Antiretroviral drugs induce oxidative stress and neuronal damage in the central nervous system
- First Online:
- Cite this article as:
- Akay, C., Cooper, M., Odeleye, A. et al. J. Neurovirol. (2014) 20: 39. doi:10.1007/s13365-013-0227-1
- 1.1k Views
HIV-associated neurocognitive disorder (HAND), characterized by a wide spectrum of behavioral, cognitive, and motor dysfunctions, continues to affect approximately 50 % of HIV(+) patients despite the success of combination antiretroviral drug therapy (cART) in the periphery. Of note, potential toxicity of antiretroviral drugs in the central nervous system (CNS) remains remarkably underexplored and may contribute to the persistence of HAND in the cART era. Previous studies have shown antiretrovirals (ARVs) to be neurotoxic in the peripheral nervous system in vivo and in peripheral neurons in vitro. Alterations in lipid and protein metabolism, mitochondrial damage, and oxidative stress all play a role in peripheral ARV neurotoxicity. We hypothesized that ARVs also induce cellular stresses in the CNS, ultimately leading to neuronal damage and contributing to the changing clinical and pathological picture seen in HIV-positive patients in the cART era. In this report, we show that ARVs are neurotoxic in the CNS in both pigtail macaques and rats in vivo. Furthermore, in vitro, ARVs lead to accumulation of reactive oxygen species (ROS), and ultimately induction of neuronal damage and death. Whereas ARVs alone caused some activation of the endogenous antioxidant response in vitro, augmentation of this response by a fumaric acid ester, monomethyl fumarate (MMF), blocked ARV-induced ROS generation, and neuronal damage/death. These findings implicate oxidative stress as a contributor to the underlying mechanisms of ARV-induced neurotoxicity and will provide an access point for adjunctive therapies to complement ARV therapy and reduce neurotoxicity in this patient population.
KeywordsAntiretroviralFumaric acid esterHIVHIV-associated neurocognitive disorderMacaqueOxidative StressReactive oxygen speciesSIV
Despite introduction of combination antiretroviral therapy (cART), HIV-associated neurocognitive disorder (HAND) continues to affect approximately 50 % of HIV(+) patients (Dore et al. 1999; Heaton et al. 2010). Furthermore, in the cART era, the underlying neuropathology has shifted from overt subcortical involvement to a more insidious cortical damage (Gannon et al. 2011). Various factors, such as poor adherence to drug regimen, emergence of resistant virus species, and residual viral DNA in the central nervous system (CNS), may contribute to these changes (Gannon et al. 2011). However, another likely contributor to HAND in the cART era is the virtually unstudied potential for antiretroviral (ARV)-related toxicity in the CNS. cART has decreased HIV-related morbidity and mortality by limiting T cell loss and controlling opportunistic infections. However, cART regimens are associated with potentially serious side effects, including dyslipidemia, lipohypertrophy, and increased risk of atherosclerosis (Vidal et al. 2010). Additionally, cART-associated toxicity in the peripheral nervous system is well documented, and it is likely that cART would trigger similar responses in the CNS. Pharmacokinetic studies suggest limited ARV penetrance into the CNS and indicate low cerebrospinal fluid (CSF) and parenchymal drug concentrations (Yilmaz et al. 2004; Yilmaz et al. 2009). However, direct blood–brain barrier (BBB) compromise by viral proteins and neuroinflammation and indirect BBB impairment due to concomitant factors, such as coexisting infections, can lead to increased CSF and parenchymal drug concentrations. Thus, the impact of ARVs in the CNS of HIV(+) patients is clinically relevant and must be examined.
