This study investigated the effects of acute exposure to clinically relevant concentrations of antiretroviral drugs, both separately and in combination, on IFIT3 and STAT1 gene expression. The main goal was to create an in vitro model for HIV infection and antiretroviral treatment to determine which concentrations of the drugs effectively suppressed IFIT3 and STAT1 gene expression. We assessed the viability of differentiated SH-SY5Y cells exposed to clinically relevant concentrations of antiretroviral drugs using the MTS assay. Concentrations tested included Tenofovir Disoproxil Fumarate (TDF) at 1.8 μM, Dolutegravir (DTG) at 20 μM, and Emtricitabine (FTC) at 21 μM. Results showed no significant cytotoxic effects at these concentrations, indicating that TDF, DTG, and FTC do not significantly affect the viability of neuronal cells. These findings supported their selection for further experimental analysis ( Figure 1A ). Two different concentrations based on their clinical evidence of antiretroviral drugs were used for the in vitro study (18). To achieve this goal, the current study used drugs at their lowest therapeutic concentrations of TDF (0.06 μM), DTG (0.8 μM), and FTC (0.7 μM), and another set of highest therapeutic concentrations of TDF (1.8 μM), DTG (20 μM), and FTC (21 μM). We initially selected the lowest concentrations of the antiretroviral drugs mentioned above and cART drugs (a combination of the three drugs mentioned above at that concentration) to observe their effects on differentiated SH-SY5Y cells. The results indicated significant downregulation of the expression of the IFIT3 gene in the presence of DTG and cART within 24 hours of exposure ( Figure 1B ). However, the expression of the IFIT3 and STAT1 genes did not significantly differ between the cells exposed to TDF or FTC and the control cells. The effects of these antiretroviral drugs at the same concentrations on STAT1 gene expression were also investigated. Our data revealed that DTG (0.06 μM) significantly altered gene expression, with an approximately 0.8-fold change compared to the control (p = 0.0077). Similarly, acute exposure to the cART drugs significantly decreased STAT1 gene expression, with a 0.6-fold change compared to that of the control (p <0.0001) ( Figure 1C ). Overall, cART has a substantially greater effect on the downregulation of IFIT3 and STAT1 gene expression than individual drugs. Consequently, we investigated the effects of higher concentrations of these antiretroviral drugs, specifically TDF (1.8 μM), DTG (20 μM), FTC (21 μM), and cART drugs, on differentiated SH-SY5Y cells. Acute exposure to all antiretroviral drugs for 24 hours resulted in significant downregulation of IFIT3 genes with individual exposure or exposure to cART. There was a significant decrease in gene expression compared to the control group (p <0.0001) ( Figure 1D ). Similarly, the expression of the STAT1 gene was also significantly decreased compared to that of the control with the same concentrations of cART treatment ( Figure 1E ). Based on these findings, further experiments were designed with optimized concentrations of TDF (1.8 μM), DTG (20 μM), and FTC (21 μM).

SH-SY5Y cells were exposed to different concentrations of the HIV-Tat protein to observe the effects of the HIV protein on cell viability. The HIV-Tat protein stock was diluted with the culture medium as a working solution to ensure proper dosing. The Tat protein used its transduction domain to cross cell membranes. Once inside the cells, it interacts with transcription factors and modulates the activity of genes like IFIT3 and STAT1 which are vital for responding to viral infections. The 24-hour exposure to the Tat protein indicated a dose-dependent cytotoxic effect of the Tat protein on differentiated SH-SY5Y cells. Compared to the control, the lowest concentration (10 ng/ml) of Tat protein did not significantly affect the viability of differentiated SH-SY5Y cells. However, higher concentrations (>10 ng/ml) of Tat protein led to significant cellular toxicity. Following treatment with 50 ng/ml Tat protein, a significant reduction in cell viability (p <0.0001) was observed after 24 h of exposure. The trend of increasing toxicity was even more evident after exposure to the highest concentration of Tat protein (100 ng/ml) ( Figure 2A ).

