Abstract
The lymphotropic and myelotropic nature of wild-type measles virus (wt-MV) is well recognized, with dendritic cells and lymphocytes expressing the MV receptor CD150 mediating systemic spread of the virus. Infection of respiratory epithelial cells has long been considered crucial for entry of MV into the body. However, the lack of detectable CD150 on these cells raises the issue of their importance in the pathogenesis of measles. This study utilized a combination of in vitro, ex vivo and in vivo model systems to characterize the susceptibility of epithelial cells to wt-MV of proven pathogenicity. Low numbers of MV-infected epithelial cells in close proximity to underlying infected lymphocytes or myeloid cells suggested infection via the basolateral side of the epithelium in the macaque model. In primary cultures of human bronchial epithelial cells, foci of MV-infected cells were only observed following infection via the basolateral cell surface. The extent of infection in primary cells was enhanced both in vitro and in ex vivo cornea rim tissue by disrupting the integrity of the cells prior to the application of virus. This demonstrated that, whilst epithelial cells may not be the primary target cells for wt-MV, areas of epithelium in which tight junctions are disrupted can become infected using high m.o.i. The low numbers of MV-infected epithelial cells observed in vivo in conjunction with the absence of infectious virus release from infected primary cell cultures suggest that epithelial cells have a peripheral role in MV transmission.
INTRODUCTION
Measles is a severe disease contributing to significant morbidity and mortality rates in many parts of the developing world (Grais et al., 2007). It is also of growing concern in some areas of the developed world where the vaccination rate has fallen below the level required to prevent periodic outbreaks of measles (Choi et al., 2008). The disease is characterized by fever and a maculopapular rash, often in conjunction with cough, coryza and/or conjunctivitis, and a generalized immunosuppression. The causative agent, wild-type measles virus (wt-MV) is spread by aerosolized droplets or direct contact (Griffin, 2007). The highly infectious nature of MV not only suggests a very efficient route of entry into the body but also an effective means of dissemination to susceptible individuals following systemic spread of the virus within the body.
The tropism of MV in vivo is determined mainly by the distribution of CD150, also known as signalling lymphocyte activation molecule (SLAM), which is the primary cellular receptor for wild-type strains of the virus. The molecule is widely expressed in lymphoid tissues on subsets of B and T lymphocytes, thymocytes, macrophages and dendritic cells (DCs) (Kruse et al., 2001; McQuaid & Cosby, 2002). This is reflected by analysis of MV-infected cell types present at different stages of the disease, which shows that, in all tissues involved, the majority of infected cells are of a lymphoid origin (Hall et al., 1971; Moench et al., 1988). Another cell-surface molecule that has been implicated in wt-MV infection is the C-type lectin DC-SIGN, which is expressed exclusively on DCs. Although it does not act as an entry receptor, this molecule enhances wt-MV infection of DCs in vitro and mediates trans-infection of lymphocytes (de Witte et al., 2006, 2008). The observation that large numbers of DC-SIGN-positive DCs are present in the epithelium of the respiratory tract provides support for an alternative mechanism through which MV may gain entry into the body.
The generation of recombinant MVs that express fluorescent proteins such as enhanced green fluorescent protein (EGFP) allows the detection of single MV-infected cells within tissue samples from infected animal systems (Duprex et al., 1999). The application of this virus to a natural model has resulted in a reassessment of some of the basic tenets of measles pathogenesis (de Swart et al., 2007; Hashimoto et al., 2002; Ludlow et al., 2007). Extensive macro- and microscopic analysis of MV-infected macaque tissue samples has shown that the major targets of MV infection in lymphoid tissues are CD150+ B and T lymphocytes, with large numbers of MV-infected CD11c+ MHC class II+ myeloid DCs observed in a number of peripheral tissues (de Swart et al., 2007). This study also emphasized the importance of MV spread and dissemination within the body through cell-to-cell contact rather than through the development of large multi-nucleated syncytia in which only a minority of the MV-infected cells present in lymphoid tissues were observed to reside. Many studies have reported extensive epithelial cell infection during measles (Kimura et al., 1975; Lightwood & Nolan, 1970; Moench et al., 1988; Olding-Stenkvist & Bjorvatn, 1976). However, in the absence of appropriate epithelial and immune cell markers, caution should be applied in using such studies to support a prominent role for epithelial cells in the pathogenesis of measles. In addition, the inherent complexity of the network of cells that comprise the respiratory epithelium can lead to erroneous assumptions on the identity of MV-infected cells. Accordingly, a new approach to dissecting the role of epithelial cells in measles pathogenesis is warranted.
