CELLULAR MECHANISMS OF FELINE IMMUNODEFICIENCY VIRUS (FIV)-INDUCED NEUROPATHOGENESIS

Elisabeth Zenger, Evelyn Tiffany-Castiglioni, and Ellen W. Collisson

Department of Medicine, UCSF AIDS Program, San Francisco General Hospital, University of California, San Francisco, California (Zenger) and the Departments of Veterinary Anatomy and Public Health (Tiffany-Castiglioni) and Pathobiology (Collisson), College of Veterinary Medicine, Texas A&M University, College Station, Texas.

Received 10/7/97 Accepted 10/23/97

4. A ROLE FOR ASTROGLIA?

The cellular basis for the neurobehavioral deficits associated with HIV infection have not been identified. Because of the paradox between the relatively small number of productively infected glia and the magnitude of clinical dysfunction (23, 24), indirect mechanisms of neurological impairment are suspected. Astroglia, which are interposed between neuronal cells and brain capillaries, with astroglial foot plates ensheathing these capillaries (25), are the most numerous cells in the brain and are essential for maintenance of normal neuronal function. Astrocytes play immunologic roles by secreting cytokines and serving as facultative antigen presenting cells and maintain the extracellular fluid homeostasis by regulating levels of certain electrolytes and neurotransmitters. Because of their unique relationship with cerebral blood vessels and neurons, and importance in the maintenance of the CNS environment, astrocytes are likely to play a very strategic role in the progression of neurological disease in lentiviral-infected patients. Although in vivo productive infection of astroglia seems to be rare, extensive persistent restricted and latent HIV infection occurs in astroglia and can be reactivated under certain conditions including cytokine activation (26, 27, 28). Elucidation of functional impairment of cells essential to neuronal function, such as astroglia, may provide insight into how low numbers of productively infected cells can elicit progressive and devastating neurological impairment.

4.1. Lymphocyte facilitated infection of astroglia by neurovirulent FIV

Recent studies have demonstrated that persistent astroglial infection by HIV in vivo may be extensive (28, 29), supporting early reports of the presence of viral particles in astrocytes (21, 30, 31). Because astroglial foot-processes surround the majority of the vascular surfaces in the mammalian brain (25) and widespread infection of astroglia may contribute to the neurologic deficit (28), the hypothesis that astroglia are the initial sites of lentiviral replication and source of infectious viral particles for other brain cells is compelling. Primary feline astroglia and the feline G355-5 astroglial cell line are readily infected with cell-free stocks of FIV-Pet but not with cell-free stocks of FIV-MD. Whether feline T-cells infected with FIV-MD might transmit the virus more effectively than exposure to cell-free virus was investigated (32). Astroglial infection is very efficient by this method, with virtually 100% of the astroglia demonstrating positive staining for FIV by immunofluorescence assay (IFA) after coculture. The importance of cell-cell transmission of HIV is well established (33, 34, 35, 36, 37). Recently, a situation similar to that we described was reported with lymphocytotropic but not monocytotropic strains of HIV (34). In this section, we describe our recent findings concerning a mechanism of FIV infection of astroglia that requires lymphocyte-facilitation.

To facilitate infection of feline astroglia with FIV-MD, FETJ cells, a feline IL-2 independent T cell line were first infected and then added to established astroglial cell cultures. FETJ cells infected with FIV-MD rapidly and specifically adhered to the astroglial cells (figure 1). Evaluation of reverse transcriptase and p26 in supernatants at intervals ranging from 20 minutes to 24 hours immediately following coculture with astroglial cells revealed no increase in FIV protein release from FIV-MD-infected FETJ cells suggesting direct viral transfer or a release of virions into the confined space between contacting cells. Further, FIV-MD infection of astroglia was non-productive, a situation that may be analogous to that observed for nonproductive infection of glial cells by HIV in vitro (26, 27, 28, 38) and in vivo (26, 29). Maryland strain FIV-infected FETJ cells in contact with astroglial cells demonstrated striking microvillus formation close to the level of contact with the astroglial cell monolayer (figure 2) but not at levels opposite this area.

