Activation of small ruminant aortic endothelial cells after in vitro infection by caprine arthritis encephalitis virus

Research in Veterinary Science 2000, 69, 225–231 doi:10.1053/rvsc.2000.0413, available online at http://www.idealibrary.com on

Activation of small ruminant aortic endothelial cells after in vitro infection by caprine arthritis encephalitis virus C. LE JAN*, T. GREENLAND, F. GOUNEL, S. BALLEYDIER, J.F. MORNEX Laboratoire Associé I.N.R.A. de Recherches sur les Lentivirus chez les Petits Ruminants, INRA U754, Ecole Nationale Vétérinaire de Lyon, 1 Avenue Bourgelat, Marcy l’Etoile 69280, France SUMMARY Small ruminants infected by the lentiviruses caprine arthritis–encephalitis virus (CAEV), originally isolated from a goat, or maedi–visna virus, originally from sheep, typically develop an organising lymphoid infiltration of affected tissues. This could reflect modulation of the migration pattern of lymphocytes in infected animals. Possible active contribution by vascular endothelial cells was investigated using an in vitro model. Low-passage cultured ovine aortic endothelium proved susceptible to productive infection by CAEV without significant cytotoxicity. Infected endothelial cells maintained expression of endothelial markers, increased MHC class I antigen expression and initiated expression of the adhesion molecule VCAM-1 and, at a late stage, MHC class II antigens. Infected endothelial cells showed a two-fold increase in binding capacity for sheep peripheral blood leucocytes over uninfected controls. Such events could contribute to the tissue distribution of lymphoid cells and local immune responses in lentiviral infections of small ruminants. © 2000 Harcourt Publishers Ltd

THE small ruminant lentiviruses, classically described as maedi–visna virus in sheep and caprine arthritis– encephalitis virus (CAEV) in goats, cause lifelong persistent infections in their hosts. Their host specificity is complex and CAEV-like virus is frequently isolated from natural infections of sheep (Leroux et al 1995a). Infection is often clinically inapparent, but infected animals usually show a greater or lesser degree of lymphoid infiltration in target organs, for instance lungs or mammary glands, which show lymphoid hyperplasia and the generation of structured lymphoid follicles. The infection follows an irregular course, with quiescent periods when isolation of virus from blood monocytes or tissue macrophages is difficult, alternating with active periods when the presence of proviral DNA in a proportion of circulating monocytes is associated with active infection of macrophages in the target tissues (Zink and Johnson 1994, Pépin et al 1998). Alteration of the pattern of distribution of mobile cells in body tissues implies changes in cellular adhesion patterns which precede migration. For blood leucocytes these changes could affect either the lymphoid cells themselves or endothelial cells at the target site. Human monocytes infected in vitro with HIV-1 express augmented β2 integrins and metalloproteases (Lafrenie et al 1996), leading to increased adhesion to endothelial cells and transmigration. Also, endothelial cells can be infected by many viruses which could alter their adhesive properties. Human endothelia may be infected in vitro by coronaviruses (Cabirac et al 1995), cytomegalovirus (Woodroffe et al 1993), Epstein–Barr herpes virus (Jones et al 1995) or HIV-1 (Poland et al 1995). HIV-1 (Bagasra et al 1996) and cytomegalovirus (Sinzger et al 1995) can be found in *Corresponding author. 0034-5288/00/030225 + 07 $35.00/0

endothelial cells of infected patients, and simian immunodeficiency virus (SIV) can infect simian endothelial cells in vitro and is present in the endothelia of infected monkeys (Mankowski et al 1994). Ruminant endothelial cells have been infected in vitro with bluetongue virus (Russell et al 1996) and sheep endothelial cultures express reverse transcriptase after infection by maedi–visna virus (Craig et al 1997). Endothelial cells can respond to viral infection by activation, as shown by the increased expression of adhesion molecules and MHC class II antigens following in vitro infection of human endothelial cells with West Nile virus (Shen et al 1997). In addition, activated endothelial cells are competent of expressing a variety of cytokines and chemokines including IL-1, IL-6 and IL-8, susceptible of influencing leucocyte behaviour (Chen and Manning 1996). In order to investigate the influence of viral infection on cellular distribution in French sheep commonly infected with a CAEV-like virus (Leroux et al 1995a), the authors examined the capacity of sheep aortic endothelial cells to support infection by CAEV in vitro and the phenotype of the infected cells with respect to their interaction with normal sheep peripheral blood leucocytes. Aortic cell cultures were used to establish the principle and modalities of interaction between infected endothelia and lymphocytes in sheep rather than cultures established from microvascular tissues (Craig et al 1997), which show a variable response depending on the tissue of origin.