To determine the most effective and nontoxic concentration of Tat protein in neuronal cells, we used three concentrations. Compared to the control, 10 ng/ml HIV-Tat significantly increased IFIT3 gene expression, which was approximately seven times greater (p < 0.0001). At 50 ng/ml, IFIT3’s expression increased nearly 2-fold (p = 0.0206), whereas at 100 ng/ml, IFIT3’s expression did not significantly change ( Figure 2B ). Therefore, 10 ng/ml was used for subsequent experiments.

In addition to IFIT3, we investigated the effects of Tat concentration on STAT1 gene expression. Using the same Tat concentrations, we found that 10 ng/ml significantly increased STAT1 expression, with levels twice as high as those of the control (p < 0.0001). Higher concentrations, such as 100 ng/ml, resulted in the downregulation of STAT1 expression (p < 0.0007) ( Figure 2C ). Thus, a Tat concentration of 10 ng/ml was chosen for the remainder of the study, as it significantly changed both IFIT3 and STAT1 gene expression.

The effects of the HIV Tat protein on IFIT3 and STAT1 gene expression were investigated in differentiated SH-SY5Y cells. Compared to the control, exposure to 10 ng/ml HIV-Tat resulted in a 5.7-fold increase in IFIT3 gene expression (p < 0.0001). However, the presence of cART drugs significantly suppressed IFIT3 gene expression both independently and in combination with the Tat protein ( Figure 2D ). Similarly, STAT1 gene expression was upregulated 3.2-fold in the presence of the Tat protein alone (p < 0.0001). When Tat was combined with cART or when cART was used alone, STAT1 gene expression was significantly downregulated (p = 0.0024) ( Figure 2E ). Western blot analysis was performed to evaluate the protein levels of IFIT3 and STAT1 following 24-hour treatments with Tat, Tat + cART, or cART alone. Tat protein exposure resulted in a significant increase in IFIT3 protein expression (p = 0.0010), whereas combined Tat and cART treatment suppressed IFIT3 protein expression. When treated with only cART, the IFIT3 protein level was significantly lower (p = 0.0068) ( Figures 2F, G ). In contrast, STAT1 protein expression was not significantly altered by any of the treatments. Western blot densitometry analysis confirmed that there was no significant expression of Tat protein levels compared to the control ( Figures 2H, I ). To assess the impact of HIV-1 on cytokines (IL-10), differentiated SH-SY5Y cells were exposed to HIV-1, HIV-1 with cART, or cART alone for 24 hours. Compared to the control treatment, HIV-1 exposure led to significant upregulation of IFIT3 gene expression, with a 20-fold increase (p < 0.0001). The combination of HIV-1 and cART significantly inhibited IFIT3 gene expression (p < 0.0001) but still remained significantly higher than control, whereas cART alone downregulated IFIT3 expression ( Figure 2J ). STAT1 gene expression was similarly affected, with HIV-1 alone causing a 5.7-fold increase (p < 0.0001). Compared to the control, HIV-1 + cART resulted in a 2-fold increase, which was less noticeable than that of HIV-1 alone (p = 0.0077). cART alone downregulated STAT1 expression (p = 0.0377) ( Figure 2K ).

Following the gene expression study, the effects of the HIV tat protein on the IFIT3 and STAT1 protein expression in differentiated SH-SY5Y cells were observed. In this regard, immunocytochemical studies were performed to examine the protein expression of the IFIT3 and STAT1 proteins after exposure to Tat, Tat + cART, or cART alone. This experiment was carried out to understand the possible regulatory impacts of HIV-Tat and cART on the expression of the IFIT3 and STAT1 proteins. Following 24 hours of HIV-Tat treatment, we observed a significant increase in the immunocytochemical staining intensity for IFIT3 (p <0.0001) ( Figures 3A, B ) and STAT1 (p <0.0001) ( Figures 3D, E ), suggesting an increase in their relative abundance. These results suggest that Tat exposure alone increased the protein expression of IFIT3 and STAT1. However, protein expression significantly differed when the cells were treated with the Tat+cART. Interestingly, we detected a moderate reduction in IFIT3 protein expression. When the cells were exposed to cART alone, the expression of the IFIT3 protein was reduced (p=0.9733) ( Figures 3A, B ), and the expression of the STAT1 protein was similar to that of the control, indicating a potential inhibitory effect of cART on the relative abundance of these proteins ( Figures 3D, E ). Additionally, automated image analysis via AI demonstrated that the classification accuracy for the IFIT3 ( Figure 3C ) and STAT1 ( Figure 3F ) treatments was approximately 100% on average across 6 runs per protein expression. As mentioned before, the improvement in classification accuracy was due to the use of a balanced dataset for training. The classification values in the IFIT3 and STAT1 test runs are 100% for all the control groups. The results revealed the importance of the balanced dataset, which can be further improved as more data is obtained.

