Repeated cocaine or methamphetamine treatment alters astrocytic CRF2 and GLAST expression in the ventral midbrain

Abstract

Dopamine neurons in the substantia nigra (SN) and ventral tegmental area (VTA) play a central role in the reinforcing properties of abused drugs including methamphetamine and cocaine. Chronic effects of psychostimulants in the SN/VTA also involve non-dopaminergic transmitters, including glutamate and the stress-related peptide corticotropin-releasing factor (CRF). In the SN/VTA, astrocytes express a variety of membrane-bound neurotransmitter receptors and transporters that influence neurotransmission. CRF receptor type 2 (CRF2) activity in the VTA is important for stress-induced relapse and drug-seeking behavior, but the localization of its effects is incompletely understood. Here, we first identified CRF2 transcript in astrocytes of the SN/VTA using RNA-Seq in Aldh1l1;NuTRAP mice, and confirmed it using in situ hybridization (RNAscope) in wild type mice. We then used immunofluorescence to quantify the astrocytic marker protein S100β, glial-specific glutamate/aspartate transporter GLAST, and CRF2 in the SN/VTA following 12 days of treatment (i.p.) with methamphetamine (3 mg/kg), cocaine (10 mg/kg), or saline. We observed a significant decrease in GLAST immunofluorescence in brains of psychostimulant treated mice compared to saline controls. In addition, we observed increased labelling of CRF2 in drug treated groups, a decrease in the number of S100β positive cells, and an increase of co-staining of CRF2 with both S100β and tyrosine hydroxylase (dopamine neurons). Our results suggest a significant interaction between CRF2, GLAST, and astrocytes in the midbrain that emerges with repeated exposure to psychostimulants. These findings provide rationale for future investigation of astrocyte-based strategies for altering cellular and circuit function in response to stress and drug exposure.

INTRODUCTION

Drug use disorders are driven by numerous chronic neuroadaptations that occur in response to repeated exposure to the drug. For psychomotor stimulants such as cocaine and methamphetamine many of these effects occur in the ventral midbrain, which is home to tyrosine hydroxylase-expressing dopaminergic cell bodies in the ventral tegmental area (VTA) and substantia nigra (SN). While the direct role of dopaminergic neurons on reward-seeking and motivated behavior has been extensively studied, much less is known about the effect of psychostimulants on other cell types in these areas. In addition to neuronal populations, the VTA and SN contain glial cells, including astrocytes.^1,2 Astrocytes have an established role in glutamate homeostasis, and glutamate input is essential for burst firing of dopamine neurons.³ Indeed, our collaborative work recently established that astrocytes can affect the activity of midbrain dopamine neurons through a local circuit and have direct effects on reward behavior. ² Thus, drug-induced alterations occurring in astrocytes in the VTA and SN could be predicted to have downstream effects on motivated behavior, and may have a significant impact on both addiction and relapse.

Use of both cocaine and methamphetamine has been linked to increases in neuromodulators of stress including corticotropin releasing factor (CRF)^4,5, and several theories link the increased CRF activity to negative reinforcement seen with addiction and relapse.⁶ In the VTA, endogenous CRF activity increases extracellular levels of glutamate and dopamine, but only in animals with previous cocaine self-administration experience.⁷ While a role of CRF receptor type 1 (CRF1) has been established in drug abuse and addiction, CRF receptor type 2 (CRF2) is less well understood. Stress produces reinstatement of cocaine-seeking in animals after extinction⁸ that can be blocked by intra-VTA administration of a CRF2 receptor antagonist⁹, suggesting a role of CRF2 in the VTA in drug-related behaviors. CRF2 mRNA has been reported in a subpopulation of VTA dopaminergic cells^10,11, and CRF2 effects have historically been attributed to receptors on this neuronal population. Together these data support a possible role of VTA CRF2 in drug addiction. The synaptic effects of CRF in the VTA are themselves complex, and alterations produced by cocaine include seemingly competing effects on glutamate, GABA, and dopamine neurotransmission^11–16 that may involve both neurons and interactions with astrocytes. Thus, while the VTA, and likely the SN, are positioned to act as a convergence point between stress and dopaminergic activity central to reward and drugs of abuse, the mechanisms involved are complex and incompletely understood.

