Integration of miRNA and protein profiling reveals coordinated neuroadaptations in the alcohol-dependent mouse brain - PubMed (original) (raw)

Integration of miRNA and protein profiling reveals coordinated neuroadaptations in the alcohol-dependent mouse brain

Giorgio Gorini et al. PLoS One. 2013.

Erratum in

Abstract

The molecular mechanisms underlying alcohol dependence involve different neurochemical systems and are brain region-dependent. Chronic Intermittent Ethanol (CIE) procedure, combined with a Two-Bottle Choice voluntary drinking paradigm, represents one of the best available animal models for alcohol dependence and relapse drinking. MicroRNAs, master regulators of the cellular transcriptome and proteome, can regulate their targets in a cooperative, combinatorial fashion, ensuring fine tuning and control over a large number of cellular functions. We analyzed cortex and midbrain microRNA expression levels using an integrative approach to combine and relate data to previous protein profiling from the same CIE-subjected samples, and examined the significance of the data in terms of relative contribution to alcohol consumption and dependence. MicroRNA levels were significantly altered in CIE-exposed dependent mice compared with their non-dependent controls. More importantly, our integrative analysis identified modules of coexpressed microRNAs that were highly correlated with CIE effects and predicted target genes encoding differentially expressed proteins. Coexpressed CIE-relevant proteins, in turn, were often negatively correlated with specific microRNA modules. Our results provide evidence that microRNA-orchestrated translational imbalances are driving the behavioral transition from alcohol consumption to dependence. This study represents the first attempt to combine ex vivo microRNA and protein expression on a global scale from the same mammalian brain samples. The integrative systems approach used here will improve our understanding of brain adaptive changes in response to drug abuse and suggests the potential therapeutic use of microRNAs as tools to prevent or compensate multiple neuroadaptations underlying addictive behavior.

PubMed Disclaimer

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Figure 1

Figure 1. Hierarchical clustering of differentially expressed miRNAs from CIE-2BC (green), Air-2BC (yellow), and Naïve (grey) mice.

Top 10 significant differentially expressed miRNAs for each comparison and brain region analyzed are shown. A-C, CTX; D-F, MB. A, D: CIE-2BC vs. Air-2BC; B, E: CIE-2BC vs. Naïve; C, F: Air-2BC vs. Naïve. Rows: individual miRNAs; columns: individual samples. Red within the heatmap represents miRNA up-regulation, and blue within the heatmap represents miRNA down-regulation. Each heat map shown contains 10 miRNAs with significant differential expression (p<5E-02, p<5E-04, p<5E-04, p<5E-03, p<5E-06, p<5E-07, respectively). Refer to Dataset S1 for p-values and fold change for individual miRNAs. The Venn diagrams indicate the number of shared and unique differentially expressed miRNAs among comparisons: between CIE-2BC versus Air-2BC, CIE-2BC versus Naïve, and Air-2BC versus Naïve groups in CTX (G), MB (H), and across the two brain regions (I). Only differences greater than 1.05 fold with p<0.05 (Bayesian two-tailed t-test) on mapped miRNAs eligible for IPA dataset filter are listed in Venn diagrams.

Figure 2

Figure 2. WGCNA analysis of miRNA expression in CTX and MB of mice subjected to CIE paradigm identified distinct modules of coexpressed miRNAs.

A and B show dendrograms produced by average linkage hierarchical clustering. Horizontal color bars represent different coexpression modules. Bar sizes correspond to the number of miRNAs in each module. SoftPower β = 10 (A), 9 (B), minModuleSize = 5, cutHeight = 0.995, deepSplit = 2 (see Methods). Tables C, D show the relative contribution of miRNAs to CIE paradigm, in terms of correlation (Corr.) between the individual top 20 coexpressed miRNAs sorted by their gene significance (GS) for the EoC trait, with relative p-values and rank. Module number and color information are also included. Full lists are reported in Dataset S1. Tables E-H show correlation between modules of coexpressed miRNAs (E, F) or proteins (G, H) and the EoC trait, or the average 2BC ethanol consumption, as well as individual proteins or miRNAs as traits. Modules are named by a number and a color. Protein names are followed by corresponding gel spot number. In correlation columns, blue represents negative and red represents positive correlations, as reported on the legend. Green p-values are <0.05. Part of the data shown in G and H is taken from our previous study .