To explore the effects of HIV exposure on cytokine expression in SH-SY5Y cells, SH-SY5Y cells were exposed to HIV, HIV+cART, or only cART. After 24 hours of exposure, cytokine expression was assessed. The results revealed a significant increase in the expression of proinflammatory cytokines (TNF-α, IL-6, and IL-1β) and anti-inflammatory cytokines (IL-10) in the HIV group compared to the control group (p <0.0001). On the other hand, in the HIV+cART group, the levels of IL-1β and IL-6 were significantly reduced (p < 0.0054 and p < 0.0014, respectively), demonstrating the efficacy of cART in mitigating the inflammatory effects induced by HIV. Compared to the control group, the cART group presented a significant change in cytokine expression, confirming that cART also independently affects cytokine levels ( Figures 4A-D ) Differentiated SH-SY5Y cells were further tested to observe the effect of HIV-Tat on Reactive oxygen species (ROS) production. The objective of this study was to determine changes in ROS generation after exposure to Tat, Tat + cART, and cART. Tat-treated cells exhibited the highest level of ROS production (19). The Relative Fluorescence Units (RFU), indicative of the level of ROS, was markedly greater in Tat-treated cells than in control cells (p <0.0001). On the other hand, the level of ROS production was lower in cells exposed to Tat + cART than in cells treated with Tat alone. However, it was still greater than that of the control (p <0.0006), indicating that cART treatment potentially reduces the ROS-generating effect of Tat in SH-SY5Y cells. The cART treatment alone did not significantly affect ROS production compared to the control ( Figure 4E ).

