A Modified ABCDE Model of Flowering in Orchids Based on Gene Expression Profiling Studies of the Moth Orchid Phalaenopsis aphrodite

Previously we developed genomic resources for orchids, including transcriptomic analyses using next-generation sequencing techniques and construction of a web-based orchid genomic database. Here, we report a modified molecular model of flower development in the Orchidaceae based on functional analysis of gene expression profiles in Phalaenopsis aphrodite (a moth orchid) that revealed novel roles for the transcription factors involved in floral organ pattern formation. Phalaenopsis orchid floral organ-specific genes were identified by microarray analysis. Several critical transcription factors including AP3, PI, AP1 and AGL6, displayed distinct spatial distribution patterns. Phylogenetic analysis of orchid MADS box genes was conducted to infer the evolutionary relationship among floral organ-specific genes. The results suggest that gene duplication MADS box genes in orchid may have resulted in their gaining novel functions during evolution. Based on these analyses, a modified model of orchid flowering was proposed. Comparison of the expression profiles of flowers of a peloric mutant and wild-type Phalaenopsis orchid further identified genes associated with lip morphology and peloric effects. Large scale investigation of gene expression profiles revealed that homeotic genes from the ABCDE model of flower development classes A and B in the Phalaenopsis orchid have novel functions due to evolutionary diversification, and display differential expression patterns.


Features and quality assessment of Phalaenopsis specialized microarray
To examine genome-wide gene expression profiling of orchids, a customized microarray chip based on sequence information in the Orchidstra database [1] (http://orchidstra.abrc.sinica.edu.tw) was designed. The Orchidstra database contains 42,661 transcript contigs of Phalaenopsis aphrodite generated by de novo assembly procedure of reads generated from next generation sequencing (NGS) technologies such as Illumia Solexa and Roche 454 platforms. These transcript contigs were submitted to Agilent e-array (Agilent, CA) for microarray probe selection. Probe design of 39,431 transcript contigs was successful after software screening. Probes for the detection of common orchid viruses, Cymbidium mosaic virus (CymMV) and Odontoglossum ringspot virus (ORSV), were included to array design.
We have performed a series of quality test on the tailor made orchid array. Technical repeats showed reproducibility with correlation coefficients higher than 99.9% (after percentile shift normalization at 75%) between repeats (Supplementary Figure S1A). On average, 64-72% whole probe-sets on the array could be considered as detectable under Agilent scanner criteria or raw intensity more than 50. The detection rate was reasonable for basal expression since we have compared various orchid tissues including leaf, root, flower and germinating seeds under normal culture environment. The array performance between the diploid Phalaenopsis aphrodite, a Taiwan native species, and the tetraploid Sogo Yukidian 'V3', a popular commercial hybrid of Phalaenopsis, was satisfactory with correlation coefficients higher than 94% when same tissues were compared (Supplementary Figure S1B). The application of this array is therefore not limited to the rather scarce source of native Phalaenopsis orchid and can be extended to the more popular commercial hybrid.  Genes for the validation are listed in Supplementary Table S2 and their primers were listed in Table S3.  Supplementary Table S3.   homologs and LIPLESS is snapdragon AP2 gene. 38 Phalaenopsis AP2 genes were included in the analysis. Numbers in the brackets indicate number of (PATC/overall) genes within each category. A complete gene list of the phylogenetic analysis is in Supplementary Table S4.  Figure S3 analysis. 28 MADS box genes and 38 AP2 genes of Phalaenopsis aphrodite (marked with *) were denoted with PATC initials used in the Orchidstra database. Additional MADS box genes, 16 others, were applied to Figure 2 (panel B and C) for detailed class A and class B functional groups.

Subcellular localization of Phalaenopsis aphrodite MADS box genes
Method GFP fusion constructs were prepared followed by particle bombardment and confocal microscopy with the same procedures on previous report [2]. Full length cDNA of target genes were PCR amplified and cloned into a smGFP vector (326-GFP) [3]. Nuclei marker construct of mCherry with NLS signal peptide fusion (E3170) was kindly provided by Dr.
Gelvin (Purdue University, IN) [4]. The GFP construct and nucleus mCherry construct were co-bombarded into the petal of Phalaenopsis Sogo Yukidian 'V3' (a commercial hybrid) for further observation with confocal microscopy and image taken (Supplementary Figure S3).
Free GFP construct was also bombarded to orchid petals and showed diffused cytoplasmic pattern (data not shown) as described before [2]. All transient expression experiments were repeated at least three times of bombardments and more than three images were taken from each bombardment.
36 Figure S4. Subcellular localization of MADS box genes according to particle bombardment.
All GFP fusion patterns exhibit nuclei localization and co-localized with nuclear marker of