|Home | About | Journals | Submit | Contact Us | Français|
We previously found that brain-derived neurotrophic factor (BDNF) haplodeficient mice exhibit greater ethanol-induced place preference and psychomotor sensitization, and greater ethanol consumption after deprivation. We further observed that, in mice, voluntary ethanol intake increases BDNF expression in the dorsal striatum (DS). Here, we determined whether BDNF within the DS regulates ethanol self-administration in Long Evans rats trained to self-administer a 10% ethanol solution. We observed a greater increase in BDNF expression after ethanol self-administration in the dorsolateral striatum (DLS) than in the dorsomedial striatum (DMS). We further found that downregulation of endogenous BDNF using viral-mediated siRNA in the DLS, but not in the DMS, significantly increased ethanol self-administration. Infusion of exogenous BDNF (0.25 μg/μl/side into the DMS; 0.25 and 0.75 μg/μl/side into the DLS) attenuated responding for ethanol when infused 3 hrs prior to the beginning of the self-administration session. Although the decrease in ethanol intake was similar in the DLS and DMS, BDNF infused in the DLS but not in the DMS induced an early termination of the drinking episode. Furthermore, the action of BDNF in the DLS was specific for ethanol, as infusion of the neurotrophic factor in the DMS but not DLS resulted in a reduction of sucrose intake. Together, these findings demonstrate that the BDNF pathway within the DLS controls the level of ethanol self-administration. Importantly, our results suggest that an endogenous signaling pathway within the same brain region that mediates drug-taking behavior also plays a critical role in gating the level of ethanol intake.
BDNF belongs to the nerve growth factor (NGF) family of neurotrophic factors (Huang and Reichardt, 2001; Chao, 2003). BDNF and its receptor TrkB are widely distributed throughout the brain (Wetmore et al., 1990; Altar et al., 1994), and the BDNF/TrkB pathway plays an important role in neuronal proliferation, differentiation and survival, as well as synaptic plasticity (Chao, 2003). More recently, BDNF has been implicated in psychiatric disorders such as depression and anxiety (Martinowich et al., 2007). In addition, a growing body of literature suggests a role for BDNF in drug addiction (Russo et al., 2009a).
Interestingly, human studies have linked BDNF to alcohol addiction. For example, the region of chromosome 11 containing the BDNF gene has been implicated as a susceptibility locus for addiction to multiple drugs of abuse, including alcohol (Uhl et al., 2001), and a single nucleotide polymorphism in the BDNF gene has been linked with an earlier onset of alcoholism (Matsushita et al., 2004). We and others, generated evidence that suggests a role for BDNF in regulating behavioral responses to alcohol (ethanol) in rodents. Specifically, a reduction in BDNF expression in BDNF heterozygous mice (Hensler et al., 2003; McGough et al., 2004) or inhibition of the BDNF receptor TrkB (Jeanblanc et al., 2006) increases ethanol consumption and preference. Moreover, we observed that both acute systemic administration of ethanol and voluntary ethanol intake increase BDNF expression in the DS of mice (McGough et al., 2004; Logrip et al., 2009). We further showed that this increase in BDNF level triggers the expression of downstream effectors, including the dopamine D3 receptor (D3R) and preprodynorphin (Jeanblanc et al., 2006; Logrip et al., 2008), and that inhibition of the D3R (Jeanblanc et al., 2006) or of the dynorphin receptor, the κ opioid receptor (Logrip et al., 2008), blocks the BDNF-mediated decrease in ethanol consumption. Taken together, these studies suggest that BDNF may act as an endogenous negative regulator of ethanol intake. However, the localization of this regulatory effect remains unknown.
As mentioned above, we found that ethanol treatment increases BDNF expression specifically in the DS (McGough et al., 2004). The DS has been implicated in the control of goal-directed behaviors and in the formation of habit (White, 1996; Yin and Knowlton, 2006). Specifically, the DMS has been shown to play a role in response-outcome learning (Yin et al., 2005), whereas the DLS is suggested to regulate stimulus-response, or habit learning (White and McDonald, 2002; Featherstone and McDonald, 2004; Yin et al., 2004). In addition, the lateral and medial parts of the DS have distinct anatomical inputs and outputs (Voorn et al., 2004). We were therefore interested in determining whether and where BDNF within the subregions of DS controls the level of ethanol self-administration.
Male Long-Evans rats (400–450 g at the time of surgery) were obtained from Harlan (Indianapolis, IN). Animals used in the studies were individually housed under a light:dark cycle of 12 hrs, with lights on at 7:00 a.m. and food and water available ad libitum. All animal procedures in this report were approved by the Gallo Center Institutional Animal Care and Use Committee and were conducted in agreement with the Guide for the Care and Use of Laboratory Animals, National Research Council, 1996. Three to 7 animals per group were used for the Reverse Transcription - Polymerase Chain Reaction (RT-PCR) and western blot analysis, and 5 to 12 animals per group for the behavioral experiments as indicated in the Figure Legends.
