Molecular and pathologic insights from latent HIV-1 infection in the human brain

METHODS

RESULTS

We analyzed 32 cases, whose demographic, clinical, virologic, and pathologic characteristics are presented in table 1. Based on the levels of HIV-1 DNA, RNA, and p24, 12 cases were classified as HIV+ controls, with no detectable HIV-1 DNA, RNA, or p24; 10 cases were designated as latent HIV+, with high levels of HIV-1 DNA and no detectable RNA or significant p24 (figure 1, A–C). In contrast, those cases classified as HIVE (n = 10) displayed high levels of HIV-1 DNA, RNA, and p24 (figure 1, A–C). Productive HIV infection in the latter group was accompanied by the presence of microglial nodules, astrogliosis, and myelin pallor (not shown). At the pathologic level, latent HIV cases displayed some rarefaction of the neuropil accompanied by astrogliosis, while in HIV+ controls there were no histopathologic alterations (data not shown). As shown in table 1, the proportion of subjects prescribed HAART was similar in the control and latent groups (50% vs 44%, respectively), whereas fewer subjects on the HIVE group received HAART (29%). Sustained virologic suppression on HAART was also similar in the control and latent groups (60% vs 67%), in contrast to the HIVE group, where none of the subjects showed evident viral suppression. The groups did not differ significantly with respect to age, estimated duration of HIV infection, year of death, CD4 nadir, current CD4, or plasma viral load at the visit closest to death.

Among the 32 cases, antemortem neurocognitive diagnosis was available for 23 subjects, representing 75% of HIV+ controls (9 of 12), 80% of HIV+ latent cases (8 of 10), and 60% of HIVE cases (6 of 10) (table 1). Severe neurocognitive impairment was only reported for cases with HIVE. Mild to moderate neurocognitive impairment occurred in 6/8 (75%) of the latent cases as compared to 5/9 (56%) of the control cases. The relatively high rate of mild impairment in the control cases may reflect brain injury that occurred before HAART became available to these subjects, who were typically infected for a decade or more.

To understand the anatomical correlates to latent HIV infection, we analyzed neurodegeneration by confocal microscopy (figure 1, D–K). Latent HIV cases displayed considerable loss of MAP2 immunoreactive dendrites (figure 1, D and H) and synaptophysin immunoreactive nerve terminals in the frontal cortex (figure 1, E and I), similar to the alterations observed in HIVE cases (figure 1, D, E, H, and I). Moreover, astrogliosis (figure 1, F and J) and microgliosis were detected in the latent HIV group when compared to controls (figure 1, G and K). As expected, HIVE cases displayed the highest levels of GFAP (figure 1, F and J) and Iba1 (figure 1, G and K) accompanied by the presence of p24 immunostained microglial cells (no p24 immunoreactivity was detected in HIV controls or in latent HIV cases, not shown).

Silencing of HIV-1 transcription in microglial cell cultures is mediated by the transcriptional regulator BCL11B, which recruits a multienzymatic chromatin-modifier complex to repress gene expression.¹¹ We therefore analyzed the levels of BCL11B and other components of the chromatin-remodeling complex in human brain samples. BCL11B protein was significantly higher in the latent group than in control and HIVE cases (figure 2, A and B). Consistently, BCL11B immunoreactivity was elevated in neuronal (figure 2, C and D) and glial cells (figure 2, C and E) in frontal cortex and white matter from latent HIV cases. Double labeling analysis confirmed that high levels of BCL11B were detected in microglial (figure 2, F and G) and astroglial cells (figure 2, H and I) only in the latent HIV cases, where p24 was not detectable, supporting the role of BCL11B in silencing HIV-1. On the contrary, in HIVE cases, p24-positive glial cells were negative for BCL11B (figure 2, G and I). No apparent changes in BCL11B mRNA levels were detected in the studied groups (figure e-1 on the Neurology^® Web site at www.neurology.org), suggesting that increased BCL11B associated with latency might be regulated at the level of protein translation or stability. To evaluate whether alterations of BCL11B in the brain might translate into peripheral fluids, similar analyses were performed in CSF samples. CSF from latent HIV cases showed increased BCL11B in comparison to the HIVE group (figure 2, J and K).