While conducting a transcriptomic analysis of cell types in the midbrain, we unexpectedly observed that CRF2 transcript is expressed not only by neurons in the SN/VTA but also by astrocytes, which has not been previously reported in the midbrain. This novel information could provide a missing piece to improve our understanding of the complex role of CRF in the midbrain, both in normal physiology and subsequent to drug exposure. Given the link between CRF and glutamate observed following drug exposure⁷, the objective of this study was to determine the effect of repeated administration of cocaine and methamphetamine on CRF2 in the SN/VTA along with the glial glutamate/aspartate transporter GLAST (EAAT1). Mice were treated sub-chronically with either cocaine or methamphetamine, followed by in situ hybridization and immunofluorescence to visualize alterations in neurons and astrocytes that could ultimately modulate dopaminergic output from the VTA and SN.

METHODS

Animals:

All experiments were approved by the Oklahoma Medical Research Foundation (OMRF) Institutional Animal Care and Use Committee, and procedures were consistent with those described in The Guide for Care and Use of Laboratory Animals. For the RNA-Seq study, adult, male Aldh1l1-Cre/ERT2⁺;NuTRAP⁺ mice (hereafter Aldh1l1;NuTRAP)¹⁷ were housed in a vivarium with a normal 12h:12h light/dark cycle (lights on at 0700). For all other studies, male C57Bl/6J mice (8-10 weeks on arrival, Jackson Laboratory, Bar Harbor, ME) were group housed in a vivarium on a reversed 12-h light/dark cycle (lights off at 0900). The vivaria were held at constant temperature (21±1°C). Mice were housed in clear, polycarbonate cages with bedding and additional materials such as Nestlets for environmental enrichment. Food (PicoLab Rodent Diet 20, Catalog #5053) and water were available ad libitum in the home cage throughout the study.

Translating ribosome affinity purification (TRAP) and RNA isolation:

Aldh1l1;NuTRAP mice were treated with tamoxifen (100 mg/kg body weight, 20 mg/ml stock solution, #T5648; Millipore Sigma, St. Louis, MO) for five consecutive days at approximately 3-months of age to induce expression of Cre recombinase in astrocytes. At approximately 6-months of age, mice (n = 5) were euthanized by cervical dislocation and a 600-μm horizontal brain section containing the ventral midbrain was collected from each mouse using a vibrating microtome (VT1000S, Leica Biosystems). For each mouse the SN/VTA were further dissected from this section and placed in a single tube. Using established protocols^17–19 with minor modifications, TRAP was then performed on this sample. The tissue was minced before being placed in a buffer and homogenized using a Kimble Pellet Pestle^™ Motor (#749540, Kimble Histology) and Kimble pestle (#41101702, Kimble Histology) for ten seconds. The resulting suspension was centrifuged, 100 μL of supernatant was saved as the input fraction, and the remainder was used for further processing. For the immunoprecipitation step, anti-GFP antibody (ab290; Abcam) was added and rotated for an hour at 4°C. Following incubation, washed Protein G Dynabeads^™ (#10003D; Thermofisher) were added to the mixture, which was rotated at 4°C overnight. The next day, beads were magnetically separated and washed with a salt-buffer. β-mercaptoethanol and Buffer RLT (Qiagen) were added to the beads and mixed at room temperature. Unbound magnetic beads were then separated and the supernatant containing the target polysomes and their RNA was loaded onto an RNeasy MinElute column, and RNA was isolated following the manufacturer’s instructions using the RNeasy Mini Kit (#74104, Qiagen). RNA quantification was performed by a DS-11 FX spectrophotometer (DeNovix) and RNA quality was analyzed with a HSRNA screentape on a 4150 Tapestation analyzer (Agilent Technologies).

Library construction, RNA sequencing, and analysis:

Previously established protocols¹⁷ were used except as noted. Briefly, 25 ng of total RNA from TRAP input and positive fractions was used to carry out library construction using the NEBNext Ultra II Directional Library Prep Kit for Illumina (#NEBE7760L; New England Biolabs Inc., Ipswich, MA) following manufacturer’s instructions. Sample mRNA was then fragmented and cDNA was synthesized. Incorporation of dUTP instead of dTTP in the second strand of cDNA synthesis allowed for strand specificity, and end repair was subsequently performed. Library size and quality was assessed by Tapestation (4150; Aligent) using HSD1000 Screentape (5067-5584; Agilent Technologies) and further quantified by QuBit 4 Fluorometer (Q33226; Invitrogen). Dual indexing and transposon mediated fragmentation of libraries using NEBNext Multiplex Oligos for Illumina (96 index Primers, E7600S) was conducted by manufacturer’s instructions. Following the addition of indices, PCR amplification of libraries was carried out to ensure integration of adapters. The libraries were then pooled at 4nM concentration and sequenced using Illumina NovaSeq 6000 system (SP 150bp) at the OMRF Clinical Genomics Core.

