HIV-1 Tat and Morphine Have Interactive Effects on Oligodendrocyte Survival and Morphology

Abstract

HIV-infected individuals who abuse opiates show faster progression to AIDS, and enhanced incidence of HIV-1 encephalitis. Most opiates with abuse liability are preferential agonists for μ-opioid receptors (MORs), and MORs are expressed on both neurons and glia, including oligodendrocytes (OLs). Tat, gp120, and other viral toxins, cause neurotoxicity in vitro and/or when injected into brain, and co-exposure to opiates can augment HIV-1 protein-induced insults to both glial and neuronal populations. We examined the effects of HIV-1 Tat +/- opiate exposure on OL survival and differentiation. In vivo studies utilized transgenic mice expressing Tat_1-86 regulated by an inducible GFAP promoter. Although MBP levels were unchanged on immunoblots, certain structural and apoptotic indices were abnormal. After only 2 days of Tat induction, OLs showed an upregulation of active caspase-3 that was enhanced by morphine exposure. Tat also upregulated TUNEL staining, but only in the presence of morphine. Tat significantly reduced the length of processes in Golgi-Kopsch impregnated OLs. A greater proportion of cells exhibited diminished or aberrant cytoplasmic processes, especially when mice expressing Tat were co-exposed to morphine. Collectively, our data show that OLs in situ are extremely sensitive to effects of Tat +/- morphine, although it is not clear if immature OLs as well as differentiated OLs are targeted equally. Significant elevations in caspase-3 activity and TUNEL labeling, and evidence of increased degeneration/regeneration of OLs exposed to Tat +/- morphine suggest that toxicity towards OLs may be accompanied by heightened OL turnover.

Introduction

The central nervous system (CNS) is extremely vulnerable to toxicity induced by the human immunodeficiency virus (HIV). HIV encephalitis (HIVE) and neurological complications are a common side effect of AIDS. Prior to the advent of highly active anti-retroviral therapy (HAART) an estimated 10-30% percent of patients experienced full blown HIV-associated dementia (HAD) (Anthony and Bell 2008; Ghafouri et al. 2006; Kandanearatchi et al. 2003). The use of antiretrovirals and protease inhibitors initially reduced the incidence of HIVE and HAD, and changed the pattern of neurological disabilities and psychological impairment (Cysique et al. 2004; Gonzalez et al. 2002; Gonzalez-Scarano and Martin-Garcia 2005). However, longer patient survival times and selection of more highly virulent and resistant forms of the virus have resulted in a recent increase in HIVE severity and prevalence (Ances and Ellis 2007; Anthony and Bell 2008; Ghafouri et al. 2006; Langford et al. 2003; Neuenburg et al. 2002). Since neurons themselves are not virally infected, neuropathological changes are likely due to combined direct effects of viral proteins on neurons, and indirect inflammatory processes mediated by infected microglia/macrophages or changes in astroglial function. Many studies, including some from our laboratories, show that neurons are either directly or indirectly targeted by HIV viral proteins including gp120, Tat, Nef, Rev, and Vpr (Aksenov et al. 2003; Bonavia et al. 2001; Fitting et al. 2008; Gurwell et al. 2001; Jana and Pahan 2004; Jones et al. 2007; Kim et al. 2003; Kruman et al. 1998; Lipton 1997; Nath et al. 2000; Nath et al. 1996; Patel et al. 2000; Singh et al. 2004; Trillo-Pazos et al. 2000). However, only a subset of patients present with symptoms of neuroAIDs, and the onset of those symptoms usually occurs as a relatively late manifestation of the disease process. This suggests that more subtle and indirect insults over a longer time period may be quite important in causing neuronal dysfunction and/or death.

One exacerbating factor in HIVE development appears to be concurrent exposure to opiate drugs of abuse. HIV infected individuals who also inject opiate drugs show accelerated progression to clinical AIDS and to HIVE, which may lead to HAD (Anthony et al. 2008; Bell et al. 2006; Burdo et al. 2006; Hauser et al. 2007; Kopnisky et al. 2007). Opiates with abuse liability largely target the μ-opioid receptor (MOR), which is widely expressed both in the peripherally and in the CNS. Interactions between opiates and HIV or related viruses and viral proteins appear to directly accelerate viral replication under many conditions (Li et al. 2003; Noel and Kumar 2006; Peterson et al. 2004; Schweitzer et al. 1991; Suzuki et al. 2002). Progression to AIDS and HIVE is undoubtedly advanced by MOR-signaling on cells of the peripheral immune system, whose function is altered by chronic opiate exposure (Carr and France 1993; Donahoe and Falek 1988; Nath et al. 2002; Peterson et al. 1993; Vallejo et al. 2004). However, there is substantial evidence in several disease models that both neurons and glia, as well as immune cells in the brain, are directly affected by co-exposure to opiates and HIV in a manner that would ultimately lead to greater compromise of CNS function. In vitro, concurrent exposure to morphine promotes Tat-induced striatal neuron death and also destabilizes Ca²⁺-signaling in astroglia, leading to enhanced secretion of pro-inflammatory chemokines and cytokines and enhanced macrophage/microglial migration (El-Hage et al. 2005; El-Hage et al. 2006b; Gurwell et al. 2001). Similar effects can be observed in some studies with co-exposure to gp120 and morphine (Hu et al. 2005; Mahajan et al. 2005; Turchan-Cholewo et al. 2006). In vivo, the intrastriatal injection of Tat into morphine treated mice results in astrogliosis and increased microglial numbers, as well as an overall upregulation in the percentage of cells expressing MORs (El-Hage et al. 2006a).