A number of recent studies investigating MV infection of epithelial cells have focused on in vitro model systems (Tahara et al., 2008; Takeda et al., 2007) or have utilized the macaque model of measles in the absence of histological or pathological analysis of MV-infected tissues (Leonard et al., 2008). These studies have identified specific amino acids of the MV haemagglutinin (H) glycoprotein that are involved in the binding of H to a putative receptor (EpR) on the basolateral surface of epithelial cells and suggest a critical role for epithelial cells in mediating MV transmission. Here, we have used a combination of in vivo, ex vivo and in vitro model systems to study the role of epithelial cells in the pathogenesis of measles.
RESULTS
Infection of epithelial cells by MV is observed very rarely in the macaque
The availability of an EGFP-expressing wild-type strain of MV in conjunction with a natural model system of measles (the macaque model) allows a sensitive in vivo assessment of both the numbers and tissue tropism of MV-infected epithelial cells. This model has already proved valuable in illuminating the role of DCs and lymphocytes in measles pathogenesis and also, in some instances, the involvement of the epithelium in the disease (de Swart et al., 2007). In the present study, we utilized a comprehensive panel of formalin-fixed MV-infected monkey tissues to determine to what extent epithelial cells in diverse organs were involved in natural measles. In concurrence with the observation of living, fluorescent MV-infected ciliated epithelial cells in a subset of bronchoalveolar lavage (BAL) cells taken from infected monkeys at 6 and 9 days p.i. (unpublished data), analogous cells were readily detectable in a variety of tissues from the respiratory tract of these animals. Analysis of immunostained tracheal and bronchial rings showed that small focal areas of MV-infected ciliated epithelial cells were present in most of the sections analysed (Fig. 1a⇓). However, the number of these cells was relatively low in comparison with the number of MV-infected lymphoid and myeloid cells that were present in subepithelial cell layers. Similarly, the large majority of MV-infected BAL cells had the phenotype of lymphoid or myeloid cells, not of epithelial cells. Strikingly, in vivo foci of MV-infected epithelial cells were mostly observed in areas of the trachea or bronchus in which MV-infected lymphocytes or DCs were present in close proximity to the epithelial cell layers (Fig. 1a⇓, insets). Foci of MV-infected epithelial cells in the respiratory tract were typically small (fewer than six cells) and did not form multi-nucleated syncytia.
Detection of MV-infected epithelial cells in formalin-fixed tissue sections obtained from an rMVIC323EGFP-infected rhesus monkey at 9 days p.i. (a) A composite image of a complete tracheal ring. MV-infected cells (green) were observed in both the epithelial and subepithelial cell layers adjacent to the tracheal lumen (TL). Inset (i) shows MV-infected ciliated bronchial epithelial cells readily detectable adjacent to the TL. Inset (ii) shows that the majority of MV-infected cells in the trachea are of lymphoid or myeloid origin and are located in subepithelial cell layers. The location of insets (i) and (ii) is indicated by boxes in the composite image. (b) A composite image of MV-infected bronchial epithelial and subepithelial cells (green) adjacent to the bronchial lumen (BL) in the lung. (c) MV-infected cells (green) expressing the epithelial cell marker CAM5.2 (red). (d) SLAM+ cells (green) are distributed in the basal epithelial cell layers and in aggregates of lymphoid tissue in subepithelial cell layers. (e) The majority of MV-positive cells (red) within large foci of infection in the tongue are in the upper non-proliferating layers of the epithelium. Only a few Ki67+ (green) cells in the basal epithelial layer were MV-infected. MV-infected cells within this cell layer stained positive for the epithelial cell marker AE1/A3 (inset). (f) Abundant MV-infected cells (green) in the duodenum are present in the submucosal regions and virus is absent from CAM5.2+ epithelial cells (red). Bar, 100 μm. Cell nuclei in (a) and (b) were counterstained with propidium iodide (red), whilst cell nuclei in (c), (d), (e, inset) and (f) were counterstained with DAPI (blue). Bars (μm), 400 μm (a); 200 μm (b); 50 μm (c); 100 μm (d); 150 μm (e).