Trimming and alignment of reads against Mm10 build of the mouse genome (2014.11.26), and filtering of criteria was performed using Strand NGS software package (Agilent). Consensus cell type marker genes^17,20, were examined for fold change enrichment (log2(TRAP+/Input)) of astrocyte markers and depletion of markers for other cells types. Statistical assessment between TRAP+ and input fractions was assessed by a moderated t-test with a Benjamini-Hochberg Multiple Testing Correction. Additionally, a statistical criterion of >|1.25|was used to ensure only transcripts that were statistically and biologically significant were included within the analysis. A subsequent heatmap of relative gene expression was generated using Morpheus (https://software.broadinstitute.org/morpheus).

In situ hybridization:

For colocalizing CRF2 and the astrocyte marker S100β within cells in the VTA/SN we used RNAscope^® manual assay with fluorescence. Group-housed adult, male C57Bl/6J mice were injected daily (0.1 ml/10 g body weight, i.p.) for 12 consecutive days with either sterile saline (n = 4) or a moderate stimulant dose of 3 mg/kg methamphetamine (n = 4). Approximately 24 h after the last injection, the mice were sacrificed humanely and whole brain tissue was immediately harvested and flash frozen using methyl-butane and dry ice. Brains were then embedded in OCT in tissue blocks and stored in an air-tight container at − 80° C. Ten-micron thick coronal sections of the brains were sectioned onto UltraFrost with UltraStick slides (Thermo Scientific #3039) using a cryostat (ThermoScientific CryoStar NX70). Slides with brain sections were kept in airtight containers at − 80° C until RNAscope was initiated. Manual fluorescent RNAscope was performed using the RNAscope Multiplex Fluorescent Reagent Kit (Cat. No. 323133) and followed the guidelines provided by the ACD Fresh Frozen Tissue Protocol. Experimental probes for S100b (Cat. No. 431731) and Crhr2 (Cat. No. 413201) were used to stain for colocalization within coronal brain sections of VTA/SN. Proper positive (Cat. No. 320881) and negative controls (Cat. No. 320871) were applied to ensure efficacy of experimental probes. Coronal sections (1 section per mouse at the same rostral/caudal level) were imaged in the Geroscience Molecular and Cellular Imaging Core using a Leica Thunder 3D microscope with appropriate filters for the fluorescent probes used. Images were analyzed using Image-J software by an experimenter blinded to the treatment. Positive cells were counted in the region of interest (VTA/SN) for S100b and Crhr2 fluorescence, and the percent of colocalization of S100b and Crhr2. Results were compared statistically with an unpaired t-test.

Immunofluorescence procedures:

Male, adult C57Bl/6J mice (group-housed) were assigned to one of the following treatment conditions (saline, n = 8; 10 mg/kg cocaine, n = 7; 3 mg/kg methamphetamine, n = 6). Cocaine and methamphetamine were provided by the NIDA Drug Supply Program (Bethesda, MD) and were dissolved in sterile saline. Daily at 1400, mice were weighed, injected (0.1 ml/10 g, i.p.), and placed in a clean cage for 5-10 minutes for observation of behavior before being returned to their home cage. Treatment continued for 12 days. Approximately 24 hours after the final injection of saline, methamphetamine, or cocaine, mice were deeply anesthetized with 2,2,2-tribromoethanol (300-500 mg/kg, i.p.) then transcardially perfused with 10% sucrose in phosphate buffered saline (PBS) followed by 4% paraformaldehyde in PBS for approximately 90 s each using a Perfusion Two system (Leica). The brain was removed and postfixed in 4% paraformaldehyde in PBS overnight at 4°C before being moved to a 30% sucrose solution in PBS for cryoprotection. A cryostat was used to section the brains coronally at 40 μm thickness for immunohistochemistry. All immunofluorescent staining was conducted on freely floating sections in 12-well plates, and incubations were conducted with gentle agitation at room temperature unless otherwise stated. Sections were processed with primary and secondary antibodies as described in Table 1. Briefly, sections were first washed in fresh PBS for at least 5 minutes before a 30-minute blocking incubation with 5% Normal Donkey Serum in PBS. Sections were next incubated with donkey anti-mouse F(ab) fragment (Jackson ImmunoResearch) in PBS for 90 minutes. A 0.3% Tween 20 solution in PBS was used to wash the slices at least three times before the addition of primary antibody mixture containing anti-tyrosine hydroxylase, anti-S100β, and one other antibody (either anti-CRF2 or anti-GLAST) overnight at 4° C. The next afternoon the sections were washed at least 3 times with PBS, and then the secondary antibody mixture was added at room temperature for 2 hours. The sections were then subjected to at least 3 final rinses with PBS before they were mounted on gelatin-coated slides, and cover slips were applied with ProLong Gold Antifade Mountant (ThermoFisher, Grand Island, NY). Slides were imaged in the OMRF Imaging Core on a Zeiss Axiovert 200M inverted fluorescence microscope using a 20x or 40x objective with filters with appropriate excitation and emission spectra for each of the secondary antibodies. For analysis, two images (Bregma: from - 2.91mm to −3.39mm²¹) were analyzed per mouse, and an average of the measures was used for statistical analysis. Using ImageJ software (NIH), a consistent threshold was set for all images, and the region of interest (ROI) was drawn based on tyrosine hydroxylase staining (cell bodies and projections) and aligned with the mouse brain atlas (Paxino and Franklin) to define the VTA or SN. The area within the ROI that was positive for staining was quantified automatically (% area fluorescence) by Image-J. Cell counting within the ROI was conducted manually by a blinded experimenter using Cell Counter add-in within Image-J. Overlap of immunofluorescence signals was analyzed with Fiji software using cell counter. Threshold was set automatically using Otsu’s method. For coexpression of TH and CRF2, we used this composite image to confirm which cells were positive for tyrosine hydroxylase, and then calculated what percentage of them were also positive for CRF2 fluorescence. For co-expression of S100β and CRF2, after setting threshold using Otsu’s method we confirmed the immunopositive cells for S100β and then determined what number of cells were also positive for CRF2. Statistics were performed using GraphPad Prism (GraphPad Software, San Diego, CA). One-way ANOVAs were used to compare saline versus cocaine and methamphetamine groups, with a Dunnett’s post-hoc to compare groups relative to saline.

