Transcriptome profiling of maize transcription factor mutants to probe gene regulatory network predictions

TL;DR

This study tests gene regulatory network predictions in maize by analyzing transcription factor mutants and their effects on gene expression and metabolism.

Contribution

The paper experimentally validates gene regulatory network predictions using maize transcription factor mutants and transcriptomic/metabolomic profiling.

Findings

Loss-of-function mutants for 22 maize transcription factors showed altered gene expression but no obvious morphological changes.

Only a small subset of predicted gene targets from yeast-1-hybrid screens showed altered expression in mutants.

Gene co-expression network predictions had limited success in identifying transcription factor targets with altered expression.

Abstract

Transcription factors play important roles in regulation of gene expression and phenotype. A variety of approaches have been utilized to develop gene regulatory networks to predict the regulatory targets for each transcription factor, such as yeast-1-hybrid screens and gene co-expression network analysis. Here we identified potential transcription factor targets and used a reverse genetics approach to test the predictions of several gene regulatory networks in maize. Loss-of-function mutant alleles were isolated for 22 maize transcription factors. These mutants did not exhibit obvious morphological phenotypes. However, transcriptomic profiling identified differentially expressed genes in each of the mutant genotypes, and targeted metabolic profiling indicated variable phenolic accumulation in some mutants. An analysis of expression levels for predicted target genes based on yeast-1-hybrid screens identified a small subset of predicted targets that exhibit altered expression levels. The analysis of predicted targets from gene co-expression network-based methods found significant enrichments for prediction sets of some transcription factors, but most predicted targets did not exhibit altered expression. This could result from false-positive gene co-expression network predictions, a transcription factor with a secondary regulatory role resulting in minor effects on gene regulation, or redundant gene regulation by other transcription factors. Collectively, these findings suggest that loss-of-function for single uncharacterized transcription factors might have limited phenotypic impacts but can reveal subsets of gene regulatory network predicted targets with altered expression.

Loss-of-function mutant alleles were characterized for 22 maize transcription factors to test the functional relevance of prior gene regulatory network predictions. The transcriptome and phenolic metabolome of these mutants were profiled to document the functional roles of these transcription factors. A subset of the prior gene regulatory network predictions were supported but had limited overall power in predicting changes in gene expression or phenolic compounds. The limitations of testing transcription factor-target predictions with single TF knockouts are discussed.

Figures