Background

The family of Orchidaceae is the largest family of flowering plants and the number of species may exceed 25,000 [1]. Like all other living organisms, present-day orchids have evolved from ancestral forms as a result of selection pressure and adaptation. They show a wide diversity of epiphytic and terrestrial growth forms and have successfully colonized almost every habitat on earth. Factors promoting orchid species richness include specific interaction between the orchid flower and pollinator [2], sequential and rapid interplay between drift and natural selection [3], obligate interaction with mycorrhiza [4], and epiphytism which is true for most of all orchids and probably two-thirds of the epiphytic flora of the world.

The radiation of the orchid family has probably taken place in a comparatively short period as compared with that of most flowering plant families, which had already started to diversify in the Mid-Cretaceous [5]. The time of origin of orchids is in dispute, although Dressler suggests that they originated 80 to 40 million years ago (Mya; late Cretaceous to late Eocene) [6]. Recently, the origin of the Orchidaceae was dated with a fossil orchid and its pollinator. The authors showed that the most recent common ancestor of extant orchids lived in the late Cretaceous (76-84 Mya) [7]. They also suggested that Epidendroideae and Orchidoideae, two of the largest orchid subfamilies, which together represent > 95% of living orchid species, began to diversify early in the Tertiary (65 Mya) [7].

According to molecular phylogenetic studies, Orchidaceae comprise 5 subfamilies: Apostasioideae, Cypripedioideae, Vanilloideae, Orchidoideae and Epidendroideae. The Apostasioideae is considered the sister group to other orchids. Vanilloideae diverged just before Cypripedioideae. Both subfamilies have relatively low numbers of genera and species. Most of the taxonomic diversity in orchids is in 2 recently expanded sister-subfamilies: Orchidoideae and especially Epidendroideae [8, 9]. Orchids are known for their diversity of specialized reproductive and ecological strategies. For successful reproduction, the production of labellum and gynostemium (a fused structure of androecium and gynoecium) to facilitate pollination is well documented and the co-evolution of orchid flowers and pollinators is well known [10, 11]. In addition, the especially successful evolutionary progress of orchids may be explained by mature pollen grains packaged as pollinia, pollination-regulated ovary/ovule development, synchronized timing of micro- and mega-gametogenesis for effective fertilization, and the release of thousands or millions of immature embryos (seeds without endosperm) in a mature capsule [12]. However, despite their unique developmental reproductive biology, as well as specialized pollination and ecological strategies, orchids remain under-represented in molecular studies relative to other species-rich plant families [13]. The reasons may be associated with the large genome size, long life cycle, and inefficient transformation system of orchids.

The genomic sequence resources currently available for orchids are limited. Very recently, a sketch of the Phalaenopsis orchid genome from sequencing the ends of 2 bacterial artificial chromosome libraries of P. equestris was reported [14]. In addition, a number of studies have developed expressed sequence tags (ESTs) resources for orchids by using Sanger sequencing [15–18]. Fewer than 12,000 ESTs, including 5,593 from P. equestris, 2,359 from P. bellina, 1,080 from Oncidium Gower Ramsey, and 2,132 from Vanda Mimi Palmer, have been deposited in public database. These studies have highlighted the utility of cDNA sequencing for discovering candidate genes for orchid floral development [19, 20], floral scent production [16, 21] or flowering time [22] in the absence of a genomic sequence. However, a comprehensive description of the full complement of gene expressed in orchids remains unavailable.

Massively parallel 454 pyrosequencing has become feasible for increasing sequencing depth and coverage while reducing time, labour, and cost [23, 24]. This technology can be used to deeply explore the nature and complexity of a given transcriptional universe. 454 sequencing of transcriptomes for model organisms has confirmed that the relatively short reads produced by this technology can be effectively assembled and used for gene discovery [54], shoot apical meristem maintenance [55], drought tolerance [56], and response to abscisic acid in Craterostigma plantagineum[57]. The AP2/ERF superfamily is defined by the AP2/ERF domain, of about 60 to 70 amino acids, and is involved in DNA binding. A combination of genetic and molecular approaches has been used to characterize a series of regulatory genes of the AP2/ERF family. The members of this family are involved in regulating various biological processes related to growth and development, as well as various responses to environmental stimuli. This family includes genes related to drought [58], high salt concentration [58], low temperature [59], diseases [60, 61], and the control of ovule development and flower organ growth [62]. Understanding the functions of these genes will advance our understanding of the great morphological diversity and successful adaptation of orchids. However, we did not find the transcription factor families LFY, M-type, STAT, VOZ, and WOX, in the Phalaenopsis transcriptome. These families might either be rarely expressed, or might not have appeared in our cDNA sampling.