Initial cART usually includes two nucleoside/nucleotide reverse-transcriptase inhibitors (NRTIs) in combination with a non-nucleoside reverse-transcriptase inhibitor (nNRTI) or with a protease inhibitor (PI) boosted with a low dose of a second PI, Ritonavir. NRTIs and nNRTIs bind to the HIV reverse-transcriptase enzyme and inhibit proviral DNA synthesis; PIs inhibit viral proteases needed for virus maturation and assembly. Despite some crossover, certain ARV classes are more highly associated with particular side effects and toxicities than are other classes. PIs alter lipid metabolism and induce the endoplasmic reticulum stress response in macrophages, linking PIs to increased risk of atherosclerosis (Touzet and Philips 2010). NRTIs inhibit DNA polymerase γ and lead to decreased mitochondrial DNA, loss of mitochondrial membrane potential, and oxidative phosphorylation, consequently precipitating oxidative stress (Nolan and Mallal 2004). Previous studies exploring possible side effects of ARVs in the CNS are scarce and mostly involve cell lines (Cui et al. 1997). Given the mutations and aberrations in immortalized cell lines, these studies may not reflect the ARV toxicity potentially occurring in biological settings. Peripheral dorsal root ganglia neurons are the only primary cell type of neural lineage previously studied for ARV toxicity (Werth et al. 1994). In this report, we examined effects of ARVs in primary CNS neurons both in vivo and in vitro. Our findings suggest that cART induces oxidative stress and neurotoxicity in the CNS, and that the patients on long-term cART regimens would benefit from adjuvant therapies that include antioxidant strategies to overcome deleterious effects of cART in the CNS.
Materials and methods
Chemicals and reagents
Chemicals and reagents comprise the following: (1) AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH, antiretroviral reagents; (2) Abcam (Cambridge, MA), mouse monoclonal NAD(P)H/quinone oxidoreductase-1 (NQO-1) antibody (A180) and mouse monoclonal synaptophysin antibody (SY38); (3) BioRad (Hercules, CA), Biosafe Coomassie stain, immunoblot polyvinylidene fluoride (PVDF) membrane and prestained broad range molecular weight ladder; (4) Cell Signaling Technology (Danvers, MA), rabbit polyclonal antibody raised against cleaved caspase-3; (5) Citifluor, Ltd. (London, UK), citifluor AF1. Covance (Princeton, NJ), mouse monoclonal microtubule-associated protein 2 (MAP2) antibody (SMI-52); (6) Dako (Carpinteria, CA), rabbit polyclonal glial fibrillary acidic protein (GFAP) antibody (Z0334); (7) Enzo Life Sciences (Farmingdale, NY), rabbit polyclonal antibody to heme-oxygenase-1 (HO-1); (8) Frontier Scientific (Logan, UT), Sn(IV) mesophorphyrin IX dichloride (SnMP); (9) Jackson ImmunoResearch Labs (West Grove, PA), fluorescein isothiocyanate-conjugated goat anti-mouse IgG and Cy3-conjugated goat anti-rabbit IgG secondary antibodies; (10) Invitrogen (Carlsbad, CA), Dulbecco's modified Eagle's medium (DMEM), tetramethyl rhodamine methyl ester, goat anti-mouse beta-lactamase TEM-1 conjugate, fluorocillin green substrate, dihydroethidium (DHE), 4′,6-diamidino-2-phenylindole (DAPI), neurobasal media, and B27 supplement; (11) New England Biolabs (Ipswich, MA), tyramide amplification system; (12) Peptide International (Louisville, KY), poly-l-lysine; (13) Sigma Aldrich (St. Louis, MO), carbonyl cyanide m-chlorophenyl hydrazone, cytosine arabinoside (Ara-C), fetal bovine serum (FBS), monomethylfumarate (MMF), oligomycin, propidium iodide, and PI cocktail; (14) ScyTek Labs (Logan, UT), normal antibody diluent; (15) Thermo Scientific (Waltham, MA), goat anti-rabbit horse radish peroxidase (HRP) antibody and goat anti-mouse HRP antibody, SuperSignal West Dura extended duration substrate; and (16) Tocris Bioscience (Ellisville, MO), thapsigargin. The antibody against calpain-cleaved spectrin was a generous gift from Dr. Robert Siman (The Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA).
