ART administration reduces neuroinflammation without restoring BDNF signaling in alcohol-administered SIV-infected macaques

Abstract

Objective:

This study examined interactions between simian immunodeficiency virus (SIV), chronic binge alcohol (CBA), and antiretroviral therapy (ART) on growth factor signaling, neuroinflammatory markers, viral loads (VL), and CD4 counts.

Design:

Adult male rhesus macaques were administered CBA (13-14 g EtOH/kg/week) or sucrose (SUC) three months prior to SIV_mac251 infection until study endpoint. At viral setpoint, a subset of CBA/SIV+ and SUC/SIV+ macaques were randomized to receive daily ART (PMPA 20 mg/kg, FTC, 30 mg/kg). Frontal cortex (FC) and basal ganglia (BG) were collected for gene and protein expression.

Methods:

Relationships between brain and plasma VL or CD4 counts were determined using linear regression. Effects of SIV, CBA, and ART on markers of neuroinflammation and brain-derived neurotropic factor (BDNF) signaling were determined by ANOVA and linear regression.

Results:

SIV increased FC and BG neuroinflammatory and glial cell gene expression (CX3CR1, B2M), and reduced FC AKT phosphorylation. CBA decreased FC and BG TrkB phosphorylation, and increased TrkB-FL and SLC1A3 expression in FC and BG, respectively. ART suppressed plasma and brain VL, reduced neuroinflammatory gene expression in FC (IBA1, CX3CR1, and GFAP), and BG (CD74 and CD11ß), and did not restore FC or BG BDNF signaling deficits.

Conclusions:

Results show ART-mediated reduction in VL and neuroinflammatory gene expression, irrespective of CBA administration. ART did not attenuate SIV− and CBA-mediated BDNF signaling deficits, suggesting these deficits, despite effective neuroinflammation suppression, may explain CBA- and SIV-associated neurocognitive deficits. Therapeutics targeting growth factor signaling may be important adjuvants in treating HIV-associated neurocognitive decline.

Introduction

Despite widespread antiretroviral therapy (ART) use, human immunodeficiency virus (HIV)-associated neurocognitive decline (HAND) remains prevalent in over 50% of people living with HIV (PLWH)^1,2. Higher cerebrospinal fluid (CSF) and plasma viral loads and lower CD4 T-cell counts are predictive of neurocognitive impairment^3,4. Preclinical studies have provided evidence for relationships between high CSF viral load and inflammatory cytokines with higher brain viral load and neuropathology^5-7. Additionally, CD4 nadir has been predictive of HAND and other psychological impairments, and reduced cortical and basal ganglia (BG) volumes^8,9. ART effectively reduces CSF and brain viral load and inflammation when initiated 12 days after simian immunodeficiency virus (SIV) infection^10,11 and reduces HIV-related neuropathology^12,13. ART drugs that readily cross the blood brain barrier (BBB) are more effective in reducing CSF viral load, HIV encephalitis (HIVE), and cognitive impairment^14,15.

Neuroinflammation is hypothesized to be the primary mechanism of HIV-associated neuropathology or cognitive impairment^13,16,17. However, it is becoming increasingly recognized that neuroinflammation alone cannot explain the neurological consequences of HIV¹⁸. In a murine model of HIVE, neuropathology persists despite suppressed neuroinflammation and central nervous system (CNS) viral replication¹⁹. Previous work from our group and others suggest disruptions in growth factor expression and signaling likely contribute to HIV neuropathology and cognitive impairments^20-24. Whether ART can protect against neuroinflammation and improve growth factor signaling remains unresolved.

The lifetime history of alcohol use disorder (AUD) in PLWH is 55%, compared to 12-29% in the general population^25-27. Heavy alcohol consumption predicts unsuppressed viral load in PLWH, even those on ART^28-30. These data underscore the potential for alcohol to increase viral load, either through decreased ART adherence, increased viral replication, or impaired viral control, increasing the risk for neurocognitive impairment.

