INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Feline immunodeficiency virus (FIV) is a lentivirus associated with immunological and neurological abnormalities in cats (34), abnormalities similar to those seen in human immunodeficiency virus type 1 (HIV-1)-infected individuals (28) and with other lentivirus infections (32). Neurological disease may develop at the onset of FIV infection, but it usually occurs as a complication of acquired immunodeficiency syndrome (AIDS) (38), which is characterized by a decline in CD4⁺/CD8⁺ lymphocyte ratios and a weight loss in adult animals (3). The most common FIV-induced neurological syndrome, FIV encephalopathy, usually occurs after AIDS develops; affected animals present with ataxia, reduced motor activity, irritability, and disorientation (38). Experimental FIV studies reveal that behavioral abnormalities occur in 20 to 40% of infected animals and are accompanied by electrophysiological abnormalities, including delayed visual and brainstem-evoked potentials, abnormal electroencephalograms, and neuroradiological abnormalities such as cerebral atrophy and white matter lesions (35, 43). Studies indicate that FIV is a neurotropic virus that infects microglia and astrocytes but not neurons in vivo (7), a finding similar to the tropism displayed by HIV-1 and simian immunodeficiency virus (SIV) in the brain. Neuropathological changes accompanying FIV encephalopathy include gliosis, white matter pallor, mineralization of the basal ganglia, and microglial nodules (1, 16). Although multinucleated giant cells have also been reported in FIV-infected cats (16), the extent of inflammation in the brain is less than with other lentiviruses, such as HIV-1, SIV, visna-maedi virus, or caprine arthritis encephalitis virus (32). As reported for HIV-1 and SIV infections, neuronal injury and death have been observed in FIV-infected animals (29, 41). These neurological and neuropathological observations suggest that FIV may share a common pathogenesis with other lentiviruses.

The biochemical mechanisms underlying lentivirus encephalopathies are uncertain but are presumed to be the consequence of direct viral infection and/or activation of glial cells, resulting in increased release of host and/or viral neurotoxic molecules (22). Neuronal injury is common to most lentivirus infections, presumably due to an excitotoxic mechanism in the brain involving increased neuronal calcium influx and/or reduced intracellular uptake of glutamate (20). These observations are supported by findings of increased glutamate levels in the brains of FIV-infected cats showing neuronal loss (41).

Individual viral strains and/or specific domains within viral genes of animal retroviruses have been shown to be responsible for neurovirulence or the development of neurological disease (15, 26, 46). Several studies indicate that some lentivirus strains may influence the extent of neurological disease (26, 36, 47), although the relationship between strain-dependent neurovirulence and systemic immune suppression remains uncertain. In the present study, we examined the extent of neurovirulence caused by different FIV strains and its relationship to systemic immune suppression. A viral strain derived from cerebrospinal fluid caused greater neuronal death as determined in an assay of neurotoxicity. Infection of neonatal animals resulted in the rapid onset of neurological disease and immune suppression in the group receiving the viral strain derived from cerebrospinal fluid.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Viruses and cell cultures. The feline immunodeficiency viruses used in this study included two primary isolates, V1CSF and Petaluma (a gift from N. C. Pedersen), that underwent no more than five in vitro passages before the present experiments. FIV V1CSF was isolated from a cat with encephalopathy in Baltimore, Md., as previously reported (41). Each virus was grown and titers of the virus were determined on feline peripheral blood mononuclear cells (PBMC) in RPMI 1640 with 15% fetal calf serum (FCS) and 1% penicillin and streptomycin, initially stimulated with concavalin A (5 µg/ml) for 3 days and subsequently treated with interleukin-2 (100 U/ml). Tissue culture infectious doses (TCIDs) of each virus were determined by limiting dilution in PBMC, as detected by reverse transcriptase (RT) assay. Feline macrophages were prepared from PBMC by adherence on polystyrene flasks for 3 days, following PBMC isolation in RPMI 1640 in 10% FCS and 1% penicillin and streptomycin, and then seeded at 10⁵ cells per well in a 96-well plate. Mixed glial cell (MGC) cultures were prepared from healthy adult cat cerebrum in Dulbecco modified Eagle medium in 10% horse serum with antibiotics, as described above and as previously reported (40), and seeded at 10⁵ cells per well in a 96-well plate. At 7 days postadherence, macrophage and MGC cultures were infected in triplicate wells with each virus at a titer of 10³ 50% TCID (TCID₅₀)/0.1 ml or heat-inactivated virus (boiled for 10 min). After infection of cultures, supernatants were harvested every 3 days for RT assay.

