Behavioral and histological assessment of a novel treatment of neuroHIV in humanized mice

Neurocognitive deficits are prevalent among people living with HIV, likely due to chronic inflammation and oxidative stress in the brain. To date, no pharmaceutical treatments beyond antiretroviral therapy (ARV) has been shown to reduce risk for, or severity of, HIV-associated neurocognitive disorder. Here we investigate a novel compound, CDDO-Me, with documented neuroprotective effects via activation of the nrf2 and inhibition of the NFkB pathways. Methods: We conducted three studies to assess the efficacy of CDDO-Me alone or in combination with antiretroviral therapy in humanized mice infected with HIV; behavioral, histopathological, and immunohistochemical. Results: CDDO-Me in combination with ARV rescued social interaction deficits; however, only ARV was associated with preserved functioning in other behaviors, and CDDO-Me may have attenuated those benefits. A modest neuroprotective effect was found for CDDO-Me when administered with ARV, via preservation of PSD-95 expression; however, ARV alone had a more consistent protective effect. No significant changes in antioxidant enzyme expression levels were observed in CDDO-Me-treated animals. Only ARV use seemed to affect some antioxidant levels, indicating that it is ARV rather than CDDO-Me that is the major factor providing neuroprotection in this animal model. Finally, immunohistochemical analysis found that several cellular markers in various brain regions varied due to ARV rather than CDDO-Me. Conclusion: Limited benefit of CDDO-Me on behavior and neuroprotection were observed. Instead, ARV was shown to be the more beneficial treatment. These experiments support the future use of this chimeric mouse for behavioral experiments in neuroHIV research