The present study explores the possibility that oligodendrocytes (OLs), in addition to other CNS cells, may be targets of lethal or sublethal effects of HIV-1 Tat. Myelin pallor has been described by neuropathology and on MRI scans from some HIV patients (Cherner et al. 2002; Glass et al. 1993; Gray et al. 1992; Gray et al. 1996; Lanjewar et al. 1998; McArthur et al. 1989; Schwartz and Major 2006), and although this has been thought to be a consequence of endothelial damage and increased blood brain barrier permeability, the direct and/or secondary effects of HIV proteins on OLs have not been studied in detail. An earlier study reported no changes in myelin basic protein levels and no demyelinated regions in brains of patients with HIV dementia (Power et al. 1993). However, there are more subtle changes in myelin, in axon-myelin interactions, or in OL function that might contribute to neuron dysfunction. Additionally, since OLs express MORs and other classes of opioid receptors both in vivo and in vitro (Knapp et al. 2001; Knapp et al. 1998; Stiene-Martin et al. 2001; Tryoen-Toth et al. 2000), and since opioid receptors modulate OL growth and viability in vitro (Knapp and Hauser 1996; Knapp et al. 2001; Knapp et al. 1998), we explored the combined effects of Tat and opiate exposure. In vivo studies utilized a transgenic mouse in which HIV-1 Tat_1-86 is produced in astroglia under control of a doxycycline-induced promoter. We report that Tat exposure in combination with morphine can variably affect indices of OL phenotype and well-being. Tat exposure alone alters process outgrowth and caspase-3 expression in OLs within the striatum and/or corpus callosum. Terminal dUTP nick end labeling (TUNEL) is increased when Tat is delivered concurrently with morphine. Some effects are exacerbated by co-exposure to morphine, while others are lessened. Thus OLs, in addition to neurons and other CNS glia, may be primary or secondary targets of Tat and opioid toxicity.

Methods

Tissue Culture

Progenitor cultures containing OL precursors were prepared from striatum of embryonic day 14-15 (E14-15) mouse pups using modifications of a previously published technique (Khurdayan et al. 2004). Briefly, timed pregnant ICR dams (Charles River, Boston, MA) were deeply anaesthetized with halothane and pups were removed from the uterus. The striata from 10-12 pups were dissected from the cortex, finely minced, and incubated with 5 ml of DNAse (0.015 mg/ml) in Dulbecco's Modified Eagle's Medium without additives (15 min, 37°C). Tissue was triturated through a 5 ml pipette, centrifuged (5 min, 800 rpm), suspended in 5 ml progenitor maintenance medium (1:1 DMEM/F12; insulin (25 μg/ml); transferrin (100 μg/ml); progesterone (20 nM); putresceine (60 μM); sodium selenite (30 nM); all from Sigma), and filtered through 70 μm pore nylon mesh. Resulting cells were counted, centrifuged and resuspended in the same medium. After a second filtration, the supernatant was plated on either poly-L-lysine coated T25 culture flasks for RT-PCR (3 × 10⁶ total cells) or poly-L-lysine coated glass coverslips for immunostaining (3 × 10⁵ cells in 150 μl). Cultures containing more mature OLs and used for immunostaining were isolated similarly, grown in progenitor maintenance medium for one week and gradually switched to differentiation medium (DMEM with 5 μg/ml insulin, 25 μg/ml transferrin, 10 ng/ml biotin, 5 ng/ml sodium selenite, and 6.0 mg % glucose, and antibiotics). Some cultures were subsequently treated with Tat_1-86 (ImmunoDiagnostics, Woburn, MA; 100nM), morphine sulphate (NIDA Drug Supply System, Rockville, MD; 500 nM), Tat+morphine, or Tat+morphine+naloxone (naloxone-1.5 μM, NIDA Drug Supply System) for 24 hrs prior to fixation and immunostaining for active caspase-3 and OL markers. All treatments were applied simultaneously in pre-warmed medium.

Opioid Receptor PCR

Non-quantitative RT-PCR was used to document the presence of mRNA for each major opioid receptor in OL progenitor cultures. At 2 and 5 days after plating, progenitors were harvested by scraping into ice-cold, deionized water from culture plates. Total RNA was isolated using GenElute™ Mammalian Total RNA kit (Sigma). cDNA was synthesized from 2 μg of total RNA using the High-Capacity cDNA Archive Kit (Applied Biosystems, Warrington, UK) per manufacturer's instructions. Primers for MOR (forward, 5'- CCGCAGCAAGCATTCAGAA -3'; reverse, 5' GCCATCAACGTGGGACAA 3'), KOR (forward, 5'- GAGCACCAATAGAGTTAGAAA -3'; reverse, 5'- CCTGGAGAACAATAAGAAGAC -3'), and DOR (forward, 5'- GCGTGTCACTGCCTGCAC -3'; reverse, 5'- TGGCCTCTGGATCACTTCACT -3') were used to assay opioid receptor expression. 18S mRNA was assayed as a control (forward, 5'-CACTTGTCCCTCTAAGAAGTTG-3'; reverse, 5'- GACAGGATTGACAGATTGATAG-3').

Opiate and Tat delivery in vivo

In vivo studies utilized a transgenic mouse engineered to express the HIV-1 Tat_1-86 protein in astrocytes in a doxycyline-inducible manner. As previously described (Bruce-Keller et al. 2008), mice expressing the Tat gene under control of a tet responsive element (TRE) in the pTREX vector (Clonetech, Mountain View, CA) were crossed with mice engineered to express the glial fibrillary acidic protein (GFAP) promoter driving the reverse tetracycline transactivator (RTTA). The inducible Tat transgenic mice (Tat⁺ mice) used in these studies express both GFAP-RTTA and TRE-Tat genes, while the control mice (Tat^- mice) express only the GFAP-RTTA gene. Individual animals were genotyped using RNA isolated from ear clip samples, by standard PCR procedures previously described (Bruce-Keller et al. 2008). Tat-induction was obtained by replacing standard chow with a formulation containing doxycycline (DOX) at 6 mg/g (Harlan, Indianapolis, IN). Some control groups were also fed DOX-containing chow to control for non-specific actions of DOX intake.

Morphine was delivered by 25 mg, timed-release pellets implanted subcutaneously in the subscapular region (NIDA Drug Supply System) under aseptic conditions using isoflurane anaesthesia. Sham pellets were similarly implanted as controls. The 25 mg morphine pellets reportedly deplete at a rate of 5 mg/day, yet steady state levels of morphine, measured in ng/ml plasma or ng/gm brain tissue reflect lower availability, probably due to the balance of delivery and metabolism (see, e.g., (Feng et al. 2006)). Both C3HeB and C57Bl mice reportedly tolerate delivery from 75 mg/5 day morphine pellets (Peart and Gross 2004; Rahim et al. 2003). Less than 5% of the mice in our studies died, and there was no relationship between morphine-induced toxicity and either genotype or DOX administration.

