If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Department of Metabolism and Aging, The Scripps Research Institute, Jupiter, FloridaDepartment of Neuroscience, The Scripps Research Institute, Jupiter, Florida
Memories associated with drugs of abuse, such as methamphetamine (METH), increase relapse vulnerability to substance use disorder by triggering craving. The nucleus accumbens (NAc) is essential to these drug-associated memories, but underlying mechanisms are poorly understood. Posttranslational chromatin modifications, such as histone methylation, modulate gene transcription; thus, we investigated the role of the associated epigenetic modifiers in METH-associated memory.
Methods
Conditioned place preference was used to assess the epigenetic landscape in the NAc supporting METH-associated memory (n = 79). The impact of histone methylation (H3K4me2/3) on the formation and expression of METH-associated memory was determined by focal, intra-NAc knockdown (KD) of a writer, the methyltransferase mixed-lineage leukemia 1 (Mll1) (n = 26), and an eraser, the histone lysine (K)-specific demethylase 5C (Kdm5c) (n = 38), of H3K4me2/3.
Results
A survey of chromatin modifications in the NAc of animals forming a METH-associated memory revealed the global induction of several modifications associated with active transcription. This correlated with a pattern of gene activation, as revealed by microarray analysis, including upregulation of oxytocin receptor (Oxtr) and FBJ osteosarcoma oncogene (Fos), the promoters of which also had increased H3K4me3. KD of Mll1 reduced H3K4me3, Fos and Oxtr levels and disrupted METH-associated memory. KD of Kdm5c resulted in hypermethylation of H3K4 and prevented the expression of METH-associated memory.
Conclusions
The development and expression of METH-associated memory are supported by regulation of H3K4me2/3 levels by MLL1 and KDM5C, respectively, in the NAc. These data indicate that permissive histone methylation, and the associated epigenetic writers and erasers, represent potential targets for the treatment of substance abuse relapse, a psychiatric condition perpetuated by unwanted associative memories.
Learned associations between the rewarding effects of a drug, such as methamphetamine (METH), and contextual stimuli present at the time of drug use are a central component of acquiring and maintaining drug-seeking behavior and are capable of inducing relapse years after the initiation of abstinence, perpetuating substance use disorder (
). Understanding the mechanisms that underlie these unwanted memories should provide insight into effective therapies aimed at reducing context-induced relapse.
By receiving inputs from brain regions implicated in reward and memory, including the ventral tegmental area, amygdala, hippocampus (HPC) and medial prefrontal cortex, the nucleus accumbens (NAc) is uniquely positioned to mediate reward-based behaviors (
), including drug seeking triggered by associative memories. Drug-induced transcriptional activation in the NAc is thought to be key to the behavioral and neurophysiological changes that occur with drug intake (
). Furthermore, the NAc has been implicated in many forms of learning, including contextual fear, spatial memory, instrumental learning and Pavlovian reward memory (
) and that the NAc is a hub for reward, memory, and drug seeking, mechanisms contributing to memory-induced transcriptional changes in the NAc may provide insight into approaches aimed at disrupting METH-associated memories.
Epigenetic modifications modulate transcriptional activity without altering the DNA sequence (
). Histone acetylation is associated with transcriptional activation. However, methylation has been implicated in both repression and activation, depending on the specific lysine residue that is modified and the number of methyl moieties that are attached (
G9a/GLP histone lysine dimethyltransferase complex activity in the hippocampus and the entorhinal cortex is required for gene activation and silencing during memory consolidation.