Neurotoxicity assay. Neuronal cultures were prepared from 12- to 15-week human gestational fetuses with the approval of the Human Ethics Committee at the University of Manitoba as previously reported (kindly provided by A. Nath [33, 40a]). This culture system was selected because FIV infection of human astrocytes and microglia is minimal (19). Hence, the neurotoxicity induced by FIV-infected supernatants could be assessed directly, avoiding infection of glial cells which may influence the extent of neurotoxicity. Briefly, the tissue was mechanically dissociated, and the cells were resuspended in Opti-MEM (Gibco) with 1% heat-inactivated FCS, 1% N₂ supplement (Gibco), and 1% antibiotic solution (10⁴ U of penicillin G per ml and 10 mg of streptomycin B per ml in 0.9% NaCl) and then seeded in 96-well microtiter plates at 10⁵ cells per well and maintained for a minimum of 4 weeks prior to experimental use. Sample wells were immunostained for the neuronal marker, microtubule-associated protein 2 (MAP-2), and only cultures in which >70% of the cells were MAP-2 immunopositive were used for experiments. The remaining cells were principally astrocytes, as indicated by glial acidic fibrillary protein immunopositivity with rare microglia (>1%), which immunostained with EBM-11 (39). Macrophage culture supernatants were harvested at 4, 8, and 12 days postinfection, saved at 80°C, mixed (50 µl) with Opti-MEM (0.1% FBS) (50 µl), and applied to the cultured neurons for 3, 6, or 12 h. The neuronal cultures were stained subsequently with trypan blue and fixed in 4% paraformaldehyde, and the number of neurons with trypan blue-positive nuclei were counted per unit area over five randomly chosen fields by an examiner who was unaware of the specific treatment (40a). Neuronal death was expressed as the percentage of trypan blue-stained neurons to the total number of neurons counted. Background neuronal death varied from 4 to 8% among cultures, depending on the age of the fetus and the duration in culture; thus, background neuronal death was subtracted from the level of toxicity for each experiment. Individual experiments were conducted in triplicate wells and repeated at least twice.

Animals and virus inoculation. Six specific-pathogen-free pregnant cats (queens) were obtained through the University of Manitoba Animal Services. All queens were negative for feline retroviruses as determined by PCR. At day 1 postdelivery, all kittens were inoculated intracranially in the right frontal lobe with 0.2 ml of titered virus (10³ TCID₅₀/0.1 ml); V1CSF, n = 8; Petaluma, n = 6) or heat-inactivated virus (control; n = 8) by a 30-gauge needle and syringe. Preliminary studies suggested that V1CSF and Petaluma differed in the extent to which neurological disease and immune suppression was induced. Hence, animals infected with Petaluma (n = 3) were treated with cyclosporin A (CyA), which is a potent inhibitor of T-cell function and proliferation, from 8 to 12 weeks postinfection. CyA was administered daily at doses of 2.5 to 7.5 mg/ml subcutaneously. Weekly measurements of renal function and CyA serum levels were performed, ensuring that CyA was maintained in the therapeutic range (ca. 500 ng/ml), as recommended (14). Only one viral strain was used for each litter to avoid cross-contamination and at least two litters were infected per viral strain. Kittens were weaned at 6 weeks and monitored until 12 weeks of age, at which time they were euthanatized. Blood samples were obtained at 8 and 12 weeks postinfection from the kittens from which PBMC had been prepared as described above.

Behavioral studies. To determine neurobehavioral and developmental features associated with FIV infection, the animals were examined weekly and weighed by animal care staff who were unaware of the infection status. The age (weeks) was recorded at which each developmental milestone was manifested; these included playful interaction, walking, running, air righting, the ability to walk along a plank, and blink reflex, as adapted from Villablanca and Olmstead (48). In addition, the height to which an animal jumped, pursuing the moving light on a wall, was measured. At 12 weeks following infection, five different parameters, including activity level, play interaction, motor ability, inquisitiveness, and general health, were scored by using the Feline Behavioral Scale (FBS). A technician, unaware of the animal's infection status, ranked each parameter depending on the level of impairment (5 [none] to 1 [extreme]) with a maximal score of 25.

