Effects of HIV and azidothymidine on Alzheimer’s-like pathology and amyloid beta transporters in hu-PBL-NSG mice, brain endothelial amyloid beta uptake and endothelial barrier integrity

Abstract

Background

Despite increased life expectancy with antiretroviral therapy, people with HIV (PWH) have significantly higher prevalence of comorbid diseases, including neurocognitive impairment. PWH and neurocognitive impairment often show features of Alzheimer’s-like pathologies, including increased brain amyloid-beta (Aβ) and phospho-Tau; it is not known if antiretroviral drugs contribute to the development of these pathologies or potentiate HIV effects. We aimed to investigate whether azidothymidine (AZT), a drug still used by PWH in many resource-limited countries, alters or potentiates HIV-induced Alzheimer’s-like pathologies.

Methods

Hu-PBL-NSG mice plasma/serum and brain tissues were analyzed to investigate the effects of HIV-1 infection and AZT treatment on viremia, immunosuppression, phospho-Tau, Aβ₄₂, neuronal NeuN, endothelial claudin-5, ZO-1, and Aβ transporters [low-density lipoprotein receptor–related protein-1 (LRP1) and receptor for advanced glycation end-products (RAGE)] transcription, expression, and proteolytic cleavage. In vitro, we assessed the effects of HIV-1 Tat and AZT on Aβ₄₂ aggregation, endothelial LRP1 and RAGE expression, Aβ uptake and transport.

Results

HIV significantly increased brain phospho-Tau (serine199, threonine181, serine396), brain and plasma Aβ₄₂, decreased LRP1, NeuN, claudin-5, ZO-1, and increased RAGE transcription and expression, increased soluble(s)LRP1 and decreased sRAGE. AZT treatment of infected animals decreased blood and brain viremia and reduced HIV-induced immunosuppression but had no effect on HIV-induced brain Aβ₄₂, phospho-Tau, NeuN, claudin-5, ZO-1, LRP1 or RAGE transcription, expression, or cleavage. AZT treatment of non-infected animals significantly increased phospho-Tau in the brain somatosensory cortex, decreased LRP1, NeuN, claudin-5, and ZO-1, and increased RAGE transcription and expression, increased sLRP1 and decreased sRAGE. Tat decreased LRP1 and increased RAGE expression in brain endothelial cells, and AZT accentuated Tat-induced effects. AZT+Tat significantly increased Aβ aggregation and AZT significantly increased endothelial Aβ₄₂ uptake/retention.

Conclusions

HIV and AZT independently dysregulate NeuN, claudin-5, ZO-1, LRP1 and RAGE transcription, expression, and proteolytic cleavage in hu-PBL-NSG mice. Such dysregulation of neuronal nuclei, endothelial tight junction proteins, and Aβ transporters could contribute to increased neurovascular injury and altered Aβ clearance following HIV infection and/or AZT treatment. In the presence of Tat, AZT increased Aβ aggregation. AZT also increased endothelial Aβ uptake/retention, which suggests that AZT may contribute to brain endothelium impairment and dysfunction.

Supplementary information

The online version contains supplementary material available at 10.1186/s12987-026-00807-4.

Background

Successful combination antiretroviral therapy (cART) over the past decades has significantly increased the life expectancy of people with HIV (PWH). However, there is currently no cure for HIV, and compared to age-matched controls, PWH have a significantly higher prevalence of comorbid diseases, including central nervous system (CNS) complications [1–3]. These CNS comorbidities include HIV-associated neurological disorders (HAND), a constellation of behavioral, motor and cognitive abnormalities associated with increased blood-brain barrier (BBB) injury and dysfunction, neuronal injury, and neurodegeneration [3–6]. Although the prevalence of the most severe form of HAND (HIV-associated dementia) has decreased in the cART era, milder forms of HAND [asymptomatic neurocognitive impairment and mild neurocognitive disorders] are still highly prevalent and occur in up to 50% of the current 40.8 million PWH [3–5, 7].

Autopsy studies of HIV-infected people showed the presence of proteopathy and Alzheimer’s disease (AD)-like CNS pathologies, including increased amyloid-beta (Aβ) deposits, Tau hyperphosphorylation and the presence of neurofibrillary tangles (NFTs)-like structures in brain tissues [8–13] associated with high viral loads [13] and neurocognitive impairment, including impairment in speed of information processing, attention, and working memory [12, 13]. These AD-like CNS pathologies are also present in people who have been on long-term cART, but it is not known whether antiretroviral drugs (ARVs) contribute to the development of these pathologies or potentiate the effects of HIV and viral factors.

Newer ARVs have been shown to induce less toxicity and adverse reactions compared to older HIV medications [14–16]; however, many long-term HIV survivors had been on older ARVs [15, 16] and it has been suggested that some comorbid diseases in PWH may be due to legacy effects from older HIV medications [16, 17]. Several such older ARVs are nucleoside/nucleotide reverse-transcriptase inhibitors, a drug class that constitute the backbone of cART and are used both for HIV treatment and pre-exposure prophylaxis. Drugs in this class approved for HIV treatment have included the nucleoside analogs azidothymidine (AZT), stavudine, zalcitabine, lamivudine, emtricitabine, didanosine, and abacavir; and the nucleotide analog tenofovir. Of these seven nucleoside analogs, three (zalcitabine, didanosine, and stavudine) have been withdrawn or discontinued, and four (AZT, lamivudine, emtricitabine, and abacavir) remain part of current FDA-approved HIV medicine [14]. Although AZT is now rarely used in developed countries [14, 18], it is still used in many resource-limited countries; the current World Health Organization (WHO) HIV treatment guideline list AZT-based regimens as preferred and alternative 2^nd line cART for adults, adolescents, and children [19]. The goal of the present study was to investigate the effects of HIV infection and AZT treatment on CNS Aβ₄₂, phospho-Tau, neuronal nuclei protein (NeuN), the brain endothelial tight junction proteins (claudin-5, ZO-1) and Aβ transporters low-density lipoprotein receptor–related protein-1 (LRP1), a receptor that mediates Aβ CNS efflux/clearance [20–22], and receptor for advanced glycation end products (RAGE), a receptor that binds and transports circulating Aβ into the brain [21, 23, 24]. We further investigated the effects of HIV-1 Tat protein and AZT on Aβ aggregation, and using an in vitro BBB model consisting of primary human brain microvascular endothelial cells, we assessed the effects of Tat and AZT on endothelial LRP1 and RAGE expression, Aβ uptake and transport.

