SARS-CoV productively infects human bronchial epithelial cells, causing CPE.
In the course of searching for a relevant cell culture model for SARS-CoV replication, we infected an established cell line of human bronchial epithelial origin, Calu-3. Monolayers of Calu-3 cells were inoculated with SARS-CoV at an MOI of 0.1. To determine whether they were susceptible to SARS-CoV infection, the titers of infectious viral particles in the supernatants harvested at indicated time points after infection were assessed. Because Calu-3 cells were grown in a medium containing 20% FCS, virally infected cells were also tested in this medium and the usual 2% FCS medium. As shown in Fig.
1, a representative of two independent experiments, minimum 100- and 1,000-fold increases in the yields of infectious viral particles were consistently observed within 24 h in infected Calu-3 cells maintained in medium supplemented with 2% and 20% heat-inactivated FCS, respectively. Although the viral yields reached a maximum at days 1 and 2 after infection, depending upon the serum content in the culture medium, the infectivity of supernatants harvested thereafter without further passage of infected cells remained prominent.
To further verify the permissiveness of Calu-3 cells to SARS-CoV, we analyzed the expression of subgenomic mRNA 5 (M protein) at different time points after infection by quantitative real-time PCRs. As shown in Fig.
2, a dramatic increase in the expression of mRNA 5 could be readily detected in Calu-3 cells within 24 to 48 h after infection, confirming viral replication in Calu-3 cells. Indirect immunofluorescence staining with confocal imaging using convalescent-phase serum from a SARS patient also demonstrated the expression of SARS-CoV-specific antigens in some, but not all, areas of infected Calu-3 monolayers at day 2 after infection (Fig.
3).
SARS-CoV lytically infects Vero E6 cells, resulting in profound CPE with characteristic cell rounding. However, it persistently infects permissive human intestinal epithelial cell lines, such as Caco-2 and LoVo, without causing any CPE (
5,
21). Thus, we also investigated the nature of productive SARS-CoV infection in Calu-3 cells by examining microscopic lesions of infected cells cultured in medium supplemented with either 2% or 20% FCS. As with Vero E6 cells, SARS-CoV infection induced CPE in Calu-3 cells. However, in contrast to Vero E6 cells, where CPE usually occurred within 24 to 36 h after infection, coincident with the maximal release of infectious viral particles (
21,
22), no visible CPE could be observed until day 4 or 8 in infected Calu-3 cells, depending on the serum content in the culture medium. Additionally, the morphology of CPE in infected Calu-3 cultures was considerably different from that observed in Vero E6 cells (
7,
21-
23,
30). Specifically, elongated or balloon-like cells, eventually followed by the cellular detachment in the absence of cellular fusion as shown in Fig.
4, appear to be a characteristic feature of SARS-CoV-associated CPE in Calu-3 monolayers. The cell detachment of infected Calu-3 cells cultured in 20% FCS-containing medium starting from day 8 eventually reached over 95% of the monolayers by day 28 after infection. Interestingly, when cultured in medium containing 2% rather than 20% FCS, more than 95% of infected Calu-3 monolayers were detached at day 14 (data not shown), suggesting that the kinetics of CPE progression are also dependent on the serum content in the medium. Although the factors and mechanisms that account for such delayed and morphologically different CPE in infected Calu-3 cells are currently unknown, the fact that SARS-CoV-induced CPE occurred considerably later than the maximal viral release suggests that a direct viral effect may not be responsible for cell death. Taken together, these results clearly demonstrate that human bronchial epithelial cells, as exemplified by Calu-3 here in this study, are susceptible to SARS-CoV infection. More significantly, the ability of SARS-CoV infection to induce CPE in Calu-3 cells appears to duplicate, to some degree, the histopathological manifestations in lung biopsies and autopsy specimens obtained from SARS patients (
7,
22,
23,
30).
Expression of ACE-2 and its correlation with SARS-CoV infection in Calu-3 cells.
A successful CoV infection starts with its interaction with specific cellular receptors (
16). ACE-2 has recently been identified as a functional receptor for mediating SARS-CoV entry into permissive cells (
20). Additionally, the induced expression of ACE-2 on nonpermissive cells via transfection with expressing plasmid often confers susceptibility to productive SARS-CoV infection (
12,
33), indicating that ACE-2 alone is sufficient to allow viral entry. Thus, we initially investigated the expression of ACE-2 on Calu-3 cells by the flow cytometry using a goat anti-human ACE-2 polyclonal antibody and an FITC-conjugated secondary antibody. The FACS analysis revealed that approximately 30 to 70% of Calu-3 cells were positive for ACE-2 expression, depending on the gating (data not shown), with various staining intensity. We also used indirect immunofluorescence staining with a conventional immunofluorescence microscopy to localize the expression of ACE-2 in Calu-3 cells. As shown in Fig.