In experiments lasting for a total of 7 days, opioid and control pellets were replaced after 5 days. Naltrexone was administered via an Alzet mini-pump (1007D) implanted in the same region. 600 mg/ml naltrexone (NIDA Drug Supply System) was prepared in 50% DMSO, a standard vehicle for mini-pump delivery (Arnot et al. 1996). 100 μl total volume was loaded per mini-pump, delivering 0.5 μl/hr for up to 7 days. Pumps containing 50% DMSO and sterile saline were used as naltrexone controls. For studies where mice were exposed to both opioids and Tat, mice were given the DOX feed starting on the night before opioid treatment in order for blood levels of DOX to stabilize. For histological studies, mice were deeply anaesthetized with halothane prior to perfusion with Zamboni's fixative (2% paraformaldehyde, pH 7.4, with 0.15% picric acid).

Immunostaining and Quantification In Vivo

After perfusion, brains were removed and postfixed in fresh fixative overnight. Brains were hemisected, rinsed several times and overnight in 15 ml changes of phosphate buffered saline, cryopreserved through graded sucrose solutions (10% and 25%), embedded in Tissue Tek OCT compound (Sacura Finetek, Torrance, CA), and stored at -80°C. 12 μm frozen sections containing the corpus callosum were thaw-mounted on SuperFrost Plus slides (VWR Scientific, West Chester, PA) and processed for immunostaining. Sections were double-stained sequentially for APC (adenomatous polyposis coli), followed by immunostaining for active caspase-3 or terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick-end labelling (TUNEL) using fluorescent markers. APC is a tumor suppressor gene whose mutation is linked to forms of colon cancer, but which is also expressed specifically in OL cell bodies in the brain (Bhat et al. 1996). It is especially useful for unambiguous quantification of cell numbers since it does not stain myelin. Sections were rinsed and permeabilized with 0.3% Triton X-100 in phosphate-buffered saline containing 5% normal goat serum and 0.1% bovine serum albumin for 30 min. APC/CC-1 antibody (Oncogene, San Diego, CA; 5 mg/l or 1:20) was applied for 1 hr at room temperature, followed by visualization with goat anti-mouse second antibody conjucated to Oregon Green 488 (Molecular Probes; 1:250; 1hr; room temperature). In some cases, APC staining was followed by overnight incubation with affinity-purified rabbit anti-human/mouse active caspase-3 antibody (R&D Systems, Minneapolis, MN; 1:2,000, 4°C). This antibody detects amino acids 163-175 of the p17 subunit of caspase-3, but does not detect or poorly detects the precursor form. Anti-caspase-3 was visualized with goat anti-rabbit IgG conjugated to CY-3 (Jackson ImmunoResearch, West Grove, PA). In other sections, APC staining was combined with TUNEL labeling to detect in situ DNA fragmentation using the TACS2-TdT Apoptosis Detection Kit (Trevigen, Gaithersburg, MD), per manufacturer's directions. Lastly, immunostained sections were incubated with Hoechst 33342 dye (1:20,000) for identification of all nuclei (Molecular Probes), mounted in ProLong Gold antifade reagent (Molecular Probes) and dried for 8 hrs in the dark. Sections were stored dessicated at -80°C and used for quantitation within 2 months. No nonspecific binding was observed when the primary antibodies were omitted.

APC⁺ OLs were identified in the corpus callosum in the region of the cingulum at 63X. The first 100 APC⁺ cells per section that were clearly distinct from neighboring cells were assessed for either caspase-3 or TUNEL staining depending on the experiment. Cells were counted only if the entire nucleus was stained with Hoechst. Two separate sections at least 40 μm apart were examined per animal and averaged in N=4-5 animals per experimental group. The relative percentage of OLs expressing active caspase-3 or TUNEL was calculated and compared between treatment and control groups. Stereological techniques were not used in this study since we were interested in the relative response of OLs within the population.

Immunostaining and Quantification In Vitro

In some experiments cultured cells were double-immunostained for μ, δ, or κ-opiate receptors and olig-2, a basic helix-loop-helix (bHLH) transcription factor that is required for OL lineage specification in the brain and spinal cord (Ligon et al. 2006; Woodruff et al. 2001). In other experiments, we double-stained cells using O₄ antibody and an antibody to active caspase-3. O₄ antibody detects sulfatide (Sommer and Schachner 1982), a sulfated galactocerebroside specifically expressed by immature and mature OLs in the brain. In terms of developmental expression, olig-2 is an earlier marker than sulfatide, but both are also found in mature OLs. Cells were immunostained using O₄ hybridoma supernatant harvested in our laboratory (1:2, 20 min, 4°C), then visualized with goat anti-mouse IgM-FITC (Jackson ImmunoResearch, 1:250, 20 min, 4°C) prior to fixation in 4% paraformaldehyde for 10 min. The antibody to active caspase-3 was then applied and visualized as described for the tissue sections above. The effect of Tat and/or opioids on the activation of caspase-3 in OLs was quantified after 24 hrs of treatment. 100 04⁺ cells were selected randomly on duplicate coverslips from each treatment group in N=3 different cultures, then assessed for active caspase-3 immunostaining.

Golgi-Kopsch Impregnation / OL Process Measurement

Whole forebrains (N = 4-8 for each treatment) were impregnated after 7 days of doxycycline and/or opioid treatment using a modified Golgi-Kopsch procedure that randomly impregnates neurons and glia (Hauser et al. 1989). Briefly, mice were deeply anesthetized by halothane inhalation and euthanatized by intracardiac perfusion with 2% potassium dichromate and 5% glutaraldehyde. Perfused forebrains were isolated and immersed in 2% potassium dichromate and 5% glutaraldehyde (v/v) in the dark at room temperature. The ratio of dichromate solution to tissue volume was at least 50:1. After 5 days, the tissue was rinsed twice in ultra-pure water three-times, blotted dry after each rinse, and placed in aqueous 0.75% silver nitrate solution for 5 days in the dark (50:1 fluid to tissue volume ratio). Intact forebrains were infiltrated with graded sucrose solutions (10, 20, and 30% sucrose in 0.75% silver nitrate) for 24 hr periods, then embedded in Tissue Tec OTC compound, frozen on dry ice, and stored at -80°C. Serial, coronal sections were cut at 120 microns, thaw-mounted on Fisher Superfrost-Plus microscope slides, dehydrated through graded ethanols, cleared in xylene and mounted in Permount (Fisher Scientific, Waltham, MA). Non-metallic handling devices were used throughout. All slides were quantified by an individual unaware of the treatment group.

