Abstract
One of the consequences of HIV infection is damage to the CNS. To characterize the virologic, immunologic, and functional factors involved in HIV-induced CNS disease, we analyzed the viral loads and T cell infiltrates in the brains of SIV-infected rhesus monkeys whose CNS function (sensory evoked potential) was impaired. Following infection, CNS evoked potentials were abnormal, indicating early CNS disease. Upon autopsy at 11 wk post-SIV inoculation, the brains of infected animals contained over 5-fold more CD8+ T cells than did uninfected controls. In both infected and uninfected groups, these CD8+ T cells presented distinct levels of activation markers (CD11a and CD95) at different sites: brain > CSF > spleen = blood > lymph nodes. The CD8+ cells obtained from the brains of infected monkeys expressed mRNA for cytolytic and proinflammatory molecules, such as granzymes A and B, perforin, and IFN-γ. Therefore, the neurological dysfunctions correlated with increased numbers of CD8+ T cells of an activated phenotype in the brain, suggesting that virus-host interactions contributed to the related CNS functional defects.
The best experimental model to mimic human AIDS is SIV infection in macaques. SIV infection shares several qualities with HIV-1 infection, such as cell type tropism, immunosuppression, and disease of the CNS (1, 2). Both SIV and HIV gain access to the CNS early in the infection (3, 4, 5). Varying degrees of CNS pathology may result, the most severe form, SIV or HIV encephalitis, being characterized by lesions bearing multinucleated giant cells, microglial nodules, infiltrating macrophages, and gliosis (2, 6, 7).
The infected cells in the CNS are of the monocytic lineage (6, 8, 9). Virus can be detected in the brain within 2 wk after infection, when acute viremia peaks. In the early stages of experimental SIV infection, viral expression in the brain is localized mainly in perivascular mononuclear cells and within the leptomeninges (4, 10). Later, viral RNA can be detected in the leptomeninges and in brain infiltrating/perivascular macrophages and multinucleated giant cells (3, 9, 10, 11). Of several mechanisms proposed to account for this CNS infection, the “Trojan horse” model, in which infected monocytes and/or T cells migrate into the brain, has received much support (12, 13).
Approximately one-third of HIV-1-infected individuals develop the associated cognitive/motor disorder, also known as the AIDS dementia or neuroAIDS (13, 14, 15). Furthermore, HIV alters sleep patterns, psychomotor activity, and sensory evoked potentials (EPs)3 (16, 17, 18). In humans, progressive motor and cognitive abnormalities are thought to be initiated by the entry of HIV into the CNS (19). Similarly, in SIV-infected rhesus monkeys, functional CNS deficiencies in sensory EPs, behavioral/cognitive ability, and motor skills have been found by us and others (20, 21, 22, 23). In the rhesus/SIV model, the reduction of virus load in the periphery by antiviral treatment restored neurophysiological normalcy, perhaps, in part, because this treatment lowered the number of infected macrophages and T cells entering the CNS (20). However, abnormalities in movement and temperature were unchanged by antiviral treatment. Because cells of the macrophage/microglia lineage (but not neurons) are infected in the CNS, the neurological consequences of the CNS infection are related to indirect mechanisms, such as viral products or molecules produced by the host in response to the virus. However, the triggering events remain unknown.
Tissue damage is frequently associated with host cell-mediated (T cell) responses (24). Virus-specific CD8+ CTLs play a central role in the immune response against viruses, in general, by eliminating virus-infected cells. Macaques additionally have CD8+ cells that lack the CD3 complex and exhibit NK cell activity (25). The importance of CD8+ cells in controlling SIV infection in monkeys was shown when the viral load increased and progression of disease was accelerated in animals whose CD8+ cells were depleted by Ab treatment (26).
Most of the specific effector CTL generated during viral infections undergo apoptosis, but a fraction of them remains in the circulation, constituting a pool of memory cells. These memory cells promptly and efficiently respond to a second specific stimulation. The expression of CD45RA can be used in rhesus macaques to define whether previously primed cells are mediating a secondary response. Resting naive CD8+ cells, not primed by Ag, express high levels of CD45RA, but memory cells express far less (27). However, this low expression of CD45RA does not discriminate memory cells in the resting state from those that are activated. Instead, the expression of receptors for homing to endothelium of inflamed sites, such as CD11a (LFA-1), give some clue about the activation status of these cells (28). The expression of CD95 (Fas) also increases on activated cells, and its binding to its ligand (FasL) may induce apoptosis or proliferation of CD8+ cells (29). CD28 can also be used to define the pattern of activation. The T cell Ag CD28 is an important provider of costimulatory signals to TCR-mediated activation, thereby regulating proliferation, preventing the induction of anergy or apoptosis, and driving the pattern of cytokine production (30). The expansion of CD8+CD28− cells in HIV infection suggests that such cells arise in response to the virus (31). In fact, the CD8+CD28− cells can present characteristics of end-stage effector CTL, such as expression of perforin, granzymes A and B, and IFN-γ, and they exhibit cytolytic activity (32).