MATERIALS AND METHODS Aortic endothelial cell cultures and virus Abdominal aortas from 3-month-old lambs were collected at a slaughterhouse directly into Eagles minimum © 2000 Harcourt Publishers Ltd

226

C. Le Jan, T. Greenland, F. Gounel, S. Balleydier, J.F. Mornex

essential medium (MEM; Gibco, Life Technologies, CergyPontoise, France) supplemented with 2 per cent fetal calf serum (FCS; Gibco) and containing penicillin (200 U ml–1; Boehringer-Mannheim, Meylan, France), streptomycin (200 µg ml–1; Boehringer-Mannheim) and amphotericin B (5 µg ml–1; Gibco). Cells were collected by gently scraping the interior surface of the aorta with a scalpel blade and were seeded into plastic culture flasks (Falcon BD, Grenoble, France) previously coated with a 1 per cent solution of bovine skin gelatin (Sigma, St Quentin Fallavier, France) in a medium consisting of equal volumes of RPMI 1640 and 199 media (Gibco) supplemented with 20 per cent heat-inactivated FCS, 1 per cent each of sodium pyruvate (100 mM; Gibco) and L-glutamine (200 mM; Boehringer), endothelial cell growth factor (ECGF, 100 µg ml–1; Boehringer), heparin (50 µg ml–1; Sigma), penicillin (100 U ml–1), streptomycin (100 µg ml–1) and amphotericin B (2.5 µg ml–1). After reaching confluence the cells were subcultured in the same medium containing 10 per cent FCS. Cork strain CAEV was propagated on goat synovial membrane (GSM) cell cultures grown in MEM medium containing a minimum of FCS at the time of viral collection. Stocks were titrated by limiting dilution, and cells infected by these stocks showed no evidence of mycoplasma infection and were negative for parainfluenza virus 3 and for bovine viral diarrhoea virus by immunofluorescence or by alkaline phosphatase/anti-alkaline phosphatase (APAAP) immunocytochemistry (Cordell et al 1984) using 1B6 or D89 primary antibodies (VRMD; Pullman, WA, USA), respectively. Supernatants of uninfected GSM cells were used for mock infection controls.

treated with equivalent dilutions of culture fluids from uninfected GSM cells, or heat-inactivated virus (56°C for 30 minutes) in place of the infectious virus.

Characterisation of endothelial cells

Leucocyte adhesion to endothelial cell monolayers

The endothelial nature of the cultured cells was confirmed by treating acetone/ethanol (50 per cent v/v)-fixed cells (20 minutes at 4°C) with a 1/200 dilution of a polyclonal rabbit antiserum to human von Willebrand factor (vWF), which cross-reacts with ovine vWF, followed by detection of antibody binding by Envision DAB (DAKO, Trappes, France). As a control, similar preparations were treated with a 1/10 dilution of a mouse monoclonal antihuman α-actin (Boehringer), which recognises ovine α-actin. Antibody binding was detected by APAAP.

Aortic endothelial cells were grown in six-well plates; two wells were left untreated as controls and the other four were infected with CAEV (Cork) at 1 tissue culture infected dose 50 per cent (TCID50) per cell or treated with rTNF-α (500 U ml–1) for 18 hours before washing and medium replacement. After 2 to 6 days in culture peripheral blood leucocytes were added from CAEV antibody-negative sheep, purified from jugular vein blood by centrifugation on Histopaque density 1.077 (Sigma). Mononuclear cells were collected from the interface, washed and counted. Each well of cultured endothelial cells was incubated with 6 × 106 blood leucocytes for 1 hour at 37°C, then rinsed four times with PBS and incubated for 30 minutes at 37°C with 1 ml EDTA at 4 per cent in PBS (Saukkonnen et al 1997). Cells were rinsed off, pooled according to treatment and counted in Mallassez chambers.