Our group previously showed chronic binge alcohol (CBA) administration unmasks neurocognitive deficits in simian immunodeficiency virus (SIV)-infected, ART-naive macaques³¹. Subsequent studies, in agreement with others^21-23, identified disruptions in growth factor expression and signaling as potential mechanisms contributing to cognitive impairment²⁰. Brain-derived neurotropic factor (BDNF) is a neuroprotective factor regulating synaptic plasticity and cognitive function. Reduced BDNF signaling, including TrkB/ERK and Akt pathways, decreases long-term potentiation and is linked to HAND³². Previously, our lab found no differences in BDNF expression in the frontal cortex (FC) or BG, but did not measure BDNF signaling proteins³³. HIV induces neuroinflammation, neuroinflammatory cytokine release, and glial activation which can dampen BDNF signaling in the brain, impairing mood and cognitive functioning³⁴. In vitro studies have indicated that HIV peptide Tat induces downregulation of BDNF and other CREB-mediated genes³⁵. Furthermore, alcohol consumption can reduce BDNF levels and signaling through TrkB receptors³⁶. Whether disruptions in BDNF signaling by HIV/SIV or alcohol are associated with neuroinflammation and glial activation or are attenuated by ART administration remains to be determined.

This study examined the interactions between SIV, CBA, and ART on growth factor signaling proteins, neuroinflammation, and their relationships with plasma and brain viral load and CD4 T-cell counts. We selected the FC and BG based on our previous findings indicating greater susceptibility of the FC to alterations in growth factor expression and signaling and of the BG to neuroinflammation³³.

METHODS

Animal model and study design

The experiments were approved by the Institutional Animal Care and Use Committee at both Tulane National Primate Research Center (TNPRC) in Covington, LA, and Louisiana State University Health Sciences Center in New Orleans, LA, and followed the National Institutes of Health Guidelines (NIH) for the Care and Use of Experimental Animals. This cohort of animals was used for studies examining glucose metabolism³⁷ and immunologic³⁸ and viral dynamics in response to ART³⁹ during the 14-month experimental timeline described below. Additional experimental details pertaining to this cohort were published elsewhere^37,39. Male rhesus macaques, 4-6 years of age, were assigned into two groups: sucrose (SUC)-administered, SIV-infected (SUC/SIV+) and CBA-administered, SIV-infected (CBA/SIV+). Throughout the 14-month experiment, animals received alcohol daily, totaling 13-14 g EtOH/kg/wk, reaching peak blood alcohol concentrations of 50-60 mM. Three months after initiation of CBA or SUC administration, animals were infected intrarectally with 100 TCID₅₀ of SIV_mac251 (provided by Dr. Preston Marx, TNPRC). Two-and-a-half months after SIV infection, subsets of CBA- and SUC-administered animals began daily subcutaneous ART treatment (CBA/SIV/ART+, n=7, SUC/SIV/ART+, n=7, CBA/SIV/ART−, n=6, SUC/SIV/ART−, n=5), consisting of two nucleoside reverse transcriptase inhibitors (generous gift from Gilead Sciences, Inc.), 20 mg/kg tenofovir (PMPA, 9-R-2-phosphonomenthoxypropyl adenine), and 30 mg/kg emtricitabine (FTC, 2',3'-Dideoxy-5-fluoro-3'-thiacytidine) as previously described³⁹. SIV disease progression was monitored throughout the study using immunological, biochemical, plasma viral load measures, and clinical observations described previously^40-42. Macaques were euthanized 11 months after SIV infection. Euthanasia protocols were in accordance with the Panel on Euthanasia of the American Veterinary Medical Association. Sections of FC and BG were removed, frozen in liquid nitrogen, and stored at −80°C until analysis. Tissue samples from a set of SIV− animals (n=6) free of neuropathology were acquired from TNPRC and used as control tissue.