We previously identified SIV-specific CD8+ CTL in the cerebrospinal fluid (CSF) of SIV-infected macaques very early after the infection and in the brain at autopsy (33). Yet the contribution of the CD8+ cells and other T cell subpopulations in the progression of CNS functional deficits remains unknown.
Therefore, the present work was designed to evaluate the brain infiltrate and the phenotype of T cells in the CNS compared with those in the periphery, at a relatively early stage of SIV-induced disease, for a possible correlation with neurological dysfunction. The results showed altered EPs and a drastic increase in the number of activated CD8+ T cells expressing predominantly a memory and a cytolytic phenotype in the brains of animals with early SIV infection. These data suggest that SIV infection induces a chronic immune response in the CNS that mediates injury to the brain.
Materials and Methods
Animals
Rhesus monkeys, free from SIV, type D simian retrovirus, and herpes B virus were obtained from Covance (Alice, TX) and Charles River Breed Laboratories (Key Lois, FL). All animal experiments were performed with approval from the Institutional Animal Care and Use Committee. Blood samples were drawn from the femoral vein, and plasma was separated by centrifugation from cells in EDTA-anticoagulated blood. CSF was obtained from the cisterna magna. CSF samples containing >0.1% blood contamination (determined by RBC counts) were discarded. Necropsy was performed after terminal anesthesia. During necropsy, animals were perfused intracardially with sterile PBS containing 1 U/ml heparin before tissue samples were taken for cell isolation, virus quantification, and formalin fixation for histology. Animals 350, 352, 353, and 354 were sacrificed at 73, 75, 77, and 80 days postinoculation (p.i.), respectively, and comprised the 11-wk p.i. group. Animals 335, 346, and 347 were sacrificed at 13 days p.i. and comprised the 2-wk p.i. group.
Viral infection
A cell-free stock of SIV (SIVmac182) obtained after serial passage of SIV-infected microglia was the inoculum used (34). Animals received 0.25 ml of SIVmac182 stock, diluted into RPMI 1640 for injection, containing 5 ng/ml p27 (gag) Ag.
Viral quantitation
To measure virus load, SIV RNA in plasma, CSF, and brain tissue was determined by using the quantitative branched DNA (bDNA) signal amplification assay. Bayer Reference Testing Laboratory (Emeryville, CA) performed this assay. Each sample was measured in duplicate and is reported as an average value. Virus was also recovered from the CSF in a coculture assay. Cells were separated from the CSF by centrifugation, and the cells and cell-free supernatant were separately added to cultures of CEMx174 cells; all cultures were maintained for at least 3 wk for examination for syncytia formation.
Electrophysiology
Eletrophysiological analysis of brainstem auditory EP (BSAEP) and auditory EP (AEP) was performed on ketamine (20 mg/kg)-anesthetized animals as described previously (35). Positive waves in BSAEP were denoted as P1, P2a, P2b, P3, P4, and P5. In AEP, two positive (P1 and P2) and two negative (N1 and N2) waves were identified. Averaged peak latencies were calculated and compared individually between animals and as group means using ANOVA.
Peripheral mononuclear cells
Buffy coats obtained from centrifugation of EDTA-anticoagulated blood, and cell suspensions from spleen, deep cervical, and inguinal lymph nodes, passed through a nylon mesh, were submitted to a Ficoll-Isopaque (Pharmacia Biotech, Uppsala, Sweden) gradient centrifugation for isolation of the mononuclear fraction. After washing, cells were enumerated in a Coulter Z2 (Coulter, Miami, FL) and resuspended in complete RPMI 1640, containing 10% FCS at a concentration of 107 cells/ml, for labeling or freezing.
Brain cells
The meninges were carefully removed, and the brain was minced and transferred to a sterile cell sieve with a 40-mesh wire screen. A sterile plunger was used to force this material through the sieve, and the disassociated material was collected by centrifugation at 470 × g for 15 min at 4°C. The supernatants were aspirated, and each pellet brought to a final volume of 30 ml with HBSS, after which 28 U/ml of DNase I and 500 U/ml of collagenase II were added. Digestion was performed at 37°C in a shaking water bath for 1 h. Afterward, the cells were pelleted at 470 × g for 10 min and then washed in HBSS containing 1% FCS. The cell pellet was resuspended in 6.4 ml of a 1.122 g/ml Percoll (Pharmacia Biotech) solution and brought to a final volume of 30 ml with HBSS, resulting in 1.033 g/ml Percoll. Eight milliliters of 1.088 g/ml Percoll was underlayed, and this gradient was centrifuged at 1200 × g for 20 min at 20°C. The cells at the 1.033/1.088 interface were collected, washed, and quantified in a Coulter Z2.
