Human immunodeficiency virus type-1 vulnerates nascent neuronal cells


Yoshio Koyanagi, Laboratory of Viral Pathogenesis, Institute for Virus Research, Kyoto University, 53 Shogoin-kawara-cho, Sakyo-ku, Kyoto 606-8507, Japan. Tel: +81-75-751-4811; fax: +81-75-751-4812; email:


Macrophages or microglial cells are the major target cells for HIV-1 infection in the brain. The infected cells release neurotoxic factors that may cause severe neuronal cell damage, especially in the basal ganglia and hippocampus. In this study, we used rat OHC to examine the region-specific neuronal cell damage caused by HIV-1-infected macrophages. When OHC was cocultured with HIV-1-infected MDM, we found that neuronal cells at the GCL of the DG were preferentially killed via apoptosis, and that projection of MF from GCL to PCL of the CA3 region was severely disturbed. We marked precursor cells around the DG region by using an EGFP-expressing retrovirus vector and found that these cells lost the ability to differentiate into neurons when exposed to HIV-1-infected MDM. In the DG, new neurons are normally incorporated into GCL or PCL, while in the presence of HIV-1-infected MDM, mature neurons failed to be incorporated into those layers. These data indicate that the neurotoxic factor(s) released from HIV-1-infected macrophages impede(s) neuronal cell repair in brain tissue. This suggests that DG is the region of the hippocampus most vulnerable to neuronal damage caused by HIV-1 infection, and that its selective vulnerability is most likely due to the highly active neurogenesis that takes place in this region.

List of Abbreviations: 

acquired immunodeficiency syndrome


antiretrovirus therapy


central nervous system


dentate gyrus


enhanced green fluorescent protein




granule cell layer


glial fibrillary acidic protein


human immunodeficiency virus-1


neurofilament protein




monoclonal antibody


monocyte-derived macrophages


mossy fiber


murine leukemia virus


multiplicity of infection


neuronal nuclei


oligodendrocyte O4


organotypic hippocampal slice culture


peripheral blood mononuclear cells


pyramidal cell layer




room temperature


transcriptional transactivator


terminal deoxynucleotidyl transferase (TdT)-mediated biotinylated UTP nick end labeling


tumor-necrosis factor α

HIV infection is well known to be a cause for development of dementia. Most HIV-associated dementia is subcortical dementia due to predominant involvement of the basal ganglia, cerebral cortex, and hippocampus (1–3). Introduction of ART has improved the immunological condition of patients and reduced the incidence of HIV severe dementia in AIDS patients (1, 4, 5). However, due to improved survival rates among ART-treated patients and healthy HIV-infected individuals, prevalence rates of HIV-1 associated neurological disease continue to rise (1). These patients display milder neurological disease than the severe neurological disease seen in non-treated patients (1, 2). Histopathological examination of autopsy samples from HIV-1 infected brains reveals the presence of multinucleated giant cells and loss of synapses and neurons (6).

Although the mechanisms involved in loss of synapses and neurons in these patients are not completely understood, some data support the contention that HIV-1 infected macrophages or microglial cells release neurotoxic factors such as viral products, excitotoxins, and/or cytokines and chemokines that damage neurons through multiple pathways (1, 7). These factors may also work to activate astrocytes which then produce chemokines and cytokines that affect neuronal function (8, 9). Such neurotoxic factors are likely to affect a diverse range of neurons in the most vulnerable areas of the CNS, including the hippocampus, frontal cortex and white matter (1). In HIV-infected individuals, viral and host factors released from HIV-1-infected macrophages might affect hippocampal mediated learning and memory formation (1, 10). It has been reported that macrophage-induced inflammation affected hippocampal plasticity in a murine model of HIV-1 encephalitis (11). However, there has been no longitudinal study on the pathogenesis of HIV-1 encephalopathy and therefore no information has been collected by in vitro assay regarding the mechanism of causation of milder HIV-1 associated neurological disease.