), suggesting a potential for therapeutic disruption of drug-associated memories by targeting chromatin-modifying enzymes. Mixed-lineage leukemia 1 (MLL1) has been identified as an HMT for the permissive methylation that occurs at lysine 4 (H3K4me2/me3) and is required for adult neurogenesis, synaptic plasticity, and HPC-dependent memory formation. It is also involved in prefrontal gamma-aminobutyric acid–ergic dysfunction associated with schizophrenia and the mechanisms of cortical spreading depression (
). De novo mutations in Mll have also recently been associated with Wiedemann-Steiner syndrome, a disorder marked by hypertrichosis cubiti and intellectual disability (
). The KDM5 family’s function in the intact brain, on the other hand, has not yet been studied. However, Kdm5c/Smcx mutations, which decrease demethylase activity, have been found in male patients diagnosed with X-linked intellectual disability and short-term memory deficits have been reported in carrier females (
To investigate the role of H3K4 methylation modifiers in METH-associated contextual memory, we used conditioned place preference (CPP), in which Pavlovian associations are formed between a specific context (conditioned stimulus, CS+) and the rewarding effects of METH (
). As activation of gene expression occurs with exposure to drugs of abuse and during memory storage, we aimed to test the hypothesis that modifiers of histone modifications supportive of active transcription (e.g., H3K4me2/me3) are involved in METH-associated memory.
Methods and Materials
Animals
Eight- to 12-week-old male C57Bl/J6 mice (Jackson Laboratories, Bar Harbor, Maine) were housed in groups of four on a 12-hour light–dark cycle with ad libitum access to food and water. All experiments were performed during the light part of the diurnal cycle. Housing, animal care, and experimental procedures were consistent with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee.
Small Interfering RNA–Mediated Focal Knockdown of Mll1 and Kdm5c
KD of Mll1 or Kdm5c was first assayed in naive mice 2 and 5 days after intra-NAc injection of each small interfering RNA (siRNA) (ON-TARGETplus SMART pool, mouse Mll1, Kdm5c, or nontargeting pool; Dharmacon, Waltham, Maryland; see Table S1 in Supplement 1 for target sequences) using JetSI transfection reagent (Polyplus Transfection, Illkrich, France) (
). Mice were anesthetized with 1.0% to 1.5% isoflurane and placed into a stereotaxic apparatus (Kopf, Tujunga, California). The appropriate siRNA-JetSI complexes were injected bilaterally into the NAc (1 µL at 200 nL/min; anterior-posterior: +1.6 mm, medial-lateral: ±2.6 mm from bregma and dorsal-ventral: –4.8 mm from skull at a 20° angle) (
) using an automated injection pump through a 10-µL syringe and 33-G stainless steel needle. The syringe was left in place for 5 minutes to allow for diffusion of the siRNA-JetSI complexes. Coordinates were first validated with 500 nL of a green fluorescent protein (GFP)-expressing adeno-associated virus (GFP-AAV). Coronal sections collected 1 week later showed GFP expression spread across 1 mm of the anterior-posterior axis of the NAc, affecting both the core and medial shell subregions; expression was not detected in the dorsal striatum. In a separate group of animals used for behavioral assessments, bilateral stainless-steel guide cannulae (26 G; Plastics One, Roanoke, Virginia) were placed 1-mm above the NAc using the same coordinates. Metacam (1.5 mg/mL) was administered immediately after surgery. Five to 7 days later, training began and mice were injected with Mll1, Kdm5c or control siRNA using infusion needles (Plastics One) that projected 1 mm beyond the tip of the guide cannula.
Selective, retrieval-independent disruption of methamphetamine-associated memory by actin depolymerization [published online ahead of print September 5].
). For additional details, please see Supplement 1.
RNA and Protein Extraction
For the microarray, RNA was extracted from NAc tissue punches using the AllPrep DNA/RNA Micro Kit (Qiagen, Hilden, Germany). For all other experiments, NAc protein and RNA were extracted using the AMBION PARIS kit (Life Technologies, Grand Island, New York).
Gene Expression Analysis Microarray
From each CPP sample (7–8 samples per group), 100 ng of total RNA was pooled, such that RNA from each animal in a given group was equally represented in the appropriate pooled sample, resulting in three samples of 1 µg each for the array: saline CPP, METH home cage (HC), and METH CPP. Transcriptome analysis was performed with the GeneChip Mouse Gene 1.0 ST Array (Affymetrix, Santa Clara, California) by the Scripps Florida Genomics Core. Briefly, in vitro transcription was performed to produce labeled complementary RNA, which was then fragmented (~100 bp), added to the chip, hybridized overnight, washed, stained, and scanned. Data were normalized using the MAS5 algorithm; signal intensities lower than 100 arbitrary units were considered not expressed. Data for the METH CPP group were expressed as fold change relative to saline CPP and METH HC control groups; hit genes were identified by fold changes ≥1.6, compared with both control groups.