Viral Load Quantification

Virus particles were isolated from 1 ml of plasma by high-speed centrifugation and RNA prepared using Trizol reagent (Life Technologies, Carlsbad, CA, USA) as described previously⁴³. RNA was extracted from frozen brain tissue with the RNeasy Universal Mini Kit (Qiagen, Valencia, CA, USA), per the manufacturer’s instructions. SIV levels were quantified using TaqMan (Life Technologies) quantitative polymerase chain reaction (qPCR) assay targeting the SIV gag gene and normalized to plasma volume or absolute levels of tissue RPS13 RNA⁴³. The absolute quantity of brain RPS13 RNA was set at 1.3 x 10⁷ RPS13 RNA copies per 10⁶ brain tissue cells. The assay limit of detection is 50 copies SIV RNA per 10⁶ cells or mL of plasma. For statistical analyses, samples with undetectable SIV levels (<50 copies) were set at 25 copies and normalized viral loads were log10-transformed.

Western Blot

Frozen tissue was homogenized by sonication in lysis buffer (320 mM sucrose, 5 mM HEPES, 1 mM EGTA, 1 mM EDTA, 1% SDS, and protease inhibitor cocktail, phosphatase inhibitor cocktails II and III diluted 1:100, (Sigma, St. Louis, MO, USA)), and protein concentration was determined using the DC Protein Assay (Bio-Rad #5000116, Hercules, CA, USA) and Benchmark Plus Microplate Spectrophotometer (Bio-Rad). 20 μg or 40 μg protein samples were electrophoresed on 10%, 4-15%, or 10-20% SDS-polyacrylamide gels (Bio-Rad) using Tris/Glycine/SDS buffer system (Bio-Rad #1610772), and immediately transferred to polyvinylidene difluoride membranes (Millipore, #IPVH00010, Billerica, MA, USA). Membranes were blocked in 5% nonfat milk in 0.1% tween-tris-buffered saline (Bio-Rad, #1706404XTU) for one hour followed by overnight incubation with primary antibody diluted in 2.5% non-fat milk solution. Membranes were washed and incubated in species-specific secondary antibody (1:10,000; Cell Signaling Technology, #7074S, Danvers, MA, USA) for one hour. Following chemiluminescent detection, membranes were stripped (Thermo Fisher Scientific, #46430, Pittsburgh, PA, USA) and reprobed for total protein levels or loading controls. Beta-actin was used as the loading control in the FC. In the BG of the SIV− control group, Beta-actin was observed to range from 0.40 to 1.68 across individual animals (not shown); therefore, we chose to use ERK as the loading control due to its more consistent expression within experimental groups, with levels in the SIV− group exhibiting a tighter range, from 0.88 to 1.10. Densitometry was performed using ImageJ (version 1.49b; NIH) under linear exposure conditions. Values for each animal are expressed as a percentage of the mean control values. Conditions and reagents for western blot analysis are summarized in Supplemental Digital Content, Table S1.

QuantiGene Plex Assay

FC and BG tissue were processed using the Sample Processing Kit (Affymetrix, #QS0106, Santa Clara, CA, USA) prior to gene quantification with the QuantiGene Plex Assay Kit (Affymetrix, #QP1014), both according to manufacturer’s instructions. The QuantiGene Plex assay is a multiplex assay to quantify mRNA expression directly from lysed tissue samples. Powdered brain tissue was weighed and treated with 300 μl of working tissue homogenate solution (Tissue homogenizing solution and proteinase K at a ratio of 60:1) per 10 mg of tissue. The samples were diluted to 16x to ensure fluorescent intensity readings were in the linear range. The tissue weight, ratio of tissue homogenizing solution to proteinase K, and sample dilutions were determined in previous optimization experiments. All incubation steps were performed in the Vortemp56 (Labnet, Edison, NJ, USA). Bead fluorescence was read on the Luminex 100/200 plate reader (Luminex Corporation, Austin, TX, USA) and Luminex xPonent for LX100/LX200 software (Version 3.1, Build 971). Fluorescence readings for each gene were normalized to housekeeping gene (RPS13) fluorescence levels and expressed as fold change compared to SIV− controls.

