Use of Confocal Immunofluorescence Microscopy To Localize Viral Nonstructural Proteins and Potential Sites of Replication in Pigs Experimentally Infected with Foot-and-Mouth Disease Virus

Abstract

Replication of foot-and-mouth disease virus in infected pig epithelium has been studied by immunofluorescence labeling of the viral nonstructural protein 3ABC and confocal microscopy. The results were correlated with viral RNA copy numbers in tissue samples from adjacent sites determined by reverse transcription-PCR (RT-PCR). Lesion formation was seen in the tongues and coronary band epithelia of infected pigs 2 days after infection. Viral replication was observed in cells of the epithelium of the tongue and coronary band but not in the associated stromal cells. Infected epithelial cells were present in the stratum spinosum, away from the lesion, with small lesions formed above the basement membrane. Viral replication was markedly reduced in tongue epithelium by day 3 postinfection but remained apparent in the coronary band tissue up to 5 days postinfection. These results were confirmed by the RNA copy number determined by RT-PCR.

The causative agent of foot-and-mouth disease (FMD) is a highly infectious virus which can infect a wide range of cloven-hoofed animals, including domestic cows, pigs, and sheep (3, 6). FMD virus (FMDV) is a member of the Picornaviridae and is a small (25-nm) nonenveloped virus with a genome consisting of a single strand of positive-sense RNA. Once the virus is inside a cell, the RNA is released from the capsid and translated into a single polyprotein which is cleaved into individual proteins. These consist of structural proteins that will form the viral capsid and nonstructural proteins which are involved in viral replication (8).

Clinical signs of the disease vary in severity between species, with cows and pigs showing greater severity of signs than sheep (6). The most obvious effects of FMDV infection are elevated temperature and the appearance of lesions on the tongue and epidermis adjacent to the hoof, namely, the coronary band and, in ruminants, interdigital skin. Lesions first appear in animals which have been infected by contact with infected animals within 1 to 5 days of contact, depending on the intensity of contact, the virulence of the virus, and the dose used to inoculate the pigs. Inoculated animals have viremia from around 1 day after inoculation, and it is assumed that the spread of the virus within such infected animals is by the general circulation.

The precise nature of the virus-cell interactions that give cell type specificity in vivo is not certain, although host cell integrins have been proposed as receptors for FMDV (11-16, 19-21). Both αvβ3 and αvβ6 integrins have been proposed as cell membrane receptors for FMDV in vivo, and we have recently demonstrated, with bovine epithelium, that αvβ6 is expressed at considerably higher levels than αvβ3 on epithelial cells. The highest levels of αvβ6 were detected in epithelial cells of the tongue and interdigital skin, which are sites most likely to show lesion formation in FMDV infection of cattle (P. Monaghan et al., unpublished data).

Previous studies of lesion formation have identified areas of increased cytoplasmic eosinophilia within the stratum spinosum of the cornified, squamous epithelium as the first stages of lesion formation. This is followed by fluid accumulation in the affected region to form a vesicle which may burst to give the large lesions characteristic of the disease (3, 6). While histological studies of lesion formation have shown that they occur within the epithelial cell layer, the precise route of infection, the cell types involved, and means of local spread are not clear. There have been attempts to study this using in situ hybridization methods (9, 10), but such methods have so far not had a sufficient morphological resolution and do not separate viral replication and accumulation. This is best achieved using immunolabeling and confocal microscopy of viral replication proteins. However, it has been exceedingly difficult to detect viral proteins resulting from viral replication within infected tissues despite many attempts. Away from the main lesion, small microscopic areas where the cells showed cytoplasmic changes have been proposed as early signs of viral replication, but without overt evidence of viral replication, these studies remain speculative (3, 6). Clearly, identifying sites of viral replication in the context of lesion formation and, particularly, individual infected cells will be important in developing our understanding of the early processes of virus replication and pathogenesis of FMDV in vivo.

We report the results of a combined confocal microscopic and reverse transcription-PCR (RT-PCR) study of foot-and-mouth disease in infected pig epithelium. Lesion formation was studied by immunofluorescence labeling for the presence of the nonstructural protein 3ABC in epithelial tissues from FMDV-infected pigs. The 3ABC protein, or rather its precursor P3 in processed form as 3A, 3BCD, 3AB, and 3CD, is expressed at high levels in infected cells, and reaction of intracellular antigens with antibodies raised against FMDV 3ABC protein is indicative of viral replication. The methods used have the sensitivity to detect viral proteins in infected cells and also have the advantage of retaining the morphology of the tissue to allow a full understanding of the processes taking place in infected epithelia. Samples of tissue adjacent to those used for microscopy were processed for RT-PCR to determine the copy numbers of FMDV RNA within the tissue, and these provide a comparative overall picture of virus replication in the tissues studied.

