Novel Avian-Origin Influenza A (H7N9) Virus Attachment to the Respiratory Tract of Five Animal Models

Abstract

We determined the pattern of attachment of the avian-origin H7N9 influenza viruses A/Anhui/1/2013 and A/Shanghai/1/2013 to the respiratory tract in ferrets, macaques, mice, pigs, and guinea pigs and compared it to that in humans. The H7N9 attachment pattern in macaques, mice, and to a lesser extent pigs and guinea pigs resembled that in humans more closely than the attachment pattern in ferrets. This information contributes to our knowledge of the different animal models for influenza.

TEXT

In March 2013, a novel avian-origin influenza A virus of the H7N9 subtype caused severe disease in humans (1, 2) and infected over 142 humans, with 45 deaths (3) to date. So far, limited but unsustained human-to-human transmission has been reported (4–6). Recently, we showed that this novel avian-origin H7N9 virus attached abundantly to epithelial cells of both the upper respiratory tract (URT) and lower respiratory tract (LRT) in humans. Unlike other avian influenza viruses, both H7N9 virus strains tested (A/Shanghai/1/2013 and A/Anhui/1/2013) attached to ciliated epithelial cells of the nasal concha, trachea, and bronchus, which is more characteristic of human influenza viruses. Furthermore, similar to other avian influenza viruses, H7N9 viruses attached to alveolar epithelial cells, alveolar macrophages, and Clara cells in the terminal bronchioli (7).

Since the emergence of this H7N9 virus, experimental infections in ferrets (Mustela putorius furo), cynomolgus macaques (Macaca fascicularis), laboratory mice (Mus musculus domesticus), domestic pigs (Sus scrofa domesticus), and guinea pigs (Cavia porcellus) have been performed to gain more insight into its pathogenesis and transmissibility (8–16). However, the pattern of attachment throughout the respiratory tract, which is known to be an important factor for both pathogenicity and transmissibility (17–19), is not known for H7N9 virus in these animal species. Previously, we showed that the attachment pattern of human and avian influenza viruses differs largely among animal species (17, 19). Therefore, we determined the attachment pattern of two H7N9 virus strains—A/Shanghai/1/2013 (H7 Shanghai) and A/Anhui/1/2013 (H7 Anhui), the latter containing a Q226L (H3 numbering) substitution in the hemagglutinin (1)—in the respiratory tracts of these five animal species and compared it to that in humans (7). Furthermore, we compared the attachment pattern of H7N9 virus with that of two human influenza viruses, A/Netherlands/213/03 (H3N2) and A/Netherlands/602/09 (pandemic H1N1), and two highly pathogenic avian influenza (HPAI) viruses, A/Vietnam/1194/04 (H5N1) and A/Netherlands/219/03 (H7N7). All viruses were concentrated, inactivated, and labeled with fluorescein isothiocyanate (FITC) as described previously (17, 19). From the respiratory tracts of ferrets (n = 3), macaques (n = 2), C57BL/6 mice (n = 3), domestic pigs (n = 3), and guinea pigs (n = 2), tissue sections of the nasal concha, trachea, bronchus, bronchiolus, and alveolus were stained with FITC-labeled viruses by virus histochemistry as described previously (17, 19). The percentage of cells to which the virus attached was blindly scored in an ordinal manner, as follows: −, attachment to <1% of epithelial cells; ±, attachment to ≥1 and <10% of epithelial cells; +, attachment to ≥10% and <50% of epithelial cells; and ++, attachment to ≥50% of epithelial cells. Finally, the median score was determined for each species at the different anatomical sites. Bronchus was not included for mice or guinea pigs, as all airways present in lung tissues were not lined by cartilage and were therefore considered bronchioli.