In vivo magnetic resonance spectroscopy (MRS). Animals were anesthetised with 1.5 to 2.0% halothane administered through a nose cone; the animals were then placed in an animal holder, and the coil was taped into place on the head. Localized ¹H MR spectra of the left frontal cortex were obtained on week 12 postinfection by using the STEAM localization method. A mixing time (TM) of 30 ms, an echo time (TE) of 20 ms, and a repetition time (TR) of 2 s were used. The voxel (VOI) size was 4 mm³ and was positioned by using a scout image. Water suppression was accomplished by using a CHESS sequence preceding the acquisition. Localized shimming and radiofrequency amplitude adjustment was accomplished by using the signal obtained from the VOI with the water suppression turned off. A total of 512 signal averages were acquired, requiring a total acquisition time per spectrum of 17.1 min. An elliptical surface coil (5 cm × 2 cm) was used for the acquisition of both scout images and spectra. The spectrometer used was a 7-T 21-cm Bruker/Biospec2 equipped with an actively shielded gradient coil set of ID 11.6 cm. Spectra were processed by anodizing with a 15-Hz exponential multiplication, Fourier transformation, and phasing, after which a modest polynomial baseline correction was necessary. Peak intensities were then measured and used to calculate metabolite ratios relative to creatine, which is a stable metabolite in the brain (42).

Flow cytometry. Blood was drawn at 8 and 12 weeks postinfection. PBMC were divided into three equal aliquots. Each aliquot (0.5 × 10⁶ to 0.7 × 10⁶ cells) was incubated for 1 h on ice with either murine antifeline CD4⁺ or CD8⁺ monoclonal antibodies (clones FE1.7B12 and FE1.10E9, respectively; LABL, Davis, Calif.) or RPMI 1640 tissue culture media supplemented with 10% FCS. The cells were then washed two times with serum-free RPMI 1640 medium to remove the unbound antibodies. The cells were resuspended in 100 µl of RPMI 1640 medium supplemented with 10% FCS, and 5 µl of goat anti-mouse immunoglobulin G1 conjugated with fluorescein isothiocyanate (Cedarlane Laboratory, Ltd., Hornby, Ontario, Canada) was added followed by incubation on ice for 1 h. The cells were washed again as described above and then resuspended in 200 µl of 0.5% paraformaldehyde in phosphate-buffered saline. Flow cytometry analysis was performed by using a Coulter Electronics EPICS 753 cell sorter with the argon ion laser excitation set at 488 nm (500 mW).

Morphological studies. At 12 weeks of age the animals were deeply anesthetized and frontal craniotomies were performed to remove the left frontal lobes. The abdomen of each animal was opened, and the spleen was removed. Each animal was then perfused with 4% paraformaldehyde, and the brain was fixed for 1 week, embedded in paraffin, sectioned, and stained with hematoxylin, eosin, and luxol fast blue. Histopathological analysis was performed on the frontal and temporal lobes, thalamus, basal ganglia, brainstem, and cerebellum by a qualified neuropathologist (M.R.D.B.). Each left frontal lobe was frozen on dry ice and stored at 80°C for the PCR studies. Fixed brain tissue from the frontal lobe, including the cortex, white matter, and deep nuclei, was prepared for the immunocytochemistry analyses as described previously (41) and immunostained with antibodies to FIV p24 (clone 43-1B9, a gift from N. C. Pedersen).

Viral detection. cDNA was synthesized from total RNA extracted directly from the frontal lobe and the spleen of animals from each group, as previously reported (50). The viral genome was detected by amplification of a conserved region of the FIV pol gene by nested PCR (41). Each reaction was carried out in a mixture containing 0.2 mM each dNTP, 2.5 mM MgCl₂, 0.2 µM concentrations of each primer, 5 mM KCl, 1 mM Tris-HCl, and 0.25 U of Taq polymerase, to which 2 µl of template DNA was added for a total volume of 25 µl. A 770-bp fragment representing positions 3361 to 4131 of the pol gene was amplified in an initial reaction of 30 cycles consisting of 95°C/min of denaturation, 40°C/min of annealing, and 72°C/2 min of elongation by using primers 3361 (5'-AAGGATCCAGAAAAGATACTATGG-3') and 4131C (5'-GGCAACATTAGCTTTACCCCTGTTGG-3'). With 2 µl of the primary PCR product as a template, a 192-bp fragment corresponding to positions 3860 to 4052 of the pol gene was amplified by using identical reagents in a second reaction of 35 cycles at an annealing temperature of 50°C with primers 3860 (5'-CCAGATATGATGGAGGGAATCT-3') and 4052C (5'-CATATCCTGCATCTTCTGAACT-3'). A 26-mer oligonucleotide probe (5'-TGTCAAACAATGATGATAATAGAAGG-3') was labeled with [-³²P]dCTP and was used to probe the transferred gel by routine Southern blot procedures that were designed to recognize positions 3986 to 4012 contained in both reaction products.