Flow cytometry
Cells isolated as described (2 × 105 to 10 × 105) were stained with 50 μl mixtures of Abs diluted according to a previous titration in staining buffer (HBSS with 2% FCS and 0.01% NaN3). The Abs used for the staining were anti-monkey CD3-biotin (clone FN-18; Biosource International, Camarillo, CA) followed by streptavidin-PerCP or streptavidin-APC (BD PharMingen, San Diego, CA), anti-human CD8-PE (clone DK25; DAKO, Carpinteria, CA), anti-CD11a-FITC (clone 25.3.1; Immunotech, Westbrook, ME), anti-CD28-FITC (clone CD28.2; Immunotech), anti-CD95-FITC (cloneDX2; Caltag, Burlingame, CA), anti-Mac-1-PE (clone M1/70(9) Roche, Indianapolis, IN), and anti-CD4-PE (clone OKT4; hybridoma obtained from the American Type Culture Collection (Manassas, VA), with the secreted Ab purified and PE-conjugated in the laboratory). Isotype controls (BD PharMingen) were also used. The cells were then processed through a FACScan flow cytometer before analysis of data with CellQuest software (BD Immunocytometry Systems, San Jose, CA).
Magnetic columns
Cells isolated from the blood or lymphoid tissues were resuspended to the concentration of 2 × 107 in 200 μl of PBS, and 50 μl of anti-CD8 MACS Microbeads (Miltenyi Biotec, Auburn, CA) were added. The cells were incubated at 4°C for 20 min, washed once, and passed twice through magnetic columns according to the manufacturer’s instructions for the obtention of a CD8-enriched population. The cells were pelleted and frozen for RNA extraction.
RNA extraction and RT-PCR
The RNA was extracted from the cell pellets on a Bead-Beater (Biospec Products, Bartlesville, OK) with an Ambion Totally RNA kit (Ambion, Austin, TX), as per the manufacturer’s instructions. A volume equivalent to 1 μg was subjected to reverse transcription using 125 ng of random primers, 5× first-strand buffer, 0.1 M of DTT, 10 mM of dNTPs, and 200 U of Superscript II (Life Technologies, Grand Island, NY), in a final volume of 20 μl. The samples were then incubated at 42°C for 50 min followed by inactivation at 90°C for 5 min. To design primers for rhesus CD8 β-chain, granzymes A and B, and perforin, fragments of cDNA derived from activated rhesus monkey lymphocytes were PCR-amplified using primers derived from human sequences available in GenBank. These products were molecularly cloned, and three to seven clones of each product were sequenced using an Applied Biosystems automated sequencer at the TSRI Protein and Nucleic Acids Core Facility. Primers were then designed from the monkey cDNA sequences obtained, as well as the rhesus IFN-γ sequence available in GenBank, with the aid of the MacVector program (Accelrys, San Diego, CA). Next, amounts of cDNA to be used in the PCR were normalized after determining the relative amounts of 18 S cDNA in the sample by real-time quantitative PCR (primers and probe from Perkin-Elmer, Foster City, CA) using a Smart Cycler (Cepheid, Sunnyvale, CA). Finally, the normalized cDNA was amplified with the following primers: CD8β, 5′-CATGG TGAAG AGGTG GCACA GGAGA and 3′-TGGAC TTCTT GGTGG GCTGG GCA (72°C, 205-bp product); perforin, 5′-ACTTT GCAGC CCAGA AGACC CAC and 3′-CCGTA GTTGG AGATG AGCCT GAGGT (71°C, 193-bp product); granzyme A, 5′-GGGTG GGGGA AGATT CACAA TAGTG C and 3′-CCCTC GGAAA ACACC GTTAC ACAAC A (60°C, 219-bp product); granzyme B, 5′-GGAAG ATCAA ACGTG CAAAT CCCG and 3′-GGAGG CTTGC CATTT CTTTG TCCAT (60°C, 177-bp product); and IFN-γ, 5′-ACGGG ATGAC TTTGA AAAGC TGACC and 3′-CACTG GGATG CTCTT CGACC TGC (72°C, 165-bp product).
Histopathology and immunohistochemistry
Tissues were fixed in 10% formalin, embedded in paraffin, and cut into 5-μm sections. After H&E staining, sections from each tissue block were examined microscopically. Representative sections of brain (plus lymph node and spleen controls) were immunohistochemically stained for leukocyte common Ag (LCA) as described (20).
Statistical analysis
After establishing the normality and homogeneity of variances, the samples were submitted to one- or two-way ANOVA, depending on the number of criteria to be tested, followed by a multiple comparisons t test (Tukey) when appropriate. Only the samples that exhibited a confidence interval of 95% or more were accepted as significant. The tests were performed by SigmaStat version 2.0 for Windows (SPSS Science, Chicago, IL).
Results
Rhesus monkeys infected with SIVmac182 were monitored for CNS function (EP activity), blood cellularity, and viral load. When evaluated by bDNA, viremia peaked in the plasma at the 14th day p.i., then stabilized at a lower level (Fig. 1⇓A). In the CSF, the temporal pattern of viral load was similar to that in plasma for two animals (monkeys 350 and 353), another animal (354) never had virus detectable by bDNA tests of the CSF at any time (Fig. 1⇓B). Early time point CSF samples from monkey 352 were too small for quantitation by bDNA analysis, but CSF from all four animals was positive in coculture assays on day 14 p.i., suggesting that virus was present in the CSF simultaneously with the peak load in the plasma of all four animals.