To observe HIV infection and the effect of cART on this model, an HIV P24 antigen assay and Long Terminal Repeat (LTR) PCR were performed on plasma and brain samples from HIV-infected and cART-treated humanized mice, respectively. In this regard, the study has used hu-PBL mice considering its timely human immune system engraftment and HIV infection in this model (Figure 5A). Our previous study has shown that this model can establish HIV infection within 12 hours of HIV inoculation and the effect of the treatment can be observed within one week of administration (20, 21). Thus, this model was preferred and adopted for the study compared to other existing models for HIV infection where infection and disease pathologies take a longer time to establish HIV infection pathologies (20–23). The results of the p24 assay revealed a gradual increase in infection from day 1 to day 5, and more prominently, on day 14, the HIV group presented a significant increase in viral load, indicating successful infection in this model. In contrast, the HIV + cART group presented a substantial reduction in p24 levels ( Figure 5B ), demonstrating the ability of cART to suppress active HIV replication in these mice. Additionally, HIV-LTR PCR analyses revealed a significant reduction in the latent HIV LTR gene in the HIV+cART group compared to the HIV group (p-value <0.0001), as shown in Figure 5C . These findings suggest that cART not only inhibits active viral replication but also helps in reducing latent HIV in the brain, an indicator of effective cART. The present study also investigated the expression of the IFIT3 gene and protein in brain tissues from the humanized mouse model. The experiment was designed to clarify the expression of IFIT3 at the gene and protein levels in response to HIV infection and its modulation by cART. To evaluate the expression of the IFIT3 gene, we conducted gene expression analysis in three different experimental groups. Control, HIV-infected, and HIV + cART-treated mice. Our PCR analysis of IFIT3 gene expression revealed significant upregulation in the HIV-infected group compared to the control group, which was almost 7-fold greater than the control group (p< 0.0001), indicating a notable response to viral infection at the genetic level. In contrast, the group treated with HIV+ cART presented a reduction in IFIT3 gene expression, which was lower than that in the control group ( Figure 5D ). Following this promising finding at the gene expression level, Western blotting was used to assess IFIT3 protein levels, providing information on the posttranscriptional impact of HIV infection and cART treatment. Compared to control mice, the protein expression of IFIT3 was significantly greater in HIV-infected mice (p= 0.0219), underscoring the upregulation of gene expression and protein synthesis in the presence of HIV. In the HIV+cART-treated group, a significant decrease in the expression of the IFIT3 protein was observed indicating the efficacy of cART in negatively regulating IFIT3 protein expression levels. ( Figures 5E, F ). However, it was not significantly different from the control group suggesting that while cART effectively controls viral replication, its impact on IFIT3 protein expression was not significantly altered during the tested conditions. To gain a deeper understanding of IFIT3 protein expression in humanized mice, we conducted a series of immunohistochemical analyses. This approach allowed us to precisely quantify the IFIT3 protein within distinct neuronal populations in brain tissues derived from both the HIV-infected and control groups. This focus was guided by evidence suggesting that neurons are particularly susceptible to HIV, which can lead to HAND. IFIT3, a crucial component of the neuronal interferon response, plays a significant role in the antiviral defense mechanisms of neurons, making it a pertinent marker for studying the effects of HIV at the cellular level. By employing the neuronal marker NeuN, we specifically targeted our examination to neuronal cells, ensuring that our findings would accurately reflect the changes in IFIT3 expression due to HIV infection and its treatment with cART (p < 0.0001). Our results revealed a significant increase in IFIT3 expression in the HIV group compared to the control group, confirming the gene upregulation at both the gene and protein levels following HIV infection. In the HIV+cART group, IFIT3 expression was significantly downregulated (p= 0.0219), indicating the effectiveness of cART in reducing the expression of this gene in the context of HIV infection ( Figures 6A, B ). These findings suggest that cART may have a beneficial effect on mitigating the impact of HIV on neuronal cells by modulating the expression of IFIT3. This evaluation was further validated with a machine-learning approach. This can impact model accuracy, particularly for IFIT3 classification. Model accuracy for different treatments was quantified, with the control, HIV, and HIV+ cART treatments showing accuracies of 100% ( Figure 6C ). The confidence interval analysis, which was based on six runs with preliminary data, indicated the impact of data balance on classification accuracy. The results revealed that this improved dataset classification resulted in approximately 100% classification accuracy.

All reagents and antibodies were obtained from the following sources: SH-SY5Y neuroblastoma cells (catalog #CRL- 2266, ATCC). Eagle’s Minimum Essential Medium (EMEM) (Catalog # 30-2003, ATCC), Fetal Bovine Serum (FBS) (Catalog # SH30071.03IR25-40, Cytiva), Dulbecco’s modified Eagle’s medium (DMEM) with high glucose (Catalog # SH30022. FS, Cytiva) Dulbecco’s phosphate buffer saline (DPBS) 1× (catalog # SH30256, Cytiva), penicillin–streptomycin-glutamine (catalog # 10378016, Thermo Fisher), and all trans retinoic acid (RA) (catalog # 302-79-4, Millipore Sigma) were used. 2 - Beta mercaptoethanol (catalog: BP176-100), Dimethyl sulfoxide (DMSO) (catalog: 67-68-5, Fisher Scientific), TDF (catalog #202138-50-9), FTC (catalog #143491-57-0) from MilliporeSigma, MA, USA), DTG (catalog #1051375-16-6, Advanced ChemBlocks Inc., Hayward, CA, USA). HIV-1 Tat protein (catalog# 7002, Immunodx Woburn, MA, USA), RNase AWAY™ Surface Decontaminant (catalog # 7002), DEPC-treated water (catalog #AM9939), RNeasy Mini Kit (Cat: 74104, Qiagen, Hilden, Germany), TaqMan™ RNA-to-CT™ 1-Step Kit (catalog # 4392938), STAT1 (catalog # HS 01013996_m1), IFIT3 (catalog #HS01922752752_s1), GAPDH (catalog # Hs02786624_g1), and TaqMan™ PCR Master mix (catalog # 4304437, Thermo Fisher). A CellTiter 96^® Aqueous One Solution Cell Proliferation Assay (Catalog # G3582; Promega, WI, USA), a BioTek Synergy HT multimode microplate reader (BioTek, VT, USA) and a Pur-A-Lyzer™ Maxi 6000 Dialysis Kit (Sigma–Aldrich) were purchased from the listed vendors. Poly-D-lysine-coated coverslips (catalog # NC0672873, Fisher Scientific), 4% paraformaldehyde (catalog # J61899. AK), Fluoromount-GTM DAPI (Catalog# 010001, SouthernBiotech), RIPA Lysis and Extraction Buffer (Catalog # 89901), Halt protease and phosphatase inhibitor (Catalog # 78440), Pierce™ BCA Protein Assay Kit (Catalog # 23227), prestained protein ladder (Catalog # 26616) from Thermo Scientific), 4–20% Mini-PROTEAN^® TGX (Catalog # 4561095, Bio-Rad), Trans-Blot Turbo RTA Mini 0.2 µm PVDF Transfer Kit (Catalog # 1704272, Bio-Rad, California, USA), beta-actin antibody (Catalog # sc-47778), IFIT3 antibody (Catalog # sc-393512), and m-IgGκ BP-HRP (Catalog # sc-516102) from Santa Cruz Biotech, Dallas, Texas, U.S.A. The STAT1 antibody (catalog # 9167s), anti-rabbit IgG, and HRP-linked antibody (catalog #7074s) were purchased from Cell Signaling Technology (Danvers, MA, USA). West Pico PLUS Chemiluminescent Substrate # 34579, Thermo Scientific).

