Accelerated neurodegeneration of basal forebrain cholinergic neurons in HIV-1 gp120 transgenic mice: Critical role of the p75 neurotrophin receptor

Abstract

Human Immunodeficiency Virus-1 (HIV) infection of the brain induces HIV-associated neurocognitive disorders (HAND). The set of molecular events employed by HIV to drive cognitive impairments in people living with HIV are diverse and remain not completely understood. We have shown that the HIV envelope protein gp120 promotes loss of synapses and decreases performance on cognitive tasks through the p75 neurotrophin receptor (p75^NTR). This receptor is abundant on cholinergic neurons of the basal forebrain and contributes to cognitive impairment in various neurological disorders. In this study, we examined cholinergic neurons of gp120 transgenic (gp120tg) mice for signs of degeneration. We observed that the number of choline acetyltransferase-expressing cells is decreased in old (12–14-month-old) gp120tg mice when compared to age matched wild type. In the same animals, we observed an increase in the levels of pro-nerve growth factor, a ligand of p75^NTR, as well as a disruption of consolidation of extinction of conditioned fear, a behavior regulated by cholinergic neurons of the basal forebrain. Both biochemical and behavioral outcomes of gp120tg mice were rescued by the deletion of the p75^NTR gene, strongly supporting the role that this receptor plays in the neurotoxic effects of gp120. These data indicate that future p75^NTR-directed pharmacotherapies could provide an adjunct therapy against synaptic simplification caused by HIV.

1. Introduction

Human immunodeficiency virus-1 (HIV)-associated neurocognitive disorders (HAND) are the collection of neurological and neuropsychiatric impairments present in people living with HIV (PLWH). Cognitive decline has been one of the more common complications among PLWH despite the use of antiretroviral therapy (Heaton et al., 2011; Sacktor, 2018). The cognitive decline observed in PLWH are likely the result of reduced cerebral white/grey matter integrity (Sanford et al., 2018) and neuroinflammation (Buckley et al., 2021) or comorbidities (Smail and Brew, 2018) and is similar to neurodegenerative diseases of aging, such as Parkinson’s (Fama et al., 2023) and Alzheimer’s diseases (AD) (Gonzalez et al., 2023). Whether persistent cerebral inflammation alone causes cognitive alterations in HAND remains inconclusive.

We have previously shown that HIV reduces the levels of brain-derived neurotrophic factor (BDNF) in human lymphocytes (Avdoshina et al., 2011) as well as in the brains (Bachis et al., 2012). BDNF is a neurotrophin (NT) crucial for synaptic plasticity of adult neurons [reviewed in (Arancio and Chao, 2007)]. The decline in BDNF levels observed in HAND is attributed to the ability of the HIV and its envelope protein gp120 to target BDNF processing (Bachis et al., 2012). In facts, both HIV and gp120 promote the accumulation of BDNF precursor or proBDNF (Bachis et al., 2012), which induces neuronal apoptosis when it binds to the p75 neurotrophin receptor (p75^NTR) (Teng et al., 2005). Activation of this receptor by the proNTs contributes also to memory loss (Buhusi et al., 2017; Woo et al., 2005), age-related brain atrophy (Cade et al., 2022), as well as HAND (Bachis et al., 2012; Killebrew et al., 2022).

We have previously shown that an established mouse model of HAND, which constitutively expresses gp120 (gp120tg) (Toggas et al., 1994), displays decreased hippocampal synapses when compared to wild type (WT) when they are ~8 month-old (Bachis et al., 2016). Most importantly, the biochemical, histological, and behavioral aberrations in this model are mitigated when gp120tg mice are crossed with mice lacking p75^NTR (Speidell et al., 2019; Speidell et al., 2020), indicating that p75^NTR plays a critical role in gp120-driven neural injury. This suggestion is supported by data from other investigators showing that LM11A-31, a p75^NTR antagonist (Knowles et al., 2013), prevents gp120-mediated cell death of cortical neurons in vitro (Meeker et al., 2016). Nevertheless, the extent to which neuronal vulnerability to gp120-related neural injury is conserved in p75^NTR-rich neural substrates critical for higher cognitive functions remains under-investigated.

P75^NTR is abundant in the basal forebrain cholinergic neurons (BFCNs) (Holtzman et al., 1995; Yeo et al., 1997). Their loss is an early and central event in pathogenesis of aging or aging-related neurodegenerative disorders [reviewed in (Chen and Mobley, 2019)]. Accumulating evidence from both animal models and human reports demonstrate that BFCNs and their connections (De Rosa et al., 2004; Fujii et al., 2002) serve critical roles in at least three cognitive domains known to be affected in PLWH (Cole et al., 2017). We hypothesized that gp120 may similarly act on the forebrain proNT environment to promote p75^NTR activation and thereby evoke death of this neuronal population. A loss of BFCNs and disruption of BFCN-modulated neuronal events may explain the impairments on complex cognitive tasks in gp120tg mice and in HAND individuals more broadly.

