Basal microRNA expression patterns in reward circuitry of selectively bred high-responder and low-responder rats vary by brain region and genotype

Abstract

Mental health disorders involving altered reward, emotionality, and anxiety are thought to result from the interaction of individual predisposition (genetic factors) and personal experience (environmental factors), although the mechanisms that contribute to an individual's vulnerability to these disorders remain poorly understood. We used an animal model of individual variation [inbred high-responder/low-responder (bHR-bLR) rodents] known to vary in reward, anxiety, and emotional processing to examine neuroanatomical expression patterns of microRNAs (miRNAs). Laser capture microdissection was used to dissect the prelimbic cortex and the nucleus accumbens core and shell prior to analysis of basal miRNA expression in bHR and bLR male rats. These studies identified 187 miRNAs differentially expressed by genotype in at least one brain region, 10 of which were validated by qPCR. Four of these 10 qPCR-validated miRNAs demonstrated differential expression across multiple brain regions, and all miRNAs with validated differential expression between genotypes had lower expression in bHR animals compared with bLR animals. microRNA (miR)-484 and miR-128a expression differences between the prelimbic cortex of bHR and bLR animals were validated by semiquantitative in situ hybridization. miRNA expression analysis independent of genotype identified 101 miRNAs differentially expressed by brain region, seven of which validated by qPCR. Dnmt3a mRNA, a validated target of miR-29b, varied in a direction opposite that of miR-29b's differential expression between bHR and bLR animals. These data provide evidence that basal central nervous system miRNA expression varies in the bHR-bLR model, implicating microRNAs as potential epigenetic regulators of key neural circuits and individual differences associated with mental health disorders.

the limited biological understanding of many mental health disorders stems from the mechanistic complexity associated with these diseases that lie at the nexus of dynamic environmental and genetic factors. Innate differences in temperament encountered in reaction to environmental situations and stressors can be observed in psychiatric diseases, like addiction, as individuals vary not only in their propensity to use drugs but also in their vulnerability to subsequently transition to drug dependence. These individual differences in susceptibility and sensitivity are a common feature throughout mental health biology, and we seek to identify the gene expression background within which these individual variations take place to understand the underlying physiologies.

Past studies have highlighted the important observation that personality traits can be strong predictors of certain psychopathologies (1, 23, 25). In humans, for example, the behavioral measure of novelty-seeking is directly correlated with the development of externalizing disorders (e.g., drug abuse) and inversely correlated with the development of internalizing disorders (e.g., anxiety and depression) (10). In efforts to further understand the individual differences of this disease vulnerability, we have employed the high-responder/low-responder (HR-LR) novelty-seeking model that defines variation in rats based on behavioral differences in an open, inescapable environment (37). HR and LR animals differ not only in spontaneous novelty-seeking behavior but also in stress reactivity, anxiety-like behaviors, and affinity toward drugs of abuse (15, 37, 38). Furthermore, HR rats also differ from LR rats in the rate and dosage of self-administered cocaine, ethanol, nicotine, and amphetamine (i.e., HR rats consume more of these substances at a faster rate than LR animals) and acquire self-administration behavior more quickly (2, 31, 38, 44). Investigations into the physiology behind these anxiety, stress, and reward processing alterations seen in the HR-LR model have also identified numerous neurochemical and molecular differences between the phenotypes, including altered dopamine and stress signaling, histone acetylation, and mRNA expression in various brain regions (6, 7, 14, 19, 20, 39). These observations combine to create an animal model potentially relevant to the individual vulnerability for a broad range of mental health disorders. To further understand the genetic basis of the HR-LR phenotypes as a model of mental health vulnerability, we have employed a selective breeding program based on the novelty-seeking trait. This approach has facilitated the enrichment of genetic variations present in outbred rats while also enabling a priori determinations regarding behavior phenotype in the bred HR (bHR) and bred LR (bLR) lines, allowing for analysis of the gene expression at which the variation in temperamental liability exists.

With little evidence of dramatic monogenic mRNA effects in many mental health diseases or within the HR-LR model, recent studies have begun to investigate other regulatory mechanisms of gene expression [e.g., DNA methylation and microRNAs (miRNAs)]. miRNAs are an intriguing mechanistic candidate for complex disease pathologies because of their ability to influence networks of genes through posttranscriptional regulation (21, 41). Hundreds of miRNAs display enriched expression in the brain, and miRNAs can influence critical pathways in brain physiology (e.g., synaptic development and plasticity) that are altered in multiple central nervous system disorders (e.g., addiction, Alzheimer's disease, and schizophrenia) (26, 29, 43). Associations between miRNAs and altered brain physiology are continuing to emerge based on the observation of direct miRNA regulation of common anxiety- and drug-related mRNAs (e.g., Mecp2, Ntrk3, Bdnf, Dnmt3a, and Creb) (5, 8, 13, 17, 24, 34). Despite these intriguing findings, the role of miRNAs in the HR-LR model as well as whether miRNAs contribute to individual differences in liability to mental health disorders are currently unknown.

In addition to genetic and epigenetic components, neural circuitry is an important factor in understanding brain physiology. Previous work has implicated modifications in the mesocorticolimbic reward system, a major dopaminergic circuit in the brain that includes the nucleus accumbens (core and shell subregions), prelimbic cortex, hippocampus, amygdala, and ventral tegmental region, in the pathology of numerous diseases (18, 22, 46). The functional association between anxiety, reward, and emotional processing and the mesocorticolimbic neural circuitry highlights its importance as a target of study to the understanding of the HR-LR model and many mental health disorders (25, 47).

