Role of Microglial Cells in Selective Replication of Simian Immunodeficiency Virus Genotypes in the Brain

Abstract

An accelerated, consistent macaque simian immunodeficiency virus (SIV) model in which over 90% of pigtailed macaques (Macaca nemestrina) coinoculated with SIV/17E-Fr and SIV/DeltaB670 developed encephalitis was used to determine whether central nervous system (CNS) lesions are associated with the replication of specific genotypes in the brain and, more specifically, in the microglia. Ten of 11 inoculated macaques had severe (n = 3), moderate (n = 5), or mild (n = 2) encephalitis at 3 months postinoculation. To compare actively replicating viral genotypes in the CNS and in microglia with those in the periphery, the V1 region of the SIV envelope gene was amplified and sequenced from RNA extracted from basal ganglia, from microglial cells isolated from the brain, and from peripheral blood mononuclear cells (PBMC) isolated from blood at the time of death. To distinguish between actively replicating with latent viral genotypes in the CNS, viral genotypes in RNA and DNA from basal ganglia were compared. Two macrophage-tropic, neurovirulent viruses, SIV/17E-Fr and SIV/DeltaB670 Cl-2, predominated in the brain RNA of macaques with encephalitis, comprising 95% of the genotypes detected. The same two viral genotypes were present at the same frequencies in microglial cell RNA, suggesting that microglia are pivotal in the selective replication of neurovirulent viruses. There was a significantly greater number of viral genotypes in DNA than there were in RNA in the brain (P = 0.004), including those of both the macrophage- and lymphocyte-tropic viral strains. Furthermore, significantly fewer viral genotypes were detected in brain RNA than in PBMC RNA at the time of death (P = 0.004) and the viral strain that predominated in the brain frequently was different from that which predominated in the PBMC of the same animal. These data suggest that many viral genotypes enter the brain, but only a limited subset of macrophage-tropic, neurovirulent viruses replicate terminally in the brains of macaques with encephalitis. They further suggest that the selection of macrophage-tropic, neurovirulent viruses occurs not at the level of the blood-brain barrier but at a stage after virus entry and that microglial cells may play an important role in that selection process.

Human immunodeficiency virus (HIV), the cause of AIDS, induces dementia in a significant proportion of HIV-infected patients (39). However, the role(s) of specific viral strains in the development of encephalitis is not well defined. Simian immunodeficiency virus (SIV) infection of macaques induces clinical diseases and pathological changes that are strikingly similar to HIV infection in humans (18, 20, 34) and provides an excellent model from which to examine the pathogenesis of HIV encephalitis (23, 42, 51, 54).

SIV and HIV exist in their hosts as quasispecies, and strains within a single host may vary in cell tropism and their ability to cause disease (12, 19, 26). By definition, lymphocyte-tropic strains replicate in primary lymphocytes and in lymphocyte cell lines but not in macrophages. These strains cause immunosuppression, leading to AIDS and opportunistic infections. In contrast, macrophage-tropic strains replicate in macrophages and primary lymphocytes and cause organ-specific disease syndromes, including encephalitis (17, 37, 65) and pneumonia (8, 16, 36), and can also lead to immunosuppression and AIDS.

Cells of macrophage lineage are the main cellular targets for HIV and SIV replication in the central nervous system (CNS) (1, 10). Several studies point to activated microglia as a major player in CNS dysfunction in HIV-infected individuals. Virus-infected microglia release neurotoxic products, such as viral proteins (9), cytokines (24), reactive oxygen species (40), and metalloproteinases (27, 35), which may contribute to motor and/or cognitive deficits.

The majority of studies comparing viral genotypes in the brain and the periphery have examined only genotypes in DNA (15, 21, 32, 45, 47, 49, 60, 64). Based on the envelope hypervariable region sequences generated from DNA extracts, a number of studies have shown that viral genotypes in DNA from brain tissue differ from those identified in peripheral blood mononuclear cells (PBMC), lymph nodes, the spleen, or bone marrow, although this finding is not universal (49). Power and coworkers showed that HIV-induced dementia was associated with the presence of distinct envelope DNA sequences in the brain and that the brain-derived HIV type 1 envelope from demented HIV-infected patients can induce neuronal death in cultured macrophages (47, 48). Based on pol gene sequence analysis, Wong and coworkers showed that genotypes from brain DNA were phylogenetically distinct from genotypes in the spleen and lymph node (62). In that study, some genotypes in RNA were examined but genotypes in RNA and DNA from the same brains were not compared, probably because of difficulties extracting brain RNA from HIV-infected individuals after substantial postmortem intervals.

