|Home | About | Journals | Submit | Contact Us | Français|
Avian H7 influenza viruses have been responsible for poultry outbreaks worldwide and have resulted in numerous cases of human infection in recent years. The high rate of conjunctivitis associated with avian H7 subtype virus infections may represent a portal of entry for avian influenza viruses and highlights the need to better understand the apparent ocular tropism observed in humans. To study this, mice were inoculated by the ocular route with viruses of multiple subtypes and degrees of virulence. We found that in contrast to human (H3N2 and H1N1) viruses, H7N7 viruses isolated from The Netherlands in 2003 and H7N3 viruses isolated from British Columbia, Canada, in 2004, two subtypes that were highly virulent for poultry, replicated to a significant titer in the mouse eye. Remarkably, an H7N7 virus, as well as some avian H5N1 viruses, spread systemically following ocular inoculation, including to the brain, resulting in morbidity and mortality of mice. This correlated with efficient replication of highly pathogenic H7 and H5 subtypes in murine corneal epithelial sheets (ex vivo) and primary human corneal epithelial cells (in vitro). Influenza viruses were labeled to identify the virus attachment site in the mouse cornea. Although we found abundant H7 virus attachment to corneal epithelial tissue, this did not account for the differences in virus replication as multiple subtypes were able to attach to these cells. These findings demonstrate that avian influenza viruses within H7 and H5 subtypes are capable of using the eye as a portal of entry.
Highly pathogenic avian influenza (HPAI) H5N1 viruses, which have resulted in over 420 documented cases of human infection to date, have generally caused acute, often severe and fatal, respiratory illness (1, 50). While conjunctivitis following infection with H5N1 or human influenza viruses has been rare, most human infections associated with H7 subtype viruses have resulted in ocular and not respiratory disease (1, 9, 37, 38). Infrequent reports of human conjunctivitis infection following exposure to H7 influenza viruses date from 1977, predominantly resulting from laboratory or occupational exposure (21, 40, 48). However, in The Netherlands in 2003, more than 80 human infections with H7N7 influenza virus occurred among poultry farmers and cullers amid widespread outbreaks of HPAI in domestic poultry; the majority of these human infections resulted in conjunctivitis (14, 20). Additionally, conjunctivitis was documented in the two human infections resulting from an H7N3 outbreak in British Columbia, Canada, in 2004, as well as in H7N3- and H7N2-infected individuals in the United Kingdom in 2006 and 2007, respectively (13, 18, 29, 46, 51). The properties that contribute to an apparent ocular tropism of some influenza viruses are currently not well understood (30).
Host cell glycoproteins bearing sialic acids (SAs) are the cellular receptors for influenza viruses and can be found on epithelial cells within both the human respiratory tract and ocular tissue (26, 31, 41). Both respiratory and ocular tissues additionally secrete sialylated mucins that function in pathogen defense and protection of the epithelial surface (5, 11, 22). Within the upper respiratory tract, α2-6-linked SAs (the preferred receptor for human influenza viruses) predominate on epithelial cells (26). While α2-3-linked SAs are also present to a lesser degree on respiratory epithelial cells, this linkage is more abundantly expressed on secreted mucins (2). In contrast, α2-3-linked SAs (the preferred receptor for avian influenza viruses) are found on corneal and conjunctival epithelial cells of the human eye (31, 41), while secreted ocular mucins are abundantly composed of α2-6 SAs (5). It has been suggested that avian influenza viruses are more suited to infect the ocular surface due to their general α2-3-linked SA binding preference, but this has not been demonstrated experimentally (30).
The mouse model has been used previously to study the role of ocular exposure to respiratory viruses (6, 39). In mice, ocular inoculation with an H3N2 influenza virus resulted in virus replication in nasal turbinates and lung (39), whereas ocular infection with respiratory syncytial virus (RSV) resulted in detectable virus titers in the eye and lung (6). These studies have revealed that respiratory viruses are not limited to the ocular area following inoculation at this site. However, the ability of influenza viruses to replicate specifically within ocular tissue has not been examined.