Virus load in the plasma (A) and CSF (B) during the course of SIV infection in rhesus monkeys followed for 11 wk following viral inoculation. Samples of blood and CSF periodically collected from SIV-infected monkeys were analyzed for the presence and quantification of virus copies by the bDNA assay, given as the log10 transformation of the number of RNA equivalents per milliliter. Values shown at the axis break are below the limits of detection (<2.7).
To detect neuronal dysfunction, the monkeys were also analyzed for alterations by the measurement of eletrophysiologic EPs. For this, auditory sensory evoked responses were obtained for analysis of latencies of the waves. As early as the first assessment (27 days p.i.) all four animals presented delays in at least one of the sensory evoked potential measures, and both the brainstem (BSAEP) and cortical (AEP) measures were abnormal in all animals at sacrifice (11 wk p.i.). Fig. 2⇓A shows that the latency of the BSAEP P5 increased during the infection in comparison to each animal’s individual preinoculation baseline, and Fig. 2⇓B similarly shows the increase in the latency of the AEP N2 auditory time following viral inoculation. Representative waveforms before and after infection for the BSAEP are shown in Fig. 2⇓C. These data are consistent with previous observations of early abnormal sensory EPs after infection.
SIV-induced EP abnormalities. A, Latency for BSAEP P5 in each SIV-inoculated monkey. Preinoculation baseline is the average of four measurements taken before infection. B, Latency for the AEP N2 wave, baseline as in A. C, Representative pattern of BSAEP waveforms from monkey 352 before and after SIV infection (day 0 and day 69). The position of the P5 latency is indicated by arrows.
During the course of SIV infection, circulating CD4+ and CD8+ T cells were monitored by lymphocyte counts and FACS analysis, and the CD4:CD8 ratio was calculated. The CD4:CD8 ratio in the blood declined acutely, reaching a nadir on day 14 p.i. (0.69), followed by a gradual rise, but never reaching the preinfection level. The main cause was an increase in the absolute number of CD8+ T cells, peaking at 2 wk p.i., whereas the number of CD4+ T cells remained stable throughout the infection. Peripheral blood CD8+ lymphocytes were also analyzed for CD11a expression, which increased to an average of 65% by the 69th day postinfection compared with the preinfection average of 35%. Therefore, SIV infection apparently favored the relative enlargement of an activated CD8+ population (Fig. 3⇓). Because the number of CD4+ T cells with CD11a phenotype did not change (Fig. 3⇓), no generalized activation of T cells occurred.
Kinetics of CD11a expression on the surface of peripheral CD4+ and CD8+ T lymphocytes after SIV infection in rhesus macaques. Blood samples were obtained periodically from SIV-infected animals and analyzed for the expression of CD11a by flow cytometry. Cells were gated on CD4 or CD8+CD3+ cells. Values represent the average ± SD of the four infected animals in each time point.
To investigate whether T cell infiltration into the brain correlated with the observed functional abnormalities, the animals were sacrificed at 11 wk p.i., when the EP assessments were abnormal and the viral load measurements indicated that infection had stabilized. Blood cells were cleared from the CNS by perfusion, and brains were collected and processed for histology as well as measurement of viral load and isolation of lymphocytes from the CNS.
According to histopathological analysis, SIV encephalitis was not present, and the CNS had no focal lesions, although infiltrating lymphoid cells were noted and subsequently verified by reactivity with LCA. The LCA-positive cells were easily detected on brain sections from infected animals but were rare in uninfected controls. Positively stained mononuclear cells formed foci in the perivascular area or were scattered diffusely within the parenchyma (Fig. 4⇓). No differences were observed in the intensity or distribution of infiltrating cells among individual infected animals.
Histopathology of the CNS of an 11-wk-SIV-infected monkey immunohistochemically stained for LCA. The presence of leukocytes can be observed in the brain sections (from animal 350) either in the perivascular area (A), forming foci (B), or as scattered infiltrates within the parenchyma (C and D). A–C are from the brainstem, observed at ×20 magnification. D is from the cortex, observed at ×40 magnification.
bDNA analysis of RNA isolated from five separate regions of the CNS revealed that three of the four SIV-infected animals had virus in at least one segment of the brain (none was detected in samples from monkey 350), but one animal (353) had virus in all five regions (Table I⇓). Calculation of the number of CD3+CD8+ cells isolated from the CNS revealed that an increased number of CD8+ T cells was found in the brains relative to controls, but the intensity of the CD8+ infiltrate was not necessarily proportional to the amount of virus detected in the brain tissue (Table I⇓). Thus, no link was evident between the amount of virus and the number of CD8+ T cells at this stage of SIV infection.