While genotypes from brain DNA reflect viral strains that have entered the brain at some time during infection and may include both latent and actively replicating strains, genotypes from RNA represent viral strains that are replicating at the time of death and that are likely responsible for the lesions seen in the brain at that time. Further, genotype studies using human tissues are complicated by the presence of blood in the vasculature at autopsy, which may confound comparisons of the relative frequency of various viral strains in various tissues. To the best of our knowledge, this is the first study to compare genotypes in DNA and RNA from fresh, saline-perfused brain samples with the genotypes replicating in PBMC and in microglia, a pivotal target cell in the brain. The data show that two macrophage-tropic, neurovirulent genotypes, SIV/17E-Fr and SIV/DeltaB670 Cl-2, predominated in RNA from the brains of macaques with encephalitis, comprising 95% of the 800 genotypes screened. Furthermore, the two genotypes were detected at the same frequencies in RNA extracted from microglia isolated from the brain. In contrast, a variety of macrophage- and lymphocyte-tropic genotypes were identified in brain DNA and found to replicate in PBMC. These data suggest that the brain is exposed to a wide variety of macrophage- and lymphocyte-tropic viruses but that SIV encephalitis is associated with the active replication of specific neurovirulent virus strains, particularly in microglial cells.

MATERIALS AND METHODS

Viruses and animal inoculation.

The SIV/DeltaB670 stock used in these experiments was originally obtained by coculturing lymph node tissue from rhesus monkey B670 that had been infected with SIVsmm grown in primary human phytohemagglutinin-stimulated PBMC (41, 63). Virus stocks were produced by infecting primary rhesus peripheral blood lymphocytes. The 50% animal infectious dose (AID₅₀) of SIV/DeltaB670 was determined by inoculating macaques with 10-fold dilutions of cryopreserved virus stocks. The AID₅₀ was calculated using software developed by John Spouge (57). The SIV/DeltaB670 stock consists of at least 22 different genotypes characterized by nucleotide sequencing of the hypervariable region V1 of the envelope gene (2) and includes both the macrophage- and lymphocyte-tropic strains (6, 7, 50). SIV/DeltaB670 induces the full range of SIV-associated disease, including immunosuppression, encephalitis (37, 65), and pneumonia (8, 36).

SIV/17E-Fr is a molecularly cloned macrophage-tropic virus obtained by inserting the entire env and nef genes and the 3′ long terminal repeat of the macrophage-tropic, neurovirulent virus strain SIV/17E-Br into the backbone of the lymphocyte-tropic clone SIVmac239 (22). SIV/17E-Br was derived from SIVmac239 by serial passage in rhesus macaques and was isolated from the brain of a macaque with fulminant encephalitis (4, 55). SIV/17E-Fr preferentially infects macrophages (22) and causes encephalitis (37, 65). Infectious virus stock was made by transfecting SIV/17E-Fr into CEMx174 cells. Eleven 3- to 5-year-old pigtailed macaques (Macaca nemestrina) were intravenously inoculated with a mix of the two virus stocks: SIV/DeltaB670 (50 AID₅₀s) and SIV/17E-Fr (10,000 AID₅₀s). Macaques were euthanized at 3 months postinoculation in accordance with federal guidelines and institutional policies. At euthanasia, macaques were perfused with sterile saline to remove blood from the vasculature, thus allowing the accurate determination of viral genotypes in the brain parenchyma. Complete necropsies were performed, and all tissues were examined microscopically by two pathologists (M.C.Z., J.L.M.).

CD4⁺-cell counts.

Peripheral blood was taken from infected and control macaques on days 7, 10, 14, and 28 postinoculation and every 2 weeks thereafter. Complete blood counts with differentials were performed using Cell-Dyne 3500 (Abbott, Dallas, Tex.). Cells were labeled with fluorochrome-conjugated monoclonal antibodies (Becton Dickinson, Bedford, Mass.) to identify CD4⁺ cells as previously described (38). The absolute CD4⁺-cell counts were determined by multiplying the percentage of CD4⁺ cells by the absolute lymphocyte count.

Pathological assessment.

Sections of frontal and parietal cortex, basal ganglion, thalamus, midbrain, and cerebellum were examined microscopically and scored as mild, moderate, or severe by using the following semiquantitative system. Sections with more than 30 perivascular macrophage-rich cuffs were given a score of 3, sections with 10 to 30 perivascular cuffs were given a score of 2, and those with fewer than 10 perivascular cuffs were given a score of 1. The scores for all sections were totaled and divided by 6 (six regions were graded for each brain) to give a mean score (out of a maximum of 3) for the severity of CNS lesions. Animals with a mean encephalitis score of 2 to 3 were listed as having severe encephalitis, those with a score between 1 and 1.9 had moderate encephalitis, and those with a score between 0 and 0.9 had mild encephalitis.