In the present study, we assessed the extent to which BFCNs are compromised by gp120, through a variety of histological, biochemical, and behavioral approaches. Our data suggests that gp120 expression promotes an age-dependent atrophy and eventual loss of the BFCN population.

2. Materials and Methods

2.1. Animals.

All four genotypes within this study were created from the intercrossing of p75^NTR+/− with gp120tg mice in our vivarium as previously described (Speidell et al., 2019; Speidell et al., 2020). The p75^NTR−/− mice used to generate our colony has a targeted knockout of exon III of the p75^NTR allele and has been previously characterized (Lee et al., 1992). In total, we used 28 WT, 23 p75^NTR+/, 27 gp120tg, and 25 p75^NTR+/−gp120tg mice in this study. Both males and females were used and were counterbalanced among our experimental groups. Homozygous p75^NTR knockouts display marked impairments in somatosensation and peripheral nervous system function (Lee et al., 1992) and were not used in this study. Animal genotypes were confirmed through an outsourced genotyping service (Transnetyx, Inc., Cordova, TN) from tail snips taken at time of weaning and euthanasia. Mice were housed with their littermates approximately 2–4 mice per cage in the Georgetown University Department of Comparative Medicine and were used for this study when were either 3–5 or 12–14-month-old. All rodents were kept on a 12-hour light-dark cycle beginning at 6:00 am and ending at 18:00 pm. Purina rodent chow (cat.no. 5001, Purina Animal Nutrition LLC., Summit, MO) and Hydropac purified water (Lab Products Inc., Rockville, MD) were available ad libitum throughout the entire study. No animals were excluded for health reasons during this study. All procedures were approved by the Institutional Animal Care and Use Committee at Georgetown University (protocol number 2016–1188).

2.2. Euthanasia and tissue processing.

At the appropriate age, animals were anesthetized with Ketamine/Xylazine (80/10 mg/kg, i.p.) mixture and them intracardially perfused with ice-cold phosphate buffered saline (PBS) via a 25G needle placed in the left ventricle. The whole brains were extracted, and immediately post-fixed in 4% paraformaldehyde in PBS for 24 hr and then placed in a 30% sucrose in PBS solution until saturated. Brains assigned to biochemical endpoints (western blotting) were instead chilled in PBS and microdissected by coarse sectioning with a 1 mm brain block (cat.no. RBMS-200C, Kent Scientific Inc., Torrington, CT). The extracted neural tissue was snap frozen on dry ice and stored at −80°C.

2.3. Immunohistochemistry and c-Fos immunoreactivity analysis.

Brains from above were placed on a sliding microtome (Thermo Fisher Scientific, Waltham, MA) and sectioned at 30μm. Free-floating sections were stored in cryoprotectant at −25°C until used for immunostaining. Anti-ChAT nickel-enhanced peroxidase immunohistochemistry (IHC) was performed using Vector Laboratories Vectastain ABC HRP (cat. no. PK4000) and 3,3’-Diaminobenzidine (DAB) Peroxidase Substrate kits (cat. no. SK4100, Vector Laboratories Inc., Burlingame, CA) kits per manufacturer’s instructions. Target proteins were probed for using goat anti-ChAT (1:1000, cat. no. AB144P, Chemicon, Japan) antibody. Immune complexes were detected with a biotinylated rabbit anti-goat IgG (cat.no. B7014, MilliporeSigma, Burlington, MA) secondary antibody. DAB-IHC Sections were dehydrated in sequential one-minute 50–70-95–100% ethanol incubations, cleared in HemoDe xylene substitute, and coverslipped with resinous media (Permount, Fisher Scientific Inc., Hampton, NH).

For fluorescent IHC, we used rabbit anti-c-Fos antibody (1:1000, cat.no. 2250S, Cell signaling Technologies, Edina, MN) with aqueous coverslipping media (cat.no. H-1200–10, Vectashield Antifade with DAPI, Vector Laboratories Inc.). For automated detection of c-Fos positive cells, we employed Rewire AI’s (Rewire AI, Portland, OR) cytoplasmic cell-counting model. The cell detection reduces false positives from the observer such that only cells designated by both the observer and model are highlighted. Cells in the dorsal hippocampal formation were considered c-Fos positive when exhibited large and round immunoreactive puncta within the pyramidal or granule cell layers and when they appeared much brighter than the surrounding signal. For c-Fos semi-quantitation, we followed a simple processing and analysis procedure within ImageJ (NIH, Bethesda, MD) involving histogram normalization, thresholding, and area fraction measurement of the positive signal.

