A Mutant H3N2 Influenza Virus Uses an Alternative Activation Mechanism in TMPRSS2 Knockout Mice by Loss of an Oligosaccharide in the Hemagglutinin Stalk Region

TEXT

Influenza A virus (IAV), a member of the Orthomyxoviridae family, affects and kills many humans worldwide. The viral hemagglutinin (HA) glycoprotein, which is responsible for receptor binding and subsequent membrane fusion, is synthesized as the inactive precursor, HA₀, and cleaved by a host cell protease(s) into HA₁ and HA₂ subunits. The cleavage is essential for HA to mediate membrane fusion.

Using mouse models, three research groups recently demonstrated that a type II transmembrane serine protease, TMPRSS2, expressed in the airway epithelium, is critically important for HA cleavage in vivo (1–3). Mice lacking TMPRSS2 expression (TMPRSS2 knockout [tmprss2^KO] mice) are highly tolerant of challenge infection by low-pathogenic (LP) IAVs possessing a monobasic cleavage site in HA. Our group (1) and Hatesuer et al. (2) showed that TMPRSS2 is essential for proteolytic activation of both H1N1 and H3N2 subtype IAVs. However, Tarnow et al. (3) reported that TMPRSS2 is not critical for activation of H3N2 IAV. Since the three groups used different H3N2 IAV strains, the discrepancy likely arose from possible variations in HA among the H3N2 IAV strains. The present study reveals a molecular determinant that modulates the TMPRSS2 dependency of H3N2 IAV in mouse lungs.

A mouse-adapted human H3N2 IAV strain, MA-A/Guizhou-X (H3^WT) (1), was used as a reference wild-type (WT) virus. The virus was a reassortant H3N2 virus of A/Guizhou/54/89 (H3N2), a Chinese human isolate from 1989, and A/Puerto Rico/8/34 (H1N1) (4). The H3^WT virus was passaged 10 times in tmprss2^KO mice, and a tmprss2^KO mouse-adapted MA-A/Guizhou-X virus (H3^p10) was obtained.

The parental H3^WT and H3^p10 viruses were inoculated intranasally into tmprss2^KO mice (n = 5). After challenge infection with 3.0 × 10³ PFU of H3^WT, tmprss2^KO mice showed no body weight loss, while all TMPRSS2^+/+ wild-type mice (tmprss2^WT) showed severe body weight loss (Fig. 1A). Conversely, both tmprss2^KO and tmprss2^WT mice infected with 1.7 × 10² PFU of H3^p10 showed severe body weight loss and required euthanasia on day 6 postinfection (p.i.) (Fig. 1B). To compare the susceptibility between the two mouse strains in detail, both mouse strains were also infected with 1.7 × 10, 1.7, and 0.17 PFU of H3^p10 (Fig. 1C to E). The H3^p10 infections exhibited pathogenicity in both tmprss2^KO and tmprss2^WT mice, although the pathological changes were slightly less severe in tmprss2^KO mice than in tmprss2^WT mice. The 50% lethal doses of H3^p10 for tmprss2^WT and tmprss2^KO mice were 0.5 and 5.4 PFU, respectively. Histopathological examination revealed severe inflammation in the lungs of both tmprss2^KO and tmprss2^WT mice infected with H3^p10, and viral antigens were detected throughout the lungs of both mouse strains (Fig. 1F).

• Open in new tab
• Download powerpoint

FIG 1

Pathogenicity of the parental (H3^WT) and tmprss2^KO mouse-adapted (H3^p10) GZX[H3N2]. (A) Body weight kinetics of tmprss2^WT (filled triangles) and tmprss2^KO (open triangles) mice infected with 3.0 × 10³ PFU of H3^WT. (B to E) Body weight kinetics of tmprss2^WT (filled triangles) and tmprss2^KO (open triangles) mice infected with 1.7 × 10² (B), 1.7 × 10 (C), 1.7 (D), or 0.17 (E) PFU of H3^p10. (F) Histopathological findings in the lungs of tmprss2^WT and tmprss2^KO mice infected with 1.7 × 10² PFU of H3^p10. Data obtained by hematoxylin and eosin (HE) staining (original magnification, ×10) and immunohistochemistry (IHC) for the IAV nucleocapsid protein (original magnification, ×10) are shown.