CD8+ T cells and virus load in the CNS of uninfected controls and SIV-infected monkeys
Because SIV has been shown to enter the brain within the first 2 wk following viral inoculation, a separate group of three (335, 346, 347) animals was sacrificed at 2 wk p.i., corresponding to the peak of viremia. High levels of plasma and CSF virus (7.79 ± 0.68 and 5.72 ± 0.42 log viral load/ml, respectively), as well as activated blood CD3+CD8+ cells (58.2 ± 7.4% CD11ahigh, compared with 28.2 ± 19.5% preinoculation) were present at this stage. Relative to the animals sacrificed at 11 wk p.i., at least 10-fold more virus was present in the brain, and virus could be found in all areas of the brain tested, yet fewer CD3+CD8+ cells were present in the brain than in the 11 wk p.i. animals (Table I⇑).
The phenotypes of lymphocytes infiltrating the CNS was determined by flow cytometry and identification of surface markers that define the subpopulations and their activation status. Fig. 5⇓A depicts the relative percentage of T cell types in the brains of individual monkeys. The most dramatic expansion in both absolute and relative counts was found in the CD3+CD8+ cell population in the 11 wk p.i. animals; such infected animals had a significant 5.4-fold increase (F = 38.797; p < 0.001) relative to the number in controls. Amounts of other lymphocyte subpopulations also increased somewhat, but not to a level of statistical significance. That is, CD3+CD4+ T cells increased 2.3-fold (F = 1.811; p = 0.218) and CD3−CD8+ cells 2.5-fold (F = 1.108; p = 0.371) in the brains of 11-wk-infected animals compared with control brains. Although the number of CD3+CD8+ T cells in the brain of the 2 wk p.i. group was approximately twice that found in uninfected control animals, such a change was not statistically significant (F = 38.797; p = 0.403).
Distribution of T cell subpopulations in the brain parenchyma of SIV-infected and uninfected control monkeys. A, Percentage of indicated lymphocyte subpopulations in the brain. B, Absolute numbers of CD8+ T cells in the brain and the expression of CD45RA. Single-cell suspensions originating from the brain were analyzed by flow cytometry, gating on lymphocytes.
A particularly notable change was the increased number of CD3+CD8+ cells expressing a low level of CD45RA, which is considered a memory phenotype. In the brain, 11 wk of infection induced an average 6.7-fold increase of CD3+CD8+CD45RAlow (memory phenotype) cells and a 4.0-fold increase in the absolute numbers of CD3+CD8+ cells expressing CD45RAhigh (naive phenotype) (Fig. 5⇑B). The increase in both memory and naive CD8+ cells in 11 wk was statistically significant in comparison to controls (F = 16.669 and F = 10.334, respectively; p < 0.001). In contrast, the smaller increase in CD3+CD8+ cells found at 2 wk p.i. was evenly divided between memory and naive cells. In the CNS of SIV-infected monkeys, a preferential enrichment of the CD8+ T cells observed at 11 wk presumably favored those expressing a memory phenotype.
Furthermore, the foregoing increase of CD8+ T cells resulting from SIV infection was also observed in peripheral sites. Fig. 6⇓ presents the CD4:CD8 ratios in the blood, spleen, and lymph nodes, all of which exhibited lower values following 2 or 11 wk of infection relative to controls. Comparison of CD4:CD8 T cell ratios in the brain vs peripheral regions showed that infected as well as uninfected monkeys had a predominance of CD8+ T cells in the brain (Fig. 6⇓).
CD4:CD8 ratio in several organs of controls and SIV-infected monkeys. Single-cell suspensions from the indicated sites were analyzed by flow cytometry, cells were gated on the CD3+ T cell subpopulation. The ratio between the subpopulations was calculated. Values represent the average ± SD of the five uninfected controls, three animals at the 2 wk and four at 11 wk after SIV infection.
The naive/memory phenotype of the CD8+ T cells was also analyzed in uninfected as well as in 11 wk p.i. animals at sites other than the CNS. At all sites except the brain, the proportion of cells expressing memory phenotype (low levels of CD45RA) was higher in infected monkeys than in controls (Fig. 7⇓), suggesting that the infection led to a systemic enrichment of memory cells. The comparison of CD8+ T cells from different sites within the same group of animals revealed that in control monkeys, the percent of memory CD8+ T cells (expressing CD45RAlow) in the brain was significantly higher (F = 10.334; p < 0.01) than in the periphery (Fig. 7⇓). However, in the SIV-infected monkeys, the percent of memory CD8+ T cells was comparable at all sites analyzed. In the CNS, 11 wk-infected and control animals had the same level of predominance of memory CD8+ T cells, but their absolute number differed greatly.
Percentage of CD8+ T cells expressing memory phenotype (CD45RAlow) in several locations from uninfected controls and 11 wk SIV-infected monkeys. Single-cell suspensions originating from the indicated sites were analyzed by flow cytometry, cells were gated on the CD8+CD3+ T cell subpopulation. Values represent the average ± SD of the five uninfected controls and four 11-wk-SIV-infected animals.