Cell culture and in vitro infection.

To culture blood-derived lymphocytes, PBMC were isolated from the blood of the uninfected pigtailed macaque by centrifugation on Percoll gradients (Amersham Pharmacia, Uppsala, Sweden), phytohemagglutinin stimulated for 3 days, and then cultivated at 2 × 10⁶ cells/ml in RPMI 1640 medium containing 10% fetal bovine serum (FBS), 0.5 mM sodium pyruvate, 1 mM l-glutamine, 20 μg of gentamicin/ml, and 10 U of human recombinant interleukin-2)/ml. Blood-derived macrophages were obtained by cultivating PBMC for 5 days in macrophage differentiation medium (RPMI medium containing 10% human AB-type serum and 1,000 U each of human recombinant granulocyte-macrophage colony-stimulating factor and macrophage colony-stimulating factor/ml [Genetics Institute, Inc., Cambridge, Mass.]). Nonadherent cells were removed by vigorous washing. Microglial cells were isolated from the uninfected macaque by using the protocol detailed below. Interleukin-2-activated lymphocytes, blood-derived macrophages, and microglial cells were infected with SIV/DeltaB670 (50,000 cpm of reverse transcriptase [RT] activity), and RT activity in the supernatant was measured every 2 days. At 14 days after infection, cells were washed to remove free virus, macrophages and microglial cells were trypsinized for 30 min at 37°C, and total RNA was isolated.

Isolation and culture of microglial cells.

Microglia were isolated from the cerebral cortexes of 6 of the 11 inoculated macaques (macaques V387, V389, V394, V708, V713, and V715) by adapting a previously described protocol (25, 56). Fresh brain tissue was washed, minced, pelleted, and then washed twice with phosphate-buffered saline solution. Tissue was digested in trypsin-DNase solution (Dulbecco's modified Eagle medium [DMEM] supplemented with 0.25% trypsin, 50 μg of DNase/ml, and 50 μg of gentamicin/ml) for 30 min at 37°C with agitation. Trypsinized tissue was filtered through 183-μm-pore-size sterile mesh followed by 100-μm-pore-size mesh, washed once with DMEM-10% FBS, and then pelleted. Cells were resuspended in phosphate-buffered saline, mixed with Percoll, and then centrifuged at 411,000 × g for 30 min. Microglial cells were harvested from the gradient layer below the myelin cap and pelleted in DMEM-10% FBS. Cells were plated at 10⁷ cells per 25-cm² flask and then incubated overnight at 37°C. The next day, cultures were washed to remove nonadherent cells; refed with DMEM supplemented with 5% FBS, 5% giant-cell-tumor-conditioned medium (OriGen, Gaithersburg, Md.), 1 mM sodium pyruvate, and 50 μg of gentamicin/ml; and cultured for 5 days. Cells were then lysed in RNAStat-60 (Tel-Test Inc., Friendswood, Tex.) buffer to extract RNA following the protocol described above. Microglial cell purity (>99%) was verified by staining with low-density lipoprotein labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate at 10 μg/ml for 12 h in the dark and examining by fluorescence microscopy.

RT-PCR and PCR.

RNA and DNA were each extracted from two separate samples of basal ganglia from each macaque. Brain tissue was homogenized, and microglia and PBMC were lysed. RNA was extracted from the homogenates and lysates by using RNAStat-60, followed by DNase treatment and purification with the RNeasy kit (Qiagen, New Orleans, La.) to remove any contaminating DNA. Five micrograms of total RNA was reverse transcribed to cDNA by using the Superscript II kit (Life Technologies, Grand Island, N.Y.), random hexamer primers, and 200 U of reverse transcriptase. DNA was extracted from brain tissue by using the Fast DNA kit following the manufacturer's protocol (Qbiogene, Carlsbad, Calif.). To amplify the V1 region of the SIV env gene, PCR was performed with 0.2 μg of cDNA or 1 μg of genomic DNA in the presence of 1 μM concentrations of each primer, 2.5 U of Taq polymerase (Perkin-Elmer, Foster City, Calif.), and 2.5 U of Pfu polymerase (Stratagene, La Jolla, Calif.). The reaction buffer consisted of 20 mM Tris-HCl (pH 8.8), 10 mM KCl, 10 mM (NH₄)₂SO₄, 2 mM MgSO₄, 1% Triton X-100, and 100 mg of bovine serum albumin/ml. The sequences of the oligonucleotide primers used in the first-round PCR were 5′-AGGAATGCGACAATTCCCCT-3′ (nucleotides 6709 to 6728) and 5′-TCCATCATCCTTGTGCATGAAG-3′ (nucleotides 7406 to 7385). The oligonucleotide primers used in the nested PCR were 5′-CAGTCACAGAACAGGCAATAGA-3′ (nucleotides 6845 to 6866) and 5′-TAAGCAAAGCATAACCTGGCGGT-3′ (nucleotides 7327 to 7305). The template was denatured at 94°C for 1 min, and PCR amplification was performed with an automated DNA thermal cycler (Applied Biosystems, Foster City, Calif.) for 40 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and primer extension at 72°C for 1 min.