Primary cortical neuroglial cultures
Primary rat cortical neuroglial, pure neuronal, and pure astrocytic cultures were isolated from embryonic day 17 Sprague Dawley rat pups, with modifications of protocols previously described (Wilcox et al. 1994). Briefly, the cortical cell suspensions isolated from rat pups following standard protocols were plated on poly-l-lysine-coated T25 tissue culture flasks (2 × 106 cells/flask), 96-well tissue culture plates (0.5 × 105 cells/well), or glass coverslips (0.25 × 106 cells/well). These cultures were maintained in neurobasal media with B27 supplement at 37 °C with 5 % CO2 for the generation of neuroglial cultures, as described previously (Wang et al. 2010; Akay et al. 2011b; White et al. 2011). Half of the media was replaced with fresh media every 7 days and the experiments were performed at 21 days in vitro (DIV), at which time the cultures contain approximately 85–90 % neurons and 10–15 % astrocytes/glia. For the preparation of pure neuronal cultures, cultures were treated with 10 μM Ara-C 48 h after plating and maintained in neurobasal media with B27 supplement at 37 °C with 5 % CO2; at 21 DIV, the age at which the experiments were conducted, no astrocytes are detectable by staining for GFAP. Pure astrocytic cultures were prepared by initially plating the cortical cell suspensions in 10 % FBS in DMEM and maintaining at 37 °C with 5 % CO2 for 7–12 days, after which time astrocytes constitute more than 90 % of the cultures. These pure astrocytic cultures are then replated in T25 flask or poly-l-lysine-coated glass coverslips in DMEM/10 % FBS and the experiments are conducted 3–5 days after replating.
SIV/pigtail macaque model of HIV CNS disease
Juvenile pig-tailed macaques (Macaca nemestrina) were inoculated with SIV/DeltaB670 and SIV/17E-Fr, as described previously (Zink et al. 1999). Beginning 12 days after inoculation, animals were treated daily with a four ARV drug combination (cART) until necropsy (range, days 161–175). The treatment consisted of the NRTI tenofovir (Gilead) at 30 mg/kg subcutaneously every day; the PIs saquinavir (Roche) and atazanavir (Bristol- Myers Squibb) at 205 and 270 mg/kg orally twice a day, respectively; and the integrase inhibitor L-870812 (Merck)(Hazuda et al. 2004) at a dose of 10 mg/kg orally twice a day(Zink et al. 2010). The tenofovir dose was determined on the basis of previous studies(Tsai et al. 1997), whereas atazanavir and saquinavir doses were determined by pharmacokinetic experiments conducted in pig-tailed macaques, and reflected those that resulted in the same area under the curve as detected in humans treated with atazanavir and saquinavir (Zink et al. 2010). The dose of the integrase inhibitor was based on a previous study conducted in rhesus macaques(Hazuda et al. 2004).
Rodent model of antiretroviral-induced neurotoxicity
All surgical procedures were performed with the approval of the Institutional Animal Care and Use Committee. Adult male Sprague Dawley rats were catheterized via jugular vein, as described previously (Thrivikraman et al. 2002). The drug cocktail composed of AZT (100 mg kg−1 day−1), ritonavir (20 mg kg−1 day−1), and saquinavir (25 mg kg−1 day−1) was administered twice daily for 7 days by continuous intravenous injection. Each drug dose was based on previously published studies in rodents (Shibata et al. 2002; Huisman et al. 2003; Manda et al. 2010; Pistell et al. 2010; Waring et al. 2010; Yang et al. 2010; du Plooy et al. 2011; Fontes et al. 2011; Lledo-Garcia et al. 2011; Wagner et al. 2011; Mak et al. 2013; Reyskens and Essop 2013a, b; Reyskens et al. 2013). The catheters were flushed with 0.3 ml heparin (50 IU/ml) in PBS until the end of treatments. At time of euthanasia, preceding decapitation and tissue harvest, catheter patency was reverified by response to pre-euthanasia sedation (100 mg/kg ketamine per 10 mg/kg xylazine). The brains were removed, the frontal cortex and the hippocampus were dissected on ice, and the tissue samples were stored at -80 °C until immunoblot analysis.