As an in vitro model of HIV-1 encephalopathy that mimics its pathogenesis in vivo, OHC were employed in the present study. OHC would provide an alternative model to examine neuronal cell function in the hippocampus (12). As cellular composition, architecture and connections within the hippocampus are well preserved in OHC, the microenvironment in which the cells are immersed is similar to that in living animals (13–15). Although a few exceptions have been found, probably as a result of afferent deprivation of neuronal signals (16, 17), neurons in slice culture maintain their physiological arrangement and the capability to transmit signals (12). In addition, neurogenesis has been reported to occur in hippocampal slice cultures (18, 19). Therefore, slice cultures have been widely used to study morphology and plasticity of the hippocampus. Since HIV does not infect cells with a neuronal lineage, it was considered appropriate to use small animal-derived OHC in combination with HIV-1-infected human macrophages to study pathological changes in the brain of HIV-1-infected individuals.

In the present study, we established an in vitro model of HIV-1 encephalopathy using rat OHC exposed to HIV-1-infected MDM. This model reproduced pathological alterations such as neuronal apoptosis, loss of synapses and activation of astrocytes or microglial cells. More importantly, granule cells in the DG, where active neurogenesis occurs, were found to be predominantly damaged at an early stage, and neuronal cell differentiation was found to be disturbed by viral and/or host factors released from HIV-1-infected MDM. This culture system will be useful for understanding the pathology of HIV-1 encephalopathy in the hippocampus.


OHC and cocultivation with HIV-1-infected macrophages

OHC were prepared from postnatal day seven Wister Hannover GALAS rats (CLEA, Osaka, Japan) as previously described (18). The experiments were carried out in accordance with the guidelines of Kyoto University for animal experimentation. After decapitation, the brain was removed and transversely sliced at the hippocampus into a 350 μm thickness on a McIlwain tissue chopper (Mickle Laboratory Engineering, Guildford, UK). The slices were cultured on porous translucent membrane (Millicell-CM, Millipore, Bedford, MA, USA) at 34°C and the culture medium was changed three times a week. The culture medium consisted of 50% OPTI-MEM (Invitrogen, Carlsbad, CA, USA), 25% heat inactivated horse serum (Invitrogen) and 25% Hank's balanced salt solution (Invitrogen) supplemented with D-glucose (5 g/L, Wako, Osaka, Japan), penicillin and streptomycin (100 units/ml, 100 μg/ml, respectively, Invitrogen). Two weeks after initiation of culture, slices were cocultured with HIV-1JRFL (20)-infected (MOI 1.0) or uninfected MDM isolated from PBMC of HIV-1 seronegative healthy donors, across the porous membrane in culture medium for another two weeks. The concentration of p24 antigen was measured by ELISA (RETROtek, ZeptoMetrix, Buffalo, NY, USA).

Retrovirus vector transduction

An MLV-based vector, SRα-EGFP (21), was derived from SRαLthy (22) by replacing the murine thy 1.2 gene with EGFP. Vector was generated as described before (21). The titer of vector was 1.7 × 107 infectious units per milliliter. Slices were inoculated with SRα-EGFP by microinjection (FemtJet and Inject Man NI2, Eppendorf, Hamburg, Germany) in the suprapyramidal region of the GCL at the DG (18). Then, slices were cultured without, or with uninfected or HIV-1-infected MDM, as described above.

Staining dead cells in OHC

After removal of culture medium, slices were incubated at 34°C with culture medium containing SYTOX Green dye (Invitrogen) for 30 min and then samples were examined using a fluorescent microscope (Nikon TS100, Nikon, Tokyo, Japan) using a GFP filter with 5× objective.


MAb against NeuN, nestin and O4 were purchased from Chemicon (Temecula, CA, USA). MAb against NFP and rat CD68 (clone ED1) were purchased from DakoCytomation (Carpinteria, CA, USA), and Serotec (Oxford, UK), respectively. A rabbit polyclonal antibody against GFAP was purchased from DakoCytomation. A rat mAb against EGFP was purchased from Nakarai (Kyoto, Japan). A biotin-conjugated goat anti-mouse IgM was purchased from Chemicon. Alexa Fluor 488-, 594-conjugated goat anti-mouse/rabbit IgG and Alexa Fluor 594-conjugated streptavidin were obtained from Invitrogen. An FITC-conjugated goat anti-rat IgG was obtained from Jackson Immuno Research Laboratories (West Grove, PA, USA).