Foci of MV-infected ciliated bronchial epithelial cells were observed at low levels and were also located in close proximity to infected lymphoid or myeloid cells in subepithelial cell layers surrounding the bronchial lumen (Fig. 1b⇑). Spread of MV from immune cells to cytokeratin-positive epithelial cells in the lung (Fig. 1c⇑) could occur via SLAM+ cells, which were present within epithelial cell layers (Fig. 1d⇑). Discrete foci of MV-infected epithelial cells in close proximity to infected lymphocytes and DCs were also present in the tongue of infected animals (Fig. 1e⇑). These cells, which stained positive for the epithelial cell marker cytokeratin, were present mainly in the stratum spongiosum and stratum corneum epithelial layers. Only small numbers of infected cells were observed in the Ki67+ proliferating stratum basal layer of the tongue, showing that MV infection of epithelial cells does not require actively dividing cells. Foci of MV-infected epithelial cells in the tongue were composed of larger numbers of cells (∼30–50) than analogous foci in the upper and lower respiratory tracts (Fig. 1a–c⇑). An extensive analysis of tissues from outside the respiratory tract including the stomach, ileum, kidneys and bladder did not show any convincing evidence of MV-infected epithelial cells. Close contact in the absence of viral spread between MV-infected lymphocytes and uninfected cytokeratin-positive epithelial cells was commonly observed in tissues such as the appendix and duodenum (Fig. 1f⇑).
Infection of the apical surface of ex vivo-cultured human corneal rim epithelium results in a limited infection
Ocular complications such as corneal inflammation (keratitis) resulting from MV infection are commonly observed in the developing world, but little is known about the susceptibility of epithelial cells in the eye to MV infection (Rima & Duprex, 2006). Therefore, in order to extend observations of epithelial cell infection in the macaque model of measles, we examined the susceptibility of the RPE cell line and ex vivo-cultured corneal epithelium to the wild-type rMVIC323EGFP. Infection of RPE cells with vaccine and wt-MV strains at an m.o.i. of 0.01 resulted in productive infections with markedly different cytopathic effects. Infection with the recombinant vaccine MVeGFP strain resulted in the formation of large multi-nucleated syncytia by 48 h p.i. (Fig. 2a⇓), whilst rMVIC323EGFP-infected RPE cell monolayers contained only small foci of infection (fewer than six cells) at equivalent time points (Fig. 2b⇓).
Susceptibility of continuous (a, b) and primary explant (c–g) epithelial cells of an ocular origin to MV infection. (a, b) RPE cells were infected at an m.o.i. of 0.01 with MVeGFP (a) or rMVIC323EGFP (b) and UV microscopy was used at 48 h p.i. to observe and image autofluorescent foci of infection. (c–g) Corneal explant cultures were infected with 106 TCID50 rMVIC323EGFP. The approximate location of virus infection is indicated by the green crosses on the schematic diagrams to the right. (c, d) Phase-contrast photomicrograph (c) and fluorescent overlay image (d) of a single focus of infection (green) observed at 5 days after infection of the apical surface of the cornea. (e, f) Phase-contrast photomicrograph (e) and equivalent fluorescent image (f) of a large focus of infection (green) observed at 4 days after infection of the peripheral surface of the corneal rim. (g) A composite image of five fluorescent photomicrographs of extensive MV infection (green) observed at 5 days after infection of the corneal rim. The dotted line indicates the approximate location of the peripheral surface of the corneal rim. MV N protein was readily detected in formalin-fixed sections of MV-infected corneal rim tissue (inset). Bars, 200 μm (a–f); 500 μm (g).