Methods

Hu-PBL-NSG mice model

Four-week-old NOD/scid–IL-2 Rγ_c^null (NSG) mice were purchased from the Jackson Laboratory (Bar Harbor, ME), maintained in sterile microisolator cages under pathogen-free conditions in accordance with the University of Nebraska Medical Center (UNMC) and National Institutes of Health (NIH) ethical guidelines for the care of laboratory animals, and bred at the UNMC animal facility to expand the colony. This study was performed under a protocol approved by the UNMC Institutional Animal Care and Use Committee. Human peripheral blood lymphocytes (PBL) were obtained by countercurrent centrifugal elutriation of leukopheresis packs from HIV-1, 2, and hepatitis B seronegative donors, as previously described [25]. Mice (4 to 6 weeks old males) were engrafted by intra-peritoneal (i.p.) injection of human PBL (30 × 10⁶ cells/mouse), and engraftment was confirmed a week later by fluorescence-activated cell sorting (FACS) quantification of human CD45+ cells in the animals’ blood (Fig. 1). Engrafted animals were randomly assigned into 4 groups (10 mice per group): PBS (non-treated and non-infected mice), AZT (non-infected mice treated with AZT); HIV (infected animals, untreated), HIV+AZT (infected animals treated with AZT). Animals were infected by i.p. injection of HIV-1_ADA (single dose of 10⁴ tissue culture infectious doses-50, 100 μl) (Fig. 1) as previously described [26–28]. Controls were mock-infected by i.p. injection of 100 μl phosphate-buffered saline (PBS). Animals’ blood samples were collected weekly post-infection (p.i.) for 3 weeks and analyzed by FACS (Additional file 1). Animals were sacrificed at week 3 p.i., tissue samples were harvested (Fig. 1) and processed as previously described [26–28].

Fig. 1.

Schematic overview of animal experiments. NOD/scid–IL-2 Rγ_c^null (NSG) mice were engrafted with human peripheral blood lymphocytes (PBL); engraftment was confirmed by FACS a week post PBL injection and engrafted animals were randomly assigned to four experimental groups: PBS, PBS+AZT, HIV, HIV+AZT. For animals receiving AZT, treatment began at 12-hour post-infection. Animals’ blood samples were analyzed weekly by FACS; animals were sacrificed at week 3 post-infection, followed by FACS and ELISA analysis of blood and plasma/serum samples; ELISA, real-time PCR, and Western blot (WB) analysis of brain tissues

AZT preparation and treatment

AZT (Camber^TM Pharmaceuticals, Piscataway, NJ) in 300 mg tablets were purchased from the UNMC pharmacy. One tablet was crushed and dissolved in 4 ml of a 1:1 mixture of dimethyl sulfoxide (DMSO) and sterile PBS, resulting in an initial stock concentration of 75 mg/ml (C1) that was stored in 600 µl aliquots at −80 °C and used within two days of preparation. For treatment, 500 µl of C1 stock solution was further diluted in PBS, yielding an intermediate working concentration of 15 mg/ml (C2). The C2 solution was then diluted in PBS to obtain a final (C3) concentration of 7.5 mg/ml. This C3 solution was passed through a 0.45 µm syringe filter to remove any undissolved drug excipients and 200 µl (60 mg/kg, 1.5 mg AZT per 25 g mouse) injected (i.p.) into each mouse twice/day. This dose and treatment schedule were based on recommended human AZT doses of 300 mg twice/day and the consensus human-mouse interspecies allometric scaling factor of 12.3 [29, 30]. Untreated animals were injected (i.p.) with 200 μl PBS containing an equal amount of DMSO as in the final AZT solution (C3) used for treatment.

Ethics and processing of brain tissues

This study was performed under a protocol approved by the UNMC Institutional Animal Care and Use Committee (IACUC protocol number 19–141-01-FC). Immediately following the animals’ sacrifice, each brain was collected and placed in ice-cold PBS and under a stereo zoom microscope, each lobe was split into three pieces: frontal cortex (FC), somatosensory cortex (SSC), and a 3^rd piece consisting of a region encompassing the insular, auditory and visual cortex, hippocampus, thalamus, and hypothalamus. Animal samples were immediately frozen at −80 °C and subsequently used for protein extraction and Western blot analysis, RNA extraction and real-time PCR (Additional file 1).