Human BDNF was purchased from Sigma-Aldrich, Inc. (Saint Louis, MO). pRNAT-H1.1/Shuttle was purchased from GenScript (Piscataway, NJ). The adenoviral vector Adeno-X and the Adeno-X Virus Purification Kit were purchased from Clontech (Mountain View, CA). Lipofectamine 2000 and TRIzol were purchased from Invitrogen (Carlsbad, CA). 2× PCR master mix and the Reverse Transcription System were purchased from Promega (Madison, WI). Anti-GFP antibodies were purchased from Abcam (Cambridge, MA), anti-NeuN antibodies from Millipore (Temecula, CA). Anti-GFAP, and anti-actin antibodies were from Sigma-Aldrich, Inc. (St. Louis, MO) and anti-BDNF polyclonal antibodies (sc-546) were purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). Secondary antibodies were obtained as follows: donkey anti-rabbit Alexa-Fluor 488 and Alexa Fluor-594 labeled donkey anti-mouse from Invitrogen (Carlsbad, CA) and donkey anti-mouse Cy3 conjugate from Jackson ImmunoResearch (West Grove, PA).
Two 19-nucleotide (19nt) BDNF siRNA sequences (Baker-Herman et al., 2004), GCTGAGCGTGTGTGACAGT and GAGCTGCTGGATGAGGACC targeting to the coding domain of the BDNF mRNA were used for vector-based small-hairpin RNA (shRNA) expression. Two complementary oligonucleotides were synthesized as follows: 5′-GATCCC (19nt, sense) TTGATATCCG (19nt, antisense), and TTTTTT CCAAA-3′ and 3′-GG (19nt antisense) AACTATAGGC (19nt, sense) AAAAAA GGTTTTCGA-5′, flanked by Bam H1 and Hind III residues. The paired oligonucleotides were annealed and ligated into Bam H1/Hind III sites of pRNAT-H1.1/Shuttle, a GFP-containing adenoviral shuttle siRNA vector. The pRNAT-siRNA recombinants sequences were confirmed before subcloning into the cloning sites of I-Ceu I and PI-Sce I of the adenoviral vector Adeno-X. A non-related 19-nt sequence ATGAACGTGAATTGCTCAA (Ptasznik et al., 2004), was cloned into Adeno-X vector as described above. Preparation of adenoviruses was initiated by transfection of recombinant adenoviral constructs into HEK293 cells using Lipofectamine 2000 according to the Adeno-X Expression System 1 User Manual. Viruses were amplified in HEK293 cells, followed by purification using Adeno-X Virus Purification Kit. Viruses were titered based on GFP-visualized infection. Viruses containing BDNF siRNA or the control sequence were used to infect SH-SY5Y neuroblastoma cells at a concentration of multiplicity of infection (MOI) of 20 (1010 TU/ml).
SH-SY5Y human neuroblastoma cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) plus 1× MEM non-essential amino acid solution (NEAA, Invitrogen). Cells were incubated in DMEM containing 1% FBS and 1× NEAA and treated with 10 μM retinoic acid for differenciation. Cells were then infected with adenoviruses expressing siBDNF-1, siBDNF-2 or siCT for 3 d before analysis of BDNF mRNA level.
Brains were collected, and slices were cut between +1.70 mm and +0.5 mm anterior to Bregma. The cortex and the ventral striatum were removed (−6.00 mm below the brain surface). The DS was then divided into the DLS and the DMS, at +2.5 mm from the middle line.
Total RNAs were isolated using Trizol reagent and reverse transcribed using a Reverse Transcription System kit (Promega Corporation) at 42°C for 30 min. BDNF, NGF and NT3 expression were analyzed by PCR with temperature cycling parameters consisting of initial denaturation at 94°C for 2 min followed by 32 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, extension at 72°C for 1 min, and a final incubation at 72°C for 7 min. PCR for the control genes, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and actin, was run with the same temperature cycling parameters for 30 cycles. The following primers were used: rat BDNF, upstream, 5′-TTG AGC ACG TGA TCG AAG AGC-3′, and downstream 5′-GTT CGG CAT TGC GAG TTC CAG-3′; rat NGF, upstream 5′-ACA CTC TGG ATC TAG ACT TCC AGG-3′ and downstream 5′-AGG CAA GTC AGC CTC TTC TTG TAG-3′; rat NT3, upstream 5′-CTA CTA CGG CAA CAG AGA CGC TAC-3′ and downstream 5′-TGT CAA TGG CTG AGG ACT TGT CGG-3′; rat GAPDH, upstream 5′-TGA AGG TCG GTG TGA ACG GAT TTG GC-3′ and downstream 5′-CAT GTA GGC CAT GAG GTC CAC CAC-3′; human BDNF, upstream 5′-CTT TGG TTG CAT GAA GGC TGC-3′ and downstream 5′-G TCT ATC CTT ATG AAT CGC CAG-3′; human actin, upstream 5′-TCA TGA AGT GTG ACG TTG ACA TC-3′ and downstream 5′-AGA AGC ATT TGC GGT GGA CGA TG-3′. PCR products were separated on 1.8% agarose gel in Tris/acetic acid/EDTA buffer with 0.25 μg/ml ethidium bromide and photographed by Eagle Eye II (Stratagen, La Jolla, CA). The images were scanned and the signals of the PCR products were quantified by densitometry using the NIH Image 1.61 program. The intensities of signal of BDNF, NGF and NT3 were normalized to GAPDH or Actin as indicated in the figure legends.