Slices were fixed by immersion in 4% PFA for one hour at 4°C, and then incubated with blocking buffer containing 5% normal goat serum Vector Laboratories (Burlingame, CA, USA) and 0.3% Triton X-100 at 4°C overnight, followed by incubation with primary antibodies against the following molecules; NeuN, EGFP, NFP, rat CD68, O4, or GFAP at 4°C overnight in blocking solution. Sample stained with antibody against O4 was further incubated with biotin-conjugated anti-mouse IgM antibody at 4°C overnight. These samples were subsequently incubated at RT for another six hours with fluorescent dye conjugated secondary antibodies; Alexa Fluor 594-conjugated anti-mouse/rabbit IgG with or without FITC-conjugated anti-rat IgG or Alexa Fluor 594-conjugated streptavidin. Nuclei were stained using Hoechst33342 (Invitrogen). Each sample was examined under a fluorescent microscope (Leica CTR6500, Leica Microsystems, Heidelberg, Germany) using a Texas Red filter with 5× and 10× objectives, and a confocal laser microscope (TCS SP2 AOBS, Leica Microsystems) using 405, 488 or 543 nm excitations with 10×, 20× and 40× objectives. The quantification of fluorescent intensity was done following the protocol supplied by the manufacturer (LCS Lite software, Leica Microsystems).

TUNEL staining

Slices were fixed by immersion in 4% PFA for one hour at 4°C, treated with 20 mg/ml proteinase K in 50 mM Tris-HCl buffer at 37°C for 15 min, and then fixed again with 4% PFA for 30 min at 4°C. Following permeabilization with 0.3% Triton X-100 for one hour at RT, the samples were incubated with TdT solution (TdT buffer, 25 mM CoCl2, TdT [Terminal Transferase recombinant, Roche Applied Science, Indianapolis, IN, USA], biotin-16-2′-deoxy-uridine-5′-triphosphate [Roche Applied Science]) at 37°C for four hours. After washing, samples were stained with anti-NeuN antibody as described above and subsequently incubated with fluorescent dye conjugated secondary antibodies: Alexa Fluor 488-conjugated anti-mouse IgG (NeuN) and streptavidin Alexa Fluor 594-conjugated (TUNEL). Nuclei were stained using Hoechst33342 at RT for six hours. Samples were examined under a confocal laser microscope using 405, 488 and 543 nm excitations with 20× and 40× objectives as described above.


Microscope analysis of OHC exposed with HIV-MDM

Since OHC is known to preserve morphological features of the CNS for at least one month, we used it to explore tissue damage caused by HIV-1 protein or humoral host factor produced by HIV-1-infected macrophages in the hippocampus. Typical cellular arrangement of granule cells was reconstituted in slices on a porous membrane 14 days after initiation of culture and slices were then exposed to HIV-1JRFL-infected or uninfected MDM across the porous membrane for another 14 days. The laminated structure and morphological organization of slices cocultured with or without uninfected MDM was maintained during the 14 days (data not shown), while the slice exposed to HIV-1-infected MDM (HIV-MDM) was found to gradually become thinner (Fig. 1A). Although cell morphology of MDM was not changed by HIV-1 infection, high concentrations of HIV-1 even greater than 200 ng p24 antigen per milliliter were found in culture supernatant (Fig. 1A and B). In addition, moderate thinning down of slices was induced by exposure to culture supernatant of lower concentrations of p24 antigen (below 20 ng per milliliter) from HIV-1-infected MDM (data not shown). These results suggest that HIV-1-infected MDM produce factor(s) that impair the maintenance of neuronal tissue architecture.

Figure 1.

Longitudinal alteration of OHC exposed to HIV-1-infected MDM. (A) OHC was cultured in the presence of HIV-1-infected MDM for two weeks. Microscopic examination was performed at days (D) 3, 9 and 14 after cocultivation. The slices gradually became thinner and cells were shed from them (a), though morphology of MDM was not affected by HIV-1 infection (b). Scale bar: 500 μm (a), 100 μm (b). (B) Large amount of HIV-1 was produced by HIV-1-infected MDM during cocultivation. X-axis is days after cocultivation and Y-axis is p24 concentration in supernatant of culture medium measured by ELISA.

