respiratory secretions were collected from 10 children (ages 1 month to 7 years old) at the Children's Hospital of Philadelphia for routine diagnostic purposes during the 2012-2013 influenza season. Leftover deidentified clinical samples were passaged twice on MDCK cells with serum-free media containing tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin and gentamicin. Supernatants were collected, and infectious titers were quantified by a 50% tissue culture infective dose (TCID₅₀) assay. Agglutination was then measured using turkey and guinea pig RBCs as previously described (13).

Although infectious viruses were detected in 9 of 10 MDCK cultures, only 5 of these virus isolates agglutinated turkey RBCs (Lampire Biological Laboratories, Pipersville, PA) (Table 1). Some H3N2 strains bind more efficiently to guinea pig RBCs (14), which express distinct sialic acid receptors (15). All nine infectious virus isolates agglutinated guinea pig RBCs, and agglutination titers correlated with infectious titers (Table 1). To determine if agglutination of each viral isolate was mediated by HA or NA, we repeated assays in the presence of 10 μM oseltamivir, a compound that binds in the sialic acid binding site of NA (16). The addition of oseltamivir reduced agglutination of four isolates (Table 1), indicating that these isolates bind to cells through an NA-dependent mechanism.

TABLE 1.

Infectious and HAU titers of expanded clinical isolates^a

Isolate no.	TCID50/ml	Turkey HAU		Guinea pig HAU		Category
		−Oselt	+Oselt	−Oselt	+Oselt
1	1.35E + 07	19	<2	48	16	NA dependent
2	4.07E + 06	<2	<2	4	4	Low
3	1.35E + 06	<2	<2	4	3	Low
4	4.07E + 06	<2	<2	6	6	Low
5	<1.00E + 02	<2	<2	<2	<2	Not detected
6	5.56E + 07	36	30	128	128	NA independent
7	6.88E + 07	3	<2	16	8	NA dependent
8	1.89E + 07	22	<2	64	12	NA dependent
9	4.07E + 06	2	<2	8	4	NA dependent
10	1.48E + 06	<2	<2	4	4	Low

^aShown are infectious (TCID₅₀/ml) and agglutination (HAU) titers after clinical isolates were passaged twice on MDCK cells. Oselt, oseltamivir.

We grouped the viral isolates into three categories: viruses that agglutinate via an NA-dependent mechanism (n = 4); a virus that agglutinates via an NA-independent mechanism (n = 1); and viruses that agglutinate poorly (n = 5) (Tables 1 and 2). We sequenced the HA1 and NA genes from primary clinical material and MDCK-grown virus using standard Sanger techniques. We compared the HA1 and NA sequences to the A/Victoria/361/2011 (A/Victoria/361/11) H3N2 vaccine strain (EpiFlu accession numbers EPI349103 and EPI349104). There were HA and NA sequence differences among the primary clinical isolates (Table 2; linear numbering used throughout the manuscript). All four viral isolates that agglutinated RBCs via an NA-dependent mechanism acquired mutations at NA residue 151 during MDCK expansion but did not acquire HA mutations. Reverse-genetics experiments have previously demonstrated that D151G and D151N NA mutations facilitate NA-dependent viral attachment (9, 10). These NA mutations were not detected in the original clinical isolates, which is consistent with previous sequencing studies (17). The isolate that agglutinated RBCs via an NA-independent mechanism (isolate 6) did not acquire an NA mutation following MDCK expansion but did acquire a single HA mutation at residue 237. This isolate also uniquely possessed an F209S HA mutation, which was present before and after MDCK expansion. We did not detect HA or NA mutations in the five MDCK-expanded viral isolates that grew to low titers and agglutinated RBCs poorly.

TABLE 2.

Differences in sequences of clinical isolates before and after MDCK expansion^a

Isolate no.	Category	Differences in NA sequence from that of the H3N2 vaccine strainb	NA mutations upon MDCK expansion	Differences in HA sequence from that of the H3N2 vaccine strainb	HA mutation upon MDCK expansion
1	NA dependent	V143 M, K258E, T329N	D151G	Q49R, N161S, N294K
7	NA dependent	N141S, K258E, T329N	D151N	Q49R, T144A, R158G, N161S, Q189H, N294K
8	NA dependent	K258E, T329N	D151G	Q49R, T144A, R158G, N161S, N294K
9	NA dependent	K258E, T329N	D151G	R158G, N161S, N294K
6	NA independent	K258E, T329N		Q49R, T144A, R158G, N161S, F209S, N294K	P237L
2	Low	K258E, T329N		T144A, R158G, N160S, N161S, N294K
3	Low	K128R, V143M, K258E, T329N		Q49R, N161S, N294K
4	Low	P79S, G93D, E221D, K258E, T329N		Q49R, N161S, N294K
10	Low	K258E, T329N		Q49R, N161S, N294K

^aMutations that arose upon MDCK expansion are in bold.