The DLS and DMS were dissected and homogenized in radio-immunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, 2 mM EDTA, protease inhibitors cocktail) at 4°C using the homogenizer TH115 PowerGen 125 (Fisher Scientific, Inc.) for 10 seconds. The homogenates (40 μg protein each) were resolved on a 16% sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) gel at constant voltage of 180 volts and transferred to a nitrocellulose membrane (Millipore), and blocked in a milk solution (5% milk in TBS-T: 20 mM Tris-HCl, pH 7.6, 137 mM NaCl, and 0.1% Tween 20). Membranes were incubated with polyclonal anti-BDNF antibodies diluted (1:1000) in a milk solution overnight at 4°C. Membranes were then washed 3 times with TBS-T, followed by incubation with anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibodies diluted (1:1000) in a milk solution for 1 hr at room temperature and washes were repeated 3 times. Immunoreactivity was detected by using enhanced chemiluminescence (ECL) detection kit (GE Healthcare, Buckinghamshire, UK) and processed by exposure to Kodak BioMax film. The film was developed using SRX-101A Medical Film Processor (Konica Minolta Medical & Graphic, Inc.). The membrane was stripped in a stripping buffer (100 mM 2-mercapto-ethanol, 62.5 mM Tris-HCl, 2% SDS, pH 6.7) at 50°C for 30 min, washed 3 times in TBS-T and reprobed with anti-actin antibodies, which was used as an internal control. The images were scanned and the immunoreactivity signal of proteins were quantified by densitometry using the NIH Image 1.61 program. The intensity of the immunoreactivity of BDNF was normalized to actin.
Rats were deeply anesthetized with Euthasol® (Virbac, Forth Worth, TX), and intracardially perfused with 0.9% NaCl for 2 min, followed by cold 4% paraformaldehyde (PFA) in phosphate buffer (PB) pH 7.4 for 10 min. Brains were removed, fixed in 4% PFA for 2 hr at 4°C, and transferred to PB at 4°C. The following day, brains were transferred into 30% sucrose in PBS until fully sunk. Coronal sections (50 μm) were cut using a cryostat Leica CM3050 (Leica Instruments, Nussloch, Germany); collected in PB and stored at 4°C. Free-floating sections were first permeabilized with 50% ethanol for 20 min, rinsed in PBS, then blocked with 10% normal donkey serum in PBS for 30 min, and incubated for 48 hrs at 4°C on an orbital shaker with antibodies for either neuronal marker (monoclonal anti-NeuN, 1:100) or glial marker (monoclonal anti-GFAP; 1:1,000) in combination with the polyclonal rabbit anti-GFP antibody (ab290, 1:10,000), diluted in PBS/0.05% Triton X-100. Sections were then washed 3 times for 5 min with PBS, and then incubated with 2% normal donkey serum for 10 min and incubated in a mixture of secondary antibodies: Alexa Fluor-594 labeled donkey anti-mouse or donkey anti-mouse Cy3 conjugate (to detect either NeuN or GFAP) and Alexa Fluor-488 labeled donkey anti-rabbit (to detect GFP) for 3 hr. After staining, sections were washed in PBS, mounted on gelatin-subbed slides, briefly air dried, and coverslipped using Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Images were acquired using LSM 510 META laser confocal microscope with multichannel excitation and detection options using optimal factory recommended filter configurations to minimize spectral bleed-through (Zeiss, Thornwood, NY). The colocalization of markers and spread of infection were analyzed using Metamorph 6.3 software (Molecular Devices, Sunnyvale, CA) and SigmaScan Pro 5.0 (Systat Software Inc., San Jose, CA).
Following 3 weeks of exposure to ethanol in the home cage, rats were trained to self-administer a solution of 10 % ethanol (v/v). The self-administration chambers contain two levers: an active lever (the ethanol lever) for which presses resulted in delivery of a 0.1 ml fluid reward (a 10 % ethanol solution), and an inactive lever, for which presses were counted but no programmed events occurred. After 3 d under a Fixed Ratio 1 (FR1, one press delivers one reward) schedule, the rats were trained on an FR3 schedule (3 presses are required to receive one reward) for 60-min sessions either 5 days a week (BDNF infusion experiments) or 7 days a week (BDNF mRNA knockdown experiments). Animals were trained for at least 5 weeks before the beginning of the experimental manipulations; this length of training results in ethanol intakes in a range of 0.3 to 0.7 g/kg in 1 hr leading to a blood ethanol concentration from 5 to 25 mg% (Suppl. Fig. 1). During the self-administration sessions, several parameters (number of presses on the levers, number of ethanol deliveries, latency to the first press, latency to the first reward, delay between presses, latency to the last press) were recorded using MED-PC IV software (Med Associates Inc.).
Rats trained to self-administer ethanol for at least 5 weeks as described above were placed into the operant conditioning chambers for a 30-min session of ethanol self-administration. Thirty min after the end of the session (i.e. 1 hr after the beginning of the ethanol self-administration session), rats were euthanized, and the DLS and DMS were dissected out and processed for RT-PCR as described above.