MATERIALS AND METHODS

Virus.

The virus inoculum was derived from a suspension of vesicular epithelium collected from a pig at Brentwood Abattoir, Essex, United Kingdom, during the 2001 epidemic in the United Kingdom, designated FMDV O UKG 34/2001 (2, 7), and passaged once in pigs, and a first pig-passaged inoculum was prepared from foot lesion epithelia. The titers of this stock virus were 10^7.2 and 10^6.2 50% tissue culture infective doses (TCID₅₀) per ml in BTY and IB-RS-2 cells, respectively (5). Each inoculated animal received approximately 10^5.6 TCID₅₀ (BTY) in 0.25 ml of a 1:10 dilution in Eagle's minimal essential medium with 20 mM HEPES buffer and antibiotics.

Animals.

The 13 pigs described in this report were from a larger experiment involving 40 Landrace cross-bred Large White pigs weighing between 20 and 30 kg and housed in biosecure animal buildings. Details of the larger experiment will be described elsewhere (C. Murphy, M. Quan, and S. Alexandersen, unpublished data). Briefly, four of the pigs were sacrificed as normal controls, and the other 36 pigs were randomly divided into six separate groups of 6 pigs, of which 3 pigs in each group were inoculated and the other 3 pigs kept as direct contacts. Two inoculated and two direct-contact pigs were selected at random and sacrificed at predetermined time points. The pigs described in this report included one control pig and two inoculated pigs sacrificed at time points 6 h and 1, 2, 3, 4, and 5 days post inoculation (p.i.), as shown in Table 1. Pigs were inoculated in the heel pad as described previously (1, 2). This method of infection, rather than using contact with an infected animal(s), was chosen for this experiment as it allows the time postinfection to be unambiguously known. The animals were examined clinically for signs of FMD, and rectal temperatures were recorded daily.

TABLE 1.

RT-PCR results

Time postinoculation	Animal	Viremia RNA copies (FMDV of serum)/ml	FMDV RNA copies/g of tissue
			Coronary band	Tongue	Soft palate
0	VB05	0a	0	0	0
6 h	VA92 and VA93	0	0	0	0
1 day	VA85	108	108	107.5	105.4
	VA86	108	106.5	107	107
2 days	VA79	109.5	1011.5	1010.7	108.3
	VA80	109.8	1011.5	1010.7	108.3
3 days	VA73	108.1	1010.4	108.6	106.7
	VA74	107.3	109.4	107.2	106.7
4 days	VA67	103.7	105.3	107.0	106.4
	VA69	103.9	109	108.3	106.8
5 days	VA81	0	1010	106.4	104.8
	VA87	0	108.3	107.0	104.8

^a0, below detectable limits (detection limits of around 10^2.4 copies of FMDV RNA per ml of serum and around 10^4.2 copies per g of tissue).

Sample collection.

Although the results are not to be described in detail here, blood samples and nasal swabs were taken daily for the first period after inoculation. The blood samples were transported to the laboratory and allowed to clot for at least 1 h at room temperature. The samples were kept at 4°C, and then the serum was separated within 24 h before storage at −80°C. For RNA extraction, an aliquot of serum was thawed and 2 volumes of serum added to 3 volumes of lysis buffer (Roche lysis solution) and stored at 4°C until nucleic acid extraction and subsequent analysis. Swabs were placed in 1 ml of TRIzol (Life Technologies, Paisley, United Kingdom) and stored at −80°C.

For RT-PCR, tissue samples were collected postmortem from pigs randomly selected and sacrificed at the indicated time after inoculation and were put immediately into RNAlater (Ambion, United Kingdom) and then stored at −20°C until required (results to be described in detail elsewhere). For microscopic analysis, samples of the heel pad inoculation site were obtained at 6 h p.i. and samples of tongue, soft palate, coronary band, interdigital skin, and flank skin were obtained at days 0 (control), 1, 2, 3, 4, and 5 p.i. (Table 1). Where a lesion could be detected macroscopically, the sample was taken to include both lesion and surrounding apparently normal skin. The samples, approximately 1 to 2 cm by 0.5 to 1.0 cm, were immediately placed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 90 to 120 min and then transferred to PBS for storage at 4°C.

