Equine Endothelial Cells Support Productive Infection of Equine Infectious Anemia Virus

Abstract

Previous cell infectivity studies have demonstrated that the lentivirus equine infectious anemia virus (EIAV) infects tissue macrophages in vivo and in vitro. In addition, some strains of EIAV replicate to high titer in vitro in equine fibroblasts and fibroblast cell lines. Here we report a new cell type, macrovascular endothelial cells, that is infectible with EIAV. We tested the ability of EIAV to infect purified endothelial cells isolated from equine umbilical cords and renal arteries. Infectivity was detected by cell supernatant reverse transcriptase positivity, EIAV antigen positivity within individual cells, and the detection of viral RNA by in situ hybridization. Virus could rapidly spread through the endothelial cultures, and the supernatants of infected cultures contained high titers of infectious virus. There was no demonstrable cell killing in infected cultures. Three of four strains of EIAV that were tested replicated in these cultures, including MA-1, a fibroblast-tropic strain, Th.1, a macrophage-tropic strain, and WSU5, a strain that is fibroblast tropic and can cause disease. Finally, upon necropsy of a WSU5-infected horse 4 years postinfection, EIAV-positive endothelial cells were detected in outgrowths of renal artery cultures. These findings identify a new cell type that is infectible with EIAV. The role of endothelial cell infection in the course of equine infectious anemia is currently unknown, but endothelial cell infection may be involved in the edema that can be associated with infection. Furthermore, the ability of EIAV to persistently infect endothelial cultures and the presence of virus in endothelial cells from a long-term carrier suggest that this cell type can serve as a reservoir for the virus during subclinical phases of infection.

Equine infectious anemia virus (EIAV) is a lentivirus that causes an acute disease during periods of frank viremia that is distinguished by fever and thrombocytopenia (6, 7, 12). Episodes of viremia can occur during subsequent months and may cause chronic anemia, ventral edema, and general wasting. In most horses, the viremia and accompanying clinical disease are eventually controlled, resulting in a clinically quiescent, seropositive carrier status.

Several studies have investigated the sites of viral replication during acute, clinically severe episodes of EIA. Early immunofluorescence studies by McGuire et al. identified viral antigen-positive cells in a number of macrophage-rich tissues such as spleen, lung, liver, and bone marrow (19). Rice et al. confirmed and extended these studies by Southern blotting analysis of proviral sequences (27). Again, proviral sequences were found in tissues rich in tissue macrophages, such as liver, lymph node, spleen, and bone marrow. The first direct demonstration that EIAV replication occurred in macrophages in vivo was performed by Sellon et al. through colocalization of active viral replication with cells that immunostained for macrophage-specific markers (31). Thus, during the acute phase of the infection, tissue macrophages in a number of different organs are productively infected with EIAV; however, other infected cell types during the acute or carrier phase of the infection have not been identified.

In tissue culture, EIAV has been found to replicate in equine monocyte-derived macrophages (14), equine fibroblasts (13, 16), equine fetal kidney cells (4), and canine and feline fibroblast cell lines (4, 13). Strains of virus previously grown in fibroblasts usually have a preference for replication in fibroblast and fibroblast cell lines, whereas strains obtained in vivo during a viremic episode or strains passaged in macrophages replicate to higher titers in primary equine macrophage cultures (5, 24). Here we demonstrate for the first time that equine macrovascular endothelial cells are productively infected with EIAV.

Endothelial cell cultures.

Equine endothelial cells were isolated from umbilical cords from term pregnancies. Morphologically, isolated umbilical vein endothelial cells (UVEC) were relatively uniform and, upon confluence, had the characteristic cobblestone appearance of macrovascular endothelial cells (Fig. 1a). First- or second-passage cultures were found to be greater than 99% vimentin positive (data not shown), cytokeratin negative (data not shown), and von Willebrand’s factor positive (Fig. 1b), indicating that the cultures were endothelial cells. As previously reported for horse endothelial cells, these cells were negative for staining with the lectin Ulex europaeus (28).

FIG. 1.