Although progress has been made in understanding genetic and neuroanatomical components of many central nervous system disorders, less is known about how these factors contribute to individual variation and vulnerability for these diseases. Given the associations between miRNAs and brain pathologies and their mechanistic potential to contribute to polygenic phenotypes through regulating networks of gene expression, we hypothesized that the bHR-bLR model would contain differential miRNA expression patterns in brain regions associated with reward and emotional processing. Here we present an extensive characterization of miRNA expression patterns in three mesocorticolimbic structures [prelimbic cortex (PL), nucleus accumbens core (NAcC), nucleus accumbens shell (NAcS)] in bHR and bLR rats. Our data offer insight into the potential role of miRNAs in an animal model relevant to individual differences in mental health liability, enabling further examination into the complexity underlying these diseases.

METHODS

Animal Handling and Sample Generation

Rats were handled and treated in accordance with the ethical guidelines put forth and protocol approved by the University of Michigan Committee on the Use and Care of Animals. The bHR and bLR male rats were obtained from in-house breeding colonies and classified based on reactivity in a novel environment as previously described (33). Rodents in this study (generation 22) consisted of 10 bHR and 10 bLR adult male rats (70–90 days of age). Animals were killed by decapitation between the hours of 8:00 AM and 10:00 AM. Brains were immediately removed and frozen in chilled isopentane (−35°C). Brains were then stored at −80°C until tissue sectioning.

Tissue Sectioning and Neuroanatomical Details

Prior to sectioning, brain tissue was embedded in M-1 embedding matrix (Thermo Shandon, Hanover Park, IL) and mounted with O.C.T. (Tissue Tek, Radnor, PA). Tissues were sectioned on a cryostat (CM1850 Leica) at a thickness of 10 μm, and sections were then thaw-mounted, two per slide, on Superfrost/Plus Microscope Slides (Fisher Scientific, Hanover Park, IL). Slides containing tissue sections were stored at −80°C until further processing. To identify the PL and NAc brain regions in all animals prior to laser capture microdissection (LCM) or in situ hybridization (ISH) studies, every 25th slide from each animal was stained with cresyl violet and compared with the Paxinos and Watson Rat Atlas (36). Additionally, PL and NAc were identified in respective sections by locating the corpus callosum, anterior commissure, and lateral ventricle as anatomical markers. Areas of abutment of NAcC and NAcS subregions were delineated by a visible change in cell density initially located at the intersection of a linear extension of the lateral ventricle in the ventral direction and in the medial direction from the center of the anterior commissure (Fig. 1).

Fig. 1.

Neuroanatomical summary of laser capture microdissection tissue acquisition. Brain atlas representations of experimental regions of interest (ROI) were used during laser capture microdissection (LCM) for prelimbic cortex (PL) (A) and nucleus accumbens core (NAcC) and nucleus accumbens shell (NAcS) (B). Top: relevant atlas coronal section including ROI with area of magnification for bottom left panel outlined in red. Bottom right: an image of an actual experimental slide post collection of ROI via LCM.

LCM

Six PL-containing slides and six NAc-containing slides were chosen from each animal (n = 20, 12 slides per animal per group with 2 sections per slide) for LCM. In preparation for LCM, tissue sections were dehydrated in alcohol washes and xylene before being air-dried. Tissue samples were collected on CapSure LCM Macro caps (Applied Biosystems, Foster City, CA) using either the AutoPix or ArcturusXT LCM instruments (Applied Biosystems) with ×4 or ×10 objectives (Fig. 1). We calibrated LCM spot size for effective collection of target tissue while avoiding contamination from adjacent tissue areas. Instrument settings ranged from 55 to 65 mW (power) for 5,000 to 5,500 ms (duration). For the majority of samples, PL captures from four sections on two consecutive slides were placed on a single LCM cap. NAcC and NAcS tissues were captured on separate caps; captures from one and a half slides (three sections) were placed on two caps (one cap for NAcC samples, one for NAcS samples).

RNA Isolation, Qualification, and Quantification

Following tissue collection, LCM films were removed from their caps and inserted into 0.5 ml tubes containing lysis solution from the RNAqueous-Micro kit (Ambion, Austin, TX) and incubated at 42°C for 30 min. Resulting extracts were either purified immediately or stored at −80°C until further use. We isolated RNA using the RNAqueous Micro procedure after adding 1.25 volumes of 100% ethanol to recover both small and large RNA species. Following elution, samples were treated with DNase I to remove contaminating DNA. RNA quantity and quality were assessed with the RiboGreen RNA assay (Molecular Probes, Carlsbad, CA) and the 2100 Bioanalyzer (Agilent Technologies, Wilmington, DE), respectively. RiboGreen assays were run on a Fluostar Optima fluorescence reader (BMG Labtech, Ortenberg, Germany) based on the low-range standard curve outlined by the manufacturer.

miRNA Quantitative PCR Arrays and Individual Assays

Quantitative real-time PCR (qPCR) was used to measure expression of >500 mature miRNAs with TaqMan Low Density Array (TLDA) cards (Rodent MicroRNA A and B v2.0, Applied Biosystems). Reverse transcription of 15 ng of total RNA was conducted using stem-loop Megaplex primer pools (Applied Biosystems) and 2.5 μl of the resulting material transferred to a preamplification reaction. The final preamplified sample was diluted to 100 μl and stored at −20°C until further processing. A reaction mixture of 9 μl diluted preamplified material, 441 μl nuclease-free water, and 450 μl TaqMan Universal PCR Master Mix was loaded onto the TLDA cards and run on a PRISM 7900HT Sequence Detection System (Applied Biosystems) or ViiA 7 Real-Time PCR System (Applied Biosystems). SDS v2.3 software (Applied Biosystems) and ViiA 7 software (Applied Biosystems) were used for operation of respective instruments.