Conclusion

Thanks to recent advances in next-generation sequencing technology, we have applied RNA-seq to facilitate transcriptome analysis of orchids which present important biological questions but lack a fully sequenced genome. Our findings represent substantial contributions to the publicly accessible expressed sequences for the Orchidaceae family. With the whole genome sequencing of P. equestris in progress, this collection of ESTs is a valuable resource that will be immediately useful for researchers, allowing for correction of assemblies, annotation, and construction of gene models to establish accurate exon-intron boundaries. Application of these resources through the common language of nucleotide sequences will greatly enhance the insights into the reproductive success of orchids.

methods

Plant materials and cDNA library construction

Phalaenopsis equestris, P. aphrodite subsp. formosana and P. bellina were grown without fungal symbiosis in greenhouses at National Cheng Kung University under natural light and controlled temperature ranging from 23°C to 27°C. To maximize the diversity and effectively collect sequences from expressed genes of orchids, we collected 10 samples from different tissues, developmental stages and treatments (Table 1). Inflorescences, flower buds, leaves and roots were sampled from the 3-year-old P. equestris. Young leaves were collected as they emerged. Old leaves were taken at the fourth leaf counting down from the newly emerged one. The cold-stressed leaves were collected from old leaves of 3-year-old plants treated for 4 hrs at 4°C. Because Erwinia chrysanthemi is one of the most serious pathogens infecting Phalaenopsis, old leaves were infected with E. chrysanthemi to induce the expression of pathogen-related genes. Protocorms were 20-day-old germinating seeds of P. aphrodite subsp. formosana grown on tissue-cultured plates without fungal symbiosis. Cool night-induced spikes were sampled from 3-year-old P. aphrodite subsp. formosana treated with cool night temperature (28°C day/20°C night) for 2 weeks to induce spike emergence [63]. P. bellina flowers with a strong fragrance were collected on day 5 post-anthesis [16]. Collected samples were frozen immediately in liquid nitrogen and stored at -80°C until used.

Total RNA from each sample was extracted separately following the method described by [19]. Poly-A RNA was prepared from each total RNA sample using the Oligotex@ mRNA Mini kit (Qiagen, Ontario, Canada). Samples of 0.5 μg mRNA from each sample were combined into a single large pool and mixed well. This single large, equally-mixed pool was the source for the cDNA library construction. The cDNA library was constructed using the SMART cDNA synthesis Kit (BD Clontech, Mountain View, CA) according to the manufacturer's instructions.

Pyrosequencing and assembly

In preparation for 454 sequencing, 5 μg of the cDNA sample was nebulized to a mean fragment size of 600 ± 50 bp, end repaired and adapter ligated according to previously published literature [23]. After streptavidin bead enrichment and DNA denaturation, single-stranded molecules were titrated onto derivatized Sepharose beads and then amplified by emulsion PCR. A second streptavidin bead enrichment followed emulsion breaking, the bead-attached DNAs were then denatured with NaOH, and sequencing primers were annealed. One 454 pyrosequencing run was carried out with use of a GS FLX sequencer. A 454 SFF file containing raw sequences and sequence quality information can be access through the SRA web site under accession number SRA030758.2.

Low quality data (base call score < 10) were trimmed from the ends of individual sequences. Sequences shorter than 50 bp after processing were excluded from the analysis. For assembly, GS FLX gsAssembler was used with minimum 40 bases overlap with at least 95% identity.

Sequence analysis and GO classification

All sequences were queried for their similarity to known sequences by use of a BLASTX algorithm [64] against the "nr" protein database. Sequence similarity was considered significant at E-value < 10-7 and the "best hits" annotation was used to represent proteins similar to those encoded by the contigs and singletons. The BLAST score (bits) used the BLOSUM 62 matrix and Existence 11, Extension 1 Gap costs for BLASTX. The GO Slim Classification for Plants, developed at TAIR (http://www.arabidopsis.org/help/helppages/go_slim_help.jsp) was used to characterize the ESTs functionally. The GO identifier of the best hit (with a cutoff of 1e-7) was attributed to the sequence. This step allowed putative functions to be assigned on the basis of the classification proposed by GO.

Characterization of ESTs by Arabidopsis Gene Family and KEGG Pathways

The TAIR9 A. thaliana annotated protein databases (ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR9_genome_release/TAIR9_sequences) was downloaded. The protein sequence set was BLAST against Phalaenopsis contigs and singletons with use of the TBLASTN programs. Sequence similarity was considered significant at an E-value < 10-7. Unique sequences with BLAST matches were mapped to TAIR Gene Families and KEGG Pathways of Arabidopsis for further analysis. The TAIR Gene Family information contains 8,693 genes in 176 gene families updated on September 26, 2009. The KEGG Pathways for Arabidopsis contains 6,756 genes in 121 pathways released on May 11, 2010.

Identification of putative transcription factor-related ESTs

The protein sequences of predicted transcription factors for rice were downloaded from the Plant Transcription Factor Database (PTFDB; http://planttfdb.cbi.pku.edu.cn/). PTFDB contains information on 2,424 rice (Oryza sativa subsp. japonica) transcription factors in 56 families. For identification of transcription factor-related ESTs from Phalaenopsis, the protein sequence set of each predicted rice transcription factor family was BLAST against Phalaenopsis contigs and singletons with use of the TBLASTN programs. Sequence similarity was considered significant at E-value < 10-7.