Brain endothelial cells culture, treatment, and Aβ trans-endothelial transport

Primary human brain microvascular endothelial cells (HBMEC) were cultured as previously described [31, 32] (Additional file 1). For Aβ transport, confluent HBMEC cultured on collagen-treated tissue culture inserts (0.4-μm pore size; Corning, Lowell, MA) were exposed to human Aβ₄₂ peptide (10 μM, added to the upper chamber of the transwell) in the presence or absence of AZT (1 μM) and/or high-affinity antagonists for LRP1 (500 nM, Kerafast, Boston, MA) and RAGE (200 nM, Tocris, Minneapolis, MN). Doses of LRP1 and RAGE inhibitors were selected based on previously published studies [26, 33] and manufacturers’ data showing that these concentrations blocked secretase activity and Aβ binding to LRP1 and RAGE without causing cellular toxicity. AZT used for in vitro studies was obtained from the NIH HIV Reagent Program (cat# 0952). The AZT dose used (1 μM) is within human plasma therapeutic ranges, has been shown to be nontoxic to human cells in vitro [34–36], and did not alter HBMEC viability (Fig. S1, Additional file 1). Controls included cells treated with DMSO (vehicle). After 24 h of culture, media from the transwell lower chamber were collected and any cell debris was removed by centrifugation (1000 ×g for 10 minutes at 4°C). HBMEC in the transwell upper chamber were harvested by trypsinization, washed three times with PBS, and lysed. Aβ₄₂ levels in cell lysates and culture supernatants were quantified by ELISA. For LRP1 and RAGE expression, HBMEC cultured in six-well plates were exposed to HIV-1 Tat proteins (100 ng/ml) and/or AZT (1 μM) for 24 h and levels of LRP1 and RAGE quantified by immunoblotting (Fig. S2, Additional file 1).

ELISA and Thioflavin T (ThT) binding assay

Cellular, brain, and plasma levels of Aβ₄₂, HIV-1 p24, soluble (s)LRP1, and sRAGE were quantified by ELISA (Additional file 1). For the ThT binding assay, 100 μl of Aβ₄₂ (1 mg/ml in PBS) (AnaSpec, Fremont, CA) was sonicated for 10 minutes on ice using a 50 Sonic Dismembrator (Fisher Scientific), followed by a 30-seconds quick spin in a tabletop centrifuge. Sonicated Aβ₄₂ was mixed with ThT (Biotium, Fremont, CA) in an ice-chilled Eppendorf tube, and 50 μl of this mixture (5 μM Aβ₄₂ and 50 μM ThT in PBS) dispensed into a black, clear-bottom 96-well plate (Corning, Glendale, AZ). AZT (1 μM) and/or HIV-1 transactivator of transcription (Tat) (100 ng/ml) were then added, the plate sealed with an adhesive film and immediately placed in a SpectraMax M5e plate reader (configured to maintain constant shaking, with the measurement chamber temperature set to 37°C). Fluorescence was recorded hourly over 48 hours (excitation = 450 nm, emission = 482 nm) to assess ThT binding to aggregated Aβ. Controls included blank wells (without Aβ, Tat, or AZT) and wells containing only Aβ.

Statistical analysis

Data were analyzed by Student’s t-test (two-tailed) or by one- or two-way analysis of variance followed by Tukey’s multiple-comparisons test using GraphPad Prism 10.1.1 (GraphPad Software, La Jolla, CA, USA). Sample size was calculated based on previous studies showing an expected effect size ≥ 5 when comparing brain pathologies in non-treated and treated infected animals [26]; this achieved 90% power to detect at least 5 standard deviation (SD) differences in a single factor ANOVA study, the threshold for the significance level was 0.05. Data are presented as mean ± SD.

Results

HIV infection increased brain phospho-Tau and AZT induced Tau phosphorylation in the brain SSC

Compared to mock-infected animals (PBS group), HIV infection increased frontal cortex pTau (S199) levels (by 2.34-fold, Fig. 2a–c, p = 0.017) but changes in pTau (T181) (Fig. 2a, b, d) and pTau (S396) (Fig. 2a, b, e) levels were not statistically significant. In the SSC, HIV increased Tau phosphorylation at S199 (by 4.13-fold, p < 0.0001), T181 (by 3.63-fold, p = 0.0001), and S396 (by 4.33-fold, p < 0.0001) (Fig. 2g–k). In the 3^rdBR, HIV increased Tau phosphorylation at S199 (by 3.16-fold, p = 0.0025), T181 (by 3-fold, p = 0.013), and S396 (by 2.61-fold, p = 0.014) (Fig. 2m–q). Compared to infected and untreated animals, AZT treatment of infected animals did not significantly change Tau phosphorylation in the FC, SSC, or 3^rd BR (Fig. 2). Compared to untreated controls (PBS group), AZT treatment of non-infected animals increased Tau phosphorylation in the SSC: at S199 (by 2.48-fold, Fig. 2g–i, p = 0.0015), T181 (by 2.82-fold, Fig. 2g, h, j, p = 0.006), and S396 (by 2.78-fold (Fig. 2g, h, k, p = 0.0012). HIV infection and/or AZT treatment did not alter total Tau levels (Fig. 2f, l, r).

Fig. 2.

HIV increased CNS phospho-Tau in infected animals and AZT also increased somatosensory cortex phospho-Tau. Levels of pTau (S199, T181, and S396) and total Tau in the frontal cortex (FC, a-f), primary somatosensory cortex (SSC, g-l), and a 3^rd brain region encompassing the insular cortex, auditory cortex, visual cortex, hippocampus, thalamus, and hypothalamus (3^rdBR, m-r) were analyzed by Western blot. For each brain region, a total of 6 animals were analyzed in each group, with densitometry quantification normalized to the sample’s total Tau (c-e, i-k, o-q) or β-actin (f, I, r) levels. *p < 0.05; **p = 0.0012 to 0.0098; ***p = 0.0001 to 0.0008; #p < 0.0001. Error bars represent SD

AZT reduced HIV-induced immunosuppression and viremia in blood and brain tissues

FACS quantification of human cells in engrafted animals showed similar levels of hCD4+ and hCD8+ T- cells pre infection (Table 1 , Additional file 1). There was gradual immunosuppression post-infection, with increasing loss of hCD4+ cells and increases in hCD8+ cells from week 1 to week 3 p.i.; AZT treatment significantly curtailed HIV-induced hCD4+ cells loss and hCD8+ cells increases (Table 1 , Additional file 1). Quantification of viremia (at week 3 p.i.) showed that AZT treatment also decreased CNS HIV mRNA levels: reduced CNS HIV-1 gag by 159-fold (p = 0.0022), polymerase by 126-fold (p = 0.0055), long-terminal repeats by 109-fold (p = 0.016), and tat by 2358-fold (p = 0.0037) (Fig. 3a), and decreased plasma HIV-1 p24 levels by 4.4-fold (Fig. 3b, p < 0.0001, two-tailed t-test, t = 8.871, df = 16).

