Influenza H5N1 virus infection of polarized human alveolar epithelial cells and lung microvascular endothelial cells

Highly pathogenic influenza (HPAI) H5N1 virus first emerged as a cause of severe human disease in 1997 in Hong Kong [1, 2]. Since then, it has become entrenched in poultry across Asia and Africa with zoonotic transmission to humans, sometimes with fatal outcome. In contrast to human seasonal influenza, H5N1 disease has a higher reported case-fatality rate ranging from 33% in Hong Kong in 1997 to 61% more recently [1, 3]. The reason for this unusual severity of human disease remains unclear. Within the lung, the alveolar epithelium is the primary target cell for influenza H5N1 virus [4‐6]. Although a novel influenza H1N1 virus of swine origin has recently emerged to cause a pandemic [7, 8], the pathogenesis of H5N1 virus remains an important public health issue because this virus remains a pandemic and public health threat, either directly or through reassortment with the novel pandemic H1N1 virus.

Epithelial cells line the major cavities of the body, functioning in selective secretion and adsorption, and providing a barrier to the external environment. In human lung, the alveolar epithelium consists of a continuous layer of tissue made up of two principal cell types: flattened type I alveolar epithelial cells and cubodial type II alveolar epithelial cells. Type I alveolar epithelial cells cover over 80% of the alveolar surface in which they function as a broad thin layer for gaseous exchange. These cells are highly polarized since the plasma membranes of these cells are divided into two discrete domains, namely, the apical domain (facing the luminal air surface) and the basolateral domain (facing the systemic circulation) [9]. And this large thin surface makes them extremely susceptible to injury from inhaled pathogens. While there is some data on H5N1 virus infection and cytokine responses in alveolar epithelial cells [10], there is no information of the effect of cell polarity on H5N1 virus replication or on virus-induced host responses.

Though influenza virus infection is localized primarily to the respiratory system, HPAI in some avian species is associated with systemic dissemination of the virus to multiple organs. There is increasing evidence that H5N1 influenza viruses are found in the peripheral blood, the gastro-intestinal tract and occasionally even the central nervous system of humans, and such dissemination may contribute to unusual disease manifestations including those of multiple organ dysfunction [11‐14]. The close anatomical relationship between alveolar epithelium and the lung microvascular endothelium, together with the distribution of putative influenza A virus receptors on the endothelial cell surface [15], make it important that parallel investigations on the lung epithelium and lung endothelium are carried out.

In the present study, we investigated the infection of polarized, primary, human type I-like alveolar epithelial cells and lung microvascular endothelial cells by influenza A virus. The low pathogenic human seasonal influenza virus, A/HK/54/98 (H1N1) and the HPAI A/HK/483/97 (H5N1) virus were studied. We found that both influenza H1N1 and H5N1 viruses efficiently infect alveolar epithelial cells from both apical and basolateral surface of the epithelium. Irrespective of the route of infection, both viruses were preferentially released at the apical surface of the alveolar epithelium. Whereas in lung microvascular endothelial cells, influenza H1N1 virus failed to replicate convincingly in contrast, influenza H5N1 virus showed evidence of replication following infection by either apical or basolateral route and new virus was released from both sides of the cell. As previously reported [10], influenza H5N1 virus was a more potent inducer of cytokines and chemokines (e.g. IP-10, RANTES, IL-6) in comparison to H1N1 virus in alveolar epithelial cells. Influenza H5N1 virus induced chemokines were secreted on both the apical and basolateral aspects of the polarized alveolar epithelium while the human influenza H1N1 virus led predominantly to apical secretion. These findings enhance the understanding of how virus infection may spread within and beyond the lung in influenza virus infection and how innate host response may contribute to modulating or aggravating tissue pathology.

In this study, we compared human influenza H1N1 virus with a highly pathogenic influenza H5N1 virus to investigate whether there are differences in the polarity of virus infection and of host cytokine responses in human polarized type I-like alveolar epithelium. The basolateral aspect of the alveolar epithelium lies in close proximity to the basolateral aspect of the lung microvascular endothelial cells raising the question of whether virus regressing the basolateral aspect of the type I-like alveolar epithelial cells can infect the lung microvascular endothelial cells by the basolateral aspect. Alternatively, since influenza H5N1 virus is believed to disseminate systemically and has been detected in the peripheral circulation, it is relevant to understand whether endothelial cells can be infected via the apical aspect, thereby allowing virus in the blood circulation to infect these cells and traffic outward to infect the lung alveolar epithelium from the basolateral aspect. As the lung endothelium covers about 20% of the total surface area of the alveoli sac, the rest being covered by the alveolar epithelium [20], the tropism of influenza A virus in both alveolar epithelium and endothelium is important in the pathogenesis of influenza H5N1 virus infection in human.