Table 1:

Vendors and concentrations for primary and secondary antibodies used for: Immunofluorescence

Primary antibody, catalog #	Source	Concentration	Secondary antibody	Source	Concentration
Mouse anti-mouse S100β, ab212816	Abcam	1:750	Donkey anti-mouse AlexaFluor488, 715-545-150	Jackson ImmunoResearch Laboratories	1:750
Donkey anti-chicken tyrosine hydroxylase, ab764421		1:1000	Donkey anti-chicken Rhodamine Red-X, 703-295-155		1:225
Donkey anti-rabbit CRF2, 720291	Invitrogen	1:250	Donkey anti-rabbit AlexaFluor647, ab150067	Abcam	1:1000
Donkey anti-rabbit GLAST, 100-1869	Novus	1:300

^*The primary mixture incubated overnight at 4°C with agitation, and the secondary mixture incubated for 2 hours at room temperature with agitation.

Data availability.

Data from the RNA-Seq experiment have been deposited in GEO repository with the GSE184497 accession code (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE184497; data used for Figure 1). Other data are available from the corresponding author (ALS) upon request.

Figure 1: Crhr2 mRNA is expressed by astrocytes in the SN/VTA.

A heatmap of cell-specific marker gene expression levels following TRAP isolation from Aldh1l1;NuTRAP SN/VTA shows enrichment of astrocytic genes and a depletion of neuronal, microglial, and oligodendrocytic genes in our sample (positive fraction) relative to the input fraction (A). Violin plots showing the enriched expression (fold change, FC) of astrocytic markers and depletion of other cell type markers in the positive vs. input fraction (B). Relative gene expression level of specific neuronal and astrocytic marker genes in the positive and input fractions confirm enrichment in astrocytic markers in the positive fraction with depletion of pan-neuronal and dopaminergic cell markers that are expected from SN/VTA whole tissue (C). Expression levels of Crhr2 (CRF2) in the SN/VTA input vs astrocyte-enriched positive fraction do not significantly differ and are actually slightly higher, suggesting that while CRF2 is expressed by astrocytes in addition to other cells types, presumably neurons, in the SN/VTA (D).

RESULTS

RNA Sequencing (RNAseq).

Astrocyte transcriptomes were enriched through collection of eGFP-bound polysomes from Aldh1l1;NuTRAP mouse midbrains via the translating ribosome affinity purification (TRAP) protocol. From the collected input and positive fractions, RNA was isolated for RNAseq. Transcriptomic data indicated significant enrichment in astrocytic marker genes and depletion of neuronal, oligodendrocytic, endothelial, and microglial markers in the positive fraction (Fig 1A, 1B). Investigation into genes coding for specific astrocyte (Gfap, S100b, Aldh1l1, Slc1a3) and region specific-neuronal markers (Slc6a3, Spink10, Kcns3, Park2, Drd2, Ntsr1, Cacnca1c) were consistent with astrocyte enrichment (Fig 1C). Surprisingly, in the positive (astrocyte) fraction, Crhr2 (which codes for CRF2) was present at levels similar to the input fraction, indicating that the receptor’s expression in SN and VTA astrocytes is similar to the bulk tissue sample that includes neuronal populations (Fig 1D).