Next, the extents of HA cleavage for H3^p10 and H3^WT in the infected lungs were compared between tmprss2^WT and tmprss2^KO mice (n = 3). On days 4 and 6 p.i., lung lavage fluids were obtained from the mice and subjected to SDS-PAGE and immunoblotting assays by using a rabbit antiserum against H3 (Fig. 2A, C, D) or a goat antiserum against whole H3N2 virions (Fig. 2B). Consistent with our previous observation in the lungs of tmprss2^KO mice infected with H3^WT, where HA remained in the uncleaved HA₀ form (1), viral antigens were barely detectable by day 6 p.i., owing to the lack of multiple replication cycles (Fig. 2D). In contrast, H3^p10 HA in the lungs of tmprss2^KO mice was cleaved into the HA₁ and HA₂ subunits (Fig. 2A to C). However, the extents of cleavage estimated by the signal intensities of HA₀ and the cleaved HA subunits (HA₁ and HA₂) were lower in tmprss2^KO mice than in tmprss2^WT mice (Fig. 2E).

• Open in new tab
• Download powerpoint

FIG 2

Proteolytic activation of H3^p10 in tmprss2^WT and tmprss2^KO mice. (A to C) HA in lung lavage fluids from tmprss2^WT and tmprss2^KO mice infected with 1.7 × 10² PFU of H3^p10 was analyzed by SDS-PAGE and immunoblotting. Data from days 4 (A) and 6 (C) p.i. using a rabbit antiserum against H3 and data from day 4 p.i. (B) using a goat antiserum against whole H3N2 virions are shown. Each lane corresponds to data from an individual mouse (n = 3). (D) HA proteins in lung lavage fluids from tmprss2^WT and tmprss2^KO mice infected with 3.0 × 10³ PFU of H3^WT were analyzed by SDS-PAGE and immunoblotting. Data on day 6 p.i. using a rabbit antiserum against H3 are shown. Each lane corresponds to data from an individual mouse (n = 3). (E) Quantification of cleavage by measuring the chemiluminescent signals for the immunoblots shown in panels A and C. The bar graph shows the ratios of cleaved HA subunit (HA₁ and HA₂) signals to total HA (HA₁, HA₂, and HA₀) signals. Gray and black bars indicate data for cleavage in tmprss2^WT and tmprss2^KO mice, respectively. Data represent means ± standard deviations (SD). *, P < 0.01, significant difference based on a t test. (F) Virus titers in lung lavage fluids from tmprss2^WT and tmprss2^KO mice infected with 1.7 × 10² PFU of H3^p10. Lung lavage fluids on days 2, 4, and 6 p.i. were untreated (Trypsin −) or treated with trypsin (Trypsin +) and used for virus titration. Data represent means ± SD. (G) Plaque formation of H3^WT and H3^p10 in MDCK cells in the presence or absence of trypsin.

The extent of infectivity activation of the progeny virus in the lungs was analyzed, as reported previously (1). Briefly, lung lavage fluids from tmprss2^WT and tmprss2^KO mice infected with 1.7 × 10² PFU of H3^p10 on days 2, 4, and 6 p.i. were treated or untreated with trypsin in vitro, and the infectivity titers (PFU) were determined using MDCK cells. Our previous study demonstrated that H3^WT replicated poorly and showed low virus titers in tmprss2^KO mice (1). Only 1 to 2% of H3^WT virus particles were activated in the lungs. In contrast, H3^p10 replicated efficiently to high virus titers in the lungs of tmprss2^KO mice. Although the mean virus titers were severalfold lower in tmprss2^KO mice than in tmprss2^WT mice, a statistical analysis showed no significant difference (Fig. 2F). The percentages of activated virus particles in tmprss2^WT and tmprss2^KO mice were 23.0 to 30.2% and 2.4 to 10.3%, respectively (Fig. 2F). To validate the assay, it was confirmed that H3^p10 did not produce plaques in MDCK cells in the absence of trypsin (Fig. 2G).