Validation by quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) was performed using the individual samples obtained before pooling a portion for the microarray. The iScript One-Step qRT-PCR Kit (BioRad, Hercules, California) was used with mouse TaqMan expression assays (Life Technologies; see Table S1 in Supplement 1) and 20 ng of RNA. All samples were run in triplicate. Data were analyzed using the ΔΔCt method.
Histone Extraction and Western Blotting
Histones were extracted using the EpiQuik Total Histone Extraction kit (Epigentek, Farmingdale, New York). Two micrograms of extracted histones or 40 µg of total protein were resolved on 15% or 7% sodium dodecyl sulfate polyacrylamide gel electrophoresis gels, respectively, and transferred onto polyvinylidene fluoride membranes, then blocked for 1 hour at room temperature with 5% bovine serum ablumin (for histones) or 5% nonfat milk (for total protein) in Tris buffered saline with Tween20 (50 mmol/mL Tris, 150 mmol/mL NaCl, 0.1% Tween20). See Table S2 in Supplement 1 for antibodies and concentrations. Membranes were washed with Tris buffered saline with Tween20 and incubated with horseradish peroxidase–coupled secondary antibodies (anti-rabbit immunoglobulin G (IgG), 1:1000; anti-mouse IgG, 1:1000; Vector Laboratories, Burlingame, California). Protein expression was assessed by chemoluminescence (Pierce ECL, Thermo Scientific, Waltham, Maryland) and exposure to autoradiography film (Denville Scientific, South Plainfield, New Jersey). ImageJ software was used to quantify band intensities. Relative protein expression was calculated by normalizing the integrated band density values to total H4 or GAPDH.
Chromatin Immunoprecipitation with qRT-PCR
Chromatin immunoprecipitation (ChIP) was performed using an EpiQuik Tissue Chromatin Immunoprecipitation Fast kit (Epigentek) and a ChIP grade rabbit anti-H3K4me3 antibody (Abcam, Cambridge, England). Normal IgG was used as a negative control. qRT-PCR was performed with SsoAdvanced SYBR Green Supermix kit (BioRad) on purified DNA samples obtained from input, anti-H3K4me3 IP and normal IgG IP samples using primers designed against the promoter regions of Fos and Oxtr (see Table S1 in Supplement 1 for primer sequences). Immunoprecipitation values were expressed relative to each sample’s input signal.
Statistical Analysis
One-way and repeated measure analyses of variance, one-sample t tests and Wilcoxon signed-rank tests were used to analyze all data sets, with Bonferroni post hoc tests for significant effects.
Results
Formation of a METH-Associated Memory Involves Global and Promoter-Specific Changes in Marks of Active Histone Methylation in the NAc
An understanding of the epigenetic and transcriptional changes induced by METH CPP was first determined to guide our selection of the most appropriate class of epigenetic writers and erasers to explore in METH-associated memory. A typical CPP protocol involves three pairings between the unconditioned stimulus (US; METH) and CS+ (context). Through this training, animals form a Pavlovian association between the rewarding effects of METH and the environmental context in which it is administered and later show a preference for that environment. To investigate the molecular mechanisms occurring during the initial stages of formation of this conditioned response to METH, we analyzed changes in the NAc occurring after the first CS-US pairing (Figure 1A). Analysis at this time point reflects mechanisms involved in the initial encoding of an association between the US and CS before behavioral manifestation of the preference. To specifically identify changes associated with METH CPP formation, two key control groups were included: saline CPP to exclude changes associated with exposure to the novel environment and METH HC to exclude changes associated with exposure to METH alone (Figure 1A).
Figure 1Induction of posttranslational histone modifications associated with active gene expression in the nucleus accumbens (NAc) during the formation of a methamphetamine (METH)-associated memory. (A) Schematic of experimental design (S, saline; M, methamphetamine). (B) Histone modifications associated with transcriptional activation (H3K4me2, me3, pan-AcK H2-3, pan-AcK H4) or repression (H3K9me2, H3K27me2) were measured in the NAc by Western blot and normalized to total H4. (C) Representative images of Western blots. *p < .05, error bars represent SEM. CPP, conditioned place preference; HC, home cage; Sal, saline.