In situ hybridization:

We next used RNAscope for in situ hybridization to visually assess and confirm expression of Crhr2 in astrocytes in the VTA. We were able to demonstrate colocalization of Crhr2 in a small number of astrocytes in the VTA using S100β as an astrocytic marker (Fig 2A). Interestingly, VTAs from mice treated with methamphetamine (3 mg/kg) for 12 days exhibited a greater colocalization of CRF2 with S100β (Fig 2B). Summary data indicate that mice treated with methamphetamine consistently had higher rates of colocalization of CRF2 with astrocytes than saline treated mice (Fig 2C; t(6) = 3.534 , P = 0.0123).

Figure 2: In situ hybridization shows colocalization of Crhr2 with VTA astrocytes in saline- and methamphetamine-treated mice.

RNAscope for Crhr2 (green) and S100b (red) signal, with DAPI (blue) to mark nuclei, in coronal VTA sections shows detection of individual signals (open arrowheads) and colocalization (closed arrowheads) in saline- (A) and methamphetamine- (B) treated mice. Methamphetamine (Meth) treated mice had an increased percentage of cells that contained Crhr2 signal that colocalized with S100b signal (C). *** P < 0.05

Immunofluorescence:

Next, we sought to confirm and extend these findings beyond transcript levels using immunostaining for multiple proteins in the SN/VTA. Adult, male C57Bl/6J mice were injected (i.p.) with saline or moderate doses of the psychostimulants cocaine (10 mg/kg) or methamphetamine (3 mg/kg) for 12 consecutive days. Mice were sacrificed on the following day, and coronal brain sections containing SN/VTA were processed for immunofluorescence. Quantification of immunofluorescence showed significant effects of treatment on astrocyte number, as well as CRF2 and GLAST staining in the ventral midbrain. Both in the VTA (F_2,18 = 11.1, P = 0.0007) and in the SN (F_2,18 = 9.0, P = 0.002) there was a main effect of psychostimulant treatment on the amount of CRF2 fluorescence, showing a significant increase from saline (Fig 3 column 3, Fig 4A, 4B). Astrocyte number (measured as number of cells positive for S100β) was decreased somewhat following psychostimulant treatment in both the VTA and SN; however, only the VTA had a statistically significant main effect of treatment (Fig 3 column 2, Fig 4C, 4D; VTA: F_2,18 = 6.9, P = 0.006; SN: F_2,18 = 2.1, P = 0.15). Posthoc analysis indicated that both cocaine and methamphetamine treatments produced a significant decrease from saline for S100β positive cell number in the VTA. In confirmation of our RNAseq and in situ hybridization results, we also observed co-labeling of CRF2 with both tyrosine hydroxylase- (TH, dopamine cells) and S100β (Fig 3 column 4). On analysis of CRF2 in relation to S100β and TH, we observed that the percentage of S100β and TH cells that colocalized with CRF2 both showed a main effect of psychostimulant treatment in the VTA and SN (S100β –Fig 4E, 4F; VTA: F_2,18 = 76.7, P < 0.0001; SN: F_2,18 = 9.7, P = 0.0014) (TH—Fig 4G, 4H; VTA: F_2,18 = 24.1, P < 0.0001; SN: F_2,18 = 10.4, P < 0.001).

Figure 3: Representative panels of immunofluorescent staining in the VTA for saline-, cocaine- and methamphetamine-treated mice.

Staining for tyrosine hydroxylase (TH) positive cells (1) is representative of the dopaminergic neurons in the VTA. Immunofluorescence for the astrocytic marker S100β (2) was decreased following treatment with psychostimulants compared to saline. CRF2 immunofluorescence (3) was significantly increased following treatment with cocaine and methamphetamine (METH). Coronal sections (40 μm) in VTA of adult C57Bl/6J mice co-labeled for TH (red), S100β (green) and CRF2 (blue) at 20x (4). Yellow arrows point to singly labelled cells (TH or S100β) and white arrows point to double labelled cells (CRF2 + either TH or S100β). Scale bar equals 100 μm.

Figure 4: Repeated cocaine and methamphetamine injections increase CRF2 signal and decrease astrocyte number in the VTA and SN.