HIV-1-induced region-specific damage of neuronal cells in the hippocampus

To examine which region was affected by HIV-1-infected MDM, OHC was cultured in the presence of a non-permeable staining reagent, SYTOX, and cell damage was examined. Two days after exposure to HIV-1-infected MDM, severe cell damage in the neuronal cell layer was found (Fig. 2A), suggesting that neurons in these regions were predominantly harmed at an early stage after exposure to HIV-1-infected MDM. Fourteen days after exposure, we found severe loss of neurons especially at the GCL of the DG as revealed by staining with antibody against NeuN (Fig. 2B). Loss of GCL neurons was also found by staining with antibody against Calbindin D28K, one of the calcium binding proteins expressed specifically in GCL neurons (data not shown). Three-dimensional analysis revealed that the number of NeuN positive neurons clearly decreased and thickness of the sample became thinner in slices exposed to HIV-MDM compared to those without cocultivation (Fig. 2C). This change could not be found when OHC was exposed to HIV-1-infected HeLa-derived cells (data not shown), indicating that loss of neurons was specific to HIV-1-infected MDM and that exposure to HIV-1-infected MDM was more effective in reproducing the pathological alteration of HIV-1 encephalopathy. These results suggest that humoral factor(s) produced from HIV-1-infected MDM preferentially damage cells in the DG.

Figure 2.

Neurotoxic effect in OHC exposed to HIV-1-infected MDM. (A) Neurons in the CA1 and DG were harmed at early stage after the initiation of the exposure to HIV-1-infected MDM. Slices were cultured without (a; control) or with uninfected (b; MDM) or HIV-1-infected MDM (c; HIV-MDM) for two days and stained with SYTOX. CA1: CA1 region, CA3: CA3 region, DG: dentate gyrus. Scale bars: 100 μm. (B) Severe loss of neurons at the GCL of the DG was found in OHC exposed to HIV-1-infected MDM. Slices were cultured without (a; control) or with uninfected (b; MDM) or HIV-1-infected MDM (c; HIV-MDM) for 14 days and stained with antibody against NeuN. Rectangular regions were enlarged. CA1: CA1 region, CA3: CA3 region, DG: dentate gyrus. Scale bars: 100 μm. (C) Thinning of slices and loss of neurons were found by three-dimensional analysis in OHC exposed to HIV-1-infected MDM. Slices were cultured without (a; control) or with HIV-1-infected MDM (b; HIV-MDM) for 14 days and stained with NeuN (neuron) and Hoechst33342 (nuclei). The x-z plane of slices was imaged using a fluorescent microscope. (D) Neuronal apoptosis was induced by HIV-1-infected MDM. Slices cultured with HIV-1-infected MDM were stained with TUNEL (red) and anti-NeuN antibody (green), and nuclei were stained with Hoechst33342 (blue). A low magnification image is shown in left panel (a), and a high magnification image is shown in right panel (b). Rectangular regions were enlarged. CA3: CA3 region, DG: dentate gyrus. Scale bars: 150 μm (a), 100 μm (b, upper panels), 25 μm (b, lower panels). (E) Neuronal apoptosis was induced especially at the GCL of the DG. The numbers of TUNEL positive neurons were counted at three areas (CA1, CA3 and DG) of the slices cultured with uninfected (MDM) or HIV-1-infected MDM (HIV-MDM). Y-axis indicates the number of TUNEL positive neurons in slice.*: P < 0.01 versus each area of MDM, #: P < 0.01 versus CA1 and CA3 in HIV-MDM. (F) Loss of synapses was induced in GCL neurons by HIV-1-infected MDM. Slices cultured with uninfected (a; MDM) or HIV-1-infected MDM (b; HIV-MDM) were stained using antibody against NFP and neuronal filaments were visualized. Low magnification images are shown in left panels, and high magnification images in right panels. The area enclosed by the dotted line indicates the DG area. Long axons were formed (filled arrowheads) in GCL neurons at DG and they projected to CA3 in slice cultured with uninfected MDM (a). Conversely, axons were severely damaged (open arrowheads) in GCL neurons of DG in slice cultured with HIV-1-infected MDM (b). Scale bars: 50 μm. (G) Pixel mean of the signal of NFP-stained samples was significantly decreased in the slices cultured with HIV-1-infected MDM (HIV-MDM). *: P < 0.01 versus uninfected MDM (MDM).