We first explored the potential for epigenetic regulation at the time of learning by measuring global levels of several transcriptionally active (H3K4me2/me3, pan-AcK H2-3, and pan-AcK H4) and repressive (H3K9me2 and H3K27me2) chromatin modifications in the NAc. METH CPP was accompanied by an increase in active histone methylation modifications (H3K4me2: F2,21 = 9.237, H3K4me3: F2,21 = 10.752; p < .001) and pan-AcK H2–3 (F2,15 = 5.97, p < .05), but not pan-AcK H4. This is consistent with a previous report demonstrating increased acetylation of H3, but not H4, in the HPC with contextual associative learning and in the NAc with cocaine administration (
). Methylation of residues associated with repressed transcription were unchanged or decreased (H3K9me2: F2,15 = .224; H3K27me2: F2,20 = 4.252, p > .05; Figure 1B,C).
Using the same design depicted in Figure 1A, we next performed an exploratory microarray to identify the general transcriptional profile of the NAc during the formation of a METH-associated memory. Results suggest that close to 100 genes are upregulated by METH-associated contextual learning (Figure 2A), with no genes downregulated below the 1.6-fold cutoff (Table S3 in Supplement 1). Further analysis revealed the upregulated genes fall into classes associated with synaptic growth and structure, memory, transcription, translation, chromatin modification, and development. We were particularly interested in genes with established roles in memory and/or addiction. Of these, we selected two to serve as a proxy for the epigenetic changes likely occurring on a number of genes during the associative process, Cebpd (CCAAT/enhancer binding protein delta) and Oxtr (oxytocin receptor). We also selected FBJ osteosarcoma oncogene, an immediate early gene and transcription factor, the expression of which has previously been shown to be increased in the NAc by CPP for cocaine and METH and is a commonly used marker of neural activation (
Figure 2Formation of a methamphetamine (METH)-associated memory is associated with transcriptional activation in the nucleus accumbens (NAc). (A) Using the same experimental design depicted in Figure 1A, microarray analysis in the NAc revealed the upregulation of 91 genes, relative to control groups (saline conditioned place preference [Sal CPP], METH home cage [HC]). (B) From the group of 91 genes selectively upregulated in the METH CPP group, Cebpd, Fos, and Oxtr were selected for quantitative reverse-transcriptase polymerase chain reaction validation. (C) Schematic of the region of the Fos and Oxtr promoters amplified from genomic DNA samples obtained from H3K4me3 chromatin immunoprecipitation (upper panel). Promoter-specific H3K4me3 at Fos and Oxtr was measured by chromatin immunoprecipitation–quantitative reverse-transcriptase polymerase chain reaction in the NAc of METH CPP and control mice (lower panel). *p < .05, error bars represent SEM. CS, conditioned stimulus; US, unconditioned stimulus.
Using qRT-PCR, we confirmed that the expression of Fos and Oxtr messenger (m)RNAs, but not Cebpd, was increased in the METH CPP group (Fos: F2,18 =15.541, p < .001; Oxtr: F2,18 = 6.512, p < .01; Cebpd: F2,18 = 2.877, p > .05; Figure 2B). Fos expression was also increased in the METH HC group compared with saline CPP (p < .05), yet Fos mRNA levels in the METH CPP group were still greater than the group that received METH in their home cage (p < .05; Figure 2B). Given the repressive effect of oxytocin in the NAc on METH CPP (
), the enhanced expression of Oxtr in the NAc may represent an endogenous, homeostatic mechanism aimed at counteracting the impact of METH. Future studies will be needed to explore this intriguing possibility.