Statistical tests. The statistical tests comparing groups were made by using nonparametric (Spearman correlation) or parametric (analysis of variance [ANOVA]; Student's t test) analyses.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Ex vivo viral tropism and neurotoxicity. To determine the tropism of the two FIV strains, adult feline macrophages and MGC cultures were infected with V1CSF or Petaluma (Fig. 1A). Viral replication in macrophage and MGC cultures, as indicated by mean fold increases in RT activity in infected cultures relative to uninfected cultures, peaked at 8 days postinfection for both V1CSF- and Petaluma-infected macrophage cultures. Infected MGC showed significantly lower peak RT activities than did the macrophage cultures infected by either virus (P < 0.05). During the period of infection, the RT values did not differ significantly between the two viruses at each time point for both macrophage and MGC cultures.

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Ex vivo macrophage and MGC tropism (A) and neurotoxicity (B and C) of V1CSF and Petaluma. (A) The fold increase in supernatant RT activity above uninfected cultures (± standard deviation) did not differ between viruses at any time point, although the peak RT activity was higher in the macrophage cultures than in the MGC cultures. Mean RT levels in uninfected macrophage and MGC culture supernatants ranged from 38 ± 10 to 56 ± 10 cpm/10 µl and did not differ significantly between time points (ANOVA, P > 0.05). (B) Neuronal death induced by supernatants from V1CSF-infected macrophages was significantly higher than Petaluma or control supernatants harvested at 4, 8, and 12 days postinfection. (C) Comparison of neuronal killing after 3, 6, and 12 h of exposure to CM revealed that V1CSF induced significantly greater neurotoxicity, although the extent of the neurotoxicity did not differ between exposure times (Student's t test; **, P < 0.001; *, P < 0.05).

Neurobehavioral studies. To determine whether individual FIV strains differed in their in vivo neurovirulence, we developed a model of FIV infection in which neurodevelopmental skills were compared in neonates infected with V1CSF or Petaluma or sham infected with heat-inactivated virus (V1CSF) (Table 1). The groups did not differ in the mean age at which they acquired neurodevelopmental skills until 6 weeks, when the V1CSF group were observed to be significantly delayed in their ability to perform the plank walk task compared to controls and the Petaluma group. In addition, the V1CSF and Petaluma-CyA groups displayed significant delays in blink reflex acquisition compared to the control and Petaluma groups. At 11 weeks, the V1CSF and Petaluma-CyA groups showed a significantly diminished mean jump height when pursuing a moving light compared to the control and Petaluma groups. At 12 weeks, the total FBS score was significantly less in the V1CSF and Petaluma-CyA groups compared to the control group. Urea and creatinine levels in the Petaluma-CyA group remained within the normal values (5 to 10 mmol/liter and 75 to 180 µmol/liter, respectively [14]) throughout the experiments, indicating that the neurobehavioral abnormalities were not due to CyA toxicity. Neurobehavioral and weight gain differences were not observed among individual litters infected with the same virus (Students t test, P > 0.05). These findings suggested that the neurodevelopmental delay induced by FIV infection might be strain specific.