For sectioning, the method was essentially as described previously (17, 18). The tissues were trimmed to an oblong shape if needed and then attached to the holder of a vibrating microtome (Leica Microsystems, Milton Keynes, United Kingdom) with superglue. They were sectioned and processed under PBS. The short fixation time used in this study provides the optimum combination of minimal damage to antigens with preservation of tissue structure. Sections (70 μm) were cut as near to 90° to the surface of the tissue as possible. The sections were handled with paint brushes trimmed to one bristle and immunolabeled in 150-μl droplets of reagent. The sections were permeabilized for 60 min in 0.1% Triton X-100, nonspecific labeling was blocked with PBS containing 0.5% bovine serum albumin (PBS-BSA) overnight, and then the sections were transferred to primary antibody for 90 min. Labeling for FMDV replication was with a rabbit antibody recognizing the nonstructural protein 3ABC (IAH) diluted 1:1,000 in PBS-BSA. Following 10 2-min washes, the bound primary antibody was detected with species-specific antibodies conjugated to Alexa 488 (Molecular Probes, Cambridge Biosciences, Cambridge, United Kingdom) diluted 1:200 in PBS-BSA for 90 min. Ten 2-min washes were followed by 30 min in DAPI (4′,6′-diamidino-2-phenylindole) (Sigma, Poole, United Kingdom) diluted 1:5,000 in PBS and mounting in Vectashield (Vector Laboratories, Peterborough, United Kingdom). Coverslips were sealed with nail varnish.

Control sections were labeled in the absence of primary antibody or with an irrelevant primary antibody. The sections were imaged in a Leica SP2 confocal microscope equipped with 405-, 488-, and 568-nm laser excitation (Leica Microsystems, United Kingdom). Differential interference contrast (DIC) images were also collected.

Quantitative RT-PCR.

Although the vibrating microtome sections provide relatively large pieces of tissue for analysis, there remains a risk of sampling error. In order to make a correlation between immunolabeling and overall viral replication, we determined the quantity of FMDV RNA in extracts of total nucleic acid from blood and tissue samples by quantitative real-time RT-PCR. The assay method detects FMDV RNA in the 5′ untranslated region and has been described in detail elsewhere (5, 7, 26-28). All extractions involved 0.200 ml of sample, the nucleic acids were finally eluted in a volume of 0.050 ml, and all estimations included standard reactions based on in vitro-transcribed, molecularly cloned FMDV RNA, using samples with a known content of FMDV genome equivalents (31). The method is influenced minimally by sample type (4-6, 22, 26, 28, 29).

RESULTS

Clinical signs and virus load in serum and tissues.

Virus load is given as the number of virus genomes per ml serum or per gram of RNAlater-stabilized tissue. Comparative titrations on serum and tissue samples indicate that for most isolates of FMDV, one BTY TCID₅₀ equals roughly 10² to 10³ genomes (5).

Briefly, an increase in body temperature, viremia, and early local lesions were detected in the inoculated pigs from around 24 h after inoculation and reached a peak at 48 h after inoculation with severe, generalized lesions. Clinical disease was severe and characterized by lameness/reluctance to stand or walk, reduced or no interest in feed, and severe vesicular lesions along the coronary bands and heel pad area and sometimes on the tongue and snout. The details for the pigs examined here will be given together with the microscopic findings below.

Microscopic findings.

No vesicular lesions were observed in the soft palate, interdigital skin, and flank skin, and no labeling was detected in samples from these regions at all stages of the experiment. The viral load in the soft palate samples as measured by quantitative RT-PCR is reported below. The viral loads in interdigital skin and flank skin were not measured for these samples; however, previous experiments indicate that the viral loads in these tissues are in general only 0.1 to 1% of the levels in skin with vesicular lesions and never exceed 10⁸ FMDV copies per gram of tissue (S. Alexandersen, M. Quan, and C. Murphy, unpublished findings).

Day 0 (control), 6 h, and day 1.