Table 1.

Blood levels of human (h)CD4+ T-cells and hCD8+ T-cells (mean ± SD) before and after infection

	PBS	HIV	HIV+AZT	AZT	p-value (PBS vs. HIV)	p-value (PBS vs. HIV+AZT)	p-value (HIV vs. HIV+AZT)	p-value (HIV vs. AZT)	p-value (HIV+AZT vs. AZT)
hCD3+hCD4+ (%hCD45+ cells)
Pre-infection	61 ± 10	57.6 ± 20.83	61.5 ± 10.8	55 ± 18.5	0.95	1	0.937	0.98	0.77
Week 1 post-infection	72.5 ± 12	48.32 ± 15.65	53.4 ± 18.3	73.76 ± 8.1	0.0033	0.03	0.874	0.002	0.017
Week 2 post-infection	54.7 ± 18	15.3 ± 8.2	51.2 ± 6.6	54.6 ± 28	<0.0001	0.952	<0.0001	<0.0001	0.957
Week 3 post-infection	55.1 ± 13	9.95 ± 6.32	56.63 ± 15.34	57.6 ± 11.45	<0.0001	0.996	<0.0001	<0.0001	0.999
hCD3+hCD8+ (%hCD45+ cells)
Pre-infection	28.3 ± 7.6	36.75 ± 12	36.75 ± 7.74	25.4 ± 13.3	0.63	0.63	1	0.38	0.384
Week 1 post-infection	24 ± 9.75	41.2 ± 24	44.53 ± 18.5	18.52 ± 5.6	0.08	0.026	0.96	0.01	0.0027
Week 2 post-infection	39.7 ± 18.6	85.5 ± 8.1	48.77 ± 6.6	40 ± 27.84	<0.0001	0.575	<0.0001	<0.0001	0.61
Week 3 post-infection	40.43 ± 8.75	84.7 ± 6.33	43.38 ± 15.32	40.46 ± 9.2	<0.0001	0.975	<0.0001	<0.0001	0.976

Fig. 3.

Effects of HIV infection and AZT treatment on blood and brain viremia and Aβ levels. Levels of HIV-1 gag, pol, LTR, and tat genes in animals brain tissues were quantified by real-time PCR and normalized to levels of human CD45+ cells in each brain tissue sample (a). Levels of HIV-1 gag p24 (b) and Aβ₄₂ in the plasma (c) and brain tissues (d) of infected and non-infected animals, AZT treated and non-treated, were quantified by ELISA. All samples were collected at week 3 post-infection. For each panel, n = 8 to 9 animals in each group. For real-time PCR, each animal sample was analyzed in triplicate and for ELISA, each animal sample was analyzed in duplicate. **p = 0.0028; ***p = 0.0009; *p < 0.0001. Error bars represent SD

HIV infection increased blood and CNS Aβ and AZT treatment had no effect on HIV-induced Aβ

Compared to control animals (non-infected with and without AZT), HIV infection increased blood/serum Aβ₄₂ levels by 3.12 to 4.07-fold (Fig. 3c) and increased brain Aβ₄₂ levels by 8.44 to 11.15-fold (Fig. 3d). AZT treatment did not induce Aβ in non-infected controls and did not significantly change HIV-induced Aβ in infected animals (Fig. 3c, d).

Dysregulation of CNS LRP1 transcription and expression in HIV infection and AZT treatment

Transcription: Compared to non-infected animals (PBS group), HIV infection decreased LRP1 transcription in the FC (by 2.57-fold), SSC (by 3.62-fold), and in the 3^rdBR (by 5-fold) (Fig. 4a–c, p < 0.0001). AZT treatment of infected animals did not change LRP1 transcript levels in the FC, SSC, or the 3^rdBR, compared to infected and non-treated animals. Compared to untreated (PBS) controls, AZT treatment of non-infected animals decreased LRP1 transcription in the FC (by 1.3-fold, p = 0.04), SSC (by 2-fold, p = 0.0003), and in the 3^rdBR (by 2.5-fold, p < 0.0001) (Fig. 4a–c)

Fig. 4.

HIV and AZT decreased CNS LRP1 and increased RAGE transcription in infected and treated animals. Levels of LPR1 (a-c) and RAGE (d-f) mRNA in the frontal cortex (FC; a, d), primary somatosensory cortex (SSC; b, e), and a 3^rd brain region encompassing the insular cortex, auditory cortex, visual cortex, hippocampus, thalamus, and hypothalamus (3^rdBR; c, f) were quantified by real-time PCR. For each panel, n = 8 animals in each group, each animal sample was analyzed in triplicate. *p < 0.05; **p = 0.0017 to 0.003; ***p = 0.0001 to 0.0008; #p < 0.0001. Error bars represent SD

Expression: Compared to non-infected animals, HIV infection decreased LRP1 expression in the FC (by 1.73-fold, Fig. 5a–c, p = 0.0015), SSC (by 2.48-fold, Fig. 5e–g, p = 0.0003) and in the 3^rdBR (by 1.8-fold, Fig. 5i–k, p = 0.0006). Compared to infected and non-treated animals, AZT treatment of infected animals did not change LRP1 expression in the FC, SSC, or the 3^rdBR, (Fig. 5). Compared to the PBS control group, AZT treatment of non-infected animals decreased LRP1 expression, but this decrease only reached statistical significance in the SSC (Fig. 5e–g, p = 0.04).