In vitro infection of endothelial cell cultures Endothelial cell cultures were infected with CAEV at 0.5 or 1.0 TCID50 (50 per cent tissue culture infectious doses) per cell overnight at 37°C in 5 per cent CO2. The cells were then washed and fresh medium was substituted. Supernatants were titrated for CAEV by inoculation of sequential dilutions onto susceptible goat synovial fibroblasts grown in 96-well microtitre plates (Costar, Brumath, France). Syncytia were scored after staining with May–Grünwald Giemsa and the titre was calculated as described by Reed and Muench (1938). Viral p30 antigen was detected using the CAEP5A1 monoclonal antibody (VMRD) and APAAP detection; simultaneous vWF expression was detected by indirect immunofluorescence using fluoresceine-conjugated anti-mouse IgG (1/40; Boehringer) or rhodamine conjugated anti-rabbit IgG (1/10; Boehringer) as required. Negative controls were

Stimulation of endothelial cell cultures The capacity of human recombinant tumour necrosis factor-α (rTNF-α) to stimulate ovine endothelial cells was established by treating fifth passage cultures in six-well Costar plates with concentrations of 50 to 1000 U ml–1 human rTNF-α (Sigma) for 18 hours at 37°C. Cells were then washed and replaced in fresh medium for 3 or 4 days’ further culture when they were trypsinised and treated in the unfixed state with anti-VCAM-1 (HAE2-1, 1/1000; kindly provided by T. F. Tedder, Duke University Medical Center, Durham, USA) and anti-MHC class I (B5C, 1/75; VRMD) or class II (CATB2A, 1/75; VRMD) antigens. Membrane antigen expression was quantified by flow cytometry (Facscan; Becton-Dickinson). Qualitative expression was estimated by APAAP immunocytochemistry of cultures in 96-well plates. Cells were similarly evaluated for the expression of the same antigens following infection with CAEV or treatment with control supernatants. For double-labelling experiments on infected cells, cultures were harvested into PBS by gentle scraping with a rubber policeman, deposited onto glass slides by cytocentrifugation and permeabilised by fixation in acetone/ethanol (50 per cent v/v). Mouse antibodies to VCAM-1 or CAEV p30 were detected by fluorescein-conjugated goat anti-mouse IgG immunoglobulins (Boehringer), and rabbit antibodies to vWF by rhodamine-conjugated goat anti-rabbit IgG f(ab′)2 fragment (Boehringer).

Statistical testing Comparisons between infected, induced and control cells were tested for significance by Student’s t test. Probabilities were considered significant if less than 0.05.

RESULTS Low-passage cultures of ovine aortic endothelial cells showed a typical cobblestone morphology (Fig 1A), but

In vitro endothelial cell activation by CAEV

A

FIG 1:

227

B

Monolayer cultures of aortic endothelial cells (× 200). (A) Primary aortic endothelial cell culture; (B) aortic endothelial cell culture, fifth passage.