Markers of glial cells and inflammation

Expression of proinflammatory immune response/cytokine/chemokine (B2M, CCL2, IL1B, ISG15, and IFNA13), anti-inflammatory cytokine (CCL3, CCL5, or IL10. IL6 and IL13), lymphocyte (CD8), monocyte/macrophage (CD74, CD11B, and CD3E), microglia-related (IBA1, CX3CR1) and astrocyte (GFAP) genes were determined in the FC and BG. Accession numbers of macaque genes determined in QuantiGene Plex assay (Supplemental Digital Content, Table S2 shows RefSeq).

Statistical analysis

To determine the contributions of SIV, CBA, ART, and interaction of CBA and ART on gene and protein expression, linear regression and analysis of variance (ANOVA) were used. The model controlled for the presence of brain viral load at necropsy, mean pre-ART plasma viral loads (mean viral loads over weeks 4, 5, 6, 9, and 10 post-infection) and plasma nadir CD4 T-cell count. For the regression model, nadir CD4 T-cell counts, defined as the lowest CD4 count measured, were coded as 0 if they were less than 200 cells/μl or 1 if they were greater than 200 cells/μl⁴⁴. The outcome variables included gene and protein expression levels expressed as values relative to SIV− controls. Implementation was performed in R using the “lm” function for linear regression and the “anova” function for ANOVA (version 3.1.1 (2014-07-10)). Linear regression between mean pre-ART plasma viral load or nadir CD4 count with brain viral load was performed using GraphPad Prism version 5.04 for Windows (GraphPad Software, La Jolla, California, USA). Mean pre-ART plasma viral load, nadir CD4 count, and brain viral load were analyzed as continuous variables. Pre-ART viral load between SUC/SIV+ and CBA/SIV+ macaques was analyzed using the Mann-Whitney test (GraphPad Software). Necropsy plasma and brain viral load comparison among SIV+ animals was performed using two-way ANOVA followed by Bonferroni post-hoc tests (GraphPad Software).

Results

Plasma and brain tissue viral load and CD4 count

Pre-ART plasma viral loads were non-significantly higher in CBA/SIV+ compared to SUC/SIV+ macaques (p = 0.08; Figure 1A). ART resulted in significant suppression of viral load at study endpoint (11 months post SIV-infection) (p < 0.05; Figure 1B) compared to ART-naive macaques. Nadir CD4 T-cell count was significantly lower in ART-naive animals (p < 0.05; Figure 1C). Pre-ART plasma viral load significantly correlated with FC viral load (spearman r = 0.4230, p < 0.05; Figure, 1D). A similar trend was observed in the BG, but did not reach statistical significance (Figure 1D). A significant correlation was detected between CD4 nadir and BG viral load (spearman r = −0.4269, p < 0.05; Figure 1E); however, this correlation did not reach statistical significance in the FC.

Figure 1.

Plasma and brain viral load, CD4 counts, and their correlation in SIV-infected macaques. A) Log plasma viral load (SIV copies/mL) in SUC/SIV+ (open circles) and CBA/SIV+ (black circles) macaques pre-ART (mean viral load from measurements obtained at 4, 5, 6, 9, and 10 weeks post SIV infection). Solid lines denote mean viral load for each group. Dotted line denotes assay limit of detection. B) Log plasma viral load (SIV copies/mL) in SUC/SIV+/ART−, CBA/SIV+/ART−, SUC/SIV/ART+, and CBA/SIV/ART+ at time of necropsy. C) Nadir CD4 T cell count in SUC/SIV+/ART−, CBA/SIV+/ART−, SUC/SIV/ART+, and CBA/SIV/ART+. Nadir value was the lowest CD4 T cell count measured at any time point after SIV infection. D) Correlation of pre-ART viral load in the plasma with SIV viral RNA in the FC and BG. E) Correlation of CD4 T cell nadir with viral RNA in the BG and FC. Viral loads are expressed as the base 10 logarithm of viral copy number. Dotted line indicates the limit of detection (50 copies). SUC/SIV+/ART− (n=5), CBA/SIV+/ART− (n=6), SUC/SIV/ART+ (n=7), and CBA/SIV/ART+ (n=7). $ p < 0.05 in ART+ compared to ART−, FC = frontal cortex, BG = basal ganglia, VL = viral load.