Human SH-SY5Y neuroblastoma cells were cultured in a 75 cm² culture flask and the cells were passaged with 0.05% trypsin and 0.02% EDTA solution. For differentiation, SH-SY5Y cells were seeded at a density of 1x10⁵ cells/cm² in a T75 flask in DMEM/F12 containing 5% FBS and 1% penicillin–streptomycin. After 24 hours, the culture medium was replaced with a fresh medium containing 10 μM RA. RA was added every 3 days for a total of 7 days to induce differentiation toward a more neuron-like phenotype (51–53). The cells were kept in a humidified incubator at 37°C and 5% CO₂ until they reached 80% confluency. Differentiated SH-SY5Y cells at passage 7 or lower were used for further experiments.

Three different concentrations of Tat protein (10, 50, and 100 ng/mL) were prepared in EMEM (54–56). SH-SY5Y cells were treated with Tat protein at the desired concentration according to the experimental conditions, and the incubation period was 24 hours for all the experiments. The concentration of 10 ng/mL Tat protein was selected based on previous studies demonstrating that this level induces molecular and functional changes in neuronal cells without causing significant cytotoxicity (55). A 1 mM stock solution was prepared for all drugs, including FTC, DTG, and TDF, using either PBS or DMSO. A dilution was also performed to obtain the desired concentration for the study. All three drugs, as mentioned above, were used in equal proportions to prepare cART.

An MTS (3-(4-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt) assay was performed to determine the cytotoxicity of the Tat protein in SH-SY5Y cells. SH-SY5Y cells were seeded in a 96-well plate for differentiation until they reached 80% confluency and subsequently treated with various concentrations of Tat protein (10, 50, and 100 ng/ml) for 24 hours. After the incubation period, 20 µL of MTS reagent with 100 µL of media was added to each well, and the plates were incubated for an additional hour. The absorbance of each well was measured at 490 nm via a microplate reader. The percentage of cell viability was calculated via the following formula: % cell viability = absorbance of treated cells/absorbance of control cells × 100 (57).

The HIV-1_ADA virus (strain NIH-ARP- 416) was obtained from the National Institutes of Health HIV Reagent Program. The virus was thawed and briefly vortexed before use. SH-SY5Y cells were exposed to HIV-1 at a concentration of 100 pg/ml. The virus was added to the cells 24 hours after seeding and incubated for 72 hours to allow for viral entry and initial replication. The control wells were treated with an equivalent volume of virus-free medium.