2.4. Stereology and Stereological Analysis.

Serial sections through the basal forebrain stained for ChAT were assessed using the optical fractionator probe within Stereo Investigator^® 9.0 (MBF Biosystems Inc., Williston, VT). We restricted our analysis to the medial septal nucleus (MS) and the vertical limb of the diagonal band (VDB) in four serial sections, 180μm apart from one another. We adapted a protocol with minor modifications from Xie and colleagues for the stereological estimation of the MS/VDB BFCN population (Xie et al., 2019). Briefly, a motorized stage mounted on a Zeiss Axiophot microscope was employed to view and image the regions of interest in the brightfield. The combined MS/VDB region was traced in Stereo Investigator using a 2.5X objective before cell counting with a 63X oil immersion objective. To estimate the cumulative ChAT+ cell population, we used the following equation for total number of particles counted in the optical fractionator workflow, where $Q^{-}$ = particles (cells) counted, $t$ = software-measured section mounted thickness, $h$ = counting frame height (35μm), $as\int$ = area sampling fraction [(35μm × 35μm)/(80μm × 80μm)], and $ss\int$ = section sampling fraction (1/6).

$$N=\sum Q^{-}\cdot \frac{t}{h}\cdot \frac{1}{as\int }\cdot \frac{1}{ss\int }$$

Gunderson’s coefficient of error (CE) was generated alongside these counts as a measure of precision. In parallel with the optical fractionator workflow, we also adapted the vertical nucleator probe within Stereo Investigator to estimate the average cross sectional area of eligible cells located within the dissector as previously described (Gundersen et al., 1988). The cross-sectional area of the immunostained cell was therefore estimated by the equation $a=\pi l^2$, with $l$ defined as the average distance of the lines from the center to the edge of the cell.

2.5. Western Blot Analysis.

The basal forebrain tissue from 12–14-month-old mice was homogenized in a mixture of RIPA buffer (cat. no. 20–188, MilliporeSigma) containing protease-phosphatase inhibitors (cat. no. 78442, Thermo Fisher Scientific). A BCA Protein Assay Reagent Kit (cat.no. 23225, Thermo Fisher Scientific) was used to determine protein content. Proteins were separated on a NuPAGE 4–12% Bis-Tris Gel (cat. no. NP0335, Invitrogen Inc.) and transferred to a PVDF membrane (cat. no. 1620117, Bio-Rad Laboratories, Hercules, CA). Membranes were blocked with 5% milk in PBS and 0.1% Tween-20. PVDF membranes were probed using an antibody against proNGF (1:500; cat. no. ANT-005, Alomone Labs, Jerusalem, Israel) and the secondary antibody goat anti-rabbit IgG HRP (1:2000, cat. no 7074P2, Cell Signaling). Membranes were stripped with Restore^™ Western Blot Stripping Buffer (cat. no. 21059, Thermo Fisher Scientific) for 15 min at room temperature before applying anti-β-actin antibody (1:5000, cat. no. A1978 Sigma-Aldrich) followed by goat anti-mouse IgG (cat. no. ab97023, Abcam, Cambridge, UK) to control for protein loading.

2.6. Fear Conditioning Extinction.

To examine extinction of fear conditioning, we adapted a strong extinction protocol from two previous studies (Boskovic et al., 2018; Lin et al., 2011). Briefly, mice were exposed to context A and B over 4 days in this procedure. Context A was a Ugo Basile fear conditioning apparatus (cat. no. 46003, Ugo Basile SRL, Gemonio, Italy). Context B was an empty reagent bucket with roughly similar floor surface area. These two contexts had different wall and ceiling patterns, floor shapes and textures, odorants (vanilla vs. lemon), and were cleaned with different alcohols (70% ethanol or 30% isopropanol) on each day. The apparatuses were lit with soft indirect light (~20 lux) and the foot shock was calibrated to 0.1mA. The conditioning stimulus (CS) in both contexts was a 50dB 3kHz sinusoidal tone. There was soft white noise played within each trial to mask outside noises and the behavior within each context was recorded from an overhead camera and analyzed via Anymaze. Briefly, extinction of fear conditioning was performed in the following manner: Day 1 (Context A): 120s baseline and 4x (120s CS co-terminating with a 1s shock) with 120s intertrial interval (ITI). Day 2 (Context B): 120s baseline and 30x (120s CS with 2s ITI). Day 3 (Context B): 120s baseline and 2x (120s CS and 120s ITI). Day 4 (Context A): 120s baseline and 2x (120s CS and 120s ITI). We recorded freezing behavior throughout the entirety of each test, but restricted our analyses to the baseline and CS presentation periods of each trial on days 1, 3, and 4. We used the software-recommended definition of freezing: greater than or equal to 1000 consecutive ms without detectable motion in the apparatus. In our analyses, raw freezing time (recorded to the nearest tenth of a second) was first converted to percentage of the interval spent freezing (i.e., “% freezing”) before statistical analysis. We excluded animals if they showed no conditioning to the CS above the a priori determined, but arbitrary threshold of 20% of the CS+ period spent freezing. Between day 3 and 4, a small and random subset of mice (n = 16) was selected and euthanized for evaluation of c-Fos immunoreactivity as described above. Therefore, in total, days 1–3 of the paradigm had n = 75 mice and day 4 had n = 59 mice.