Glial cell markers and inflammation in the FC

SIV infection significantly increased CX3CR1 expression, and ART administration significantly reduced its expression, irrespective of CBA (p < 0.05; Supplemental Digital Content, Table S3). Furthermore, while there was no significant effect of SIV or CBA, ART significantly reduced CCL5, CD8, IBA1, and GFAP expression (p < 0.05; Supplemental Digital Content, Table S3). A statistically significant association was detected between FC viral load and CD4 gene expression (p < 0.05; Supplemental Digital Content, Table S3). A statistically significant reduction in SLC1A3 associated with FC viral load was also detected (p < 0.05; Supplemental Digital Content, Table S3). CD68 expression was significantly greater in animals with positive FC viral load (p < 0.05; Supplemental Digital Content, Table S3). Expression of proinflammatory cytokines and macrophage/monocyte genes CD74, CD68, and CD11B were not significantly altered by SIV, CBA, or ART administration (Supplemental Digital Content, Table S3). The expression of CCL3, IL10, IL13 and CD3e was not significantly different among SIV-infected or CBA-administered animals (Supplemental Digital Content, Table S3).

Glial cell markers and inflammation in the BG

B2M gene expression was significantly greater in SIV-infected macaques compared to SIV− controls, and ART decreased B2M expression irrespective of CBA (p < 0.05; Supplemental Digital Content, Table S4). No significant differences in gene expression of the other inflammatory cytokines or the anti-inflammatory cytokines CCL3, CCL5, or IL10 were detected (Supplemental Digital Content, Table S4).

While there were no significant effects by SIV, gene expression of CD74 and CD11B was significantly lower in the BG of CBA- and SUC-treated animals receiving ART (p < 0.05; Supplemental Digital Content, Table S4). No significant differences were detected in CD3E, CD8, GFAP, IBA1, or CX3CR1 (Supplemental Digital Content, Table S4). CD68, IL6 and IL13 gene expression was below the limit of quantification (Supplemental Digital Content, Table S4). Finally, there was a statistically significant increase in BG SLC1A3 expression associated with CBA, with no effects of ART or SIV (p < 0.05; Supplemental Digital Content, Table S4).

BDNF signaling cascade

To determine the effects of SIV, CBA, and ART on downstream BDNF signaling, phosphorylation of several proteins in the BDNF signaling cascade, including TrkB, AKT, and ERK, were determined in the FC and BG.

TrkB phosphorylation

Phosphorylation of TrkB receptors was significantly reduced in both the FC (p < 0.05; Fig 2A) and BG (p < 0.05; Fig 2B) of CBA/SIV+ macaques, irrespective of ART administration. Post hoc analyses demonstrated CBA-administration decreased TrkB phosphorylation fivefold in the FC (Figure 2A) and twofold in the BG (Figure 2B) irrespective of ART, compared to levels in SIV− macaques (p < 0.05, Mann-Whitney U = 44.50). Total full-length TrkB (TrkB-FL) levels were significantly greater in the FC of CBA/SIV+ macaques (p < 0.01; Figure 2C) but not changed in the BG (Figure 2D) relative to SIV− animals. Levels of truncated TrkB (TrkB-T, lacking the phosphorylation domain) were unchanged in the FC and BG between groups (Figure 2E, F), although SIV reduced TrkB-T in the BG (p < 0.05; Figure 2F).