We next investigated the chromatin state of Oxtr and Fos, focusing on transcriptionally permissive H3K4me3 in the promoter of these genes for several reasons. Despite being one of the more persistent histone modifications, methylation has received little attention in the memory field (
). Furthermore, trimethylation of H3K4 occurs concomitantly with active transcription, whereas dimethylation can be present at poised, inactive genes (
). Finally, the writers and erasers for this modification appear to participate in cognition, yet function in the adult brain is poorly understood. Consistent with increased Fos in the METH HC and METH CPP groups, ChIP-qPCR analysis revealed a trend toward increased H3K4me3 levels at the Fos promoter in both groups (F2,29 = 2.41, p = .108; Figure 2C). At the Oxtr promoter, levels of H3K4me3 were specifically increased by METH CPP training (F2,29 = 4.483, p < .05; Figure 2C), consistent with its transcriptional profile (Figure 2B). These results provide evidence that increased methylation of H3K4 may indeed be an important epigenetic modification involved in supporting METH-context associations, warranting further investigation of its writers and erasers.
The epigenetic writer of H3K4me, Mll1, is necessary in the NAc for METH-associated memory
We next examined modifiers of permissive methylation, focusing first on the role of a writer of H3K4 methylation, MLL1, in the modulation of METH-associated memory. Mll1 was specifically upregulated in the NAc 1 hour after training and downregulated in animals that received METH in the HC, compared with saline controls (F2,14 = 9.52, p < .005; Figure 3A). Interestingly, Mll1 mRNA levels were reduced in METH HC mice, whereas H3K4me3 levels were upregulated at the Fos promoter in this same control group (Figure 2C). This suggests that other HMTs involved in H3K4 methylation (e.g., myeloid/lymphoid or mixed-lineage leukemia 2, SET domain containing 1A and 1B) may be upregulated and/or related KDMs downregulated by nonassociative administration of METH.
Figure 3The H3K4 histone methyltransferase mixed-lineage leukemia 1 (MLL1) is increased during methamphetamine (METH) conditioned place preference (CPP) acquisition and its loss results in H3K4 demethylation. (A) Using the same experimental design depicted in Figure 1A, Mll1 expression was measured by quantitative reverse-transcriptase polymerase chain reaction. (B) Representative coronal section of a mouse brain injected in the nucleus accumbens (NAc) core with an adeno-associated virus expressing the enhanced green fluorescent protein. The same brain coordinates were used for subsequent small interfering RNA (siRNA) delivery. (C)Mll1 levels in the NAc 2 and 5 days after intra-NAc siRNA injection. (D) Intra-NAc KD of Mll1 resulted in demethylation of H3K4 in the NAc. (E) Effect of Mll1 KD on Fos and Oxtr messenger RNA levels. *p < .05, error bars represent SEM.
To determine the behavioral importance of this increase in Mll1, we used siRNA-mediated focal KD of the gene within the NAc. As can be seen from injection of a GFP-AAV, stereotaxic coordinates targeted the NAc core with spread limited to the shell (Figure 3B). Using naive adult mice, we next confirmed effective KD of Mll1 mRNA in vivo 2 and 5 days after intra-NAc siRNA injection compared with a nontargeting, control siRNA (2 days: t3 = −4.54, p < .05, 5 days: t3 = −3.63, p < .05; Figure 3C). Consistent with MLL1’s role as an H3K4 methyltransferase, its knockdown resulted in a 50% reduction in H3K4me3 as well (t7 = 2.53, p < .05; Figure 3D). Finally, we found that Mll1 KD dramatically reduced the transcript levels of Fos and Oxtr within the NAc (Fos: F1,9 = 20.38, p < .005; Oxtr: F1,9 = 10.00, p < .05; Figure 3E), suggesting that MLL1-mediated H3K4 methylation contributes to their transcriptional regulation.
To determine the role of NAc MLL1 in the formation of METH-associated memory, Mll1 was knocked down before training (Figure 4A). Animals that received intra-NAc infusions of the Mll1 siRNA failed to develop a preference for the METH-paired side of the CPP apparatus (Figure 4B; control siRNA: n = 13, Z = −2.34, p < .05; Mll1 siRNA: n = 13, Z = −1.15, p > .05). After testing, reduction of Mll1 mRNA was confirmed (F1,14 = 4.83, p < .05; Figure 4C).
