Extensive, transient, and long-lasting gene regulation in a song-controlling brain area during testosterone-induced song development in adult female canaries

Seasonal song production in canaries, influenced by gonadal hormones, is a well-documented phenomenon. We explored testosterone-induced song development in adult female canaries—a behavior rarely exhibited naturally. Gene regulatory networks in the song-controlling brain area HVC were compared at multiple time points (1 hour to 14 days) post-treatment with those of placebo-treated controls, paralleling HVC and song development. Females began vocalizing within four days of testosterone treatment, with song complexity and HVC volume increasing progressively over two weeks. Rapid transcriptional changes involving 2,739 genes preceded song initiation. Over two weeks, 9,913 genes—approximately 64% of the canary's protein-coding genome—were differentially expressed, with 98% being transiently regulated. These genes are linked to various biological functions, with early changes at the cellular level and later changes affecting the nervous system level after prolonged hormone exposure. Our findings suggest that testosterone-induced song development is accompanied by extensive and dynamic transcriptional changes in the HVC, implicating widespread neuronal involvement. The data reveal extensive transcriptomic changes, including alterations in steroid receptor expression and numerous transcription factors, coinciding with significant neural transformations. These changes underpin the gradual emergence of singing behavior, providing insights into the neural basis of seasonal behavioral patterns.

Methods
We compared the expression data of song nuclei HVC dissected from adult female canaries (Serinus canaria) implanted with 7-mm SilasticTM tubes filled with testosterone for 6 time durations (1h, 3h, 8h, 3d, 7d, and 14d) with control birds (implanted with empty tube) to identify testosterone effects on gene expression.

We performed microarray analysis using the Custom Affymetrix Gene Chip® MPIO-ZF1s520811 Exon Array. The CEL files were generated using Affymetrix GeneChip Command Console® Software (AGCC). Differential expression analyses were performed using the R package limma, following the workflow of Klaus and Reisenauer (F1000Research 2016).​

The HVC volumes, plasma androgen levels, and physiological measurements of testosterone-treated and control birds were also deposited. 

The song parameters of birds in T7d and T14d Multi_Channel_Analyser (MCA), a custom program with a graphical user interface written in MATLAB (version R2016b, Mathworks), as described previously (Ko et al., 2020) and available to the public (https://doi.org/10.5281/zenodo.1489098).