^bSequences were compared to that of the A/Victoria/361/11 2012-2013 vaccine strain (linear numbering).

We completed additional reverse-genetics studies to determine if the P237L or F209S HA mutation promoted enhanced growth and agglutination of isolate 6. We used a QuikChange kit (Stratagene, La Jolla, CA) to introduce each of these mutations into the HAs of the A/Victoria/361/11 H3N2 vaccine strain and isolate 2. We chose isolate 2 for reverse-genetics experiments, since this isolate agglutinates poorly (Table 1). We rescued virus using A/Puerto Rico/8/1934 (PR8) internal genes and the NA of either A/Victoria/361/11 or isolate 2 after transfecting 293T/MDCK cocultures as previously described (18). Viruses engineered to have the P237L mutation, but not the F209S mutation, grew to high titers and agglutinated RBCs in the presence of oseltamivir (Table 3). P237 is located in the HA trimer interface (Fig. 1A), and it is possible that this mutation affects sialic acid binding by stabilizing the HA trimer. We determined the relative receptor avidities of viruses possessing either P237 or L237 by completing agglutination assays with RBCs previously treated with neuraminidase (receptor-destroying enzyme, cholera filtrate; Sigma, St. Louis, MO), which removes sialic acid receptors (3, 19). A/Victoria/361/11 viruses with the P237L HA mutation efficiently agglutinated RBCs treated with larger amounts of receptor-destroying enzyme (Fig. 1B), indicating that these viruses have higher receptor binding avidities. We completed HAI assays using antisera generated against A/Victoria/361/2011 (Influenza Reagent Resource, Atlanta, GA) and found that viruses with the P237L HA mutation had reduced HAI titers (Fig. 1C), which is consistent with previous studies demonstrating that viruses with higher avidity are difficult to inhibit in HAI assays (3, 19, 20). Using Sanger sequencing, we did not detect the P237L HA mutation in the original clinical material of isolate 6 (Table 2); however, we were able to detect low levels of this mutation in the original clinical isolate after sequencing individual clones following TA cloning (TOPO TA Cloning kit; Life Technologies, Carlsbad, CA) of PCR products (Fig. 1D).

TABLE 3.

Infectious and HAU titers of reverse-genetics viruses^a

Virus	TCID50/ml	Turkey HAU		Guinea pig HAU
		−Oselt	+Oselt	−Oselt	+Oselt
A/Victoria/361/11 WT HA	1.20E + 06	2	2	16	16
A/Victoria/361/11 F209S HA	2.58E + 04	<2	<2	<2	<2
A/Victoria/361/11 P237L HA	1.76E + 07	192	192	256	256
Isolate 2 WT HA	1.20E + 05	<2	<2	3	3
Isolate 2 F209S HA	1.76E + 02	<2	<2	<2	<2
Isolate 2 P237L HA	2.58E + 06	48	48	64	64

^aShown are infectious (TCID₅₀/ml) and agglutination (HAU) titers. Oselt, oseltamivir.

FIG 1.

Characterization of the P237L HA mutation. (A) Residue 237 is highlighted in orange on the H3 structure and sialic acid is colored white (Protein Data Bank entry 1HGG). (B) Relative receptor binding avidities of reverse-genetics-derived viruses were determined by agglutination of guinea pig RBCs pretreated with increasing amounts of neuraminidase. Data are expressed as maximal amounts of neuraminidase that allowed full agglutination. Means and standard errors of the means of the results for triplicate samples are shown. WT and P237L viruses were compared using one-tailed, unpaired t tests. Data are representative of two independent experiments. (C) HAI assays were completed using anti-A/Victoria/361/11 sera generated in ferrets. (D) P237 and L237 HA frequencies were determined after TA cloning the PCR product amplified from original clinical material. We sequenced 21 plasmids following TA cloning.

Taken together, our data indicate that recent clinical H3N2 strains replicate poorly in MDCK cells and rapidly acquire either HA or NA mutations when propagated in vitro prior to antigenic testing. These mutations increase virus growth and agglutination titers and can potentially skew HAI assays. The goal of HAI assays is to detect HA antigenic changes, and therefore it is difficult to interpret data when agglutination is mediated by mutated NAs. The addition of oseltamivir directly into HAI assays may circumvent this problem (9). We have previously shown that HA mutations that alter receptor binding avidity can skew antigenic maps based on HAI assays (20), and recent computational methods have been developed to create antigenic maps that account for variations in viral avidity (20, 21). Future studies should focus on alternative cell-derived systems for propagating primary influenza isolates, such as MDCK cells engineered to express different types of sialic acid receptors (12, 22). Additional studies should also address if cell culture-derived mutations in vaccine strains impact immunogenicity, since recent studies suggest that egg-derived mutations in vaccine strains can have this effect (23).

Nucleotide sequence accession numbers.

The nucleotide sequences determined in this study have been deposited in GenBank under accession numbers KJ734801 to KJ734838.