Figure 2.

TrkB phosphorylation in the FC (left panel) and BG (right panel) of SIV-infected macaques. Relative levels of pTrkB in FC (A) and BG (B); Full length TrkB (TrkB-FL) levels in the FC (C) and BG (D); truncated TrkB (TrkB-T) levels in the FC (E) and BG (F) of SUC/SIV+/ART−, CBA/SIV+/ART−, SUC/SIV/ART+, and CBA/SIV/ART+ macaques. Data are presented as fold-change relative to SIV− (indicated by dashed line) for individual sucrose- (open circles) and alcohol-treated (black circles) SIV-infected animals without (ART−) and with (ART+) treatment. Solid lines represent mean of each group. G) Representative blot of FC pTrkB, TrkB-FL, TrkB-T, and Beta-actin. H) Representative blot of BG pTrkB, TrkB-FL, TrkB-T, and ERK. * p < 0.05 for main effect of alcohol. # p < 0.05 for main effect of SIV infection. FC = frontal cortex, BG = basal ganglia. 1) SIV− (n=6), 2) SUC/SIV/ART− (n=5), 3) CBA/SIV/ART− (n=6), 4) SUC/SIV/ART+ (n=7), 5) CBA/SIV/ART+ (n=7)

AKT phosphorylation

AKT phosphorylation in the FC was significantly reduced by SIV (p < 0.05; Figure 3A), but not altered by CBA or ART (Figure 3A). In contrast, there were no significant differences in AKT phosphorylation due to SIV, CBA, or ART in the BG (Figure 3B). Additionally, total levels of AKT were not different in the FC (Figure 3C) or BG (Figure 3D) due to SIV, CBA, or ART.

Figure 3.

AKT phosphorylation in the FC (left panel) and BG (right panel) of SIV-infected macaques. Relative levels of phosphorylated (pAKT) in the FC (A) and BG (B) and total AKT in FC (C) and BG (D) of SUC/SIV/ART−, CBA/SIV/ART−, SUC/SIV/ART+, and CBA/SIV/ART+ macaques. Data are represented as fold-change relative to SIV− (indicated by dashed line) for individual sucrose- (open circles) and alcohol-treated (black circles) SIV-infected animals. Solid lines represent mean of each group. E) Representative blot of FC pAKT, AKT, and β-actin. F) Representative blot of BG pAKT, AKT, and ERK. # p < 0.05 for main effect of SIV infection. FC = frontal cortex, BG = basal ganglia, 1) SIV− (n=6), 2) SUC/SIV/ART− (n=5), 3) CBA/SIV/ART− (n=5), 4) SUC/SIV/ART+ (n=6), 5) CBA/SIV/ART+ (n=7).

ERK phosphorylation

There were no significant differences in phosphorylated or total ERK in SIV, CBA, or ART-treated macaques in the FC (Figure 4A, C). Phosphorylation of ERK was significantly increased in the BG of SIV-infected macaques (p < 0.05; Figure 4B), while total ERK levels remained unchanged (Figure 4B).

Figure 4.

ERK phosphorylation in the FC (left panel) and BG (right panel) of SIV-infected macaques. Relative levels of phospho-ERK (pERK) in the FC (A) and BG (B) and total ERK levels in the FC (C) and BG (D) of SUC/SIV/ART−+, CBA/SIV/ART−+, SUC/SIV/ART+, and CBA/SIV/ART+ macaques. Data are represented as fold-change relative to SIV− (indicated by dashed line) for individual sucrose- (open circles) and alcohol-treated (black circles) SIV-infected animals. Solid lines represent mean of each group. E) Representative blot of FC pERK, ERK, and Beta-actin. F) Representative blot of BG pERK and ERK. ERK was used as the loading control for basal ganglia samples. FC = frontal cortex, BG = basal ganglia, 1) SIV− (n=6), 2) SUC/SIV/ART− (n=5), 3) CBA/SIV/ART− (n=6), 4) SUC/SIV/ART+ (n=7), 5) CBA/SIV/ART+ (n=7)