Although the Golgi procedure randomly impregnates neurons and glia, additional criteria were used to determine whether an OL was sampled, and OLs that fulfilled all of the following criteria were arbitrarily selected and analyzed: 1) OLs had to be discernable from other cell types. This included having the unique cellular morphology characteristic of OLs (e.g., small, ~10 micron-diameter cell body, myelin-forming cytoplasmic processes, cytoplasmic processes aligned along white matter tracts) and a location appropriate for OLs (e.g., associated with focal striatal white matter bundles, or in the corpus callosum adjacent to the striatum); 2) OLs had to be entirely impregnated throughout all processes; 3) OLs had to be located in the middle of the 120 μm section thickness so that they could be measured in their entirety, and 4) OL cytoplasmic processes could not overlap those from nearby neurons or glia, and could not contact blood vessels. A MicroBrightfield Neurolucida System (Williston, VT) configured with a Zeiss Axiomat microscope and x, y, and z-axes stage-encoder was used to measure and analyze OL processes in 3-dimensions. For individual OLs, estimates of: (i) total process length, (ii) the number of intervening segments (number of processes + number of times they branch), and (iii) mean segment length (total length/number of process segments) were obtained. We estimated the total process length for each individual OL. At least 20 OLs were measured from each animal, with 4-8 animals/group. Photomicrographs were obtained using a Nikon Diaphot microscope with 100x oil-immersion objective and Spot 2 digital camera (Diagnostic Instruments, Sterling Heights, MI) or a Zeiss Axio Observer Z1 microscope with an AxioCam MRm (Carl Zeiss, Inc.), and 63x or 100x oil-immersion objective. In the case of the Zeiss Axio Observer microscope images, multiple z-stack images were typically acquired through the same OL at 0.25-0.45 μm intervals. Different z-plane images from the same OL were projected into a single, sharp 2-dimensional image using the Extended Focus Module (AxioVision 4.6.3, Carl Zeiss, Inc., Thornwood, NY) with “Normal” alignment and the “Wavelets” algorithm to compress the images (Fig. 5 D,E,H,I,L,M).

Figure 5.

Tat and morphine induce changes in OL morphology. Golgi-Kopsch impregnation was used to visualize the structure of OLs within the corpus callosum and striatum of Tat⁺ and Tat^- mice undergoing various opiate treatments during a 7 day exposure to doxycycline (DOX). Sections were examined using a Zeiss Axio Observer microscope and multiple z-stack images were acquired through the same OL at 0.25-0.45 μm intervals. Z-stacks from the same OL were projected into a single, sharp 2-dimensional image using the Zeiss AxioVision Extended Focus Module (see Methods for details). Brightfield photomicrographs (panels A-M) show cells from both Tat⁺ and Tat^- mice, with treatment groups as labeled. Although normal appearing, myelinating OLs are present in both genotypes, there are increased examples of OLs with aberrant morphology in the Tat⁺ mice. This is especially true in Tat⁺ mice receiving treatments with DOX or DOX co-administered with morphine. Panels (A-C) show phenotypically normal OLs in the striatum and corpus callosum of wild-type mice receiving unmodified chow. Most OLs in morphine-treated, wild-type mice (D-E) displayed normal morphology (D). However, it was more common than usual to see smaller OLs that were potentially degenerating or regenerating (arrow in E), and OLs with phenotypic abnormalities such as severely stunted processes (cell in E without arrow). Expression of the HIV-1 Tat gene dramatically increased the presence of OLs with abnormal processes or cell bodies (F-G, J-K), and combined morphine exposure and Tat induction further aggravated the pathology (H-I, L-M). Pathologic changes included abnormal varicosities along processes or large, club-like cytoplasmic endings that could be several microns in diameter (arrowheads in G, I, J, K) and/or quantitative reductions in the number and extent of cytoplasmic processes (E-G, J, L, M). Note that diminished myelin contributions per cell were frequently associated with white matter tracts within the striatum following Tat induction, and especially when morphine was co-administered (M). It is important to note that despite overt pathologic changes accompanying Tat expression, many OLs appeared relatively normal indicating a non-uniform response among cells (H); arrowheads indicate abnormal cytoplasmic/myelin processes, arrows indicate apparent degenerating cells, dashed-lines indicate approximate boundaries of white matter tracts in the striatum; scale bars = 10 μm).

Immunoblotting

Total proteins were extracted from cortical tissue by homogenization in RIPA buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% Nonidet, 0.5% deoxycholic acid, 0.1% SDS, one tablet of protease inhibitors from Roche in 10 ml RIPA). Protein concentration was measured using the BCA Protein Assay Kit (Pierce Inc., Rockford, IL). 20 μg of total protein was separated on 15% Tris-HCl Ready Gels (Bio-Rad, Hercules, CA) and blotted to Hybond-P membrane (Amersham Biosciences, Piscataway, NJ). MBP was detected using a monoclonal antibody that identifies all major isoforms except for 14 Kda (SMI 99, 1:1000; Covance Research Products, Princeton, NJ). Blots were detected using the ECL-Plus detection system (Amersham) and visualized on a Kodak Image Station 440.

Statistical Analysis

All results, except those for the TUNEL labeling, were analyzed by ANOVA and Duncan's post-hoc testing using either JMP 5 (SAS, Cary, NC) or StatSoft (Statistica, Tulsa, OK) software. In the case of the TUNEL analysis, Kruskal-Wallis non-parametric ANOVA and post-hoc multiple comparisons tests were used to assess statistical relevance.

Results

Overall, our results suggest that OLs are extremely sensitive to the effects of HIV-1 Tat exposure, and that the consequences of Tat exposure can be either selectively exacerbated or alleviated by concurrent exposure to morphine.