Figure 4Nucleus accumbens (NAc) mixed-lineage leukemia 1 (MLL1) is required for methamphetamine (METH)-associated memory. (A) Schematic of experimental design. (B) Effect of focal knockdown of Mll1 in the NAc on METH-associated memory. (C)Mll1 knockdown in the NAc was confirmed after behavioral testing. *p < .05, error bars represent SEM. CPP, conditioned place preference; CS, conditioned stimulus; mRNA, messenger RNA; siRNA, small interfering RNA.
The epigenetic eraser of H3K4me, Kdm5c, is the most abundant H3K4 demethylase in the NAc and is necessary for expression of an established METH-associated memory
Given the loss of function achieved through KD of the H3K4me writer MLL1 (Figure 4), we next explored the potential for a gain of function (memory enhancement) through manipulation of an H3K4me eraser. We focused on KDM1A and the KDM5 family (KDM5A-D, also known as RBP2, PLU-1, SMCX, and SMCY, respectively), because of their suggested roles in cognition (
) and their expression pattern remains uncharacterized, with the exception of two earlier studies that compared whole brain levels of Kdm5c and Kdm5d (
). Thus, we first determined the abundance of Kdm5a–d and Kdm1a within the NAc of naive, adult male mice. Although transcript for all demethylases was detected in the NAc, the X-linked intellectual disability gene, Kdm5c, was expressed at relatively high levels, as its expression was approximately 20% of glyceraldehyde-3-phosphate dehydrogenase an abundant housekeeping gene (Figure 5A). The other Kdm5 family members and Kdm1a were expressed at levels 16 to 100 times less than Gapdh (Figure 5A). We then compared Kdm5c mRNA expression levels across several brain regions relevant to substance abuse and memory and found them to be comparable (Figure 5B).
Figure 5Kdm5c is the most abundant H3K4 demethylase in the nucleus accumbens (NAc). (A) Expression of the KDM5 (Kdm5a, Kdm5b, Kdm5c, and Kdm5d) and KDM1 (Kdm1a) families of H3K4 demethylases was assayed in the NAc by quantitative reverse-transcriptase polymerase chain reaction and expressed relative to Gapdh expression. (B)Kdm5c messenger RNA expression was also measured across several brain regions relevant to psychostimulants and memory (NAc, dorsal striatum [dStr], medial prefrontal cortex [PFC], amygdala [AMY], and anterior hippocampus [HPC]). Error bars represent SEM.
Kdm5c levels were equal to controls in the NAc 1 hour after training (Figure 6A; p > .05). However, this did not preclude the possibility that a loss of demethylase activity might lead to a gain of memory function through an accumulation of H3K4 methylation. Therefore, using naive adult mice, we next confirmed that intra-NAc injection of Kdm5c siRNA effectively knocked down the demethylase’s mRNA in vivo 2 days later (F1,14 = 13.23, p < .005; Figure 6B). Interestingly, transcript levels had returned to baseline by 5 days after siRNA, indicating a fairly rapid turnover rate for Kdm5c (p > .05; Figure 6B). Because so little is known about this demethylase in the adult brain and because the siRNA had a modest effect on the gene’s mRNA levels, we also confirmed KD at the protein level (F1,12 = 16.59, p < .005; Figure 6C). Consistent with reports from cell culture and peripheral tissue, KMD5C loss also led to an accumulation of H3K4me3 (F1,10 = 5.99; p < .05; Figure 6D), extending its role as an H3K4 demethylase to the brain and indicating that KDM5C contributes to the active maintenance of H3K4me3 in the NAc (
Figure 6Loss of Kdm5c in the nucleus accumbens (NAc) results in H3K4 hypermethylation. (A) Using the same experimental design depicted in Figure 1A, Mll1 expression was measured by quantitative reverse-transcriptase polymerase chain reaction. (B)Kdm5c levels in the NAc 2 and 5 days after intra-NAc small interfering RNA (siRNA) injection. (C) KDM5C protein levels in the NAc 2 days after intra-NAc siRNA injection. (D) Intra-NAc knockdown of Kdm5c resulted in hypermethylation of H3K4 in the NAc. *p < .05, error bars represent SEM. CCP, conditioned place preference; HC, home cage; METH, methamphetamine; Sal, saline.