Fig. 5.

HIV and AZT decreased CNS LRP1 and increased RAGE expression in infected and treated animals. Levels of LPR1 (a-c, e-g, i-k) and RAGE (a, b, d, e, f, h, i, j, l) proteins in the frontal cortex (FC, a-d), primary somatosensory cortex (SSC, e-h), and a 3^rd brain region encompassing the insular cortex, auditory cortex, visual cortex, hippocampus, thalamus, and hypothalamus (3^rdBR, i-l) were analyzed by Western blot followed by densitometry quantification normalized to samples’ β-actin levels. Representative data from 2 immunoblots (6 animals in each group) for each brain region are shown. *p < 0.05; **p = 0.0015 to 0.007; ***p = 0.0001 to 0.0006; #p < 0.0001. Error bars represent SD

Dysregulation of CNS RAGE transcription and expression in HIV infection and AZT treatment

Transcription: Compared to non-infected animals, HIV infection increased RAGE transcription in the FC (by 3.14-fold, p < 0.0001), SSC (by 2.33-fold, p < 0.0001), and in the 3^rdBR (by 3.89-fold, p = 0.0004) (Fig. 4d–f). Compared to infected and non-treated animals, AZT treatment of infected animals did not significantly change RAGE transcript levels in the FC, SSC, or the 3^rdBR (Fig. 4d–f). Compared to the PBS control group, AZT treatment of non-infected animals increased RAGE transcription in the FC (by 2.6-fold, p = 0.003), SSC (by 1.93-fold, p = 0.0008), and in the 3^rdBR (by 2.16-fold) (Fig. 4d–f)

Expression: Compared to the PBS group, HIV infection increased RAGE expression in the FC (by 2.44-fold, Fig. 5a, b, d), SSC (by 4.18-fold, Fig. 5e, f, h), and in the 3^rdBR (by 3.25-fold, Fig. 5i, j, l) (p < 0.0001). Compared to infected and non-treated animals, AZT treatment of infected animals did not change RAGE expression in the FC, SSC, or the 3^rdBR (Fig. 5). Compared to untreated (PBS) controls, AZT treatment of non-infected animals increased RAGE expression in the FC (by 1.9-fold, Fig. 5a, b, d; p = 0.007), SSC (by 4.36-fold, Fig. 5e, f, h), and in the 3^rdBR (by 3-fold, Fig. 5i, j, l) (p < 0.0001).

HIV and AZT dysregulate LRP1 and RAGE proteolytic cleavage

Proteolytic cleavage of full-length LRP1 or RAGE on the brain endothelial membrane release truncated isoforms (sLRP1 and sRAGE) into extracellular space and impair the ability of remaining cell-bound fragments to bind and transport Aβ [37–39]. ELISA quantification of sLRP1 and sRAGE in animals’ serum showed that compared to non-infected (PBS) animals, blood sLRP1 increased in infected animals (by 2.98-fold), in infected animals treated with AZT (by 3-fold), and in non-infected animals treated with AZT (by 2.6-fold) (Fig. 6a, p < 0.0001). Compared to the PBS group, blood sRAGE decreased in the HIV group (by about 2-fold), in infected animals treated with AZT (by 2.23-fold), and in non-infected animals treated with AZT (by 2.4-fold) (Fig. 6b, p < 0.0001).

Fig. 6.

HIV and AZT increased sLRP1 and decreased sRAGE in infected and treated animals. Levels of sLRP1 (a) and sRAGE (b) in animals’ serum were quantified by ELISA. All samples were collected at week 3 post-infection. For each panel, n = 8 animals in each group, each animal sample was analyzed in duplicate. *p < 0.02; **p = 0.003; #p < 0.0001. Error bars represent SD

HIV-1 Tat dysregulate brain endothelial LRP1 and RAGE expression

Western blot analysis showed that compared to untreated controls, exposure of primary HBMEC to Tat proteins decreased LRP1 expression by 1.6-fold and increased RAGE by 1.4-fold; Tat+AZT decreased LRP1 by 1.95-fold and increased RAGE by 1.85-fold, whereas AZT alone did not significantly change LRP1 or RAGE levels (Fig. S2, Additional file 1).

HIV infection and AZT treatment decreased CNS NeuN transcription and expression

Real-time PCR quantification of the neuronal nuclei marker NeuN in animals’ brain tissues showed that compared to non-infected animals, NeuN mRNA levels decreased in HIV-infected animals [decreased in the FC by 2.26-fold (Fig. 7a, p = 0.0002) and in the SSC by 16.24-fold (Fig. 7b, p = 0.0008)], decreased in infected and non-infected animals treated with AZT [in the FC by 2-fold (Fig. 7a, p = 0.0003) and in the SSC by 4.6-fold (Fig. 7b, p = 0.005)]. Compared to non-infected animals, NeuN expression in infected and AZT-treated animals decreased by 1.2 to 1.8-fold (Fig. 8a–d).

Fig. 7.

HIV and AZT decreased CNS NeuN and claudin-5 transcription in infected and treated animals. Levels of NeuN (a, b) and claudin-5 (c, d) mRNA in the frontal cortex (FC; a, c) and primary somatosensory cortex (SSC; b, d) were quantified by real-time PCR. For each panel, n = 8 animals in each group, each animal sample was analyzed in triplicate. **p = 0.006 to 0.0083; ***p = 0.0002 to 0.0009; *p < 0.0001. Error bars represent SD

Fig. 8.