Expression of Opioid Receptor Genes in Cells of the OL Lineage

We used RT-PCR and immunocytochemistry to determine whether glial progenitors and immature OLs express opioid receptors. mRNA for MOR, DOR and KOR was detected in whole tissue samples taken from E15 cerebral cortex and striatum (Fig. 1). We next prepared highly purified glial progenitor cells from E14-15 striatum, and cultured them for 2 or 5 days prior to performing RT-PCR. mRNA for all three receptors was detected at both times in culture (Fig. 1). Since these were mixed progenitors, we used immunohistochemistry to determine the types of receptors expressed in immature cells committed to an OL fate. Double-labeling showed that most immature glia expressing the olig-2 transcription factor also expressed one or more opioid receptors at both 2 and 5 days in culture (Table 1). Many cells that immunostained for opioid receptors did not express olig-2, suggesting that glial precursors at many stages of development could interact with, and be targets of, opioid ligands.

Figure 1.

OL progenitors express opioid receptors. A and B. Real time RT-PCR detection of mRNA for MOR, DOR and KOR in (A) striatum and cerebral cortex from embryonic day 15 (E15) mice, and in (B) progenitor cells (PC) from E15 cortex after 2 and 5 days in culture. C and D. Progenitors in E15 cortical cultures stained for MOR (red) and Olig-2 (green), a transcription factor indicating commitment to the OL lineage in these cultures. Arrows indicate double-labeled cells, many of which are Olig-2^-, showing that MORs are expressed in both committed and multipotential cells.

Table 1.

Percent of Olig-2⁺ Cells Expressing μ-, δ-, and κ-Opioid Receptors in Vitro

	2 DIV	5 DIV
MOR	71.2 ± 0.038	72.0 ± 0.029
DOR	67.6 ± 0.119	64.7 ± 0.041
KOR	69.7 ± 0.090	71.4 ± 0.025

Cells were cultured from E14-15 striatum and assessed after 2 or 5 days in vitro (DIV). N=4 for MOR; N=3 for DOR and KOR.

Active Caspase-3 and TUNEL Expression

In order to determine whether Tat and/or opioid exposure might ultimately affect the survival of OLs, we assessed their effects on two potential indicators of apoptosis - activation of caspase-3 and TUNEL labeling. The calcium-dependent protease caspase-3 is a major effector of programmed cell death. An antibody that detects activated human/mouse caspase- 3 was used in combination with markers for OLs both in vitro and in frozen sections. Since it is now recognized that caspase-3 may be activated under certain non-lethal conditions, such as during stress, development or tissue remodeling (Campbell and Holt 2003; Jarskog et al. 2007; McLaughlin 2004; Oomman et al. 2004; Oomman et al. 2005; Yang et al. 2004a; Yang et al. 2004b), we also evaluated TUNEL labeling as a more certain indicator of nuclear fragmentation and cell death.

We first examined glial progenitor cells that had been in culture under differentiating conditions for 7 days prior to 24 hr treatment with exogenous HIV-1 Tat_1-86, morphine, Tat+morphine, or Tat+morphine+naloxone. Exposure to Tat or morphine alone increased active caspase-3 expression in O₄⁺ cells, and the combination was synergistic (Fig. 2). The exacerbation caused by concurrent Tat and morphine exposure was blocked by naloxone, suggesting that the additivity was an opioid-specific response. TUNEL was not examined in cultured cells because of background DNA fragmentation in non-phagocytized cells over time.

Figure 2.

Tat and morphine (Morph) elevate active caspase-3 expression in cultured OLs. A 24 hr exposure to Tat_1-86 (100 nM) or morphine (500 nM) alone caused a 2-fold increase in immunohistochemical expression of activated-caspase-3 in O₄⁺ OLs. Concurrent exposure to Tat and morphine resulted in an exacerbation of activated caspase-3 expression as compared to either treatment alone. Naloxone (NAL; 1.5 μM) eliminated the Tat+Morph exacerbation, but did not block the effects of morphine alone.* p ≤ 0.005 vs. control; # p ≤ 0.005 vs. Tat+Morphine. N = 3 separate cultures.

Inducible Tat-transgenic mice were used to examine caspase-3 and TUNEL expression in OLs within the CNS. Tat⁺ and Tat^- mice were treated with combinations of DOX, morphine, and naltrexone for 2 days. OLs were identified by immunostaining with the APC marker. OLs in all groups of Tat^- mice, both untreated mice and those receiving DOX and/or opioid treatments, had low levels of active caspase-3 expression (≤ 5%) and none of the groups were significantly different from one another (Fig. 3). In contrast, levels of caspase-3 expression were significantly elevated within OLs in all groups of Tat⁺ mice. Both DOX-induced and uninduced Tat⁺ mice had similarly elevated expression levels, suggesting that the DOX-induced promoter has some constitutive activity, and that caspase-3 activation occurs with even low levels of Tat production. In Tat⁺ mice given DOX, concurrent morphine exposure resulted in a significantly increased number of APC⁺ OLs expressing active caspase-3 (Fig. 3). However, morphine administration did not exacerbate caspase-3 expression in the uninduced Tat⁺ mice, suggesting that opioid-Tat interactive effects on caspase-3 expression must require a threshold level of Tat that was not reached without DOX administration. Naltrexone (NTX) eliminated Tat+morphine additivity, showing that the caspase-3 response was opioid specific.

Figure 3.

Active caspase-3 expression is increased in OLs from Tat⁺ mice and is enhanced by co-treatment with morphine. Tat⁺ and Tat^- mice were treated with combinations of DOX, morphine (Morph), and naltrexone (NTX) for 2 days. Cells of the OL lineage were identified by APC staining within the corpus callosum. The percent of APC⁺ OLs expressing active caspase-3 was significantly increased in all Tat⁺ groups vs. all Tat^- groups (* P<0.01 vs. all Tat^- groups). Tat⁺ mice receiving DOX+morphine showed synergistic toxicity (# P<0.01 vs. all other Tat⁺ groups). Simultaneous treatment with Naltrexone (NTX) blocked the Tat-morphine synergism.