In all species included in this study, both H7 Shanghai and H7 Anhui viruses attached to the epithelium of the URT and LRT; however, the degree of attachment among the different anatomical sites and animal models varied (Table 1; Fig. 1). In general, the attachment pattern did not resemble that of either the human influenza or the HPAI viruses (Table 1). The attachment pattern of H7 Shanghai and H7 Anhui in the macaque and mouse, and to a lesser extent the pig and guinea pig, respiratory tracts resembled the attachment pattern observed in humans (Table 1; Fig. 1). In these five species, both H7N9 viruses attached to ciliated epithelial cells in the nasal concha, trachea, and bronchus, to ciliated and nonciliated cells in the bronchioles, and to both type I and type II pneumocytes in the alveoli. In contrast, in the ferret respiratory tract, both H7 viruses attached to <10% of ciliated epithelial cells in the trachea, bronchus, and bronchioles, whereas in the human respiratory tract, H7 Shanghai and H7 Anhui viruses attached to ≥10% of the ciliated epithelial cells in the trachea and to ≥50% of ciliated epithelial cells in the bronchus and bronchioles.

TABLE 1.

Attachment of influenza viruses to the respiratory tracts of humans and animal models^a

Virus	Species	Nasal concha		Trachea		Bronchus		Bronchiole		Alveolus
		Score	Cell type	Score	Cell type	Score	Cell type	Score	Cell type(s)	Score	Cell type(s)b
H3N2	Human	++	Cil	++c	Cil	++c	Cil	++	Cil	+	I
	Ferret	++	Cil	+	Cil	+	Cil	+	Cil and Clara	+	I
	Macaque	±	Cil	−	ND	−	ND	−	ND	±	I
	Mouse	−	ND	−	ND	ND	ND	−	ND	−	ND
	Pig	±	Cil	+c	Cil	++	Cil	++	Cil and Clara	+	I
	Guinea pig	±	Cil	−	ND	ND	ND	−	ND	±	I
pH1N1	Human	+	Cil	++c	Cil	++c	Cil	+	Cil	+	I
	Ferret	++	Cil	±	Cil	±	Cil	±	Clara	++d	I and II
	Macaque	−	ND	−	ND	−	ND	±	Cil	+d	I
	Mouse	±	Cil	−	ND	ND	ND	±	Cil	++	I and II
	Pig	±	Cil	+c	Cil	++c	Cil	++	Cil and Clara	+	I
	Guinea pig	++	Cil	−	ND	ND	ND	±	Clara	+	I
H7N9 (Shanghai)	Human	+	Cil	+	Cil	++c	Cil	++	Cil and Clara	++d	I and II
	Ferret	+	Cil	±	Cil	±	Cil	±	Clara	++d	I and II
	Macaque	+	Cil	++	Cil	++	Cil	++	Cil and Clara	++d	I and II
	Mouse	++	Cil	++	Cil	ND	ND	++	Cil and Clara	++d	I and II
	Pig	±	Cil	+c	Cil	±	Cil	+	Cil and Clara	++	I and II
	Guinea pig	+	Cil	++	Cil	ND	ND	+	Cil and Clara	++d	I and II
H7N9 (Anhui)	Human	++	Cil	+	Cil	++c	Cil	++	Cil and Clara	++d	I and II
	Ferret	++	Cil	±	Cil	±	Cil	±	Clara	++d	I and II
	Macaque	±	Cil	+	Cil	+	Cil	++	Cil and Clara	++d	I and II
	Mouse	++	Cil	++	Cil	ND	ND	++	Cil and Clara	++d	I and II
	Pig	±	Cil	+	Cil	+	Cil	++	Cil and Clara	++d	I and II
	Guinea pig	+	Cil	++	Cil	ND	ND	+	Clara	++d	I and II
H5N1	Human	−	ND	±	ND	±	ND	+	Clara	+d	II
	Ferret	−	ND	−	ND	−	ND	−	ND	+	II
	Macaque	−	ND	±	Cil	+	Cil	+	Cil and Clara	++	I and II
	Mouse	++	Cil	++	Cil	ND	ND	++	Cil and Clara	+	II
	Pig	−	ND	−	ND	±	Cil	±	Cil and Clara	±	ND
	Guinea pig	−	ND	++	Cil	ND	ND	+	Clara	+d	II
H7N7	Human	−	ND	±	ND	±c	ND	+	Clara	+d	II
	Ferret	±	Cil	−	ND	−	ND	±	Clara	+	II
	Macaque	+	Cil	++	Cil	++	Cil	+	Cil	+	I and II
	Mouse	++	Cil	++	Cil	ND	ND	+	Cil and Clara	+	II
	Pig	−	ND	±	Cil	±	Cil	+	Cil and Clara	+	II
	Guinea pig	−	ND	+	Cil	ND	ND	−	ND	+d	II