2 1. Light microscopic appearance of FIV-MD infected FETJ cell/astroglial coculture. A. Note the spherical lymphocytes specifically attached to the barely visible nonconfluent monolayer of feline astroglial cells but not to the plastic of the tissue culture flask. (x250). B. Cross-section of FIV-MD-infected FETJ cell/astroglial coculture fixed in paraformaldehyde. (x500).

Figure 2. Transmission electron micrographaphs of FIV-MD-infected FETJ cell/astroglial cocultures. A. Transverse section cut though an FIV-MD-infected FETJ cell at a level close to its association with the astroglial monolayer. Note the striking microvillus formation at this level. B. Cross-section through both cell types showing an FIV-MD-infected FETJ cell (top) tightly attached to the surface of an astroglial cell (lower right). Bar = micrometer.

In the absence of reagents that recognize feline adhesion molecules, adherence assays were performed to investigate the mechanism of cell-cell adhesion by evaluating the degree of cell-cell interaction between uninfected and FIV-infected FETJ cells with other cells. For comparison of adhesion to feline astroglial G355-5 cells with adhesion to other cell types, similar experiments were performed with cocultures of FIV-MD-infected or uninfected FETJ cells with primary feline astroglial cells, primary astroglial cells from rats, CRFK, C-6, and SY5Y cell lines. Adherence of uninfected or FIV-MD infected feline PBMC and FIV-Pet-infected FETJ cells to G355-5 cells was also examined. Despite similar levels of infection, feline leukocytes infected with the neurovirulent Maryland strain, but not the non-neurovirulent Petaluma strain, adhered tightly and specifically to astroglial cells of feline origin but not to neural cells from other species nor non-astroglial feline cells, suggesting specific cell-cell interaction between these FIV-MD-infected FETJ cells and astroglia. In contrast, uninfected and FIV-Pet infected FETJ cells adhered only loosely, if at all. Results of experiments with primary cells (primary feline astroglia or primary feline lymphocytes) were similar to those with established cell lines. The role of specific proteins expressed on the cell surface or free in the cell culture medium of FIV-MD-infected lymphocytes in the observed interaction with astroglia was investigated by preincubation of FIV-MD-infected FETJ cells with specific polyclonal or monoclonal antibodies. Although the anti-FIV antibodies used are neutralizing for FIV-MD in other systems (e.g., infection of feline primary lymphocytes), preincubation of FIV-MD-infected lymphocytes with these antibodies did not block adhesion in the described experiments. Similarly, preincubation with antibodies to CD18 or CD69 did not prevent adhesion.

The BBB is a highly selective barrier whose structure and control is intricately linked to astrocytes (39, 40). The presence of this selective barrier between the blood and the neuropil that impedes the passive diffusion of solutes and yet allows the entry of selected substances and cells necessitates a specific transport mechanism for neurotropic viruses entering the CNS. Little is know about lymphocyte/astroglial interaction in vivo although it has been demonstrated that the CNS is continuously patrolled by small numbers of T cells and monocytes (41). These hematogenous cells may serve as a source of infectious virions for resident glia. Our findings are consistent with this generally accepted belief. We have demonstrated that infection of astrocytes via FIV-MD-infected lymphoblastoid cells is much more efficient than by free virus, a finding perhaps more meaningful in terms of viral pathogenesis than the specific cell-cell adhesion we reported (32). Immune surveillance mechanisms may be evaded by FIV-MD by direct transmission of virus across the confined space between infected lymphocytes and astroglia. Although macrophage tropism has long been a purported requirement for tissue dissemination of lentiviruses (18, 42, 43, 44, 45, 46, 47, 48), recent studies contradict this widely held belief (38, 49). Our findings and those recently reported by others necessitate investigation into the role of lymphocytes in CNS infection by neurotropic lentiviruses. Although the specific mechanism of cell-cell interaction and viral transmission is yet to be elucidated, our observations may offer insight into the cellular sequence of events which may occur during transmission of FIV from blood to brain tissue and may further the understanding of FIV- and HIV-associated neuropathogenesis.