Despite repeated instances of conjunctivitis associated with H7 subtype infections in humans, the reasons for this apparent ocular tropism have not been studied extensively. Here, we present a murine model to study the ability of human and avian influenza viruses to cause disease by the ocular route. We found that highly pathogenic H7 and H5 influenza viruses were capable of causing a systemic and lethal infection in mice following ocular inoculation. These highly pathogenic viruses, unlike human H3N2 and H1N1 viruses, replicated to significant titers in the mouse corneal epithelium and primary human corneal epithelial cells (HCEpiCs). Identification of viruses well suited to infecting the ocular surface is the first step in better understanding the ability of influenza viruses of multiple subtypes to use this tissue as a portal of entry.
Influenza A viruses were propagated in the allantoic cavity of 10-day-old embryonated hen's eggs at 37°C (H5N1, H7N7, and H7N3 viruses) or 35°C (H7N2 and H3N2 viruses) for 26 to 48 h, with the exception of H1N1 viruses, which were grown in Madin-Darby canine kidney (MDCK) cells as previously described (42). Allantoic fluid pooled from multiple eggs was clarified by centrifugation and frozen in aliquots at −70°C. The 50% egg infectious dose (EID50) titer for each virus stock was calculated by the method of Reed and Muench (32), following serial titration in eggs. Viruses were additionally tested by plaque assay in MDCK cells for determination of PFU titer. Concentrated virus stocks for virus immunohistochemistry were prepared as previously described (24). All experiments with HPAI viruses were conducted under biosafety level 3 containment, including enhancements required by the U.S. Department of Agriculture and the Select Agent Program (33).
Female BALB/c mice (Jackson Laboratories) 6 to 8 weeks of age were deeply anesthetized with 2,2,2-tribromoethanol in tert-amyl alcohol (Avertin; Sigma-Aldrich, St. Louis, MO) prior to ocular inoculation. The right eye of each mouse was lightly scarified by three twists of a 2-mm corneal trephine (Katena Products, Denville, NJ), followed by 106 EID50 (H7, H5, or H3 subtype) or 106 PFU (H1 subtype) of virus in 5 μl diluted in phosphate-buffered saline (PBS) dropped onto the corneal surface and massaged in with the eyelids as described previously (4). Five to 10 mice per virus were monitored daily for 14 days postinoculation (p.i.) for morbidity, as measured by weight loss, and mortality. The corneal opacity of mouse eyes was examined on alternate days with a dissecting biomicroscope as described previously (43). Any mouse that lost greater than 25% of its preinfection body weight was euthanatized. Replication and systemic spread of each virus were determined by harvesting the right eye, nose, lung, and brain of mice (n = 3 to 6) on indicated days p.i. Tissues were titrated in eggs as described previously (25). All clarified homogenates were titrated starting at a 1:2 dilution (limit of detection, 100.8 EID50/ml). Statistical significance was determined using the standard error of the mean.
The isolation of corneal buttons and corneal epithelial sheets from BALB/c mice was performed as described previously (16, 44). To infect, five corneal buttons or corneal epithelial sheets were placed in 1 well of a 48-well plate and were infected with a total of 5 × 103 PFU of each virus for 1 h at 37°C, washed at least three times with medium, and brought to a final volume of 1 ml Dulbecco's modified Eagle's medium with penicillin-streptomycin, 0.3% bovine serum albumin (Gibco, Carlsbad, CA), and 300 μg/liter N-p-tosyl-l-phenylalanine chloromethyl ketone (TPCK)-treated trypsin (Sigma-Aldrich). Culture supernatant was removed at indicated times postinfection and titrated for the presence of virus by plaque assay as described previously (53).
Primary HCEpiCs were obtained from ScienCell (San Diego, CA) at passage 1 and were grown on poly-l-lysine-coated flasks per the manufacturer's instructions. Cells were grown to confluence in six-well plates in serum-free keratinocyte medium (ScienCell). HCEpiC monolayers were infected with virus at a multiplicity of infection (MOI) of 0.01 for 1 h and then washed at least three times with medium before being brought to a final volume of 1 ml medium with 300 μg/liter TPCK-trypsin. Culture supernatant was removed at indicated times p.i. and titrated for the presence of virus by plaque assay.
