Genome-Wide Identification of the Castanea mollissima MADS-Box Gene Family and Its Potential Roles in Female Flower Development Based on Transcriptomic and Morphological Analyses

Abstract

Castanea mollissima is an important woody food crop in China. Late, sparse, and unstable female flower differentiation constrains the development of the chestnut industry. Here, two more-female cultivars, ‘Longquanzhuang No. 2’ (LQ) and ‘Yanqiu’ (YQ), and the control cultivar ‘Yanshanzaofeng’ (ZF) were sampled at five stages after budbreak (7–35 d) for morphological observation and transcriptome sequencing. Morphological divergence became evident at 21 d after budbreak. Transcriptome analysis identified 2223 differentially expressed genes specific to the LQ–ZF comparison at 21 d and 791 inter-cultivar differentially expressed genes shared across 21–35 d. GO enrichment analysis showed significant enrichment in transcription factor activity and related pathways. Transcription factor prediction identified MADS-box genes as candidates, and 60 MADS-box members were identified in Chinese chestnut. These comprised 29 Type I members and 31 Type II members, with the latter classified into 13 subfamilies. Integrated analysis of gene-expression patterns and predicted protein interactions showed that more Type II genes were differentially expressed in female flowers and that expression patterns differed between the more-female cultivars and the control. The contrasting expression patterns of putative positive and negative CmMADS regulators may be associated with differences in female flower developmental capacity among these cultivars. RT-qPCR analysis of four representative CmMADS genes showed that genes expressed at higher levels in female flowers generally increased during development, consistent with the transcriptome data. These findings reveal morphological divergence and dynamic CmMADS expression during female flower development in two more-female chestnut cultivars and a control cultivar, providing a basis for investigating the underlying molecular regulatory network, but functional validation is currently limited to transcriptome-level prediction, requiring further experimental verification.