TUNEL results were somewhat different. In adjacent sections from the same mice, TUNEL expression was not increased by Tat alone, but was significantly increased in OLs treated with both Tat and morphine (Fig. 4). As would be expected, only a small percentage of APC⁺ OLs were also TUNEL⁺ in any of the groups. This is because cells at the stage of apoptosis where DNA is nicked are very rapidly phagocytized in vivo. Of great interest, the percentage of APC⁺ cells that were TUNEL⁺ was increased after 2 days of DOX treatment, but only in the Tat⁺ mice that also received morphine (*p = 0.03). NTX at this dose (60 mg/100 μL 50% DMSO) partially blocked morphine-induced synergy. The TUNEL labeling index of the Tat⁺/DOX/morphine/NTX group was reduced to a level that was no longer significantly different than Tat^-, Tat⁺, Tat⁺/DOX or Tat⁺/DOX groups, but at the same time was not different than the Tat⁺/DOX/morphine group. This partial blockade of morphine-induced synergy was in contrast to the complete blockade seen with caspase-3 expression.

Figure 4.

TUNEL expression is not increased by Tat alone, but is increased in OLs treated with both Tat and morphine. Tat⁺ and Tat^- mice were treated with combinations of DOX, morphine (Morph), and naltrexone (NTX) for 2 days. Dying OLs were identified using APC immunostaining and a fluorescent TUNEL marker. Only a small percentage of APC⁺ OLs also TUNEL⁺ in any of the groups. Cells at this stage of apoptosis are rapidly phagocytized in vivo. The percentage of APC⁺ cells that are also TUNEL⁺ was only increased in Tat⁺ mice receiving both DOX and morphine treatments, suggesting synergy between the morphine and Tat treatments (*p ≤ 0.03 vs. Tat^-, Tat⁺, Tat⁺/Morph and Tat⁺/DOX, non-parametric ANOVA; N=5). The TUNEL levels with NTX treatment were intermediate between the control and the Tat⁺/DOX/Morph groups, and were not significantly different from either. Thus NTX at least partially blocked the morphine-induced synergy.

Oligodendrocyte Morphology and Process Length

Mature OLs typically elaborate multiple processes, each of which has the potential to contact and form myelin around a separate axonal segment. The measurement of total process length per OL is therefore one indication of whether the OL population is phenotypically normal. OLs were impregnated using the Golgi-Kopsch technique (Hauser et al. 1989) to visualize and measure their process network. This staining technique results in random impregnation of the cell body as well as processes of individual neurons and glia and, as can be appreciated from Figure 5, detailed measurements of process extensions can be made by focusing through the body of the tissue section. Impregnation was done at 7 days after DOX administration since we initially thought that phenotypic alterations might not occur immediately after the transgene was activated. In OLs from the striatum and adjacent corpus callosum of control mice without the Tat transgene (Tat^-), the average total process length per cell was approximately 1,000 μm. This length was reduced significantly in mice with the Tat transgene (Tat⁺) regardless of the presence of DOX (Fig. 6). Since both the uninduced and the DOX-induced Tat⁺ mice showed equivalent reductions in process length, it is likely that the DOX-induced promoter has some constitutive activity and that process outgrowth is quite sensitive to continuous, low levels of Tat production. Evidence of such constitutive activity was also observed in the caspase-3 data (Fig. 3). Morphine appeared to dampen the Tat-induced reductions in process length in Tat⁺ mice both with and without DOX. However, the dose of NTX used here did not block the morphine-induced changes in process length.

Figure 6.

The total length of all cell processes was measured in random, well impregnated OLs within the corpus callosum. This total length averaged 1,000 μm per OL in mice without the Tat transgene (Tat^-), but was reduced significantly in mice with the Tat transgene (Tat⁺) regardless of the presence of doxycycline (DOX) (ANOVA; *p<0.05 vs Tat^-, Tat⁺/Morph, Tat⁺/DOX/Morph, Tat⁺/DOX/Morph/NTX). This finding of reduced process length in Tat⁺ mice without DOX induction suggests that the DOX-induced promoter has some constitutive activity. Morphine dampens this response in Tat+ mice both with and without DOX (*p<0.05 vs Tat⁺/Morph, Tat⁺/DOX/Morph), although overall process length in the Tat⁺/DOX/Morph group is still significantly less than in the Tat- group (^#p<0.05 vs. Tat^-). The dose of NTX used (60 mg/100 microL 50% DMSO) did not block the morphine effect. This may reflect the action of NTX on multiple opioid receptors, as well as endogenous secretion of multiple opioid peptides by OLs, as discussed in the text.

Figure 5 shows examples of Golgi-impregnated OLs from different genotypes and treatment groups. Large, mature, and well-arborized OLs were present in both Tat^- and Tat⁺ mice (e.g. Figs. 5A, B, H, and I). However, the Tat⁺ groups also had a population of mature OLs with an aberrant phenotype. These abnormal cells were more evident in Tat⁺ mice receiving doxycycline, and especially so when morphine was co-administered. Some OLs had reductions in total number of cytoplasmic processes, and/or stunted or abbreviated processes (Fig. 5F, G, J, L, and M). Others had abnormal varicosities along their length or large, club-like cytoplasmic endings that could be several microns in diameter (Figs. 5F, G, I, J, and K). Normal OLs typically have a smooth-surfaced round or ovoid cell body (Figs. 5A, C, and D), yet in many Tat⁺ OLs the cell body was irregular in shape, and/or had a rough, wrinkled appearance (Figs. 5I, J, and M). Others had cytoplasmic/myelinating processes that failed to align with processes of other OLs in white matter tracts (Figs. 5C, D, and M). Occasional cells were so severally degenerated that their identity could not be determined with certainty, although the size of the soma and location relative to white matter tracts suggested an OL identity. Abnormal OLs did not appear to correlate with a particular subregion of the striatum, but may coincide with heterogeneity in Tat expression among astrocytes since we noted a tendency for aberrant cells to cluster in foci, with otherwise normal appearing OLs in the surrounding areas. OLs with aberrant morphology were only occasionally observed in Tat^- mice, and these abnormal cells might represent normal glial turnover processes (Fig. 5E).

MBP Levels

As an index of overall myelination, levels of MBP protein were assessed by immunoblotting in cortical samples taken from Tat^- and DOX-exposed Tat⁺ mice. The Tat⁺ mice had received DOX in their drinking water for 4 weeks prior to tissue harvesting. The results therefore reflect the effects of a more prolonged exposure to elevated Tat than other endpoints reported here, and which were taken after 2 days of DOX exposure. Quantitative densitometric analysis of MBP immunoblots from DOX-exposed Tat⁺ mice (n=4) and Tat^- mice (n=4) were essentially identical (ratio 1.034) after correction for loading. This suggests that any alterations in myelin integrity that occur with Tat exposure are more subtle than can be detected as the loss of a major myelin protein.