2.7. Experimental Design and Statistical Analysis.

As the effect size for our measures of interest was unknown prior to initiating this study, we could not perform a formal power analysis. Instead, we inferred group size from previous studies by our lab and others with similar endpoints. Because in other studies we have never observed an effect for sex, or genotype vs sex interaction in these animals, we did not randomize with counterbalance for sex. We assessed our data sets for normality via Wilk-Shapiro before proceeding with parametric measures. If assumptions of parametric measures were not met within these data sets, we elected to use the corresponding non-parametric measure Mann-Whitney U or Kruskal-Wallis ANOVA. Two-sample Kolmogorov-Smirnov test with a manually calculated Bonferroni correction was used for analysis of frequency distribution curves. All of our statistical measures were performed within GraphPad Prism 9.0 (Graphpad Software Inc., San Diego, CA). Figures were created within GraphPad Prism 9.0, Adobe Photoshop CS 11.0 (Adobe Inc., San Jose, CA), and BioRender (https://biorender.com).

3. Results

3.1. gp120 transgenic mice exhibit fewer ChAT+ neurons in two basal forebrain subnuclei.

BFCNs express a functional p75^NTR (Yeo et al., 1997). These neurons undergo an age-associated reduction in total number (Xie et al. 2019) and are also lost in selected models of neurodegenerative disease (Granholm et al. 2000; Kerbler et al. 2013). Therefore, in concordance with previous studies that gp120 promotes neurodegeneration through p75^NTR (Speidell et al., 2020), we hypothesized that this receptor may mediate an accelerated age-associated loss of ChAT+ neurons in gp120tg rodents.

To evaluate loss of BFCNs, we employed an unbiased stereological analysis of sections from WT, gp120tg, and p75^NTR heterozygous (+/−) intercrossed with gp120tg mice. We have previously shown that in p75^NTR+/−gp120tg mice the levels of p75^NTR in synaptosomes is significantly reduced when compared to either WT or gp120 (Speidell et al., 2020); thus, these animals are valuable to examine a potential effect of the removal of one p75^NTR allele. Using a nickel-intensified DAB immunohistochemical approach, we visualized ChAT-expressing cells in basal forebrain (Fig. 1A) of young (3–5-month-old) and aged (12–14-month-old) mice. The optical fractionator, a standard stereological method, was used to generate estimates of the total number of ChAT in the combined MS and VDB in four serial forebrain sections. Young gp120tg mice, who exhibited a behavior similar to WT (Bachis et al., 2016; Speidell et al., 2020) did not display a significant reduction in ChAT-expressing neurons in the BFCN when compared to WT (Figs. 1B–C, Student’s two-tailed t-test; t=0.2721, p=0.7990). On the contrary, older gp120tg mice exhibited a reduced number of BFCNs vs aged WT counterparts (Figs. 1D–E, One-Way ANOVA and Tukey HSD post-hoc; F_(3,15) = 24.43, p <0.001). Interestingly, experimental groups heterozygous for p75^NTR had an increased number of ChAT-expressing neurons in this region vs aged WT (Figs. 1D–E). This finding is not unexpected, given the critical role for p75^NTR in this neuronal population (Boskovic et al., 2014). Importantly, we observed no reduction in ChAT-expressing cells in p75^NTR+/−gp120tg vs the p75^NTR+/− group (Figs. 1D–E), suggesting that heterozygosity at p75^NTR may protect against gp120-associated reductions in this neuronal population.

Figure 1. The population of ChAT-expressing cells in the basal forebrain of aged gp120tg rodents is reduced.

A. Stereological workflow and approximate anatomical area analyzed in gp120 mice. Medial septal (MS) and vertical limb regions of the diagonal band (VDB) were assessed using the optical fractionator stereological probe. B. Representative photomicrographs of the ChAT+ MS/VDB neurons in young mice immunostained with an anti-ChAT antibody. Scale bar 200μm. C. Quantification of the ChAT+ cells in young mice via the optical fractionator. D. Representative photomicrographs of the MS/VDB region in the aged cohorts (four experimental groups). Scale bar 200μm. E. Quantification of ChAT+ cells in aged experimental groups via the optical fractionator. Data shown as mean ± SEM. **p<0.01, ***p<0.001, One-way ANOVA with Tukey HSD post-hoc. WT (2F, 4M); gp120tg (3F, 2M); p75^NTR+/− (4F, 0M); p75^NTR+/−gp120tg (1F, 3M).