Macroscopically, no lesions were observed in the control pig. In the pigs at 6 h p.i. the inoculation sites were visible on the heel pads as needle tracks but without any vesicular lesions. In the pigs at 1 day p.i. (VA85 and VA86) local vesicular lesions were observed at the sites of inoculation on the heel pads, but no generalized lesions were evident and their body temperatures were not elevated (38.6°C and 38.9°C, respectively). No labeling was detected by confocal immunofluorescence microscopy at any location in tissues from the control animal or animals at 6 h or 1 day p.i. This does not prove that there were no infected cells within the coronary band or tongue, just that the lack of lesions made the selection of the site of the epithelium for microscopy an arbitrary process.

Examination of coronary band epithelia, tongue, and soft palate for FMDV RNA showed that all the day 0 samples and the 6-h-p.i. samples were negative (below the detection limit of around 10^4.2 copies per gram), while the 1-day-p.i. samples contained from 10^5.4 to 10⁸ copies of FMDV RNA per gram of tissue, except for the inoculated heel pad tissue from pig VA85, which contained around 10^10.7 copies of FMDV RNA per gram. Both VA85 and VA86 had viremia of around 10⁸ copies of FMDV RNA per ml (Table 1).

Day 2.

Macroscopically, the pigs at 2 days p.i. (VA79 and VA80) had generalized lesions involving all four feet and the tongue, and their body temperatures were elevated to above 40°C.

Lesions of the tongue were well developed, and samples were taken which contained both lesion and adjacent apparently uninvolved tissue. In the tongue, the epithelium contains many cell layers, from the basement membrane through the spinous layer to the upper cornified layer, which has specialized “hairs” on the surface. The basal region of the epithelium has deep invaginations (papillae), which are long processes of the dermis consisting of capillaries and stromal cells and extend almost to the outer cornified layer of the epithelium. Immunofluorescence labeling of the edge of the lesion showed that FMDV 3ABC protein was present in most layers of the tongue epithelium, extending from the basal area up to close to the cornified layer (Fig. 1a and b). Large numbers of infected cells were present in the region of lesion, but the infected cells were detected only within the cells of the epithelium. No positively labeled stromal cells were seen. The lesion on the tongue consisted of a split within the epithelial layer which had occurred some way up away from the basal level in the region of the stratum spinosum. A small number of FMDV-positive cells remained below the lesion. This is more clearly seen in Fig. 1c and d, which show the center of a lesion. The cells of the epithelium separated from the dermis (upper region in Fig. 1) are almost all FMDV positive, and below the lesion, clusters of FMDV-positive cells are also seen in the epithelium remaining in contact with the dermis. The tissue architecture in this region was distorted, and from this time postinfection onward, the whole area of the lesion along with its associated dermal tissue was infiltrated with a massive influx of small cells with a high nuclear/cytoplasmic ratio. These were interpreted as being lymphocytic cells.

FIG.1.

Tongue epithelium from an animal at 2 days p.i. (pig VA79), processed as described in the text. Sections were labeled with antibodies against FMDV 3ABC, detected with species-specific fluorescent conjugate. FMDV replication is indicated by labeling in green, nuclei are blue, and morphology is shown in DIC images (b and d). (a and b) Area at the edge of a lesion showing labeling in most of the epithelium. The lesion is formed part way up the epithelium above the lowest level of the basal cells. (c and d) Region at the center of a lesion. The upper tissue had separated from the epithelium, with almost all cells showing viral replication. Below the lesion, areas of labeling indicate infected epithelial cells. (e) Section of epithelium away from the lesion shows the spread of the lesion laterally through the epithelium. (f) Higher power of the extreme edge of infected area. In one area just one cell shows labeling in the cytoplasm (arrow). Bars, 200 μm (a, b, and e), 80 μm (c and d), and 40 μm (f).

The extent of infection as measured from the basal region of the epithelium towards the cornified layer was maximal in the area of the lesion and extended almost to the keratinized layer (Fig. 1a and b), reducing with distance away from the lesion. At the edge of the lesion the epithelial cells between the negative papillae showed small lesions (Fig. 1e). Whether these lesions form from cell lysis of infected cells or fluid increase in the infected area was not clear. At the limits of the infection away from the lesion, or at least the limits of our ability to detect infected cells, labeling consisted of a few relatively normal-appearing positive cells in the stratum spinosum region of the epithelium (Fig. 1f).