To determine the role of NAc KDM5C in the formation of METH-associated memory, Kdm5c was knocked down before training by two siRNA infusions to ensure suppression of the demethylase for the entire training period (Figure 7A). Control and Kdm5c siRNA-treated groups developed an equal preference for the METH-paired side of the CPP apparatus (Figure 7B; Control siRNA: n = 8, Z = –2.24, p < .05; Kdm5c siRNA: n = 10, Z = –1.99, p < .05), with Kdm5c KD showing no memory-enhancing effect. After the behavioral task, we observed that KDM5C and H3K4me3 levels had returned to baseline by the time of expression testing (KDM5C: F1,13 = .02, P > .05; H3K4me3: F1,12 = .79, p > .05; Figure 7C), again indicating rapid recovery of KDM5C levels after siRNA treatment. Importantly, both locomotion and velocity were normal in the Kdm5c siRNA-treated mice during training (Figure S1 in Supplement 1), indicating that a lack of memory enhancement could not be attributed to a change in the degree of METH-induced locomotor activation.
Figure 7Focal knockdown of Kdm5c in the nucleus accumbens (NAc) after consolidation disrupts methamphetamine (METH)-associated contextual memory. (A) Schematic of experimental design for panels B and C. (B) Pretraining intra-NAc Kdm5c small interfering RNA (siRNA) had no effect on METH-associated memory. (C) KDM5C protein and H3K4me3 levels in the NAc were measured after behavioral testing. Lower panel contains representative images of Western blots for KDM5C (~170 kDa), H3K4me3 (~17 kDa), and GAPDH (~36 kDa). (D) Schematic of experimental design for panels E and F. (E) METH-associated memory was assayed before (Test 1) and after (Test 2) Kdm5c siRNA infusion. (F) KDM5C protein and H3K4me3 levels in the NAc were measured after behavioral testing. Lower panels contain representative Western blot images of KDM5C, H3K4me3, and GAPDH. *p < .05, error bars indicate SEM. CPP, conditioned place preference; CS, conditioned stimulus.
We next explored the potential for posttraining manipulation of KDM5C to affect expression of METH-associated memory because disruption of established drug-associated memories is an emerging approach to the treatment of substance abuse (
). To achieve the necessary temporal control over KDM5C, we infused Kdm5c siRNA into the NAc 1 day after the first METH CPP test (p > .05; Figure 7D,E). At the post-siRNA injection test (Test 2), Kdm5c KD mice no longer showed a preference for the METH-paired chamber (Test × siRNA interaction F1,20 = 5.061, p < .05; Test 2 control versus Kdm5c siRNA p < .005; Figure 7E). Suspecting that this effect may have been due to a rapid acceleration of extinction (
), we examined place preference expression during the first 5 minutes of the test because accelerated extinction would be defined by an initial preference that rapidly dissipated during the 15-minute test period. This lack of a preference for the METH-paired compartment in Kdm5c siRNA animals was apparent even in the first 5 minutes (F1,21 = 6.90, p < .05; Figure S2 in Supplement 1), suggesting that a lack of CPP expression in Test 2 was not due to accelerated extinction. Following behavioral testing, reduction of KDM5C protein and hypermethylation of H3K4 by Kdm5c siRNA were confirmed (KDM5C: F1,12 = 7.24, p < .05; H3K4me3: F1,18 = 4.85, p < .05; Figure 7F).
Discussion
METH-associated memory formation is accompanied by the induction of a NAc transcriptional program linked to permissive histone modifications. Consistent with this, we found that loss of the H3K4 HMT, Mll1, within the NAc impaired METH-associated memory, whereas postconsolidation loss of Kdm5c disrupted the expression, but not formation, of the same memory. This indicates that a balance of histone methylation must be maintained in the NAc to successfully support memory.
The upregulation of several epigenetic marks associated with active transcription correlated with the increase in expression of close to 100 genes by METH-context associative learning. Although our study does not fully address this possibility, these chromatin modifications are likely a driving influence in the transcriptional changes reported, as suggested by the impact of Mll1 KD on baseline Fos and Oxtr expression. However, these epigenetic changes are also expected to place a number of genes in a transcriptionally primed state, such that the threshold for a given gene to respond to future stimulation (
), such as subsequent CPP training trials, is altered. Therefore, these chromatin modifications may influence subsequent gene transcription, enabling a context to produce a behavioral outcome (a place preference or context-induced drug seeking) in the absence of reinforcement (METH). Additional mechanistic studies will be needed to test such a hypothesis.