3.2. ChAT-expressing neurons display shrunken morphology in aged gp120tg rodents.

Shrinking of BFCNs (i.e., reduction of soma volume) has been reported frequently and has been shown to be a common outcome of normal, non-pathological aging processes (Banuelos et al., 2013; Martinez-Serrano et al., 1995). Reductions of neuronal volume precedes outright cell death (Schliebs and Arendt, 2011) and indicates that neurites and distal processes in targeted regions are likely retracting or degenerating. To examine whether ChAT-positive neurons in old gp120tg mice exhibit features of neuronal processes simplification, we utilized a parallel stereological probe to evaluate cross-sectional areas of ChAT+ neurons. Within the nucleator probe, we collected cross-sectional areas (CSAs) of this cell populations as they appeared in our sections using a random and software-generated procedure. As with the optical fractionator probe above, we first evaluated CSAs in young WT and gp120tg mice. These ChAT+ neurons had large cell bodies with distinct neurites visible in both groups (Fig. 2A). To assess the distribution of all CSAs measured in these experimental groups, we constructed cumulative frequency distribution curves from all CSAs, normalized to the maximal value observed in the WT group. We found there to be minimal difference between the curves constructed from 3–5-month-old WT and gp120tg mice (Fig 2B), although the Kolmogorov-Smirnov (K-S) test on these distributions surpassed our experimental alpha (K-S “D statistic” = 0.08856; p = 0.0436).

Figure 2. ChAT-expressing cells in the medial septal nucleus of aged gp120tg rodents adopt a shrunken somal morphology.

A and C. Representative images taken from the MS region of the indicated experimental groups of young (A) and old (C) mice. Extended depth-of-field composite of at least 12 images with z-increment of 1 μm. Scale bar 50 μm. B and D. Cumulative frequency distribution curves of all cross-sectional areas of the indicated groups of young and old mice, respectively, given by measurements using the vertical nucleator stereological probe. Data are shown normalized to the maximum WT cross-sectional area within each cohort. Area under curve (AUC) in arbitrary units: young: WT (7684), gp120tg (7906); aged: WT (7829), gp120tg (8453), p75^NTR+/− (7464), p75^NTR+/−gp120tg (7822). Note leftward shift of aged gp120tg curve (red) indicated by greater AUC. *p<0.05, ***p<0.001 Kolmogorov-Smirnov test with Bonferroni post-hoc. WT (2F, 4M); gp120tg (3F, 2M); p75^NTR+/− (4F, 0M); p75^NTR+/−gp120tg (1F, 3M).

We next evaluated CSAs in all four 12–14-month-old groups. In general, all four experimental groups at this age had smaller ChAT+ soma than the younger counterpart, but neurites were still plainly visible emerging from these cell bodies. Compared to the old WT mice, the 12–14-month-old gp120tg mice had occasional neurites visible near cell bodies, but ChAT+ bodies in the groups appeared reduced in area (Fig. 2C). When CSA cumulative frequency distribution curves were generated for all 12–14-month-old animals, we observed a leftward shift of the gp120tg CSA curve vs other groups, indicating that there is a greater proportion of smaller CSAs in gp120tg rodents (Fig. 2D). There was a very modest leftward shift in p75^NTR+/−gp120tg vs p75^NTR+/−, but this did not approach the magnitude of the gp120tg vs WT shift (Fig. 2D). Sequential K-S tests with Bonferroni post-hoc correction indicated significant differences between the frequency distributions for WT and gp120tg (D = 0.2539; p < 0.001) and gp120tg and p75^NTR+/−gp120tg (D = 0.2078; p < 0.001), but not for WT and p75^NTR+/−gp120tg (D= 0.0458, p = 0.5033). Overall, our data indicates that histopathological features of BFCN degeneration are accelerated in gp120tg animals and that this damage may be reduced through downregulation of p75^NTR.

3.3. Old gp120tg mice exhibit increased proNGF in the basal forebrain.

The main ligands for p75^NTR are the proNTs. Previous studies have shown an accumulation of proBDNF in several brain areas of HAND subjects combined with a decrease of furin and tissue plasminogen (Bachis et al., 2012), two endoproteases that cleave proBDNF to mature BDNF (Seidah et al., 1996). A decrease in furin accompanied by an increase in proBDNF have been reported in the gp120tg mice used in this study (Bachis et al., 2016). ProNGF, like proBDNF, harbors the R-X-K/R-R furin cleavage motif at its pro/mature domain boundary (Pagadala et al., 2006). We therefore hypothesized that a reduction in furin expression in this animal model could similarly affect also the levels of proNGF. Accordingly, we utilized Western blot analysis to measure the content of proNGF immunoreactivity in the basal forebrain of the four aged genotypes. The antibody against proNGF detected a single immunoreactive band (~42kDa), indicative of proNGF (supplementary Fig. 1), supporting other studies showing that proNGF is the most abundant form of NGF in the adult brain (Fahnestock et al., 2001). The intensity of proNGF immunoreactivity was stronger in gp120tg mice when compared to age-matched WT or the other two genotypes (supplementary Fig. 1), confirming previous data that gp120 increases the accumulation of proNTs in vivo (Bachis et al., 2016).