Cloning and sequencing.

PCR products from five independent amplifications from brain tissue, microglia, and PBMC were extracted from an agarose gel by using the QIAQuick gel extraction kit (Qiagen), pooled, and cloned into the blunt-end pT7 blue vector by using the Perfectly Blunt cloning kit (Novagen, Inc., Madison, Wis.). Individual colonies were transferred to ampicillin-agar plates. After overnight incubation at 37°C, 400 colonies were transferred to nitrocellulose filters for hybridization with the ³²P-end-labeled oligonucleotide 5′-GAGACTAGTTCTTGTATAGCCCAGG-3′ to probe for the SIV/17E-Fr V1 sequence. Membranes were incubated overnight with the probe in hybridization buffer consisting of 50% formamide, 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 2 μg of sheared salmon sperm DNA/ml, 50 mM HEPES buffer, 1× Denhardt's reagent, and 20 μg of yeast tRNA/ml. The membrane was incubated overnight with the probe in hybridization buffer at 42°C and then washed three times for 20 min each at 42°C with 6× SSC. At these conditions, the positive colonies were always SIV/17E-Fr clones. Rare negative colonies representing nonviral clones were excluded from the statistical analysis. Eighty to 100 randomly selected negative colonies, representing SIV/DeltaB670 clones, were grown from the master plate, and plasmid DNA was purified by using a miniprep kit (Qiagen) and prepared for sequencing with Sequenase and primers T7 (5′-CTAATACGACTCACTATAGGG-3′) and U19 (5′-GTTTTCCCAGTCACGACGT-3′) in an automatic-sequencer ABI PRISM 3700 DNA analyzer (Applied Biosystems). Viral envelope sequences from the V1 region were analyzed using MacVector 7.0 sequence analysis software (Oxford Molecular Ltd., San Diego, Calif.).

Quantitation of plasma and brain viral RNA.

Viral RNA in brain tissue (basal ganglia) and plasma was quantitated using a protocol previously described (67). Briefly, RNA was extracted from 50 mg of brain tissue by using RNAStat-60 buffer followed by DNase treatment with the RNeasy kit (Qiagen) to remove any contaminating genomic DNA. Virus was pelleted from 0.5 ml of plasma, and RNA was extracted using the PureScript kit (Gentra Systems, Minneapolis, Minn.) according to the manufacturer's protocol. Two-step RT-PCR was performed in 96-well plates (MJ Research, Watertown, Mass.) with 1 μg of RNA isolated from brain and RNA extracted from 0.5 ml of plasma by using the primers and probe to the SIV gag gene on an ABI Prism 7700 sequence detection system (Applied Biosystems) as previously described (29, 58, 67).

Statistical analysis.

A Mann-Whitney test was used to determine whether there was a relationship between lesion severity and viral RNA in the basal ganglia. Paired t tests were used to determine the statistical significance of the differences in frequency of SIV/17E-Fr, SIV/DeltaB670 Cl-2, and SIV/DeltaB670 Cl-12 in the brain and PBMC and to compare the number of genotypes detected in brain RNA versus DNA and in brain versus PBMC RNA. Paired t tests were used since the samples of measurements to be compared were taken from the same macaques (i.e., the samples were not drawn independently).

RESULTS

Immune depletion and development of CNS lesions.

The immune status of inoculated macaques was tracked by counting peripheral blood CD4⁺ lymphocytes by flow cytometry before inoculation and throughout infection. CD4⁺-T-cell counts dropped dramatically in infected macaques during the first week postinoculation and then partly rebounded in some animals in the second week. During the third week after inoculation, CD4⁺-lymphocyte counts again declined rapidly in all 11 macaques; thereafter, CD4⁺-lymphocyte counts declined more gradually. By 3 months postinoculation, the CD4⁺-cell counts were 56 to 97% below preinoculation values in all infected macaques (Table 1).

TABLE 1.