Eighteen female Hu-NSG-PBMC mice (strain 745557, 6–7 weeks old) were obtained from Jackson Laboratory and assigned to three groups (n=6 each): a control group, which received a solvent mixture; an HIV group, which received the virus; and a third group, which was infected with HIV for 24 hours, followed by cART drugs exposure (22, 23, 58, 59). In this regard it is worth mentioning that our study was designed based on the hu-PBL mice that were obtained from Jackson Laboratory as mentioned in the manuscript. The reason for developing this model exclusively in female is due to more stable engraftment and immune reconstitution (60). In addition, the H-Y antigen which is lacking in female mice can trigger immune response in the transplanted male mice and cause “Graft-versus-Host reaction”. Thus, this model is largely based on the female mice. Experimentally, the mice were acclimatized for 4 days under controlled conditions (12-hour light/dark cycle, 20–22°C, 50–60% humidity) at The University of Texas Rio Grande Valley facility upon arrival. Baseline weights were recorded two days before experimentation (Day -2). On day 1, the HIV and HIV+cART groups were injected intraperitoneally with 200 pg of HIV. On day 0, the HIV+cART group received a cART drugs that was a combination of TDF (182 mg/kg), DTG (11.6 mg/kg) and FTC (9.3 mg/kg), reflecting human equivalent doses (22, 59). Blood samples were taken on days 1, 5, and 14. At the end of the study (day 14), the mice were euthanized with 30-40% Carbon dioxide displacement per minute in the animal’s home cage and heart perfusion to clear the organ blood, followed by immediate organ harvesting, flash freezing, and storage at -80°C in 24-well plates wrapped in paraffin. All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of The University of Texas Rio Grande Valley. Standard laboratory conditions and ad libitum access to diet and water were maintained throughout the study.

Following HIV infection in humanized mice, blood samples were collected on days 1, 5, and 14 to assess the viral load. Viral p24 antigen levels were quantified via an ELISA (ZeptoMetrix, Catalog No. 0801111). To confirm the results of the ELISA, quantitative PCR was performed to target the HIV-1 3-LTR region. The PCR utilized the following primers and probes: forward primer (F-Primer) GCCTCAATAAAGCTTGCCTTGA, reverse primer (R-Primer) GGCGCCACTAGAGATTTT, and a fluorescence-labeled probe (59FAM/AAGTAGTGTGCCCGTCTGTTGTGTGACT/3IABkFQ) (61). This combination allows for specific amplification and detection of HIV-1 genetic sequences, which facilitates the accurate quantification of viral RNA within the samples.

Total RNA was extracted from the SH-SY5Y cell line via the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. The purity and yield of RNA were evaluated via a DeNovix DS-11 fluorometer, and cDNA was amplified with the TaqMan™ RNA-to-CT™ 1-Step Kit gene-specific primers for human IFIT3 (Assay id Hs01922752_s1, Human STAT1 (Hs01013996_m1), IL-6 (Hs00174131_m1), TNF alpha Hs00174128_m1), IL-10 (Hs00961622_m1), IL1B (Hs01555410_m1), and GAPDH (Hs99999905_m1). Mouse IFIT3 (Mm01704846_s1) and mouse GAPDH (Mm99999915_g1) obtained from Thermo Fisher. The PCR assays were then performed on an RT–PCR system (Applied Biosystems) under the following conditions: 95°C for 10 minutes, 40 cycles of 95°C for 10 seconds, and 60°C for 1 minute. The reaction specificity was confirmed by running appropriate negative controls. The fold change was calculated via normalization of the expression to that of GAPDH via the ΔΔCt method.