3.4. Extinction of fear conditioning is impaired in aged gp120 transgenic mice.

To examine whether the synaptic simplification of ChAT neurons observed in old gp120tg mice has a behavioral consequence, we tested animals for consolidation of extinction of conditioned fear. This complex behavior is modulated by p75^NTR-expressing cells of the basal forebrain (Boskovic et al., 2018). We included young p75^NTR heterozygous groups, which do not display any behavioral abnormalities (Bachis et al., 2016) as a comparison. Within a four-day conditioning-extinction-recall paradigm (Fig. 3A), all young and aged genotypes displayed equivalent freezing behavior within pre-conditioning and conditioning phases (Figs. 3B and C, Young: F_(3,24) = 0.6967, p = 0.5631; Aged: F_(3,43) = 0.9293, p = 0.4348). Freezing behavior in the pre-conditioned stimulus (CS) interval on day 3 was equivalent among all genotypes (Fig. 3D; Young: F_(3,24) = 0.7291, p = 0.5447; Aged: F_{(3, 43)} = 1.961, p = 0.1341). While young groups freezing behavior in extinction-recall trials was comparable (Fig. 3E; One-way ANOVA: F_(3,24) = 0.7523, p=0.5318), aged gp120tg mice displayed increased freezing during CS presentations vs WT (Fig. 3E, F_(3,43) = 5.482, p = 0.0028, Tukey HSD post-hoc WT vs gp120tg p <0.01). This increase in freezing behavior was rescued in aged p75^NTR+/−gp120tg rodents (Tukey HSD post-hoc p75^NTR+/−gp120tg vs gp120tg p<0.05). Contextual recall (Fig. 3F) was found to be equivalent between all genotypes of either young or aged cohorts, indicating that continued freezing following extinction training could not be attributed to broader hippocampal dysfunction. Overall, our results suggest that the function of neurons which subserve consolidation of extinction training in rodents may be interrupted, likely through a p75^NTR-mediated process.

Figure 3. Extinction of conditioned fear is impaired in aged gp120tg mice.

A. Experimental schematic of the four-day fear conditioning and extinction paradigm conducted in young and aged mice. B and C. Freezing activity during the pre-conditioning and conditioning periods of day 1, respectively. Kruskal-Wallis: young N/A (no non-zero values); aged: H = 4.497, p = 0.2126. D and E. Freezing during the pre-CS and CS periods of day 3, respectively. F. Total freezing activity during the CS presentation of day 4. Freezing behavior in each panel is expressed as a percentage of the cumulative duration of the CS presentation(s). Pre-CS denotes the two-minute interval before presentation of the first CS. **p<0.01 vs WT, ^p<0.05 vs gp120tg, one-way ANOVA performed within each age group with Tukey HSD post-hoc as needed (Comparisons not shown are non-significant). Young: WT (4F, 3M); gp120tg (2F, 4M); p75^NTR+/− (2F, 4M); p75^NTR+/−gp120tg (8F, 1M). Aged: WT (5F, 7M); gp120tg (6F, 6M); p75^NTR+/− (7F, 5M); p75^NTR+/−gp120tg (7F, 4M).

3.5. The immunoreactivity of a marker of neuronal activation is reduced in the hippocampus of aged gp120tg mice.

BFCNs innervate the hippocampus and facilitate an array of cell- and network-level function in this target region, including modulation of theta and gamma oscillations (Apartis et al., 1998) and alteration of synaptic transmission (Dasari and Gulledge, 2011). More broadly, cholinergic neurotransmission modulates neuronal excitability during learning (Picciotto et al., 2012). We hypothesized that discrepancies in activation of cells in the hippocampus targeted regions between our experimental groups may underlie the inability to consolidate extinction of conditioned fear.

To examine activation of BFCN-targeted regions during extinction of conditioned fear, we assessed neuronal expression of the immediate early gene product c-Fos in a subset of rodents sacrificed just after the day 3 extinction recall trial (Fig. 4A). We manually highlighted c-Fos immunoreactive cells and refined our highlighting through a machine learning (Rewire AI) model. Large and bright puncta, which rose above the intensity of background c-Fos expression, were taken as a proxy measurement for “activated” neurons (i.e. active during the extinction recall phase). In naïve animals, diffuse c-Fos immunoreactivity was observed with occasional and sparse activation of cells in the pyramidal layer of CA1–3 and the hilar region of the dentate gyrus (Figs. 4B and C), likely due to exposure to a novel environment just before euthanasia. Following the recall of extinction probe from our experimental paradigm, WT and p75^NTR+/− had greater c-Fos immunoreactivity in all hippocampal subregions when compared to naïve controls (Figs. 4C and D and supplementary Fig. 2). In contrast, c-Fos immunoreactivity in gp120tg animals appeared more similar to the naïve condition (Fig. 4D and supplementary Fig. 2), possibly representing under activation of an extinction-associated engram in this group. This effect was reversed in gp120tg mice lacking a p75^NTR allele (Fig. 4D and supplementary Fig. 2). Overall, our results establish a correlation between loss of neuronal activation and reduced behavior as a consequence of gp120 exposure.