Percent decline in CD4⁺ counts and viral load in plasma of SIV-infected and mock-inoculated macaques

Animal no.	Severity of brain lesions	Terminal decline in CD4+-cell counts (%)	Terminal plasma viral load (106 SIV copy eq/ml)
18292	Severe	89	69
18033	Severe	85	77
V389	Severe	81	110
18031	Moderate	83	64
18242	Moderate	64	300
V708	Moderate	56	9.8
V387	Moderate	65	13
V394	Moderate	97	41
V713	Mild	80	2.7
V715	Mild	70	6.7
17834	None	89	45
18013	Uninfected	8	0
18130	Uninfected	2	0
18180	Uninfected	10	0

Microscopic examination of brain sections from the 11 coinoculated macaques revealed CNS lesions characteristic of SIV encephalitis in 10 of the 11 macaques. Three macaques had severe, five had moderate, and two had mild SIV encephalitis (Table 1). One infected macaque had no CNS lesions. Lesions were typical of SIV encephalitis and consisted of perivascular cuffs of mononuclear cells and multinucleated giant cells, isolated multinucleated giant cells scattered throughout the brain parenchyma, multifocal glial nodules, macrophage-rich infiltrates in the meninges, multifocal myelitis, and ganglioneuritis of the trigeminal and cervical dorsal root ganglia. The lesions were seen in the basal ganglia, thalamus, midbrain, cerebellum, brain stem, and spinal cord but were most severe in the subcortical white matter at the level of the basal ganglia.

Viral RNA levels in plasma and brain are independent.

All infected macaques had high viral RNA levels in plasma at death, ranging from 2.7 × 10⁶ to 3.0 × 10⁸ copy eq/ml of plasma (Table 1). There was no statistically significant correlation between plasma viral load and the severity of CNS lesions, although macaques with mild encephalitis (V713 and V715) had the lowest plasma viral loads (2.7 × 10⁶ and 6.7 × 10⁶ copy eq/ml of plasma, respectively). The single macaque without encephalitis, however, had a relatively high plasma viral load (45 × 10⁶ copy eq/ml of plasma).

Macaques with moderate to severe CNS lesions had high levels of viral RNA in the basal ganglia, ranging from 1.7 × 10⁶ to 4.5 × 10⁷ copy eq/μg of total brain RNA (Fig. 1). The two macaques with mild SIV encephalitis had significantly lower levels of viral RNA (1.4 × 10⁵ and 1.7 × 10² copy eq/μg of brain RNA) in brain homogenates than macaques with moderate or severe encephalitis (P = 0.05). Viral RNA was not detected in the brain of macaque M17834, the only macaque that did not have SIV encephalitis.

FIG. 1.

Viral RNA levels in the basal ganglia of SIV-infected macaques at necropsy. Viral RNA quantitation was performed by real-time RT-PCR of total RNA extracted from homogenates of basal ganglia with gag-specific primers and probe. Bars represent median RNA levels for macaques with severe (n = 3), moderate (n = 5), and mild (n = 2) encephalitis.

Selective replication of SIV/DeltaB670 genotypes in vitro.

To determine the tropism of specific genotypes of SIV/DeltaB670 for primary target cells, peripheral blood lymphocytes, monocyte-derived macrophages, and microglial cells were isolated from uninfected pigtailed macaques and inoculated with the SIV/DeltaB670 viral swarm. Infected cells were harvested at 14 days postinoculation to provide sufficient time for productive virus replication, RNA was isolated from infected cells, and the V1 region of the env gene was amplified by RT-PCR, cloned, and sequenced. The SIV/DeltaB670 swarm was previously shown to consist of many genotypes identified by the nucleotide sequence analyses of the V1 region of the envelope (2, 3). The consensus amino acid sequences of the V1 region of the envelope of the SIV/DeltaB670 genotypes and SIV/17E-Fr are shown in Fig. 2. Eleven different genotypes were detected in the cultured peripheral blood-derived lymphocytes (Fig. 3). In contrast, only three genotypes (SIV/DeltaB670 Cl-2, Cl-3, and Cl-12) were detected in monocyte-derived macrophages. All three of the macrophage-tropic genotypes also replicated in microglial cells in addition to a fourth genotype, which was present at a frequency of less than 5%. This suggests that SIV/DeltaB670 Cl-2, Cl-3, and Cl-12 are macrophage-tropic.

FIG. 2.

Amino acid sequence alignment of the SIV/17E-Fr and SIV/DeltaB670 envelope V1 region. Dots indicate amino acid identity, and dashes indicate gaps inserted to maintain alignment.

FIG. 3.