While there were areas of infection where the basal cells were clearly labeled (Fig. 2a and b) in the majority of lesions observed, the basal cells were not, at least at this stage, infected (Fig. 2c). Whether any or all of these smaller lesions were interlinked is unclear, as the papillae gave the impression of interrupting the infected area. Our interpretation of the images is that it is likely that the infected areas formed interlinked lesions rather than a group of separate lesions.

FIG.2.

Tongue and coronary band epithelium from an animal at 2 days p.i. (pig VA79), processed as described in the text. Sections were labeled with antibodies against FMDV 3ABC, detected with species-specific fluorescent conjugate. FMDV replication is indicated by labeling in green, nuclei are blue, and morphology is shown in DIC image. (a and b) Occasionally a small lesion was seen away from the center of the main lesion where basal cells were positively labeled. (c) In a region similar to that in panel a, the more common observation was labeling in the spinous cell layer; the basal cells were uninfected (arrow). (d) In the coronary band almost all epithelial cells were strongly labeled. (e) Interdigital skin showed no positive labeling. (f) Flank skin showed no positive labeling Bars, 40 μm.

The cells within the coronary band lesion showed similar patterns of infection as seen in the tongue lesions, with large areas of positively labeled epithelium (Fig. 2d). In contrast, both the flank skin and interdigital skin were negative for FMDV labeling (Fig. 2e and f).

Examination of coronary band epithelia, tongue, and soft palate for FMDV RNA showed that the 2-day-p.i. samples (VA79 and VA80) contained from around 10^8.3 copies of FMDV RNA per gram for the soft palate tissue up to 10^10.7 to 10^11.5 copies per gram for the tongue, coronary band, and heel pad. Both VA 79 and VA80 had viremia of around 10^9.5 to 10^9.8 copies of FMDV RNA per ml (Table 1).

Day 3.

Macroscopically, the pigs at 3 days p.i. (VA73 and VA74) had severe generalized and ruptured lesions involving all four feet and tongue, and their body temperatures were elevated to above 40°C. By confocal immunofluorescence microscopy, the labeling intensity in the region of the lesion in the tongue was dramatically reduced compared to that for the day-2-p.i. animals, and positive cells were only rarely detected at the edge of the lesion. There was a clear dividing line at the edge of the lesion between the necrotic tissue of the lesion and the apparently normal residual epithelium (Fig. 3a and b).

FIG.3.

Tongue and coronary band epithelium from animals at 3 (pig VA73), 4 (pig VA67), and 5 (pig VA81) days p.i., processed as described in the text. Sections were labeled with antibodies against FMDV 3ABC, detected with species-specific fluorescent conjugates. FMDV replication is indicated by labeling in green, nuclei are blue, and morphology is shown in DIC images. (a and b) Edge of tongue lesion at 3 days p.i. There is no labeling for FMDV replication, and there is a clear demarcation between the lesion and the residual normal epithelium. (c and d) Coronary band from pig at 3 days p.i. The basal area of the epithelium is strongly labeled (c), and the lesion has caused major destruction of tissue architecture (d). (e) Coronary band epithelium from pig at 4 days p.i. The ruptured cornifed layer is seen at the top, and significant numbers of labeled cells are present within the lesion. (f) Coronary band epithelium from pig at 5 days p.i. Despite extensive necrosis, there are positively labeled cells in the lesion. Bars, 200 μm (a, b, c, and d) and 80 μm (e and f).

In the coronary band, the majority of the cells within the lesion were strongly positive, and these infected cells covered the whole of the epithelium from the basal region towards the cornified layer. The structure of the epithelium within the lesion was badly disrupted, and a large number of small lymphocyte-like cells were present (Fig. 3c).

Examination of coronary band epithelia, tongue, and soft palate for FMDV RNA showed that the 3-day-p.i. samples (VA73 and VA74) contained around 10^6.7 copies of FMDV RNA per gram for the soft palate tissue, while the tongue samples contained from 10^7.2 (VA74) to 10^8.6 (VA73) copies per gram and the coronary epithelia up to 10^9.4 (VA74) and 10^10.4 (VA73) copies per gram (Table 1). VA73 had viremia of around 10^8.1 and VA74 of around 10^7.3 copies of FMDV RNA per ml.

Day 4.

Macroscopically, the pigs at 4 days p.i. (VA67 and VA69) had severe generalized and ruptured lesions involving all four feet and tongue, and their body temperatures, which had been 40.0 and 39.8°C on days 2 and 3 p.i., had dropped to 39.6 and 39.3°C, respectively.