HIV and AZT decreased CNS NeuN, claudin-5, and ZO-1 expression in infected and treated animals. Levels of NeuN (a-d), claudin-5 (a-c, e), and ZO-1 (a-c, f) proteins in brain tissues (frontal cortex) were analyzed by Western blot followed by densitometry quantification (using data from 3 immunoblots, 9 animals in each group) normalized to samples’ β-actin levels. *p < 0.05, compared to untreated controls. Error bars represent SD

HIV infection and AZT treatment decreased CNS endothelial tight junction proteins

Real-time PCR quantification of claudin-5 in animals’ brain tissues showed that compared to non-infected animals, claudin-5 mRNA levels decreased in HIV-infected animals [in the FC by 2.3-fold (Fig. 7c, p = 0.0002) and in the SSC by 7.6-fold (Fig. 7d, p = 0.0009)], decreased in infected animals treated with AZT [in the FC by 2-fold (Fig. 7c, p = 0.0004) and in the SSC by 3.3-fold (Fig. 7d, p = 0.0083)], and in non-infected animals treated with AZT [in the FC by 1.97-fold (Fig. 7c, p = 0.0009) and in the SSC by 3.5-fold (Fig. 7d, p = 0.0065)]. Western blot analysis showed that compared to non-infected animals, claudin-5 expression in infected and AZT-treated animals decreased by 1.4 to 1.7-fold although this did not reach statistical significance (Fig. 8a–c, e), and ZO-1 expression decreased by 1.3 to 2-fold (Fig. 8a–c, f).

AZT increased brain endothelial Aβ uptake and retention

Transwell quantification of Aβ₄₂ endothelial uptake/retention and trans-endothelial transport using HBMEC from two different donors showed that, compared to untreated controls, AZT increased Aβ endothelial uptake by 2.6 to 4.4-fold (Fig. 9a, c, p < 0.0001). LRP1 and RAGE inhibitors decreased Aβ uptake, but only partially reduced AZT-induced Aβ uptake. Compared to uptake in untreated controls, AZT increased endothelial Aβ uptake by 2-fold in the presence of LRPinh or RAGEinh (Fig. 9a, c). AZT did not significantly alter Aβ transendothelial transport (Fig. 9b, d); both LRPinh or RAGEinh significantly decreased Aβ transendothelial transport (by 2.4 to 2.7-fold in the presence of AZT and by 2.6 to 4.4-fold in the absence of AZT) (Fig. 9b, d).

Fig. 9.

AZT increased brain endothelial Aβ uptake and retention. HBMEC from two different donors were cultured to confluence in a transwell system and treated with Aβ₄₂ (10 µM) in the presence or absence of AZT (1 μM) and/or inhibitors (inh) of LRP1 and RAGE. Levels of Aβ₄₂ in trypsinized cells (upper chamber of the transwell) (a, c) and in the lower chamber media (b, d) were quantified by ELISA as detailed in Methods. For each of the two donors (biological replicates, cells at passage 4) four to five technical replicates were used for each treatment condition. *p < 0.033; **p = 0.0017 to 0.0042; ***p = 0.0001 to 0.001; #p < 0.0001. Error bars represent SD

HIV-1 Tat and AZT increased Aβ aggregation

In two independent experiments, HIV-1 Tat proteins increased Aβ₄₂ aggregation at 24 h and 48 h respectively by 34 and 32% (Fig. 10a, b) and by 43.2 and 13% (Fig. 10c, d). Compared to Aβ alone, AZT did not induce a major change in Aβ aggregation, but AZT+Tat increased Aβ aggregation at 24 h and 48 h respectively by 70 and 51% (Fig. 10a, b) and by 54% and 45.3% (Fig. 10c, d) (p < 0.0001). Compared to AZT alone, AZT+Tat also increased Aβ aggregation at 24 h and 48 h respectively by 42 and 32% (Fig. 10a, b) and by 43 and 40% (Fig. 10c, d) (p < 0.0001).

Fig. 10.

HIV-1 Tat and AZT increased Aβ aggregation. In two independent experiments, Tat increased Aβ₄₂ aggregation by 34% (a, b) and 43.2% (c, d) at 24 h, and by 32% (a, b) and 13% (c, d) at 48 h. AZT+Tat increased Aβ₄₂ aggregation by 70% (a, b) and by 54% (c, d) at 24 h, and by 51% (a, b) and 45.3% (c, d) at 48 h, compared to Aβ₄₂ alone. AZT+Tat increased Aβ₄₂ aggregation by 42% (a, b) and by 43% (c, d) at 24 h and by 32% (a, b) and 40% (c, d) at 48 h, compared to AZT alone. For each panel, each experimental condition included three technical replicates. *p < 0.03; **p = 0.002 to 0.007; #p < 0.0001. Error bars represent SD

Discussion

PWH show increased incidence and prevalence of HAND and other CNS comorbidities, compared to demographically matched controls [1–3]. HIV and viral proteins have been shown to damage brain cells and the BBB, and these viral-induced injuries contribute to the pathogenesis of HAND [4–6]. Although many PWH with HAND have controlled viremia and have been on long-term treatment, it is not known if cART or some ARV drugs are contributing factors. It has also been suggested that legacy effects from initial HIV-induced brain injury and older HIV medications such as AZT could be contributing factors [17]. The current WHO HIV treatment guideline list AZT-based regimens as preferred and alternative 2^nd line cART for adults, adolescents, and children [19]. Over 120 low and middle income countries (87% of the 40.8 million PWH) follow WHO HIV treatment guidelines [40]; with persisting poor adherence, PWH failing 1^st line cART remain a major issue, and those who fail 1^st line therapy are on 2^nd line cART, several of which are AZT-based regimens [19, 41, 42]. Thus, although currently recommended cART regimens in high income countries don’t include AZT [14, 18], this drug is still in use in numerous countries and study of its effects is important.