To better characterize these CD8+ T cells, we compared the phenotype of activation markers expressed on the CD3+CD8+ cells obtained from different sites in uninfected control animals to those at 11 wk p.i. Generally, the percent of cells expressing CD11a, which defines activation status, tended to be higher in infected animals than in controls (Fig. 8⇓A), especially in the spleen and lymph nodes, where the differences were statistically significant (F = 8.733; p < 0.001). Additionally, both infected and control animals had a significantly higher (F = 26.521; p < 0.001) percentage of CD8+ T cells expressing CD11ahigh in the CNS than in the periphery. Close to 100% of the cells in the CNS expressed high levels of CD11a, but in peripheral sites of control as well as infected monkeys, the proportion of cells was always much below that level. The level of CD95 expression did not differ statistically between controls and infected monkeys for each individual site analyzed, but CD8+ T cells from the CNS tended to have a higher level of CD95 expression than those in the periphery (Fig. 8⇓B). The expression of CD28 was also evaluated on CD8+ cells from the CNS and other sites in SIV-infected monkeys and controls (Fig. 8⇓C) but the differences were not significant. CD8+ T cells in the CNS seem to differ phenotypically, then, from those in the periphery, and the SIV-induced changes in the CNS do not involve differences on the expression of the surface markers that we analyzed. Similar to the animals infected for 11 wk, the CD8+ T cells in the brain of 2-wk-infected animals did not differ from uninfected controls for the level of expression of the activation markers analyzed (Fig. 8⇓D).
Phenotypic characterization of the CD8+ T cells in different sites of SIV-infected and control monkeys. A, Percent of CD8+ T cells expressing high levels of CD11a. B, Percent of CD8+ cells expressing high levels of CD95. C, Percent of CD8+ T cells negative for expression of CD28. D, Percent of CD8+ T cells in the brain expressing high levels of CD11ahigh levels of CD95, negative for expression of CD28, and low levels of CD45RA. Single-cell suspensions originating from the indicated sites were analyzed by flow cytometry, cells were gated on the CD8+CD3+ T cell subpopulation. Values represent the average ± SD of the five uninfected controls, three animals at 2 wk (D only), and four animals at 11 wk after SIV infection.
Fig. 9⇓ represents examples of fluorescence intensity histograms of the surface markers we investigated. The differences between controls (Fig. 9⇓A) and SIV-infected monkeys (Fig. 9⇓B) were strongest for CD11a and CD95 expression. Clearly, the uninfected controls expressed lower levels of these markers and fewer cells with an abundance of these markers than SIV-infected animals. This was true at all sites analyzed, except the brain. Although control animals varied considerably in CD28 expression, CD8+ T cells from infected animals had less CD28 than CD8+ cells from three of five control animals, regardless of site.
Representative FACS histograms exemplifying the expression of activation markers. The values shown are for one representative animal from each group (control uninfected and SIV-infected). Single-cell suspensions originating from the indicated sites were analyzed by flow cytometry, cells were gated on the CD8+CD3+ T cell subpopulation. The numbers between keys (defining M2) represent the percentage of cells situated in that range of expression. The numbers above each histogram represent the geometric mean of the fluorescence intensity for the whole subpopulation, indicating the average level of expression.
Because CD8+ T cells from the brain presented a more highly activated phenotype than those in the periphery, we determined whether these brain cells produced cytolytic and potentially neurotoxic molecules. The cells isolated from the brain were enriched for CD8+ cells by a magnetic positive selection. RNA extracted from these CD8+ cells was then analyzed for detection of gene expression by RT-PCR. As a result, mRNA for granzymes A and B, perforin, and IFN-γ, all agents of tissue destruction, was identified in the CD8+ cells from the brains of SIV-infected animals at both 2 and 11 wk p.i. (Fig. 10⇓). In contrast, the mRNA of CD8+ cells obtained from the brains of control animals contained heterogeneous and variable expression patterns of the analyzed molecules.
mRNA expression of CD8β, perforin, granzymes A (Gzm A) and B (Gzm B), and IFN-γ in CD8+ cells obtained from the brains of uninfected controls and of SIV-infected animals at 2 and 11 wk following infection (analysis of cells from 2 wk p.i. animal 347 was not performed due to insufficient sample). The expression of the molecules was evaluated by RT-PCR on samples normalized for amounts of 18S cDNA, followed by agarose gel electrophoresis. Comparison to the molecular size markers (100-bp ladder, brightest band corresponds to 600 bp) indicates the PCR products were of the expected size (with the exception of the IFN-γ signal from uninfected animal 331, which is of an incorrect size and thus does not represent the IFN-γ mRNA transcript). Control, Negative control.
Discussion
Here we used a microglia isolate of SIV to infect rhesus macaques for the purpose of correlating relevant immune cells within and outside the brain with CNS infection and dysfunction. The brain of each infected animal was examined for the infiltration of lymphoid cells. Immunohistochemical analysis of several brain segments from infected animals showed the presence of mononuclear leukocytes in the parenchyma and in the perivascular area. These infiltrating cells were predominantly CD3+CD8+ T cells, a subset that was dramatically increased in the brain tissue of the infected animals at 11 wk p.i. compared with uninfected controls.