Tail blood was collected in heparinized capillary tubes from rats immediately at the end of a 60-min ethanol self-administration session. Serum was extracted with 3.4% trichloroacetic acid followed by a 5-min centrifugation at 420 g, and assayed for ethanol content using the Nicotinamide Adenine Dinucleotide (NAD-NADH) enzyme spectrophotometric method (Zapata et al., 2006). BEC was determined by using a standard calibration curve.
Rats were initially trained in 2 overnight sessions under an FR1 schedule using 0.1 ml of an 8% sucrose solution as the reinforcer. Subsequently, rats were trained 5 days a week in 60-min sessions, with the FR schedule progressively increased to FR3 and the sucrose concentration progressively decreased to 2%.
Stereotaxic injections of the viruses were conducted once the rats had reached a stable baseline of ethanol self-administration for at least 7 consecutive days. Rats were continuously anesthetized with Isoflurane (Baxter, IL, USA) during the surgery. Two holes were drilled above the sites of injection to allow the introduction of the injectors (Acute Internal Cannula, C315IA, diameter: 33G, PlasticsOne, VA, USA). The injectors were connected to Hamilton syringes (25 μl #1702) and infusion was controlled by an automatic pump (Harvard Apparatus, Holliston, MA). As the size of the targeted structures is not uniform, rats received 3 injections of 1 μl of solution per side in the DLS and 2 injections of 1 μl per side for the DMS at a rate of 0.5 μl/min and the injectors remained in place for 10 additional min. The coordinates used for the injections into the DLS were as follows: Injection #1: +1.2 mm anterior to Bregma, 3.5 mm lateral the medial suture and −4.5 mm ventral to the skull surface; Injection #2: +0.2 mm anterior to Bregma, 3.5 mm lateral to the medial suture and −5.5 mm ventral to the skull surface; Injection #3: +0.2 mm anterior to Bregma, 3.5 mm lateral to the medial suture and −4.5 mm ventral to the skull surface. The coordinates used for the injections into the DMS were as follows: Injection #1: +1.2 mm anterior to Bregma, 1.5 mm lateral to the medial suture and −4.5 mm ventral to the skull surface; Injection #2: +0.2 mm anterior to Bregma, 1.5 mm lateral to the medial suture and −4.5 mm ventral to the skull surface. After one day of recovery the subjects were returned to self-administration training until the end of the experiment.
Rats were continuously anaesthetized with Isoflurane (Baxter, IL, USA) during the surgery. Four holes were drilled for screws and 2 other holes were drilled for the placement of the cannulae (DLS: single cannula C315GA; DMS: double cannulae C235G, 3 mm between the cannulae; shell of the NAc: double cannulae C235G, 2 mm between the cannulae; 26G diameter, PlasticsOne, Roanoke, VA). The coordinates for the DLS were +1.2 mm anterior to Bregma and 3.5 mm lateral to the medial suture. The cannulae were implanted into the lateral part of the dorsal striatum (−4.2 mm from the skull surface) and fixed with dental cement. The coordinates for the DMS were +1.0 mm anterior to Bregma, 1.5 mm lateral to the medial suture and −4.2 mm from the skull surface. The coordinates for the shell of the NAc were +1.4 mm anterior to Bregma, 1 mm lateral to the medial suture and −6.8 mm from the skull surface. Subject weights were monitored daily after the surgery to ensure a healthy recovery of each rat. One week after recovery, subjects returned to self-administration training and were habituated to the microinjection procedure with two sham injections. The experimental microinjections began upon acquisition of a stable level of responding. The injectors used for each groups extended 0.5 mm below the tip of the cannula.
BDNF doses were chosen according to Lu et al. which showed that 0.25 or 0.75 μg/μl of BDNF microinfused into the VTA are sufficient to alter cocaine seeking (Lu et al., 2004). BDNF or PBS was infused via a 25 μl Hamilton syringe (#1702) 3 hrs prior to the beginning of the self-administration session. BDNF (0.75 μg/μl) was also infused into the DLS 10 min prior to the beginning of the session. For the DMS and DLS injections, the volume infused was 1 μl and the infusion lasted 2 min; due to the small size of the shell of the NAc, and to limit the possible diffusion of BDNF to the DMS, a volume of 0.5 μl was microinfused into the shell of the NAc over 2.5 min. The injectors remained in position for an additional 2 min. Rats were returned to the home cage until the beginning of the self-administration session. The order of injections was counterbalanced across all subjects for each brain region.
Rats implanted with cannulae were perfused transcardially with the fixative (4% PFA), then 75 μm coronal slices were cut and examined for cannulae placements. The placement of cannulae are shown in Suppl. Fig. 2. Animals in which the cannulae placements were not in the appropriate brain area were removed from the study. Specifically, 2 rats were removed from the experiment in which BDNF was infused into the DLS and 1 rat was excluded from the group in which BDNF was infused into the shell of the NAc.
Biochemical and behavioral data were analyzed by one- or two-way ANOVA with repeated measures, depending on the experiment, followed by the Student-Newman-Keuls test when indicated by significant effects of treatments or interactions. For simple comparisons, data were analyzed by a Student’s t-test. Significance for all tests was set at p < 0.05.