Previously, the polarity of influenza virus infection and release have only been studied with low pathogenic influenza A virus (subtype H3N2) and low pathogenic avian influenza A virus (subtype H5N3 and H4N6) [21] in human airway trachea-bronchial epithelial cells. It was demonstrated that newly forming influenza virus was released from the apical surface of respiratory epithelium [21, 22]. However, the mouse-adapted influenza H1N1 virus (WSN strain) and Sendi virus have been shown to bud from the apical and basolateral domains [23, 24]. Vesticular stomatitis virus and retroviruses [25] are released from the basolateral domain of polarized cells, the symmetrical or asymmetrical binding, internalization and budding of virus from polarized epithelial cells in culture therefore would have potential implications for viral pathogenesis. For example, with coronavirus, the apical release of transmissible gastroenteritis virus resulted in a local infection in vivo while the basolateral release of mouse hepatitis virus (MHV) in epithelial cells resulted in systemic infection [26]. In mouse studies, MHV initially replicates in the nasal epithelium before being disseminated throughout the body. The basolateral release of MHV from epithelial cells into the animal's circulation was postulated as the first step in the establishment of a systemic infection.

We showed that both influenza H1N1 and H5N1 viruses preferentially infect type I-like alveolar epithelial cell from its apical surface with higher levels of viral M gene expression as well as higher percentages of cells being infected, when compared to basolateral infection (Figure 3 and 4). This is expected since respiratory viruses need to be adapted to efficiently infect cells via the apical surface, which is the surface that is exposed to the respiratory lumen, and therefore accessible to infection. With influenza H1N1 virus, release of newly formed virus was restricted to the apical aspect, irrespective of whether the alveolar epithelial cells were infected by the apical or basolateral route (Figure 5). Again this is expected with a virus that appears not to disseminate beyond the lung. Similar observations have been reported in parainfluenza virus infected epithelia with the virus preferentially entering and exiting via the apical surface [27, 28]. Given its propensity to disseminate beyond the lung, we initially hypothesized that influenza H5N1 virus may be released from both apical and basolateral aspects. But this proved not to be the case and H5N1 was similar to H1N1 in this respect, i.e. virus was released predominantly via the apical aspect, irrespective of the route of infection of the cell (Figure 5B). While the efficiency of infection via the basolateral aspect was lower than that from the apical aspect for both viruses, cells infected with H5N1 virus via the basolateral aspect resulted in a greater than 10 fold higher virus yields on the apical surface than cells comparably infected with H1N1.

We then investigated virus replication in polarized lung microvascular endothelium which is anatomically in close proximity to the alveolar epithelium. There was no convincing evidence of replication of H1N1 virus in the polarized HLMVE cells (Figure 6A). In contrast, H5N1 virus could initiate productive replication of these cells from either aspect and virus release also occurred from apical or basolateral aspect of the cell (Figure 6B). Although neither influenza H5N1 nor H1N1 viruses are efficiently released via the basolateral aspect of the alveolar epithelium, virus replication is likely to lead to weakening of the tight-junctions and to cell death, thus providing these viruses access to the underlying tissues and the basolateral aspect of microvascular endothelial cells. As the lung microvascular endothelium also comprises 20% of the total surface area of the alveoli [20], influenza virus entry via the basolateral aspect of HLMVE cells, replication within them and release from the apical aspect of these cells could lead to viremia and dissemination of infection. The fact that HPAI H5N1 virus H0 precursor form can be cleaved by proteases not restricted to the lung [29] facilitates disseminated virus infection.

The fact that influenza H5N1 (but not H1N1) virus can infect the HLMVE cells from the basolateral aspect would facilitate dissemination of this virus via the blood stream. Furthermore, the observation that HLMVE cells can be infected via the apical aspect and release virus from the basolateral aspect (as well as the apical side) suggests that systemically circulating virus can initiate infection in the lung parenchyma via the endothelial route. This is particularly relevant since recent studies in mice have shown that HPAI H5N1 virus experimentally injected into muscle can led to fatal virus infection with virus establishing infection in the lungs and brain [30]. Furthermore, there has been speculation and anecdotal evidence that H5N1 virus can initiate infection via ingestion and the gastrointestinal tract [31]. The possibility that virus in the systemic circulation can establish a foothold in the lung is therefore an important observation.