For determination of expression of surface sialoligosaccharides, uninfected mouse corneal epithelial sheets or HCEpiCs grown on glass coverslips until confluent were fixed in 3.7% formaldehyde for 10 min at 37°C. Corneal epithelial sheets were blocked with an avidin-biotin blocking kit (Vector Laboratories, Burlingame, CA) per the manufacturer's instructions, and HCEpiCs were blocked with 3% BSA in PBS for 30 min. Tissues and cells were then sequentially incubated with biotinylated Maackia amurensis lectin I (MAAI; 20 μg/ml) or II (MAAII; 40 μg/ml), which bind α2-3-linked SAs, or biotinylated Sambucus nigra lectin (SNA; 20 μg/ml), which binds α2-6-linked SAs, for 1 h at room temperature (RT), washed with PBS, followed by the addition of fluorescein-conjugated avidin D (Vector Laboratories) for 30 min at RT. Additional HCEpiCs on glass coverslips were treated with 25 mU/ml of Vibrio cholerae neuraminidase (Roche, Indianapolis, IN) for 60 min prior to fixation to confirm the specificity of lectin binding as previously described (53). Tissues and cells were further counterstained with 4′,6′-diamidino-2-phenylindole (DAPI) to visualize nuclei and then mounted on glass slides with Vectashield (Vector Laboratories) and examined under a Zeiss Axioskop 2 fluorescence microscope.
Labeling of virus was modified and adapted from previously published methods (11, 47, 52). Concentrated virus was labeled using the FluoroTag fluorescein isothiocyanate (FITC) conjugation kit (Sigma-Aldrich) following the manufacturer's protocols. Briefly, an equal volume of 1 mg/ml inactivated virus was gently mixed with 0.1 mg/ml of FITC in 0.1 M sodium carbonate-bicarbonate buffer (pH 9.0) for 2 h at RT. This reaction mixture was then applied to a Sephadex G-25 M gel filtration column to separate labeled virus from unbound FITC. The labeled virus fractions yielding more than an A280 of 0.6 were pooled and stored in 0.1% sodium azide at 4°C until use.
Slides of paraffin-embedded and formalin-fixed normal BALB/c mouse lung and ocular tissue sections were purchased from Comparative BioSciences, Inc. (Sunnyvale, CA), or prepared in house. Slides were heated at 60°C for 30 min to melt paraffin and then deparaffinized with xylene and rehydrated through an ethanol series to deionized water (dI H2O). Next, tissues were digested in 200 μl of 0.1 mg/ml proteinase K (Applied Biosystems/Ambion, Inc., Austin, TX) at 37°C for 30 min and then rinsed in dI H2O. Tissues were blocked with normal goat serum for 20 min at RT to minimize nonspecific staining or binding, and rinsed.
Stock FITC-labeled virus was initially diluted 1:10 in Tris-buffered saline with fish gelatin (BioFX Laboratories, Inc., Owings Mills, MD) and then passed through a 0.45-μm-pore Millipore Millex-HV syringe filter unit (Fisher Scientific, Pittsburg, PA). Each virus was run through a series of dose-response titrations to determine the optimum staining dilution. Hemagglutination titers were determined on the final use dilutions of each virus to ensure standardization of each virus concentration at 8 hemagglutination (HA) units per virus. Tissues were incubated overnight at 4°C with each dilution of virus before rinsing in PBS. Tissues were treated with 3% hydrogen peroxide for 10 min at RT to remove endogenous peroxidase activity and then rinsed in PBS.
To reveal bound FITC-labeled virus, tissue sections were incubated with a peroxidase-labeled rabbit anti-FITC antibody (Dako North America, Carpinteria, CA) overnight at 4°C, followed by rinsing in PBS. Tissues were incubated with 3-amino-9-ethylcarbazole (AEC; Dako) containing hydrogen peroxide for 10 min at RT to reveal the presence of peroxidase. Sections were washed in dI H2O and thereafter counterstained with Mayer's hematoxylin for 5 min at RT followed by washes in dI H2O. Tissues were coverslipped in an aqueous mounting medium (Vector Laboratories, Burlingame, CA) and examined by light microscopy. Bound virus was indicated by the presence of a reddish precipitate on a bluish background of tissue. Tissues were treated with 1 U/ml neuraminidase (Roche Applied Science, Indianapolis, IN) for 1 h at 37°C to remove the presence of SA receptors to serve as a negative control for virus binding. Stock FITC-labeled virus was passed through a 0.22-μm-pore filter to serve as a negative control for this staining protocol. The images shown are representative of at least three independent slides examined.
The BALB/c mouse model has been used extensively to study the pathogenicity of influenza viruses administered by the intranasal route (15, 24, 25). We modified this animal model to assess the pathogenicity of influenza viruses of multiple subtypes following ocular inoculation (Table (Table1).1). The right eye of each mouse was lightly scarified, and 106 EID50 of the indicated virus was deposited onto the corneal epithelial surface and massaged into the eye with the eyelid. The Eurasian lineage H7N7 viruses NL/219 (from a fatal case), NL/230 (from a conjunctivitis case), and Ck/NL/1 were isolated from an HPAI outbreak in The Netherlands in 2003. Following ocular inoculation, NL/219-infected mice showed the greatest morbidity and mortality, with 30% of mice succumbing to virus infection (Fig. 1A and B); NL/230 and Ck/NL/1-infected mice did not exhibit morbidity p.i. All H7N7 viruses tested replicated to a detectable titer in the eye and nose (Fig. (Fig.2A).2A). NL/219 virus was also present in the lungs on days 3 and 6 p.i. (P < 0.05). By day 9 p.i., NL/219 virus in lung tissues reached a mean titer of 6.2 ± 1.5 log10 EID50/ml and infectious virus was also detected in 4 of 4 brain tissues (mean titer of 2.7 ± 0.9 log10 EID50/ml; data not shown). NL/219 virus was further detected in the eye and nose day 9 p.i. in 2 of 4 mice (1.6 ± 0.5 and 1.5 ± 0.4 log10 EID50/ml, respectively).
The North American H7N3 viruses Can/504 and Can/444 were isolated from an HPAI virus outbreak in British Columbia, Canada, in 2004 (Table (Table1).1). The low-pathogenicity avian influenza (LPAI) H7N2 virus NY/107 was isolated from an individual with respiratory illness in 2003 (8). These viruses did not cause pronounced morbidity or mortality following ocular inoculation in the mouse (data not shown), similar to what is observed following the traditional intranasal route of inoculation (4). However, both LPAI and HPAI H7N3 viruses tested replicated to substantial titer in the mouse eye by day 6 p.i. (Fig. (Fig.2B)2B) and also were detected in the nose tissue. Can/504 virus was additionally present in the lung on day 6 p.i., albeit at a low mean titer of 1.5 ± 0.4 log10 EID50/ml. In contrast to the H7N3 viruses, NY/107 virus was not detected in any tissue at any time point following ocular inoculation (Fig. (Fig.2B2B).
We next examined the pathogenesis of selected HPAI H5N1 viruses when introduced by the ocular route. The HPAI viruses HK/483, Thai/16, and VN/1203 cause lethal disease in mice following intranasal infection (Table (Table1)1) (24, 25). Strikingly, inoculation by the ocular route with either HK/483 or Thai/16 virus resulted in weight loss and 60% mortality of mice (Fig. 1C and D) in comparison with VN/1203 virus, which did not cause severe disease. HK/483 and Thai/16 viruses were both detected in the eye by day 6 p.i.; however, substantial H5N1 titers were only observed in the lung and nose tissues of Thai/16-infected mice (Fig. (Fig.2C).2C). Thai/16 virus was additionally detected in 2 of 4 mouse lungs (mean titer of 2.8 ± 1.2 log10 EID50/ml) and 4 of 4 mouse brains (mean titer of 3.5 ± 0.9 log10 EID50/ml) on day 9 p.i. (data not shown). VN/1203 virus was not detected in the eye or lung following ocular inoculation and was only detected at low titer in the nose (mean titers of 1.4 ± 0.3 and 1.2 ± 0.3 log10 EID50/ml on days 3 and 6 p.i., respectively) (Fig. (Fig.2C).2C). Inoculation by the ocular route with the HPAI H5N1 viruses A/Hong Kong/486/97 (HK/486) and A/Hong Kong/213/03 (HK/213) did not result in morbidity or mortality of mice, and virus was not found at detectable titer in any tissue tested with these viruses (data not shown).
For comparison, we examined the ability of H3N2 and H1N1 viruses to infect mice following ocular inoculation. Two early human H3N2 viruses, Aichi/68 and Mem/72, that replicate efficiently in the lungs of mice following intranasal inoculation were chosen for this study (19; data not shown). Additionally, the H1N1 1918 pandemic influenza virus (possessing the A/South Carolina/1/18 [SC18] HA) which results in a lethal infection following intranasal administration (42), and a modified 1918 virus with avian receptor specificity, AV18 (45), were also evaluated in this system. Mice inoculated with any of these viruses by the ocular route did not exhibit morbidity or mortality (Table (Table11 and data not shown). Furthermore, these viruses failed to replicate to detectable titer in the eye or nose of mice and were found only at low titers (<2 log10 EID50/ml) in the lung on day 6 p.i. (Fig. (Fig.2D;2D; Table Table11).
In summary, we found that H7 viruses were detected more frequently and replicated to higher titers in the eye following ocular inoculation compared with the H5N1 viruses, which replicated to more modest titers in the eye and to higher titers in the respiratory tract of mice. Three HPAI viruses, NL/219, HK/483, and Thai/16, were found to be capable of causing lethal disease in mice when introduced by the ocular route. In contrast, human influenza viruses did not replicate to a detectable titer in the eye or nose and were only detected at low levels in the lung.
Influenza viruses bind to terminal SA prior to cell entry and replication; however, the composition of SA on the mouse cornea had not been previously determined. Biotinylated lectins specific for α2-6 and α2-3 SAs were used to stain mouse corneal epithelial sheets. Based on staining with both SNA and MAAI and II lectins, the mouse corneal epithelial sheet was found to express both α2-6- and α2-3-linked SAs (Fig. (Fig.3A).3A). This expression pattern differs from human corneal tissue, which predominantly expresses α2-3-linked SAs (41), as shown here with primary HCEpiCs (Fig. (Fig.3B3B).
The occurrence of conjunctivitis following human infection with H7 viruses led us to examine if the ability to bind ocular tissue is restricted to selected influenza virus subtypes. Influenza viruses of multiple subtypes were FITC labeled and exposed to BALB/c corneal or lung tissue overnight. As expected, NL/219 virus, which demonstrated predominant attachment to cells in the mouse alveoli (Fig. (Fig.4B),4B), also displayed virus attachment to mouse corneal epithelium tissues, as did Can/504 virus (Fig. (Fig.4A).4A). H5N1 (Thai/16) virus attachment to mouse corneal epithelium was also detected. Binding was observed with both the reconstructed 1918 SC18 virus and the mutant 1918 AV18 virus with altered receptor binding specificity, demonstrating the ability of viruses with either α2-6 or α2-3 SA binding preference, respectively, to bind to mouse ocular tissue. In contrast to NL/219 virus, SC18 virus attachment was less abundant in the alveoli and more abundant in the bronchi and bronchioles (Fig. (Fig.4B).4B). All other influenza viruses tested did bind to mouse lung tissue (not shown). These results indicate that, in the mouse model, the ability of influenza viruses to bind ocular tissue is not predictive of their ability to mount a productive infection in vivo.