We selected a subset of miRNAs as candidates for validation by a more sensitive and accurate qPCR detection methodology using TaqMan individual miRNA assays (specific miRNA assays corresponded to TLDA rodent pools A and B, Applied Biosystems). Individual miRNA assays were prepared in a reaction mix containing diluted preamplified material, nuclease-free water, and TaqMan Universal PCR Master Mix. Reactions were run in 96-well 0.1 ml plates.

miRNA ISH

ISH probes were generated from RNA oligonucleotides (Invitrogen, Carlsbad, CA) with minor changes to the labeling reaction as follows: 0.75 μl T4 DNA Kinase, 1 μl 10× Polynucleotide Kinase Buffer (Affymetrix, Santa Clara, CA), 2 μl RNA (20 pmol/μl), and 6.25 μl ³³P-γATP (Perkin Elmer, Waltham, MA) and incubated at 37°C for 30 min (45). Antisense and control (two interior mismatched nucleotides) RNA oligonucleotides (Invitrogen) were synthesized based on sequences obtained from miRBase.org and hybridization specificity as determined through comparative ISHs failing to yield autoradiographic signal above background (Fig. 2). Two slides (4 total sections) from each bHR and bLR animal derived from the PL or NAc were used per probe. For signal detection, optimal exposure time to Kodak BioMax MR film (Fisher Scientific) was determined empirically for each miRNA (6–30 days).

Fig. 2.

Specificity of microRNA (miRNA) in situ hybridization (ISH) probes. A comparison of neuroanatomical expression pattern of microRNA (miR)-128 (A), miR-484 (C), and miR-29b (E) to its 2MM (2MM = 2 bp mismatched probe) to ensure specificity, displayed in generic rat brain tissue containing the PL (A + C) and NAc (E). AS, antisense.

Pathway Analysis of miRNA-mRNA Target Relationships

Functional analyses were generated through the use of Ingenuity Pathway Analysis (IPA; Ingenuity Systems, http://www.ingenuity.com) as a general filtering mechanism to identify experimentally proven miRNA-mRNA interactions in the literature. A feature within IPA allows users to examine literature-based miRNA-mRNA relationships (e.g., validated mRNA targets of miRNAs). Hence, our results were limited solely to the downstream pairings of miRNAs and mRNAs to increase the degree of confidence in these interactions (“experimentally validated”). Further identification of mRNA targets and investigations of the IPA analysis were conducted through extensive literature review.

mRNA qPCR Methods

Purified RNA samples (30–60 ng total mass) were converted to cDNA using a High Capacity Reverse Transcription cDNA kit (Applied Biosystems). The resulting cDNA created was used at a constant volume per well in qPCR reactions with Absolute Blue SYBR Green reagents according to manufacturer protocol (Thermo Scientific). mRNAs were selected for analysis based on 1) experimentally validated regulation by one or more of the miRNAs differentially expressed in the study and 2) reported association to central nervous system function. Primers were designed using Primer Blast software (National Center for Biotechnology Information; amplicons 70–250 nucleotides, spanned exon-exon junction if possible; Fig. 9B). The mRNAs were measured in samples derived from brain regions that demonstrated differential expression of associated miRNA(s). NAcC mRNA expression was not assessed due to lack of sufficient sample from this brain region.

Fig. 9.

qPCR analysis of specific miRNAs and a subset of predicted mRNA targets in PL and NAcS of bHR-bLR samples identifies inverse expression of miR-29b and DNMT3a mRNA. A: graph displaying FED for several miRNAs whose genotypic differential expression was validated by qPCR individual assay and a subset of respective mRNA targets in both PL and NAcS. List of mRNAs targeted by validated miRNAs along with related pathways is contained in Table 1. *P < 0.05, error bars ± SE. B: primer sequences and respective FED for mRNA qPCR validation experiments.

Methods of Statistical Analysis

TLDA.

miRNA expression data were compiled with SDS RQ Manager (v2.3, Applied Biosystems) and ViiA 7 software and then analyzed with RealTime StatMiner software (Integromics, Madison, WI). The 2^−ΔΔCt method was employed to calculate relative quantification values and log fold expression difference (FED) values (expression normalized to MammU6 and U87) (28). We defined absolute miRNA expression as detectable based on 1) reliable precision of measurement (i.e., C_t values <34), and 2) results indicating differential expression determined by Linear Models for MicroArray were considered significant when P < 0.05 without false discovery rate. Biological replicates with poor correlation were identified using RealTime Statminer, and samples with correlation flags >20 were excluded from all further analyses. To maintain consistency between differential expression comparisons, we also excluded samples identified as outliers in one comparison from all subsequent comparisons involving that sample (e.g., if sample PL HR-X was identified as an outlier in comparison of HR PL vs. LR PL, it was also excluded in comparison of HR PL vs. HR NAcC). The largest number of samples excluded from either genotype per brain region was two (i.e., sample numbers in all analyses ≥8). All reported instances of differential expression were derived from postexclusion analyses unless otherwise noted.

