Since the time of Darwin, biologists have studied the origin and evolution of the Orchidaceae, one of the largest families of flowering plants. In the last two decades, the extreme diversity and specialization of floral morphology and the uncoupled rate of morphological and molecular evolution that have been observed in some orchid species have spurred interest in the study of the genes involved in flower development in this plant family. As part of the complex network of regulatory genes driving the formation of flower organs, the MADS-box represents the most studied gene family, both from functional and evolutionary perspectives. Despite the absence of a published genome for orchids, comparative genetic analyses are clarifying the functional role and the evolutionary pattern of the MADS-box genes in orchids. Various evolutionary forces act on the MADS-box genes in orchids, such as diffuse purifying selection and the relaxation of selective constraints, which sometimes reveals a heterogeneous selective pattern of the coding and non-coding regions. The emerging theory regarding the evolution of floral diversity in orchids proposes that the diversification of the orchid perianth was a consequence of duplication events and changes in the regulatory regions of the MADS-box genes, followed by sub- and neo-functionalization. This specific developmental-genetic code is termed the “orchid code.”

The floral transcriptome of Orchis italica, a wild orchid species, was obtained using Illumina RNA-seq technology and specific de novo assembly and analysis tools. More than 100 million raw reads were processed resulting in 132,565 assembled transcripts and 86,079 unigenes with an average length of 606 bp and N50 of 956 bp. Functional annotation assigned 38,984 of the unigenes to records present in the NCBI non-redundant protein database, 32,161 of them to Gene Ontology terms, 15,775 of them to Eukaryotic Orthologous Groups (KOG) and 7,143 of them to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The in silico expression analysis based on the Fragments Per Kilobase of transcript per Million mapped reads (FPKM) was confirmed by real-time RT-PCR experiments on 10 selected unigenes, which showed high and statistically significant positive correlation with the RNA-seq based expression data. The prediction of putative long non-coding RNAs was assessed using two different software packages, CPC and Portrait, resulting in 7,779 unannotated unigenes that matched the threshold values for both of the analyses. Among the predicted long non-coding RNAs, one is the homologue of TAS3, a long non-coding RNA precursor of trans-acting small interfering RNAs (ta-siRNAs). The differential expression pattern observed for the selected putative long non-coding RNAs suggests their possible functional role in different floral tissues.