Immunofluorescence staining of tissue
Tissue slides of paraffin-embedded tissue sections from hippocampus of male and female pig-tailed macaques (M. nemestrina) were prepared for immunofluorescent staining, as described previously (Akay et al. 2011a) . Briefly, glass slides containing paraffin-embedded sections (10 μm) were heated to 55 °C for at least 30 min, deparaffinized in Histoclear, and rehydrated with consecutive 100, 95, 90, and 70 % ethanol washes. Endogenous peroxidase activity was blocked with 3 % H2O2 in methanol and antigen unmasking was achieved with target retrieval solution at 95 °C for 1 h. Tissue sections were blocked with 10 % normal goat serum in phosphate-buffered saline solution (PBS). Mouse monoclonal antibodies to synaptophysin, MAP2, and rabbit polyclonal antibody to GFAP were used at empirically defined dilutions (synaptophysin at 1:500; MAP2 at 1:100; and GFAP at 1:80), and DAPI was used to stain nuclei. The tyramide amplification system was used to detect synaptophysin. Slides were mounted in Citifluor AF1, and for each specimen, five to ten randomly selected areas within the CA1–CA3 layer of the hippocampus were scanned along the z-axis to create z-stack images at high magnification (×600) by laser confocal microscopy on a Biorad Radiance 2100 equipped with Argon, Green He/Ne, Red Diode, and Blue Diode lasers (Biorad, Hercules, CA). Post-acquisition analysis was conducted using MetaMorph 6.0 (Universal Imaging, Downingtown, PA). Total intensity for synaptophysin, MAP2 and GFAP were determined by the measurement of integrated pixel intensity for synaptophysin, MAP2 or GFAP per z-stack image, where the integrated pixel intensity is defined as total pixel intensity per image times the area of pixels positive for synaptophysin, MAP2 or GFAP. Averages are expressed as mean ± SEM. All data was analyzed by Prism 5.0 software (GraphPad Software, San Diego, CA).
MAP2 cell-based ELISA
A MAP2 cell-based ELISA was performed to quantify neuronal damage/death as previously described (White et al. 2011). Briefly, primary rat cortical neuroglial cultures plated in 96-well plates were fixed for 30 min with 4 % paraformaldehyde in 4 % sucrose at the conclusion of treatments. After blocking for 1 h with 5 % normal goat serum in PBS, the plates were incubated with monoclonal MAP2 antibody overnight at 4 °C, followed by washes with PBS with 0.1 % Tween-20 (PBS-T). The plates were then incubated for 30 min with goat anti-mouse secondary antibody conjugated to beta-lactamase TEM-1 at room temperature (RT), washed with PBS-T, and incubated in the dark at RT for 1 h in fluorocillin green substrate. Fluorescence intensity was measured using a Fluoroskan Ascent fluorometer plate reader (Thermo Electron, Waltham, MA) with excitation at 485 nm and emission at 527 nm.
Hand counting of MAP2-positive cells
As a complementary method to MAP2 ELISA, neuronal survival was verified by hand counting of MAP2-positive neurons; 15 μM propidium iodide was added to primary cortical neuroglial cultures grown on coverslips 15 min before the end of the treatments. The coverslips were washed once with PBS and fixed for 30 min with 4 % paraformaldehyde in 4 % sucrose, followed by blocking and permeabilization in 0.2 % BSA + 0.1 % Triton X-100 in PBS for 1 h at RT. The coverslips were then washed twice with PBS and incubated with anti-MAP2 antibody (1:100) in normal antibody diluent for 2 h at RT. After two washes with PBS-T, the coverslips were incubated in a fluorescein isothiocyanate-conjugated goat anti-mouse IgG secondary antibody (1:200) for 30 min at RT. The coverslips were mounted on slides and alive/dead neurons were hand-counted based on MAP2 and PI-positive staining using a Nikon Eclipse E400 fluorescent microscope (Nikon Corp, Tokyo, Japan) equipped with Olympus DP70 digital camera (Olympus Corp, Tokyo, Japan). Live cells stain negative for PI, while dead cells retain it in their nuclei. The number of surviving MAP2-positive neurons will be positive for MAP2 and negative for PI staining. The percentage of MAP2-positive cells ± SEM were calculated from blinded counting of six fields at ×200 in two adjacent vertical columns through the center of each coverslip, proceeding from top to bottom and bottom to top. For each condition, three coverslips were counted from two or more independent experiments.