A drawback in the use of continuous epithelial cell lines is that the characteristics of such cell lines are very different from those of polarized primary epithelial cells. Therefore, in order to obtain a more informative assessment of the susceptibility of primary human epithelial cells derived from the ocular system to MV, we monitored rMVIC323EGFP-infected human corneal ex vivo explant cultures over a period of 5 days. Following infection of the apical surface of the cornea, only a single fluorescent epithelial cell was observed at 5 days p.i. (Fig. 2c, d⇑). Further monitoring of this single focus of infection showed that viral cell-to-cell spread in the cornea was completely restricted. In contrast, large numbers of MV-infected cells were observed following application of virus to the edge of the corneal rim, the integrity of which had been compromised during surgical removal of the tissue (Fig. 2e, f⇑). Despite extensive virus replication and cell-to-cell spread, the absence of any cytopathic effect meant that foci of infection could only be observed by exposing the tissue to UV light. In some instances, extensive MV cell-to-cell spread from the wounded surface of the corneal rim resulted in large foci of infection penetrating deep within the tissue (Fig. 2g⇑).
Infection of human epithelial cells by MV occurs exclusively via the basolateral surface, is enhanced by wounding of the cell monolayer and does not result in the generation of multi-nucleated syncytia
Upon observing that epithelial cells in the respiratory tract of monkeys and human corneal tissue are susceptible to MV, we developed a model system that would allow us to take a mechanistic approach to the study of MV infection of epithelial cells. This was achieved through the growth of air–liquid interface cultures of primary human bronchial epithelial cells, which were assessed after maintenance for 28 days on Transwell inserts for the presence of beating cilia, mucous production and a high electrical resistance. Confocal microscopy, optical sectioning and three-dimensional reconstructions showed that two to three layers of epithelial cells were present at this time point (Fig. 3a⇓). Whilst large foci (70–100 cells) were observed at 3 days after rMVIC323EGFP infection (m.o.i. of 1) onto the basolateral side of bronchial epithelial cells, analogous infection of the apical side of cultures did not result in the detection of a single fluorescent cell (Fig. 3b⇓, inset). In order to ensure that the above observations were applicable to other strains of MV, bronchial epithelial cultures were infected with the non-recombinant wild-type MVDub strain at an m.o.i. of 1 (Fig. 3c⇓). This confirmed that foci of infection were present following infection of the basolateral cell surface; MV-infected cells were never observed at any time point after virus was applied to the apical surface of the cultures. Whilst the application of MV to the basolateral surface of epithelial cell cultures resulted in large foci of infection by 3 days p.i., infectious virus was never detected in either the apical or the basolateral compartment at any time point. The absence of MV-infected cells following application of virus to the apical cell surface indicated that cilia on the surface of bronchial epithelial cells are unable to mediate virus entry (Fig. 3c⇓). Mucus-secreting goblet cells were poorly susceptible during basolateral MV infection, although occasional examples of mucin-positive cells were observed within large foci of infection (Fig. 3d⇓).
Susceptibility of human bronchial epithelial cells to infection by wild-type strains of MV. (a) Three-dimensional reconstruction of ciliated bronchial epithelial cells (red) detected by immunostaining with an anti-tubulin mAb. (b) A large syncytium of rMVIC323EGFP-infected cells (green) at 3 days after infection of the basolateral surface of a bronchial epithelial cell monolayer. No MV-infected cells were observed following rMVIC323EGFP infection of the apical surface of the cell monolayer (inset). (c) A large syncytium of MVDub-infected cells (green) observed 3 days after infection of the basolateral surface of a bronchial epithelial cell monolayer. No MV-infected cells were observed following MVDub infection of the apical surface of the cell monolayer (inset). (d) Incorporation of mucin-expressing goblet cells (red; arrow) within a focus of rMVIC323EGFP-infected cells (green). (e) Cell wounding increases the efficiency of apical infection by rMVIC323EGFP. A composite image of 12 fluorescent photomicrographs obtained 3 days after infection of the apical surface of a wounded bronchial epithelial cell monolayer. Extensive foci of infection were observed along the line of cell wounding (dotted line). (f) MVDub-infected bronchial epithelial cells (green) observed at 4 days p.i. do not express the cell-proliferation marker Ki-67 (red). Bars, 200 μm (b, f); 50 μm (c); 25 μm (d); 500 μm (e).