Frequency of SIV/DeltaB670 genotypes that replicated in lymphocytes and macrophages cultured from PBMC and in microglia isolated from the brain of an uninfected pigtailed macaque.

Selective replication of SIV genotypes in the brain.

RNA was extracted from two different samples of basal ganglia to ensure adequate sampling. The region containing the basal ganglia was chosen because of its high viral load and the consistent severity of the lesions in this region. In previous studies comparing the frequency of viral genotypes in the basal ganglia to that in a section of thalamus 1.5 cm posterior to the basal ganglia, the same viral genotypes were identified at similar relative frequencies. The V1 region of the env gene was used as a signature sequence to identify specific viral genotypes because it distinguishes between SIV/17E-Fr and the various substrains of SIV/DeltaB670 and is the most variable region of the SIV envelope (Fig. 2) (2, 12, 44). In addition, the V1 variable region of the SIV envelope can induce high-titer neutralizing antibodies (30).

Two macrophage-tropic genotypes, SIV/17E-Fr and SIV/DeltaB670 Cl-2, accounted for 95% of the genotypes identified in RNA from the brain parenchyma of all animals with encephalitis (44 and 51%, respectively) (Fig. 4 and Table 2). There was no relationship between the severity of encephalitis and whether SIV/17E-Fr or SIV/DeltaB670 Cl-2 predominated in the brain. SIV/DeltaB670 Cl-12 was the only other viral genotype identified in RNA from the brains of macaques with SIV encephalitis, representing 5% of the genotypes identified in that tissue. No viral RNA was detected in the single macaque that did not have encephalitis. In contrast, SIV/DeltaB670 Cl-12 was the most common genotype in the PBMC, representing 41% of all genotypes replicating in PBMC (Table 2).

FIG. 4.

Comparison of the percentages of viral genotypes in RNA and DNA from brain homogenates and of viral genotypes in RNA of microglial cells and PBMC at 3 months postinoculation in 11 pigtailed macaques coinoculated with SIV/17E-Fr and SIV/DeltaB670. (a) Genotypes for RNA of brains from 11 infected macaques; (b) genotypes for RNA from microglial cells isolated from 6 macaques; (c) genotypes for DNA extracted from the brain; (d) genotypes for RNA isolated from PBMC.

TABLE 2.

Mean viral genotype frequencies in the brain, microglial cells, and PBMC

	Genotype frequency (%)a
	SIV/17E-Fr	SIV/DeltaB670 Cl-2	SIV/DeltaB670 Cl-12	Other genotypes
Brain RNA	441	51	5	0
Brain DNA	34	35	19	12
Microglial cell RNA	50	46	4	0
PBMC RNA	22	25	41	12

^aNumbers represent the mean percentage of each genotype detected in all macaques with encephalitis by screening over 800 clones amplified from RNA or DNA extracted from basal ganglia or RNA extracted from isolated microglial cells and PBMC.

SIV/17E-Fr was detected more often in the brain than in PBMC (P = 0.057) (Table 2). There also was a trend towards a higher frequency of SIV/DeltaB670 Cl-2 in the brain than in PBMC (P = 0.09). SIV/17E-Fr and SIV/DeltaB670 Cl-2 were each the sole genotype replicating in the brain of a single macaque with encephalitis (macaques 18242 and 18031, respectively), indicating that both of these viruses are neurovirulent (Fig. 4). In contrast, SIV/DeltaB670 Cl-12 was detected significantly less frequently in the brain than in PBMC (P = 0.0001). This suggests that SIV/DeltaB670 Cl-12 is better adapted for growth in PBMC than in the brain environment.

The brain is clearly a selective site for SIV replication since significantly fewer viral genotypes were detected in brain RNA (mean = 2) than in PBMC RNA (mean = 3.9) (P = 0.004). In all, there were eight viral genotypes present in the PBMC, including SIV/17E-Fr at a frequency of 22%, SIV/DeltaB670 Cl-2 at 25%, SIV/DeltaB670 Cl-12 at 41%, SIV/DeltaB670 Cl-13 at 5%, SIV/DeltaB670 Cl-3 at 4%, SIV/DeltaB670 Cl-17 at 1%, SIV/DeltaB670 Cl-6 at 0.5%, and SIV/DeltaB670 Cl-14 at 0.1%. Taken together, these data suggest that there was selective replication of neurovirulent genotypes in the brains of macaques with SIV encephalitis. Further, virus replication in the CNS progressed independent of the viral genotypes replicating in the PBMC or the level of peripheral virus replication.

Selective replication of SIV genotypes in microglia.