Quantification of synaptophysin
Cultures grown on coverslips, as described above, were immunofluorescently stained for synaptophysin (1:1,000), MAP2 (1:100), and DAPI. Tyramide amplification was used for the detection of synaptophysin. For each treatment condition, images of 10 randomly selected areas from three coverslips from three independent experiments were captured by fluorescence microscopy. Post-acquisition analysis was performed using NIH ImageJ program (V 1.36b, Bethesda, MD). Briefly, synaptophysin-positive puncta were detected by background subtraction and manual thresholding. The number of synaptophysin-positive puncta was determined by dividing the total number of puncta within a dendrite segment by the length of the segment.
Measurement of reactive oxygen species
The superoxide indicator dihydroethidium (DHE, Invitrogen) was used to detect the presence of reactive oxygen species (ROS) in vitro. 3 μM DHE was added to culture media 15 min prior to conclusion of treatments. Cells were then washed with PBS, fixed with 4 % paraformaldehyde in 4 % sucrose at RT for 8 min, and stained for DAPI. The coverslips were then mounted on slides with CytoSeal and visualized using fluorescent microscopy, as described above. Post-acquisition analysis was performed using MetaMorph to determine the fluorescence intensity of DHE normalized to the area of DAPI signal.
Whole cell extracts of rat tissue samples were prepared by homogenization in ice-cold tissue extraction buffer (50 mM Tris at pH 7.5, 0.5 M NaCl, 1 % NP-40, 1 % SDS, 2 mM EDTA, 2 mM EGTA, 5 mM NaF, 0.4 mM Na3VO4, 1 mM dithiothreitol (DTT), and 1:100 PI cocktail), followed by centrifugation at 12,000×g at 4 °C for 20 min. Whole cell extracts of primary rat cortical cultures were prepared with ice-cold cell lysis buffer (50 mM Tris at pH 7.5, 120 mM NaCl, 0.5 % NP-40, 10 mM EDTA, 0.4 mM Na3VO4, 100 mM DTT, and 1:100 PI cocktail), followed by centrifugation at 14,000×g at 4 °C for 10 min. The protein concentrations of the collected supernatants were determined using the Bradford method and 25–50 μg of protein was loaded into each lane of a 4–12 % Bis-Tris gradient gel for separation. A broad range molecular weight ladder was run on each gel. Subsequent to separation, proteins were transferred onto PVDF membranes, and blocked in Tris-buffered saline with 0.1 % Tween-20 (TBS-T) and 5 % bovine serum albumin (BSA) for 1 h at RT. The membranes were incubated with the primary antibodies in TBS-T with 5 % BSA at 4 °C overnight, washed with TBS-T, followed by incubation with corresponding HRP-conjugated secondary antibodies. The membranes were developed using SuperSignal West Dura extended duration substrate. Loading controls were obtained by staining the membranes and the gels with the Biosafe Coomassie Stain for 20 min, followed by destaining with deionized water for 30 min. For densitometric analysis, autographs were scanned into Adobe Photoshop (Adobe Systems, San Jose, CA) and regions of interests (ROI) of equal size were determined for each band. The pixel intensities of ROIs were quantified using the NIH ImageJ program (V 1.36b, Bethesda, MD). Target band intensities were normalized to gel and membrane coomassie stain controls to account for protein degradation and loading discrepancies, and the normalized target band intensities were used to quantify fold changes over controls.
Quantitative reverse transcription polymerase chain reaction
The expression of HO-1 and NQO-1 genes in rat neuroglial cells was quantified by quantitative reverse transcription polymerase chain reaction (qRT-PCR). Custom TaqMan® Gene Expression Assays were purchased from Applied Biosystems for the genes: NQO-1 (Rn00566528_m1) and HO-1 (Rn01536933_m1). Approximately 5 ng cDNA was used per reaction. StepOne™ Software v2.0 was used to construct the experimental protocol and the qRT-PCR took place in the StepOne Real-Time PCR System (Applied Biosystems, Carlsbad, CA). Data was normalized using both β-actin (Rn00667869_m1) and 18S (Hs99999901_s1) and was analyzed according to the ΔΔCT method. All samples were run in triplicate from three biological replicates.
All data was analyzed by Prism 5.0 software (GraphPad Software, San Diego, CA). Values are expressed as mean ± SEM, and values of p < 0.05 were considered significant for all statistical analyses performed.