became more elongated at higher passage (Fig 1B). Intracytoplasmic von Willebrand factor was present in over 90 per cent of cells in primary or subsequent cultures, indicating an endothelial character, but less than 5 per cent of cultured cells expressed α actin, an indicator of a myoepithelial origin (not shown). On infection with CAEV, at day 0 the cells expressed neither viral p30 antigen nor VCAM-1 (Fig 2A), but by day 3 post-infection when viral p30 antigen began to be expressed (Fig 2B), VCAM-1 was clearly detectable in a proportion of the cells (Fig 2C). By day 5 most cells were positive for viral antigen and a few were heavily stained (Fig 2D); VCAM-1 expression was maintained (not shown). Viral p 30 antigen expression peaked around day 8, with heavy staining of some 90 per cent of cells and visible cytopathic effect, whereas VCAM1 expression declined somewhat concomitantly with the cellular damage (not shown). The fragility of infected monolayers prevented direct double-labelling experiments so cells were harvested by gentle scraping with a rubber policeman and deposited onto slides by cytocentrifugation. Despite loss of cell architecture compared with the original cultures, double immunofluorescence investigations of permeabilised cells indicate co-expression of vWF and VCAM-1 (Figs 3A, B) and of vWF and CAEV p30 antigen (Figs 3C, D) in the infected cultures. Traces (103 ml–1) of infectious virus were present in the supernatant as early as 1 day post-infection, and titres rose to 106 to 107 ml–1 by day 10 (Fig 4). Syncytia were only rarely observed in the infected cultures; zones of cell lysis were not seen. As shown in Figure 3B, VCAM-1 was strongly expressed by infected cells by day 5 post-infection. The amount of expression of VCAM-1 and MHC class I antigens was compared on cells infected with CAEV or induced with TNF-α, as compared with controls, using mean fluorescence intensity determined by flow cytometry after treatment with the relevant antibodies and specific labelled anti-immunoglobulins. The histograms shown in Figures 5A and B indicate that membrane expression of both antigens was increased significantly in treated cells compared with controls, and that viral infection induced a quantitatively similar expression of both VCAM-1 and MHC class I to that resulting from optimal (500 U ml–1) rTNF-α stimulation. The fluorescence profiles show that an increased expression of MHC class I antigens was induced on virtually all

cells in cultures infected at 0.5 or 1 TCID50 per cell (Fig 6A), whereas never more than about 50 per cent of the cells expressed VCAM-1 at any time (Fig 6B). MHC class II antigens only appeared late after infection, and were expressed by approximately 20 per cent of cells (Fig 6C). Alkaline phosphatase/anti-alkaline phosphatase staining indicated that MHC class I antigens appeared at day 1 post-infection and were maintained until days 9 to 10, when they began to decline. VCAM-1 persisted virtually unchanged from day 1 to day 10, and MHC class II antigens only appeared around days 8 to 9. The adhesion of freshly-prepared peripheral blood monocytes from healthy sheep to the cultured endothelial cells was significantly enhanced by either treatment with rTNF-α or infection with CAEV. The number of leucocytes eluted by EDTA from the washed cultures rose by a factor of 2.3 (n = 10; P < 0.001) after treatment with rTNF-α, and by a factor of 1.9 (n = 10; P < 0.001) after infection by 1 TCID50 of CAEV-cork (Fig 7).

DISCUSSION Naturally-infected French sheep are more often found to harbour lentiviruses molecularly related to CAEV than to maedi–visna virus (Leroux et al 1995a). The cork strain of CAEV, which can infect sheep experimentally, was chosen to examine potential interactions with ovine endothelial cells. Unfortunately no other well-characterised CAEV-related strains were available for more extensive comparisons. Aortic endothelial cells were chosen because they are in plentiful supply and have been shown to be susceptible to small ruminant lentivirus infection, although microvascular endothelial cells may show a greater, but often more variable, degree of susceptibility (Craig et al 1997). The technique of gentle scraping of the inner surface of ovine aortas was found to be more satisfactory for the recovery of endothelial cells for culture than the alternative method of collagenase digestion, because of the presence of numerous collateral vessels. At least 90 per cent of the cultured cells were endothelial, as judged by the presence of intracytoplasmic vWF, which can also be demonstrated by immunohistochemistry on macrovascular and microvascular endothelium in tissue samples. After infection by CAEV,

C. Le Jan, T. Greenland, F. Gounel, S. Balleydier, J.F. Mornex

228

A

B

C

D

FIG 2: APAAP labelling of aortic endothelial cell cultures infected in vitro by CAEV at 1 TCID50 per cell (× 170). (A) Labelling by anti-VCAM-1 mAb, at day 0; (B) labelling by anti-viral p30 at day 3; (C) labelling by anti-VCAM-1 mAb, at day 3; (D) labelling by anti-viral p30 at day 5.