We previously showed that both ethanol consumption and systemic injection of 2 g/kg ethanol leads to an increase in BDNF expression in the DS in mice (McGough et al., 2004). Therefore, we tested whether an increase in BDNF mRNA expression can be detected within the DLS and/or the DMS of rats after ethanol self-administration. The average number of ethanol deliveries for this group of rats during a 30-min session was 23.5 ± 6.65, which corresponds to a mean estimated g/kg of 0.34 ± 0.097. Interestingly, we found that ethanol self-administration induced a dramatic increase in BDNF expression within the DLS but only a small increase within the DMS (Fig. 1).
To study the possible role of endogenous BDNF in the DLS in the regulation of ethanol self-administration, we utilized the adenovirus-mediated delivery of siRNA (Davidson and Breakefield, 2003) to knockdown the level of the neurotrophic factor. Two distinct BDNF siRNA sequences (Baker-Herman et al., 2004), were cloned into an adenoviral shuttle vector containing green fluorescent protein (GFP), that was recombined into an adenoviral vector. BDNF siRNAs recombinant adenoviruses (siBDNF-1 and siBDNF-2) significantly reduced the expression level of BDNF in SH-SY5Y neuroblastoma cells as compared with uninfected cells and cells infected with a control non-specific RNA sequence-recombinant adenovirus (siCT) (Suppl Fig. 3). Next, we injected siBDNF-1 into the DLS of rats and monitored the level of viral infection. As shown in Fig. 2a, a high level of virus infection was detected in DLS neurons as measured by co-staining of GFP with the neuronal marker NeuN, while a lower level of infection was detected in glia as measured by co-staining of GFP with the glial marker GFAP. The infection spread was within a radius of ~500 μm from the infusion needle track (Suppl. Fig. 4a), indicating that it is unlikely that the infection extended laterally from the DLS to the DMS. Next, we measured the level of BDNF mRNA in the DLS after the administration of siBDNF-1, and observed a decrease in BDNF expression 5 d post-infusion of siBDNF-1, which was still observed 15 d but not 25 d later (Fig. 2b). A two-way ANOVA revealed a significant effect of Treatment (F(1, 35) = 5.2, p < 0.05) and an interaction between Treatment and Time points (F(2, 35) = 4.8, p < 0.05). Furthermore, the down-regulation of BDNF mRNA levels was specific, since infection of the DLS did not result in a decrease in the mRNA levels of the related neurotrophic genes, NGF (Suppl. Fig. 4b) and neurotrophin-3 (NT3) (Suppl. Fig. 4c). We also tested the level of the BDNF protein in the DLS after infection with siBDNF-1 and found a reduction 15 d after virus infusion (Fig. 2c). A Student’s t-test revealed a significant difference between the siBDNF-1 group and the siCT group for the time point of 15d post-infusion (p < 0.001). Although the anti-BDNF antibodies recognize both the mature form of BDNF and pro-BDNF, we were unable to detect pro-BDNF in our experiments, which is likely to be due to the fast-processing of pro-BDNF to the mature form of the protein. The reduction in the expression of the protein was localized to the site of infusion (i.e., the DLS), as the level of BDNF was unaltered in the DMS (Suppl. Fig. 4d). We also analyzed the levels of the internal controls used for the mRNA (i.e. GAPDH) and for the protein (i.e. actin) to ensure that the levels of these controls were not altered after infection with siBDNF-1. As shown in Suppl. Fig. 4e and 4f, the densities of the bands of GAPDH and actin are not modified by siBDNF-1 infection as compared to siCT (p > 0.05). Together, these results show that adenovirus-mediated BDNF siRNAs decrease BDNF expression, which corresponds with a decrease in BDNF protein level.
Operant self-administration of ethanol was used next to test the behavioral consequences of BDNF knockdown in the DLS. Rats were trained on a fixed-ratio 3 schedule to lever-press for delivery of 0.1 ml aliquots of a 10% ethanol solution in daily 1-hr sessions. Lever-press responding and estimated ethanol consumption in g/kg body weight were measured before, during and after infusion of siBDNF-1 or siCT. Because we found significant decreases in BDNF mRNA and protein 15 d after viral infusion, we tested whether ethanol self-administration would be altered at this time point. We found that BDNF knockdown in the DLS significantly increased ethanol self-administration 15–16 d post-viral infusion (p < 0.05; Fig. 2d). When ethanol self-administration was examined across the 29 sessions after viral infusion, we observed a progressive increase in the number of ethanol deliveries beginning 8–9 d after infusion of the siBDNF-1 virus, and reaching a maximum at 16–17 d post-infusion (Fig. 2e); the number of deliveries then gradually returned to baseline levels 24–25 d post-virus infusion. A Student’s t-test did not reveal any differences in the number of ethanol deliveries between the two groups before the infusion. A two-way repeated-measures ANOVA on the time course of the behavior after viral infusion found a main effect of Time (F(13,221) = 2.5, p < 0.01), no main effect of Virus Treatment (F(1,221) = 0.3, ns), and a significant interaction between Treatment and Time (F(13,221) = 3.6, p < 0.001). The post-hoc analysis revealed a significant difference between siBDNF-1 and siCT groups for the time points 14–15 and 16–17 d post-infusion.