The dual observation that the integrity of the ciliated epithelial cell layer lining the respiratory tract is often disrupted following viral or bacterial infection or by mechanical damage (Message & Johnston, 2001) and that tissue damage may result in high levels of MV-infected epithelial cells (Fig. 2e–g⇑) raises the possibility that disruption of the tight junctions at cell–cell contacts may lead to an increase in susceptibility to MV infection. In order to test this hypothesis, cultures of differentiated bronchial epithelial cells, which had been damaged by drawing a pipette tip over the surface of the cell monolayer in the shape of a sinusoid curve, were infected with rMVIC323EGFP. Large numbers of foci of infection were observed along the lines of cell wounding at 3 days p.i. (Fig. 3e⇑). Foci of infection were never observed in areas of the culture in which cell–cell contacts had not been disrupted. These areas of disruption in the cell monolayer were rapidly repaired due to cell division, and this raises the possibility that the increased number of foci of infection is due to the preferential infection by MV of replicating cells. However, MV-infected cells in these cultures did not stain positively for the cell proliferation marker Ki-67, demonstrating that this is not the case (Fig. 3f⇑).
A characteristic feature of wild-type MV infection of SLAM+ cell lines is the formation of large multi-nucleated syncytia due to fusion of the cell membranes of MV-infected cells with adjacent uninfected cells. This was commonly observed in vivo in sections from the lungs of patients with acute measles (data not shown) and in lymphoid tissues, such as the tonsils, of monkeys infected with rMVIC323EGFP (Fig. 4a⇓). One consequence of the fusion of adjacent cell membranes is that tight junction proteins such as ZO-1 are downregulated within large syncytia (Fig. 4b⇓). In contrast, ZO-1 was not downregulated from the cell membranes of adjacent MV-infected primary bronchial epithelial cells (Fig. 4c⇓) or from adjacent MV-infected epithelial cells lining the trachea of an infected monkey (Fig. 4d⇓), showing that breakdown of cell membranes through the fusion of adjacent cells is not a prerequisite for MV spread in vitro or in vivo.
Effect of MV cell-to-cell spread on epithelial cell tight junctions. (a) ‘Classical' syncytia (arrows) of MV-infected lymphoid cells observed in a germinal centre of a haematoxylin and eosin-stained section from the tonsil of an rMVIC323EGFP-infected rhesus monkey. (b) The tight junction marker ZO-1 (red) is absent in rMVIC323EGFP-infected VeroSLAM cells (green) due to the MV-induced fusion of adjacent cells. Cell nuclei were counterstained with DAPI (blue). The inset shows the same focus of infection in the absence of EGFP fluorescence. (c) No downregulation of ZO-1 (red) is observed at the cell membranes of rMVIC323EGFP-infected human bronchial epithelial cells (green). (d) ZO-1 (red) is detectable at the cell membranes (arrows) of adjacent MV-infected epithelial cells observed in an immunostained section from the trachea of an rMVIC323EGFP-infected rhesus monkey. Bars, 100 μm (a); 50 μm (b); 25 μm (c); 10 μm (d).
DISCUSSION
The respiratory tract provides a number of unique challenges, which viruses such as MV have to overcome in order to gain entry into the body. Initially, a virus encounters a multitude of barriers formed by the presence of mucus, surfactant, enzymes and antimicrobial molecules. Once the virus succeeds in reaching the epithelium, tight junction proteins at epithelial cell boundaries maintain a physical barrier preventing pathogens from accessing the internal environment (Guttman & Finlay, 2008). Despite these barriers, it has been proposed that the entry of MV into the body is initiated following infection of the apical side of respiratory epithelial cells, with subsequent spread to DCs leading to amplification of the virus in a regional lymph node (Blau & Compans, 1997; Griffin, 2007). However, in vivo models of measles have so far provided no direct evidence in support of this model. In this study, we have attempted to delineate the role of epithelial cells in measles pathogenesis through the use of complementary model systems.
We showed that wild-type MV cannot infect primary bronchial epithelial cells or corneal epithelium via the apical surface, indicating that epithelial cells are unlikely to be involved in the initial stages of MV infection. The close proximity of MV-infected epithelial cells in the respiratory tract of macaques to underlying infected leukocytes suggested that virus infection in the epithelium may arise following spread from underlying infected leukocytes. Interestingly, a number of other viruses that enter the body through the respiratory tract, such as lymphocytic choriomeningitis virus, Junín virus and vaccinia virus, have also been shown to infect epithelial cells preferentially via the basolateral surface (Dylla et al., 2008; Vermeer et al., 2007). An alternative model of MV entry into the body has been proposed by de Witte et al. (2008) in which respiratory DCs capture the virus via DC-SIGN, and transmit the virus to lymphocytes in the regional lymph nodes by in cis or in trans infection (de Witte et al., 2008). Examination of tissues from infected macaques at earlier time points could provide evidence in support of this model or, alternatively, highlight a role for alveolar macrophages and lymphocytes in BAL in mediating MV entry, as these cells are also known to traffic from the respiratory epithelium to the lymph node (Corry et al., 1984; Lehmann et al., 2001).