Discussion

Our data indicate that OLs are cellular targets of the HIV-1 Tat protein, both in vitro and within the CNS. Findings in culture suggest that some effects could be direct, although it is likely that effects observed in Tat-transgenic mice also have an indirect component. Many, but not all, of these effects are exacerbated by concurrent exposure to morphine. Others, such as increased TUNEL labeling, are only observed in the presence of morphine. Thus, exposure to opioid agonists with abuse potential may have profound effects on the relationships between OLs and neurons/axons in HIV-infected individuals, especially considering the extended time course of the disease process.

Our motive for exploring this issue was to determine whether HIV-1 Tat and opiates might have synergistic effects on OLs, as shown previously by our laboratories for neurons, astroglia, and microglia (Bruce-Keller et al. 2008; El-Hage et al. 2005; El-Hage et al. 2006b; Gurwell et al. 2001). Subpopulations of both glial progenitors and OLs express several types of opioid receptors, suggesting that they might be subject to opiate-induced synergism. Our initial in vitro studies showed that exposure of young, O₄⁺ OLs to Tat alone or morphine alone caused an increased immunohistochemical expression of active caspase-3. When exposed simultaneously, the effect was additive and additivity was blocked by the opioid antagonist naloxone (Fig. 2). This led to a further examination of potential toxic and other effects in vivo, using transgenic mice that conditionally express HIV-1 Tat_1-86 in astroglia. We examined activation of caspase-3, a pro-apoptotic protease associated with release of caspase-activated DNAse and terminal DNA fragmentation. However, since our lab and others have shown that caspase-3 can be activated to high levels in situations where cells do not die (Adjan et al. 2007; Campbell and Holt 2003; Garrido and Kroemer 2004; Jarskog et al. 2007; Launay et al. 2005; McLaughlin 2004; Oomman et al. 2004; Oomman et al. 2005; Yang et al. 2004a; Yang et al. 2004b), we additionally examined TUNEL labeling as a more unambiguous indicator of cell death. We also measured the arborization of OL processes as an index of whether Tat and/or morphine might cause non-lethal changes that impact OL function. Our results showed that all parameters had a clear-cut response to Tat and/or morphine. These responses were very endpoint specific, varying in sensitivity and threshold, and suggesting that a complex set of signaling responses is initiated by Tat and morphine exposure. All groups of Tat⁺ mice, even those that were not exposed to DOX, showed significant upregulation of caspase-3 expression versus all Tat^- groups in APC+ OLs (Fig. 3). This likely indicates some constitutive expression of Tat due to promoter “leakiness”, especially since OLs in Tat+ mice also displayed changes in overall process length prior to DOX exposure (Fig. 6). Tat mRNA transcripts and immunostaining had previously been detected at low levels in the absence of DOX in a similar transgenic mouse (Kim et al. 2003). Of great interest, OLs from the same Tat+ mice with upregulated active caspase-3 expression showed no increase in TUNEL labeling, except when treated with both DOX and morphine (Fig. 4). This suggests that caspase-3 expression did not necessarily lead to cell death. Further, it shows that Tat and morphine have additive effects on OLs in vivo, as previously described in the brain and/or in vivo for neurons (Gurwell et al. 2001; Hu et al. 2005; Turchan-Cholewo et al. 2006), astroglia (El-Hage et al. 2005; El-Hage et al. 2006a; El-Hage et al. 2006b; Khurdayan et al. 2004) and microglia (Chao et al. 1994; El-Hage et al. 2006b). It is not clear whether all cells in the OL lineage are equally affected since APC staining does not distinguish between mature and immature stages.

The small amounts of Tat that are constitutively released over the lifetime of Tat⁺ mice appear to have a profound effect to reduce the average total process network in OLs (Fig. 6). Induction with DOX for 7 days does not increase this effect, suggesting an ongoing response at a very low threshold of sensitivity. This reduction might be due to the actual retraction or loss of processes in mature cells. However, it could also reflect a population of less well arborized OLs whose differentiation has been compromised, or a proportional increase in a population of more immature OLs. The effects of Tat on glial population dynamics are of great interest and are under investigation. The ability to maintain multiple, myelinating processes that ensheathe an adequate number of axons is fundamental to OL function. Thus, while not a lethal response in itself, a reduction in process formation might seriously impact OL function.

Morphine co-administration seemed to mitigate some of the Tat effect on OL process length. This is contrary to our observations for caspase-3 and TUNEL in OLs, and for other endpoints in astrocytes and microglia, where morphine generally exacerbates the effect of Tat alone. While morphine might be stabilizing OL processes or encouraging outgrowth, there is also a possibility that the population has been shifted towards larger cells with a larger process network through toxic effects directed towards immature OLs and/or their progenitors.

It is difficult to explain why the opiate antagonist NTX did not completely reverse the observed effects of morphine on all outcome measures. It did reverse morphine effects on caspase-3 in vitro (Fig. 2), and in vivo (Fig. 3), and significantly attenuated TUNEL effects (Fig. 4). In the same mice, total OL process length was measured after DOX±morphine±NTX treatments. There was a robust effect of morphine to antagonize Tat effects, both in the absence and presence of DOX, but this was not reversed by NTX (Fig. 6). Although there are potential explanations for this observation, it may be extremely difficult to convincingly prove or disprove any of them, because opioids can modulate normal OL development. OLs express all major opioid receptors (Knapp et al. 2001; Knapp et al. 1998; Stiene-Martin et al. 2001; Tryoen-Toth et al. 2000) and in addition synthesize KOR and DOR selective peptides (dynorphins, proenkephalins) (Knapp et al. 2001). KOR and MOR agonists have documented effects on OL differentiation and survival (Knapp and Hauser 1996; Knapp et al. 2001; Knapp et al. 1998; Sanchez et al. 2008). While total process length reflects the process outgrowth of individual cells, it also reflects population shifts due to net changes in OL survival or maturation. For example, if mature OLs represent less of the total population, either because they die, or because there is increased progenitor formation, then average arborization may decrease. Vice versa, if there are proportionately more mature OLs, average total process length may increase. Since neither morphine nor NTX are entirely selective for particular opiate receptor types, both compounds have the potential to interfere with multiple endogenous opioid signals that affect measurements of average OL process length. Thus, one possible explanation for the paradoxical effects of NTX in this study is related to the complexity of opioid signaling in OLs. A broad-spectrum opioid receptor antagonist like NTX may disrupt endogenous opioid interactions that affect OL survival and differentiation, either in a primary or secondary manner, and have effects unrelated to blocking exogenous morphine. Competing effects mediated through multiple opioid signaling pathways that converge on the same endpoint could mask NTX effects at MORs. Studies with more specific agonists and antagonists might not clarify this situation, since long-term exposure would almost certainly disrupt normal OL function. For now, the mechanism behind this particular result with NTX remains an unresolved point.