Introduction
Human immunode ciency virus type1 (HIV) infection can result in signi cant neurobehavioral impairments.Among people with HIV (PWH), the spectrum of cognitive and functional de cits is termed HIV-associated neurocognitive disorders (HAND), which range from mild neurocognitive de cits, with limited impact on day-to-day functioning, to a rare and debilitating dementia (Antinori et al., 2007).
Estimates vary widely, from 40%-60% in case-control studies (Heaton et al., 2011;McArthur et al., 2005) to less than 8% in cohort studies (Sacktor et al., 2016;Vance et al., 2016;Wang et al., 2021;Wang et al., 2019) within the United States, where antiretroviral (ARV) therapy is widely available and serves as the only effective prophylactic measure against the most severe forms of HAND (Kolson, 2022b).Considering that there are over one million PWH in the United States and almost 40 million worldwide, including large regions where access to ARV is problematic, HAND represents a signi cant public health concern both locally and globally.While a complete understanding of HAND pathogenesis is lacking, the role of monocytes is welldocumented.Both infected and uninfected monocytes cross the blood-brain barrier, driven by both increased chemokine release in the central nervous system (CNS) and an enhanced peripheral immune response (Ancuta et al., 2004;Kraft-Terry et al., 2009;Peluso et al., 1985).This increased tra cking of monocytes into the CNS further increases the expression of chemokines, promoting the migration of additional monocytes in a feed-forward manner (Persidsky et al., 1999;Persidsky et al., 2000).Once inside the CNS, monocytes differentiate into perivascular macrophages (Cosenza et al., 2002), where their role in the HAND pathogenesis includes the release of pro-in ammatory cytokines, chemokines, reactive oxygen species, interferons, and viral proteins, each of which can injure neighboring cells (Adle-Biassette et al., 1999; Glass et al., 1995;Kaul & Lipton, 2006;Kedzierska & Crowe, 2002;Kraft-Terry et al., 2009;Lindl et al., 2007).Clinico-pathological analysis has found that the density of perivascular macrophages is strongly associated with neurocognitive status, further implicating this mechanism a key driving force behind HAND (Glass et al., 1995).
Initially, a subset of monocytes was implicated in the severe form of HAND (i.e., HIV-associated dementia) (Ellery et al., 2007;Pulliam et al., 1997); however, more recent studies indicated that milder forms of HAND were the result of chronic neuroin ammation and low-grade viral replication driven in part by monocyte in ux into the CNS (Langford et al., 2003;Yadav & Collman, 2009).In addition to in ammation, oxidative stress in the brain is an early and persistent consequence of HIV infection (Louboutin et al., 2010;Nath, 2002;Ronaldson & Bendayan, 2008), leading to neuronal dysfunction and death.Oxidative stress as a contributor to the pathogenesis of HAND through its effects on the brain is well established (Aksenov et al., 2001;Aksenov et al., 2003;Aksenova et al., 2006;Buckley et al., 2021;Sacktor et al., 2004;Toborek et al., 2003).In addition, HIV-induced oxidative stress may promote HAND pathogenesis via mechanisms outside of the CNS via activation of peripheral blood monocytes (Levine et al., 2013a;Williams et al., 2014).In vitro exposure of monocytes to HIV induces a proteomic response characterized by cellular activation and oxidative stress (Kadiu et al., 2009).Further, human studies have provided evidence for direct links between oxidative stress in peripheral blood monocytes and HAND.For example, anti-oxidant activity is diminished in monocytes and the cerebrospinal uid of women living with HIV with HAND (Velazquez et al., 2009).Findings from a large gene expression study of participants in the Multicenter AIDS Cohort Study indicate that oxidative stress within peripheral blood monocytes is strongly correlated with neurocognitive dysfunction in PWH (Levine et al., 2013b).In sum, oxidative stress acting on monocytes/macrophages both inside and outside of the CNS contributes to HAND pathogenesis; therefore, treatments that mitigate this response in both body compartments are likely to be more effective.
Under conditions of oxidative stress, a robust endogenous anti-oxidant stress response is induced throughout the body, including within the brain (Morris et al., 2019).This response is largely mediated through the transcriptional activation of a cis-regulatory element known as the antioxidant response element (ARE), which is located in the promoter region of numerous genes that modulate cytoprotective responses against oxidative injury.Regulation of this host response depends upon the nuclear transcription factor erythroid 2p45-related factor 2 (nrf2), which in turn is under tight regulation by the nrf2 binding protein Kelch-like erythroid CNC homologue (ECH)-associated protein 1 (KEAP1).KEAP1 normally binds to and sequesters nrf2 in the cytoplasm, where it cannot act as a transcriptional activator (McMahon et al., 2003).Under oxidative stress, KEAP1 detaches from nrf2, translocates into the cell nucleus, and then binds to and transcriptionally activates the ARE in the promoter region of numerous antioxidant response genes (Ma, 2013).This ARE-modulated pathway is recognized as a target for neuroprotection (Barone et (Lee & Surh, 2005;Surh et al., 2008;Surh et al., 2005), and anti-aging (Volonte et al., 2013).It has also been indirectly implicated as a target for HAND (Gill & Kolson, 2013; Kolson, 2022a) (Ambegaokar & Kolson, 2014;Garza et al., 2020;Gill et al., 2018), and among those molecules that inhibit the release of nrf2 is GSK3-β.GSK3-β inhibitors have shown modest promise for improving neurocognitive functioning in PWH (Ances et al., 2008;Letendre et al., 2006).Modi cation of KEAP1 functioning also in uences transcriptional activity of NF-κβ (Itoh et al., 1999;Lv et al., 2013), the most potent inducer of HIV-1 replication and an inducer of in ammatory factors.Targeting NF-κβ and nrf2 via this pathway can both suppress pathological over-activation of NF-κβ signaling and activate cytoprotective genes (Wakabayashi et al., 2010;Z. Wang et al., 2014) (Cross et al., 2011).The relevance of the KEAP1/nrf2 pathway to HAND has been directly demonstrated; HIV-1 gp-120 upregulates nrf2 expression in human astrocytes, and this in turn stimulates gene and protein expression of the antioxidants heme oxygenase-1 (HO-1) and NAD(P)H quinone oxidoreductase-1 (NQO1).Further, expression of proin ammatory factors TNF-α, NF-κβ, and matrix metalloproteinase-9 is elevated in astrocytes in which nrf2 expression is suppressed (Reddy et al., 2012;Reddy et al., 2011), and upregulation of nrf2 protein expression in astrocytes reduces oxidative damage (Reddy et al., 2010).Nrf2 was also found to be suppressed in HIV transgenic rats as demonstrated by decreased nrf2 and HO-1 expression (Davinelli et al., 2014).It is also notable that HO-1 expression is induced via the KEAP1/nrf2 pathway (McDonagh, 1990;Ryter & Choi, 2009).HO-1 is highly expressed in CNS cells (primarily astrocytes, macrophages, microglia, and endothelia) particularly during brain injury in several neurodegenerative disease states (Browne et al., 1999;Castellani et al., 1995;Castellani et al., 1996;Fagone et al., 2013;Mateo et al., 2010;Mehindate et al., 2001;Schipper et al., 1995;Schipper et al., 1998;Smith et al., 1994;Takeda et al., 2000).HO-1 is signi cantly reduced in prefrontal cortex of PWH with HAND (Gill et al., 2014), and this HO-1 de ciency correlates with brain HIV-1 RNA load, macrophage activation, and type-I interferon response.Further, an in vitro HIV neurotoxicity model demonstrates that HIV infection of monocyte-derived macrophages markedly reduces HO-1 expression and that this de ciency is linked to release of glutamate, a HAND-associated excitotoxin (Drummond et al., 1987;Huang et al., 2011;Jiang et al., 2001).This HO-1 de ciency and associated excitotoxin production are a generally conserved feature of infection with macrophage-tropic HIV-1 strains and correlate closely with the extent of virus replication (Gill et al., 2015), suggesting a potential therapeutic bene t of restoring HO-1 expression in HIV-infected brain macrophages.Induction of HO-1 expression in HIV-infected macrophages suppresses the release of excitotoxic levels of glutamate and thereby protects neurons against HIV-induced excitotoxic injury (Gill et al., 2014).Considering these ndings from a variety of independent groups of investigators and research models, therapeutics that activate the KEAP1/nrf2 pathway as a preventative measure for HAND are worth examination.
One such nrf2 pathway activator is Bardoxolone methyl ester (CDDO-Me), an orally-available semisynthetic triterpenoid that acts as an activator of the nrf2 pathway and an inhibitor of the NF-κB pathway.Several preclinical and Phase I and Phase II clinical studies have delineated the mechanistic, safety, and tolerability aspects of this compound (Hong et al., 2012; Y. Y. Wang et al., 2014;Warnock et al., 2012).Phase III clinical trials of CDDO-Me continue for other conditions such as pulmonary arterial hypertension and chronic kidney disease.Based on its mechanism of action, the studies described above suggest that CDDO-Me holds promise for preventing or treating HAND.Our present study described several experiments engineered to test the hypothesis that CDDO-Me could prevent or treat HAND in PWH.We rst assessed the effects of oral CDDO-Me administration on behavioral functioning of humanized mice engineered to express human monocytes and T-cells.We then examined histopathology and expression of neuroprotective factors (e.g., HO-1) in the mouse brains.Our hypothesis was that CDDO-Me administration following HIV-infection would prevent HIV-induced behavioral de cits and neuropathology via increases in neuroprotective factors.