Individual miRNA assay qPCR.

miRNA validation candidacy was derived from 1) average C_t value and standard deviation among biological replicates and 2) magnitude of observed FED as reported on the original TLDA card analysis. miRNAs with 1) the lowest standard deviation, 2) lowest average C_t values (indicating highest expression), and 3) largest FED were assigned priority for validation efforts. With few exceptions, validation was not attempted for differentially expressed miRNAs where FED was <2 or where reliable detection was not achieved in both sample groups (i.e., average C_t value >34). All cDNA samples (e.g., n = 10), assayed in technical triplicate, were analyzed with the 2^−ΔΔCt method normalized to MammU6. Differential miRNA expression was considered validated when a two-tailed Student's t-test resulted in P < 0.05.

miRNA ISH.

Selection of miRNAs for ISH testing was based on the following criteria: 1) differential expression in TLDA analysis, 2) differential expression validated by individual qPCR analysis, and 3) reliable detectability (i.e., C_t values < 30 in all qPCR experiments). Autoradiograms from ISH were scanned with a ScanMaker 1000XL Pro Flatbed Scanner (Microtek, Santa Fe Springs, CA) using SilverFast Ai Imaging Software (LaserSoft Imaging, Sarasota, FL). Blinded optical density (OD) measurements were generated from the scanned images using ImageJ software (v1.45S, NIH). Measurements were collected separately by hemisphere and averaged within section prior to background subtraction. The OD mean of the area of interest was then subtracted from a background value of a nontissue aspect of the slide (Background = 3.5*SD of Background + Background mean) to calculate a final normalized OD (average from two slides per animal). Comparison between bHR-bLR animals was conducted with a two-tailed Student's t-test with statistical significance assigned when P < 0.05.

mRNA qPCR.

Differential expression analysis was performed with the 2^−ΔΔCt method by averaging technical replicates per animal, normalized to β-actin to account for variations in mass of total RNA input for reverse transcription. Analyses utilized β-actin as the lone housekeeping gene due to the longstanding use of this mRNA in our group and previous literature as an unregulated mRNA in nonmanipulated brain samples. Outlier measurements were identified by the Grubbs test and excluded from downstream analyses (12). Statistical significance was determined by a two-tailed Student's t-test with significance assigned when P < 0.05.

RESULTS

bHR Animals and bLR Animals Demonstrate Global miRNA Expression Patterns That Vary by Genotype and Brain Region

We measured the basal expression of >500 miRNAs between bHR rats and bLR rats in three brain regions associated with reward processing (PL, NAcC, and NAcS); 187 unique miRNAs demonstrated differential expression by genotype in at least one brain region (Fig. 3). Between bHR animals and bLR animals, 119 miRNAs were differentially expressed in PL, 73 miRNAs were differentially expressed in NAcS, and 23 miRNAs were differentially expressed in NAcC (Fig. 3). In general, the trend of miRNA expression by genotype demonstrated lower expression levels of miRNAs in bHR animals compared with bLR animals. Complete qPCR array data for differential miRNA expression by genotype are listed (Supplemental Table S1).¹ Additionally, 101 unique miRNAs demonstrated differential expression by brain region that was consistent between the two genotypes (Fig. 4). In both bHR animals and bLR animals, 94 miRNAs were differentially expressed between PL and NAcC, 81 miRNAs were differentially expressed between PL and NAcS, and 34 miRNAs were differentially expressed between NAcS and NAcC. There was extensive overlap among miRNAs differentially expressed between these three brain regions, particularly between miRNAs differentially expressed between PL and either region of the nucleus accumbens (Fig. 4B). In general, the comparison of individual miRNA expression differences between the three brain regions yielded the following general trend of relative miRNA expression: PL > NAcS > NAcC. The miRNAs demonstrating consistent differential expression by brain region independent of genotype (Fig. 4C) and complete qPCR array data for differential miRNA expression by brain region are listed (Supplemental Table S2). The complete list of C_t values generated from qPCR array experiments prior to any differential expression analyses is listed (Supplemental Table S3).

Fig. 3.

Multiple miRNAs are differentially expressed by bred high responder/bred low responder (bHR-bLR) genotype. A: triangular diagram representing miRNA expression comparisons made between bHR and bLR animals across 3 brain regions. B: proportional Venn diagram summary of overlap in miRNAs demonstrating differential expression identified by qPCR arrays by bHR-bLR genotype per brain region. Blue-shaded area represents the number of miRNAs differentially expressed by genotype in PL only (98 miRNAs), yellow-shaded area represents miRNAs differentially expressed in NAcS only (48 miRNAs), and red-shaded area represents miRNAs differentially expressed in NAcC only (15 miRNAs). The full summary of Venn diagram areas and miRNAs contained therein is listed in Supplemental Table S6. C: directional fold expression difference (FED) data displaying the greatest magnitude miRNA expression differences from qPCR arrays between bHR and bLR animals.