In the region of the tongue lesion there was no clear labeling for FMDV by confocal immunofluorescence microscopy. In the coronary band lesion, there were still positively labeled cells within the lesion. (Fig. 3e).

Examination of coronary band epithelia, tongue, and soft palate for FMDV RNA showed that the 4-day-p.i. samples (VA67 and VA69) contained from around 10^6.4 to 10^6.8 copies of FMDV RNA per gram for the soft palate tissue, while the tongue samples contained from 10^7.0 (VA67) to 10^8.3 (VA69) copies per gram and the coronary epithelia up to 10^9.0 copies per gram (Table 1). VA67 had viremia of around 10^3.7 and VA69 of around 10^3.9 copies of FMDV RNA per ml.

Day 5.

Macroscopically, the pigs at 5 days p.i. had different severities of lesions. Pig VA81 had severe generalized and ruptured lesions involving all four feet and tongue and a body temperature of 39.6°C, while pig VA87 had lesions only on the inoculated foot and on the tongue and had a body temperature of 39.5°C.

By confocal immunofluorescence microscopy examination, the lesion on the tongue of pig VA81 showed no positive labeling. The lesions on the coronary bands of both animals showed an overall picture of complete loss of epithelial architecture with mostly necrotic epithelium. However, there was still extensive labeling in the lesion, and this was clearly associated with individual cells (Fig. 3f). Both tongue and coronary band lesions contained a large number of small cells with round nuclei that were considered to be lymphocytes. In the tongue lesion, there was, however, a hint of a return of a papillary structure, which may indicate the emergence of a few uninfected basal epithelial cells from within the stroma below the lesion.

Examination of coronary band epithelia, tongue, and soft palate for FMDV RNA showed that the 5-day-p.i. samples (VA81 and VA87) contained from around 10^4.8 copies of FMDV RNA per gram for the soft palate tissue, while the tongue samples contained from 10^6.4 (VA81) to 10^7.0 (VA87) copies per gram and the coronary epithelia contained 10^8.3 (VA87) and 10¹⁰ (VA81) copies per gram (Table 1). Neither VA81 nor VA87 had detectable viremia (detection limit around of around 10^2.4 copies of FMDV RNA per ml).

DISCUSSION

These results demonstrate, for the first time, the cellular replication of FMDV in infected animals. Despite the fact that the two systems described here detect different aspects of FMDV replication—immunolabeling, which detects viral nonstructural protein 3ABC, and RT-PCR, which provides numerical data on actual viral RNA copy numbers—there is a strong correlation between the two sets of data. Although some of the RNA detected in the tissues could have come from the blood present, the data in Table 1 clearly demonstrate higher levels of viral RNA in certain tissues than in blood (Table 1). This suggests that a significant replication of FMDV occurs in coronary band and tongue epithelia, peaking at day 2, which is consistent with the findings using immunolabeling for 3ABC. In the RT-PCR data, the time course of the infection as judged by viral RNA both in the blood (viremia) and within most tissues peaked at day 2 and declined thereafter, albeit faster in the blood than in the tissues. The one exception to this was the coronary band, where the viral RNA values remained high for longer. The results for the immunolabeling for the presence of the nonstructural protein 3ABC indicated that by day 2, generalized lesions had started to form in both tongue and coronary band epithelium, and there was a clear region of infected cells extending from the edge of the lesion. However, within a further 24 h, viral replication appeared to have slowed down and was below the detection limit in the tongue. In contrast, in the coronary band epithelium, cells clearly labeled with the antibody were seen up to day 5 p.i. despite the lesion containing large areas of necrosis.

Generalized skin lesion formation begins after the animals become viremic and it must be assumed that the susceptible cells in the cornified squamous epithelia become infected by the virus traversing the basement membrane. The mechanism for this is not known, and it is not clear if this is an active or passive process (6). Lesion formation takes place rapidly within infected animals, but this is not too surprising as it has been reported that the cytopathic effect of FMDV can be detected in BHK-38 cells, which are a particularly susceptible tissue culture cell line, as soon as 1.5 h p.i. (17). While this is an extreme case, the doubling rate of FMDV in pigs has been estimated to around 2.3 h (Quan and Alexandersen, unpublished). Extrapolation of data from tissue culture to the in vivo situation is clearly an uncertain exercise and supports the need for experimental information as described here.