Antiretroviral drugs lead to neuronal damage in vivo
As studies of uninfected, cART-treated macaques have not been performed to determine the contribution of cART to neuronal damage independent of SIV infection, we administered combinations of ARVs intravenously to adult rats. In patients, initial cART usually includes two nucleoside/NRTIs in combination with a nNRTI or with a PI boosted with a low dose of a second PI, ritonavir (Rit). Thus, we used zidovudine (AZT), an NRTI, along with two PIs, saquinavir (Saq) and Rit, at doses based on previously published pharmacokinetic studies of these ARVs (Busidan and Dow-Edwards 1999; Kageyama et al. 2005; Pistell et al. 2010). In agreement with previous studies (Waring et al. 2010), the animals showed no overt signs of distress during the course of the treatment. However, via immunoblotting, we observed decreases in hippocampal synaptophysin expression in cART-treated rats compared with vehicle-treated rats (Fig. 1g), complementing our findings of synaptic damage in SIV(+)/cART macaques and further supporting a role for ARV-associated neuronal injury in the CNS. Of note, we also detected decreases in MAP2 levels in the cART-treated rat hippocampus (Fig. 1g), which likely reflects the acute drug toxicity in this treatment paradigm, compared to the subtler synaptic injury observed in the cART-treated macaque brain.
Antiretroviral compounds in therapeutically relevant combinations are neurotoxic in vitro
Combination antiretroviral drug treatments induce oxidative stress in neurons
Endogenous antioxidant response activation by combination antiretroviral drugs
Monomethyl fumarate protects against ROS generation and induces activation of the endogenous antioxidant response
Virus-related factors, such as resistant virus species and persistent viral DNA in the CNS, may contribute to the persistence of HAND in the post-cART era. One recent focus in HIV neurovirology is the development of ARVs with greater CNS penetrance (Letendre et al. 2008; Marra et al. 2009; Tozzi et al. 2009; Edén et al. 2010; Heaton et al. 2010; Garvey et al. 2011; Smurzynski et al. 2011). However, potential ARV toxicity in the CNS remains largely unexplored. Our study supports a possible contribution of the ARVs themselves to neuronal and synaptic damage observed in patients with HAND.
In this study, we show cART-induced synaptophysin loss, indicative of synaptic injury in two animal models. In our first model, in an in vivo model of SIV-infected pig-tailed macaques, we report decreased synaptophysin and CaMKII levels in the SIV(+)/cART group compared with uninfected or SIV(+)/placebo groups, indicating synaptodendritic damage. Interestingly, MAP2 levels did not change significantly across groups, which may be because of the relatively short duration of infection in this retrospective study cohort. While these data demonstrate potential effects of cART drugs in the presence of viral infection, there are three variables to consider in interpretation of these findings: (1) the time to euthanasia from the start of the experiments is different between SIV(+)/placebo and SIV(+)/cART groups, (2) persistent viral DNA in the CNS of SIV(+)/cART group, and (3) the lack of SIV(-)/cART group. It is not yet known whether brain SIV DNA is replication competent. As we utilized post-mortem samples obtained from a cohort of macaques enrolled in a previous study addressing the efficacy of CNS penetrant cART in reducing viral loads in the CNS, further experiments that include an additional control group receiving cART but not inoculation with SIV will be instrumental to more clearly determine the contribution of viral DNA and cART to synaptic damage in this model.
In the second in vivo model presented here, adult rats received a therapeutically relevant combination of ARVs (NRTI+PI+Rit boost). In the small number of studies where pharmacokinetics and effects of ARVs in the CNS were examined, pathological read-outs of neuronal damage, such as MAP2 or synaptophysin loss, were not determined (Huisman et al. 2003; Anthonypillai et al. 2004; Anthonypillai et al. 2006). Synaptic injury is a known indicator of neuronal damage and dysfunction in various neurodegenerative diseases, including HAND (Gupta et al. 2010) and synaptodendritic injury persists in HIV-infected individuals in the post-cART era (Xu and Ikezu 2009). We observed decreases in synaptophysin and MAP2 protein levels in the hippocampus in response to ARV administration over 7 days. Thus, our model demonstrates that ARV-associated neurotoxicity warrants consideration in developing therapeutic regimens for HIV-infected patients.