MRS studies. To determine whether the neurodevelopmental delays observed above were associated with neuronal injury, we examined N-acetyl aspartate (NAA) levels in the brain, since this has been shown to be a reliable indicator of neuronal integrity (42). Spectra were recorded from the left frontal cortex (Fig. 2A) at 6 and 12 weeks postinfection. Comparison of the spectra at 12 weeks revealed that the NAA levels were reduced in the V1CSF group but that the Petaluma and control groups did not differ (Fig. 2B). When NAA and choline (Cho) levels were expressed as a ratio, relative to creatine (Cr), the mean NAA/Cr and Cho/Cr ratios did not differ among groups at 6 weeks (Fig. 3A). Conversely, the mean NAA/Cr ratio was significantly less in the V1CSF and Petaluma-CyA groups compared to the control and Petaluma groups, but the mean Cho/Cr ratios did not differ between groups at 12 weeks postinfection (Fig. 3B). Comparisons of NAA/Cr and Cho/Cr ratios in the thalamus did not reveal significant differences among the groups (data not shown). When all of the animals were compared at 12 weeks, NAA/Cr ratios were found to be significantly correlated with FBS scores (Spearman, r = 0.83; confidence interval = 0.43 to 0.95; P < 0.005). These observations implied that neurodevelopmental delay caused by different strains of FIV may occur due to neuronal injury and/or death but that systemic immune suppression also influences neurovirulence.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   In vivo MRS of left frontal cortex. (A) Spectra were recorded from deep in the mid-frontal sulcus. (B) Comparison of representative spectra from each group of animals shows peaks corresponding to NAA, Cho, and Cr peaks. NAA peaks were lower in the V1CSF group than in the Petaluma and control groups.

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Comparison of NAA/Cr and Cho/Cr (± standard errors of the mean) ratios in the frontal cortex at 6 (A) and 12 (B) weeks postinfection. (A) The mean NAA/Cr and Cho/Cr ratios did not differ significantly between the groups at 6 weeks. (B) In contrast, the mean NAA/Cr ratios were significantly lower in the V1CSF (n = 4) and Petaluma-CyA (n = 3) groups than in the control (n = 4) and Petaluma (n = 4) groups (Student's t test, P < 0.01), but the mean Cho/Cr ratios did not differ among the groups. Two control animals were excluded from the analysis since they represented extreme outliers (greater than 2 standard deviations, P < 0.025).

CD4⁺ and CD8⁺ lymphocyte counts. Since enhanced systemic immunosuppression appeared to influence neurological disease, fluorescence-activated cell sorter analysis of PBMC from the animals was performed at 8 weeks postinfection, revealing that the mean percentage of CD4⁺ cells did not differ among the control, V1CSF, and Petaluma groups (Fig. 4A). At 12 weeks, however, the mean percentage of CD4⁺ cells was significantly lower in the V1CSF animals compared to the control and both Petaluma groups (P < 0.001). Comparison of the mean percentage of CD8⁺ cells at 8 weeks (Fig. 4B) indicated that both the Petaluma and the V1CSF groups were significantly higher than the control group (P < 0.002). At 12 weeks, the mean percentage of CD8⁺ cells was significantly higher in the Petaluma group compared to the V1CSF and control groups (P < 0.001). However, the percentage of CD8⁺ cells in the Petaluma-CyA group was significantly lower than the Petaluma group at 12 weeks, indicating that low CD8⁺ cell counts were common to both groups of animals with neurodevelopmental delay and reduced NAA/Cr ratios.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Mean CD4+ (A) and CD8+ (B) cell percentages in PBMC from control, V1CSF, and Petaluma groups at 8 and 12 weeks postinfection. CD4+ percentages did not differ at 8 weeks among the groups but were significantly lower in the V1CSF group at 12 weeks postinfection than in the control and Petaluma groups (ANOVA, P < 0.001). CD8+ percentages were significantly higher in the Petaluma group than in the V1CSF and control groups at both 8 weeks (P < 0.001) and 12 weeks (P < 0.001). The Petaluma-CyA and Petaluma CD8+ cell counts did not differ at 8 weeks, but they differed significantly at 12 weeks postinfection (P < 0.001; Student's t test).

Systemic illnesses. To examine other systemic features of FIV infection, weekly determinations of body weights indicated that weight gain occurred in all of the groups but that in the V1CSF group. The mean weights were significantly lower at 6, 9, and 12 weeks postinfection than those of the control and the Petaluma-infected groups (Fig. 5). Two animals in the V1CSF group died at 4 and 6 weeks postinfection of pneumonia and diarrhea, respectively. Similarly, three animals infected with V1CSF at an input titer of 10⁴ TCID₅₀ were euthanatized at 4, 5, and 6 weeks postinfection after being judged by the University Veterinary Services to be too ill, as evidenced by weight loss, gait ataxia, and/or inability to feed, to continue in the studies. These findings also suggested that marked systemic disease accompanied neurodevelopmental delay and reduced NAA/Cr ratios.