METHODS
The Chancellor's Animal Research Committee at the University of California, Los Angeles, approved the research protocols described here.

Behavioral Testing of HIV-Infected Humanized Mice
The mice were initially grouped and treated as follows: 1. Pre-suppression ARV + CDDO-Me: Daily oral administration of CDDO-Me (10 mg/kg) via drinking water beginning immediately after viral setpoint was reached along with suppressive ARV therapy until viral suppression was reached, typically after 4 weeks.2. Post-suppression ARV + CDDO-Me: The same approach as above, but with CDDO-Me administration beginning after viral suppression had occurred, typically 4 weeks after ARV initiation.Once viral loads were determined to be below the detection levels, treatment with a daily oral administration of CDDO-Me (10 mg/kg) in drinking water was initiated and continued along with the suppressive ARV therapy for 10 weeks.3. ARV + Placebo condition: Same as above but without CDDO-Me.after HIV infection and treatment.Prior to testing, mice were handled daily for eight minutes over six days, followed by two days of habituation to a three-chamber open eld (63.6 x 43 x 24.5 cm) for 12 minutes each day.All tests were run consecutively over approximately one week, beginning with SIT and SMT, followed by NOR and OPR.Stimulus location was randomly alternated between left and right.Background noise, humidity, and 70% ethanol scent were kept consistent throughout the habituation and test days.Testing apparatus was centered on a lab bench to minimize gradients in light, temperature, and other environmental conditions that could produce a side preference.
All equipment was cleaned between conditions with 70% isopropyl alcohol to remove residue scents.Exploration times were calculated by the same human observer (C.T.) across all animals, and validated with video recordings.The observer was blind to the mouse group membership.Additional information concerning the behavioral tests can be accessed in Supplemental Materials.