Fig. 4.

miRNAs are differentially expressed by brain region in the mesocorticolimbic circuit independent of bHR-bLR genotype. A: triangular diagram representing the miRNA expression comparisons made between PL, NAcC, and NAcS independent of genotype. B: proportional Venn diagram summary of overlap in miRNAs demonstrating differential expression identified by qPCR arrays between brain regions independent of bHR-bLR genotype. Purple-shaded area represents the number of miRNAs differentially expressed between PL and NAcC only (9 miRNAs), green-shaded area represents miRNAs differentially expressed between PL and NAcS only (2 miRNAs), and orange-shaded area represents miRNAs differentially expressed between NAcC and NAcS only (3 miRNAs). Areas of overlap represent the number of miRNAs differentially expressed in >1 brain region comparison e.g., area where purple and green overlap represents miRNAs differentially expressed between both PL and NAcC, and PL and NAcS (56 miRNAs), and the area where all 3 regions overlap represents miRNAs differentially expressed between PL and NAcC, PL and NAcS, and NAcC and NAcS (21 miRNAs). Full summary of Venn diagram areas and miRNAs contained therein is listed in Supplemental Table S6. C: directional FED data from qPCR arrays for 20 miRNAs demonstrating consistent bHR-bLR-independent differential expression across all 3 brain regions.

Differential miRNA Expression by 1) Genotype and 2) Brain Region Validated by Individual miRNA qPCR Assays and ISH Analysis

Based on analysis of differential expression between bHR and bLR genotypes in TaqMan qPCR array data, we selected 21 unique miRNAs involving 29 instances of differential expression for validation analyses with individual qPCR assays. In terms of neuroanatomy, these selected miRNAs contained 12 instances in the PL, 11 instances in NAcS, and six instances in NAcC (Fig. 5). Fifteen of 29 miRNAs demonstrated differential expression by qPCR via individual assay consistent with the array data (6/12 bHR:bLR PL, 6/11 bHR:bLR NAcS, 3/6 bHR:bLR NAcC; percentage of differential bHR:bLR instances validated = 51.7%). All miRNA validations exhibited lower expression in bHR relative to bLR (Supplemental Table S4). We validated differential expression of four miRNAs by genotype across multiple brain regions [Fig. 6; microRNA (miR)-544, miR-598, miR-484, miR-192]. Conversely, differential expression in a single brain region was validated for miR-26b, miR-128a, miR-194, and miR-340-5p (in PL, Fig. 5A), as well as miR-29b and miR-592 (in NAcS, Fig. 5B). Using individual miRNA qPCR assays, we examined the expression of seven miRNAs (miR-126-5p, miR-128, miR-26b, miR-484, miR-194, miR-192, miR-188-5p) across all three brain regions in both genotypes. These miRNAs were chosen because they were reported as differentially expressed in TLDA miRNA array analysis in at least one brain region independent of genotype (i.e., consistent expression difference in both bHR and bLR animals). Differential expression by brain region was observed for 36 of 42 comparisons (13/14 PL:NAcC, 11/14 PL:NAcS, 12/14 NAcC/NAcS; percent of comparisons exhibiting differential brain region expression = 85.7%; Fig. 7, Supplemental Table S5). Of 42 comparisons, 28 were differentially expressed in the original TLDA analysis, all of which validated by individual assay qPCR (12/12 PL:NAcC, 8/8 PL:NAcS, 8/8 NAcC/NAcS; percent of differential brain region expression instances validated = 100%; Fig. 7, Supplemental Table S5). Of note, eight out of the 36 instances of differential expression identified by individual assay were not significant in the original TLDA card data, including miR-192 in the NAcC vs. NAcS (bHR, bLR), miR-484 in PL vs. NAcS (bHR, bLR), miR-194 in NAcC vs. NAcS (bLR), and miR-188-5p in the PL vs. NAcC (bHR, bLR), as well as NAcC vs. NAcS (bHR). This discrepancy is likely due to lower variability in the individual assay validation studies based on greater sample measurements (i.e., data derived from singleton assays in TLDA experiments vs. average of technical triplicates in validation studies).

Fig. 5.

Differential miRNA expression in bHR-bLR genotype validates by individual qPCR assays. Graph of miRNA expression differences represented in FED magnitude comparing bHR animals to bLR in the PL. bHR n = 8, bLR n = 10, and P values were as follows: miR-26b: 0.005, miR-484: 0.026, miR-199a-3p: 0.052, miR-194: 0.011, miR-192: 0.017, miR-340-5p: 0.047, miR-128a: 0.044, miR-369-3p: 0.333, miR-126-5p: 0.252, miR-1: 0.250, miR-450a: 0.580, miR-188-5p: 0.607 (A), NAcS, bHR n = 10, bLR n = 9 (bHR = 9 and bLR = 9 for miR-592, a TLDA pool B miRNA), and P values are as follows: miR-496: 0.083, miR-484: <0.001, miR-544: 0.036, miR-592: 0.013, miR-192: 0.002, miR-29b: 0.003, miR-598: 0.038, miR-509-3p: 0.110, miR-93: 0.089, miR-1: 0.717, miR-31: 0.562 (B), and NAcC, bHR n = 9, bLR n = 10, and P values are as follows: miR-544: 0.050, miR-598: 0.002, miR-192: 0.010, miR-1: 0.608, miR-496: 0.608, miR-872: 0.412 (C) measured via individual TaqMan qPCR assays. *P < 0.05, error bars ± SE.

Fig. 6.