To identify the genotypes that were replicating in microglia, the V1 region of the env gene was amplified from RNA extracted from microglial cells isolated from 6 of 10 macaques with encephalitis. The frequency of detection of SIV/17E-Fr, SIV/DeltaB670 Cl-2, and SIV/DeltaB670 Cl-12 in microglial cells paralleled that in the brain parenchyma, with frequencies of 50, 46, and 4%, respectively (Table 2). SIV/17E-Fr and SIV/DeltaB670 Cl-2 were the only genotypes detected in the microglia from five of six macaques; SIV/DeltaB670 Cl-12 was detected in the sixth macaque at a frequency of 25% (Fig. 4). Although SIV/DeltaB670 Cl-12 replicates in microglia, it appears to be less adapted for replication in these cells than SIV/17E-Fr or SIV/DeltaB670 Cl-2. Together, these data demonstrate that there is selective replication of macrophage-tropic, neurovirulent genotypes in microglia and suggest that microglia may play an important role in selective viral replication in the brain. Nonetheless, susceptibility of microglial cells to infection with macrophage-tropic virus is apparently conditioned by the in vivo environment. Thus, SIV/DeltaB670 Cl-3 and Cl-13, while capable of replicating in microglia in vitro, were not detected in the microglia of macaques with encephalitis at necropsy.

Comparison of RNA and DNA genotypes in the brain.

To distinguish between actively replicating and latent genotypes, both RNA and DNA were extracted from two different samples of basal ganglia of the brain and the V1 region of the env gene was amplified and sequenced. All viral genotypes detected in brain RNA were also identified in DNA, confirming that these viruses were established in the brain (Fig. 4). Although only three viral genotypes (SIV/17E-Fr, SIV/DeltaB670 Cl-2, and SIV/DeltaB670 Cl-12) were detected in brain RNA, a significantly greater number of viral genotypes were detected in DNA (P = 0.004). These genotypes in the brain DNA likely represented latent viruses or viruses replicating at a level below the level of detection for real-time RT-PCR. Most of the viral genotypes that were not present as replicating virus were lymphocyte-tropic (i.e., did not replicate in macrophages or microglial cells) (Fig. 3). This suggests that the brain is exposed to both lymphocyte- and macrophage-tropic viruses, probably via trafficking cells, but that only the macrophage-tropic viruses replicate terminally. The lymphocyte-tropic viruses do not replicate but leave their footprints as viral DNA. Interestingly, despite the presence of SIV/17E-Fr and SIV/DeltaB670 Cl-2 DNA in the brain of macaque M17834, the single macaque without encephalitis, these viruses did not actively replicate.

DISCUSSION

This is the first study to directly compare the viral genotypes replicating in the CNS with those that are present in DNA in the brain. Furthermore, this is also the first direct comparison between viral genotypes that replicate in the brain, microglial cells, and the periphery. Importantly, this study demonstrates the selective replication of distinct neurovirulent viral genotypes in the brain and, specifically, in the microglia of macaques with SIV encephalitis. At 3 months postinoculation, two macrophage-tropic, neurovirulent viruses, SIV/17E-Fr and SIV/DeltaB670 Cl-2, comprised 95% of the genotypes identified in RNA from the brains of macaques with encephalitis. These same two viruses were present at the same relative frequencies in microglial cells isolated from the brain. In contrast, the predominant virus in the PBMC of infected macaques was SIV/DeltaB670 Cl-12, which represented 41% of the genotypes identified in PBMC, but only 5% of the isolates from brain. This study further demonstrated that, while many lymphocyte-tropic viruses are present in the brain as viral DNA, these viruses are not actively replicating in the brains of macaques at the time of death, supporting the notion that lymphocyte-tropic viruses that may enter the brain in trafficking cells do not contribute directly to CNS disease.

A number of studies have demonstrated more limited genetic variation of viruses in the brain than in peripheral blood and lymphoid tissues (21, 32, 45, 62). The majority of those studies examined viral genotypes present in DNA, not RNA. The data presented here differ in that we found no apparent selection of viral genotypes in the brain DNA. There was, however, selective replication of viruses in the brain compared to that in PBMC. Many viral genotypes were present in the brain DNA, representing both lymphocyte- and macrophage-tropic viruses, yet only macrophage-tropic, neurovirulent viruses replicated in the brains of macaques with encephalitis. These data suggest that many viruses enter the brain but that the selection for replication of specific virus strains in macaques that develop encephalitis occurs at some stage after viral entry to the brain. They further suggest that the examination of viral genotypes in brain DNA does not accurately reflect the population of actively replicating genotypes.