The entire genome sequences of H3^WT and H3^p10 were analyzed by next-generation sequencing, as reported previously (5). Data for the RNA-seq reads of H3^WT and H3^p10 are available in the DDBJ Sequence Read Archive under accession number DRA002411. The data clearly demonstrated that the cytosine-to-adenine substitution at nucleotide position 86 in the HA gene was the most noticeable change throughout the viral genome between H3^WT and H3^p10 (Fig. 3; see also Fig. S1 to S8 in the supplemental material). This substitution was not detected (0.0%) in the H3^WT genome, while 97.6% of the H3^p10 genome possessed this substitution (Fig. 3; see also Fig. S4 in the supplemental material). The substitution was predicted to cause an asparagine-to-lysine substitution at amino acid position 8 (N8K [H3 numbering]) in the HA protein. As the asparagine residue constitutes a possible N-linked glycosylation motif [N-X-(S/T)] (6), this substitution was expected to cause loss of the oligosaccharide at the HA stalk. The predicted sites of oligosaccharide side chains in H3^WT are shown on the structure of A/Aichi/2/1968 (PDB 5HMG) (Fig. 4). The oligosaccharide site at position 8 was located more distantly from the cleavage site, compared with positions 22 and 38 (Fig. 4). However, under physiological conditions, the relative location of the N-terminal region possessing an oligosaccharide at position 8 may be different.

• Open in new tab
• Download powerpoint

FIG 3

Acquisition of a single amino acid substitution in the HA protein of H3^p10. (A, B) The read depths at nucleotide position 86 for H3^WT and H3^p10 were 44,124 and 914 reads, respectively. All 44,124 reads of the HA gene of H3^WT show a cytosine at nucleotide position 86, whereas 892 reads of the HA gene of H3^p10 show an adenine, and the remaining 22 reads show a cytosine at this position. An asparagine-to-lysine substitution at amino acid position 8 (N8K; H3 numbering) was predicted for this nonsynonymous substitution. The alignment was performed using BWA-SW software.

• Open in new tab
• Download powerpoint

FIG 4

Predicted positions of the oligosaccharides in H3^WT. Based on the structure of A/Aichi/2/1968 (H3N2) (PDB 5HMG), the predicted positions of the oligosaccharide modifications in H3^WT are shown. An HA molecule in a homotrimer is shown in blue. Circles with red dashed lines on the blue HA molecule indicate the predicted positions of oligosaccharide modifications. The N terminus of the HA protein and the cleavage site are indicated by red arrows.

No significant changes between H3^WT and H3^p10 were observed in the PB1, NP, NA, and M genome segments (see Fig. S2 and S5 to S7 in the supplemental material). On the other hand, the PB2, PA, and NS genome segments showed minor variations in the ratio of mixed nucleotides at several positions: 1091 and 1346 in PB2; 1893, 2014, and 2022 in PA; and 578 and 739 in NS (see Fig. S1, S3, and S8 in the supplemental material). For the PB2 and PA genes, the minor subpopulations found in H3^WT became even less abundant in H3^p10 (see Fig. S1 and S3 in the supplemental material). The reduction in minor variants in the PB2 and PA genes was very unlikely to account for the difference in HA cleavability. On the other hand, for the NS gene, the minor subpopulations found in H3^WT were enriched in H3^p10: 58.9% of the H3^p10 NS2 protein and 22.3% of the H3^WT NS2 protein were predicted to have a glycine-to-asparagine substitution at amino acid position 27 (G27N), and 28.8% of the H3^p10 NS2 protein and only 0.9% of the H3^WT NS2 protein were predicted to have a glutamine-to-lysine substitution at amino acid position 81 (Q81K) (see Fig. S8 in the supplemental material). Therefore, it was possible that the G27N and Q81K substitutions in the NS2 protein may partly contribute to the growth of H3^p10 in mice, although it was unlikely that these changes in NS2 contributed to the cleavability of HA.