We next investigated the levels of other components of the HIV-silencing complex (figure 3). HP1α, MeCP2, and HDAC1 were significantly increased in latent cases in comparison to HIVE samples (figure 3, A, B, D, and F), while similar levels of HDAC2, SP1, and SUV39H1 were observed among all groups (figure 3, A, C, E, and G). As reported for BCL11B, upregulation of chromatin modifiers seems to operate at the protein level, without significant changes in gene expression (figure e-1).

BCL11B is an important transcriptional regulator in the brain.^14,18 Since transcription of several genes, including those involved in apoptosis and cell death, is altered in HIV-1 infected brains,¹⁹ we investigated whether this functional group of genes might represent a cluster of BCL11B targets by in silico analysis of the presence of BCL11B consensus sites.²⁰ Interestingly, 18 out of 25 candidate genes contained at least one canonical binding site for BCL11B in their proximal promoter region (table e-1). Among them, CEBP, CD74, CXCR4, SOX9, CFLAR, and PDCD6 have the highest representation of sites, and therefore are the most likely targets for BCL11B-mediated regulation. Analysis of their relative abundance in frontal cortex samples by qPCR revealed that CEBP was reduced (figure 4A) and CD74 and CXCR4 were elevated in the HIV+ latent group (figure 4, B and C). Since these genes are involved in proinflammatory responses, we extended our analysis to include prominent inflammatory markers. We detected a decrease in IL-1 transcription (figure 4D), while significant upregulation of IL-6 was observed in latent HIV cases (figure 4E). Furthermore, CD74 and CXCR4 proteins were increased in the HIV latent group and, to a lesser extent, in HIVE cases when compared to HIV controls (figure 4, G–I). Similarly, CD74 and CXCR4 immunoreactivity were elevated in neurons and microglial cells from latent HIV and HIVE cases, in comparison to control samples (figure 4, J, L, and M). IL-6 immunoreactivity was also elevated in the astroglial cells in the latent group (figure 4, J and K).

DISCUSSION

We investigated the neuropathologic and molecular alterations associated with CNS latent HIV-1 infections, defined as cases displaying proviral DNA in the brain with no productive infection. Our results suggest that patients with latent HIV display cognitive deficits, neurodegenerative alterations, and neuroinflammatory changes, supporting the notion that viral latency is not innocuous for the brain and might represent a distinct condition with particular pathologic features. Although HIV+ latent and HIVE cases displayed similar clinical and neurodegenerative traits, the extent of the cognitive and pathologic alterations was greater in the HIVE group. At the molecular level, latent cases showed increased BCL11B protein levels, a transcriptional regulator that mediates silencing of the integrated provirus, while this epigenetic modulator of HIV infection was reduced in HIVE cases.

Previous studies suggest that CNS is an important reservoir for HIV where proviral DNA and productive infections can be detected.^8,21 Although HAART greatly benefited infected patients, it is unable to cure the infection due to the persistence of long-lived transcriptionally silent proviruses that can integrate into the host genome in subpopulations of susceptible cells, generating viral reservoirs in peripheral blood and organs. The brain, in particular, represents a sanctuary for HIV-1 latency, where the provirus can persist due to the variable and poor penetration of antiretroviral drugs.^22,23

While viral replication was considered necessary for inflammation and neurodegeneration, recent studies showed a high prevalence of HAND even in long-standing aviremic patients.²⁴ We provide here a novel explanation for brain damage in the absence of viral replication, mediated by deregulation of BCLL1B that contributes to sustain HIV-1 latency and might also alter the expression of inflammation-related genes.

Neuroinflammatory responses play important roles in neurodegeneration in HIV+ latent and HIVE groups, but based on BCL11B analysis, the pathways leading to inflammation might differ among groups. In HIVE, it has been proposed that HIV-infected monocytes cross the blood–brain barrier early on during infection and subsequently initiate a cascade of inflammatory mechanisms through the release of progeny virus/viral protein and cytokines/chemokines.^25,26 Migrating infected monocytes express IL-1, IL-6, tumor necrosis factor (TNF)–α, tumor growth factor–β, GM-CSF, and prostaglandin E₂,²⁷ which bind glia receptors and activate additional inflammatory genes through a positive feedback mechanism. In addition to neuroinflammation mediated by physiologic cellular molecules, monocyte-expressed HIV proteins, vpr, tat, nef, and gp120 activate interferon, apoptosis, and MAPK pathways in microglia and astrocytes contributing to chronic inflammation and neurotoxicty.^28–31 In contrast, we propose that in patients with latent HIV infections, increased levels of BCL11B might directly trigger neuroinflammatory cascades by transcriptional deregulation of CD74, IL-6, and TNF-α. Although we cannot exclude that the cases we classified here as HIV+ latent could have previously been HIVE manifestations in which effective HAART suppressed viral replication, this possibility seems unlikely, since 50% of the subjects of the latent group never received HAART and the remaining only followed the treatment intermittently, favoring the idea that absence of viral RNA resulted from bona fide silencing mechanisms.