One reason for selective virus replication in the brain may be that certain virus strains induce stronger or more effective antiviral immune responses in the brain, resulting in their elimination (5, 11, 43). Alternatively, virus mutations that impair recognition of certain virus strains by the immune system may develop. There is evidence that virus selection in the brain occurs under immune pressures different from those in the periphery. When viral sequences in the brain corresponding to four known epitopes that can serve as targets for major histocompatibility complex class I-restricted cytotoxic lymphocyte activity were examined in HIV-infected patients, there was a clustering of mutations in the brain that differed from those identified in the spleen and lymph nodes (62). The V1 variable region of the SIV envelope can induce the production of high-titer neutralizing antibodies, and it is possible that the high variability of the V1 region is one mechanism by which the virus may escape the immune system (14, 30, 46, 52).

Another explanation for our frequent demonstration of SIV/17E-Fr and SIV/DeltaB670 Cl-2 in the brain may be the propensity of these two viruses to replicate in microglial cells. Neurovirulent quasispecies have also been obtained by several intermacaque passages of SIV isolated from microglia (33). It should be noted, however, that there are several populations of cells of monocyte/macrophage origin in the brain, including parenchymal microglia and perivascular macrophages, of which are likely resident populations of cells that are susceptible to HIV and/or SIV infection (61). As the technique for microglial culture in this study may not solely select for parenchymal microglia, it is possible that perivascular macrophages may also be present. Because activated parenchymal microglial cells can share immunophenotypic markers with perivascular macrophages, it is impossible to differentiate between these two populations of isolated cells at this time.

It is also possible that one or both of these virus strains may replicate in other susceptible neural cell types such as astrocytes (28, 53, 59). Replication of virus strains in different target cells may result in the expression of different profiles of proinflammatory molecules, such as chemokines and cytokines, that modulate the patterns and severity of disease.

Regardless of the reason, selection for the replication of specific virus strains appears to be regulated differently in the CNS than it is in the periphery. This suggests that the brain can act as a reservoir for virus strains that are not replicating in the periphery. Cells in the perivascular (Virchow-Robin) spaces of the brain drain into the cerebrospinal fluid of the leptomeninges and then travel via the perineural spaces to the lymphatics of the head and their draining lymph nodes (13, 31, 66). This provides a potential route for the escape of virus-infected cells from the brain to the periphery, introducing viruses that replicate in the CNS into the periphery.

Viral RNA was not detected in the brain of macaque 17834, the only macaque that did not have SIV encephalitis, despite high levels of viral RNA in plasma (10⁶ to 10⁸ copy eq/ml during 3 months of infection). This demonstrated that perfusion was extremely efficient in removing contaminating blood virus from the brain and suggested that virus-infected cells were not trafficking from the peripheral blood into the brain of this macaque in significant numbers at the time of death. Alternately, if virus-infected cells were entering the brain terminally, they were quickly eliminated by the animal's immune system or were below our level of detection. Interestingly, DNA from SIV/17E-Fr and SIV/DeltaB670 Cl-2 and Cl-3 was detected in the brain of macaque 17834. It will be important to examine in detail the innate and adaptive immune responses of other macaques like 17834, in which the brain is exposed to neurovirulent virus strains but does not develop SIV encephalitis.

In a previous study using the same infection protocol, SIV/17E-Fr, SIV/DeltaB670 Cl-12, and a mixture of other SIV/DeltaB670 clones were detected by PCR amplification of DNA from the brains of macaques with encephalitis (65). The present study differed in that viral genotypes from RNA were examined to differentiate between latent viral DNA and the actively replicating virus strains that are more likely to be responsible for the encephalitis. The predominance of SIV/17E-Fr and SIV/DeltaB670 Cl-2 in the brains of macaques with encephalitis is consistent with the known neurovirulent phenotype of SIV/17E-Fr (37) and suggests that SIV/DeltaB670 Cl-2 is also a neurovirulent virus strain. This was supported by the finding of SIV/17E-Fr and SIV/DeltaB670 Cl-2 as the only strain present in the brain of each of the two animals with extensive SIV encephalitis (macaques 18242 and 18031, respectively).

Although we identified strains replicating in the brains of macaques with encephalitis by sequencing the hypervariable region V1, the encephalitogenic potential of SIV/17E-Br and SIV/DeltaB670 Cl-2 is likely also associated with other regions of the virus. Previous studies have identified the nef gene and the transmembrane portion of the envelope gene as two regions that determine neurovirulence (22, 37). Further studies are needed to define the exact role of these and other viral genes in the development of encephalitis and to dissect the likely complex mechanisms by which specific viral genes promote or control CNS inflammation.