Immunostaining characterization of an equine UVEC culture. (a) The UVEC culture as observed with transmitted light (magnification, ×100) showing the typical cobblestone appearance of macrovascular endothelial cells. (b) The same field of cells observed under UV light (magnification, ×100). All cells in this field were von Willebrand’s factor positive, indicating that they were from the endothelium. Acetone-water (3:1)-fixed UVEC were immunostained for von Willebrand’s factor with a 1:800 dilution of rabbit anti-human von Willebrand’s factor antisera (Dako) followed by fluorescein isothiocyanate-conjugated goat anti-rabbit (1:500). Negative controls included the use of normal rabbit serum as the primary antiserum. To obtain UVEC, umbilical cords were extensively washed in Dulbecco modified Eagle medium containing antibiotics and fungizone prior to cell isolation. A trypsin-versene solution (0.05% trypsin, 0.02% EDTA) was injected into vessels, which were tied off and incubated at 37°C for 1 to 2 h. Trypsinized cells were collected from the vessels, and trypsin incubation of the vessels was repeated. Cells were pooled, washed, and plated at a density of approximately 2 × 10⁵/T25 flask in Dulbecco modified Eagle medium with 40% fetal calf serum and antibiotics and fungizone. Cells were trypsinized and subdivided 1:4 every 4 to 5 days.

In vitro EIAV infections.

UVEC cultures were infected with either 225 or 2,250 tissue culture infectious units (TCIU) of the MA-1 strain of EIAV as titered in the equine dermal fibroblast cell line, ED (5). Cultures were maintained for 2 weeks, and supernatants were monitored during the infection for reverse transcriptase (RT) activity. At the higher infectious titer, RT activity was detectable by day 6 postinfection (Fig. 2a). Over the 2-week infection period, RT activities continued to increase, suggesting that active virus replication was occurring in the cultures.

FIG. 2.

Characterization of EIAV infection of UVEC cultures. (a) UVEC culture supernatant RT activity. Actively dividing UVEC cultures were infected with 2,250 or 225 TCIU of MA-1 stock obtained from infected ED cell cultures. Medium was changed the following day, and cultures were subdivided every 3 to 6 days. UVEC culture supernatants were taken daily during the course of the experiment. Each day, half of the supernatant was removed from the culture for sampling and replaced with an equivalent volume of fresh media. Supernatants were assayed at the termination of the experiment for RT activity by the dot blot RT assay as previously described (35). Following autoradiography of the RT assay, ³²P counts per minute of each RT sample was counted in a scintillation counter. Counts per minute of an uninfected UVEC culture supernatant were subtracted from the RT data. (b) EIAV antigen immunostaining of an MA-1-infected UVEC culture 6 days postinfection by phase-contrast microscopy. EIAV antigen immunostaining was performed with horse anti-EIAV antiserum (1:800) as previously described (magnification, ×100) (18). (c) Productive infection of UVEC cultures by the EIAV strain MA-1. RT-positive supernatants from an MA-1-infected UVEC culture were titered onto ED cells. TCIU (2 × 10²) were added to a fresh UVEC culture, and the cells were monitored for RT activity over a 2-week period as described for panel a.

Viral infection of the UVEC cultures was also demonstrated by immunostaining of EIAV antigens in cells from infected cultures. Numerous foci of viral antigen-positive cells could be observed by day 6 postinfection, as detected by horse anti-EIAV antisera (Fig. 2b) or by anti-EIAV Gag p26 monoclonal antibody (data not shown). The antigen-positive cells had the characteristic cobblestone appearance of macrovascular endothelial cells. Infected endothelial cell cultures became progressively more positive for viral antigen staining over the course of an infection until greater than 95% of the culture was EIAV antigen positive (data not shown).

To determine if infectious virions were present in supernatants of MA-1-infected UVEC cultures, RT-positive supernatants obtained from MA-1-infected UVEC cultures were passaged onto fresh UVEC cultures and supernatants were monitored over 2 weeks for RT activity (Fig. 2c). Supernatants became RT positive over the course of infection. These supernatants contained viral titers equivalent to that found in infected ED cell supernatants (data not shown). Subsequent EIAV antigen staining of the UVEC infected for 2 weeks also demonstrated viral antigen positivity of greater than 95% of the cells (data not shown). In total, the three experiments shown in Fig. 2 demonstrated that UVEC cultures could be productively infected with EIAV.

EIAV infection of renal artery endothelial cells.

Phenotypic and infectivity studies similar to those described above for UVEC cultures were also performed with equine renal endothelial cells (EREC), and these macrovascular endothelial cells were also found to be von Willebrand’s factor positive and readily infectable with the MA-1 strain of EIAV (data not shown). This result indicated that EIAV infection is not unique for UVEC cultures and suggested that all equine macrovascular endothelial cells may be infectable with EIAV.

