A cross-species transcriptomics approach to identify genes involved in leaf development

Abstract

Background: We have made use of publicly available gene expression data to identify transcription factors and transcriptional modules (regulons) associated with leaf development in Populus. Different tissue types were compared to identify genes informative in the discrimination of leaf and non-leaf tissues. Transcriptional modules within this set of genes were identified in a much wider set of microarray data collected from leaves in a number of developmental, biotic, abiotic and transgenic experiments.

Results: Transcription factors that were over represented in leaf EST libraries and that were useful for discriminating leaves from other tissues were identified, revealing that the C2C2-YABBY, CCAAT-HAP3 and 5, MYB, and ZF-HD families are particularly important in leaves. The expression of transcriptional modules and transcription factors was examined across a number of experiments to select those that were particularly active during the early stages of leaf development. Two transcription factors were found to collocate to previously published Quantitative Trait Loci (QTL) for leaf length. We also found that miRNA family 396 may be important in the control of leaf development, with three members of the family collocating with clusters of leaf development QTL.

Conclusion: This work provides a set of candidate genes involved in the control and processes of leaf development. This resource can be used for a wide variety of purposes such as informing the selection of candidate genes for association mapping or for the selection of targets for reverse genetics studies to further understanding of the genetic control of leaf size and shape.

Figure 1

Principal component analysis overview of gene expression in different tissues of Populus tremula × tremuloides 'T89. PCA overview of the expression of the 14446 genes (A) and of the only the 955 transcription factors (B) represented on the POP2 microarray in different tissue types of Populus tremula × tremuloides 'T89'.

Figure 2

Digital northern heat map representation of leaf genes in PopulusDB EST libraries. Heat map representation of a digital northern to examine the library distribution of the identified leaf gene set. The results are based on the data contained in PopulusDB [53] and genes are arranged on the basis of cluster analysis to aid visual interpretation.

Figure 3

Gene ontology category over-representation in leaf genes. Statistically over-represented Gene ontology categories within the leaf gene set. Statistical significance was calculated using an hypergeometric test with an FDR-adjusted cut off of 0.05. The size of each circle indicates the number of genes within a category and circles are shaded based on significance. Categories in white were non-significant.

Figure 4

Transcriptional module identification and network visualisation in the UPSC-Leaf microarrays. A Average linkage hierarchical clustering dendrogram of the UPSC-Leaf dataset. Five transcriptional modules were detected. Modules were assigned colours as indicated by the bar below the dendrogram. Rows and columns in the heat map represent genes in a symmetric fashion. Colour intensity represents connection strength between two genes from red (the strongest connection) to pale yellow (no connection). B Network visualisation based on pair-wise Pearson correlations. Nodes represent genes and are connected by edges where significant correlation between genes exists. Nodes are coloured based on the five modules detected in A. Edges are drawn for correlations > 0.7. Unconnected nodes are not plotted and nodes with only individual pair-wise connections that were not connected to the central network were removed.

Figure 5

Cross-experiment validation of leaf development candidate transcription factors and modules. A-D Expression of selected transcription factors with developmental expression patterns in four Populus microarray experiments (Seasonal data of [31], UPSC-TC experiment, UPSC-LP experiment, Pt-TC experiment) and of the best BLAST hit ortholog in the Arabidopsis AtGenExpress developmental baseline experiment [56]. The genes are gw1.VII.2982.1, estExt_Genewise1_v1.C_290455, gw1.XII.82.1, gw1.41.334.1 respectively. E-F Expression of the most highly-connected gene for transcriptional modules with a developmental expression trend identified by Topological Overlap Matrix weighted co-expression analysis. Genes are coloured based on their module colour in Figure 4. Blue, estExt_fgenesh4_pm.C_LG_V0721; Red, estExt_fgenesh4_pg.C_LG_XII1021; Green, estExt_fgenesh4_kg.C_LG_X0057; Turquoise, estExt_fgenesh4_pg.C_440087.

Figure 6

Candidate gene collocation to QTL for leaf development. Example extended application of identified leaf candidate genes. A – Collocation of genes that were highly helpful for the discrimination of leaf tissue from other tissue types to QTL for leaf length in the control condition from the results of [9]. Linkage groups are shown as grey bars and chromosomes as blue bars. SSR markers are indicated by black text and AFLP markers by grey text. Dashed grey lines connect the location of SSR markers in the linkage group and chromosome. QTL are plotted to the left of the linkage group with a horizontal line showing the peak F score location and confidence intervals plotted as verticals lines. The location of genes is indicated by horizontal lines on the chromosome with transcription factors represented by red lines and other genes by green lines. B – Expression of the two transcription factors located between flanking markers from A in four Populus microarray experiments (Seasonal data of [31], UPSC-TC experiment, UPSC-LP experiment, Pt-TC experiment) and of the best BLAST hit ortholog in the Arabidopsis AtGenExpress developmental baseline experiment [56]. The transcription factor from LG III (grail3.0018017701, AT2G45190) is plotted in black and that from LG V (gw1.V.4670.1, AT5G65410) in red. C – Chromosomal location of genes (green vertical lines) and transcription factors (red vertical lines) that were informative in the discrimination of leaves from other tissues identified using Orthogonal Projection to Latent Sqaures.