Neuronal apoptosis in the DG induced by HIV-1-infected MDM

To investigate the mechanism involved in the reduction in number of neurons especially at the DG region, we next carried out TUNEL staining to examine whether these cells were killed via apoptosis. There were many TUNEL positive cells in NeuN+ neurons localized at the DG region in slices exposed to HIV-1-infected MDM (Fig. 2D and E). By contrast, in slices exposed to uninfected MDM, there were less TUNEL positive neurons in PCL and GCL. Furthermore, in slices exposed to uninfected MDM, there was no difference in the number of TUNEL positive cells between PCL (CA1 and CA3) and GCL (DG) (Fig. 2E). These data indicate that granule cells in the DG area are highly susceptible to apoptosis which is probably induced by neurotoxic humoral factor released from HIV-1-infected MDM.

Synaptic loss of GCL neurons induced by HIV-1-infected MDM

We assumed that other mechanisms besides neuronal apoptosis induced neuronal damage at the DG in OHC. To confirm this assumption, we used NFP as a marker to visualize neuronal filaments such as axons and dendrites. In slices exposed to HIV-1-infected MDM, expression of NFP was severely disturbed at the DG region (Fig. 2F and G). Furthermore, the MF, collateral sprouting of dentate granule cell axons which form powerful excitatory synapses onto the proximal dendrites of CA3 pyramidal cells (23), were severely damaged in slices exposed to HIV-1-infected MDM (Fig. 2F, open arrowheads) as compared to uninfected MDM (Fig. 2F, filled arrowheads), indicating that HIV-1-infected MDM induce malformation of neuronal filaments and abnormality of synaptic projection onto neurons.

Activation of astrocytes or microglial cells and depletion of oligodendrocytes by HIV-1-infected MDM

We reproduced neuronal tissue damage through neuronal apoptosis and axonal malformations by exposure to HIV-1-infected MDM in the region of the DG. Other types of cells in OHC, which also includes astrocytes, microglial cells and oligodendrocytes, were also examined. It has been reported that astrocytes and microglial cells are extraordinarily activated in the brains of HIV-1 encephalopathy patients (2). From examination of GFAP (astrocyte marker), rat CD68 (microglial cell marker) and O4 (oligodendrocyte marker) expressions using immunofluorescence techniques, we found enlarged astrocytes and microglial cells in the slices exposed to HIV-1-infected MDM, indicating that astrocytes and microglial cells were activated (Fig. 3A and B). In addition, in the slices exposed to uninfected MDM, astrocytes and microglial cells were slightly enlarged, suggesting that both uninfected and HIV-1-infected MDM produce the factor(s) that induce activation of astrocytes and microglial cells (Fig. 3A and B). By contrast, the number of oligodendrocytes was clearly decreased in slices exposed to HIV-1-infected MDM, indicating that oligodendrocytes were preferentially damaged as seen in neurons by HIV-1-infected MDM (Fig. 3C). These results indicate that OHC cocultured with HIV-1-infected MDM is an adequate model for investigation of the molecular mechanism of HIV-1 encephalopathy.

Figure 3.

Abnormality in astrocytes, microglial cells, and oligodendrocytes. Slices were cultured without (a; control) or with uninfected (b; MDM) or HIV-1-infected MDM (c; HIV-MDM) for 14 days and stained using antibodies against GFAP (A), ED1 (B) and oligodendrocyte O4 (C). The number of astrocytes increased in the presence of HIV-1-infected MDM (A), and the number of microglial cells increased and morphological changes were found when exposed to HIV-1-infected MDM (B). Conversely, oligodendrocytes were damaged by HIV-1-infected MDM (C). Scale bars: 50 μm.

Disablement of neuronal cell differentiation by HIV-1-infected MDM

It has become known that active neurogenesis occurs in the DG region of the hippocampus throughout the lifespan and that it is crucial for formation of neuronal plasticity (24). Many neuronal and glial progenitor cells or stem cells reside in the region and intrinsic and spontaneous neurogenesis are reported to take place at the DG in OHC (18, 19). In a healthy hippocampus, damaged neurons are continuously replaced by new cells generated at the DG region. However, in HIV-1-infected MDM-exposed slices, the damaged cells as shown in Figure 2 might accumulate as a result of weakened replacement. There could be two explanations for this observation. One is that the total number of precursors including progenitor cells and stem cells in DG decrease as a result of exposure to HIV-1-infected MDM. Another is that HIV-1-infected MDM interferes with the ability of precursor cells to differentiate and migrate.