Histopathological and Western Blot Analyses of Mouse Brains
This experiment assessed the e cacy of CDDO-Me for increasing nrf2-induced cytoprotective factors and for mitigating HIV-induced injury in humanized mouse brains.Mice were sacri ced at approximately 7-8 months of age, immediately after completing behavioral testing.A nal blood draw was taken, spleens removed, blood vessels tied off, and the animals underwent a saline perfusion protocol prior to brain removal.Brain hemispheres were then separated.One hemisphere was ash-frozen in isopentane, stored at -80 o C until being shipped on dry ice in one complete batch to the lab of DLK.After thawing, each specimen was dissected for ve brain regions (frontal cortex, parietal cortex, occipital cortex, striatum, and hippocampus) (Spijker, 2011).Brain tissue lysates were prepared by homogenization (~ 100 mg of tissue) by silica bead beating and sonication in 5 volumes buffer (10 mM Tris-HCl pH 7.8, 0.5 mM Dithiothreitol, 5 mM MgCl 2 , 0.03% Triton X-100) containing a phosphatase inhibitor cocktail set II (EMD Millipore) and a protease inhibitor cocktail (Sigma-Aldrich).Protein was quanti ed using the DC ™ (detergent compatible) protein assay (Bio-Rad).Equivalent amounts of proteins were added to Laemmli sample buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 12.5 mM EDTA, 0.002% bromophenol blue) with 2.5% 2-Mercaptoethanol and denatured at 95°C for 10 minutes.Proteins were resolved on an SDS-PAGE gel and transferred over night to poly(vinylidene uoride) (PVDF) membranes (4 o C).Membranes were blocked with Odyssey Blocking Buffer (PBS) (LI-COR Biosciences) and incubated with primary antibody overnight (4 o C).TRDye-conjugated secondary antibodies (LI-COR Biosciences) were used to detect the primary antibody.Quanti cation of protein bands was determined using Image Studio Lite software (LI-COR Biosciences).One sample prepared from mixing equal volumes of all samples (Mix) was used as running and transfer control in all membranes.Each blot was normalized to that sample in each membrane making possible to compare all brain regions and animals (Garcia-Mesa et al., 2020a).
The antibodies used in this study are listed in Supplemental Materials.To assess neuronal injury, antioxidant protein expression, and potential protection by ARV, with and without CDDO-Me, we quanti ed expression of postsynaptic density-95 (PSD-95 (Garcia-Mesa et al., 2020a)), heme oxygenase-1 and − 2 isoforms (HO-1, HO-2, respectively), glutathione peroxidase 1 (GPX1), glutathione peroxidase 4 (GPX4), and peroxiredoxin 1 (PRDX1) proteins by Western blotting in the ve brain regions listed above.We have used this approach (individual brain regions and grouped brain regions) to quantify synaptic injury and recovery associated with simian immunode ciency virus (SIV) infection in individual and grouped brain in SIV-infected rhesus macaques (Garcia-Mesa et al., 2020a).
For histopathological analysis, the remaining brain hemisphere was xed in 4% paraformaldehyde/PBS (4°C, 3 days) at the time of animal sacri ce.The brain samples were sent to the lab of VS for evaluation of synaptodendritic degeneration with antibodies against synaptophysin (SYP) and microtubule associated protein 2 (MAP2); gliosis with antibodies against glial brillary acidic protein (GFAP) and ionized calcium-binding adaptor molecule 1 (IBA1) and HIV-1 p24 burden in the hippocampus (dorsal hippocampal formation), striatum, and frontal cortex separately on parasagittal brain sections (Soontornniyomkij et al., 2012).Hematoxylin and eosin histopathology and chromogenic (3,3'diaminobenzidine, DAB) immunohistochemistry were conducted on adjacent 5-µm-thick para nembedded tissue sections (two technical replicates on single microscope slides).By means of twodimensional computer-assisted image analysis, immunoreactivity for SYP, MAP2, GFAP, and IBA1 was quanti ed on DAB tissue slides.In brief, the brain sections were digitally scanned with a microscope slide scanner (Aperio ScanScope GL, Leica Biosystems, Buffalo Grove, IL, USA) equipped with a (doubled) 20x objective lens.Using Aperio ImageScope software, the hippocampus, striatum, and frontal cortex separately were digitally drawn on each of brain images.For each marker, color segmentation was set to select the speci c signal and then consistently applied to all the brain samples based on the Aperio Positive Pixel count algorithm.The quantitative analysis results were used to calculate the immunoreactivity density, i.e., ([0.x Number of Strong Positive]) / Area (µm 2 ).For each brain sample, the average immunoreactivity density of two technical replicates was used for data analysis.The investigators conducting histopathological assessments of the hemispheres were blind to the experimental conditions.See Supplemental Materials for detail.