Review of CNS pharmacokinetics and pharmacodynamics of different ARVs and cART regimens showed that AZT was among the ARVs with high CNS penetration and cerebrospinal fluid/plasma ratios [43, 44]. In a prospective study of PWH on various cART regimens, AZT-based regimens were associated with most adverse reactions, including CNS adverse effects [45]. Previous studies showed increased CNS proteopathy and AD-like pathologies in HIV-infected humans [8–11] and animal models [26], including increased accumulation of amyloid precursor protein, Aβ deposits and Tau hyperphosphorylation associated with neurocognitive impairment [8–11]. Our current study in a validated HIV animal model confirmed these findings and showed that AZT treatment did not significantly change CNS Aβ or phospho-Tau levels. AZT did not alter trans-endothelial Aβ transport, but it significantly decreased the transcription and expression of the neuronal nuclei marker NeuN, significantly decreased the transcription of the endothelial tight junction proteins claudin-5, also reduced claudin-5 and ZO-1 expression, and significantly increased endothelial Aβ uptake and retention; this could affect neuronal and brain vascular health and function. In fact, intracellular accumulation of Aβ₄₂ induce apoptosis [46] and intracellular Aβ₄₂ can further undergo oligomerization within cells, which accentuate Aβ-induced cellular toxicity [47]. Studies in humans and AD animal models showed that intracellular accumulation of Aβ₄₂ is an early event in AD development and predicts synaptic dysfunction and neurodegeneration [47, 48]. The effects of AZT-induced increase in endothelial Aβ uptake/retention are not known. Our future studies will investigate whether such ARV-induced intracellular accumulation of Aβ₄₂ in brain endothelial cells effect the brain endothelium properties and function. Our current study also showed that AZT increased Aβ aggregation in the presence of Tat. Tat proteins are secreted by infected cells in vitro and in vivo and can be detected in body fluids and tissues of HIV-infected humans [49–51]. The Tat concentration used in this study (100 ng/ml) has been used in previous studies [27, 50] and is within concentration ranges detected in PWH [49–51]. The fact that in the presence of Tat, AZT increased Aβ aggregation suggests an increased risk of Aβ deposit/accumulation in HIV infection and AZT treatment.

Our present study showed that increased CNS Aβ following HIV infection was associated with increased RAGE and decreased LRP1, two endothelial receptors that regulate Aβ BBB transport. Aβ is produced by brain cells and LRP1 binds CNS Aβ on the BBB abluminal side and induces its efflux/clearance to maintain brain Aβ homeostasis [20–22]. Aβ can also be produced by peripheral cells and RAGE binds peripheral Aβ on the BBB luminal side and induces its influx into the brain [21, 23, 24]. Studies in humans [20, 52–54] and AD animal models [20, 53] showed that AD was associated with drastic LRP1 reduction and major RAGE increase in brain tissues and capillaries, and this correlated with increased CNS Aβ and severe AD pathology [55, 56], increased Aβ influx into the brain, neuroinflammation, Tau hyperphosphorylation, and neurodegeneration [57, 58]. Our current data suggests that HIV and viral proteins may induce brain Aβ pathology by impairing brain endothelial Aβ transporters. HIV-induced decrease in LRP1 levels would impair and diminish CNS Aβ efflux and an increase in RAGE would increase influx of peripheral Aβ into the brain.

How AZT increases brain endothelial Aβ uptake/retention is not known. Dysregulation of endothelial Aβ transporters could be contributing factors. Although AZT treatment did not alter plasma or brain Aβ levels, it significantly decreased CNS LRP1 and increased RAGE and accentuated Tat-induced LRP1 decrease and RAGE increase in brain endothelial cells. AZT- or AZT+Tat- induced upregulation of RAGE (endothelial Aβ influx transporter) could contribute to increased endothelial Aβ uptake, impairment in BBB properties and function. AZT- or AZT+Tat- induced downregulation of LRP1 (endothelial Aβ efflux transporter) could further reduce endothelial Aβ CNS efflux and increase Aβ retention. Proteolytic cleavage of full-length LRP1 and RAGE releases sLRP1 and sRAGE into circulation, leaving truncated cell-bound LRP1 fragments that lack the LRP1 binding domain and cannot bind or transport/efflux Aβ [37, 59], and truncated cell-bound RAGE fragments that lack the extracellular domain essential for RAGE signaling and function [38, 39]. The current study showed that HIV infection drastically increased sLRP1 and decreased sRAGE, suggesting that dysregulation in LRP1 and RAGE proteolytic cleavage may contribute to the functional impairment of these brain Aβ transporters in HIV infection and increased CNS Aβ. Significantly, AZT treatment also drastically increased sLRP1 and decreased sRAGE in both infected and non-infected animals. sLRP1 can bind Aβ but loses the ability to endocytose and transport Aβ [37, 59]. Therefore, in addition to the fact that HIV- and AZT-induced decrease in LRP1 transcription and expression would reduce Aβ efflux, HIV- and AZT-induced increase in LRP1 proteolytic cleavage (sLRP1) will further reduce Aβ endocytosis, transport and CNS efflux. sRAGE antagonizes the activation and function of full-length RAGE and reduces Aβ BBB transport and CNS influx [60, 61], and inhibits Aβ aggregation [62]. Thus, in addition to the fact that HIV- and AZT-induced RAGE upregulation would increase Aβ CNS influx, HIV- and AZT-induced decrease in sRAGE could decrease the scavenging and clearance of circulating Aβ, and further increase the binding of peripheral Aβ to cellular RAGE and Aβ influx into the brain. However, in our current study, AZT-induced sLRP1 increase and sRAGE decrease were not associated with altered Aβ levels, and compared to non-treated animals, AZT-treatment of infected and non-infected animals did not change plasma or brain Aβ levels, suggesting that HIV infection was primarily responsible for the increased Aβ. It is not known if other transporters could have played a role. ATP-binding cassette (ABC) transporters such as P-glycoprotein/ABCB1, ABCG2, ABCA1, ABCA2, ABCA7, and ABCC1 have been shown to play a role in Aβ efflux and clearance [63, 64]. AZT has been shown to increase ABCB1, ABCC4, and ABCC5 in human macrophages [65], but it is not known if AZT also alter/increase these efflux transporters in brain cells, or whether these transporters may have facilitated CNS Aβ efflux in AZT-treated animals, despite increased RAGE and decreased LRP1.