Although at present the MV EpR on the basolateral surface of epithelial cells remains elusive, progress has been made in the analysis of the interaction of the MV H protein with this putative receptor. Tahara et al. (2008) analysed a number of MV H glycoprotein mutants and identified three amino acids (L482, Y541 and Y543) that are critical for mediating fusion of epithelial cells. Interestingly, mutation of these residues enabled an ‘epithelial-blind' virus to be generated, confirming that these residues are critical for interacting with the EpR. Leonard et al. (2008) extended these observations by studying the virulence of an analogous recombinant MV in macaques. This virus was virulent but was not shed from the airways of infected macaques, suggesting that epithelial cell infection during measles is critical for virus transmission rather than for primary infection. However, in order to confirm the absence of epithelial cell infection, it would be useful to perform a detailed examination of the in vivo tropism of this virus due to the absence of any histopathological analysis of tissues from infected animals in this study. Interestingly, infectious virus was not detected in BAL samples, which have previously been reported as positive in infected macaques before the onset of viraemia at 3 days p.i. (de Swart et al., 2007). Importantly, the majority of MV-infected cells in the BAL of macaques are SLAM+ alveolar macrophages (CD11c+ MHC II+) and lymphocytes, which should be infected with the same efficiency by both epithelial-blind and wild-type MV.
In contrast to previous studies in which MV-infected epithelial cells were detected in peripheral tissues from human cases of measles (Moench et al., 1988; Olding-Stenkvist & Bjorvatn, 1976), infection of epithelial cells in the macaque model was restricted to the respiratory tract. One distinctive feature of measles, which has been attributed to epithelial cell infection in peripheral tissues, is the detection of MV antigen and RNA in the urine of patients. However, whilst a number of studies have reported the isolation of MV (Gresser & Katz, 1960) or detected MV-infected giant cells in the urine of measles patients (Bolande, 1961; Boyd & Nedelkoska, 1967) or in bladder mucosal tissue (Sherman & Ruckle, 1958), information regarding the origin of infectious virus and/or MV-infected cells remains sparse. MV antigen was detected in the bladder urothelium from one patient, although specific details, such as the morphology or number of infected cells, were not reported (Moench et al., 1988). The lack of appropriate cellular markers in such studies, in conjunction with the absence of MV-infected epithelial cells in the bladder urothelium of infected macaques, does not rule out an alternative source, such as leukocytes, being responsible for the spread of MV to the urine of measles patients.
High levels of epithelial cell infection are observed in peripheral tissues in other morbillivirus infections, for example acute distemper in ferrets (von Messling et al., 2004). This may be due to differences in the dynamics of virulent canine distemper virus (CDV) infection in the ferret in comparison with MV in the macaque model. Whilst macaques recover rapidly from measles between days 9 and 15 p.i., systemic spread of CDV to peripheral sites in the ferret results in an increasing level of virus replication, with death occurring from 14 to 35 days p.i. Thus, the higher level of virus-infected epithelial cells observed in distemper may be attributable to a longer time period during which virus can spread within the epithelium, and to an increased virus burden and impairment of the host immune response in comparison with MV infection in the macaque.
Another animal model, such as the cotton rat, may help to dissect the role of the putative epithelial receptor in entry in a non-lymphotropic in vivo infection. It has recently been proposed that a prerequisite for MV transmission through epithelial cell barriers is a loss of integrity of the epithelium through the disruption of tight junctions at cell boundaries (Leonard et al., 2008). However, we found that the tight junction protein ZO-1 was present at the membranes of in vitro and in vivo MV-infected epithelial cells, suggesting that, at least in small foci of infection, the epithelial barrier remains intact and that virus spreads between cells without the breakdown of cell membranes, similar to the spread of wild-type MV in endothelial cell monolayers (Andres et al., 2003). Thus, exfoliated epithelial giant cells in swab samples taken from patients – a historical diagnostic feature of measles (Lightwood & Nolan, 1970; Scheifele & Forbes, 1972) – may not arise as a consequence of direct MV infection but instead may occur as result of an inflammatory immune response. Interestingly, similar epithelial giant cells are often reported in cases of non-specific chronic inflammation of the nasopharynx (Ali, 1965).