Figure 4. Hippocampal c-Fos immunoreactivity is reduced following extinction recall in gp120tg mice.

A. Sections through the dorsal hippocampal formation were prepared from a small subset of mice euthanized 3 days after the extinction recall probe. Sections were immunostained for c-Fos. B. Representative images of the WT hippocampus taken at 4x indicating c-Fos immunoreactivity is largely restricted to the hilar region and pyramidal and granular layers. Boxes indicate approximate areas examined for c-Fos immunoreactivity (see also supplementary Fig. 2). Counter-clockwise from top: CA1, CA3, hilus. Scale bar 200 μm. C. Magnified (20x) images of the specified regions in experimental-naïve rodents. Scale bar 50 μm. D. Images from indicated regions in experimental groups following extinction recall probe. Arrows indicate observer-designated c-Fos immunoreactive cells.

4. Discussion

Forebrain cholinergic neurodegeneration has a significant contributing role in driving the cognitive impairments in dementias (Schliebs and Arendt, 2011). Atrophy and eventual loss of BFCNs in these neurological diseases are likely the result of the action of neurotoxic molecular stimuli on a variety of cellular systems. One critical family of ligands known to be affected by these stimuli are the NTs, whose trophic activity is especially critical for BFCN development and stability (Holtzman et al., 1992). To our surprise, virtually no reports to our knowledge had examined BFCN injury or loss in HAND. Therefore, we employed behavioral, biochemical, and histological approaches to evaluate complementary measures of BFCN health and function in an animal model of HAND.

BFCNs exhibit a robust expression of the pro-apoptotic p75^NTR (Boskovic et al., 2014). Accordingly, altered proNT processing, proNT overabundance, and NT withdrawal are among several major hypotheses explaining the early and selective degeneration of BFCNs in neurocognitive diseases such as AD (Chen and Mobley, 2019; Fleitas et al., 2018). Our previous studies have described decreased proBDNF processing that promotes more neurotoxic proBDNF in HAND individuals (Bachis et al., 2012), as well as in gp120tg mice (Bachis et al., 2016; Speidell et al., 2020), an animal model of HAND. In these animals, gp120 decreases neuronal proconvertases, such as furin and tissue plasminogen activator (Bachis et al., 2012), which catalyze the cleavage of the NT prodomain (Seidah et al., 1996). We utilized p75^NTR heterozygous mice intercrossed with gp120tg mice because heterozygosity at p75NTR downregulates the protein expression of p75NTR in synapses by up to 75% vs WT (Speidell et al., 2020) to probe the contribution of p75^NTR to gp120-associated outcomes. We found that, in gp120tg rodents, BFCNs exhibited signs of neuronal degeneration, accompanied by an underactivation of regions subserving a BFCN-facilitated behavior. These outcomes were not seen in p75^NTR+/−gp120tg rodents, indicating that the neurotoxic effect of gp120 on these neurons was at least partially mediated through p75^NTR.

To our knowledge, these are the first preclinical data describing cholinergic forebrain degeneration in a model of HAND. A decrease in ChAT+ cells may represent a loss of the cholinergic phenotype rather than wholesale apoptosis of the neuron, as has been suggested in other models (Crews et al., 2021; Granholm et al., 2000). However, our data still compare well with reports employing rodent models and BFCN degeneration. In fact, Xie and colleagues described similar age-associated reductions in BFCN volume in the MS/VDB (Xie et al., 2019). These decrements in volume were delayed by daily administration of a p75^NTR-modulating compound (Xie et al., 2019), further supporting a critical role for p75^NTR in BFCN longevity in the face of age-related neurotoxic insults. Degeneration and eventual loss of BFCNs are also frequently observed in rodent models of AD, possibly through a mechanism involving activation of p75^NTR by amyloid beta oligomers (Perez et al., 2007; Qian et al., 2019; Sotthibundhu et al., 2008). These reports support a growing hypothesis in the HAND field that exposure to HIV or its viral proteins drives AD-like pathology in the brain of PLWH (Overton et al., 2013; Vance and Brew, 2021).