Differentiated SH-SY5Y cells cultured in glucose-supplemented DMEM were seeded in 6-well tissue culture plates at a density of 7.5×10⁵ cells/well; cell lysates were prepared in ice-cold RIPA buffer supplemented with 100X Halt protease and phosphatase inhibitor cocktail for 15 min and then centrifuged for 10 min at 14,000 × g and 4°C. The supernatant was removed, and protein concentrations were determined via a BCA protein assay according to the manufacturer’s protocol. A total of 15 µg of protein was loaded into each well. The samples were prepared by diluting the lysates in 4X laminar loading buffer containing 1% sodium dodecyl sulfate with β-mercaptoethanol and heating for 5 min at 90°C. The samples were processed on 4–12% TGX protein gels (Bio-Rad) at 100 V for 1 h. The proteins were blotted onto PVDF membranes via the Trans-Blot^® Turbo™ Transfer System. The membranes were blocked for 1 h at room temperature in buffer consisting of TBST (0.1% Tween-20, 10 mM Tris-HCl, 150 mM NaCl) and 5% BSA. The membranes were probed with the following primary antibodies: mouse anti-IFIT3 (1:500) and the beta-actin antibody phospho-Stat1 (Tyr701) overnight at 4°C. The membranes were washed 3 times for 5 minutes at room temperature with TBS-T and then incubated with horseradish-conjugated anti-mouse or anti-rabbit secondary antibodies. The membrane was developed using West Pico PLUS Chemiluminescent Substrate. Images were captured using an Alliance Q9 Advanced.

Immunohistochemical analysis of humanized mouse brain tissue sections was performed according to the protocol (62). In summary, for fixation, mouse tissue was fixed in 10% neutral buffered formalin (NBF) (EprediaTM HiPurTM, Catalog No. 22-110-873) for 48 hours. Embedding: Fixed samples were immersed in melted paraffin (paraplast Plus, melting point 60 to 65°C, Fisher Scientific) three times (1 hour each) and embedded in paraffin with microcassettes and molds. Paraffin-embedded tissue samples were sectioned at a thickness of 4 μm on a rotary microtome and affixed to glass slides (SuperfrostTM Plus, Thermo Fisher, Waltham, MA, USA) in a 40°C hot water bath. Deparaffinization and rehydration: Tissue sections were deparaffinized with xylene and rehydrated through a progressive series of ethanol (50%, 75%, 95%, and 100%). Washing: The sections were washed three times (10 min each) in phosphate-buffered saline (PBS, Fisher Chemicals, Hampton, NH). For blocking, 1% bovine serum albumin (BSA; Fisher Scientific) was added to the samples, which were then incubated with PBS at RT for 1 hour to block nonspecific interactions. For primary antibody incubation, the slides were incubated with a mouse anti-IFIT3 primary antibody (Cat no: 15201-1-AP, Proteintech, Rosemont, IL, USA) at a 1:100 dilution and an anti-NeuN antibody, clone A60 (MAB377, MilliporeSigma, Burlington, MA, USA), at 4°C overnight. For secondary antibody incubation, after primary antibody incubation, the slides were allowed to reach RT for 15 minutes and then washed with PBS. The tissue sections were then incubated with anti-rabbit IgG (H+L) (catalog # A-11008) conjugated with Alexa Fluor 488 or Alexa Fluor 594 at a dilution of 1:500 at RT for 2 hours. After secondary antibody incubation, the slides were washed three times (10 min each) in PBS, and additional washes with PBS were performed to remove any unbound secondary antibodies. Nuclear staining and mounting of the cell nuclei were performed with Fluoromount-GTM DAPI (catalog# 010001, Southern Biotech), and the coverslips were mounted on microscope slides. Image acquisition and analysis: Image acquisition was performed via a Fluoview fV10i fluorescence confocal microscope with 60X Zoom for image capture. The analysis and quantification of the stain intensity, as well as the colocalization studies, were performed via ImageJ (NIH) software.

The following steps were performed to prepare the slides for immunofluorescence staining. Cells were seeded on poly-L-coated glass coverslips in 24-well plates and allowed to differentiate with RA, as previously described. The cells were treated with Tat and cART or with cART alone for 24 hours, then fixed with 4% paraformaldehyde in PBS for 10 minutes at RT and washed three times with PBS. The cells were permeabilized, blocked, and washed with 0.1% Triton X-100 in PBS for 10 min and incubated in blocking buffer at RT for 30 min. Following blocking, the cells were washed four times with 0.1% Triton X-100 in PBS for 5 min. The cells were incubated in 0.1% Triton X-100 in PBS containing primary antibodies (1:200 to 1:500 dilution) overnight at 4°C. The slides were subsequently washed three times with 0.1% Triton X-100 in PBS. For secondary antibody incubation, secondary antibodies conjugated with Alexa Fluor 488 were added at a 1:500 dilution to cover glasses and incubated for 1 hour in the dark at RT. For counterstaining and mounting, the cell nuclei were counterstained with DAPI. Images were acquired with an Olympus fluoview fv10i confocal microscope using a 60 × objective. Image analysis was performed with ImageJ (NIH) software to analyze protein expression.