Select miRNAs are differentially expressed by bHR-bLR genotype across multiple brain regions. A: graph representing the FEDs for 4 miRNAs with differential expression identified by individual assays of bHR animals compared with bLR in multiple brain regions. NAcC bHR n = 9, bLR n = 10. NAcS bHR n = 10, bLR n = 9. PL bHR n = 8, bLR n = 10. P values are as follows: miR-544 NAcC: 0.050; NAcS: 0.036, miR-598 NAcC: 0.002; NAcS: 0.038, miR-484 NAcS: <0.001; PL: 0.026, miR-192 NAcC: 0.010; NAcS: 0.002; PL: 0.017. B: proportional Venn diagram displaying relationships between miRNAs with validated differential expression by genotype in multiple brain regions. Blue-shaded area represents the number of miRNAs differentially expressed by genotype in PL only (4 miRNAs), yellow-shaded area represents miRNAs differentially expressed in NAcS only (3 miRNAs), and red-shaded area represents miRNAs differentially expressed in NAcC (all miRNAs whose genotypic differential expression validated in NAcC also validated in other brain regions). Areas of overlap represent the number of miRNAs differentially expressed in >1 brain region e.g., area where blue and yellow overlap represents miRNAs differentially expressed between bHR and bLR animals in both PL and NAcS (1 miRNA). *P < 0.05, error bars ± SE.

Fig. 7.

Differential miRNA expression between PL and NAc validates by individual qPCR assays. Graph of miRNA expression differences represented in FED magnitude in both bHR animals and bLR animals comparing miRNA differences between PL and NAcC, PL and NAcS, and NAcC and NAcS measured via individual TaqMan qPCR assays. Sample n are as follows: PL bHR = 8, bLR = 10; NAcS bHR = 10, bLR = 9; NAcC bHR = 9, bLR = 10. P values are as follows, listed by miRNA in order of comparison (bHR PL vs. NAcC, bLR PL vs. NAcC, bHR PL vs. NAcS, bLR PL vs. NAcS, bHR NAcC vs. NAcS, bLR NAcC vs. NAcS): miR-126-5p: <0.001, <0.001, 0.075, <0.001, 0.023, 0.007; miR-128a: <0.001, <0.001, <0.001, <0.001, 0.001, 0.001; miR-26b: 0.012, <0.001, 0.007, 0.015, 0.118, 0.035; miR-484: 0.001, <0.001, 0.007, 0.015, 0.118, 0.035; miR-194: 0.014, <0.001, 0.390, 0.007, 0.072, 0.001; miR-192: 0.002, <0.001, 0.096, 0.004, 0.021, 0.033; miR-188-5p: 0.025, 0.032, 0.001, 0.001, 0.006, <0.001. *P < 0.05, error bars ± SE.

To examine the spatial neuroanatomical patterns of miRNA variation in the bHR/LR model, we used radioactive ISH to examine a subset of the miRNAs whose genotypic differential expression we had previously validated by qPCR. Seven miRNAs were assessed by ISH (PL only: miR-26b, miR-192, miR-194, miR-128a; NAcS only: miR-29b; NAcC and NAcS: miR-598; PL and NAcS: miR-484). In control experiments, we detected measurable signal for three of the seven miRNAs (Fig. 2; miR-484, miR-128a, miR-29b). Densitometry analysis of miR-484 and miR-128a expression revealed that both microRNAs were expressed at lower levels in the PL of bHR compared with bLR rats (Fig. 8A). miR-484 and miR-29b were measured in the NAcS but were not observed to be differentially expressed by genotype via ISH.

Fig. 8.

miRNA-128 and miR-484 demonstrate neuroanatomical validation of differential expression by genotype in prelimbic cortex of bHR-bLR animals. A: graph of the expression difference as quantitated by qPCR individual assay and radioactive ISH semiquantitative analysis. PL bHR n = 8, bLR n = 10, and P values are as follows: miR-128 ISH: 0.031, miR-128 qPCR: 0.044; miR-484 ISH: 0.009, miR-484 qPCR: 0.026. *P < 0.05, error bars ± SE. B: AS and MM probe sequences with respective FED for ISH experiments. No difference in neuroanatomical expression detected per miRNA or across experimental groups.

Select bHR:bLR miRNA Expression Differences are Associated With Experimentally Validated miRNA-mRNA Relationships, but Majority of Assessed mRNAs Associated With These miRNAs are not Differentially Expressed by Genotype

To examine the pathways potentially modified by the differential expression of specific miRNAs between bHR animals and bLR animals, we utilized IPA and a review of the literature to identify mRNAs reported as targets of miRNAs differentially expressed by genotype (Table 1). We selected a subset of these miRNA:mRNA relationships for further analysis (miR-29b:Dnmt3a, miR-29b/miR-93:Pten, miR-192/miR-194:Mdm2, miR-93:Stat3, miR-26b:Bdnf). This subset of miRNAs were differentially expressed between bHR and bLR animals in different brain regions; miR-29b in NAcS, miR-93 in NAcS prior to exclusion of sample outliers, miR-192 in PL/NAcC/NAcS, miR-194 in PL, and miR-26b in PL. To determine if these mRNA targets were differentially expressed by genotype in a manner consistent with miRNA regulation, we performed mRNA qPCR assays in brain regions that exhibited differential expression of their associated miRNAs (NAcS: Dnmt3a, Pten, Stat3; PL: Bdnf, Pten, Mdm2). mRNA qPCR experiments employed β-actin as a housekeeping gene; no difference was detected in β-actin mRNA C_t values between bHR-bLR animals in the NAcS (P value = 0.123). One mRNA, Dnmt3a, was expressed at a higher level in the NAcS of bHR animals compared with bLR animals (Fig. 9). However, the other mRNAs measured (Pten, Mdm2, Bdnf, and Stat3) were not differentially expressed in their corresponding brain regions by genotype (Fig. 9).