Immunofluorescence for CRF2 in the VTA (A) and SN (B) is significantly increased from saline following repeated cocaine or methamphetamine (METH) injections. Conversely, the number of S100β positive cells was significantly decreased in the VTA of cocaine and methamphetamine injected mice compared to the saline group (C) but this did not reach statistical significance in the SN (D). Colocalization of S100β and CRF2 signal expressed as the percentage of total S100β cells that showed co-staining with CRF2 was significantly increased in the cocaine and methamphetamine groups compared to the saline group in both the VTA (E) and the SN (F). Likewise, the colocalization of CRF2 and TH (expressed as the % of TH-fluorescence that colocalizes with CRF2-fluorescence) was also increased in the VTA (G) and SN (H) of psychostimulant-treated groups compared to saline. * P < 0.05, ** P < 0.01, *** P < 0.001 compared to saline

Astrocytes play a central role in glutamate homeostasis, and in the midbrain CRF signaling is linked to extracellular glutamate levels in rats with cocaine experience⁷. Examination of the effect of repeated psychostimulant treatment on expression of the glial glutamate/aspartate uptake transporter GLAST (EAAT1) showed a significant main effect of treatment in both the VTA and SN (Fig 5; VTA: F_2,19 = 10.2, P = 0.0011; SN: F_2,19 = 11.2, P = 0.0006). Dunnett’s post-hoc revealed a significant decrease from saline in the VTA for methamphetamine, and in the SN for both psychostimulants (Fig 6).

Figure 5: Immunofluorescent labeling of TH (dopaminergic cell) and the glial glutamate transporter GLAST in the VTA of mice treated with saline, cocaine, or methamphetamine.

Representative images (20 x magnification) for TH (red), GLAST (blue), and merged images in the VTA of mice treated with saline, cocaine, and methamphetamine (METH) showing that GLAST signal was decreased following cocaine- and methamphetamine treated mice.

Figure 6: Expression of the glial glutamate transporter GLAST is decreased following repeated treatment with cocaine and methamphetamine.

In the VTA (A) there was a main effect of treatment for GLAST fluorescence (expressed as the % of the area measured that was fluorescent), with a post-hoc test indicating a significant decrease from saline in the methamphetamine (METH) treated group only. Both cocaine and methamphetamine treatment significantly decreased GLAST immunofluorescence compared to the saline treated group in the SN (panel B). ** P < 0.01, *** P < 0.001 compared to saline

DISCUSSION

Dopamine neurons in the ventral midbrain have a central role in the processing of natural and drug rewards, and modulators of dopamine neuron activity contribute to multiple aspects of addiction including relapse in abstinent individuals. Here we found, in mice, that 12 days of treatment with non-contingent cocaine or methamphetamine produces significant alterations in astrocytes in both the SN and VTA. We observed an increase in CRF2 immunofluorescence accompanied by an increase in co-staining with the astrocyte marker S100β and dopaminergic neuron marker TH. We also observed an overall decrease in the apparent number of VTA astrocytes, measured by S100β immunofluorescence as well as a decrease in the expression of the glial glutamate transporter GLAST. The ability of repeated drug exposure to alter glutamatergic and/or CRF signaling in astrocytes and neurons may contribute to cellular and circuit function within the SN/VTA and affect dopaminergic output from the region.

CRF2 in the VTA and SN

We recently published a report validating Cre-dependent NuTRAP mouse lines for cell type-specific investigations into transcriptome and epigenome changes in the brain.¹⁷ Here we extended that work by showing that Aldh1l1;NuTRAP mice can be used to isolate astrocytic mRNA in the SN/VTA, demonstrated by an enrichment of typical astrocytic genes and a depletion of genes from other cell types. A surprising result in our initial screen was that Crhr2 mRNA was present in the positive (astrocytic) fraction following the TRAP preparation. Previous results with single cell RT-PCR had identified Crhr2 expression in midbrain dopamine neurons^10,11, but these receptors had not previously been observed in midbrain astrocytes. Indeed, to our knowledge the only other observation of CRF2 expression in astrocytes has been in Bergman glia of the cerebellum.²² We therefore sought to confirm this unexpected finding in the midbrain with a second technique. Our findings with RNAscope confirmed the localization of Crhr2 mRNA to astrocytes, and additionally indicated alterations following methamphetamine treatment consistent with an increase in astrocyte expression of the transcript. We then pursued evidence at the protein level with immunofluorescence, and expanded our investigation to include the psychostimulant cocaine in addition to methamphetamine. Although only low levels of CRF2 staining were evident in the midbrains of drug-naïve mice, cocaine or methamphetamine treatment produced a dramatic increase in CRF2 signal as well as a substantial increase in co-localization with both TH and S100β positive cells. While CRF2 function in the VTA has previously been linked to cocaine experience, our knowledge of the precise cellular location of these receptors has been incomplete. Our findings here suggest an upregulation of CRF2 in both dopamine neurons and astrocytes occurs in the midbrain following psychostimulant treatment that is not unique to either cocaine or methamphetamine. Future work will be required to determine the functional consequences of dynamically increased CRF2 expression in these individual cell types.