^aScores are medians. Attachment of influenza viruses was ranked on an ordinal scale as follows: −, <1% cells positive; ±, ≥1 and <10% of cells positive; +, ≥10 and <50% of cells positive; ++, ≥50% of cells positive. Cil, ciliated epithelial cells; ND, not determined.

^bType I and II pneumocytes.

^cIntracellular attachment to goblet cells.

^dAdditional virus attachment to alveolar macrophages.

FIG 1.

Attachment of novel avian-origin influenza H7N9 Anhui to different parts of the respiratory tract in humans and five animal models.

Overall, there were three interesting observations regarding the attachment pattern of both H7N9 viruses in the respiratory tract of the different animal models. First, there was limited variation between the attachment patterns of H7 Shanghai and H7 Anhui in all animal models studied, despite the Q226L amino acid substitution in H7 Anhui. It is known that both viruses recognize human and avian influenza virus receptors, α-2,6- and α-2,3-linked sialic acid residues, respectively (8, 13, 20–23). In contrast to H7 viruses, the Q226L amino acid substitution is responsible for a switch in the receptor binding preference and attachment pattern in H5N1 viruses (24–26). In addition, unlike H7 Anhui, H7 Shanghai contains a serine instead of alanine at position 138 (H3 numbering) which is known to induce a slight increase in binding to α-2,6-linked sialic acids on an H5 backbone (27, 28). This suggests that multiple sites within the hemagglutinin can contribute to the attachment of influenza viruses, depending on the hemagglutinin subtype.

Second, the attachment pattern of H7 Shanghai and H7 Anhui in the macaque, mouse, and to a lesser extent pig and guinea pig respiratory tracts resembled the attachment pattern in the human respiratory tract most closely, with attachment to epithelial cells of the URT and LRT. In vivo, in vitro, and ex vivo studies confirmed that the observed attachment of H7N9 virus results in a productive infection in human bronchi (29), in macaque nasal concha, trachea, and lungs (8), in pig nasal concha, trachea, bronchi, and lungs (12, 30), and in mouse bronchi and lung parenchyma (8, 11, 13, 14). Taken together, these data suggest that the tropism of H7N9 viruses within the mammalian respiratory tract is more widespread than the tropism of, for example, HPAI H5N1 virus in humans, ferrets, macaques, pigs, and guinea pigs (31–33).