HIV-1 latency requires a successful integration event, where the host chromatin environment plays a key role.⁸ In microglial cells, BCL11B assembles a multienzymatic complex by recruiting SP1, HP1α, HDAC1, HDAC2, and SUV39H to the viral LTR region, and establishing a heterochromatic environment at the viral insertion site silencing HIV-1 transcription.¹¹ In the present work, we investigated for the first time the status of BCL11B and associated epigenetic factors in the human brain in the context of HIV-1 infection. We report increased BCL11B, HP1α, MECP2, and HDAC1 in latent HIV infection, most probably due to posttranscriptional mechanisms, including protein translation or turnover. BCL11B is a common regulator of gene transcription in the brain¹⁴ implicated in the negative regulation of BDNF signaling, which is altered in several neurodegenerative disorders.¹⁸ Our results suggest that recruitment of BCL11B to the viral insertion sites during latency might alter the transcription of its target proinflammatory genes, triggering chronic inflammatory responses. We suggest that the presence of silent provirus in the brain might trigger chronic pathology and might be associated with neurocognitive impairment in latently infected patients.

Diagnosis of HIV-1 infection in the CNS can be achieved by detection of HIV in the CSF; still, the presence of latent viral forms can only be corroborated postmortem, hindering efforts to detect resident virus earlier to determine possible therapeutic interventions. An important outcome of the present work is the observation of abnormal BCL11B levels in the CSF of latently infected patients, in correlation with alterations in the brain. Further investigation will determine the potential use of BCL11B as biomarker of silent HIV-1 infections in the CNS.

The recent discovery of involvement of histone methylases and deacetylases in HIV latency fueled research aimed to reactivate latent reservoirs and to facilitate eradication by targeting actively replicating virus.³² Valproic acid and trichostatin A were unable to reactivate HIV in latently infected primary CD4 T cells,³³ raising concerns about their potential use in neuroAIDS. Modulation of BCL11B could represent an alternative therapeutic strategy for eradication of reservoirs.

Taken together, our results suggest that latent HIV-1 infection in the CNS might be associated with increased levels of chromatin modifiers, including BCL11B. Alteration of these epigenetic factors might activate proinflammatory cascades associated with neurodegeneration and neurocognitive impairment. BCL11B and other silencing factors might represent novel targets to ameliorate the effects of HIV-1 infections in the CNS.

Supplementary Material

AUTHOR CONTRIBUTIONS

Dr. Desplats and Dr. Masliah designed the study. Dr. Smith performed the DNA and RNA studies. Dr. Everall, Dr. Letendre, Dr. Ellis, Dr. Cherner, and Dr. Grant clinically assessed the patients. Dr. Desplats, Dr. Masliah, Dr. Dumaop, Dr. Smith, and Dr. Adame performed the experimental work. Dr. Desplats and Dr. Masliah analyzed the data and wrote the manuscript. Dr. Desplats, Dr. Masliah, Dr. Grant, Dr. Ellis, and Dr. Letendre contributed to the discussion.

STUDY FUNDING

Supported by NIH grants NS057096, AG03197, AG5131, MH62962 (to E.M.), and MH076681, MH79881, MH45294, MH5974, MH58164, and DA12065, California NeuroAIDS Tissue Network U01 MH83506 (to I.G.). This work was also funded by the HIV Neurobehavioral Research Center (HNRC) CSPAR Developmental Core Grant HNRC-859 (P.D.). The HNRC is supported by Center award MH62512 from NIMH.

DISCLOSURE

The authors report no disclosures relevant to the manuscript. Go to Neurology.org for full disclosures.