Stress-induced reinstatement following abstinence is a feature of abuse for multiple classes of drugs. While the role of CRF in the VTA has been explored relative to depressants such as morphine and ethanol^23–25, it has also been studied for its role in stress-induced cocaine seeking.²⁶ The most important source of CRF in the VTA for cocaine reinstatement is likely from neurons in the bed nucleus of the stria terminalis²⁷, although other inputs such as the lateral hypothalamic area and paraventricular hypothalamic nucleus could also play a role.^28,29 In the VTA, anatomical evidence suggests that CRF peptide is most prominently expressed by glutamate terminals, but it is also seen in GABA terminals.^28,30 Of the two main CRF receptor subtypes found in the VTA, behavioral and electrophysiological evidence supports a central role for the CRF2 subtype in cocaine reinstatement (but see ³¹). The first evidence of CRF2 action in the VTA was from Ungless et al.¹¹ who discovered that CRF enhances NMDA-type glutamate receptor currents in dopamine neurons through a CRF2-dependent mechanism. Behavioral studies subsequently showed that footshock stress causes release of CRF in the VTA, and in cocaine-experienced rats this was linked to reinstatement of cocaine seeking through CRF2 activation and an increase in glutamate release.^7,9 Hahn et al.¹⁴ observed that repeated non-contingent cocaine treatment recruits additional synaptic mechanisms involving CRF1 and enhancement of AMPA-type glutamate receptor currents. Also in that study, non-contingent cocaine exposure increased excitation to midbrain dopaminergic neurons by augmenting CRF enhancement of excitatory glutamate input ¹⁴, suggesting that cocaine’s effects could be pharmacological and not depend on contingency of administration. While the effects reported here were seen after non-contingent methamphetamine and cocaine administration, the doses used (3 mg/kg and 10 mg/kg respectively) are within the range of doses that are self-administered in rodent studies and are typical for inducing locomotor stimulation. For example, we previously showed that wild type mice on a fixed ratio 3 schedule of reinforcement can self-administer up to 4.4±0.7 mg/kg of methamphetamine in 4-hour daily sessions³² and 14.3±1.3 mg/kg cocaine in 2-hour sessions.³³ In both of those studies we also observed robust locomotor stimulation subsequent to drug self-administration, confirming the behavioral relevance of the doses administered. Future work will be needed to determine if psychostimulant self-administration produces similar effects on CRF2 expression in the VTA and SN, and therefore whether they can be attributed to pharmacology alone.

Stress-induced reinstatement of cocaine seeking is complex and involves heterosynaptic plasticity of both GABAergic and glutamatergic synapses in the VTA.¹⁶ Additionally, CRF2-induced enhancement of glutamate receptor signaling can be inhibited with the A1 adenosine receptor antagonist DPCPX through a mechanism involving an increase in GABA tone.¹⁶ The in situ functional evidence is consistent with the notion that CRF2 receptors are expressed in GABA terminals in the VTA¹⁶, which has also been inferred from studies using electron microscopy³⁰ and synaptosomes.²⁸ As mentioned above, single cell RT-PCR studies have suggested that CRF2 receptors may be expressed by dopamine neurons^10,11, although this was not previously observed using in situ hybridization.³⁴ The present findings provide additional evidence to support the presence of CRF2 expression in a population of midbrain dopamine neurons, and add astrocytes to the list of known cellular locations of CRF2 receptors in the midbrain. It is likely that the increase in CRF2 expression observed following repeated psychostimulant exposure may contribute in some manner to drug-related behaviors such as stress-induced reinstatement.