To examine the first possibility, progenitor cells in the slices were stained with antibody against nestin, which is expressed predominantly in precursors including stem or progenitor cells of the CNS. We found similar numbers of nestin+ precursor cells in slices exposed to either uninfected or HIV-1-infected MDM (Fig. 4A-a, b), indicating that factor(s) produced from HIV-1-infected MDM did not preferentially kill the precursors. To examine the second possibility, we labeled endogenous precursors in the slices using EGFP-expressing MLV vector (SRα-EGFP), which was microinjected into the suprapyramidal region of the GCL. Since MLV only infects dividing cells, and its DNA is incorporated into the host DNA, newly divided cells would specifically express EGFP. After inoculation with SRα-EGFP, slices were exposed to HIV-1-infected or uninfected MDM or not exposed to MDM at all. The total numbers of EGFP-expressing cells in all slices were similar irrespective of exposure of HIV-1-MDM or not for 2 weeks post inoculation (data not shown). To examine the ability of neuronal cell to differentiate, slices were fixed and stained with anti-NeuN antibody (neuron marker) and anti-EGFP antibody. In the slices cultured with uninfected MDM, some EGFP+ cells also expressed NeuN and were incorporated into the GCL or PCL (Fig. 4B-a), and those cells had axon-like processes (Fig. 4B-a, filled arrowheads). Measurement of the numbers of EGFP+ NeuN+ neurons incorporated into the GCL or PCL revealed that half of EGFP+ cells were neurons and those cells were incorporated into neuronal cell circuit in slices cultured with or without uninfected MDM (Fig. 4C). By contrast, in slice exposed to HIV-1-infected MDM, the majority of EGFP+ cells were not incorporated into the GCL or PCL (Fig. 4B-b and C). Furthermore, EGFP+ cells which were not incorporated into those layers did not express NeuN (Fig. 4B-c and C), indicating that EGFP+ cells could not differentiate into mature neurons and that they might have differentiated into other cell types. These results indicate that maintenance of neuronal cell circuits is disturbed by exposure to HIV-1-infected MDM through inhibition of neuronal cell differentiation, but not by elimination of their progenitors in the DG.

Figure 4.

Abnormality in differentiation ability of precursor cells. (A) The number of precursors was not affected by HIV-1-infected MDM. Slices were cultured with uninfected MDM (a; MDM) or HIV-1-infected MDM (b; HIV-MDM) for 14 days and stained using antibody against nestin to visualize endogenous precursors. CA3: CA3 region, DG: dentate gyrus. Scale bars: 50 μm. (B) Maturation of precursor cells into neurons was strongly inhibited in the presence of HIV-1-infected MDM. Precursor cells were labeled using an EGFP-expressing MLV vector and slices were cultured with uninfected (a; MDM) or HIV-1-infected MDM (b and c; HIV-MDM). Slices were stained using antibody against EGFP (green) and NeuN (red). EGFP+ cells also expressed NeuN and had axon-like processes (filled arrowheads) in slices exposed to uninfected MDM (a, GCL and PCL). Conversely, EGFP+ cells did not express NeuN and the number of EGFP+ cells, which were incorporated into GCL or PCL clearly decreased in slices exposed to HIV-1-infected MDM (b, c). granule cell layer (GCL), pyramidal cell layer (PCL), CA3: CA3 region, dentate gyrus (DG). Scale bars, 100 μm, 20 μm (GCL right panels and PCL panels). (C) HIV-1-infected MDM induced abnormality of neuronal cell differentiation. Slices were cultured and stained using antibodies against EGFP and NeuN as described in B. The number of total EGFP+ cells and EGFP+/NeuN+ cells which were incorporated into GCL or PCL were counted and classified into mature neurons (incorporated to neuronal circuit) or cells other than mature neurons. Y-axis indicates percentage of EGFP+ cells in slice.