To better identify the capacity of influenza viruses to replicate in ocular tissue, whole corneas (containing epithelial, stromal, and endothelial layers) were excised from naïve BALB/c mouse eyes and infected ex vivo with influenza viruses of multiple subtypes (Fig. (Fig.5A).5A). Culture supernatants were collected at the indicated times p.i., and titers were determined for the presence of infectious virus. NL/219 virus was found at significantly higher titers 24 to 96 h p.i. compared with all other viruses examined (P < 0.05). Thai/16 and HK/483 viruses also were detected at elevated mean titers over 104 PFU/ml. In contrast, all other H7 and H3 influenza viruses tested did not replicate efficiently in the ex vivo tissue over the 96-h p.i. period. As epithelial cells are the most abundant cell population in the excised mouse cornea (23), we isolated corneal epithelial sheets from whole corneas to further assess the ability of these viruses to replicate specifically within this tissue (Fig. (Fig.5B).5B). We found that NL/219 virus achieved titers that were at least 100-fold higher than those of the other viruses tested over the 24- to 48-h p.i. period (reaching peak mean titers of >106 PFU/ml) and significantly higher than those of HPAI H5N1 viruses 24 to 96 h p.i. (P < 0.05). Thai/16 and HK/483 viruses were additionally detected at elevated titers above 104 PFU/ml by 36 h p.i. Other H7 viruses tested persisted in culture up to 48 h p.i., but did not exhibit efficient replication in the corneal epithelial sheets. Virus titers isolated from culture supernatant of infected corneal epithelial sheets were generally higher for all viruses than that from excised mouse corneas. These findings suggest that specific HPAI H7 and H5 viruses have the capacity replicate efficiently in mouse ocular tissue and identify the corneal epithelial sheet as the surface most likely to support influenza virus replication in this tissue.
As we found the mouse corneal epithelium to express both α2-3- and α2-6-linked SAs, we compared the ability of influenza viruses to productively infect cultured corneal epithelial cells that, like human ocular tissue, predominantly express α2-3-linked SAs. Primary HCEpiCs, isolated from human cornea, were found to express a high level of α2-3-linked SAs with low α2-6 SA expression (Fig. (Fig.3B),3B), similar to observed expression on normal human eyes (41). These cells were infected with influenza viruses of multiple subtypes at an MOI of 0.01 and supernatants were collected at indicated times p.i. to quantify infectious virus (Fig. (Fig.6).6). Similar to what was observed with mouse corneal tissues, NL/219, Thai/16, and HK/483 viruses replicated to significantly higher titers (P < 0.005) than all other viruses tested. The HPAI H7 viruses Can/504 and NL/230 also replicated to titers over 105 PFU/ml by 48 h p.i., whereas the LPAI viruses NY/107 and human Aichi/68 virus did not replicate to titers greater than 103 PFU/ml. In summary, we found that avian influenza viruses of multiple subtypes, unlike the human influenza viruses tested, are capable of replicating to high titers in HCEpiCs.
Recent reports of conjunctivitis following human infection with H7 influenza viruses prompted us to conduct this study to better understand the ability of influenza viruses to use the eye as a portal of entry. Identification of influenza virus subtypes that readily infect ocular tissue may help us to better predict the risk of ocular infection and disease following exposure to influenza virus. In this study, we established mammalian in vivo, ex vivo, and in vitro models to assess the ability of influenza viruses of multiple subtypes to mount a productive infection following ocular inoculation and further identified corneal epithelial cells as a likely target for influenza virus replication. We found that, unlike human influenza viruses, specific avian influenza viruses within the H5 and H7 subtypes could successfully mount a systemic and lethal infection in the mouse following ocular inoculation and replicate to high titers in ocular tissues and cells.