4.2. FIV-associated cytotoxicity and neurotoxicity

Astroglial functions in the maintenance of CNS homeostasis include regulation of ion concentrations, uptake and metabolism of certain neurotransmitters such as the excitotoxic amino acid glutamate, and development and maintenance of the BBB. Although in vivo productive infection of astroglia seems to be rare, nonproductive infection of astroglial cells may play a major role in the development and progression of lentiviral-associated neurologic dysfunction. Prominent nonexclusive hypotheses to explain possible indirect mechanisms of retrovirus-associated, neuroglial cell-mediated neuronal dysfunction include direct toxicity of viral proteins, excitotoxicity from accumulation or potentiation of excitatory neurotransmitters such as glutamate, and immunologically-mediated damage from toxic products such as TNF-alpha, nitric oxide, or free radicals (23, 27, 50).

4.2.1. Laser cytometric analysis of FIV-associated cytotoxicity

The importance of astrocyte functions in neuronal homeostasis suggests that virally-induced changes in astroglia might ultimately impair neuronal function. This postulate, if proven to be true, may have direct relevance to neuropathology of lentiviral-infection in cats and humans. In consideration of this hypothesis, we performed experiments to investigate the mechanism of FIV-induced alterations in primary feline astroglial cell cultures via vital fluorescence bioassays which examine specific indicators of cellular dysfunction. The indicators of cellular dysfunction investigated and found to be abnormal in FIV-Pet-infected astroglia are mitochondrial membrane potentials, cell-cell communication via gap junctions, calcium homeostasis, plasma membrane fluidity, and intracellular glutathione concentration (3, 51).

Envelope proteins of HIV and FIV have been implicated in causing cytotoxicity and cytopathic effects (CPE) (52, 53, 54, 55, 56). Interactions between the host cell and viral proteins are postulated to cause changes in cellular membranes, interfering with normal homeostasis (57, 58). Interestingly, the primary toxic effects of a number of more classical toxins such as mycotoxins, heavy metals, and polychlorinated biphenyls (PCBs) also appear to involve cellular membranes (59, 60, 61, 62). It seems that although cellular insults are diverse, the ways that a cell can react to injury are limited (61). Direct lipid peroxidation has been implicated in cytotoxicity due to many cellular insults. Alternatively, cellular thiols may be depleted directly, sensitizing the cell to oxidative stress. In either instance, altered membrane structure may lead to release of calcium from internal stores, activation of membrane phospholipases, altered membrane permeability barriers, depletion of mitochondrial membrane potential and cellular ATP, and a decrease of cell-cell communication. Results of cytotoxicity assays suggest that membrane changes may occur very early following exposure to FIV, prior to detectable virus production, and that there are alterations of vital cellular constituents and functions associated with established persistent infection when high titers of FIV-Pet are produced. Petaluma strain FIV is cytotoxic based on all parameters tested. Interestingly, although exposure of astroglia to non-infectious cell-free neurovirulent FIV-MD causes evidence of cytotoxicity, infection per se does not. An in vivo relationship may exist when considering that there is CNS inflammation following FIV-Pet but not with FIV-MD. With FIV-Pet, astroglia may be cytopathically infected and die causing microfoci of inflammation and gliosis. If FIV-MD is not cytopathic, no inflammation would be expected. Although not conclusive, this correlation is at least consistent with the paradoxical findings of inversely related lesions and severity of clinical neurologic dysfunction. The next section discusses a functional abnormality that strengthens the evidence for this scenario.