EIAV infection of UVEC cultures results in a chronic infection.

By light microscopy, no obvious differences in cell numbers in the uninfected and infected cultures were apparent over the course of an infection, suggesting that EIAV infection of macrovascular endothelial cells was a chronic infection rather than a cytolytic one. MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) assays were performed to detect if more subtle changes in cell numbers were occurring during infection. By 2 weeks postinfection, when >90% of the cells were antigen positive in the infected cell cultures, the MTT assay values for the infected cultures were about 70% of those of the uninfected cultures (data not shown). These infected UVEC cultures continued to replicate over a period of several months. Thus, viral infection did not result in massive cell death of the UVEC cultures; however, infection may have slowed cell growth, since a modest decrease in cell numbers in infected cultures compared to uninfected cultures was observed over time.

Cell tropism studies.

To characterize the replication competence of EIAV strains having differing cell tropisms, four different strains of virus were titered for infectious virus in UVEC, EREC, equine dermal fibroblasts (ED), and equine macrophage cultures (Table 1). MA-1 is a fibroblast-tropic strain of EIAV that is avirulent in vivo (5). Th.1 is a macrophage-tropic strain of unknown virulence (5). Wyoming is a highly virulent strain that has been maintained by serial passage through horses, with stocks taken during the initial disease episode. WSU5 is a mildly virulent strain that is a Wyoming derivative and has been passaged through fetal kidney cells and back passaged through a series of three horses to reestablish virulence (23). WSU5 and Th.1 were found to replicate to high titers in all cells tested, whereas MA-1 replicated in the ED, UVEC and EREC cultures and the Wyoming strain of virus replicated only in macrophages. This limited cell tropism of the MA-1 and Wyoming strains had previously been reported (5, 24). Thus, the macrovascular endothelial cell cultures support replication of the same viral strains as ED cells and produced similar levels of infectious virus as detected by TCIU assays. A possible explanation for the absence of Wyoming infectivity in endothelial cultures is that virions from serum may be associated with inhibitors that decrease the infectivity of the virus. Alternatively, this strain of virus that is highly virulent in horses and moderately infectious in equine macrophages may not infect in endothelial cells. To determine if serum products were responsible for the absence of infection of endothelial cells, we infected UVEC cultures with a stock of virulent Wyoming virus that had been passaged once in vitro through primary equine macrophages (these stocks were a kind gift of Fred Fuller). No RT or viral antigen staining was detected in these UVEC cells, indicating that the UVEC cultures were not infectible with Wyoming (data not shown). It is possible that the Wyoming strain of EIAV may infect only macrophages; however, recent in situ hybridization studies of tissues from acutely ill Wyoming-infected horses revealed von Willebrand’s factor-positive cells that were strongly positive for EIAV RNA, suggesting that in vivo endothelial cells are infectible with the Wyoming strain of EIAV (22a). It is possible that other types of endothelial cells, such as microvascular and high venule, that were not studied here, may replicate the Wyoming strain in vitro.

TABLE 1.

Cell tropism of different strains of EIAV

EIAV strain	Source of stock	Infectious titer/ml inb,c:
		ED	UVEC	EREC	Macrophages
MA-1	ED cells	3.0 × 104	2.4 × 104	2.5 × 104	0
WSU5	Fetal equine kidney cells	5.8 × 104	4.2 × 104	1.9 × 104	1.4 × 104
Th.1	Equine macrophages	9.7 × 102	6.8 × 102	6.4 × 103	2.5 × 104
Wyoming	Horse seruma	0	0	0	8.1 × 103

^aThe infectious titer of this Wyoming stock in horses is ≥10⁸ infectious Units/ml. 

^bTCIUs were determined by the immunostaining assay as previously described (18). ED cultures were fixed for immunostaining at 44 h postinfection. Endothelial18) cultures were fixed at 50 h postinfection. Viral titrations in macrophages were performed at day 6 postinfection with end point dilutions. 

^cTiter values as determined from the means of two independent experiments. Serial dilutions performed in duplicate were analyzed in each experiment. 

Virus positivity in endothelial cells obtained from an in vivo infection.

