The viral determinants that underlie human immunodeficiency virus type 1 (HIV-1) neurotropism are unknown, due in part to limited studies on viruses isolated from brain. Previous studies suggest that brainderived viruses are macrophage tropic (M-tropic) and principally use CCR5 for virus entry. To better understand HIV-1 neurotropism, we isolated primary viruses from autopsy brain, cerebral spinal fluid, blood, spleen, and lymph node samples from AIDS patients with dementia and HIV-1 encephalitis. Isolates were characterized to determine coreceptor usage and replication capacity in peripheral blood mononuclear cells (PBMC), monocyte-derived macrophages (MDM), and microglia. Env V1/V2 and V3 heteroduplex tracking assay and sequence analyses were performed to characterize distinct variants in viral quasispecies. Viruses isolated from brain, which consisted of variants that were distinct from those in lymphoid tissues, used CCR5 (R5), CXCR4 (X4), or both coreceptors (R5X4). Minor usage of CCR2b, CCR3, CCR8, and Apj was also observed. Primary brain and lymphoid isolates that replicated to high levels in MDM showed a similar capacity to replicate in microglia. Six of 11 R5 isolates that replicated efficiently in PBMC could not replicate in MDM or microglia due to a block in virus entry. CD4 overexpression in microglia transduced with retroviral vectors had no effect on the restricted replication of these virus strains. Furthermore, infection of transfected cells expressing different amounts of CD4 or CCR5 with M-tropic and non-M-tropic R5 isolates revealed a similar dependence on CD4 and CCR5 levels for entry, suggesting that the entry block was not due to low levels of either receptor. Studies using TAK-779 and AMD3100 showed that two highly M-tropic isolates entered microglia primarily via CXCR4. These results suggest that HIV-1 tropism for macrophages and microglia is restricted at the entry level by a mechanism independent of coreceptor specificity. These findings provide evidence that M-tropism rather than CCR5 usage predicts HIV-1 neurotropism.

Most human immunodeficiency virus type 1 (HIV-1) viruses in the brain use CCR5 as the principal coreceptor for entry into a cell. However, additional phenotypic characteristics are necessary for HIV-1 neurotropism. Furthermore, neurotropic strains are not necessarily neurovirulent. To better understand the determinants of HIV-1 neurovirulence, we isolated viruses from brain tissue samples from three AIDS patients with dementia and HIV-1 encephalitis and analyzed their ability to induce syncytia in monocyte-derived macrophages (MDM) and neuronal apoptosis in primary brain cultures. Two R5X4 viruses (MACS1-br and MACS1-spln) were highly fusogenic in MDM and induced neuronal apoptosis. The R5 viruses UK1-br and MACS2-br are both neurotropic. However, only UK1-br induced high levels of fusion in MDM and neuronal apoptosis. Full-length Env clones from UK1-br required lower CCR5 and CD4 levels than Env clones from MACS2-br to function efficiently in cell-to-cell fusion and single-round infection assays. UK1-br Envs also had a greater affinity for CCR5 than MACS2-br Envs in binding assays. Relatively high levels of UK1-br and MACS2-br Envs bound to CCR5 in the absence of soluble CD4. However, these Envs could not mediate CD4-independent infection, and MACS2-br Envs were unable to mediate fusion or infection in cells expressing low levels of CD4. The UK1-br virus was more resistant than MACS2-br to inhibition by the CCR5-targeted inhibitors TAK-779 and Sch-C. UK1-br was more sensitive than MACS2-br to neutralization by monoclonal antibodies (2F5 and immunoglobulin G1b12 [IgG1b12]) and CD4-IgG2. These results predict the presence of HIV-1 variants with increased CCR5 affinity and reduced dependence on CCR5 and CD4 in the brains of some AIDS patients with central nervous system disease and suggest that R5 variants with increased CCR5 affinity may represent a pathogenic viral phenotype contributing to the neurodegenerative manifestations of AIDS.

Astrocyte infection with human immunodeficiency virus (HIV) is considered rare, so astrocytes are thought to play a secondary role in HIV neuropathogenesis. By combining double immunohistochemistry, laser capture microdissection, and highly sensitive multiplexed polymerase chain reaction to detect HIV DNA in single astrocytes in vivo, we showed that astrocyte infection is extensive in subjects with HIV-associated dementia, occurring in up to 19% of GFAP+ cells. In addition, astrocyte infection frequency correlated with the severity of neuropathological changes and proximity to perivascular macrophages. Our data indicate that astrocytes can be extensively infected with HIV, and suggest an important role for HIV-infected astrocytes in HIV neuropathogenesis.

HIV-1 persists in peripheral blood monocytes in individuals receiving highly active antiretroviral therapy (HAART) with viral suppression, despite these cells being poorly susceptible to infection in vitro. Because very few monocytes harbor HIV-1 in vivo, we considered whether a subset of monocytes might be more permissive to infection. We show that a minor CD16+ monocyte subset preferentially harbors HIV-1 in infected individuals on HAART when compared with the majority of monocytes (CD14highCD16−). We confirmed this by in vitro experiments showing that CD16+ monocytes were more susceptible to CCR5-using strains of HIV-1, a finding that is associated with higher CCR5 expression on these cells. CD16+ monocytes were also more permissive to infection with a vesicular stomatitis virus G protein-pseudotyped reporter strain of HIV-1 than the majority of monocytes, suggesting that they are better able to support HIV-1 replication after entry. Consistent with this observation, high molecular mass complexes of apolipoprotein B mRNA-editing enzyme, catalytic polypeptide-like 3G (APOBEC3G) were observed in CD16+ monocytes that were similar to those observed in highly permissive T cells. In contrast, CD14highCD16− monocytes contained low molecular mass active APOBEC3G, suggesting this is a mechanism of resistance to HIV-1 infection in these cells. Collectively, these data show that CD16+ monocytes are preferentially susceptible to HIV-1 entry, more permissive for replication, and constitute a continuing source of viral persistence during HAART.