Discussion

The present study determined the alterations in markers of neuroinflammation and BDNF signaling associated with SIV, CBA, and ART, and the relationship between peripheral markers of disease and viral load in the FC and BG during the asymptomatic stage of SIV infection. ART effectively reduced plasma viral load (Figure 1B), maintained CD4 T-cell counts (Figure 1C), and normalized some but not all markers of neuroinflammation, irrespective of CBA administration (Supplemental Digital Content, Tables S3 and S4). HAND impairs several aspects of cognitive functioning, including executive function, working memory/attention, fine motor skills, and information processing speed⁴⁵. The FC plays a significant role in executive function, working memory, and information processing, making it a region of interest when investigating HAND symptomatology. Before ART implementation, BG inflammation was a hallmark feature of HIVE, and represents a major site of viral replication and neuropathology, leading to significant atrophy that may correspond to motor disturbances^46-49. Even after ART implementation, BG inflammation remains prevalent among PLWH, particularly in individuals not adherent to ART or over the age of 50^50,51. Furthermore, in PLWH, both the FC and BG display reductions in markers of neuronal activity and integrity, indicative of neuroinflammation and injury^52,53, which might also impair communication between these two areas to produce HAND symptoms.

In the FC of both CBA- and SUC-treated animals, ART reduced expression of genes associated primarily with CNS resident microglia and astrocytes – IBA1, CX3CR1, and GFAP – as well as CCL5, a chemokine responsible for leukocyte recruitment (Supplemental Digital Content, Table S3). In the BG, ART did not significantly reduce microglia or astrocyte gene expression as seen in the FC, but significantly reduced expression of CD74 and CD11B, markers of microglial activation (Supplemental Digital Content, Table S4). Surprisingly, ART did not ameliorate the suppression of BDNF signaling proteins in either the FC or BG (Figures 2-4). These findings suggest that despite ART-mediated attenuation of viral load and some neuroinflammation, the inability to normalize levels and activity of BDNF signaling proteins may reflect an enhanced susceptibility to neuronal damage and neurocognitive impairment in SIV-infected macaques. These findings agree with clinical studies reporting persistent HAND despite ART adherence, suggesting more research is needed to develop additional therapeutics to reduce neuroinflammation and HAND in PLWH^9,54. Similar results have been reported in which ART reduces CNS viral load and inflammation when initiated 12 days after SIV infection^10,11. In the FC, ART reduced expression of CD8, a T-cell marker, which could indicate decreased CD8 T-cells^55,56. Clinical studies have shown that subjects with HIVE have increased CD8 T-cell infiltration in both FC and BG⁵⁵, suggesting ART may be protective against, or delay the onset of HIVE. ART also reduced CCL5, a HIV-suppressive and neuroprotective factor, in the FC (Supplemental Digital Content, Table S3), which has also been associated with markers of neurodegeneration in CSF from PLWH^57-59.

The microglia-associated gene CX3CR1 was increased (1.5-fold) by SIV in the FC, and reduced by ART, as was the microglial marker IBA1 (Supplemental Digital Content, Table S3). However, neither were reduced by ART in the BG (Supplemental Digital Content, Table S4). Potential reasons could include differences in other signaling factors or BBB permeability in these brain regions, as opposed to differential effects of ART⁶⁰. This interpretation is consistent with studies indicating greater levels of inflammation in the BG compared to FC⁶⁰ and increased BBB permeability near the BG in those with HIV-associated dementia⁶¹. Together, the data on FC and BG inflammatory gene expression indicate that ART initiated at viral setpoint reduces some, but not all, markers of neuroinflammation.