Limitations

Because we did not have brain samples from PWH who had specifically been treated with AZT, we don’t know if AZT also dysregulates LRP1 and RAGE expression and proteolytic cleavage in humans. Soluble amyloid oligomers and protofibrils are considered the most toxic to cells, although the ThT fluorescence assay is a robust test for detecting amyloid aggregation, it cannot separately quantify amyloid fibrils and other forms of Aβ aggregates (oligomers and protofibrils). With the total number of samples from the four animal groups (8–10 animals per group) and analysis of 3 different regions (FC, SSC, 3^rdBR) in each animal, we could not separately analyze more specific brain regions and had to combine several brain regions into the 3^rdBR samples. HIV infection is now a chronic condition and PWH can live for decades on cART; thus, our current study of hu-PBL-NSG mice HIV-infected and AZT-treated for three weeks (equivalent of 2.5 human years) [66] may not fully recapitulate the effects of long-term/chronic infection and cART. Because we could not perform behavioral assessment, it is not known if any HIV- and/or AZT-induced effect observed in our current study could be associated with altered cognition. There are several other ARV drugs and different drug classes, and their effects on these Aβ transporters, CNS Aβ and Tau pathology remain to be examined. Our future studies will include older animals, longer-term HIV infection and treatment, including treatment with newer ARVs, and assessment of both pathological/cerebrovascular and behavioral effects.

Conclusions

Results of the current animals’ studies suggests that HIV increased CNS pTau and Aβ₄₂, and blood Aβ₄₂. Both HIV infection and AZT treatment independently decreased brain endothelial tight junction proteins (claudin-5 and ZO-1) and NeuN, and dysregulated LRP1 and RAGE transcription, expression, and proteolytic cleavage. This dysregulation of Aβ transporters could contribute to altered Aβ clearance in HIV infection (Fig. 11). In addition to decreasing tight junction proteins and neuronal marker NeuN in infected and treated animals, AZT increased Aβ uptake and retention; in the presence of Tat, AZT increased Aβ aggregation and accentuated Tat-induced dysregulation of endothelial LRP1 and RAGE. This may suggest an increased risk of Aβ deposit/accumulation in HIV infection and AZT treatment and suggest that AZT may contribute to neuronal and BBB impairment/dysfunction (Fig. 11).

Fig. 11.

Schematic illustration of HIV and/or AZT-induced effects. In optimal health conditions, there is reduced CNS tau phosphorylation (pTau), RAGE, and Aβ₄₂ production, and increased LRP1, which would increase CNS Aβ efflux, reduce influx and maintain CNS Aβ homeostasis. HIV infection increases Aβ₄₂ production and pTau (S199, T181, S396) (in the FC, SSC, and 3^rdBR). AZT also increase SSC pTau. HIV and AZT dysregulate LRP1 and RAGE transcription, expression, and proteolytic cleavage (decrease LRP1 and increase sRLP1, increase RAGE and decrease sRAGE), which would increase Aβ CNS influx and reduce Aβ efflux, resulting in accumulation of Aβ in the brain and neuronal injury. HIV and AZT also decrease endothelial tight junctions (TJs) proteins and AZT increase endothelial Aβ uptake and retention, which would increase the risk of BBB injury and dysfunction. ↑: increase; ↓: decrease

Electronic supplementary material

Below is the link to the electronic supplementary material.

Below is the link to the electronic supplementary material.

Abbreviations

ABC

ATP-binding cassette

AD

Alzheimer’s disease

Aβ

Amyloid-beta

cART

Combination antiretroviral therapy

ARVs

Antiretrovirals

AZT

Azidothymidine

BBB

Blood-brain barrier

PBS

Phosphate-buffered saline

CNS

Central nervous system

DMSO

Dimethyl sulfoxide

ELISA

Enzyme-linked immunosorbent assay

FACS

Fluorescence activated cell sorting

FC

Frontal cortex

SSC

Somatosensory cortex

3rdBR

3rd brain region (encompassing the insular, auditory, and visual cortex, hippocampus, thalamus, and hypothalamus)

HAND

HIV-associated neurocognitive disorders

HBMEC

Human brain microvascular endothelial cells

HIV-1

Human immunodeficiency virus type-1

i.p.

Intraperitoneal

LRP1

Low-density lipoprotein receptor–related protein-1

NSG

NOD/scid–IL-2 Rγcnull

PBL

Peripheral blood lymphocytes

PWH

People living with HIV

RAGE

Receptor for advanced glycation end products

SD

Standard deviation

sLRP1

Soluble LRP1

sRAGE

Soluble RAGE

Tat

Transactivator of transcription

ThT

Thioflavin T

WHO

World Health Organization

ZO-1

Zonula occludens-1

Author contributions

BB carried out NSG mice infection, AZT treatment and regular monitoring of animals’ weight, specimen collection and processing, FACS, ELISA, real-time PCR, Western blot, HBMEC cultures, TEER, Aβ⁴² transwell assays, and participated in the making of figures, data analysis and writing the methods. NA participated in HBMEC culture and Western blot. GDK conceived, designed, and supervised the study, participated in mice infection, specimen collection, data analysis, making of figures and tables, and wrote the manuscript.

Funding

This work was supported by grants from the NIH, NIMH 1R21 MH123303 and 1R01 MH132517 (to GDK). The funder had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Data availability

The data generated and/or analyzed during the current study are included in this published article and are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.