Thus, both BDNF levels and ethanol deliveries returned to baseline 25–26 d after the injection (Fig. 2c, e). These results were further confirmed by the use of the siBDNF-2 which produced a comparable decrease in BDNF mRNA levels (Suppl. Fig. 5a) and a similar increase in ethanol consumption (Suppl. Fig. 5b) 15 d after infusion of the virus.
Analysis of the inter-ethanol delivery interval (IEDIs) distribution (a measure of the time between each successive ethanol reinforcer delivery) for day 15 post-infusion for both treatment groups found no main effect of Treatment (F(1,120) = 2.6; ns), and no interaction between Treatment and Time (F(20,120) = 1.031; ns) (Suppl. Fig. 6). In addition, the mean latency to the first reward did not differ significantly between the two treatment groups (F(1,21) = 0.0012; ns, Suppl. Table 1). Together, these results demonstrate that reduction in BDNF levels within the DLS significantly increases ethanol operant self-administration without changing the initiation of the drinking episode.
Next, we tested the contribution of endogenous BDNF in the DMS to the regulation of ethanol self-administration. Although infusion of the siBDNF-1 significantly decreased BDNF expression (Fig. 3a) and BDNF protein levels (Fig. 3b) within the DMS, we did not observe a change in ethanol self-administration 15–16 d after infusion (Fig. 3c). Moreover, BDNF mRNA knockdown within the DMS did not have any effect on the number of ethanol deliveries (Fig. 3d). No difference was observed between the siBDNF-1 group and the siCT group before the infusion of the virus (Student’s t-test, p > 0.05) and a two-way repeated-measures ANOVA on the behavioral time course found no effect of Treatment (F(1,100) = 1.0, ns), Time (F(10,100) = 0.9, ns), nor an interaction between these factors (F(10,100) = 1.3, ns). These findings suggest that the actions of endogenous BDNF to control the level of ethanol self-administration are localized to the DLS.
If the site of action of BDNF to regulate ethanol consumption is indeed the DLS, then activation of the BDNF pathway within this region should result in the opposite effect, i.e., the reduction of responding for ethanol. To test this possibility, the BDNF polypeptide was infused directly into the DLS, and ethanol operant self-administration was monitored. We found that BDNF administration dose-dependently decreased ethanol operant self-administration when injected 3 hrs, but not 10 min, before the beginning of the session, and no significant effects were observed on inactive lever responding (Fig. 4a). A two-way repeated-measures ANOVA revealed significant main effects of Lever (F(1,28) = 28.9; p < 0.001) and Treatment (F(4,28) = 7.7; p < 0.001), as well as an interaction between both factors (F(4,28) = 7.9; p < 0.001). This decrease in ethanol presses led to a decrease in ethanol deliveries as shown in Suppl. Fig. 7a. As a consequence of the reduction of the number of ethanol deliveries, consumption of ethanol during the self-administration session was significantly decreased by BDNF (Fig. 4b), as indicated by a significant main effect of Treatment (F(3,18) = 9.5, p < 0.001) on mean g/kg.
Although the total number of responses was decreased after infusion of BDNF into the DLS, this treatment did not alter the distribution of IEDIs during the session. After the injection of 0.75 μg/μl BDNF, only 2 rats pressed at least 6 times, the minimum required to receive 2 rewards and thus to measure an IEDI. Hence, the IEDI analysis was conducted on the results for 0.08 and 0.25 μg/μl BDNF and vehicle. We found that the distribution of the IEDIs was unchanged (Suppl. Fig. 7b). Next, we analyzed the latency to the 1st reward for rats meeting the minimum requirement for at least 3 responses (i.e., one reward delivery). As shown in Suppl. Table 2, BDNF did not alter the latency to the 1st reward. On the contrary, BDNF induces a premature termination of responding confirmed by the analysis of the time at which the final lever-press response of the session occurred. As shown in Fig. 4c, the time of the last press occurred significantly earlier after treatment with the 2 highest doses of BDNF (0.25 and 0.75 μg/μl/side) as compared to PBS. A one-way repeated-measures ANOVA on the time of the last press found a main effect of Treatment (F(3,15) = 3.9, p < 0.05). The post-hoc analysis revealed significant differences between the PBS treatment and both the 0.25 and 0.75 μg/μl/side BDNF injections (p < 0.05 for both comparisons).
Finally, to ensure that the attenuation of ethanol self-administration was specific, we tested the effect of heat-inactivated BDNF on ethanol self-administration. As shown in Suppl. Fig. 8, infusion of the denatured protein into the DLS did not alter responding for ethanol. Together, these results strongly suggest that the DLS is the site of BDNF’s action to regulate ethanol self-administration.
Next, we tested whether the alteration of ethanol operant self-administration by BDNF is due to a reduction of the reinforcing effects of rewarding substances in general. BDNF was therefore infused into the DLS, and operant responding for the natural reward sucrose was measured. We found that microinjection of BDNF into this brain region did not alter sucrose self-administration (Fig. 4d), suggesting that BDNF’s actions are specific for ethanol and are not due to a general attenuation of responding for reward. A two-way repeated-measures ANOVA found no effect of Treatment (F(2, 22) = 1.2, ns) and no interaction between Treatment and Lever (F(2,22) = 2.2; ns).
Next, we tested whether the BDNF signaling pathway in the DMS also plays a role in the regulation of ethanol consumption. We found that administration of the neurotrophic factor significantly decreased the number of presses on the ethanol lever (Fig. 5a; main effect Treatment (F(2,18) = 21.01, p < 0.001), main effect of Lever (F(2,18) = 27.9, p < 0.001), Treatment × Lever interaction (F(2,18) = 9.3, p = 0.002)), and the number of ethanol deliveries (Suppl. Fig. 9a), as well as the quantities of consumed ethanol (Fig. 5b; main effect of Treatment (F(2,18) = 16.4, p < 0.001). The post-hoc analyses revealed a significant difference between PBS and 0.25 μg/μl/side of BDNF for both number of responses and for g/kg (p < 0.001 for both comparisons). However, BDNF injected in the DMS did not alter the pattern of drinking since it had no effect on either the latency to the 1st reward (Suppl. Table 3) or the distribution of IEDIs (Suppl. Fig. 9b). Interestingly, the decrease in overall responding for ethanol was not associated with a decrease in the latency to the last press of the session (Fig. 5c; no effect of Treatment (F(2, 19) = 1.5, ns)), suggesting that the rats continued to press for ethanol during the entire session. To test whether the effect of BDNF in the DMS is specific to ethanol, the neurotrophic factor was infused into this brain region of rats trained to self-administer sucrose 3 hrs prior to the beginning of the session. We found that infusion of BDNF into the DMS significantly decreased the number of presses on the sucrose lever (Fig. 5d). A two-way repeated-measures ANOVA revealed main effects of Treatment (F(1, 6) = 11.2, p < 0.05) and of Lever (F(1, 6) = 35.2, p = 0.001), and an interaction between Treatment and Lever (F(1, 6) = 10.7, p < 0.05). The post-hoc analysis revealed a significant difference in presses on the active lever between the two treatments (p < 0.001). These results suggest that the activation of the BDNF pathway in the DMS results in a general reduction in responding for reward.
Finally, we tested the specificity of BDNF’s actions on ethanol self-administration in the DS by infusion of BDNF into the shell of the NAc. As a primary component of the brain’s reward circuitry (Hyman et al., 2006), the NAc (ventral striatum) in general, and the shell of the NAc in particular, play a crucial role in ethanol reinforcement (Gonzales et al., 2004). However, we observed no change in the number of presses on the ethanol lever (Treatment: F(1, 4) = 0.1, ns; Treatment x Lever interaction: F(1, 4) = 0.9, ns) after the infusion of the neurotrophic factor (Fig. 6a), no change in the number of ethanol deliveries (Suppl. Fig. 10), and thus no change in the amount of ethanol consumed (Treatment: F(2,8) = 0.5, ns) (Fig. 6b). These results suggest a specific role for BDNF in the DS and, in particular, in the lateral part of the DS, to control the level of ethanol self-administration.
Here we show that the expression level of BDNF is increased in rats administering ethanol in an operant self-administration paradigm. These results are in line with our previous findings in mice consuming ethanol in a 2-bottle choice paradigm (McGough et al., 2004; Logrip et al., 2009). Therefore, although the basal level of BDNF in the striatum is relatively lower than in other brain regions it can be clearly detected in the DS and is altered in response to ethanol. Interestingly, the ethanol consumption-mediated induction of BDNF expression is much greater in the DLS than in the DMS. These results correspond with the findings that knockdown of BDNF within the DLS, but not in the DMS, resulted in an increase in ethanol self-administration. Therefore, using RNAi, we were not only able to decrease the basal level of the BDNF expression in the DLS, but additionally we blocked the endogenous homeostatic BDNF system that normally is up-regulated in response to ethanol.
The decrease in BDNF mRNA was detected earlier than the protein, at 5 d post-injection of the siBDNF. As BDNF protein levels are much more stable than the RNA levels (Nawa et al., 1995), and since BDNF can be recycled (Santi et al., 2006) and thus reused, the delay between the decrease in BDNF mRNA and BDNF protein is likely due to the time required to process and down-regulate the pool of the polypeptide. BDNF can be produced by glial cells (Riley et al., 2004), and therefore we cannot rule out that
BDNF in astrocytes contributes to the protective actions of the neurotrophic factor. However, this possibility is unlikely since BDNF expression in brain was detected mainly in neurons but not in astrocytes (Ernfors et al., 1990; Hofer et al., 1990; Maisonpierre et al., 1990; Wetmore et al., 1990), and only minor infection by the adenovirus was detected in astrocytes.