In conclusion, this study proves that only limited numbers of MV-infected epithelial cells are observed in the respiratory tract in vivo and that these cells are infected via the basolateral side following contact with underlying MV-infected leukocytes. Spread of MV between epithelial cells without downregulation of tight junctions at cell boundaries suggests that the breakdown of epithelial cell barriers and subsequent transmission of virus may not necessarily occur as a direct result of MV infection. Further analysis of in vivo infections at earlier time points is necessary to clarify the mechanism(s) utilized by MV to initiate infection, and systematic analysis of tissue samples from measles patients will help to clarify the involvement of epithelial cells in virus transmission.
METHODS
Cell-lines and viruses.
VeroSLAM cells (a gift from V. von Messling, INRS-Institut Armand-Frappier, University of Quebec, Canada) were grown in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 8 % (v/v) fetal calf serum (FCS; PAA Laboratories) and 0.1 % (v/v) Zeocin (Invitrogen). Retinal pigment epithelial (RPE) cells (kindly provided by Alan Stitt, Queen's University, Belfast, UK) were grown in DMEM supplemented with 10 % (v/v) FCS. The recombinant wild-type strain rMVIC323EGFP was rescued following transfection of VeroSLAM cells with a full-length infectious anti-genomic clone of p(+)rMVIC323EGFP (kindly provided by Yusuke Yanagi, Kyushu University, Fukuoka, Japan), and helper plasmids expressing the MV nucleocapsid (N), phospho- (P) and large proteins of MV. MVDub is a clinical isolate of MV that was isolated following inoculation of B95a cells with a throat swab obtained from a patient during an outbreak of measles in Dublin in 2000. Virus stocks of both wt-MV strains were generated in B95a cells using Opti-MEM (Invitrogen) following plaque purification, and titres up to a maximum of approximately 106 50 % tissue culture infectious doses (TCID50) ml−1 were obtained. Viral titres were obtained by a 50 % end-point dilution assay and are expressed in TCID50 ml−1 as determined by the method of Reed & Muench (1938).
Isolation, culture and virus infection of primary human bronchial epithelial cells.
Bronchial brushings were obtained from healthy volunteers undergoing bronchoscopy for research purposes. They were performed after informed consent and with approval from the Research Ethics Committee of Queen's University of Belfast. Primary bronchial epithelial cells were isolated and cells were cultured as described previously (Doherty et al., 2003). Air–liquid interface cultures for establishment of well-differentiated, mucociliary epithelium were produced based on previously described methods (Fulcher et al., 2005; Gray et al., 1996). Cultures were maintained for 28 days to ensure full differentiation, which was assessed by the presence of beating cilia, mucus production and a transepithelial electrical resistance (TEER) of between 600 and 800 Ω×cm2. The TEER was measured using an STX3 electrode and EVOM metre device (World Precision Instruments). Virus infection of the apical and basolateral surfaces of epithelial cells was carried out for 90 min at 37 °C, after which time inoculum was removed and cultures were washed six times in PBS to remove unattached virus. Cultures were monitored daily using a DM IRBE UV microscope fitted with appropriate filter blocks (Leica Microsystems) to detect fluorescent MV-infected cells. A Leica DM600B microscope equipped with a Leica DFC350 FX digital camera was used to acquire phase-contrast photomicrographs, which were captured using Leica FW4000 software (Lemon et al., 2007).
Ex vivo culture and virus infection of human corneal tissue.
Human corneal tissue was used in virus infection studies after informed consent and with approval from the Research Ethics Committee of Queen's University of Belfast. The recombinant virus rMVIC323EGFP (106 TCID50 ml−1) was spotted onto the apical, basolateral or peripheral rim of the ex vivo cornea tissue and incubated at 37 °C for 90 min. Following removal of the virus inoculum, tissue samples were maintained in DMEM supplemented with 2 % (v/v) FCS and 1 % (v/v) penicillin/streptomycin (Gibco-BRL/Invitrogen). Virus-infected explants were monitored daily for the appearance of fluorescent cells as described above.