In this study, we established a cocultivation system employing HIV-1-infected MDM and rat OHC that reproduces pathologic changes of HIV-1 encephalopathy. Neuronal dysfunctions including neuronal apoptosis and synaptic loss were produced especially in the GCL of the DG region in the hippocampus. Furthermore, in the DG, the differentiation of precursor cells into neurons was inhibited by exposure to HIV-1-infected MDM. These results highlight that the DG is profoundly susceptible to HIV-1 caused neuronal damage, and that HIV causes hippocampal dysfunction through damaging preexisting mature neurons and inhibiting the production of new neurons.

It has been reported that culture systems of a combination of OHC and HIV-1-encoded proteins, such as HIV-1-envelope glycoprotein 120 (gp120) or Tat, are useful for the study of HIV-1-induced neuronal damage (25–27), and that neuronal toxicity is induced by those viral proteins. Those culture systems are satisfactory for understanding the mechanisms by which a single protein affects maintenance of neuronal tissue. However, in the brain of HIV-1-infected patients, there are several viral proteins as well as host factors released from HIV-1-infected macrophages or microglial cells. Our coculture system using OHC and HIV-1-infected macrophages may reflect the brain damage which is concomitantly induced by several viral proteins and/or many host factors released from HIV-1-infected MDM. The combination of OHC and HIV-1-infected MDM may be more effective for identifying the mechanisms of HIV-1 encephalopathy. Moreover, in our coculture system, OHC might be exposed to high concentrations of viral and host factors and neuronal damage or degeneration thus induced, though neuronal damage was also found from exposure to a lower concentration of p24 antigen of culture supernatant. In the HIV-1-infected brain, HIV-1-infected macrophages penetrate across the blood-brain-barrier (2). More severe damage is induced around the area invaded by infected macrophages compared to uninvaded areas, as shown in simian immunodeficiency virus-infected brain samples (28). Our coculture system might reflect alterations which are closely related to those in areas invaded by HIV-1-infected macrophages.

Infected macrophages and microglial cells are known to be major HIV producers. They are also known to release viral proteins that can be deleterious to the CNS. HIV-1gp120, Tat and viral protein R have been shown to be toxic to neurons (29–33). HIV-1 gp120 interacts with cellular receptors and alters the signaling of the glutamate pathway. The signal induces cytokine production that can damage large number of neurons and affect the activation state of microglial cells and astrocytes (1). Nanomolar concentrations of gp120 have been reported to interact with the glycine binding site of the NMDA receptor (34). One possible molecular mechanism for gp120-induced neurotoxicity is that gp120 might induce glutamate-mediated excitotoxicity and initiate caspase cascades (35). Another viral protein Tat directly enters granule cells in the hippocampus and causes neurotoxicity (30). The neurotoxic effects of Tat seem to be mediated by interactions with a polyamine-sensitive site of the NMDA receptor in OHC (27). Infected cells also produce other neurotoxic factors, such as cytokines including TNF-α, quinolinic and arachidonic acid, platelet-activating factor and nitric oxide (2, 36–38). These factors promote further activation of macrophages and/or microglial cells, as well as proliferation and activation of astrocytes (2). In addition, activated astrocytes release intracellular Ca2+ and glutamate while reducing the uptake of glutamate. This raises the extracellular concentration of glutamate and other neurotoxins that can induce excitotoxic death of neurons (39, 40). As a result, neurodegeneration may be accelerated in the HIV-1 infected brain (2, 6). We found that HIV-1-infected MDM induces damage in neuronal cells at an early stage of exposure (Fig. 2). At a later point, neuronal apoptosis, synaptic loss, impairment of oligodendrocytes and activation of astrocytes and microglial cells are induced by HIV-1-infected MDM (Figs 2, 3). These phenomena are probably caused by several virion-free viral proteins, such as gp120 and Tat, which are known to interact with several receptors and induce neurotoxic signaling, causing release of many host factors, such as TNF-α, from HIV-1-infected MDM. As a result, observed changes in OHC after exposure to HIV-1-infected MDM are very similar to some of the pathological alterations detected in the brain with HIV-1 encephalopathy.