We also show that the PIs, Rit and Saq, alone or in combinations with the NRTI, AZT, induce oxidative stress, and neuronal damage/death in primary cultures at clinically relevant doses. Previous studies examined ARV-induced toxicity in cell lines, and Robertson et al. have provided the first evidence for ARV-induced neurotoxicity in primary rat neurons (Robertson et al. 2012). Here, we provide further evidence that PI-induced oxidative stress and neuronal death in primary neurons can be blocked by the activation of the endogenous antioxidant response. Interestingly, in our experimental paradigm, the NRTI, AZT, neither induced neuronal damage/death by itself, nor augmented PI-induced damage/death when used in combination. We observed similar effects from a combination using another NRTI, stavudine (d4T; not shown). In agreement with our observations, a previous study presented similar findings, specifically that neither AZT nor d4T inhibited cell growth or neurite regeneration in PC-12 cells after long-term drug exposure (Cui et al. 1997). It should be noted that NRTIs are unequivocally tied to peripheral neuropathy, where the underlying pathology is mitochondrial toxicity and oxidative stress. However, NRTIs affect only certain cell populations, as they are formulated as pro-drugs and, to become active, need to be phosphorylated by two kinases, thymidine kinase 1 and 2 (TK1 and TK2), and the expression of the cytoplasmic TK1 is cell cycle dependent (Bazzoli et al. 2010). Thus, in our model utilized to study post-mitotic neurons, AZT is most likely not converted to its active form, and thus does not contribute to neuronal damage/death. It is of note that our in vitro model of ARV-induced neurotoxicity utilized primary neuroglial cultures rather than cell lines. Primary cells are untransformed, and therefore more accurately reflect and predict ARV-associated effects occurring in the brains of patients on cART than would immortalized cell lines. The molecular pathways we investigated in this study are highly conserved from yeast to human cells; thus, the results obtained here in cells of rodent origin are likely conserved in human cells as well.
The drug concentrations used in this study are based on the plasma and CSF levels reported by various in vitro and in vivo studies(Huisman et al. 2003). As reported in such studies, AZT can be detected in the CSF at concentrations that are similar to those measured in the plasma (Wynn et al. 2002). Contrarily, as backed by various in vitro and in vivo studies (Wynn et al. 2002), both Rit and Saq are predicted to have limited CNS penetrance due to the strong tendency of these drugs to bind plasma protein because of their lipophilic nature and pharmacokinetic properties. However, a comprehensive study conducted in an in situ guinea pig model suggests that Rit can achieve high concentrations in choroid plexus and parenchyma through diffusion via the choroid plexus; in fact, the levels of Rit were comparable to levels measured in plasma (Anthonypillai et al. 2004). Furthermore, the study showed that, surprisingly, the CSF levels of Rit were lower than the levels measured in the choroid plexus and the parenchymal compartments. Thus, the authors concluded that CSF concentrations of Rit may not necessarily reflect the parenchymal levels, which are indeed a better indicator of effective drug levels in the CNS. Rit and Saq concentrations in our experiments fall well within plasma ranges reported in patients receiving cART.
Evidence of oxidative stress has long been associated with the most severe forms of HAND (Ngondi et al. 2006; Mielke et al. 2010). Interestingly, despite systemic control of viral replication in cART-treated patients, markers of oxidative stress, such as elevated levels of lipid and protein oxidation, are still detectable in the brains of these individuals (Ances et al. 2008). Of note, oxidative stress is one of the underlying mechanisms involved in NRTI and PI-induced toxicity in the periphery. Our data suggest that the sustained ROS accumulation in neurons due to prolonged exposure to ARVs might induce the oxidative stress associated with cART-induced toxicity in the CNS, leading to the observed changes in synaptophysin levels and the subsequent neuronal death.