4.2.2. Excitotoxicity in the neuropathogenesis of FIV

An excess of glutamate can be toxic to neurons. Under normal conditions, the post synaptic action of glutamate is quickly terminated by its rapid uptake by neurons and astrocytes surrounding the synaptic cleft. Glial glutamate transporters provide the majority of this uptake. Astroglial play a critical role in maintaining low extracellular glutamate and controlling the balance between physiologic excitatory transmission and excitotoxicity (63). In particular, overstimulation of NMDA receptors by glutamate leads to a series of neurotoxic events including excessive influx of calcium, generation of nitric oxide, superoxide anion, and peroxynitrite (64, 65). Glutamate also competitively inhibits neuronal uptake of cysteine, which is the limiting precursor for glutathione (66), rendering neurons more vulnerable to oxidative injury. Infusion of glutamate into the mesopontine area reduces REM sleep in cats similar to FIV-associated sleep alteration mentioned in section 2 (10). Exposure to FIV enhances sensitivity to the excitotoxic effects of glutamate in primary feline mixed neuronal-glial cultures (67). Whether this effect results from impaired glutamate uptake by astroglia or a synergistic effect of the virus and the excitatory amino acid is uncertain and may actually be a combination of both. In this section results of experiments demonstrating impaired astroglial uptake of glutamate following exposure to FIV are described. A mechanism of neurotoxicity whereby FIV infection may promote an increase in extracellular glutamate within the brain, thereby leading to neuronal excitotoxicity and increased neuronal sensitivity to oxidative stress is suggested.

To study the effects of FIV on glutamate uptake, astroglia were either exposed to cell-free virus stocks of FIV-Pet or FIV-MD for a short time period (24 hours), or infected and analyzed after persistent infection was established. Glutamate uptake into feline astroglia, both primary and the G355-5 cell line, was linear over the uptake interval studied. Because of the greater variability of response and the more heterogeneous nature of the primary cells, the G355-5 established cell line was used for most of these studies. Figure 3 demonstrates the extremely significant (p< 0.0001) inhibition of glutamate uptake by FIV-Pet within 1 day of infection compared to uninfected G355-5 cells. Effects of one day of exposure to FIV-MD are also shown in figure 3. Although there is also a statistically significant (p=0.013) decrease in glutamate uptake in G355-5 cells exposed to FIV-MD, the drop is not nearly as profound as that observed with FIV-Pet. Figure 4 illustrates the persistent defect in glutamate uptake in FIV-Pet-infected G355-5 cells through 10 days of observation postinfection.

Figure 3. The effect of one day of exposure to FIV-MD or FIV-Pet on uptake of L-[G-³H]glutamate by feline G355-5 glial cells. The bars are the mean uptake at 5 minutes ± S.D. for each group. *Significantly different from control: control differs from FIV-MD exposed (p=0.013) and FIV-Pet exposed (p=0.004), and FIV-MD exposed differs from FIV-Pet exposed (p=0.0107).

Figure 4. The effect of FIV-Pet infection on uptake of L-[G-3H]glutamate by feline G355-5 glial cells. The bars are the mean uptake at 5 minutes ± S.D. for each group. *Significantly different from control: control differs from FIV-PET day 5 (p=0.0001) and day 10 (p=0.01) and FIV-Pet infected day 5 differs marginally from FIV-PET day 10 (p=0.0888).

To study glutamate uptake in G355-5 cells infected with, rather than just exposed to, FIV-MD, coculture FIV-MD infected G355-5 cells were prepared as described above. Additionally, G355-5 cells were mock infected by coculture with uninfected FETJ cells and results of glutamate uptake were compared between untreated G355-5 cells, G355-5 cells mock infected via coculture with uninfected FETJ cells, and G355-5 cells infected via coculture with FIV-MD-infected FETJ cells. Infection of G355-5 cells with FIV-MD via coculture (as opposed to exposure to free virus) did not cause a significant decrease in glutamate uptake early (day one postinfection), but did cause a statistically significant decrease (p=0.0085) later in infection (day 10 postinfection, see figure 5). Although statistically significantly different, the glutamate uptake of cells infected with FIV-MD was still approximately 85% of control levels compared to a drop of as much as 55% for FIV-Pet infection.