View larger version (21K): [in this window] [in a new window]   FIG. 5.   Mean body weights of FIV-infected and control animals over 12 weeks. The control and both Petaluma groups did not differ over the entire experimental period, but at 6, 9, and 12 weeks postinfection the V1CSF group showed significantly lower mean weights than did the other groups (Student's t test; *, P < 0.05; **, P < 0.001).

Viral detection. To determine whether viral RNA was present in the tissues of infected animals, nested RT-PCR revealed that the FIV genome was detectable in the brains of all of the animals infected with V1CSF or Petaluma at 12 weeks postinfection but not in the control animals (Fig. 6). Analysis of RT-PCR products after one round of PCR at 30 cycles revealed detectable FIV genome in two of six animals in the V1CSF group, three of six animals in the Petaluma group, all animals in the Petaluma-CyA group, and in none of the controls. In matched spleen samples, the viral genome was detectable by nested PCR in all infected animals and in none of the controls. Immunodetection of FIV p24 in sections from the frontal lobe revealed no p24-positive cells in the four controls. In contrast, p24 was detected in parenchymal and perivascular glia in frontal-lobe sections from both the V1CSF (3 of 4) and Petaluma (3 of 4) groups at a frequency of 1 to 2 immunopositive cells (usually perivascular) per section (1.5 cm²). These results suggest that the viral burden in the brain was low and did not differ between groups.

View larger version (24K): [in this window] [in a new window]   FIG. 6.   Southern blot detection of nested PCR amplification of FIV pol cDNA in the left frontal lobe from sham-inoculated control (n = 6), V1CSF-infected (n = 6), and Petaluma-infected (n = 6) animals at 12 weeks postinfection. Representative animals from each group are shown except for the Petaluma-CyA group. The FIV genome was detectable in the brains of all Petaluma-, Petaluma-CyA-, and V1CSF-infected animals but not in the controls.

Neuropathology. The control sections revealed no abnormalities. Sections from the V1CSF, Petaluma-CyA, and Petaluma groups showed small collections of lymphocytes and monocytes in the meninges, choroid plexus, and brain parenchyma of infected animals in each group, but no marked differences between groups were observed. Multinucleated cells and opportunistic processes were not observed in any group.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In the present study, we have shown that neurodevelopmental delay and reduced NAA/Cr ratios occurred in V1CSF-infected but not in Petaluma-infected animals, suggesting that neurovirulence in the present model is viral strain specific. These in vivo observations are supported by ex vivo studies in which V1CSF induced greater neurotoxicity than did Petaluma despite a similar tropism in macrophages and MGC. However, systemic abnormalities, including a decline in CD4⁺ and CD8⁺ cells in the blood and a reduced weight gain, were associated with the development of neurovirulence among the V1CSF-infected animals. Increased systemic immunosuppression with CyA treatment of Petaluma-infected animals also enhanced neurodevelopmental delay and reduced NAA/Cr ratios in brain. Taken together, these findings indicate that individual FIV strains differ in their capacities to induce neuronal injury in vivo and ex vivo but that systemic immune suppression is a requisite feature of in vivo FIV-mediated neurovirulence in neonates.

Several host and viral proteins released by cells of macrophage lineage have been proposed as neurotoxins in lentiviral infections (20). An indirect mechanism of neuronal injury and/or death is plausible because of the lack of evidence to suggest that FIV infects neurons directly and the present findings, in which molecules released from FIV-infected macrophages were neurotoxic, results similar to those of studies of other lentiviruses (11, 31). We used CM from 4, 8, and 12 days postinfection because preliminary studies from our laboratory and earlier studies (10) suggested that viral protein levels in CM varied over time. However, the peak neurotoxicity was observed with CM from day 4, suggesting that viral production from macrophages is not correlated with the extent of neurotoxicity. To some extent, these ex vivo observations were also reflected in the current in vivo findings of low viral burden in the brains of animals infected with either virus. Nevertheless, it is possible that neuronal death occurred due to the release of viral regulatory proteins expressed early in infection (33), but these proteins were not measured in the present studies. Neuronal death rates were lower in the present studies than the levels reported for other similar neurotoxicity assays (11), which may be due to the use of fetal neurons, a heterogeneous neuronal population of which only a subpopulation may be vulnerable to neurotoxins (27). While it is conceivable that FIV-infected astrocytes in the human fetal brain cultures released neurotoxic compounds, other studies from our group indicate that FIV infection of human fetal astrocytes is minimal (19). The findings of viral strain-specific induction of neurotoxicity may account for the in vivo differences in neurovirulence observed in the present study.