Table 1.

mRNAs targeted by miRNAs differentially expressed by bHR:bLR genotype

Gene Symbol	Definition	Pathway Related	miRNA	Ref. List Nos.
PTEN	phosphatase and tensin homolog	AKT cell proliferative/survival	miR-26b and miR-93	19, 38, 40, 47
STAT3	signal transducer and activator of transcription 3	AKT cell proliferative/survival	mir-93	3, 13
MDM2	murine double minute 2 homolog	p53	miR-192 and miR-194	21, 45
DNMT3A	DNA methyl transferase 3A	epigenetic process + spine plasticity	miR-29b	27, 52
DNMT3B	DNA methyl transferase 3B	epigenetic process	miR-29b	52
SOX2	sex determining region Y-box 2	transcription factor/neuronal stem cells	miR-340	8
BDNF	Brain-derived neurotrophic factor	neurotrophic/dopamine signaling	miR-26b	5, 49
BAG2	BCL2-associated athanogene 2	AKT cell proliferative/survival	miR-128	33, 38
FIS1	mitochondrial fission 1 protein	mitochondrial fission	miR-484	54
E2F3a	E2F transcription factor 3a	transcription	miR-128	57
BMI1	BMI1 polycomb ring finger oncogene	stem cell renewal/oncogene	miR-128	14

miRNA, microRNA; bHR, bred high responder; bLR, bred low responder; miR, microRNA.

DISCUSSION

The investigation of mental health disorders is confounded by the complex mixture of genetic and environmental factors as well as the relatively uncharacterized mechanisms of disease pathology and individual variation in vulnerability and sensitivity. Previous studies have established altered drug self-administration, anxiety, stress reactivity, and overall emotionality as relevant behavioral links to a broad range of mental health diseases (10). To understand these individual differences that persist throughout many mental health disorders, we have employed the bHR-bLR animal model to gain a molecular perspective on its unique temperamental and behavioral divergence. As potent posttranscriptional regulators of gene expression, miRNAs present a unique target of study that has recently come under scrutiny in the mechanisms of numerous central nervous system disorders (e.g., Alzheimer's disease and autism spectrum disorder) (9, 32). In this study, we present evidence for extensive differential expression of miRNAs by brain region and between bHR animals and bLR animals in multiple brain regions associated with dysfunction in reward and emotional processing as well as mental health disorders.

Our approach using LCM and qPCR-based arrays enabled us to obtain precise neuroanatomical resolution of miRNA expression patterns in the brain not achievable by other dissection methods. A key advantage of the qPCR-based analysis method is that multiple miRNAs can be efficiently and sensitively monitored. In the combination of miRNA expression data regardless of genotype, we identified a substantial number of miRNAs that were differentially expressed by brain region (Fig. 4). These data provide clear evidence for miRNA enrichment by brain region, which may mean that miRNA regulation of gene expression varies by brain region under both basal and experimental conditions. In terms of relative miRNA expression, these data present general trends in which 1) the PL exhibits higher expression of miRNAs compared with the NAc and 2) within the NAc, the NAcS exhibits higher expression of miRNAs compared with the NAcC.

In the examination of our original hypothesis that miRNAs may contribute to the biology of the bHR-bLR model, we identified numerous miRNAs differentially expressed in the mesocorticolimbic circuit between bHR and bLR animals (Fig. 3). Among the miRNAs differentially expressed by genotype, most were altered in a single brain region, suggesting a neuroanatomical specificity to the observed differential miRNA expression. Of the three brain regions examined, the PL demonstrated the greatest number of miRNAs differentially expressed by genotype, and there was also differential miRNA expression by genotype between the two subregions of the nucleus accumbens. Between genotypes, there was a unidirectional trend for higher miRNA expression in bLR animals compared with bHR animals. One potential mechanism for the unidirectional miRNA differences observed could be additional epigenetic factors, such as DNA methylation or acetylation. Of note, our analysis of one specific DNA methyltransferase seems consistent with this possibility, although more comprehensive studies will be required to fully examine this issue. In terms of the animal model, these data translate to lower expression of select miRNAs in the group associated with increased vulnerability to behaviors associated with externalizing disorders (e.g., drug abuse).

To validate our miRNA qPCR array findings, we applied individual assay technology for an enhanced sensitivity and stringency to the data analysis. Remarkably, of the miRNAs measured by individual qPCR assay, a substantial proportion of comparisons between brain regions showed significant differential expression (36/42, Fig. 7), and over half of the instances of differential miRNA expression between genotypes were validated (15/29, Fig. 5). The extensive agreement between these data provides strong support of the brain-region enrichment and genotype-dependent miRNA expression patterns identified by qPCR arrays. Some discrepancy in FED magnitude between the TLDA, individual assay, and ISH was observed, but these differences likely result from the differing sensitivities of the technologies employed. Overall, directionality of expression changes was extremely consistent.

