PhDEF has a much stronger molecular footprint in the petal epidermis than in the mesophyll, according to a study that combined single-cell gene-expression data with DNA-binding measurements. The pattern suggests that the same regulator can operate differently depending on the cell layer, although the experiments do not show that every gene touched by PhDEF is directly regulated by it.
The single-cell analysis examined 11,632 wild-type petal cells, 3,875 star cells and 3,737 wico cells. The researchers then compared PhDEF mutant tissue with corresponding wild-type tissue in each layer.
The two layers carried different molecular signatures
The expression data separated epidermal and mesophyll cells into distinct groups, indicating different layer-level transcriptomic signatures. That separation allowed the researchers to compare gene-expression changes by layer.
The epidermis also had the larger set of genes that were more specifically active there. Using the study's thresholds, 695 genes were classified as epidermis-enriched, compared with 151 genes enriched in the mesophyll. The thresholds required a base-2 expression fold-change above 0.75 and a layer-specificity difference above 10 percent, a measure based on the share of cells in each layer expressing a gene.
The mutant-versus-wild-type comparisons identified 628 expression changes found only in the epidermis, compared with 364 found only in the mesophyll, while 286 changes appeared in both layers. The authors classified 75 percent of the changes as activated and 25 percent as repressed under their analysis.
DNA binding was concentrated in the epidermal landscape
The researchers then mapped PhDEF-associated DNA regions using ChIP-seq, a method that captures DNA regions associated with a chosen protein. The assay used wild-type, star and wico petals at developmental stage 8. Each genotype was tested in four ChIP experiments, and the two strongest replicates were sequenced.
After filtering for reproducibility and removing peaks attributed to input chromatin, the analysis found 3,134 PhDEF binding peaks in wild type, 4,276 in wico and 2,216 in star. A peak is a stretch of DNA where the assay detects an enriched binding signal.
When the peaks were linked to genes and grouped by layer, the epidermis-associated category was far larger: 2,041 peaks linked to 1,896 genes, compared with 407 peaks linked to 398 genes in the mesophyll-associated category. Another 1,176 peaks, linked to 1,110 genes, were shared between the layers. The study therefore found a broad epidermal bias in PhDEF-associated binding, not a complete separation between the two tissues.
Most genes fell into a single binding-profile class. The analysis counted 1,694 genes with epidermis-only sites, 977 with shared-only sites and 325 with mesophyll-only sites. Mixed profiles were less common, with 153 genes carrying both epidermis-specific and shared sites and 23 carrying both mesophyll-specific and shared sites.
The pattern is suggestive, not a complete mechanism
The link between binding and changed expression was strongest in one specific comparison rather than universal. Binding-profile distributions were broadly similar across the differentially expressed gene groups, but epidermis-specific expression changes were enriched for genes with both epidermal and shared PhDEF binding sites. The reported one-sided test gave a p-value of 0.033, a statistical measure used here to describe that enrichment. This association does not establish direct regulation.
Sequence patterns under the binding sites offered clues about what might shape differences between layers. CArG-box motifs were enriched under gene-associated PhDEF peaks in wild type, wico and star. Predicted TCP binding sites occurred in 25.2 percent of epidermis-specific peaks and 25.0 percent of shared peaks, compared with 11.8 percent of mesophyll-specific peaks. The motif findings are suggestive, but they do not identify the exact bound protein or prove a functional PhDEF-TCP partnership.
The study's own caveats narrow the interpretation. The comparison of enriched genes depends on its chosen expression thresholds and on sampling mature petals. Some cell clusters could not be assigned to precise spatial locations, and possible misclassification of star epidermal cells could have led to an underestimate of epidermis-specific differentially expressed genes.
Binding occupancy alone does not establish transcriptional regulation, and the enrichment of epidermis-plus-shared sites is an association rather than proof of direct control. The motif analysis also does not prove that PhDEF physically partners with a TCP protein. Those questions require direct tests of regulation and protein interaction.
A map for further testing
Taken together, the results support a layer-dependent association for PhDEF in petals: the epidermis had more enriched genes, more layer-exclusive expression changes and many more gene-associated binding sites. They point to a molecular difference between the two layers, but not to a complete causal mechanism.
The paper states that its data and code were publicly available at publication, with bulk RNA-seq, single-cell RNA-seq and ChIP-seq datasets listed under GEO accessions GSE336494, GSE290697 and GSE308255. Those resources could help test whether the binding and expression patterns hold across more developmental stages and biological replicates.
Paper data and sources
Original title: The petal identity gene PhDEF has a major binding and regulatory action in the epidermis
Authors: Désert E, Cavallini-Speisser Q, Duvernois-Berthet E et al.
Journal/Repository: eLife
Status: Peer-reviewed
First online: 2026-10-19
DOI: Not available
Original paper · Full text