Deep Sequencing Reveals Differences in the Transcriptional Landscapes of Fibers from Two Cultivated Species of Cotton

Jean-Marc Lacape,Marcelo F Carazzolle,Martial Pré,Danny Llewellyn,Manuel Ruiz,Stéphanie Huguet,Ramon O Vidal,Yves Al-Ghazi,Claire Lanaud,Alexis Dereeper,John Jacobs,Marc Giband,Michel Claverie,Gonçalo A Guimarães Pereira

doi:10.1371/journal.pone.0048855

Abstract

Cotton (Gossypium) fiber is the most prevalent natural product used in the textile industry. The two major cultivated species, G. hirsutum (Gh) and G. barbadense (Gb), are allotetraploids with contrasting fiber quality properties. To better understand the molecular basis for their fiber differences, EST pyrosequencing was used to document the fiber transcriptomes at two key development stages, 10 days post anthesis (dpa), representing the peak of fiber elongation, and 22 dpa, representing the transition to secondary cell wall synthesis. The 617,000 high quality reads (89% of the total 692,000 reads) from 4 libraries were assembled into 46,072 unigenes, comprising 38,297 contigs and 7,775 singletons. Functional annotation of the unigenes together with comparative digital gene expression (DGE) revealed a diverse set of functions and processes that were partly linked to specific fiber stages. Globally, 2,770 contigs (7%) showed differential expression (>2-fold) between 10 and 22 dpa (irrespective of genotype), with 70% more highly expressed at 10 dpa, while 2,248 (6%) were differentially expressed between the genotypes (irrespective of stage). The most significant genes with differential DGE at 10 dpa included expansins and lipid transfer proteins (higher in Gb), while at 22 dpa tubulins, cellulose, and sucrose synthases showed higher expression in Gb. DGE was compared with expression data of 10 dpa-old fibers from Affymetrix microarrays. Among 543 contigs showing differential expression on both platforms, 74% were consistent in being either over-expressed in Gh (242 genes) or in Gb (161 genes). Furthermore, the unigene set served to identify 339 new SSRs and close to 21,000 inter-genotypic SNPs. Subsets of 88 SSRs and 48 SNPs were validated through mapping and added 65 new loci to a RIL genetic map. The new set of fiber ESTs and the gene-based markers complement existing available resources useful in basic and applied research for crop improvement in cotton.

Highlights

We recently reported the development of an inter-specific G. hirsutum x G. barbadense recombinant inbred line (RIL) population [32] which was used for QTL mapping of fiber quality traits [33] as well as for the mapping of expression QTLs using cDNA-AFLP [34] and microarray hybridization to identify genes underlying fiber quality
In this study we examined the transcriptomes of developing fibers of two cotton accessions representing the most important fiber producing Gossypium species, G. hirsutum and G. barbadense, focusing on two key developmental stages, peak fiber elongation and the onset of secondary cell wall biogenesis, i.e. 10 and 22 dpa respectively
An Expressed Sequence Tag (EST) collection has been generated by 454 pyrosequencing of cDNA from the 2 species at 2 fiber development stages, 10 and 22 days post anthesis

Summary

Introduction

Few transcriptome studies have investigated the cellular mechanisms and genes underlying the important fiber developmental and phenotypic differences between the two major cultivated species G. hirsutum and G. barbadense [16,17,18], despite these being important in many cotton improvement programs. It has been reported previously [18] that fiber development proceeds with relatively similar timelines in representative accessions of Gh and Gb under glasshouse conditions

Methods

Results

Discussion

Conclusion