Indirect immunofluorescence.
Primary bronchial epithelial cells were rinsed twice in PBS to remove the mucous present from the apical cell surface and fixed using 4 % (w/v) paraformaldehyde. Cells were permeabilized by incubation for 2 h at 37 °C with PBS containing 0.2 % (v/v) Triton X-100 to allow the detection of intracellular and viral antigens, which were examined by incubating cells with the appropriate antiserum diluted in PBS with 0.1 % (v/v) Triton X-100. Viral antigens were detected using a polyclonal antiserum (diluted 1 : 200) that recognizes the MV P protein (a gift from Sibylle Schneider-Schaulies, University of Würzburg, Germany). Goblet cells and ciliated epithelial cells were detected using anti-mucin 5AC (1 : 100; NeoMarkers) and anti-α-tubulin (1 : 1000; Sigma) monoclonal antibodies (mAbs). The integrity of cell membranes in virus-infected cells was examined using a polyclonal antibody (1 : 200; Invitrogen) to the tight junction marker ZO-I. Proliferating cells were detected using anti-Ki67 clone MIB-1 (1 : 100; Dako) mAb. Cells were incubated for 1 h at 37 °C in the presence of the primary antibodies, after which time unbound antibodies were removed by three successive PBS washes. Secondary antibodies (Molecular Probes) of chicken anti-mouse–Alexa Fluor 488, chicken anti-rabbit–Alexa Fluor 488 or goat anti-mouse–Alexa Fluor 568 (1 : 400) diluted in PBS with 0.1 % (v/v) Triton X-100 were applied to the cells for 1 h at 37 °C to detect bound primary antibodies. Cells were rinsed several times in PBS and in some instances counterstained with either propidium iodide (Sigma) or 4′,6-diamidino-2-phenylindole (DAPI) mounting medium (Vector) before mounting on glass slides using Citifluor (Amersham).
Immunohistochemical and immunofluorescence analysis of MV-infected tissues.
MV antigen in formalin-fixed human corneal rim tissue was detected using a mAb (diluted 1 : 1000) to the MV N protein (Harlan Seralabs) as described previously (McQuaid et al., 1998). Formalin-fixed tissue samples taken from rMVIC323EGFP-infected rhesus and cynomolgus macaques euthanized at 9 days p.i. were analysed as described previously (de Swart et al., 2007) to determine the extent of epithelial cell infection. Single- and dual-labelling immunofluorescence in MV-infected macaque tissue was performed using a mAb to the MV N protein (see above), with a polyclonal antibody to the tight junction marker ZO-1 (see above), using an anti-EGFP (1 : 400; Invitrogen) polyclonal antibody with a mAb (1 : 50; Novacastra) to the MV cellular receptor CD150, a mAb (1 : 100; Dako) to the proliferating cell marker anti-Ki67 clone MIB-1 or the epithelial cell-specific markers CAM 5.2 (1 : 2; Becton Dickinson) and AE1/A3 (1 : 100; Dako). Antigen-binding sites were detected with chicken anti-mouse–Alexa Fluor 488 or anti-rabbit–Alexa Fluor 488 and goat anti-mouse–Alexa Fluor 568 or anti-rabbit–Alexa Fluor 568 (1 : 500). In some instances, sections were counterstained with propidium iodide (Sigma) or DAPI mounting medium (Vector).
Confocal laser-scanning microscopy.
Photomicrographs of immunofluorescent-stained cell cultures and microtome-cut tissue sections were acquired using a Leica TCS SP2 confocal laser-scanning microscope as described previously (Duprex et al., 1999).
Acknowledgments
We would like to thank Paula Haddock and the staff of the Tissue Core Technology Unit and Ally Lyons at the Bioimaging Unit, Queen's University of Belfast, for expert technical assistance. We also wish to thank Bert Rima and Ingrid Allen for helpful discussions. This work was funded by the Medical Research Council Models of Disease Initiative (G0801001) and supported by a Wellcome Trust Equipment Grant. R. L. d. S. was supported by the VIRGO consortium, an innovative cluster approved by the Netherlands Genomics Initiative and partially funded by the Dutch Government (BSIK 03012).