Statistical Analysis
Behavioral testing: Two-way ANOVA and t-test were performed to compare the exploration times among 3 groups between the two objects on the test day.Due to small group numbers resulting from mouse attrition, the 2 CDDO-Me treatment groups were combined, resulting in three groups: No ARV + Placebo (5 animals: 2 males, 3 females), ARV + Placebo (9 animals: 3 males, 6 females), and ARV + CDDO-Me (19 animals: 9 males, 10 females).Data were expressed as % of total exploration time.For Experiment 2, we maintained the original 4 groups.We performed two-way ANOVA, with sex and treatment group as covariates using the lm routine of the R statistical package (version 4.0.3).We tested 24 outcomes and used an FDR of 0.05 as our signi cance cutoff to adjust for multiple comparisons.

Western blotting:
A processed brain tissue 'mix', sample made by mixing equal volumes of all samples was used as a running and transfer control in all membranes.Each blot was normalized to that sample in each membrane, allowing for comparisons between all brain regions and animals.Tubulin was used as a loading control in all membranes.Using GraphPad Prism software, Western blot band signal intensities were compared by two-way ANOVA with repeated measures and Tukey's multiple comparisons.Values were expressed as means with standard errors of the means.For these analyses, we preserved the original four groups.

Histopathological analysis:
Two-way ANOVA were performed to compare the absolute levels of IBA-1, GFAP, MAP2, and SYP in each of three regions between the three groups, as described above.

RESULTS
Behavioral testing reveals rescue of social interaction de cits in animals receiving ARV and CDDO-Me Social Interaction-SIT: Animals within the ARV + CDDO-Me group spent signi cantly more time exploring the social cup compared to the empty cup (n = 19; P < 0.0001, t = 6.48) (Fig. 1A, data expressed as % of total exploration time).No differences in time exploring the social and empty cup were observed for either the ARV + Placebo group (n = 9; P = 0.81, t = 0.23) or the No ARV + Placebo group (n = 4; P = 0.83, t = 0.22).These results suggest that the combination of ARV + CDDO-Me is able to rescue social interaction de cits in the HIV-infected humanized mouse model.

Social Recognition Memory-SMT:
In contrast with the Social Interaction-SIT test, the Social Recognition Memory-SMT test showed no signi cant different effect on animals within the ARV + CDDO-Me group (n = 19; P = 0.63, t = 0.47) or No ARV + Placebo group (n = 4; P = 0.39, t = 0.92) (Fig. 1B).However, the ARV + Placebo group did spend signi cantly more time exploring the novel mouse than the familiar mouse (n = 8; P < 0.001, t = 4.56).These results suggest that ARV treatment is able to rescue the social memory de cits shown in our mouse model.Nonetheless, the administration of CDDO-Me may have attenuated the positive effect of ARV in social memory.

Novel Object Recognition-NOR:
No signi cant effect on Novel Object Recognition (NOR) was observed, as none of the grouped animals explored the novel object signi cantly more than the familiar object: ARV + CDDO-Me (n = 19; P = 0.13, t = 1.54);ARV + Placebo (n = 9; P = 0.071, t = 1.92);No ARV + Placebo (n = 4; P = 0.39, t = 0.9) (Fig. 1C).These results indicate that neither the administration of ARV alone nor ARV + CDDO-Me is able to rescue the novel object recognition de cits in our mouse model.