Lastly, some caution should be exercised when interpreting studies such as this with long-term exposure to opiates. Although NTX completely blocks morphine's acute effects (Bruce-Keller et al. 2008; El-Hage et al. 2006a; El-Hage et al. 2006b), chronic administration of NTX, in general, adaptively increases opioid peptides and receptors, and can alter coupling to second messenger systems, thus modifying the response to morphine (Christie 2008). Chronic morphine administration itself causes a decline in MOR signaling due to tolerance and dependence, while concurrently upregulating DOR and altering receptor heterodimerization (Cahill et al. 2001; Rozenfeld and Devi 2007). These and other chronic, adaptive changes will likely alter cellular responses to morphine and NTX in unpredictable ways.

At one level, finding that OLs were sensitive to the Tat protein was unexpected, since earlier work in HIV patients showed no gross demyelination or reduction in myelin staining intensity in the CNS (Power et al. 1993). In similar immunoblotting studies to examine the abundance of CNS myelin basic protein, we found no differences between Tat⁺ and Tat^- control mice. In light of the results showing OL abnormalities, several possibilities are suggested. Tat and/or morphine exposure may affect OL process formation and caspase-3/TUNEL expression, but those effects may not disrupt myelination or reduce myelin protein levels to an extent detected by immunoblotting or staining. These more subtle changes might still disrupt normal OL-axon relationships and eventually perturb neuron function. OL loss, or aberrant myelin production by some OLs might also be compensated by increased myelin production per cell, or by recruitment of cells from an adult progenitor pool. We did note a population of smaller OLs with fewer processes in the Tat⁺ groups, and although we believe these to be OLs retracting their processes, some could also be immature OLs. It is not clear that results observed in the transgenic mice are all due to direct actions of Tat and/or morphine on OLs, but cell culture studies do show that Tat/morphine can directly activate caspase-3 within OLs. And while the effects of Tat and morphine on OL lineage cells could be mouse-specific, this seems unlikely since the majority of effects observed for astroglia, microglia and CNS neurons in the Tat transgenic model bears out what is believed to be the situation in the human disease.

What meaning do the present results have in terms of human disease and neurological function in HIV patients? White matter “pallor” has been frequently described on MRIs from HIV patients with neuroAIDs (Cherner et al. 2002; Glass et al. 1993; Gray et al. 1992; Gray et al. 1996; Lanjewar et al. 1998; McArthur et al. 1989; Schwartz and Major 2006), although this has been attributed largely to edema associated with protein leakage through a compromised endothelium. Our results do not exclude this effect, but rather suggest that OLs are also direct and/or indirect targets of the Tat protein and/or inflammatory changes that Tat initiates within the brain. HAART treatment has improved the life expectancy of many HIV patients. In these patients, where the disease is a more chronic condition, the consequences of subtle changes in myelination over a longer time course may become a more important part of the clinical picture. Our previous work showed that Tat and opioids have synergistic effects on neurons, implying that concurrent opioid abuse by HIV patients may directly increase the risk of neurological deficits. The present results show that OLs are also vulnerable to this synergy, suggesting an additional basis for the prevalence of neuroAIDs among opioid users. We also noted that immature OLs derived from glial progenitor cultures were synergistically affected by exposure to Tat and morphine. These results extend our previous findings, which showed that more immature progenitors from cortex and spinal cord were targets of Tat and opioid toxicity (Buch et al. 2007; Khurdayan et al. 2004). Since glial production and myelination continue for several years after birth in humans, these findings on glial progenitors and young OLs may be particularly relevant to pediatric HIV patients, who have a higher prevalence and accelerated rate of onset of neurological complications compared to adult HIV patients (Drotar, Olness 1997; Ensoli, Wang 1997; Schwartz & Major, 2006; Van Rie, Harrington 2007). The adult CNS also contains a population of glial progenitors (Gensert and Goldman 2001; Wolswijk and Noble 1989), which might be vulnerable to HIV protein/opioid exposure. Studies to examine the response of glial progenitors to opiates and viral proteins at various developmental points in both the immature and mature CNS are currently underway.

Overall, our results suggest that OLs are extremely sensitive to HIV-1 Tat exposure. The consequences of Tat exposure can be either selectively exacerbated or alleviated by concurrent administration of morphine, depending upon the endpoint in question. Tat and opioid signaling may thus be integrated through convergence on multiple signaling pathways within OLs. When active caspase-3 expression and process length were used as endpoints, OLs were sensitive even to the low levels of Tat produced without DOX induction. We have not seen this in other CNS cells studied in these transgenic mice, including astroglia, striatal neurons and microglia/monocytes, using any of several endpoints (Bruce-Keller et al. 2008)(Knapp and Hauser, unpublished data). Thus OLs, in contrast to other CNS cell types, are either more sensitive to low levels of Tat or do not habituate as readily to its presence. Assuming that OLs in the human CNS behave similarly, the endpoints that we have studied suggest a deleterious effect of Tat exposure on OLs, and thus on OL-neuron relationships, in both the adult and developing CNS. The Tat and morphine effects on OLs described in our studies may provide an alternative or additional explanation for white matter abnormalities observed in some HIV patients. As HAART treatment lengthens survival time, functional changes due to altered glial populations may contribute increasingly to the neurological complications observed in HIV patients, and especially in those exposed to opioid drugs of abuse.