Periodic evaluation of T cell subpopulations in the circulation of the infected animals over an 11-wk-period showed an enlargement of the CD8 compartment, peaking at 2 wk p.i., although the CD4+ cells remained unaltered. Indeed, a burst of proliferation is found in the CD8+ T cell compartment within the first few weeks of SIV infection (27, 36). Our study as well as those of others show that these CD8+ T cells express high levels of CD11a on the surface (37, 38), indicating that these cells become activated and able toadhere to endothelial cells, possibly followed by transmigration to tissue parenchyma. In parallel with the changes in blood CD8+ T cells at 2 wk p.i., a corresponding mild increase in the number of cells is found in the brain at 2 wk p.i. However, at 11 wk p.i., while the number of CD8+ T cells in the circulation has declined from the 2 wk p.i. peak, the CD8+ T cell presence in the brain has increased to over 5-fold in number compared with controls.
In peripheral sites such as the blood, spleen, and lymph nodes, the CD4:CD8 ratio of the infected animals decreased in comparison to control animals. This finding suggests that the virus infection induced an expansion of the CD8 compartment relative to CD4 that was not restricted to the brain, but was a systemic phenomenon, confirming previous observations in the SIV model and also in HIV (39). However, the CD4:CD8 ratio in the brains of SIV-infected monkeys and their uninfected counterparts was not statistically different and, in both, was lower than in the periphery. This evidence points to the predominance of CD8+ T cells in the CNS even under nonpathological conditions, as has been found in the CSF of monkeys (40) and brains of mice (41).
Naive CD8 cells in monkeys express high levels of CD45RA, but memory cells lose these molecules from the surface (27). However, an effector CD8+ T cell does not necessarily present low levels of CD45RA, because cells with naive phenotype can have effector activity, with expression of high levels of CD11a. Effector CD8+ T cells initiate transcription of cytokines such as IFN-γ early in the activation process. Other effector molecules such as granzymes A and B are expressed upon acquisition of cytotoxic function, and their expression is a useful marker of CTL activation in vivo (32, 42). Here, we found that brain-infiltrating CD8+ cells expressed mRNA for the cytotoxic granule proteins perforin, granzyme A and granzyme B, as well as for IFN-γ, indicating an effector phenotype at 2 and 11 wk after infection. CD8+ T cells can regulate IFN-γ expression in an Ag-specific manner, expressing it when in contact with their cognate Ag (42). The expression pattern we found here is consistent with our previous immunohistochemical and in situ hybridization studies identifying granzyme B and IFN-γ expression in the brains of SIV-infected monkeys (20, 43), and our finding of infiltrating SIV-specific cells with CTL activity in the brains of infected monkeys (33).
Despite having a similar pattern of surface marker expression, CD8+ cells in the brains of uninfected animals have an inconsistent or mixed pattern of mRNA expression for the proinflammatory and cytolytic molecules. This suggests that the few CD8 cells in the brain of uninfected animals are circulating cells not carrying out an effector function, but may represent cells in what is termed a surveillance mode (44). Highly activated cells, silent in terms of effector molecule production, can circulate in the blood and tissues, switching to an effector mode upon Ag stimulation. At the second week after infection, although only a limited increase in the number of CD8+ T cells in the brain was observed, the cells that got into the brain were indeed qualitatively different, because the expression of these molecules was consistent and homogeneous. At 11 wk, a highly significant increase in the CD8+ T cell population occurred with maintenance of a consistent and homogeneous level of production of effector molecules. These findings suggest that the CD8 cells that are present in the brain of SIV-infected animals are qualitatively as well as quantitatively distinct from controls.
Certainly, the increased number of CD3+CD8+ T cells at 11 wk, a relatively early stage of infection, was a most remarkable finding in terms of brain cell populations. However, the fact that we identified 1) far greater expression of activation markers on brain-derived CD8+ T cells than on those in the periphery and 2) expression of RNA to cytolytic and proinflammatory molecules adds a functional dimension to the quantitative changes. Immunohistochemical and FACS studies by others have indeed shown that CD8+ T cells are present in the brains of SIV-infected macaques both at the acute, and later stages of the disease (3, 45, 46). The CD8+ T cells that migrate into the brain can have a protective antiviral role (45), as suggested by the lowering of brain viral load in the 11 wk vs 2 wk p.i. animals and as described for other CNS viral infections such as influenza encephalitis (47). However, a pathogenic role of CD8+ T cells in virus-induced CNS disease has been suggested not only by us for SIV-infected monkeys (33), but for HTLV-I in humans (48), as well as mouse hepatitis virus, Semliki, Borna, and Theiler’s virus infections in rodents (49, 50, 51, 52).