Our data also suggest that astrocytes do not show neither ROS accumulation nor the endogenous antioxidant response in vitro. We also observe that astrogliosis in SIV-infected animals is resolved in cART-treated animals. While these findings suggest that astrocytes may not be highly impacted by ARVs in the short term, it is possible that prolonged exposure to ARVs might overwhelm astrocytes, which help buffer ROS accumulation in neurons under normal conditions, precipitating further neuronal damage.
We further show that the ARV drug-induced effects observed in this study can be blocked by MMF via activation of an endogenous antioxidant response. Cells respond to oxidative stress by activating the transcription of a subset of genes in an effort to clear excess ROS within the cell. Among these genes are NQO-1 and HO-1 (Chen and Kong 2004). Here, we show increased levels of NQO-1 and HO-1 mRNA and protein in ARV-treated neuronal cultures. Additionally, pretreatment of cultures with MMF led to further increases in HO-1 protein levels in ARV-treated cultures and provided protection against ARV-induced toxicity. Finally, our finding that the protection provided by MMF is reversed with chemical inhibition of HO-1 further provides evidence that ARV toxicity is mediated via induction of oxidative stress. The neuronal damage and death that occurs despite the cellular initiation of the endogenous antioxidant response following exposure to ARVs suggests that this response may be insufficient or too delayed to protect cultures from ARV toxicity. In support of this explanation, we observed that MMF-mediated augmentation of antioxidant responses is strongly protective against the neurotoxic effects of ARVs. Further studies using this in vitro model are warranted to determine whether other clinically relevant ARV drug combinations induce neurotoxicity and, if so, to establish whether similar pathways are involved.
DMF, a psoriasis treatment used in Europe since 1994, is currently being tested as a disease-modifying agent for multiple sclerosis (MS) (Anonymous 2011; Krieger 2011; Linker et al. 2011). MMF is the active DMF metabolite in vivo. Our report is the first to show MMF as a neuroprotectant against ARV-induced damage. Data from studies in MS patients show that DMF/MMF has good tolerability, can cross the BBB efficiently, and has relatively few and minor side effects. Furthermore, we have recently shown that DMF suppresses HIV replication, induces the antioxidant response in macrophaghes, and blocks neurotoxin release from macrophages (Cross et al. 2011). Overall, our findings make this immunomodulatory and antioxidant agent a good potential adjunctive therapeutic for use in HAND.
One of the paradoxical outcomes of cART is the persistence of HAND, despite successful viral control. Moreover, there is a shift from overt dementia to subtler neurocognitive impairments. Based on data presented here, we propose that different mechanisms by which HIV infection and cART induce neuronal damage underlie the changing neuropathology of HAND. Extensive studies have shown that during lentiviral infection, viral proteins and soluble factors secreted by infected cells lead to neuronal and synaptic damage via several direct as well as indirect mechanisms involving several cell types, including astrocytes. However, our data suggest that cART-mediated synaptic damage may involve direct mechanisms occurring specifically in neurons, such as oxidative stress, and that neurons, and not astrocytes, are the primary targets of cART-mediated damage in the CNS.
It is most likely that chronic exposure to cART regimens including neurotoxic ARVs over many years is associated with a slower, nonetheless insidious changes including synaptic damage in the absence of neuronal loss, and these neuronal perturbations may contribute to, and may even precipitate, some of the clinical and pathological changes observed in the chronic course of HAND in the cART era. While detrimental, such slow damage also suggests that alterations in cART regimens to include ARVs with low neurotoxicity profiles, such as NRTIs or nNRTIs may halt synaptic damage, and provide a point where previous damage may be reversed, either due to the withdrawal of the toxic drug, or with the help of an adjuvant, such as fumaric acid esters. Future studies, first in primates, then in humans, will be crucial to explore the specific impact of treatment interruption on recovery from cART-mediated neuronal damage and to determine the eddicacy of potential adjunctive therapies necessary to mitigate the side effects of cART in the CNS.
We would like to thank Margaret Maronski for her help in the preparation of cortical cultures.
This work was supported by the following National Institutes of Health Grants: MH083517 (K.J-S), NS043994 and NS27405 (D.L.K), AG000255 (S.A.C), DA22339 and DA18678 (R.C.P), CA133470 (M.C-S), and MH070306 (J.E.C.).
Conflict of interest
All authors declare that they have no conflict of interest.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.