Renal artery endothelial cells were isolated at necropsy from a horse infected 4 years earlier with the WSU5 strain of EIAV. This horse had been seropositive for EIAV long term and was clinically healthy prior to necropsy. Cultures were maintained for six passages prior to staining to generate enough cells for analysis of EIAV RNA by in situ hybridization and immunopositivity for von Willebrand’s factor. The presence of endothelial cells in the culture was demonstrated by positivity upon immunostaining for von Willebrand’s factor (Fig. 3C). In situ hybridization of the culture revealed cells that were actively synthesizing EIAV-specific RNA (Fig. 3E). A combination of immunostaining for von Willebrand’s factor and in situ hybridization for EIAV-specific RNA revealed dually positive cells in the culture (Fig. 3A, arrowhead). The culture also contained infected cells which were not expressing von Willebrand’s factor (Fig. 3A, open arrow); however, maintenance of endothelial cells in culture can result in loss of von Willebrand’s factor expression (11, 25). In addition, von Willebrand’s factor-positive cells that were negative for viral RNA were observed (Fig. 3A, solid arrow). Infectious virus was detected in supernatants of these cultures, indicating that a productive infection was ongoing in the culture (data not shown). The dual positivity of some cells for von Willebrand’s factor and EIAV RNA confirms the identity of infected cells as endothelium. In addition, these findings strongly suggest that endothelial cells were infectible not only in vitro but also in vivo. However, these studies cannot exclude the possibility that another in vivo-infected cell type such as a fibroblast or a macrophage could be a low-level contaminant in our endothelial cultures and have passed the infection to endothelial cells in vitro.

FIG. 3.

Photomicrographs of cytospin preparations of cultured endothelial cells demonstrating colocalization of von Willebrand’s factor and EIAV gag RNA. This EREC culture was obtained from a horse experimentally infected with 10⁶ TCIU of WSU5 as titered on fetal equine kidney cells. Following initial disease, this horse was seropositive and subclinical for EIA for 4 years and at the time of necropsy. (A) An EIAV-infected cell double-labeled by immunohistochemistry for von Willebrand’s factor and in situ hybridization for EIAV gag RNA is shown (arrowhead). Other cells positive for only von Willebrand’s factor (solid arrow) or EIAV gag RNA (open arrow) are also present in the micrograph. Bar = 10 μm. (B) Uninfected cells double labeled as in panel A. Lack of purple staining demonstrates the specificity of in situ hybridization for EIAV RNA in cells positive for von Willebrand’s factor. Bar = 10 μm. (C) EIAV-infected cells labeled by immunohistochemistry for von Willebrand’s factor alone. Bar = 10 μm. (D) EIAV-infected cells labeled by immunohistochemistry with normal rabbit immunoglobulin G, indicating specificity of the immunohistochemical assay. Bar = 10 μm. (E) EIAV-infected cells labeled by in situ hybridization for EIAV gag RNA alone. Bar = 10 μm. (F) Uninfected cells labeled by in situ hybridization of EIAV gag RNA, indicating specificity of the in situ hybridization assay. Bar = 20 μm. Cytospin slides of cultured endothelial cells were fixed in 4% paraformaldehyde. Immunohistochemical detection of von Willebrand’s factor was first performed by antigen retrieval with 0.1% pronase, blocking of nonspecific protein binding with 5% normal goat serum, and rabbit anti-von Willebrand’s factor (Dako A082) (1:1,000) as the primary antibody. Normal rabbit immunoglobulin G fraction (Dako X903) (1:1,000) served as a negative control. Bound primary antibody was detected with biotinylated goat anti-rabbit immunoglobulin, an avidin-biotin-peroxidase complex (Vectastain Elite Rabbit kit; Vector Labs), and 3,3-diaminobenzidine. Solutions were prepared in diethylpyrocarbonate-treated water, and heparin (4,000 units/ml) was added to all antisera to prevent RNase degradation of probe in the subsequent in situ hybridization. In situ hybridization was then performed. Cells were further permeabilized with 0.2 N HCl and 5 μg of proteinase K per ml. Nonspecific nucleic acid binding was blocked by treatment for 1 h at 42°C with 50% formamide containing 310 μg of sheared herring sperm DNA per ml, 310 μg of polyadenosine per ml, 31 mM EDTA, 25 mM HEPES, 1 M NaCl, 0.25% sodium dodecyl sulfate, 125 mM dithiothreitol, and 1× Denhardt’s solution. Viral RNA was detected by hybridization for 16 h at 42°C with a 450-nucleotide, dioxygenin-labeled, cRNA probe antisense in polarity to EIAV gag RNA. Unbound probe was digested with 50 μg of RNase A per ml, and the slides were washed in decreasing concentrations of saline sodium citrate. Bound probe was detected with antidioxygenin Fab fragments conjugated with alkaline phosphatase and 5-bromo-4-chloro-3-indolylphosphate nitroblue tetrazolium chromogen (Boehringer Mannheim). Specificity was demonstrated by in situ hybridization performed with uninfected cells with and without prior immunohistochemistry for von Willebrand’s factor.