Object Place Recognition-OPR:
A positive effect of ARV was observed in the Object Place Recognition-OPR test (Fig. 1D).The ARV + Placebo (n = 8; P < 0.001, t = 4.16) group spent signi cantly more time exploring the object in the novel location than the object in the familiar location.In contrast, neither the ARV + CDDO-Me group (n = 18; P = 0.94, t = 1.71) nor the No ARV + Placebo group (n = 4; P = 0.38, t = 0.94) spent more time exploring the object in the novel location compared to the object in the familiar location.These results support the idea that ARV treatment is able to rescue the spatial memory de cits shown in our mouse model.However, similar to results observe in Social Recognition Memory-SMT test (Fig. 1B), the administration of CDDO-Me may have attenuated the bene cial effect of ARV in this test.
Limited preservation of PSD-95 by CDDO-Me treatment in animals receiving ARV as evidence for neuroprotection.
We used western blotting to quantify expression of PSD-95 as a marker of postsynaptic neuronal structural integrity and multiple antioxidant enzymes as markers of the host anti-oxidant response.This combined analysis provides an assessment of potential neuroprotection afforded by ARV and CDDO-Me treatment.Antioxidant enzymes examined include HO-1, HO-2, GPX1 and GPX4 (Huang et al., 2018), and PRDX1 (Lee, 2020).Among these enzymes each is driven at least in part by nrf2, except for HO-2, which is constitutively expressed and which expresses the same enzymatic function as HO-1 (Ma, 2013).Overall, enhanced expression of these antioxidant enzymes is associates with effective cytoprotection.
Initiation of CDDO-Me treatment concurrently with ARV, but not after ARV-induced HIV suppression, was associated with preserved PSD-95 expression in total brain and grouped cortex (frontal, parietal, occipital) regions (Fig. 2).This neuroprotective effect of ARV + CDDO-Me was most pronounced in the occipital cortex and was not observed in the striatum or hippocampus (Fig. 2A).When regional brain protein expression values for all regions were averaged ('total brain'), or as 'cortex' (frontal, parietal, and occipital), a pattern of preserved PSD-95 expression under conditions of concurrent initiation of ARV and CDDO-Me (group 1) or ARV and placebo (group 3) was observed (Fig. 2B, C).