Figure 5. The effect of FIV-MD-infection on uptake of L-[G-3H]glutamate by feline G355-5 glial cells. "Control cocult" represents uptake by G355-5 cells cocultured with uninfected FETJ cells as per the procedure described for infection of G355-5 cells with FIV-MD by coculture FIV-MD-infected FETJ cells. The bars are the mean uptake at 5 minutes ± S.D. for each group. *Significantly different from control: control does not differ from day 1 but differs significantly from day 10 (p=0.002).

The mechanism by which FIV induces neurologic disease remains poorly understood but is believed to involve, at least in part, dysregulation of the intricate balance of intercellular signals between glia and neurons. The data presented herein indicate that FIV is able, at least in vitro, to alter the glutamate cycle. Astroglia infected by FIV have impaired glutamate uptake capabilities but the less neurovirulent FIV-Pet strain causes more profound effects than the more neurovirulent FIV-MD strain. Excitotoxicity associated with glutamate has generally been induced by high levels of glutamate but it has recently been demonstrated that even modest increases in extracellular glutamate can have deleterious effects (68). Therefore, it is very likely that the subtle decrease in glutamate uptake in FIV-MD-infected astroglia may relate to neuropathogenesis.

The mode of neuronal cell death after damage initiated by excitotoxins has been controversial, with some groups finding features of necrosis and others reporting characteristics of apoptosis. Both these mechanisms of cell death may be initiated by the same factors depending on the initial severity of the insult and the microenvironment present at the time of injury. For example, otherwise non-toxic levels of glutamate may be potentiated by reactive oxygen species (ROS), peroxynitrite, or other toxic substances, potentially including viral envelope proteins. A major difference between the two types of cell death is the generalized involvement of neighboring cells. To prevent leakage of excitatory amino acids, proteolytic enzymes, DNA, and oxidized lipids, apoptotic cells condense their chromatin, shrink, and shield their intracellular milieu. In a sense, apoptosis is "silent" cell death. On the other hand, necrosis is caused by catastrophic injury leading to membrane lysis, release of cellular constituents, and resulting inflammation. Depending on the intensity of the initial insult, excitotoxic insults can result in either apoptotic or necrotic neuronal cell damage (68). Because FIV-MD-infected cats have evidence of neuron death without appreciable inflammation detected, it is possible that a modest decrease in glutamate uptake by noncytopathically infected astroglia leads to mild rather than large increases in extracellular glutamate with subsequent neuronal apoptosis rather than necrosis. Widespread neuronal apoptosis certainly would cause much more profound clinical disease than small focal areas of necrosis and inflammation associated with FIV-Pet.

Analysis of glutamate uptake by FIV-Pet infected astroglia showed that day one infected cells displayed the most dramatic drop in uptake and the uptake seemed to improved over the 10 days of infection. These results parallel the lipid mobility data mentioned in section 4.2.1. and cannot be correlated with the level of extracellular virus. The extracellular fluid at day one contained only one tenth the p26 present at 10 postinfection which contained the highest level of p26. This result might imply that disruption of the cell membrane structure associated with viral entry, ROS generated during infection, or some other process associated with disruption of the plasma membrane integrity is at work rather that a straightforward toxic effect of the virus or it's associated proteins. However, cells exposed to FIV-MD and astroglia infected with FIV-MD for 10 days also display decreased uptake of glutamate. Cell-free FIV-MD is not infectious for astroglia and astroglia infected via coculture as described above do not produce detectable virus in the extracellular fluid so the mechanism of glutamate uptake inhibition is not readily apparent and requires further study.

The consequences of altered glutamate uptake by FIV-infected astroglia are likely important in the understanding of AIDS associated neuropathogenesis. Excitotoxic injury of neurons, either by necrosis or apoptosis, is one possible sequela and is consistent with the reported data about these two strains of FIV. Clinical trials with NMDA antagonists are underway in the hope that blocking the receptors associated with excitotoxicity may abrogate the neurologic manifestations of HIV infection.