Inhibition of miR-128 Enhances Vocal Sequence Organization in Juvenile Songbirds

Abstract

The molecular mechanisms underlying learned vocal communication are not well characterized. This is a major barrier for developing treatments for conditions affecting social communication, such as autism spectrum disorder (ASD). Our group previously generated an activity-dependent gene expression network in the striatopallidal song control nucleus, Area X, in adult zebra finches to identify master regulators of learned vocal behavior. This dataset revealed that the two host genes for microRNA-128, ARPP21 and R3HDM1, are among the top genes whose expression correlates to how much birds sing. Here we examined whether miR-128 itself is behaviorally regulated in Area X and found that its levels decline with singing. We hypothesized that reducing miR-128 during the critical period for vocal plasticity would enhance vocal learning. To test this, we bilaterally injected an antisense miR-128 construct (AS miR-128) or a control scrambled sequence into Area X at post-hatch day 30 (30 d) using sibling-matched experimental and control pupils. The juveniles were then returned to their home cage and raised with their tutors. Strikingly, inhibition of miR-128 in young birds enhanced the organization of learned vocal sequences. Tutor and pupil stereotypy scores were positively correlated, though the correlation was stronger between tutors and control pupils compared to tutors and AS miR-128 pupils. This difference was driven by AS miR-128 pupils achieving higher stereotypy scores despite their tutors' lower syntax scores. AS miR-128 birds with tutors on the higher end of the stereotypy spectrum were more likely to produce songs with faster tempos relative to sibling controls. Our results suggest that low levels of miR-128 facilitate vocal sequence stereotypy. By analogy, reducing miR-128 could enhance the capacity to learn to speak in patients with non-verbal ASD. To our knowledge, this study is the first to directly link miR-128 to learned vocal communication and provides support for miR-128 as a potential therapeutic target for ASD.

Keywords: learning; memory; microRNA; neurodevelopment; songbird; stereotypy.

FIGURE 1

(A) Predicted miR-128 targets were analyzed for pathway enrichment and then clustered by similarity to generate an enrichment map. Cluster annotations are summarized with adjacent labels in black. (B) miR-128 targets are enriched in ASD-related gene datasets. Heat map shows gene enrichment analysis for miR-128 target genes, ARPP21 target genes, or both [Overlap] in autism-related gene lists: all autism risk genes from the Simons Foundation Autism Research Initiative (SFARI) database [SFARI (total)], SFARI genes expressed in Area X [SFARI (Area X)], CHD8 targets at enhancers [CHD8- enhancers] and promoters [CHD8- promoters], Fragile-X mental retardation protein target genes [FMRP targets]. Neural genes differentially expressed in Down syndrome [DS DEGs] and all genes expressed in the adult zebra finch basal ganglia [All ZF BG expressed genes] are included as negative controls. FDR corrected p-value indicated by *p < 5.00E–02, **p < 5.00E–12. (C) Activity-dependent regulation of miR-128 in adult zebra finch Area X as measured by qPCR. miR-128 undergoes activity-dependent regulation in the vocal communication region of the basal ganglia in adult songbirds (R² = 0.695, p = 0.01). miR-128 levels were normalized to loading control U6 snoRNA.

FIGURE 2

(A) Timeline of experimental procedures. Developmental critical periods are shown in top two boxes. (B) Region of interest (Area X) and miR-128 antisense viral construct. (Left) Diagram showing viral targeting of Area X. (Right) Schematic characterizing the antisense construct used to inhibit miR-128. Adeno-associated virus 1 vector (AAV1) with inverted terminal repeats to maintain construct integrity (ITR). The type III RNA polymerase III promoter U6 was used to drive expression of the miR-128 siRNA sponge. A proprietary supercore promotor designed by Rachael Neve was used to drive GFP. (C) Exemplar motifs from an adult male (Tutor), a juvenile son who received the miR-128 antisense construct [Pupil (miR-128 AS)], and another juvenile son from the same clutch who received the scramble control construct [Pupil (Scramble)]. (D) Analysis of song sequence organization in 75 d sibling-matched miR-128 AS and scramble control birds (N = 10/group), *p = 0.014, Wilcoxon signed-rank test. Symbol colors denote pupils that shared the same tutor. (E) Tutor and control pupil syntax stereotypy scores are positively correlated (R² = 0.49, *p = 0.025), whereas the correlation between tutors and AS miR-128 pupils does not reach significance (R² = 0.37, p = 0.061).

FIGURE 3

(A) Tempo data divided into two bins based on the syntax score each bird’s tutor. AS miR-128 birds from tutors with higher syntax scores had faster tempos (tempo score < 0.13, Odds Ratio 1.5) relative to AS miR-128 birds from tutors with lower syntax scores (OR 0), or SCR birds from both bins (OR 0.67). (B) Schematic illustrating the non-singing vs. singing (NS-UD) paradigm. On separate days a test subject is either allowed to sing by himself for 4 h (UD-UD) or is prevented from singing for 2 h and then allowed to sing for the subsequent 2 h (NS-UD). Our prior work revealed that song variability is positively correlated to the amount of time singing (Heston and White, 2015; Burkett et al., 2018). (C–E) Coefficient of variation (CV) for spectral features [(C) Amplitude, (D) Entropy, (E) Pitch Goodness] of song following 2 h of singing (UD) or non-singing (NS). The experimental AS miR-128 group is shown in green, and the control SCR group is in black. The average trendline is shown in red. Syllables from five AS miR-128 birds (N = 26) and five sibling-matched control SCR birds(N = 25) were analyzed, *p < 0.05 bootstrap t-test.