Interactions with other neurotransmitter systems

Here we show that repeated exposure to cocaine or methamphetamine produces a decrease in expression of the glutamate transporter GLAST in the ventral midbrain. GLAST (or EAAT1) is one of two excitatory amino acid transporters (along with GLT-1/EAAT2) that are primarily expressed by astrocytes and play a central role in synaptic transmission by regulating glutamate homeostasis³⁵. Indeed, our collaborative work recently identified a glutamate transporter-dependent mechanism through which astrocytes can affect midbrain dopamine neuron excitability and motivated behavior.² Dopamine neurons receive a large number of glutamatergic inputs from cortical and subcortical areas.³⁶ Glutamate input is required for burst firing of dopamine neurons, which is a key learning signal implicated in error prediction and seeking of both natural and drug rewards.^3,37,38 A drug-induced decrease in GLAST expression might contribute to the elevated extracellular glutamate levels that have previously been observed using VTA microdialysis subsequent to footshock in cocaine-experienced rats.⁷ Additionally, work out of the Kalivas lab has linked decreased astrocyte glutamate uptake in the nucleus accumbens to drug seeking behavior and neuroplasticity.³⁹ As increased dopamine neuron firing is thought to link stressful events to reinstatement of cocaine seeking,²⁹ it is possible that decreased astrocyte glutamate transport in the midbrain could also play a key role in substance abuse and related behaviors. This decline in glutamate uptake could be reinforced by the parallel decrease in astrocyte number that we observed here subsequent to drug exposure. Alternatively, the decreased GLAST expression we observed could be secondary to decreased astrocyte number in the midbrain. It is unclear at this time how repeated administration of psychostimulants leads to changes in CRF2 and GLAST expression. One possibility is through direct interaction with dopaminergic receptors on astrocytes, which have been reported in the midbrain as well as in other brain regions.⁴⁰ This could provide a plausible mechanism through which increased extracellular dopamine induced by cocaine and methamphetamine could signal alterations in gene expression. Future studies will be needed to identify the precise mechanisms responsible for alterations in astrocytic markers, including GLAST.

An additional and intriguing possibility is that the increased expression of CRF2 on astrocytes could directly link stress and relapse through an adenosine-dependent mechanism. Adenosine tone was identified nearly three decades ago as a mediator of heterosynaptic plasticity.⁴¹ In the VTA, cocaine exposure enhances adenosine tone, altering both GABAergic and glutamatergic transmission.^16,42 There is no de novo adenosine synthesis pathway⁴³; rather, it is thought that the main source of extracellular adenosine is through enzymatic conversion of ATP released from astrocytes, which can regulate the extent that synaptic plasticity can occur.^44,45 Indeed, data from our Aldh1l1;NuTRAP preparation indicate a 2.1-fold enrichment of the enzyme adenosine kinase in the positive versus the input fraction (data not shown), suggesting that astrocytes are central to adenosine metabolism in the VTA. A role for adenosine is consistent with work showing that heterosynaptic plasticity following reinstatement of cocaine self-administration is dependent on both A1 adenosine receptors and CRF2 activation.¹⁶ It is therefore possible that the blockade of cocaine-induced reinstatement observed following intra-VTA infusion of a CRF2 antagonist⁹ is a process that is initiated on astrocytes rather than presynaptic terminals, and it is then adenosine release from astrocytes that induces plasticity at the terminals. Additionally, adenosine tone is known to decrease astrocyte proliferation⁴⁶, which is consistent with our observation of fewer cells staining for S100β following drug treatment. The addition of cell types and molecular players to the known effects of stress and drug exposure could provide new targets to prolong abstinent periods and prevent relapse in individuals with stimulant use disorders.

Similarity across psychostimulants and brain regions

Although both the SN and VTA exhibit a large number of dopaminergic cell bodies, there is considerable heterogeneity both at the cellular level and in circuit connectivity.^36,47–48 While we did observe minor differences in astrocyte alterations between the VTA and SN, the overall effects of methamphetamine and cocaine on both of these regions were very similar. This result is not unexpected since a major mechanism of action of both methamphetamine and cocaine is a decrease in dopamine uptake through the plasma membrane dopamine transporter. Indeed, our previous work indicates that dendritic dopamine neurotransmission in both the SN and VTA are similarly affected by acute exposure to these drugs.⁴⁹ Previous studies have also shown similar increases in extracellular dopamine in response to cocaine and amphetamine administration.⁵⁰ Interestingly, global CRF2 knockout mice show a divergence in acute locomotor response to cocaine versus methamphetamine.⁵¹

Overall, we have shown several alterations in protein expression that occur in astrocytes of the SN and VTA following non-contingent treatment with cocaine or methamphetamine, as well as a decrease in astrocyte number. Therapeutics targeted specifically to membrane proteins on astrocytes in the midbrain could represent a novel strategy for preventing or reversing circuit adaptations that contribute to stimulant use and vulnerability to relapse following abstinence.