Studies of perinatal HIV-1 infection show neurodevelopmental delay in as many as 50% of infected children (2) and that it may be accompanied by diminished NAA/Cr ratios in pediatric brains (6). However, the occurrence of HIV-1 and SIV infection in children and young animals is usually accompanied by the entry of inflammatory cells into the brain (44, 45). Studies with adult SIV-infected animals or HIV-infected adults indicate that inflammatory changes in the brain are not necessarily correlated with neurocognitive dysfunction (12, 30). In the present study, there was limited inflammation observed in the brains of animals with neurobehavioral and MRS evidence of neuronal dysfunction, suggesting that in the FIV model inflammation is a limited predictor of neurobehavioral abnormalities. In addition, unlike our earlier report of V1CSF-infected adult animals (41), neuronal loss was not apparent on the histological sections, despite the abnormalities observed by MRS. Detailed neuronal morphological studies and cell counts of the present animals may reveal differences between groups, as reported for SIV and HIV encephalopathies in adults (8, 23). However, neuronal loss has not been a principal feature of HIV-infected children with neurodevelopmental delay, perhaps reflecting an increased capacity of neurons in the developing brain to resist structural injury even though they may manifest chemical abnormalities. This supposition is confirmed by the significant improvement in total intelligence quotient scores among HIV-infected children following aggressive treatment with zidovudine (37). The present findings taken together with earlier clinical-pathological correlations of both HIV and SIV infection suggest that careful analysis of neurocognitive function is a sensitive parameter by which the relative neurovirulence of individual viruses may be compared.

The development of primary HIV-induced neurological disease, such as encephalopathy, usually occurs in the setting of marked immune suppression (28). A limited correlation between brain virus load and HIV encephalopathy has been observed (18, 21). In contrast, systemic and/or cerebrospinal fluid viral loads were stronger predictors of the severity of encephalopathy, implying that systemic factors such as immune suppression are also important determinants of neurological disease (5). Reduced CD4⁺ cell counts have been reported for adolescents and adults infected with different lentiviruses, including FIV (3), SIV (25), and HIV-1 (13), although the relationship between systemic immune suppression and neurological disease is not well defined. Among FIV-infected adult animals, an inversion of CD4/CD8 ratios has been associated with the development of neurological abnormalities (38a). The current studies showed a decline in CD4⁺ cells with the occurrence of neurovirulence among the V1CSF-infected animals. As in earlier studies (9), we observed a rise in CD8⁺ cell counts among infected animals at 8 weeks postinfection compared to controls. However, CD8⁺ levels were significantly lower in the V1CSF and Petaluma-CyA groups, both of which showed neurological abnormalities, compared to the Petaluma group at 12 weeks. CD8⁺ T cells, acting as cytotoxic lymphocytes, may determine the extent to which FIV is cleared from the blood or brain and to which the disease progresses (17). Future studies focusing on the mechanism(s) by which systemic immune suppression, especially the CD8⁺ cell decline (49), influences the development of neurovirulence are likely to reveal clues to lentivirus neuropathogenesis.

Among several retroviral models of neurovirulence, individual viral strains and specific domains within viral genes are responsible for the development of neurological disease (15, 26, 46). The present study confirms that the viral strain is also an important determinant of lentivirus-induced neurological disease, although the viral gene(s) responsible for inducing neurovirulence remains uncertain. It is notable that Petaluma and V1CSF were derived from different tissue compartments and that sequence analysis of the surface unit envelope region revealed multiple differing residues between the two viruses used in the present studies (19). Although Petaluma and V1CSF did not differ in their tropism for macrophages, MGC, or PBMC, the variation in env may have other effects, such as influencing posttranslational processing and the transport of envelope proteins, which has been correlated with neurovirulence in murine retroviruses (24). Studies mapping the env domains responsible for neurodevelopmental delay are currently in progress and may provide insight into the mechanisms of FIV-induced neurovirulence.