The loss of the oligosaccharide was confirmed by SDS-PAGE and immunoblotting assays. HA polypeptides of H3^WT and H3^p10 grown in MDCK cells in the presence of trypsin were analyzed. The molecular sizes of HA₀ and HA₁ of H3^p10 were smaller than those of H3^WT (Fig. 5A and B).

• Open in new tab
• Download powerpoint

FIG 5

Loss of an oligosaccharide at the stalk of HA of H3^p10. (A, B) MDCK cells were infected with H3^WT and H3^p10, and the cells (A) and their culture supernatants (B) were subjected to SDS-PAGE and immunoblotting to detect HA.

Previous studies demonstrated that an H5 LP avian IAV acquired a highly pathogenic phenotype and trypsin-independent cleavability of HA by loss of the oligosaccharide at position 21 (H3 numbering) in the stalk region (7–10). The oligosaccharide, located in the vicinity of the cleavage site, likely prevented access of the multibasic cleavage site by ubiquitous host proteases. The hindrance could also be overcome by the addition of multiple basic amino acids to the cleavage site (11–13). A recent study showed a similar effect of the oligosaccharide on HA cleavage in an H9 LP avian IAV (14). The H9 HA contained an R-S-K-R cleavage motif, which was not cleavable by furin-like ubiquitous proteases when the oligosaccharide was present at the stalk region but became cleavable by these proteases when the oligosaccharide was removed (14).

Similar studies have been performed on human H3N2 IAV strains with a single arginine residue at the cleavage site (6, 15, 16). Ohuchi et al. (15) and Kawaoka (16) tested the effects of removing the oligosaccharide at position 22 (H3 numbering), located in the vicinity of the cleavage site (17, 18). Simple removal of the oligosaccharide did not render H3 HA cleavability in cultured cells. However, the oligosaccharide removal reduced the number of inserted arginine residues required to acquire trypsin-independent cleavability of the HA (15).

Rott et al. (6) reported that the HA of the human H3N2 X-31 strain acquired trypsin-independent cleavability in MDCK cells. The oligosaccharide loss at position 8 caused by the N8K substitution contributed to the increased cleavability in MDCK cells (6). The present study showed that although H3^p10 had acquired the same N8K substitution as X-31, it failed to produce plaques in MDCK cells in the absence of trypsin (Fig. 2G). The difference in phenotypes between X-31 and H3^p10 may be caused by heterogeneity within MDCK cells (19). While the oligosaccharide at position 22 (H3 numbering) is conserved among many HA subtypes, the sugar chain at position 8 is unique to the H3 subtype (20). The great majority of H3 IAV field isolates conserve the oligosaccharide at position 8, suggesting that most wild-type H3N2 viruses hold limited cleavability by TMPRSS2.

The present study demonstrated that H3^WT was not activated proteolytically in tmprss2^KO mouse lungs and that H3^p10, which had lost the oligosaccharide side chain at HA position 8, underwent cleavage activation in tmprss2^KO mouse lungs. H3^p10 acquired the cleavability by an alternative HA activation mechanism/protease(s). It remains unclear whether the H3N2 strains used in other studies (2, 3) possessed the oligosaccharide at position 8. Since loss of the oligosaccharide could occur even in MDCK cells (6), the different passage histories of H3N2 viruses might result in different cleavability in vivo in the lungs. The differential cleavability of H3N2 viruses, possibly caused by loss of the sugar chain, may explain the discrepancy in the TMPRSS2 dependency of H3N2 IAV strains in previous studies (1–3).

In addition to TMPRSS2, many proteases, such as tryptase Clara (21), human airway trypsin-like protease (22), TMPRSS4 (23), TMPRSS13 (24), mosaic seine protease large-form (24), and matriptase (25, 26), have been shown to activate various IAV strains at different levels (27). Activation by these proteases may also be modulated by the oligosaccharide at position 8. It is of great interest to clarify the advantage for H3 IAV strains to conserve this oligosaccharide in nature.