We identified the HPAI H7N7 virus NL/219 and the HPAI H5N1 viruses HK/483 and Thai/16 as capable of causing a lethal infection in mice when introduced by the ocular route. These viruses also resulted in the highest lung virus titers observed among all strains tested and, following ocular inoculation, replicated to the highest titer in ex vivo mouse corneal tissue and in vitro HCEpiCs. Interestingly, the mortality observed in mice following ocular inoculation was delayed 3 to 4 days p.i. compared with mortality following intranasal inoculation with a similar dose of virus (4, 25). This relative delay in time to death following ocular inoculation may, in part, be attributed to the additional time necessary for virus to travel from the eye to the lung and establish a productive and ultimately fatal respiratory infection. Replication-independent spread of virus inoculum from the eye to the lung has been implicated previously in an ocular murine RSV model (6), and a similar phenomenon could be occurring here. It should be noted that previous studies have used slightly lower inoculum volumes for ocular inoculation (6, 43); however, a volume of 5 μl was the minimum volume possible to achieve the high virus challenge dose (106 EID50 or PFU) examined in this study. It is reasonable to speculate that influenza virus travels from the eye via the lacrimal and nasolacrimal ducts to the nose, trachea, and finally lung. This transport appears to be unidirectional as intranasal inoculation of mice with these viruses does not result in viral titers in the eye (data not shown). Future studies examining ocular virus travel to the respiratory tract are warranted.
Despite a higher incidence of conjunctivitis following human H7 influenza virus infection in comparison with other virus subtypes, we found that HPAI H5N1 viruses replicated to substantial titers in ocular tissue ex vivo and HCEpiCs in vitro. However, upon in vivo ocular inoculation, H7N7 and H7N3 viruses were detected more frequently and at higher titer in the eye than H5N1 viruses, which were more frequently detected in respiratory tract tissues. Two additional HPAI H5N1 viruses that did not result in severe disease in the mouse model, VN/1203 and HK/486, replicated to titers comparable to those of other HPAI H5N1 viruses in HCEpiCs (data not shown) and may in part be due to the higher density of α2-3-linked SAs on the surface of the HCEpiCs compared with mouse corneas, as shown in Fig. Fig.3.3. Not all H7 viruses replicated efficiently in ocular cells, and in the case of the LPAI NY/107 virus, this may be due, at least in part, to an increase in α2-6 SA binding in comparison to the classic avian-binding preference for α2-3-linked SAs observed with NL/219 or NL/230 virus (3). While the precise molecular determinants that are responsible for this apparent ocular tropism have yet to be elucidated, this work reveals that highly pathogenic viruses of multiple subtypes are capable of replicating in ocular cells in vitro.
The human H1N1 1918 virus, which is highly lethal in mice when administered intranasally (42), did not result in severe infection of mice when administered by the ocular route. Moreover, human influenza viruses of the H3N2 subtype were not detected in the eyes of mice following ocular inoculation and did not replicate to significant titers in vitro or ex vivo ocular cultures. While we did not observe efficient replication of these human viruses in vivo or in vitro, ocular symptoms occur sporadically following influenza infection with seasonal viruses (37). Seasonal influenza vaccination has been associated rarely with the onset of oculo-respiratory syndrome in individuals with no prior history of ocular disease, and there have been sporadic reports of individuals rejecting corneal transplants following vaccination (12, 34, 35, 49). These adverse effects demonstrate the importance of studying the ocular environment in the context of both avian and human influenza virus infection and vaccination.
The presence of α2-3-linked SAs on the mouse and human cornea suggests that avian influenza viruses with this binding specificity would be well suited to infect this tissue and cause ocular disease. However, it appears that the ability of influenza viruses to bind to or replicate in ocular tissue cannot be explained by SA binding specificity alone. Both avian and human FITC-labeled viruses were capable of binding to ocular tissue regardless of SA binding preference. While the H5N1, H7N7, and H7N3 viruses included in this study all exhibit binding to α2-3-linked SAs, only those viruses within the H7 subtype replicated to substantial titer in the mouse eye in vivo, suggesting that an α2-3 SA binding preference alone is not sufficient to mount a productive ocular infection (3, 36). Furthermore, neither the reconstructed H1N1 1918 virus SC18 (α2-6 binding) nor the 1918 virus possessing the “avianized” HA (AV18; α2-3 binding) replicated in the eye, further indicating that properties other than SA binding preference contribute to the ability of selected influenza viruses to replicate in ocular tissue (45). As we observed heightened α2-6 SA expression on the mouse corneal epithelium compared to HCEpiCs, subsequent study investigating the binding of these viruses on human tissue is warranted. Additionally, detailed characterization of the SAs present on the corneal epithelial sheet of the human eye is needed to better understand and identify receptors that are preferentially bound by influenza viruses. Recent studies examining the structural topology of surface glycans in the upper respiratory tract of humans has revealed greater complexity in the binding of influenza virus HA to α2-3 and α2-6 SAs than previously understood (10); a similar analysis of ocular tissue may shed light on the precise composition of SAs on this surface and any correlation with preferential binding of influenza viruses, such as those within the H7 subtype, that are most frequently associated with conjunctivitis. A better understanding of the H7 and H5 subtype HA may further help elucidate why viruses within these subtypes, but not H3 or H1 viruses, were capable of replicating to high titer in the mouse lung following ocular inoculation; the lack of morbidity or mortality observed with either SC18 or AV18 viruses would suggest that, similar to replication within ocular tissue, properties in addition to SA binding preference are involved.