An emerging theme for drug-associated memory is that inhibition of repressive epigenetic modifying enzymes (G9a, DNMT3a, HDAC5) in the NAc enhances the acquisition of CPP (
). This can now be extended to include the activating HMT, MLL1, as loss of the enzyme disrupted METH CPP. Interestingly, KDM5C belongs to the former category of repressors of gene expression (
). However, unlike the abovementioned examples, KDM5C KD in the NAc impaired METH CPP only after conditioning had taken place. This result is especially intriguing because KD of KDM5C and dysregulation of H3K4me3 levels occurred after the memory had stabilized. Given the timing of the manipulation, a blockade of reconsolidation is unlikely to account for the effect (
). Another possibility is accelerated extinction (within the Test 2 session), but our time-bin analysis of the preference during Test 2 (Figure S2 in Supplement 1) revealed that Kdm5c KD prevented expression of the place preference from the beginning of the test. Rather, the result predicts that this KDM controls the maintenance or ability to retrieve memory by regulating stability of the trace, motivation for drug seeking, and/or ability to access the memory. KDM5C has been implicated in the suppression of REST (RE1-Silencing Transcription Factor) (
), a transcriptional repressor. A number of memory-related genes have RE1 binding sites, including the N-methyl-D-aspartate receptor subunits NR1, NR2A, and NR2B. NR1 expression is required for the long-term maintenance of memory and NR2A is required for memory retrieval (
). Thus, loss of KDM5C may disrupt established memory through an accumulation of H3K4 methylation that drives REST transcription, leading to a repression of key components of the N-methyl-D-aspartate receptor. Regardless of the exact mechanism, these results represent a new potential target for disrupting context-induced METH seeking behavior.
Although our results and those of others indicate that Kdm5c continues to be expressed in the adult brain (
), virtually nothing is known about its functional role in this context. Our demonstration that in vivo KDM5C KD results in enhanced levels of H3K4me3 indicates that the protein continues to actively function as a demethylase in the adult brain and may be the predominant KDM in the NAc. Elevation of H3K4me3 levels with KDM5C KD also indicates the presence of active HMTs in NAc cells because methyl groups accrue on the H3K4 residue in its absence, even under baseline conditions (Figure 6D). Thus, the levels of this epigenetic mark appear to be regulated in mature neurons by a dynamic interplay between H3K4 HMTs, such as MLL, and KDMs. Such dynamic control of H3K4me3 levels in the adult brain suggests that this residue can rapidly respond to environmental cues in the control of gene expression. Importantly, in addition to the disruption of METH-associated memory following pretraining loss of MLL, knockdown of KDM5C disrupted an already-established association, representing the first evidence for a behavioral regulatory role of any member of the KDM5 family of histone demethylases in the adult brain.
This work was supported by the National Institute on Drug Abuse to CAM (Grant Nos. DA024761, DA033499, and DA034116) and Fonds de la Recherche en Santé du Québec to AAV.
We thank Ms. Vivian Hemmelder for her technical assistance, Dr. Alicia Faruzzi-Brantley and the TSRI Behavior Core, and Dr. Yang Shi at Children’s Hospital Boston and Harvard Medical School for providing the antibody for KDM5C detection.
The authors report no biomedical financial interests or potential conflicts of interest.
G9a/GLP histone lysine dimethyltransferase complex activity in the hippocampus and the entorhinal cortex is required for gene activation and silencing during memory consolidation.
Selective, retrieval-independent disruption of methamphetamine-associated memory by actin depolymerization [published online ahead of print September 5].
00855-X&id=gr1.jpg){kind=link}
00855-X&id=gr2.jpg){kind=link}
00855-X&id=gr3.jpg){kind=link}
00855-X&id=gr4.jpg){kind=link}
00855-X&id=gr5.jpg){kind=link}
00855-X&id=gr6.jpg){kind=link}
00855-X&id=gr7.jpg){kind=link}