Finally, the attachment pattern of H7 Shanghai and H7 Anhui in the ferret respiratory tract showed some remarkable differences from the attachment pattern in the human respiratory tract. This discrepancy between the attachment patterns of H7N9 viruses in ferrets and humans was unexpected, because in general the attachment patterns of human and avian influenza viruses in the ferret respiratory tract resemble those observed in humans (17, 19). In addition, the ferret is commonly used to study the pathogenesis, transmissibility, and intervention strategies of this H7N9 virus. To determine whether the sparse attachment of H7 Anhui in the ferret trachea would result in lower numbers of infected cells and inefficient virus production, ferret tracheal rings from four donors were infected ex vivo with 10⁷ 50% tissue culture infective doses (TCID₅₀)/ml H7N9 virus (A/Anhui/1/2013) and incubated at 37°C in 95% O₂ and 5% CO₂. At 1, 6, 24, and 48 h postinfection (hpi), supernatants were collected to determine virus titers, and at 24 and 48 hpi, tissues were collected for histological analysis. For comparison, replicate tracheal rings from each donor ferret were infected with H3N2 virus (A/Netherlands/213/03), which attaches moderately to ferret tracheal epithelial cells, or with H5N1 virus (A/Vietnam/1194/04), which, like H7N9 Anhui virus, attaches sparsely to ferret tracheal epithelial cells. In these ex vivo cultures, H7N9 and H5N1 viruses had a trend toward lower virus titers than H3N2 virus, although these differences were not statistically significant due to donor variations (Fig. 2A). In addition, H7N9 and H5N1 viruses infected significantly fewer epithelial cells (P < 0.05, Kruskal-Wallis with Dunn's multiple-comparison test), as determined by immunohistochemistry for influenza virus nucleoprotein as described previously (34), at both 24 and 48 hpi (Fig. 2B and C). Additionally, lung explants containing alveolar tissue were infected ex vivo with H7N9 Anhui virus, which attach to both type I and II pneumocytes in these tissues. For comparison, replicate lung explants were infected with either HPAI H5N1, which attaches predominantly to type II pneumocytes, or H3N2, which attaches predominantly to type I pneumocytes. In the supernatant of the ex vivo-infected lung explants, H7N9 Anhui virus titers were approximately 100× higher than titers of H5N1 and H3N2 viruses at 48 hpi, although these differences were not statistically significant (Fig. 3A). When we determined the number of virus-infected cells per 10 mm² in the ex vivo-infected lung pieces, H7N9 Anhui virus infected significantly more epithelial cells than H3N2 virus 24 hpi (P < 0.05). At 48 hpi, there was a trend for both H5N1 virus and H7 Anhui to infect more cells than H3N2 virus, although these differences were not statistically significant (Fig. 3B and C).

FIG 2.

Ex vivo infection of ferret tracheal explants with H3N2, H7N9 Anhui, or H5N1 virus. (A) Replication kinetics of all three viruses in ferret tracheal explants. Geometric mean titers were calculated from 4 independent experiment; error bars indicate standard deviations. (B) Calculated averages of influenza virus nucleoprotein-positive cells observed per ×400 high-power field (HPF), in ferret tracheal explants from 4 independent experiments. Error bars indicate standard deviations; asterisks indicate statistical differences with P values of <0.05. (C) Representation of influenza virus nucleoprotein-positive cells in the ferret trachea 24 h postinfection at an original magnification of ×400.

FIG 3.

Ex vivo infection of ferret lung explants with H3N2, H7N9 Anhui, or H5N1 virus. (A) Replication kinetics of all three viruses in ferret lung explants. Geometric mean titers were calculated from 4 independent experiments; error bars indicate standard deviations. (B) Calculated averages of influenza-virus-nucleoprotein-positive cells observed in ferret lung explants from 4 independent experiments. Error bars indicate standard deviations; asterisks indicate statistical differences with P values of <0.05. (C) Representation of influenza virus nucleoprotein-positive cells in the ferret lung, 24 h postinfection, at an original magnification of ×400.

Together with the virus attachment study, these ex vivo infection experiments show that limited attachment of H7N9 Anhui to ferret tracheal epithelial cells correlated with inefficient infection. The impact of limited virus attachment in the airways of ferrets on the pathogenicity and transmissibility remains to be elucidated. Airborne transmission of H7N9 viruses in ferrets is relatively inefficient (8, 9, 12), which might reflect limited attachment to and replication in tracheal and bronchial epithelium. However, transmission of H7N9 viruses in humans is limited and therefore considered inefficient (4, 5), despite the ability of this virus to attach moderately to cells in the human trachea and bronchus.

Taken together, our data show that H7 Shanghai and H7 Anhui attached to epithelial cells in both the URT and LRT of all animal species tested. Unexpectedly, the pattern of attachment to the respiratory tracts of macaque, mouse, and to a lesser extent pig and guinea pig resembled that of humans more closely than that observed in ferrets. Although attachment is not the only determinant of the cell tropism of an influenza virus, it is important to take these observations into account when designing experimental H7N9 virus infections to study and evaluate transmission, pathogenesis, and intervention studies.