Here we show that a natural perceptual experience, hearing the sound of another bird singing, alters the profile of miRNAs in parts of the songbird brain responsible for auditory perception, integration and memory. The song-regulated population includes both known (conserved) and novel miRNAs. We highlight one sex-linked song-responsive miRNA and identify mRNAs that are potential targets of its action during song exposure. Thus miRNAs may have roles in the information processing functions of the brain, in addition to their roles in brain development and evolution.
To demonstrate this, we first catalogued the miRNAs expressed in the adult zebra finch auditory forebrain. We used massively parallel Illumina sequencing of small RNAs to perform this cataloguing efficiently. In addition to known conserved miRNAs, our analysis identified 48 small RNA sequences that meet the structural criteria for miRNAs but had not been described in miRBase in any organism at the time of our analysis. Fourteen of these are detected in the chicken or human genome assemblies and may give rise to miRNAs that have not yet been described elsewhere due to low copy number, restricted tissue distribution or other factors. The remaining novel miRNAs, 34 in number, may be unique to the zebra finch or the songbird lineage. Few studies have attempted de novo identification of miRNAs from the brain  and ours is the first to report direct sequencing of songbird brain miRNAs. A previous study did identify precursor sequences for five conserved miRNAs in the developing zebra finch brain . Also, in parallel with our own Illumina analysis, Li and her colleagues used 454 sequencing to identify miRNAs in the brain and liver of adult zebra finches. These different sets of annotations are compared and collated in a supplement to the analysis of the zebra finch genome assembly .
By comparing birds hearing novel song playbacks or silence, we found evidence for experience-dependent fluctuations in large numbers of miRNAs in the auditory forebrain. We performed three separate pairwise comparisons by Illumina, where all aspects of the experimental conditions were carefully counterbalanced between the two groups in each comparison. The three comparisons were not direct replications of each other, as each had a different sex ratio. Our reasons for varying the sex ratio were partly pragmatic (limited numbers of birds of the same sex that could be removed from our aviary) and partly analytical (males and females have different behavioral responses to songs). Some of the differences between the three sets of results may reflect real biological differences in the responses of males and females. Indeed, our Northern analysis of the tgu-miR-2954-3p confirms a sex difference in expression of this Z-linked miRNA gene. This is especially intriguing because we also obtained TaqMan evidence for both sense and antisense transcripts of this miRNA. One can imagine scenarios where different ratios of sense and antisense transcription occur in males (two copies of the gene) and females (one copy of the gene) with different consequences on the transcriptional networks affected by song exposure in the two sexes.
Ignoring the potential effects of sex, we identified five miRNAs that showed significant and consistent changes in response to song across all three Illumina comparisons. Three miRNAs consistently decreased after song (tgu-miR-92, tgu-miR-124, tgu-miR-129-5p) and two increased (tgu-miR-25, tgu-miR-192). The down-regulated miRNAs are at much higher abundance (> 1000 reads in each run) and perhaps for this reason we were more successful at detecting them and replicating their song regulation by TaqMan assay in subsequent experiments with additional groups of birds. The most abundant miRNA in our regulated set, tgu-miR-124, consistently met the statistical test for significant down-regulation by song, in each of six separate experiments (three Illumina comparisons, two TaqMan analyses in Additional File 1, Table S4, and the TaqMan comparison of song vs. SEN in Additional File 2, Figure 3).
In studies in other species, miR-124 has been linked to brain plasticity and development in several contexts. Chronic cocaine administration results in down-regulation of miR-124 in the rodent mesolimbic dopaminergic system . In the developing chick neural tube, miR-124a is a component of a regulatory network that controls the transition between neural progenitors and post-mitotic neurons . miR-124 also regulates adult neurogenesis, and its overexpression promotes neuronal differentiation [42, 43] and neurite outgrowth . Intriguingly, in songbirds neurogenesis continues in the forebrain throughout adulthood, from a population of precursor cells that line the walls of the lateral ventricles and have the characteristics of neural stem cells [45–47]. The net rate of neuronal addition and loss in the adult songbird has been shown to depend on social and environmental influences [48–51]. Perhaps tgu-miR-124 is a regulatory link between experience and neurogenesis - further study of this fascinating possibility is clearly warranted.
Although miRNAs can have diverse functions, they often act by altering the concentrations of specific mRNAs they target via complementary base pairing. We used the TargetScan algorithm  to predict binding sites of tgu-miR-2954-3p in chicken genes, and then we confirmed the presence of the same conserved target sequence in the zebra finch genome assembly. We found eight targets that met these criteria and were also regulated by song in the Dong et al. microarray data . These eight genes have a provocative coherence in their function, as they are all implicated in control of cell proliferation and neuronal differentiation. Six operate by affecting gene expression and chromatin remodeling as we briefly review here. ELAVL2 is a member of a protein family that binds AU-rich regions in the 3'UTR of genes such as c-fos and promotes the shift from cell proliferation into cellular differentiation [53–57]. TLK2 is a kinase tightly associated with DNA replication during cell division . At least one of its targets, the histone chaperone Asf1, controls chromatin assembly, thus TLK2 activity can regulate transcription and elongation [59–61]. BTG1 is also regulated during the cell cycle . It acts as a cofactor for Hoxb9, a transcription factor that controls cell proliferation and differentiation, and BTG1 reduces rates of cell proliferation [62–64]. CHD2 can potentially affect transcription of many genes by remodeling chromatin [65, 66]; disruption of CHD2 has profound consequences for development and is implicated in many human diseases [67–69]. HMGB1 is another DNA binding protein that facilitates transcription by altering chromatin structure to ease promoter binding [70–73]. Some of the genes regulated by HMGB1 may play a role in cell proliferation and migration [74, 75]. Neuronal migration and neurite outgrowth are affected by CRKL, a transcriptional activator that is a component of the reelin pathway [76–79]. Unlike the other six genes, NEGR1 and LINGO2 do not seem to alter transcription but they do have established roles in neuronal differentiation. NEGR1 affects cell-cell adhesion to modulate neurite outgrowth and synapse formation [80–82]. LINGO2 is one member of a family of transmembrane proteins that are involved in neural and axonal regeneration [83, 84]. The function of LINGO2 is untested, but expression of a related protein, LINGO1, is attenuated in cortical areas deprived of sensory input and is a partner in a signaling pathway that correlates with neuronal activity during a learning paradigm [85, 86].