5A to C, the surface expression of ACE-2 could be detected in approximately 30% of Calu-3 cells by conventional immunofluorescence microscopy. At a higher magnification, clusters of cells that expressed ACE-2 on their surface could be readily visualized. The expression of ACE-2 on the surface of Calu-3 cells was also confirmed by studies with confocal microscopy. Figure
5D to F shows the clear demonstration of granular ACE-2 expression of different intensities on the surface of some, but not all, Calu-3 cells. The expression of ACE-2 in various tissues and cells has been extensively investigated with RT-PCR, Western blotting, immunohistochemistry, and flow cytometry (
5,
11,
12,
20). To our knowledge, this is the first time that the surface expression of ACE-2 has been clearly demonstrated on the surface of cells permissive for SARS-CoV.
The observation that only a fraction of Calu-3 cells was susceptible to SARS-CoV, as evidenced by the expression of viral antigens (Fig.
3), along with the demonstration of ACE-2 expression in some, but not all, Calu-3 cells (Fig.
5), prompted us to investigate the relationship between ACE-2 expression and SARS-CoV susceptibility of Calu-3 cells by two-color immunofluorescence staining. Uninfected and virally infected Calu-3 cells grown on the chamber slides were subjected to staining with pairs of goat anti-human ACE-2 antibody/FITC-conjugated rabbit anti-goat IgG and rabbit anti-SARS-CoV NP/Texas red-conjugated donkey anti-rabbit IgG. As shown in Fig.
6A and B, the expression of ACE-2 (green) was again restricted to some clusters of both uninfected and virally infected cells, whereas SARS-CoV-specific NP (red) could be detected only in a subset of cells infected with SARS-CoV for 2 days. Remarkably, a great majority of, if not all, Calu-3 cells showing the expression of viral antigen also coexpressed ACE-2 on their surface, as shown in the superimposed confocal imaging. Although studies of tissue and cellular distribution of ACE-2 have raised the possibility that ACE-2 may not be the only determinant of tissue and cellular tropism for SARS-CoV (
5,
11), the highly colocalized expression of ACE-2 and viral antigen in infected Calu-3 cells strongly indicates that ACE-2 plays a central role in SARS-CoV infection.
To further verify the role of ACE-2 in the viral replication, we treated Calu-3 cells with control antibody or various concentrations of anti-ACE-2 antibody before infection with SARS-CoV. The viral titer in supernatants harvested at day 3 after infection was determined. As shown in Fig.
7, addition of anti-ACE-2 antibody to the Calu-3 cell cultures resulted in a dose-dependent reduction of the infectious viral yields. Taken together, these results clearly demonstrate that the expression of ACE-2 is essential for SARS-CoV replication in Calu-3 cells.
SARS-CoV enters and is released from polarized Calu-3 cells preferentially through the apical membrane domain.
The pulmonary epithelial cell lining serves as a primary protective barrier against the entry of airborne respiratory pathogens. Plasma membranes of epithelial cells are highly polarized into two discrete domains (e.g., apical and basolateral membranes) that are separated by tight junctions that encircle the cells. The interaction of viruses with polarized epithelial cells has been extensively studied (
31). Having shown that human bronchial epithelial Calu-3 cells were highly susceptible to SARS-CoV, we investigated the polarity of viral entry and release from Calu-3 cells, aimed at understanding the role of pulmonary epithelial cells in respiratory SARS disease. We initially examined the expression of ACE-2 and viral antigens in infected Calu-3 cells grown on chamber slides using
Z sections of confocal imaging, representing the longitudinal cross section of the infected monolayer. Fig.
8 shows the
Z-section images of infected Calu-3 cells that have been immunostained with either an anti-SARS-CoV antibody alone or a combination of anti-ACE-2 and anti-SARS-CoV NP antibodies, as described above. Intense staining for SARS-CoV antigens (green) detected in infected cells was predominately restricted to the apical surface, with a very diffuse staining at the basolateral domain, as shown in Figure
8A. The expression of ACE-2 (green) correlated nicely with that of SARS-CoV-NP (red) at the apical surface of infected Calu-3 cells (Figure
8B). Thus, this polarized accumulation of viral antigens and/or ACE at the apical surface might suggest that SARS-CoV preferentially infects and/or is released from the apical surface of Calu-3 cells.
To further validate the polarity of SARS-CoV infection, Calu-3 cells were grown in filter inserts with a transwell system (Transwell-clear; Costar). Confluent Calu-3 monolayers were subjected to SARS-CoV infection at an MOI of 1 via either the apical or basolateral surfaces. At day 2 after infection, supernatants were collected from either the apical or basolateral chamber and assayed for infectious progeny virus, whereas the filter inserts containing infected monolayers were fixed and prepared for TEM. As shown in Fig.
9, infection through the apical surface resulted in the production of infectious progeny viruses, indicating the efficient entry of SARS-CoV into polarized Calu-3 cells through the apical surface. Strikingly, there was 5-log more infectious SARS-CoV particles recovered from the apical chamber than from the basolateral chamber of infected cultures, suggesting that the release of progeny SARS-CoV occurred almost exclusively via the apical membrane. The minute amounts of released virus present in the basolateral cultural medium may reflect either a limited basolateral release of virus from polarized cells or a leakage of the culture system, possibly due to the presence of a small number of incompletely polarized cells.