The targeting of FMDV exclusively to the epithelial cells in the tongue and coronary band epithelium is clear from this study; however, the mechanism is not clear. The outer capsid of FMDV consists of 60 copies of each of four proteins, VP1, VP2, VP3, and VP4. The external loop of VP1 has a highly conserved RGD motif, and this is a well-recognized ligand for the class of cell adhesion and signaling molecules known as integrins (11-16, 19-21). These are complex molecules which consist of heterodimers of two integrin families, α and β, and both αvβ6 and αvβ3 have been proposed as receptors for FMDV, thus providing a possible mechanism for the cell type specificity observed. Indeed, tissue culture studies have indicated that FMDV binds to and is internalized into susceptible cells by means of these cell surface integrin molecules (12-16). The pattern of expression of integrin molecules αvβ6 and αvβ3 observed within the epithelium of bovine tissues is in line with the results reported here (Monaghan et al., unpublished data), but further work will be needed to determine if the integrin expression pattern of porcine epithelium is similar to that seen in bovines.

Our interpretation of the data presented here suggests that the first site of infection in the epithelium is either a basal epithelial cell on the basement membrane or an epithelial cell further “up” the epithelium within the stratum spinosum. Certainly in infected regions away from an active lesion, the infected cells do not always include the basal cells, and this location of infection above the basement membrane correlates well with the formation of the lesion some way above the lower limit of the basement membrane. This lesion formation above the basement membrane left patches of epithelial cells below the lesion. While many of these cells were positively labeled, it was not possible to determine if all the epithelial cells became infected. The apparent beginnings of restoration of tissue structure in the day 5 p.i. tongue sample might suggest that a small number of epithelial cells escape infection and act as a source of new epithelial cells or, alternatively, that the FMDV infection may not necessarily be cytolytic in basal cells.

Away from the lesion, the infected cells were seen around the epithelial papillae, and at no stage was any cell seen outside the basement membrane which labeled positive for the presence of 3ABC. This may indicate that the spread of the virus within the epithelium is predominantly from epithelial cell to cell, and this may be encouraged by cellular stress or trauma (6).

The results of the microscopic examination and the RT-PCR are in agreement on the rapid decline in FMDV replication in the tongue compared with the coronary band. On day 2, when both tongue and coronary band show high levels of 3ABC expression, the RT-PCR gives values of 10^10.7 to 10^11.5 viral genomes per gram. In contrast, on day 3 p.i., the microscopy of the coronary band showed strong labeling, with almost no labeling present in the tongue lesion. The RT-PCR of coronary band showed similar levels to the day 2 samples (10^9.4 to 10^10.4), whereas the tongue levels were 10^7.2 to 10^8.6. Why the viral replication appears to decline more slowly in coronary band lesions is not clear. However, one possibility is the influence of the host response to the infection. The edge of the lesion in the tongue where it abuts the remaining normal epithelium in particular was very distinct, and it may be that the onset of host defenses limits lesion growth by inhibition of viral replication and that these host responses are more acute in the oral region. Certainly the lymphatic system is well developed in this region compared to epithelium in the hoof region. In addition, there is some evidence that FMDV antigen and genome are maintained longer in foot lesions than in tongue lesions, which may be due to various mechanisms, e.g., a prolonged period of active replication, increased stability, decreased clearance, or simply differences in cell renewal rate (6, 23-25, 30).

The soft palate has been proposed as a possible route of entry of the virus, and the failure to detect viral replication in the soft palate may have a number of causes. While it may be a matter of detection sensitivity, in that no cells expressed sufficient 3ABC to be detected, it may be that infection is patchy and the sections prepared for confocal microscopy did not contain infected cells. Comparison of the RT-PCR data and the immunolabeling results as a whole does indicate that the microscopic detection limit is around 10⁸ copies per g of tissue, although coronary band tissue of pig VA67 was positive by immunolabeling and contained only 10^5.3 genomes per gram. This may be due to a focal distribution of infected cells. As already discussed, these two systems detect different aspects of viral replication, but on this basis, the levels in the soft palate were not significantly above the detection limit. Another possibility is that the RNA detected by RT-PCR could result from passive uptake of virus within the tissue or, alternatively come from a cell type other than the epithelium itself, such as mucus glands or nasal-associated lymphoid tissue.

In all cases with lesion formation there was a massive lymphocytic infiltrate throughout the lesion both above and below the basement membrane, and further studies will follow this process in more detail.