-FIGURE 2-
No effect of CDDO-Me treatment on expression of antioxidant enzymes.
We next examined expression of heme oxygenase (two isoforms, HO-1 and HO-2).The HO-1 isoform is robustly inducible in response to oxidative stress, and is modestly inducible by nrf2 activating agents (Abraham & Kappas, 2008).In contrast, the HO-2 isoform is considered to be constitutively expressed, and not readily inducible.Figure 3 shows expression of each isoform in brain regions examined in Fig. 2. Expression of HO-1 was signi cantly higher in animals not receiving ARV when compared to those receiving ARV with or without CDDO-Me.This signi cant difference was observed in frontal cortex when untreated animals were compared with those receiving ARV + placebo, or concurrent ARV + CDDO-Me (Fig. 3A).Signi cantly higher HO-1 expression was also observed in untreated animals in total brain and grouped cortex (frontal, parietal, occipital) regions (Fig. 3B, C).In similar analyses of the constitutive isoform HO-2 was signi cantly elevated only in the striatum in untreated animals (Fig. 3D), and not elsewhere (Fig. 3E, F).These data suggest that unsuppressed HIV-1 replication, in the absence of ARV, may enhance endogenous expression of HO-1.
Finally, and in contrast, PRDX1 expression was signi cantly lower in animals not receiving ARV when compared to those receiving ARV with or without CDDO-Me (Fig. 5) -FIGURE 5 HERE-Histopathological differences were due to ARV treatment, not CDDO-Me P-values provided in Table 1 are the nominal per test p-values.With regard to treatment effects, four of the outcomes were signi cant at an FDR of 0.05.These were: hippocampus GFAP, striatum IBA1, hippocampus IBA1, and frontal cortex IBA1.For these 4 outcomes, we compared the individual treatment groups in post-hoc analyses.We determined that for all 4 outcomes, the difference was due to the effect of ARV and not to CDDO-Me.Animals treated with ARV had increased IBA1 values (~ 0.15) over animals without treatment.Animals treated with ARV and CDDO-Me had essentially the same IBA1 values as animals treated with ARV alone.We observed no sex effects on any of the outcomes (data not shown).Across four behavioral tasks, we observed that the combination of ARV + CDDO-Me was able to rescue only the social interaction de cits demonstrated by our No ARV + Placebo control group (Fig. 1A).Importantly, this social Interaction de cit was not rescued by the ARV treatment alone (Fig. 1A).
Conversely, CDDO-Me treatment did not rescue memory de cits, including the social recognition (Fig. 1B), novel object recognition (Fig. 1C), or object place recognition (Fig. 1D) de cits exhibited by the control group.Notably, ARV treatment alone (without CDDO-Me) was able to rescue the social recognition (Fig. 1B) and object place recognition (Fig. 1D) de cits.These results suggest that the ARV treatment has a positive impact in hippocampal dependent tasks such as social recognition and object place recognition, but not in cortical dependent tasks such as novel object recognition, in this humanized mouse model.
Notably, we observed some evidence for a neuroprotective effect (i.e., preservation of PSD-95 expression) of CDDO-Me when administered with ARV (Fig. 2A, occipital cortex), which was nonetheless not as consistent with the protective effect of ARV alone (Fig. 2B, C).Expression of PSD-95 is largely con ned to the post-synaptic regions of neuronal synapses, and it is a sensitive marker for synapse function and integrity (Vallejo et al., 2017), and synaptic injury is a common neuropathological feature of HIV infection (Gelman & Nguyen, 2010).Whether this modest CDDO-Me effect is directly related to the rescue effect on the social interaction behavioral test is unknown.Furthermore, neither of the effects observed with CDDO-Me treatment can be attributed to induction of antioxidant gene responses, as no signi cant changes in antioxidant enzyme expression levels (HO-1, HO-2, GPX1, GPX4) were observed in CDDO-Me-treated animals, with the exception of reduced expression of PRDX1.
The signi cantly higher levels of HO-  2009).Expression of each of these is regulated, at least in part, by the nrf2 transcription factor, in response to various cellular stressors.Despite this, we observed no differences in expression between CDDO-Me-treated and untreated animals.Only ARV use seemed to affect PRDX1 levels, indicating that it is ARV rather than CDDO-Me that is the major factor providing neuroprotection in this animal model.
Because ARV suppression of HIV replication in the mice revealed a brain antioxidant expression pattern that likely re ects a consistent response to HIV infection in the brain, the ndings support the use of this animal model for the further assessment of neuroprotection strategies based upon modulation of endogenous antioxidant enzyme responses.
Finally, immunohistochemical analysis found that several cellular markers in various brain regions differed between the groups: hippocampus GFAP and IBA1, striatum IBA1 and PRDX1, and frontal cortex IBA1.Similar to the behavioral and Western blot analyses, post-hoc analyses determined that the difference in the outcomes was largely due to the effect of ARV and not to CDDO-Me.Mice treated with ARV had increased IBA1 values (~ 0.15) over animals without treatment, and animals treated with ARV and CDDO-Me had essentially the same values as animals treated with ARV alone.
Several factors limit the interpretability and generalizability of our ndings.Firstly, we lacked groups that might have been useful, including a CDDO-Me-only group and an HIV-uninfected group.The complexity of engineering the humanized mice limited the number of animals available, which was in fact lower than we had anticipated, thus resulting in the combining of the CDDO-Me treatment groups for the behavioral and histopathological analyses.Secondly, the behavioral experiments, while validated and useful, might not have captured underlying neuropathological processes in the mice.Future studies that employ a wider range of behavioral tests could be useful, but they also need to consider the relatively fragile health and short lifespans of humanized mice.
In summary, while we did observe signi cant bene t from CDDO-Me administration together with ARV in rescuing the social interaction de cits demonstrated in our humanized mouse model of neuroHIV, we did not observe a bene t from CDDO-Me administration in maintaining other behaviors.Further, CDDO-Me did not elicit neuroprotection against HIV-mediated neuropathology.ARV use, however, did seem to preserve cognitive functioning and cellular integrity.These experiments do support the future use of this chimeric mouse for behavioral experiments.