Interestingly, infusion of exogenous BDNF into both subregions of the DS resulted in the attenuation of ethanol self-administration. However, the actions of BDNF in the DLS and DMS are clearly different. Application of BDNF in the DLS, but not the DMS, decreases the time of the final lever press response resulting in early termination, whereas this measure did not change after BDNF infusion into the DMS. In addition, after BDNF infusion into the DLS, only 2 rats out of 8 pressed enough to get 2 rewards, whereas 10 out of 11 rats pressed for ethanol after infusion of the neurotrophic factor into the DMS. Thus, the early termination of responding and the low number of rats pressing for ethanol after BDNF infusion into the DLS is consistent with a possible decrease in the reinforcing effects of ethanol rather than a satiety effect. Furthermore, the inhibition of ethanol self-administration by BDNF in the DLS is specific to ethanol, whereas exogenous BDNF in the DMS also affects responding for a natural reward, i.e., sucrose. Therefore, BDNF in the DMS might have a general effect on instrumental responding for reward or on the motivation for reward in general, whereas our results suggest that BDNF in the DLS has a specific role in ethanol reinforcement.
The functional differences described to the two DS subregions in the control of instrumental responding is particularly intriguing: the DLS portion controls responding supported by stimulus-response associations (White and McDonald, 2002; Featherstone and McDonald, 2004; Yin et al., 2004), whereas the DMS controls responding supported by response-outcome associations (Yin et al., 2005). These studies, in conjunction with the demonstration of distinct anatomical inputs and outputs to and from the lateral and medial parts of the DS (Voorn et al., 2004; Belin et al., 2009) have led to the possibility that the DLS mediates habit/stimulus-response learning and memory processes that may underlie addictive phenotypes such as compulsive drug and ethanol administration (Everitt and Robbins, 2005). Interestingly, ethanol self-administration has been described to be a habitual rather than a goal-directed behavior (Dickinson et al., 2002). Thus, our results suggest that the action of endogenous BDNF on ethanol self-administration takes place in the same brain region most strongly implicated in habit learning, i.e., the DLS. Because BDNF appears to control termination rather than initiation of responding for ethanol, we suggest that the reinforcing properties of ethanol that may normally serve to support maintenance of the stimulus-response association underlying lever-pressing for ethanol, are negatively regulated by BDNF.
The regulation of responding for ethanol by BDNF only occurs when BDNF is injected 3 hrs, but not 10 min, prior to the beginning of the session. This delay between the injection and the effect of BDNF suggests that the actions of the neurotrophic factor are mediated by a transcription/translation event. This possibility is further supported by our previous studies suggesting that BDNF-mediated increases in the D3R and dynorphin in the striatum are required for the reduction in ethanol consumption (Jeanblanc et al., 2006; Logrip et al., 2008). It is however, plausible that other proteins could be of equal importance in controlling the level of ethanol consumption via BDNF in the DLS, an area of investigation which we are actively pursuing.
BDNF may dampen various phenotypes associated with exposure to ethanol and stimulants via actions within specific brain regions. Pandey et al. reported that the reduction in BDNF levels in the central and medial nuclei of the amygdala resulted in increased anxiety and higher ethanol consumption in rats (Pandey et al., 2006). Pandey et al. further showed that the anxiogenic effects of ethanol withdrawal were linked to decreases in the expression of BDNF, and that infusion of BDNF in the central nucleus of the amygdala reversed ethanol withdrawal-induced anxiety (Pandey et al., 2008). In addition, Berglind et al. showed that a single microinjection of BDNF into the medial prefrontal cortex immediately following the final cocaine self-administration session attenuated cocaine seeking 22 hrs after the infusion (Berglind et al., 2007). A recent study by the same group found that the single injection of BDNF into the prefrontal cortex normalized cocaine-induced increases in glutamatergic neurotransmission in the NAc (Berglind et al., 2009). However, several lines of evidence indicate that BDNF may have opposite effects on cocaine self-administration and seeking via its actions in the VTA and NAc. Specifically, BDNF protein levels were found to be increased in the NAc, VTA and amygdala during the “incubation of craving” period following cocaine self-administration (Grimm et al., 2003). In line with these results, Lu et al. reported that a single infusion of BDNF into the VTA increased cocaine-seeking (Lu et al., 2004), and Graham et al. showed that multiple injections of BDNF into the NAc shell increased cocaine self-administration, cocaine seeking and reinstatement (Graham et al., 2007). The strikingly different role that BDNF appears to play in mechanisms that underlie or attenuate self-administration of cocaine and ethanol is intriguing, and might reveal very important signaling and/or brain region specificity differences that are valuable for understanding the mechanisms that underlie or protect against addiction to ethanol and cocaine.
Our data suggest a novel role for BDNF in the DLS in gating ethanol self-administration. Specifically, we propose that endogenous BDNF in the DLS controls the level of ethanol self-administration by reducing or holding in check the reinforcing effects of ethanol, possibly thereby preventing the escalation to uncontrolled use and abuse of the drug. Therefore, increasing BDNF expression or activation of this pathway may be a new strategy to combat alcohol abuse and alcoholism.
This work was supported by NIH-NIAAA R01 AA016848 (D.R.), the State of California for Medical Research on Alcohol and Substance Abuse through the University of California, San Francisco (D.R. and P.H.J.), and the Wheeler Center for the Neurobiology Addiction (J.J.). The authors thank Dr. Marian Logrip for her technical and editorial contributions.