Convolutional neural networks (CNNs) are well known for productive image classification in various fields and it also provided the best accuracy of the model (63). In addition, we have also employed rigorous validation strategies to ensure reliability of this method, and they are cross-validation, data augmentation and generalization assessment. Experimentally, we utilized such CNN models to perform image classification for the Tat experiments. An architecture of 4 blocks is used, where the typical elements in order are the convolutional layer of filter size 3, batch normalization, the rectified linear unit (ReLU) layer, and an average pooling layer of pool size 2. The convolutional layer channels were 8, 16, 32 and 32. For the last two blocks, the average pooling layer is removed. For the last block, additional layers were added after the ReLU layer, which are the 20% dropout layer, a fully connected layer with several neurons matching the size, a Softmax layer, and a classification layer. This type of architecture has been tested in other works, is practical for many classification tasks and is robust to noise (64). In this study, we employed two primary data augmentation methodologies. The first data augmentation methodology involves rotating images at specific angles (30, 15, 10, and 5 degrees) to increase the dataset size by 18 times the original number. The dataset was then partitioned, with 80% allocated for training and 20% allocated for validation. The validation set sizes were varied (20, 40, 60, and 120), corresponding to each angle partition. Initial experiments were conducted using a single angle for data augmentation, which was adjusted in subsequent trials to improve memory efficiency. All the experiments were performed with 100 epochs and a learning rate of 0.0001, except for the last attempt, which utilized 500 epochs and a learning rate of 0.00001. The second data augmentation methodology consists of parameters like those used in the first methodology for preliminary augmentation. A notable difference was the introduction of a secondary augmentation performed at each epoch during training, where MATLAB’s data augmenter was used to rotate training set images randomly. This approach aims to achieve high augmentation levels without significant memory resource sacrifices. The learning rate for the second methodology was fixed at 0.0001 since it performed better and converged adequately. The second methodology was chosen with the preliminary 15-degree rotation since it demonstrated a nuanced impact on model accuracy and validation performance, with higher angles having varied levels of effectiveness and finer angles increasing the computational time. The learning rate and the number of epochs yielded notable differences in model outcomes. Hyperparameter validation was carried out, and it was found that a learning rate of 0.0001 and setting the number of epochs to 100 models converged with better performance. Importantly, after data augmentation, a partition of 80% for training and 20% for testing was carried out. Owing to the dataset being four times larger than in the previous experiment, images were reshaped to a resolution of 256x256 to address computational constraints. This adjustment significantly reduced the processing time from approximately 10 hours to 45 minutes per run without compromising the accuracy of the predictions, which achieved approximately 100% accuracy across all classes.

SH-SY5Y cells were differentiated in 96-well plates with 10 μM RA. SH-SY5Y cells were incubated for 2 hours with 25 μM H2DCF-DA. After incubation, the H2DCF-DA solution was removed, and the cells were treated with TDF (1.8 μM), DTG (20 μM), FTC (21 Mm) H₂O₂, or catalase in their respective wells. The plate was immediately placed in a Bio Tek plate reader (Biotek, Synergy HTX) to measure ROS production in RFU via 480/20 excitation emission of 528/20 (65).

Statistical analyses were carried out with multiple replicates, and the data presented as means ± standard errors of the means. Before the primary statistical analysis, the normality of the data distribution was assessed. Upon confirmation of normality, one-way ANOVA and t tests were used to determine whether there were statistically significant differences between the means of the study groups. To determine the statistical significance of each experiment, one-way analysis of variance (ANOVA) and Dunnett’s multiple comparison tests were conducted via GraphPad Prism version 9 (GraphPad, Inc., CA, USA). A p value of <0.05 was considered to indicate statistical significance.