In Borna disease and Alzheimer's disease, the greatest tissue damage is reported to be localized in the DG of hippocampus (41, 42). Thus, the DG region appears to be predisposed to neuronal damage, including that induced by HIV-1 infection. The DG is one of the unique regions where new neurons are continuously generated throughout life (24, 43). Neural stem cells derived from the adult hippocampus differentiate into neurons and form synaptic networks (24). These morphological and physiological features strongly suggest that new neurons are incorporated into the local circuitry of the hippocampus and that they are involved in hippocampus-dependent memory formation and brain repair (24, 44, 45). We found that differentiation of precursors (general progenitor cells and stem cells) into neurons was significantly impaired at the DG region by exposure to HIV-1-infected MDM, and that new neurons were scarcely incorporated into neuronal circuits (Fig. 4). As a result, the number of neurons might be reduced because new neurons are not being supplied from the DG region. Although the number of new neurons was significantly reduced, their precursors were not affected by HIV-1-infected MDM, suggesting that precursors may have been directed toward the differentiation pathway that leads to the production of astrocytes rather than neurons. Then, it can be assumed that apoptosis or synaptic loss caused by exposure to HIV-1-infected MDM results in progressive neuronal degradation because the injured neurons are not replaced (Fig. 5). Selective inhibition of neuronal differentiation might be mediated by several viral proteins or many neurotoxic host factors. HIV-1 gp120 interacts with NMDA receptors, and induces disruption of glutamate and glycolytic pathway signaling and calcium homeostasis (1). NMDA receptors are frequently expressed on immature granule neurons, and signals via NMDA receptors regulate neurogenesis and neuronal migration in the hippocampus both in vitro and in vivo (46–48). When HIV-1 gp120 interacts with NMDA receptors in immature neuronal precursor cells, neurogenesis is impaired. It has been considered that TNF-α released from infected macrophages plays a role in the process of neuronal damage. TNF-α inhibits phosphorylation of three signaling molecules, serine/threonine protein kinase, extracellular signal-related kinase and glycogen synthase kinase 3β, which are suggested to have a role in neuronal differentiation and neuronal survival (49), thus, TNF-α might inhibit neuronal differentiation. However, it is possible that other humoral factors potentially have the ability to induce dysfunction in neurogenesis.

Figure 5.

Proposed model of HIV-1-infected macrophage-induced neuronal damage at the DG region. (A) In a normal OHC, MFs of GCL neurons (open dots) project to PCL neurons (open triangles) at CA3. New neurons differentiated from precursors were incorporated and configured in neuronal circuit (filled dots and triangles). (B) Conversely, in OHC exposed to HIV-1-infected macrophages, neurons underwent apoptosis (crossed dots and triangles) and MFs of GCL neurons (open dots) could not project to PCL neurons (open triangles) at CA3. However, precursors, which have the ability to replace damaged neurons, lose their ability to differentiate into neurons and can not be incorporated into GCL or PCL. Therefore, the damaged neurons and lost neuronal connections may accumulate in the DG region of hippocampus.

Furthermore, MF projected from GCL neurons to CA3 pyramidal neurons were disturbed in slices exposed to HIV-1-infected MDM (Fig. 2, NFP). It is well known that the DG provides the main input to the hippocampus, and that neuronal information reaches the CA3 region through MF generated by GCL axons at the DG (23). It is also well known that the hippocampus plays an important role in learning and memory (23, 44). Our findings suggest that loss of MF with HIV-1 infection might explain the clinical observation that reduction in the ability to learn and form memory is frequently observed in HIV-1 encephalopathy patients.

In conclusion, the culture system that we report here reproduces the pathological changes of HIV-1 encephalopathy. Neuronal damage especially at the DG is induced by viral and/or host factors released from HIV-1-infected macrophages or activated microglial cells. Under the influence of these factors, the hippocampus can not recover from neuronal damage because neuronal progenitor cells can not differentiate into neurons and be incorporated into neuronal cell circuits. Finally, vulnerability of the DG might explain HIV-induced progressive pathological changes and reduction in the ability to learn and form memory in HIV-infected patients. As our data shows, this culture system will also be an adequate model to investigate and answer important questions regarding the involvement of host factors in HIV-1 encephalopathy. This system enables longitudinal studies in the pathology of HIV-1 encephalopathy, which is a great advantage of this in vitro model over existing in vivo models.


We thank Naoko Misawa, Hiroshi Okada, Chuanyi Nie and Hiromu Yawo (Tohoku University) for helping with our study.

This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Sciences, and Technology of Japan and by grants for Research on HIV-AIDS and Health Science from the Ministry of Health, Labor, and Welfare of Japan.