It is notable that there was a temporal coincidence between peripheral and CNS events related to the peak of virus load, but not to the increase of CD8+ T cells. There is a time lag between the expansion of CD11ahigh CD8+ T cells in the periphery and their accumulation into the brain. This suggests that the increase of CD8+ T cells in the brain is not exclusively due to a systemic circumstance of activation and expansion of the CD8 subset, although such a change is potentially reflected in the mild increase in brain CD8+ T cells found at 2 wk p.i. It is recognized that activated T cells can cross the unaltered blood-brain barrier into the brain, independent of Ag specificity, as part of general immune surveillance of the CNS (53). Naive T cells can also enter an inflamed CNS (54). However, the retention and accumulation of T cells in the CNS requires Ag recognition (55, 56), as does effector molecule production (54). Thus, following infection and the consequent Ag stimulation, the CD8+ T cell population observed in the brain has changed from one mainly composed of cells in the surveillance mode to one of effector activity, capable of producing molecules that can target infected cells, but also with the capacity of inducing tissue damage. The CNS is a specialized tissue with a very low degree of functional redundancy and is thus quite sensitive to cellular injury that can compromise the ability of the brain to carry out its critical role.
Overall, our data support the concept that activated cells are present in the CNS. With respect to the expression of CD45RA in our experiments, the percentage of naive and memory CD8+ T cells in the brain was not statistically different between SIV-infected and control animals. This was due to the fact that uninfected controls presented a higher percentage of CD8+ T cells with memory phenotype in the brain in comparison to the periphery. Interestingly, recent data indicate that specific CD8+ memory and effector T cells migrate to nonlymphoid tissues (57). In other experimental models, particularly in rodents, specificity and even MHC restriction were not crucial factors for T cells to infiltrate the CNS, suggesting that SIV-specific cells are not the only ones to enter the brain (53, 54). However, the CD8+ lymphocytes we found in the brain of either infected or control monkeys were close to 100% CD11ahigh and CD95high, indicating an activation status independent of the expression of memory markers (28).
Although studies have indicated that low levels of CD28 expression correlate with effector function and infection (31), we saw no such differences, at least not to the level of significance, for the expression of this marker in the infected monkeys. In part, this may have resulted from the uneven expression of CD28 in the uninfected animals. However, not all effector cells exhibit low levels of CD28, as SIV-specific CTL, identified by tetramer binding studies, are divided for the expression of this marker (38). Others have shown a population of CD8+CD28+ cells that expanded during viral infection, produced IFN-γ and perforin, and had effector function. This cell subset, as opposed to the CD28− population, retained the ability to proliferate (58). The nature of these cells in the periphery as well as the brain has yet to be examined in SIV-infected subjects. Our results emphasize that alterations in CNS T cell subpopulations soon after infection result in an abnormally large number of activated CD8+ T cells in the brain; this may increase the concentrations of cytolytic and proinflammatory cytokines to a pathological level.
The purified brain CD8+ cells obtained from SIV-infected animals expressed the mRNA for perforin, granzymes A and B, important molecules involved in the apoptotic process and cell lysis, as well as for IFN-γ. Proinflammatory cytokines, among them IFN-γ, are known to induce an increase in the expression of adhesion molecules like ICAM-1 on endothelial cells (59). ICAM-1 is the ligand for CD11a (LFA-1) expressed on the leukocyte cell surface. The up-regulation of this molecule may play a role in the adhesion of leukocytes to the endothelial cell, a step that precedes the transmigration to the tissue, especially if LFA-1 expression is also increased on the cell surface, as we observed here on CD8+ T lymphocytes. Thus, IFN-γ can contribute to the migration of CD11ahigh CD8+ T cells to the parenchyma. Additionally, IFN-γ can regulate cell traffic into the brain by inducing chemokine synthesis (60) and can function as a potent macrophage/microglia activator that induces the production of potentially neurotoxic molecules.
Our results suggest that the SIV infection promotes a numerical increase of CD8+ T lymphocytes with the capacity to enter the CNS. The accumulation of such cells in the brain, despite having a protective role, can potentially induce cellular damage in the brains of SIV-infected monkeys and incite the functional CNS abnormalities characteristic of this and similar infections.
Acknowledgments
We thank Curtis Phillipson and Cathie York-DeFalco for technical assistance, Joe Trotter, Allan Saluk, and Mary Cleary for helpful suggestions on FACS techniques, Dr. Lisa Madden for help with statistical analysis and for a critical review of the manuscript, Dr. Michael Buchmeier for stimulating discussion, and Phyllis Minick and Michael Gilliatt for editorial assistance.
Footnotes
-
↵1 This work was supported by National Institutes of Health Grants MH59468, MH61224, and MH62261 (to H.S.F.) and DA12444 (to S.J.H.). This is manuscript number 14001-NP from The Scripps Research Institute.
-
↵2 Address correspondence and reprint requests to Dr. Howard S. Fox, Department of Neuropharmacology, The Scripps Research Institute, 10550 North Torrey Pines Road, CVN 8, La Jolla, CA 92037. E-mail address: hsfox{at}scripps.edu
-
↵3 Abbreviations used in this paper: EP, evoked potential; CSF, cerebrospinal fluid; p.i., postinoculation; bDNA, branched DNA; BSAEP, brainstem auditory EP; AEP, auditory EP; LCA, leukocyte common Ag.
- Received May 10, 2001.
- Accepted August 31, 2001.
- Copyright © 2001 by The American Association of Immunologists