Declarations
Author Contribution      See image above for gure legend.
75 x Number of Weak Positive] + [Number of Moderate Positive] + [1.25

Figure 1 See
Figure 1

Figure 2 See
Figure 2

Figure 3 See
Figure 3

Figure 4
Figure 4 al., 2011; de Vries et al., 2008; Joshi & Johnson, 2012; Scapagnini et al., 2011; Tanji et al., 2013; van Muiswinkel & Kuiperij, 2005; Zhao et al., 2011), chemoprevention and chemoprotection 4. No ARV + Placebo: DietGel Booast without ARV therapy or CDDO-Me.Behavioral functioning was assessed via four tests: Social Interaction Test (SIT), Social Recognition Memory Test (SMT), Novel Object Recognition (NOR), and Object Place Recognition (OPR).Details of these tests can be found in Supplemental Materials.We have previously shown these tests to share characteristics of other hippocampus-dependent spatial and object memory (Kogan et al., 2000), and similar tests have previously been used in HIV-related mouse studies (Kalkonde et al., 2011; Kesby et al., 2015; Soontornniyomkij et al., 2016).For each experimental group, we compared exploration time in familiar vs. novel conditions 24-hours after training.Speci cally, exploration of a familiar vs. novel mouse (in SMT), object (in NOR), and location (in OPR).Mice underwent training and testing at 7 months of age, 4 months
(Ingram et al., 2019)ety of injurious insults.Our failure to observe even higher levels of enzyme expression in the presence of CDDO-Me may re ect our inability to detect any effects above that caused by HIV infection, lack of effective in vivo drug levels, or other unknown factors.The signi cant decrease in PRDX1 expression in CDDO-Me treated animals without ARV administration (decreased striatum PRDX1 values ~ 1.1 lower) compared to mice treated with ARV alone indeed suggests a pharmacological effect within the brains of these animals.The remarkable discordance in expression between PRDX1 and the other enzymes in the setting of unsuppressed HIV infection suggests that responses to HIV infection itself (e.g., neuroin ammation) may be driving selective antioxidant gene expression.Previous studies have shown differential induction of PRDX1, HO-1, and other nrf2-driven antioxidant genes in response to pro-in ammatory signaling(Ingram et al., 2019).We have observed the same discordance in the relative expression levels of GPX1, HO-1, and PRDX1 in the brains of rhesus macaques with unsuppressed SIV infection (Garcia-Mesa et al., 2020b), which suggests a consistent association of these antioxidant enzyme expression patterns with both HIV and SIV infections in these model systems.
1, GPX1 and GPX4 in animals not receiving ARV with or without CDDO-Me may re ect the natural host endogenous antioxidant response to unsuppressed HIV infection in the brain (Gill et al., 2014; Gruenewald et al., 2020).Such responses (induction of antioxidant response enzymes, particularly HO-1, GPX1, GPX4, and PRDX1 have both distinct and overlapping functions.For example, HO-1 and HO-2 break down free heme, a major intra-and extra-cellular pro-oxidant, into carbon monoxide, biliverdin, and ferrous iron, with cytoprotective effects in numerous model systems (Basuroy et al., 2006; Chang et al., 2003; Chen, 2014; Chen et al., 2005).GPX1 is a detoxi er of peroxides and a suppressor of reactive oxygen species production and oxidative damage (Huang et al., 2018).It is particularly important for mitochondrial H 2 O 2 scavenging (Starkov et al., 2014).GPX4 is speci cally important in suppressing lipid peroxidation in the cellular membrane, and it thereby is a key suppressor of cellular injury by ferroptosis (Conrad et al., 2018; Yang et al., 2014).PRDX1 also detoxi es peroxides, including H 2 O 2 , and peroxynitrite (Goemaere & Knoops, 2012), and it suppresses microglial activation (Kim et al., 2013; Neumann et al., Andrew Levine, Scott Kitchen, Valerie Rezek, Manual Lopez-Aranda, Virawudh Soontornniyomkij, Alcino Silva, Yoelvis Garcia Mesa, and Dennis Kolson contributed to the study conception and design.Humanized mice were engineered and maintained by Valerie Rezek and Scott Kitchen.Behavioral experiments were conducted by Manual Lopez-Aranda and Chirag Thadani.Immunohistochemical and histopathological experiments were conducted by Virawudh Soontornniyomkij, Dennis Kolson, and Yoelvis Garcia Mesa.Data analyses were conducted by Manuel Lopez-Aranda, Yoelvis Garcia Mesa, and Dennis Kolson.All authors read and approved the nal manuscript.