A

B

C

D

FIG 3: Immunofluorescence double labelling of aortic endothelial cells 5 days after in vitro infection by CAEV (× 100). Rabbit anbtibodies to vWF were detected with rhodamine-conjugated goat anti rabbit immunoglobulins and mouse antibodies to either viral p 30 or VCAM-1 with fluorescein-conjugated goat anti-mouse immunoglobulins (see Materials and Methods). (A, B) Expression of von Willebrand factor (red) and VCAM-1 (green); (C, D) expression of von Willebrand factor (red) and CAEV p30 antigen (green).

In vitro endothelial cell activation by CAEV

Virus titre in supernatant

7 6 5 4 3 2 1 0 0

1

2

3 4 6 Days post-infection

8

10

FIG 4: Time course of viral titres in supernatants from aortic endothelial cell ◆ ); 1 TCID50 per cell (▲ ▲ )]. cultures infected with CAEV [0.5 TCID50 per cell (◆

some 90 per cent of these cells rapidly produced virus as indicated by their production of the p30 antigen, which is a late component of the infectious cycle and constitutes a major element of the structural capsid. Viral titres were detectable in supernatants only 1 day post-infection. Although residual virus from the inoculum might account for some of this activity, the authors believe active virus production does start early because cell supernatants at time 0, immediately after washing, were negative, CAEV is unstable in culture at 37°C and would be rapidly degraded, and faint immunofluorescent staining for p30 was already visible in cells 24 hours post-infection (not shown). The actin-positive cell population also stained positive for CAEV antigen by immunofluorescence. Infection did not cause endothelial cell lysis, as is the case in fibroblasts or sheep smooth muscle cells (Leroux et al 1995b), but was productive, with titres rising to more than 106 at 8–10 days post-infection. Endothelial cells have been shown to support maedi–visna virus infection. Craig et al (1997) detected reverse transcriptase activity in supernatants of in vitro-infected sheep endothelial cultures, and the level of replication depended on both the strain of virus and the tissue origin of the vasculature used to establish the cultures. HIV-1 can sometimes infect vascular endothelial cells, depending critically on the viral strain, the tissular origin of the cells and their status as quiescent or actively dividing cultures (Poland et al 1995). It was concluded that, although vascular endothelium can

support HIV replication, it does not constitute an important viral reservoir. HIV was shown to infect human endothelial cells in vivo (Wiley et al 1986, Bagasra et al 1996), but the extent of the presence of virus in endothelial cells from lentivirus-infected small ruminants is at present unknown. Ovine endothelial cells expressed markers of activation following infection with CAEV to a similar extent as after treatment with rTNF-α. VCAM-1 appeared by 24 hours postinfection and expression was maintained throughout the 10day culture period, concomitantly with the production of infectious virus. MHC class I and II antigens were also induced. These markers might be expected to be accompanied by production of chemotactic factors and cytokines by the activated cell (Chen and Manning 1996). It appears that all markers are not uniformly expressed by every infected cell because, although some 90 per cent of cells in the cultures expressed the viral p30 antigen, no more than 50 per cent positivity for VCAM-1 was observed. Mock infection using uninfected culture supernatant or heat-inactivated CAEV virus did not activate endothelial cells. Human endothelial cells are activated in vitro after infection by flaviviruses (Shen et al 1997) and serum levels of VCAM-1, ICAM-1 and E-selectin are raised in HIV-infected individuals (Seigneur et al 1997). Therefore, activation may be a general consequence of viral infection of endothelia. Another consequence of CAEV infection of ovine aortic endothelial cell cultures was an increase in the adhesion of peripheral blood leucocytes from naive sheep. Human endothelial cells infected with HIV-1 show increased adhesion of lymphocytes (Saukkonen et al 1997) with enrichment of the CD8+ population. CAEV does not target lymphocytes, and the phenotypic characteristics of the adhering cells is not known. Further studies will establish this, and the activation status of the adhering leucocytes. It will also be necessary to establish whether microvascular endothelial cells behave in the same manner as the macrovascular cells studied here to establish the role played by endothelial cells in leucocyte homing and lesion formation in small ruminant lentiviral disease. ACKNOWLEDGEMENTS The authors thank Dr T. F. Tedder for his gift of the monoclonal antibody HAE2-1. This work was supported by a grant of the A.I.P. INRA ‘mammary gland biology’.