A role for p75^NTR in driving cholinergic degeneration in response to hypoxia or inflammatory-related stimuli has been recently suggested (Sankorrakul et al., 2021). A recent intriguing report showed BFCN degeneration in a rodent model of binge drinking through a dual proinflammatory- and epigenetic-related mechanism (Crews et al., 2021). It is important to note, however, that degeneration of BFCNs may not totally explain the behavioral effects observed in this HAND model. In fact, gp120tg mice exhibit synaptodendritic degeneration and/or neuronal simplification in other brain areas, including the hippocampus and the striatum (Fields et al., 2016; Speidell et al., 2020; Steiner et al., 2015; Xie et al., 2021). However, it is important to point out that both biochemical and behavioral effects of gp120 were not observed in younger mice of either genotype, suggesting the collection of observed data is associated with age-related processes, likely including a chronic dysregulation of NT synthesis and neuroprotective signaling. In fact, the brain of gp120tg mice is characterized by high levels of proBDNF, proNGF and p75^NTR, and lower TrkB immunoreactivity in synapses than WT (Speidell et al., 2020). Conversely, we observed lower activation of neurons in the dorsal hippocampal formation as assessed by reduced c-Fos immunoreactivity following extinction recall. c-Fos is an immediate early response gene that is expressed in neurons following activity-dependent depolarization or other stimuli (Ceccatelli et al., 1989; Morgan et al., 1987). A reduction in expression following a stimulus could signify loss of hippocampal synapses, thus supporting previous data that gp120tg mice exhibit loss of dendritic spines (Speidell et al., 2019). This suggestion may explain why hippocampal neurons in gp120tg are underactivated upon reintroduction of the CS in an alternate context.

Our work was aimed at exploring the connection between HAND, cholinergic dysfunction and NT environment dysregulation. Although our experiments only briefly touched upon cholinergic compromise in a model of HAND, another publication has described upregulation of the α7 nicotinic acetylcholine receptor (nAChR) in gp120tg mice (Ballester et al., 2012). Moreover, post-mortem striata of HAND individuals show an increase of CHRFAM7 (Ramos et al., 2016), the gene encoding for the α7 nAChR, although the significance of this upregulation remains unknown. Other intriguing reports detail effects of HIV viral proteins on the cholinergic anti-inflammatory pathway present in macrophages and some T lymphocytes (Delgado-Velez and Lasalde-Dominicci, 2018). Thus, cholinergic dysfunction in HAND could also encompass components of the neuroimmune axis and thereby serve to exacerbate inflammatory-related processes in the CNS of PLWH.

Our data do not exclude that gp120 may have effects on the cholinergic system which are independent of NT activation of p75^NTR. BFCNs possess long-range projections to the cortical mantle, the hippocampal formation, and other distant neural structures. For this reason, both local and remote trophic and apoptotic stimuli must be considered to understand the factors driving neuronal degeneration. The NT family is well-known to utilize anterograde and retrograde trafficking in neurons (Altar and DiStefano, 1998). Thus, due to their robust expression of NT receptors, BFCNs may suffer from NT withdrawal at altered thresholds compared to other neuronal populations. Our previous work has demonstrated that gp120 compromises retrograde transport of BDNF in rat primary cortical neurons (Wenzel et al., 2019). We have also shown that the hippocampus, a major target of MS neurons, exhibits downregulation of BDNF and upregulation of proBDNF in both gp120tg mice and severe HAND individuals, thereby potentially aggravating impaired transport of NTs (Bachis et al., 2016).

Pharmacological blockade of p75^NTR or its activity may likewise attenuate apoptosis of BFCNs. LM11A-31 is an emerging p75^NTR-directed small molecule. When given to rodents over months, LM11A-31 is able to partially reverse age-related reductions in BFCN volume in mice (Simmons et al., 2014) and has shown limited effectiveness in protecting against mitochondrial dysfunction in the gp120tg model of HAND (Meeker et al., 2016; Xie et al., 2019). Finally, acetylcholinesterase (AChE) inhibitors are one of the few symptomatic pharmacological compounds available for cognitive complaints in AD and may likely offer some promise in HAND. Interestingly, some AChE inhibitors have been shown to not only increase the availability of Ach in BFCN-efferent regions in animal models, but also to support recovery of cholinergic markers in these vulnerable neurons (Crews et al., 2021). Only one clinical trial to date has assessed an AChE inhibitor in a HAND population, but reports from this trial revealed modest improvements on tests of processing speed and executive function in the treated group vs controls (Simioni et al., 2013). We remain interested in pharmacological approaches to limit p75^NTR activation in models of HAND for their potential use in PLWH.

In conclusion, we have shown that prolonged exposure to gp120 is sufficient to induce a degeneration of BFCNs and decrease the activation of BFCN-targeted regions during behavioral tasks which rely on this unique cell population. Importantly, through use of p75^NTR heterozygotes, we show that both biochemical and behavioral outcomes can be partially or fully rescued in aged animals, further supporting a critical role for p75^NTR in this region and the growing importance of this receptor in the pathogenesis of HAND.