FC CD4 gene expression negatively correlated with FC viral load (Figure 1E). This pattern could result from greater trafficking of peripherally infected CD4 T-cells to the CNS, as CD4 T-cells are the primary target of SIV infection⁶². This seems to be more likely than the alternative interpretation that viral RNA is driving the expression of CD4 in the brain, especially given the lack of an association between brain viral load and CNS glial gene expression (IBA1, CX3CR1, GFAP).

Growth factor signaling protein suppression due to CBA administration was observed during the asymptomatic stage and may be relevant to previous findings from our lab showing cognitive deficits prior to end-stage disease⁶³. The significantly reduced phosphorylation of TrkB in the FC of CBA/SIV+ compared to VEH/SIV+ animals supports our previous findings in CBA/SIV+ animals at end-stage SIV disease³³. Importantly, ART did not ameliorate the suppressed phosphorylation of TrkB in either the FC or BG (Figure 2A, B), despite reducing inflammatory gene expression (Supplemental Digital Content, Tables S3 and S4). Although our study design did not allow for direct brain region comparisons, the magnitude of reduced TrkB phosphorylation compared to SIV− animals was significantly greater (fivefold lower than SIV−) in the FC (Figure 2A) compared to BG (twofold lower than SIV−) (Figure 2B) of CBA/SIV+ macaques. These differences in TrkB phosphorylation and presumptive activity may indicate greater susceptibility of the FC to alcohol-induced damage within the context of SIV. The findings presented here also agree with reports that acute alcohol administration reduces TrkB phosphorylation⁶⁴ and provide support for the reported shift from subcortical to cortical cognitive deficits that have occurred during the ART era⁶⁵. Suppressed phosphorylation of TrkB detected at study endpoint could also enhance vulnerability to acute alcohol administration, which could account for the unmasking of cognitive deficits previously observed⁶³. Although no behavioral studies were incorporated into the current design, previous studies from our group have investigated the independent and synergistic consequences of CBA and SIV on neurocognitive behaviors. Winsauer et al.³¹ discovered significant interactions between SIV and alcohol to disrupt operant performance in a food-reinforced task. Unfortunately, similar studies might have confounded metabolic studies that were the primary focus in this cohort of animals (see Ford et al.³⁷). Ongoing studies in our current macaque cohorts are collecting additional cognitive behavioral data (novel object recognition tests) in SIV− and CBA-treated animals.

Previous studies from our group aimed to determine whether administration of CBA would attenuate the effectiveness of uninterrupted ART therapy in suppressing viral load in SIV-infected macaques (Molina et al.³⁹). Our findings indicated that CBA does not significantly decrease the effectiveness of ART to suppress viral loads and does not interact with ART to produce toxicity. In this same cohort of animals, Ford et al.³⁷ determined whether CBA or ART would independently, or in combination, affect insulin-glucose dynamics in SIV-infected macaques. Those studies discovered reduced circulating adiponectin and resistin levels, as well as a reduced acute insulin response to glucose in CBA-treated animals, irrespective of ART. The brains collected at necropsy from that study were used for the analyses presented in the current manuscript, and the time elapsed from collection of brains to analysis was approximately 5 years. Brain tissues were stored at −80°C under conditions where we have previously been successful in measuring changes in mRNA and proteins/phosphoproteins for up to a decade. We are confident that passage of time did not impact the results of this study, as time-matched appropriate controls were included in the analyses.

In summary, SIV infection decreased expression of growth factor signaling proteins in the FC, which were exacerbated by CBA administration, and not attenuated by ART. ART did, however, attenuate expression of microglial markers of neuroinflammation in the FC and monocyte/macrophage markers of neuroinflammation in the BG. These findings suggest that while ART may be effective in reducing some of the neuroinflammation due to SIV and CBA, it is not sufficient to attenuate the BDNF signaling deficits. This lack of attenuation in growth factor signaling may explain the persistence of HAND despite widespread ART use. These results support the need for continued research into strategies to prevent HAND, including the potential to target neuroinflammation and growth factor signaling as new therapies are developed.