A

B 1000 800 600 400 200 0 Control

TNF

-α

Mean fluorescence (arbitrary units)

Mean fluorescence (arbitrary units)

229

Infected

50 40 30 20 10 0 Control

TNF

-α

Infected

FIG 5: Expression of MHC class 1 (A) antigens and VCAM-1 (B) by third passage endothelial cell cultures infected with CAEV or stimulated with TNF-α compared to control cultures. Mean fluorescence intensity (in arbitrary units) determined by flow cytometry on trypsinised cultures of cells labelled by treatment with relevant antibody then fluorescent second antibody as described in materials and methods (n = 3 for each group; differences between controls and CAEV-infected or TNF-α induced cells: P < 0.05 in each case).

C. Le Jan, T. Greenland, F. Gounel, S. Balleydier, J.F. Mornex

230

A

100

a b

0 0

60

102

10

103

104

B

a

b

0 0

10

102

103

104

C 90

a

b

0 0

10

102

103

104

FIG 6: Expression of MHC class 1 antigen, VCAM-1 and MHC class 2 antigen by CAEV-infected, third passage aortic endothelial cell cultures: flow cytometric fluorescence histograms. (A) MHC class 1 expression at day 5 post-infection; (B) VCAM-1 expression at day 5 post-infection; (C) MHC class 2 expression at day 10 postinfection; (a) uninfected cultures (shaded); (b) cultures infected with CAEV at 1 TCID50 cell–1.

Percentage of adhering leucocytes

In vitro endothelial cell activation by CAEV

10 8 6 4 2 0

Control

TNF-α

Infected

FIG 7: In vitro adhesion of peripheral blood leucocytes to untreated aortic endothelial cell cultures compared to those stimulated with TNF-α or infected with CAEV. Results are expressed as the mean percentage of adhering leucocytes (n = 10; control versus TNF-α: P < 0.001; control versus CAEV infection: P < 0.001).

REFERENCES BAGASRA, O., LAVI, E., BOBROSKI, L., KHALILI, K., PESTANER, J.P., TAWADROS, R. & POMERANTZ, R.J. (1996) Cellular reservoirs of HIV-1 in the central nervous system of infected individuals: identification by the combination of in situ polymerase chain reaction and immunohistochemistry. AIDS 6, 573–585 CABIRAC, G.F., MURRAY, R.S., MCLAUGHLIN, L.B., SKOLNICK, D.M., HOGUE, B., DOROVINI-ZIS, K. & DIDIER, P.J. (1995) In vitro interaction of coronaviruses with primate and human brain microvascular endothelial cells. Advances in Experimental Medicine and Biology 380, 79–88 CHEN, C.C. & MANNING, A.M. (1996) TGF-beta, IL-10 and IL4 differentially modulate the cytokine-induced expression of IL-6 and IL-8 in human endothelial cells. Cytokine 8, 58–65 CORDELL, J.L., FALINI, B., ERBER, W.N., GHOSH, A.K., ABDULAZIZ, Z., MACDONALD, S., PULFORD, K.A., STEIN, H. & MASON, D.Y. (1984) Immunoenzymatic labeling of monoclonal antibodies using immune complexes of alkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAP complexes). Journal of Histochemistry and Cytochemistry 32, 219–229 CRAIG, L.E., NEALEN, M.L., STRANDBERG, J.D. & ZINK, M.C. (1997) Differential replication of ovine lentivirus in endothelial cells cultured from different tissues. Virology 238, 316–326 JONES, K., RIVERA, C., SGADARI, C., FRANKLIN, J., MAX, E.E., BHATIA, K. & TOSATO, G. (1995). Infection of human endothelial cells with Epstein–Barr virus. Journal of Experimental Medicine 182, 1213–1221