Detection of EIAV in EREC isolated from a horse infected long term with WSU5 suggests that endothelial cells may harbor the virus for long periods of time in vivo. Not only did the outgrowth cultures contain EIAV-positive cells but also infectious virus could be detected in the supernatants of these cultures. Whether an active infection in vivo was occurring in this long-term carrier horse or whether the infection was reactivated by in vitro conditions is not known. This finding, combined with our tissue culture studies demonstrating that endothelial cells are persistently infected with EIAV, supports the idea that endothelial cells may be persistently infected with EIAV in infected animals. Thus, unlike tissue macrophages, which are presumptively killed by EIAV, endothelial cells may act as a reservoir for the virus. Future in situ studies will be needed to determine if endothelial cells actively produce virus during long-term persistent infection.

EIAV infection of endothelial cells can possibly explain a number of pathological consequences of EIAV infection. For instance, vessel leakage resulting from endothelial cell infection may be responsible for the ventral edema and hemorrhage frequently associated with EIAV infection. In addition, at least with some strains of virus, production of cell-free virions from infected endothelial cells may, along with macrophage infection, be responsible for the high-titered viremia associated with acute infection. Finally, endothelial cells may serve as a conduit for brain infection. Brains of EIAV-infected horses are known to contain EIAV antigen-positive cells (19), although neurologic disease is not generally associated with the infection. Potentially, microvascular endothelial cells composing the blood brain barrier may be responsible for EIAV transmission to the brain as has been proposed for primate lentiviruses (20, 26).

The ability of endothelial cells to support EIAV replication in vitro is an important finding for a number of reasons. First, the observation that both fibroblast-tropic (MA-1) and macrophage-tropic (Th.1) strains replicated in endothelial cells suggests that UVEC cultures could potentially serve as an index cell for titration of these strains of virus. Until now, such a cell has not been available, since macrophage-tropic strains of virus do not always replicate well in fibroblasts and, conversely, fibroblast-tropic strains do not replicate well in macrophages (5). Furthermore, unlike equine macrophages, these primary cultures are relatively easy to isolate and maintain and can be passaged numerous times.

The ability of lentiviruses to infect endothelial cells in vivo and in vitro varies. Feline immunodeficiency virus, ovine lentiviruses, and the primate lentiviruses, human immunodeficiency virus (HIV) and simian immunodeficiency virus (SIV), have been found to infect microvascular endothelial cells from some tissues. Steffan et al. showed that feline immunodeficiency virus infected cultured microvascular endothelial cells derived from both brain (32) and liver (33). Craig et al. have demonstrated that North American strains of ovine lentivirus have a decided preference for microvascular endothelial cells, whereas Icelandic strains have a preference for macrovascular endothelial cells (9). Both HIV and SIV have been reported to infect cultured microvascular endothelial cells from the brain (17, 20, 26), and HIV has been reported to infect endothelial cells derived from bone marrow (21). Consistent with the in vitro observation that microvascular cells are infectible with primate lentiviruses are in situ data with HIV (3) and SIV (17) demonstrating infection of brain microvascular cells in vivo. Other groups have not been able to demonstrate endothelial infection in HIV-positive individuals (10, 34) or in purified brain microvascular endothelial cells (2, 22). Macrovascular endothelial cells such as umbilical cord endothelial cells do not appear to be productively infectible with the primate lentiviruses (1). While it is possible that human UVEC cultures can support HIV entry resulting in abortive infection (8) which can be rescued by cocultivation with peripheral blood mononuclear cells (30), other groups have not been able to find even transient replication of HIV in human UVEC cultures (1, 15). Studies with bovine leukemia virus (29) and bovine immunodeficiency virus (5a) have demonstrated that bovine peripheral blood high-venule endothelial cells support replication of these retroviruses.

The interesting possibility that different types of endothelial cells may play unique roles in EIAV infection of horses has yet to be explored. Further in vitro and in vivo studies will be required to clearly define the role of endothelial cells in EIAV infection.