Infected type I alveolar epithelial cells undergo either cytolytic or apoptotic death. The shedding of infected type I alveolar epithelial cells may further the inflammation and the underlining interstitial cells may then be exposed to the alveolar lumen fluid which contains high concentrations of virus. Reconstitution of the alveolar epithelial surface depends on the regeneration type I alveolar epithelial cell from its progenitor - the type II alveolar epithelial cells [32]. However, an intact basement membrane is essential for epithelial cell proliferation to occur. Alveolar basement membrane with denuded alveolar epithelial cell will accelerate the epithelial proliferation until the epithelial layer becomes confluent [33]. Nevertheless, type II alveolar epithelial cells could dominate the epithelial surface and prevent the reappearance of type I alveolar epithelial cell when injury signal of type I alveolar epithelial cells persists in the microenvironment [34].

Previous reports on human lung epithelial cell line A549 infected with human influenza H3N2 virus showed a low production of interferons and TNF-α [35]. We have previously shown, compared with influenza H1N1 virus, influenza H5N1 virus differentially upregulated cytokine and chemokine gene expression in alveolar epithelial cells [10] and macrophages [36]in vitro experiments and that the profile of differentially upregulated cytokines corresponds with the elevated IP-10 and MIG levels of H5N1 patients serum [37]. Interestingly, IP-10 and MIG have been reported to play roles in the pathogenesis of tissue necrosis and vascular damage associated with certain EBV-positive lymphoproliferative processes [38]. These results again may dictate the different pathogenesis of the downstream cytokine and chemokine response events and contribute to the unusual adverse pathology in H5N1 patient.

This study is the first demonstration of polarity secretion of cytokines in influenza H5N1 infected alveolar epithelium. The secretion of chemokines, notably IP-10, from both the apical and basolateral aspect, was found in influenza H5N1 virus infected type I-like alveolar epithelial cell but not in influenza H1N1 virus infected cell. This could potentially be relevant to the pathogenesis of influenza H5N1 virus infected patients. The basolateral release of chemotactic IP-10 from the influenza virus infected type I-like alveolar epithelial cells recruit lymphocytes from the capillary circulation into alveoli. The binding of chemokines to the receptor of the recruiting leukocytes is specific and leads to a rapid change in the cell shape and behavior of the subpopulation of the leukocyte. This makes them capable of migrating from the blood through the vascular endothelium into the site of inflammation [39, 40]. Previous studies on IP-10 and transendothelial migration indicated that IP-10 retained on endothelial cells could induce transendothelial chemotaxis of activated T cells [41]. Another investigation on the biological activity of human recombinant IP-10 investigation further verified its chemotactic properties towards human peripheral blood monocytes and stimulated human peripheral blood T lymphocytes, but not neutrophils [42, 43]. One of the studies used endothelial cell adhesion assay to demonstrate the effect of IP-10 in potentiating T cell adhesion to endothelium [43]. The potent secretion of IP-10 from both apical and basolateral side of the infected alveolar type I-like alveolar epithelial cell that we observed would suggest a possible directional recruitment and hence, migration of T cells and monocytes from the lung blood capillaries through microvascular transendothelial migration. Thus macrophages, differentiated from the recruited monocytes, may dominate the alveolar space [13, 17, 37] and T lymphocytes may occupy the interstitial space [13, 17, 37, 44, 45] as previously documented in autopsy reports of patients dying with influenza H5N1 virus infection. As cytokine secreting macrophages accumulate within the alveoli, further augmentation of the cytokine and chemokine cascades may result. Since influenza H5N1 virus is reportedly resistant to the anti-viral effects of interferons and TNF-α [46] and can lead to delayed apoptosis of infected macrophages [47], the clearance of the virus and lung inflammation would take a longer period of time than with seasonal influenza infection. Such prolonged inflammation would eventually result in pathological features with diffuse alveolar damage, hemorrhage [48] and finally interstitial fibrosis [13, 45, 49, 50], which are some key observations in the H5N1 patients.

In this study, we demonstrate that both influenza H1N1 and H5N1 viruses efficiently infect alveolar epithelial cells from both apical and basolateral surface of the epithelium but release of newly formed virus is mainly from the apical side of the epithelium. In contrast, influenza H5N1 virus, but not influenza H1N1 virus, efficiently infected polarized lung microvascular endothelial cells from either apical or basolateral aspect and also be released from either aspect of these polarized cells. This is likely to be of relevance to the pathogenesis and provides a possible explanation for the entry to the respiratory tract via the blood stream, as proposed by some who suggest that the gastro-intestinal tract can be a portal of entry for this virus. In addition, the release of inflammatory mediators such as IP-10 may be important contributors to the pathogenesis of the disease. More detailed studies on the mechanisms of alveolar epithelial cell damage and regeneration and the mediators involved in this process will be important in understanding the pathogenesis of human H5N1 disease.