231

LAFRENIE, R.M., WAHL, L.M., EPSTEIN, J.S., HEWLETT, I.K., YAMADA, K.M. & DHAWAN, S. (1996). HIV-1-Tat modulates the function of monocytes and alters their interactions with microvessel endothelial cells. A mechanism of HIV pathogenesis. Journal of Immunology 156, 1638–1645 LEROUX C., VUILLERMOZ, S., MORNEX, J.F. & GREENLAND, T. (1995a) Genomic heterogeneity in the pol region of ovine lentiviruses obtained from bronchoalveolar cells of infected sheep in France. Journal of General Virology 76, 1533–1537 LEROUX C., CORDIER G., MERCIER I., CHASTANG J., LYON M., QUERAT G., GREENLAND T., VIGNE R. & MORNEX J.F. (1995b) Ovine aortic smooth muscle cells allow the replication of visna-maedi virus in vitro. Archives of Virology 140, 1–11 MANKOWSKI, J.L., SPELMAN, J.P., RESSETAR, H.G., STRANDBERG, J.D., LATERRA, J., CARTER, D.L., CLEMENTS, J.E. & ZINK M.C. (1994) Neurovirulent simian immunodeficiency virus replicates productively in endothelial cells of the central nervous system in vivo and in vitro. Journal of Virology 68, 8202–8208 PEPIN, M., VITU, C., RUSSO, P., MORNEX, J.F. & PETERHANS, E. (1998) Maedi–Visna virus infection in sheep: a review. Veterinary Research 29, 341–346 POLAND, S.D., RICE, G.P. & DEKABAN, G.A. (1995) HIV-1 infection of human brain-derived microvascular endothelial cells in vitro. Journal of Acquired Immunodeficiency Syndrome and Human Retrovirology 8, 437–445 REED, L. & MUENCH, H. (1938) A simple method for estimating fifty per cent points. American Journal of Hygiene 27, 493–497 RUSSELL, H., O’TOOLE, D.T., BARDSLEY, K., DAVIS, W.C. & ELLIS, J.A. (1996) Comparative effects of bluetongue virus infection of ovine and bovine endothelial cells. Veterinary Pathology 33, 319–331 SAUKKONEN, J.J., FURFARO, S., MAHONEY, K.M., STRIETER, R.M., BURDICK, M., WRIGHT, E.A., KORNFELD, H. & BERMAN, J.S. (1997) In vitro transendothelial migration of blood T lymphocytes from HIV-infected individuals. AIDS 11, 1595–1601 SEIGNEUR, M., CONSTANS, J., BLANN, A., RENARD, M., PELLEGRIN, J., AMIRAL, J., BOISSEAU, M. & CONRI, C. (1997) Soluble adhesion molecules and endothelial cell damage in HIV infected patients. Thrombosis and Haemostasis 77, 646–649 SHEN, J., T-TO., S.S., SCHRIEBER, L. & KING N.J. (1997) Early E-selectin, VCAM-1, ICAM-1, and late major histocompatibility complex antigen induction on human endothelial cells by flavivirus and comodulation of adhesion molecule expression by immune cytokines. Journal of Virology 71, 9323–9332 SINZGER, C., GREFTE, A., PLACHTER, B., GOUW, A.S., THE, T.H. & JAHN, G. (1995) Fibroblasts, epithelial cells, endothelial cells and smooth muscle cells are major targets of human cytomegalovirus infection in lung and gastrointestinal tissues. Journal of General Virology 76, 741–750 WILEY, A.C., SCHRIER, R.D., NELSON, J.A., LAMPERT, P.W., OLDSTONE, M.B. (1986) Cellular localization of human immunodeficiency virus infection within the brain of acquired immune deficiency syndrome patients. Proceedings of the National Academy of Sciences of the USA 83, 7089–7093 WOODROFFE, S.B., GARNETT, H.M. & DANIS, V.A. (1993) Interleukin-1 production and cell-activation response to cytomegalovirus infection of vascular endothelial cells. Archives of Virology 133, 295–308 ZINK, M.C. & JOHNSON, L.K. (1994) Pathobiology of lentivirus infection of sheep and goats. Virus Research 32, 1439–1454 Accepted July 12, 2000