Similar to ocular models for RSV and herpes simplex virus (6, 17, 44), we lightly scarified the mouse cornea prior to virus inoculation to reflect eye abrasions which have been reported to accompany ocular infection with avian strains in some studies. Multiple cases of human conjunctivitis have resulted from poultry workers involved in the culling of birds during avian influenza outbreaks, or others with exposure to poultry, reporting possible eye abrasions prior to onset of illness (21, 46). However, similar levels of virus replication and disease outcome were obtained in mice when influenza virus was administered by the ocular route in the absence of corneal scarification (data not shown), indicating that corneal scarification may not be required for avian influenza virus replication in the eye and subsequent spread to the respiratory tract. Unlike other ocular models of virus infection (6, 28), mice inoculated with influenza virus by the ocular route did not exhibit corneal disease, as measured by dissecting biomicroscope, with the exception of one NL/219-infected mouse that displayed moderate corneal opacity.
The observation of higher titers of replicating virus in mouse corneal epithelial sheets as compared with excised mouse corneas, in addition to the high titers observed following infection of HCEpiCs, suggests that the epithelial sheet is the primary site of virus replication in the cornea. However, SAs are also present on human conjunctiva, and H7N7 influenza viruses have been shown to bind to human conjunctiva epithelium in vitro (27, 41). With regard to the H7 viruses NL/230 and Can/504, which replicated to high titer in mouse eyes but did not replicate to significant titer in excised mouse corneas or corneal epithelial sheets, the conjunctiva may represent an additional in vivo site of virus replication which is absent in our ex vivo system. Future studies examining the role of this tissue in ocular infection are warranted. Definition of the role of ocular mucins following influenza virus infection could additionally allow for a more complete understanding of influenza virus infection of the ocular area.
While the majority of avian influenza virus infections in humans currently result from direct contact with infected or dead poultry, it is still unclear in some situations exactly how the virus travels to the respiratory tract (1). Several H7 virus-infected individuals, including those from both the H7N7 outbreak in The Netherlands in 2003 and the H7N3 outbreak in Canada in 2004, reported both conjunctivitis and influenza-like respiratory illness following ocular exposure to virus (20, 46). Our findings of avian influenza viruses using the eye as a portal of entry to mount a productive and lethal infection in the mouse suggest that ocular exposure to some HPAI viruses could potentially result in the development of respiratory symptoms, similar to what was observed with these cases. In this study, H7 influenza viruses were most frequently associated with ocular disease, but avian viruses of the H5N1 subtype were also capable of using the eye as a portal of entry to initiate a productive infection. These results underscore the importance of wearing eye protection during possible exposure to avian influenza viruses of multiple subtypes (7).
We thank Ron Fouchier (Erasmus Medical Center), Yan Li (Canadian Center for Human and Animal Health), and the Hong Kong, Vietnam, and Thailand Ministries of Health for providing viruses used in this study and Hans Grossniklaus (L. F. Montgomery Eye Pathology Laboratory, Emory Eye Center) for technical assistance.
The findings and conclusions in this report are those of the authors and do not necessarily reflect the views of the funding agency.
Published ahead of print on 20 May 2009.