The preferential release of SARS-CoV from the apical surface was also demonstrated by TEM. As shown in Fig.
10, polarized Calu-3 cells were jointed to one another by typical tight junctions, and vesicles containing multiple nucleocapsids with electron-translucent and electron-dense cores can be seen aligned underneath the apical surface. Additionally, SARS-CoV particles present at the apical surface of a polarized epithelial cell and at the apical junction between two epithelial cells were also readily visible. Although a small proportion of released virus was also present in the basolateral chamber, attempts to localize viral particles at this domain by TEM were unsuccessful.
In contrast to the apical surface, infection through the basolateral surface of polarized cells did not result in the recovery of infectious viral particles from either chamber of the transwell system (Figure
9), suggesting that SARS-CoV may not properly interact with cells at the basolateral surface. Taken together, these results demonstrated that SARS-CoV efficiently infected polarized Calu-3 cells from the apical surface and also preferentially exited from the same surface.
A successful coronavirus infection of target cells relies heavily on its interaction with its cellular receptor (
16). Thus, the preferential expression of ACE-2, a functional receptor for SARS-CoV, on the apical surface of Calu-3 cells, as shown in Figure
8, may provide a molecular basis for effective SARS-CoV infection through the apical membrane domain. It is possible that a critical receptor density may be a prerequisite for a successful SARS-CoV entry to succeed, and this receptor density may be present only on the apical surface. The insufficient expression of ACE-2 on the basolateral membrane domains may thus account for the unsuccessful SARS-CoV infection in polarized cells. Alternatively, the failure of SARS-CoV to basolaterally infect polarized cells could be due to the absence or insufficient expression of coreceptors, as indicated by recent reports (
5,
11). These hypothetical coreceptors may be strictly polarized to the apical surface. Recently, CD209L (L-SIGN) was shown to be an additional receptor for SARS-CoV (
14). It will be interesting to determine what additional role CD209L might have in mediating SARS-CoV entry into Calu-3 cells.
From our studies described above, the apical surface appears to be the primary site for SARS-CoV entry into and release from Calu-3 lung epithelial cells, a pattern shared by other respiratory viruses, including human CoV 229E, measles virus, and human parainfluenza virus, interacting with polarized epithelial cells (
2,
3,
32). Although the exact mechanism specifying this apical-surface-to-apical-surface entry and release of this virus is not clear, the polarized distribution of SARS-CoV receptor on the apical surface may be critically involved. This directional release of viral particles provides an excellent model for viral transmission and has interesting implications for viral pathogenesis; that is, when SARS-CoV is brought into contact with the airways via inhalation, it efficiently enters via the apical surface and replicates within bronchial epithelial cells. Once the replication is completed, the great majority of progeny viruses are released back into the lumen, where they could apically infect additional epithelial cells, be released back into the environment in respiratory droplets, or be taken up by alveolar MΦ, which may be ultimately responsible for taking the virus to the bloodstream. While most of the viruses are released into lumen via the apical surface, a small amount of basolaterally released progeny viruses might enter the circulation and propagate by infecting other non-respiratory-tract cells, including immune cells and intestinal epithelial cells, resulting in viremia and systemic infection. Moreover, we have recently shown that interactions of SARS-CoV with MΦ and dendritic cells (DC) can modulate various key functions of these two key innate cells, including phenotypic and functional maturation, phagocytosis, and inflammatory cytokine production (C. K. Tseng et al., unpublished data). Since SARS is believed to stem from an acute, profound, and unregulated inflammatory response in the lungs (
7,
10,
17,
23,
30), active virus interactions with alveolar MΦ in the lumen and with DC that are dispersed between epithelial cells may be highly relevant to SARS pathogenesis. It is important that the polarity of SARS-CoV in our study was assessed at day 2 after infection, at which time no CPE could be observed. The fact that SARS-CoV infection eventually leads to the detachment of Calu-3 monolayers allows the argument that the strong apical polarization expected to dominate the earlier days of infection will be breached, thus enabling the virus to intimately interact with intraepithelial DC and other innate inflammatory cells.
In summary, we have characterized a SARS-CoV-susceptible epithelial cell line of human origin, Calu-3. We found that entry and release of SARS-CoV likely occurs at the apical surface in highly polarized Calu-3 cells. ACE-2, the receptor of SARS-CoV, is expressed predominately on the apical surface, which may play a crucial role in dictating viral infection. In contrast to other permissive human intestinal epithelial cell lines, with which SARS-CoV causes a persistent infection in the absence of CPE, Calu-3 cells respond to SARS-CoV by showing a delayed CPE, making it unique and potentially useful for studying the SARS-CoV-related pathogenesis of the respiratory system.