To examine the spatial patterns of differential miRNA expression in the mesocorticolimbic neural circuitry, we examined miRNA expression differences validated by qPCR using ISH. The ISH signal of miR-128a, miR-29b, and miR-484 show robust and ubiquitous expression of these miRNAs across many areas of the brain including those involved in the mesocorticolimbic reward pathway (representative images of these probes are shown in Fig. 2). Additionally, ISH analysis of the expression levels of miR-128 and miR-484 demonstrated a significantly lower signal in bHR compared with bLR animals. The sensitivity of the ISH procedure was ultimately a limiting factor as a number of miRNAs adequately detected by qPCR (i.e., TLDA C_t values >28) yielded undetectable ISH autoradiographic signals in the brain regions of interest. Together with the qPCR assays, these data support the hypothesis that miRNAs may contribute to the biological underpinnings of the bHR-bLR model in a brain region-dependent manner.

With extensive evidence for differential miRNA expression between bHR-bLR rats in the mesocorticolimbic reward pathway, we sought to determine which mRNAs may be impacted by these miRNA differences and contribute to the phenotype. Using bioinformatics analyses, we identified a number of experimentally determined interactions between these differentially expressed miRNAs and mRNA targets previously linked to central nervous system functions (Table 1). Of note, miR-128a and miR-26b have been linked to neurotrophic factor processes in the brain through regulation of the mRNAs for Ntrk3 and Bdnf, respectively (4, 30). These associations are potentially interesting given the important role that is emerging for both Ntrk3 and Bdnf in anxiety disorders and drug abuse biology (27, 42). Furthermore, miR-598 expression has been linked to antidepressant treatment (fluoxetine and electroconvulsive therapy) in the hippocampus (35). Among the seven mRNAs assessed by qPCR, we identified a higher level of expression of Dnmt3a mRNA in the NAcS of bHR animals compared with their bLR counterparts. Together with lower expression of the associated miR-29b, these data are consistent with canonical miRNA regulation of Dnmt3a by miR-29b in the NAcS. Given the role of Dnmt3a in emotional processing and spine plasticity specific to the NAc, these data suggest a putative link between miRNA regulation (miR-29b) and reward and emotional processing (3, 16, 24). The numerous miRNA differentially expressed between the bHR and bLR rats and their downstream mRNA targets identifies multiple putative pathways that could be linked to the altered experience of novelty and antecedent vulnerability to mental health diseases. Our research focused on mRNAs and not the resulting proteins due to the anatomical specificity of our dissection methods (e.g., LCM) limiting the tissue collection mass to a suboptimal level for protein analysis. Hence, while this manuscript does not include protein analyses, prior literature does reveal that alterations in the microRNAs like miR-29b do alter the levels of the protein (Dnmt3a), albeit in other model systems (11).

Interestingly, the remaining mRNA qPCR experiments did not demonstrate differential expression as predicted (Fig. 9). One possible explanation for these data is that miRNAs do not always act through degradation of mRNA targets but can also act through inhibition of translation (40). This observation should be considered when examining our mRNA qPCR data as a lack of differential expression may not rule out the possibility for altered protein levels. This complexity has been observed in recent work demonstrating alterations of BDNF protein levels but not mRNA levels in response to cocaine withdrawal as well the role of the let-7 miRNA in translation initiation blockade with minimal mRNA degradation (27, 40). Future experiments will need to investigate protein levels as a possible marker of miRNA posttranscriptional inhibition in these brain regions in both bHR and bLR animals.

Although the potential implications of these findings are interesting, we must note that these data do not demonstrate a causative role for miRNAs in the phenotypic differences characterized in the bHR/bLR animals. However, our data does highlight novel miRNA expression differences in this animal model that can be used as a foundation on which to build a putative mechanism for the phenotypic observations. Additional studies will be needed to more directly identify the role of miRNA and their interaction with mRNAs, proteins, and epigenetic interactions relevant to the molecular mechanisms that have been implicated in the bHR-bLR animal model. Individual differences in sensitivity to mental health diseases could also be investigated through exposure of animal models to environmental stimuli and monitoring the molecular interaction (e.g., drug abuse exposure, antianxiety treatment, and enriched environment). Further investigation into other neuroanatomical regions of reward pathways such as the ventral tegmental area and the hippocampus as well as investigation into cell-type specificity of the miRNA regulation will also be crucial to complete the neuroanatomical investigation of miRNA, novelty-seeking behavior, and the implications on vulnerability to mental health disorders.

Individual differences influence the occurrence and manifestation of many mental health disorders and can be investigated as a broad variable in the mechanisms of these conditions. Variation in the acquisition of drug-self administration, anxiety, and broad-term temperament illuminate the potential clinical relevance of the bHR-bLR model to many mental health diseases. Through identification of numerous miRNAs and an mRNA that are differentially expressed between genotype and brain region in the bHR-bLR model, we have made significant contributions to our understanding of the genetic background to which these individual differences manifest. These data allow us to further the connection between specific miRNAs and the biology of individual variation and mental health disorder liability.

GRANTS

Funds used to support this work come from National Institute on Drug Abuse Grant R01-DA-025973 (R. C. Thompson and S. J. Watson).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: D.E.H. and C.L.C. performed experiments; D.E.H., C.L.C., B.S.C., and R.C.T. analyzed data; D.E.H., C.L.C., B.S.C., and R.C.T. interpreted results of experiments; D.E.H., C.L.C., and B.S.C. prepared figures; D.E.H., C.L.C., B.S.C., and R.C.T. drafted manuscript; D.E.H., C.L.C., B.S.C., H.A., S.J.W., and R.C.T. edited and revised manuscript; D.E.H., C.L.C., B.S.C., H.A., S.J.W., and R.C.T. approved final version of manuscript; H.A., S.J.W., and R.C.T. conception and design of research.