Figure 3

Figure 3. Integrative networks for CTX and MB.

To highlight molecular mechanisms underlying the transition from alcohol consumption to dependence in the CIE paradigm, we integrated information from miRNA and protein differential expression and coexpression analyses with currently available miRNA target predictions (IPA database), as well as known and predicted PPIs (String database). The resulting networks for CTX (A) and MB (B) were analyzed with several topology-based scoring methods (C). Nodes with highest score for the corresponding network attribute are listed. Blue, down-regulation; red, up-regulation; diamonds, miRNAs; circles, mRNAs (genes encoding for the identified proteins); dashed lines, target predictions; continuous lines, PPIs; dotted lines, coexpression. Refer to Methods for further details.

Figure 4

Figure 4. Provisional representation of miRNA-mRNA modular network derived from miRNA and protein WGCNA and correlation analysis in mouse CTX (A) and MB (B).

Module colors reflect those from WGCNA analysis. Increasing ethanol consumption levels can be described as “0” for the Naïve group, “1” for Air-2BC, and “2” for CIE-2BC (EoC trait, sketched in C). Protein modules (circles) that are more relevant to alcohol actions are shown based on their correlation to the EoC trait. Proteins in these modules showed an increasing up-regulation across experimental groups, parallel to the EoC trait. miRNA modules (diamonds) negatively correlated to the EoC trait are also shown; the miRNAs in these modules showed an increasing down-regulation across experimental groups, inversely proportional to the EoC trait. These miRNAs are also oppositely correlated to the protein modules which are in turn important for the EoC trait. Therefore, miRNAs and proteins listed respond to 2BC drinking and/or CIE with opposite directional changes in their expression levels, and might thus play a crucial role in excessive ethanol drinking associated with dependence. Protein modules related to endocytic pathways and energy metabolism appear to be important for the effects caused by CIE paradigm. Blue and red outline, negative and positive correlation with the EoC trait.

Figure 5

Figure 5. General classification of brain molecular changes underlying the escalation of ethanol consumption associated with dependence.

We propose three overall patterns of change and list examples of molecules specific to each category (in purple). Y-axis represents the level of expression, and x-axis shows the progression of the disease, in terms of patterns of changes and experimental groups (grey, Naïve; yellow, Air-2BC; green, CIE-2BC). Changes associated with escalation of consumption (A) were identified by WGCNA analysis, and are positively (red) or negatively (blue) correlated with the amounts of alcohol consumed (inset); therefore, these modifications are more pronounced in the CIE-2BC than the Air-2BC group. Changes important for the transition to dependence (B) are almost exclusively present in Air-2BC but not CIE-2BC mice, compared to Naïve; The resulting CIE-/Air-2BC ratio from the differential expression analysis (DE) shows down- (blue) or up- (red) regulation, although the changes in Air-2BC follow the opposite direction. Finally, changes associated with the maintenance of alcohol dependence (C) include those molecules regulated in CIE-2BC but not in Air-2BC mice; their expression levels remain unmodified in non-dependent animals, and thus the differential expression ratio follows the same direction of the respective expression levels.

References

    1. Kril JJ, Halliday GM, Svoboda MD, Cartwright H (1997) The cerebral cortex is damaged in chronic alcoholics. Neuroscience 79: 983–998. - PubMed
    1. Matsumoto I (2009) Proteomics approach in the study of the pathophysiology of alcohol-related brain damage. Alcohol Alcohol 44: 171–176 10.1093/alcalc/agn104 - DOI - PubMed
    1. Sutherland GT, Sheedy D, Kril JJ (2013) Using Autopsy Brain Tissue to Study Alcohol-Related Brain Damage in the Genomic Age. Alcohol Clin Exp Res. doi:10.1111/acer.12243. - PMC - PubMed
    1. Adinoff B (2004) Neurobiologic Processes in Drug Reward and Addiction. Harv Rev Psychiatry 12: 305–320 10.1080/10673220490910844 - DOI - PMC - PubMed
    1. Ballard IC, Murty VP, Carter RM, MacInnes JJ, Huettel SA, et al. (2011) Dorsolateral prefrontal cortex drives mesolimbic dopaminergic regions to initiate motivated behavior. J Neurosci 31: 10340–10346 10.1523/JNEUROSCI.0895-11.2011 - DOI - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources