Vicia sativa (the common vetch) possesses a predominant zygomorphic flower and belongs to the subfamily Papilionoideae, which is related to Arabidopsis thaliana in the eurosid II clade of the core eudicots. Each vetch flower consists of 21 concentrically arranged organs: the outermost five sepals, then five petals and ten stamens, and a single carpel in the center.
We explored the floral transcriptome to examine a genome-scale genetic model of the zygomorphic flower of vetch. mRNA was obtained from an equal mixture of six floral organs, leaves and roots. De novo assembly of the vetch transcriptome using Illumina paired-end technology produced 71,553 unigenes with an average length of 511 bp. We then compared the expression changes in the 71,553 unigenes in the eight independent organs through RNA-Seq Quantification analysis. We predominantly analyzed gene expression patterns specific to each floral organ and combinations of floral organs that corresponded to the traditional ABC model domains. Comparative analyses were performed in the floral transcriptomes of vetch and Arabidopsis, and genomes of vetch and Medicago truncatula.
Our comparative analysis of vetch and Arabidopsis showed that the vetch flowers conform to a strict ABC model. We analyzed the evolution and expression of the TCP gene family in vetch at a whole-genome level, and several unigenes specific to three different vetch petals, which might offer some clues toward elucidating the molecular mechanisms underlying floral zygomorphy. Our results provide the first insights into the genome-scale molecular regulatory network that controls the evolution and development of the zygomorphic flower in Papilionoideae.
Citation: Liu Z, Ma L, Nan Z, Wang Y (2013) Comparative Transcriptional Profiling Provides Insights into the Evolution and Development of the Zygomorphic Flower of Vicia sativa (Papilionoideae). PLoS ONE 8(2): e57338. https://doi.org/10.1371/journal.pone.0057338
Editor: Tai Wang, Institute of Botany, Chinese Academy of Sciences, China
Received: September 27, 2012; Accepted: January 21, 2013; Published: February 21, 2013
Copyright: © 2013 Liu et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the National Basic Research Program of China (2007CB108904), the National Natural Science Foundation of China (31072072 and 31272492) and the Fundamental Research Funds for the Central Universities (lzujbky-2012-94). The funders had no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The pollinator-driven morphological diversification of flowering plants is closely associated with changes in the number, expression levels and interactions of a number of functional transcription factors. Studies in two core eudicot species, Arabidopsis thaliana and Antirrhinum majus, have led to the development of the classic ABC genetic model, which explains how three classes of genes (A, B, and C) interact spatially to specify floral organ identity , . In Arabidopsis, A (AP1, AP2) determines sepals; A and B (AP3, PI) specify petals; B and C (AG) specify stamens; and C determines the carpel. Transcription expression profiling provides broader and deeper insight by extending beyond these key transcription factors commonly identified via mutant analyses to the genes downstream of the transcription factors that are involved in related pathways or networks, which may ultimately explain a given process. Using microarray techniques, this approach has been applied in the following angiosperms: the core eudicot Arabidopsis, at a whole-genome level ; the basal eudicots California poppy (Eschscholzia californica), using 6,446 ESTs , and Aquilegia formosa, using 17,246 unigenes ; and the basal angiosperms avocado (Persea americana), with 6,068 ESTs , and water lily (Nuphar advena), with 6,220 ESTs . All other angiosperms are phylogenetically distant from model plants species (e.g., Arabidopsis), and have been analyzed using small or middle transcriptome. Therefore, previous studies might have been unable to recover transcriptome characteristics at a whole-genome scale due to the limited data available. Additionally, because only Arabidopsis belongs to the core eudicots, previous studies have tended to miss signatures of floral development and evolution within the core eudicots.
The Papilionoideae, most of which (28 tribes) exhibit specialized zygomorphic flowers, are distantly related to Arabidopsis in the eurosid II clade of core eudicots , . This subfamily includes 30 tribes, 455 genera and approximately 12,000 species. Since the 1850s, the unique shape and the zygomorphic papilionoid flowers has been the subject of intense research, both due to its importance in Mendel’s groundbreaking work regarding genetic laws and as a model for studying organ differentiation and morphogenesis –. Papilionoid flowers differ greatly from those of other well-studied eudicots in terms of their floral structure, with most of these flowers possessing five sepals, five petals, ten stamens in two whorls and a single carpel. The organs within each whorl are initiated unidirectionally from the abaxial to the adaxial side. Additionally, in contrast to the strict timing order observed in Arabidopsis, the developmental timing of one whorl may overlap with that of the next whorl in some papilionoids , . Recently, resources for transcriptome atlases have been developed in Papilionoideae, including Lotus japonicus , Medicago truncatula  and Glycine max . In these studies, the flower was sampled as a whole organ without dissection to investigate gene expression. Therefore, a gap remains in the available datasets that would allow description the genome-wide transcriptional expression patterns in Papilionoideae flowers in terms of individual organs, such as the sepals, dorsal petals, lateral petals, ventral petals, stamens, and carpels.
Recent RNA deep sequencing technologies, such as Illumina transcriptome assembly analysis and RNA-Seq Quantification analysis, provide new opportunities to unravel the possible depths of transcriptome sequencing, even for species without a sequenced genome. Transcriptome assembly analysis generates millions of short cDNA reads, providing short, overlapping fragments that cover the entire transcriptome. RNA-Seq Quantification analysis provides a digital measure of the presence and prevalence of transcripts from known genes. The combination of these two approaches generates quantitative gene expression data and overcomes many of the inherent limitations of microarray-based systems –.
Here, we have developed the first transcriptome expression analysis for individual floral organs in Papilionoideae using Illumina assembly technology and RNA-Seq Quantification analysis. The common vetch (Vicia sativa L.) belongs to the Papilionoideae and is a member of the Fabales clade, a sister clade to eurosid II (Arabidopsis) in phylogenetic reconstructions of the Rosids. We investigated the gene expression levels in the sepals, dorsal petals, lateral petals, ventral petals, stamens and carpels relative to the leaves and roots in V. sativa to provide an assessment of the spatial gene expression patterns in vetch flowers. Based on comparative analyses using data from a similar flower stage in Arabidopsis from a publically available dataset , we present insights into the general and unique molecular signatures of the evolution and development of the zygomorphic flower of the Papilionoideae at a whole-genome scale.
Materials and Methods
Plant materials and RNA extraction
In this study, eight organs from the common vetch cultivar Lanjian 3 were investigated: the sepals, dorsal petals, lateral petals, ventral petals, stamens, carpels, leaves, and roots (Figure 1). Leaves and roots were collected from 2-week-old seedlings. The other organs were harvested from plants grown at Lanzhou University in Lanzhou, China for approximately 45 days in a greenhouse under a 16 hr light/8 hr dark cycle at 22°C. A total of 300 flowers were harvested at late pre-anthesis, from 100 individual plants and dissected into the six floral organs (sepals, dorsal petals, lateral petals, ventral petals, stamens, and carpels). The floral stage of vetch was identical to stage 12 in Arabidopsis . All of the tissues were harvested between 9:00–12:00 a.m., placed immediately in liquid N2 and stored at −70 °C until RNA was extracted. Total RNA was extracted from the eight organs using the RNeasy Plant Mini Kit (Qiagen, Cat. #74904). RNA quantity and quality were assessed using a NanoDrop ND1000 instrument (Thermo Scientific).
(a–i) Sepal, dorsal petal, lateral petal, ventral petal, stamen, carpel, leaf, root and transverse section (2 µm) of a vetch flower at late pre-anthesis, which is identical to stage 12 in Arabidopsis . C, carpel; DP, dorsal petal; LP, lateral petal; SE, sepal; ST, stamen; VP, ventral petal. Scale bars = 1.0 mm in (a–g). 5.0 mm in (h) and 500 µm in (i).
Bright-field photographs of individual floral organs were obtained using a dissecting microscope (Leica, M205). Vetch flowers at late pre-anthesis were fixed in FAA (10% formalin, 50% ethanol, and 5% acetic acid) and dehydrated in a series graded ethanol. For histological analysis, the tissues were infiltrated with xylene and embedded in cold-polymerizing resin (Technovit 7100). The materials were then sectioned into 2 µm thick sections, stained with toluidine blue (Bio Basic) and observed using a light microscope. Transverse sections were photographed using a Nikon E600 microscope and a Nikon DXM1200 digital camera.
Illumina sequencing, de novo assembly and functional annotation
For Illumina sequencing, equivalent quantities of total RNA isolated from the eight tissues were pooled. After poly(A) mRNA was purified and fragmented into small pieces, we used random hexamer primers and reverse transcriptase (Invitrogen) to carry out first-strand cDNA synthesis. Second-strand cDNA synthesis was performed with RNase H (Invitrogen) and DNA polymerase I (New England BioLabs). We constructed a cDNA library with average insert sizes of 200–500 bp and conducted cDNA sequencing using the Illumina HiSeq™ 2000 system according to the manufacturer’s protocols, with a read length of 100 bp. The average proportion of clean reads for the library was 96.8%. The transcriptome sequence was assembled into distinct contigs using the short reads with SOAPdenovo software  (http://soap.genomics.org.cn). The paired-end relationships between the reads were employed to construct scaffolds between the contigs. We next filled the intra-scaffold gaps and constructed a non-redundant unigene set from all three of the assembled datasets using the EST assembly program TGICL . We annotated the sequences based on protein databases, such as nr, Swiss-Prot, KEGG, and COG (E-value <10E-5) by retrieving the proteins with the highest sequence similarity to the given unigenes, along with their functional protein annotations. The Blast2GO program  was employed to obtain GO annotations for the unigenes.
RNA-Seq Quantification analysis
Eight independent cDNA libraries were constructed for the eight organs in parallel according to the RNA-Seq protocol. Raw image files were collected using the Illumina HiSeq™ 2000 sequencing platform in BGI Shenzhen (http://en.genomics.cn/navigation/index.action). The data analyzed have been deposited on the NCBI Gene Expression Omnibus under accession no. GSE35437.
Total RNA was isolated from six floral organs as well as leaves and roots with the RNeasy Plant Mini Kit. For first-strand cDNA synthesis, we used 2.5 µg of total RNA and followed the manufacturer's protocol (TaKaRa Biotechnology, Dalian, China). The cDNA samples were diluted to a concentration of 2.5 ng/µl. Quantitative RT-PCR (qRT-PCR) was performed in triplicate using the 7500 Real-time PCR System (Applied Biosystems) with 4 µl of the cDNA dilution and Power SYBR Green Master mix (Applied Biosystems), according to the manufacturer's protocol. Unigene68614, which has an unknown function, was used as an internal control because of its relatively invariable expression across the transcription profiles. The mRNA fold changes in the different samples were calculated by the 2ΔΔCt method, as described previously . In a semi-quantitative PCR analysis to confirm the sequencing data, we employed the same primers as for the qRT-PCR.
The expression level of each gene was measured via RNA-Seq Quantification analysis as the number of reads per kilobase of an exon region in a given gene per million mapped reads (RPKM) . To identify differentially expressed unigenes among the eight organs, the Z-score transformation normalization method was applied to compare the expression levels of the unigenes from the floral organs, leaves and roots and to directly calculate the differences in unigene expression levels between the different samples , . Z-scores were calculated by dividing the difference between the target organ or combination of organs (Xi, organ or mean of organs) and all of the other organs (µ, mean) by the standard deviation (SD) of all of the other organs using the following equation:
Additionally, we measured the relative Ratio among samples using the following equation:
where X represents the expression level in the target organ or combination of organs, and O is the expression level in all of the other organs. The expression levels in both the common vetch and Arabidopsis were analyzed using the above equations.
Comparative floral transcriptome
To compare the evolutionary patterns of the floral transcriptomes, we compared the floral expression patterns of vetch and Arabidopsis. Microarray expression data for the leaves, roots and floral organs of stage 12 in Arabidopsis were extracted from datasets . To provide new insights into the development of the dorsal petal, we performed a homologous comparison between genomes of vetch and M. truncatula. The genome data M. truncatula were from Mt3.5 Release (http://tofu.cfans.umn.edu/). Based on nucleotide sequence similarity, a reciprocal BLAST search was performed with a cutoff E-value<10E-5 using blast-2.2.23-ia32-win32. The expression data were visualized using MeV v4.6 (http://www.tm4.org/mev/).
Gene Ontology analyses and cis-elements analyses
The molecular function and cellular component term "enrichment status" and "hierarchy" of unigenes were analyzed via agriGO software (http://bioinfo.cau.edu.cn/agriGO/analysis.php) . The PLACE database (http://www.dna.affrc.go.jp/PLACE/)  was used for cis-element search in the promoter regions of homologous genes in M. truncatula. We selected an approximate 2,000 bp promoter region in each gene to search for the possible cis-element.
Transcriptome characteristics in floral organs
After using paired-end information to join contigs into scaffolds and local assembly, we generated a total of 71,553 high-quality unigene sequences with lengths greater than 200 bp, and the average length of the unigenes was 511 bp. The size distribution of the assembly is displayed in Figure S1. All unigene sets were then annotated based on similarities to known or putative sequences available in public databases. The COG function classification of the vetch unigenes is displayed in Figure S2.
Using 71,553 unigenes, we analyzed the genome-scale gene expression profiles in eight organs, i.e., the sepals, dorsal petals, lateral petals, ventral petals, stamens, and carpels, using the leaves and roots as controls (Figure 1). The qRT-PCR analysis and semi-quantitative PCR analysis were used to confirm the sequencing results (Figure S3A and Table S1). A total of 60 unigenes displaying biased or unbiased expression in the RNA-Seq Quantification analysis was examined (Table S1). The Illumina sequencing and qRT-PCR results displayed a significant positive correlation for 49 of the 60 unigenes (R>0.707, P<0.05), which demonstrated that approximately 81.7% of the Illumina sequencing data could be directly validated using qRT-PCR. Furthermore, the semi-quantitative PCR results (Figure S3B) correlated well with the expression patterns of the 49 unigenes detected via Illumina sequencing.
Additionally, based on the unigene expression values obtained for each organ, we analyzed the Spearman correlation coefficients (r-value) between the pairs of organs (Table S2A) and obtained the highest correlation coefficients for the pairs of the three types of petals: dorsal petals/lateral petals (0.9399), lateral petals/ventral petals (0.8552) and dorsal petals/ventral petals (0.8119). The lowest correlation coefficient (0.2492) was observed when comparing the carpels/lateral petals. After combining the three kinds of vetch petals into a single category (the petals), the highest correlation coefficient (0.7197) was found between the petals and stamens in vetch (Table S2B). Hierarchical clustering of the total of 71,553 vetch unigenes sorted the floral organs into groups based on the similarity of the unigene expression patterns. The lateral and dorsal petals clustered together, followed by the ventral petals and the stamens, while the sepals and the carpels clustered distantly from the other floral organs (Figure S4).
Identification of enriched and downregulated unigenes in the floral organs
We identified unigenes that were enriched in the floral organs by comparing the expression levels of the unigenes in individual floral organs and in other tissues with cutoffs of a Ratio >2.0 and Z-score >3.75 . These analyses revealed that 9,502 unigenes were enriched in the individual floral organs in vetch. The carpels (4,011) exhibited the greatest number of the enriched unigenes (Table S3A). To achieve correspondence with the floral structure of Arabidopsis, we combined the three types of vetch petals into a single category and obtained 3,541 unigenes that were preferentially expressed in the petals. We used similar methods to screen for genes enriched in the floral organs of Arabidopsis  and determined the stamens possessed the greatest number of the enriched genes (Table S3A).
The genes that were downregulated in the floral organs showed a similar characteristic to the genes that were enriched in the floral organs. We chose parameters with a relatively low stringency (Ratio<0.5 and Z-score<−1.5) to screen for unigenes that were downregulated in the floral organs. We determined that 495 unigenes were downregulated in the sepals, 624 in the petals, 3,855 in the stamens and 12,504 in the carpels in vetch (Table S3B). Consistent with analysis of unigenes that were enriched in individual organs in vetch, the carpels also exhibited the greatest number of the downregulated unigenes. Using a similar method (Ratio<0.5 and Z-score<−1.5), we screened the genes that were downregulated in the floral organs of Arabidopsis  (Table S3B). Similar to our previous results in Arabidopsis, the stamens also displayed the greatest number of downregulated genes.
Furthermore, we analyzed the functional features of the enriched and downregulated transcripts in the investigated organs according to Gene Ontology terms in the transcriptome annotation for vetch (Figure S5 and Table S4). The enriched and downregulated unigenes in the floral organs of vetch were classified into 14 functional groups, and the distribution of these unigenes showed a clear variance among distinct floral organs (Figure S5). Both the enriched and the downregulated transcripts showed an obvious skew toward metabolic processes, cellular processes and responses to stimuli. However, the functional features of the enriched and the downregulated unigenes were strikingly different in some categories, such as cell wall organization or biogenesis, death-related transcripts, stimulus and death. These findings indicated that organ-dependent enrichment or downregulation of different unigenes is essential for floral organ differentiation, development, and functional specification.
To further analyze the molecular regulation of floral organ development, we searched the differentially expressed unigenes that were both enriched in individual organs and enriched in three combinations of organs (representing the A, B, and C domains) using cutoffs of a Ratio>2.0 and Z-score>3.75. This screening identified 24,472 unigenes that were enriched in vetch floral organs (Table S3A): 9,502 displayed enrichment in individual organs, whereas 14,970 were enriched in a combination of organs. It is conceivable that the unigenes enriched in the individual organs and those enriched in the combinations of organs might play important roles or be involved in pathways downstream of the key regulators involved in floral organ development. Therefore, we focused on the expression characteristics of six functional groups (4,501 of the 24,472 unigenes enriched in the floral organs): transcription, phytohormones, reproduction, kinase systems, stimulus responses and signaling transduction (Table S5A).
For example, the unigenes belonging to 41 distinct transcription factor families displayed a heterogeneous distribution among the four individual organs and the three organ combinations. However, these 41 transcription factor families exhibited a significant skew for the following 13 families: MYB/MYB related, MADS, NAC, bHLH, C3HC4, WRKY, LIM, BTB/POZ, bZIP, GRAS, ARF, HMG and SBP (Table S5B). Among the single organ, the number of transcription factors in the carpels was the greatest, indicating that the development of the carpels may require more transcription factors compared to the other organs. This result also explained why the carpels possessed the greatest numbers of both enriched unigenes and downregulated unigenes among the vetch floral organs (Tables S3A and S3C). The enriched and downregulated unigenes might be the downstream target genes of the 13 transcription factor families. Additionally, a total of 1,385 unigenes that were enriched in individual organs or in a combination of organs encoded kinase domain-containing proteins (Table S5D). Consistent with calcium signaling playing an important role in pollen development, the calcium signal-related kinase unigenes were mainly distributed in the stamens, which were characterized by containing abundant mature pollen , .
Comparative floral transcriptome analysis between vetch and Arabidopsis
We compared potentially conserved and diverse evolutionary features of the vetch and Arabidopsis floral transcriptomes based on published data for the transcriptomes of Arabidopsis flowers . Using BLAST searches with a cutoff E-value<1.0E-05, we found that 23.6% (6,152) of the 26,194 unigenes (1,819 enriched unigenes in the three individual petal types subtracted from 28,013 total unigenes) enriched in individual organs or a combination of organs in vetch had homologous transcripts in the Arabidopsis genome. A total of 3,367 (56.4%) of these 6,152 homologous unigenes were expressed in individual organs or a combination of organs in Arabidopsis. Furthermore, 1,895 (27.5%) of these 6,152 unigenes were enriched in individual organs or a combination of organs in Arabidopsis (Tables 1, S6A, and S6B). Interestingly, among the individual organs, the unigenes enriched in the vetch stamens displayed the greatest number (473) of enriched homologs in Arabidopsis organs.
In contrast, further BLAST reaches against the vetch database using the Arabidopsis sequences determined that 3,496 (56.1%) of the 6,412 genes enriched in individual organs or a combination of organs in Arabidopsis exhibited homologous transcripts in the vetch genome. A total of 1,890 (58.1%) of these 3,496 homologous genes were expressed in individual organs or a combination of organs in vetch, whereas 1,291 (31.6%) of these 3,496 genes displayed enriched expression in individual organs or a combination of organs in vetch (Tables S6A and S6C). Consistent with the results described above, the stamens (365) also presented the greatest number of enriched homologs in the two plants, which might suggest the existence of non-conserved expression patterns in the counterpart floral organs of vetch and Arabidopsis, with the exception of the stamens.
Additionally, we compared the expression patterns of the homologous transcripts that were enriched in both vetch and Arabidopsis. Only 1,052 unigenes (after deleting duplicates from 1,895 unigenes) displayed enrichment in counterpart organs in both vetch and Arabidopsis; these transcripts were mainly involved in hydrolase activity, transcription factor activity, transferase activity, protein binding, DNA or RNA binding, and kinase activity.
We further compared the spatial distribution of floral-biased expression in vetch and Arabidopsis flowers . Using log2 floral organ/non-adjacent-organ gene expression values (Ratio>2.0 and Z-score>3.75), we obtained a total of 3,903 Arabidopsis transcripts and 20,714 vetch unigenes enriched in the floral organs. Similar to the results for Arabidopsis , the enriched vetch transcripts demonstrated sharp boundaries among the different floral organs (Figure 2A). Based on all of the transcript expression data for Arabidopsis (21,972 genes) and vetch (71,553 unigenes), we constructed scatter plots for the adjacent floral organ categories. Among the floral organwise categories, the Pearson correlations were 0.55, 0.72, and 0.35 for vetch, and 0.71, 0.51, and 0.46 for Arabidopsis (Figure 2B). These results illustrated that the adjacent floral organs are divergent from each other in both species, suggesting that vetch flowers may be shaped following a strict ABC model without a “fading border” .
(a) Log2 floral organ/non-adjacent-organ gene expression ratios in vetch and Arabidopsis. Transcripts showed enriched expression in floral organs in vetch and Arabidopsis using cutoffs of a Ratio>2.0 and Z-score>3.75. C, Carpel; P, petal; SE, sepal; ST, stamen. The scale of the log2 ratios ranges from −1.5 (black) to 1.5 (yellow). The analyses are based on 3,903 Arabidopsis and 20,714 vetch transcripts enriched in the floral organs. (b) Scatter plots and Pearson correlations (r) of the gene expression levels in adjacent floral organs in vetch and Arabidopsis. The analyses are based on all of the Arabidopsis genes (21,972) and all of the vetch unigenes (71,553). The Arabidopsis data were obtained from Schmid et al. .
Furthermore, we examined transcription factors, which are key regulators of the events that involve in organ development to compare the conserved features of the transcriptomes of the two species. Among the 655 transcription factors unigenes enriched in the vetch floral organs (Table S6), 120 homologs from 16 transcription factor families were found to also display enriched expression in Arabidopsis organs (with an E-value<1.0E-05), including the MYB/MYB related, NAC, MADS and HOMEOBOX families, which suggested that these transcription factors share conserved functions in vetch and Arabidopsis. Interestingly, 102 unigenes from 17 transcription factor families, such as C4HC4, LIM, TLP, SWI/SNF, HSF, CCAAT, and AS2, appeared to be specific to vetch flowers, and none of these famlies were enriched in Arabidopsis. However, eight genes from six transcription factor families, including DDT, VP1/ABI3, and S1FA-like, that displayed enriched expression in Arabidopsis were not enriched in vetch.
We further compared the spatial expression profiles of one-to-one homologous enriched transcripts in vetch and Arabidopsis (Figure 3 and Table S7). Compared with the organ-enriched unigenes (6,152) in vetch, most of the Arabidopsis homologs skewed toward equivalent expression among the different organs, whereas only a few homologs in Arabidopsis, such as a subset of genes in the stamens and the carpels, displayed enriched expression patterns (Figure 3A). The genes in the C domain exhibited relatively conservative expression patterns between the two species. In contrast, most of the homologs of the genes (3,496) enriched Arabidopsis showed unbiased expression patterns in vetch, except for the genes enriched in the stamens and the carpels (Figure 3B). These results suggested that the expression patterns of downstream homologs in the two species display weak conservation among the floral organs, despite the fact that vetch and Arabidopsis both belong to the Rosids in the core eudicots.
(a) Expression patterns of the unigenes (5,891) enriched in vetch and their homologs (arranged in the same row) in Arabidopsis (E-value<1.0E-05). (b) Expression patterns of the genes (3,367) enriched in Arabidopsis and their homologs in vetch (E-value<1.0E-05). The bar represents the scale of the expression levels of the transcripts. The scale of the log2 ratios ranges from −2.5 (black) to 2.5 (yellow). The abbreviations for the different tissues are the same as in Figure 1. The Arabidopsis data were obtained from Schmid et al. .
Floral zygomorphy of vetch revealed by transcriptional profiling
We first identified the unigenes that were preferentially expressed in the three types of the vetch petals using cutoffs of a Ratio>2.0 and Z-Score>3.75. We then performed BLAST searches against the Arabidopsis database using a cutoff E-value<1.0E-05. A total of 31, 25, and 80 unigenes showing specific expression in the dorsal, lateral and ventral petals, respectively, were found to have homologous transcripts in the Arabidopsis genome (Figure 4). Among the unigenes presenting specific expression patterns in the three different types of vetch petals, most of the Arabidopsis homologs displayed equivalent expression levels among the different organs. The unigenes that are specifically expressed in vetch petals may contribute or respond to morphological differences in the petals between vetch and Arabidopsis. To provide fresh insights into the transcription regulation of gene expression in the dorsal petal development, we performed a homologous comparison via BLAST searches (E-value<1.0E-05) against the M. truncatula genome database using 462 transcripts specific in vetch dorsal petals. A total of 100 homologous genes were obtained from M. truncatula. We then researched cis-acting regulatory elements from approximate 2,000 bp regions upstream of start codon using the PLACE database. Of 237 identified cis-elements, the five highest binding sites were YACT, AAAG, CAAT, GRWAAW, and NGATT (Table S8A). Impressively, 104 putative CYC or TCP binding sites GGNCCC were identified in 49 genes in M. truncatula (Table S8B). The gene AC233140_70.1, also named CYC1A, was included . Additionally, analyses of the 462 dorsal petal specific unigenes in biological process and molecular functional terms "enrichment status" and "hierarchy" showed that ion transmembrane transport and localization had significant importance in the development of the dorsal petals (Figure S6A and S6B). Transcripts involved in internal side of plasma membrane seemed be more important in the dorsal petal specific data (Figure S6C).
The unigenes enriched in the three types of petals and their homologs in Arabidopsis (E-value<1.0E-05) are arranged in the same row. Most of the Arabidopsis homologs display equivalent expression among the floral organs. The bar represents the scale of the expression levels of the transcripts. The scale of the log2 ratios ranges from −2.5 (black) to 2.5 (yellow). The abbreviations for the different tissues are the same as in Figure 1. The Arabidopsis data were obtained from Schmid et al. .
Transcription factors in the TCP (TEOSINTE BRANCHED1 from maize, CYCLOIDEA from snapdragon, and PCF from rice) family regulate cell division in vegetative and reproductive structures and are essential for establishing flower zygomorphy. To understand the evolutionary relationship between the TCP genes of vetch and Arabidopsis, we identified 16 and 24 TCP genes in vetch and Arabidopsis, respectively . Through performing a phylogenetic analysis of the TCP protein domain, we grouped these TCP genes into three clades: PCF, CYC/TB1, and CIN (Figure 5A). There were seven unigenes in the PCF clade, three unigenes in the CYC/TB1 clade and six unigenes in the CIN clade. In the NJ phylogenetic tree, we observed seven well-supported clades containing both vetch and Arabidopsis genes, suggesting that the most recent common ancestor of vetch and Arabidopsis had at least seven TCP genes. Because there are a few additional clades including only vetch or Arabidopsis genes, the number of TCP genes in the most recent common ancestor was probably greater than seven. Furthermore, the expression pattern of the TCP genes in vetch was analyzed, and the genes in the CYC/TB1 clade were preferentially expressed in vetch petals, while the genes in the PCF and CIN clades exhibited unbiased expression among the eight vetch organs (Figure 5B). This interesting result might indicate that only the CYC/TB1 genes in the TCP family are involved in floral zygomorphy in vetch.
(a) Unrooted protein tree for the TCP protein domain data set produced using MEGA 4.0.1 . The tree summarizes the evolutionary relationships among the 24 AtTCPs , 16 VsTCPs, and other representative sequences. The bootstrap values are denoted above the nodes; bootstrap values of less than 50 are not shown on the phylogenetic tree. The representative sequences include OsPCF1 (NP_001052212) and OsPCF2 (NP_001062409) in rice; ZmTB1 (AF377742) in maize; and AmCYC (Y16313) in snapdragon. (b) The expression patterns of the TCP genes in vetch organs. The abbreviations for the different tissues are the same as in Figure 1.
Genome-scale gene expression patterns in vetch
Based on de novo assembly and RNA-Seq Quantification analysis, our examination of the genome-scale gene expression profiles in eight organs in vetch revealed characteristics of the transcriptomes of the floral organs and provided novel insights into the evolution and development of the zygomorphic flower. Comparative transcriptional analyses of wild-type and mutant flowers of Arabidopsis have distinguished a large number of genes that are not expressed in ABC gene mutants and are likely downstream of these key transcription factors , . In this study, we identified unigenes enriched in individual floral organs and combinations of organs in vetch. The unigenes enriched in the different organ combinations may be regulated by and associated with the homologs of AP1, AP3, PI, AG, and SEP1 in vetch. For the AP1 homolog, expression peaks were mainly detected in the sepals and the dorsal, lateral, ventral petals (Figure S7). Vetch homologs of AP3 and PI were strongly expressed at late pre-anthesis in the three types of petals and the stamens, similar to the pervious findings that persistent activity of AP3 and PI through the late stages of petal development was necessary to maintain the expression of characteristic petal features –. Additionally, the AG homolog was highly expressed in the stamens and carpels, while expression of SEP1 homolog was observed throughout the vetch flower. Therefore, our findings related to the ABC regulators in vetch were consistent with what is observed in other angiosperms, such as Arabidopsis , , L. japonicus , Petunia hybrida , Oryza sativa –, Gerbera hybrida , Zea mays , Pisum sativum  and Eschscholzia .
Among the A domain unigenes (Figure 3), all of the transcripts were strongly expressed in the three types of petals, whereas the expression levels were not as high in the sepals. Considering the dubious role of a “true” A function in the control of perianth identity –, our results have provided a few valuable clues for future molecular studies. The B domain genes, AP3 and PI, are best understood from an evolutionary standpoint, and these genes appear to be associated with a gene duplication event that occurred prior to the radiation of angiosperms . Although the B domain unigenes were found to be expressed at higher levels in the petals and stamens in the present study, there were apparently also some organ-specific transcripts, especially in the three different petal types (Figure 3), which is much different from what is observed in Arabidopsis and may be important for the development of petals in vetch. It was therefore tempting to speculate that the B domain genes of the Papilionoideae have undergone gene duplication and differentiation, leading to the genes controlling the identity and development of the three complex types of petals following the radiation of the B domain genes in angiosperms.
Interestingly, when the relationships among the organs were examined based on global transcription expression levels (Figure S4), the three types of vetch petals clustered tightly with the stamens. We provided further support for this finding by calculating the Spearman correlation coefficients between pairs of vetch organs (Table S2). The results indicated that the dorsal, lateral, and ventral petals shared some common genetic programs and were process homologous. Also, this finding is in agreement with the traditional hypothesis of the petals morphologically originating from the stamens in core eudicots , . The clustering of the petals with the stamens in vetch might be explained by the following findings. An interesting characteristic is provided by the description of the common primordia in some Papilionoideae flowers from which the petals and the stamens differentiate , , , .
Characteristics of the floral transcriptomes of vetch and Arabidopsis
Arabidopsis is the best-characterized experimental model among the eudicot plants, and global gene expression data for specific floral organs are available for this species. Our analyses suggested that there are both conserved and divergent features of the floral transcriptomes of vetch and Arabidopsis. Homologs showing similar enriched expression patterns in the two species were mostly found to be involved in basic processes, such as hydrolase activity, transcription factor activity, transferase activity, protein binding, DNA or RNA binding, and kinase activity. The expression of transcription factors was conserved between the two species for the MYB/MYB-related, NAC, MADS, and HOMEOBOX families. Additionally, consistent with other angiosperms, homologous ABC regulator genes exhibited similar expression patterns in the two species (Figure 3), suggesting that major components of their functions may be conserved throughout angiosperm evolution . The genes participating in the transcriptional programs of vetch and Arabidopsis flowers were tightly constrained to individual floral organs (Figure 2A), and organwise comparisons between adjacent floral organ categories showed similar correlations in the two species. Therefore, we concluded that vetch flowers display strict spatial ABC domains and sharp transcriptional boundaries among the floral organ categories, as observed in Arabidopsis flowers. A similar result was established in Eschscholzia , , suggesting that the flowers of more recently derived angiosperm lineages with highly differentiated perianths, such as those of the core eudicots (represented by vetch and Arabidopsis) and some basal eudicots (represented by Eschscholzia), are more likely to be characterized by a strict ABC model with sharp transcriptional boundaries among adjacent floral organ categories, despite the “fading borders” observed in other basal angiosperms (represented by Aquilegia, because of its distinctive floral anatomy) .
The percentage of transcripts enriched in vetch sepals (1.20%) is similar to that in Arabidopsis sepals (1.33%), implying that vetch tends to share a similar evolutionary scenario with Arabidopsis in terms of sepal development (Table S3A). With respect to the different categories of petals, the greater number of transcripts enriched in vetch (3,541) compared with the number of genes enriched in Arabidopsis (368) suggested that the petal developmental program in vetch is more complex than that of in Arabidopsis. Moreover, the genes showing enriched expression in vetch petals displayed a broader expression domain (4.95%) than is observed in Arabidopsis petals (1.67%) (Table S3A). Considering the three highly differentiated types of petals observed in vetch, this result can be explained by the greater complexity of the petal organ category in vetch versus Arabidopsis. Additionally, the vetch stamens and carpels displayed a larger number of enriched unigenes, which is consistent with the model eudicot Arabidopsis, and the basal angiosperms, Nuphar and Persea . This conservation of transcript numbers might be explained by the diversity of cellular events taking place in the stamens and carpels. In vetch, the carpels displayed the largest number of enriched unigenes and the most downregulated unigenes, whereas the stamens presented the greatest number of enriched and downregulated unigenes in Arabidopsis. This finding suggested possible differences between stamen and carpel development in the two species. Furthermore, we analyzed the evolutionary conservation of the transcripts enriched in vetch and Arabidopsis based on the expression patterns of their homologous transcripts (Figure 3). Compared to the sepals and petals, the homologs of the transcripts enriched in the stamens and the carpels exhibited relatively more highly conserved expression patterns in vetch and Arabidopsis, suggesting conservation of gene expression between the two species in the molecular regulation of stamen and carpel development.
Because both vetch and Arabidopsis display strict spatial ABC domains, we compared the spatial expression patterns of one-to-one homologous transcripts between vetch and Arabidopsis (Figure 3). It was determined that the downstream transcriptional regulatory machinery responding to the ABC domains in the two species was not conserved with regard to the expression patterns of the genes involved. With respect to the genes displaying enriched expression in the floral organs of vetch, a high percentage of homologs did not show enriched expression in the counterpart organs of Arabidopsis and vice versa (Table 1 and Table S7). Our results were supported by a previous study in which it was determined that most of the genes displaying enrichment in Arabidopsis floral organs were downregulated in Nuphar and Persea, whereas those showing enrichment in Nuphar, Persea, or both, were downregulated in Arabidopsis. Recent large-scale studies of evolutionary changes in gene expression among several non-plant species have led to the proposal that there is a substantial amount of divergence in orthologous gene expression –. However, the divergence of orthologous gene expression levels is significantly lower for pairs of randomly chosen genes. Furthermore, divergence in the expression of essential genes appears to be strongly constrained by natural selection, and other genes tend to be under minimal selective constraints . Therefore, we assumed that the key regulatory genes involved in the ABC domains in eudicot plants are likely subjected to stronger selection and display a conserved expression pattern among diverse floral organs. A large proportion of downstream orthologous transcripts in ABC domains have been subjected to minimal selection and display divergence in spatial expression during flower evolution and diversification.
Control of floral zygomorphy in vetch
One of the most distinct aspects of Papilionoideae species is their zygomorphy, which originated indendently several times from actinomorphic ancentors (showing radial floral symmetry) and subsequently expanded to many large and successful angiosperm lineages , . Vetch flowers, like most zygomorphic flowers, possess two types of asymmetries: dorsoventral asymmetry in the floral plane and organ internal asymmetry in the floral organ plane. The internal asymmetry is different among the various petals: the dorsal petal is internally symmetric, and two lateral petals and two ventral petals are internally asymmetric. In P. sativum and L. japonicus, the floral zygomorphy depends on the activity of the TCP transcription factors. CYC2 plays a role in establishing the identity of the dorsal petals, while CYC3 plays a key role in controlling the identity of the lateral petals , . Both CYC1 and CYC2 are specifically expressed in the dorsal petals to promote the growth of the single dorsal petal, while CYC3 is strongly expressed in the dorsal and lateral petals , . In the present study, the expression patterns of VsCYC1, VsCYC2, and VsCYC3 in vetch were found to be similar to their patterns in other species (Figure S7), which was reconfirmed using qRT-PCR and semi-quantitative PCR (Figure S3 and Table S1). This finding suggested that the three VsCYC genes might also be crucial in controlling floral zygomorphy in vetch.
A pair of dorsal-specific CYC genes is usually redundantly or partially redundantly involved in the establishment of floral zygomorphy, with one of the copies exhibiting no or transient expression in floral zygomorphic regions, such as the CYC and DICH genes in snapdragon , , CYC1 and CYC2 in Mohavea confertiflora , CYC1A and CYC1B in Lupinus nanus (Fabaceae) , CYC1 and CYC2 in L. japonicus , CYC1 and CYC2 in pea , CYC1C and CYC2A in Opithandra dinghushanensis , CYC2A and CYC2B in Byrsonima crassifolia (Malpighiaceae) , CYC2A and CYC2B in Lonicera (Caprifoliaceae) , and CYC1C and CYC1D in Primulina heterotricha . Furthermore, it seems very likely that a double positive autoregulatory feedback loops have been employed by CYC genes in different zygomorphic clades, such as in M. truncatula and G. max . Therefore, the expression of VsCYC1 and VsCYC2 might be sufficient to generate morphological zygomorphy in vetch, and these genes may be regulated by the double positive autoregulatory loops, which is a possibility that deserves further research.
Because the three vetch CYC genes are members of the CYC/TB1 clade in the TCP family, we further examined the expression patterns of TCP family genes using genome-scale analysis to investigate the relationship of floral zygomorphy with the other TCP members in vetch. Based on amino acid sequence similarity in the TCP domain, the TCP family is divided into two major clades: Class I and Class II. The latter clade is further subdivided into two additional clades: the CYC/TB1 clade and the CIN clade . Among the three clades, we found that only genes in the CYC/TB1 clade were preferentially expressed in vetch petals (Figure 5). This finding provided useful clues regarding the function of these genes and may indicate that only the CYC/TB1 genes in the TCP family are involved in floral zygomorphy in vetch. The three genes in the CYC/TB1 clade observed in vetch also supported the conclusion that most papilionoid legumes exhibit three kinds of CYC genes (LEGCYC IA, IB, and II), among which the LEGCYC I genes experienced a duplication near the base of or below the papilionoid legumes .
In vetch, 462 unigenes were found to be specifically expressed in the dorsal petals with the similar expression patterns as the VsCYC1 and VsCYC2 genes (Table S7D), indicating that some of the 462 unigenes might be downstream targets of the VsCYC1 and VsCYC2 transcription factors, which could be identified based on the presence or absence of putative CYC binding sites in upstream promoter sequences in the 462 unigenes . By the GO term analyses of 462 dorsal petal specific unigenes in vetch (Table S8A) and the identification of the 104 putative CYC or TCP binding sites in 49 homologous genes in M. truncatula (Table S8B), our results might offer some important clues to the potentially putative downstream genes of the CYC genes in the zygomorphic species. Internal asymmetry is the other important trait associated with zygomorphy, including the internal asymmetry of the lateral and ventral petals in Papilionoideae. A SYP1 (SYMMETRIC PETALS 1) locus was identified from P. sativum, whose function is to establish the internal asymmetry of the petals, and syp1 mutant flowers bearing all symmetric petals without normal internal asymmetry . However, the SYP1 gene has not been cloned from the SYP1 locus. In vetch, we found 435 and 922 unigenes that were specifically expressed in the lateral and ventral petals, respectively (Table S7D). These data provided a number of candidate genes that might allow further elucidation of the molecular mechanisms underlying the internal asymmetry in Papilionoideae.
In the present study, based on de novo assembly and RNA-Seq Quantification analysis, we generated genome-wide gene expression profiles for six floral organs in vetch to investigate the characteristics of the evolution and development of zygomorphic flowers. We isolated 71,553 unigenes with an average length of 511 bp and then compared the changes in the expression of these unigenes in the sepals, dorsal petals, lateral petals, ventral petals, stamens, and carpels, using the leaves and roots as controls. By identifying enriched genes in organwise floral categories, our comparison of the gene expression patterns in vetch and Arabidopsis revealed that the vetch flower conforms to a strict ABC model, similar to the core eudicots Arabidopsis and the basal eudicots Eschscholzia, in contrast to the fading transcriptional boundaries observed in the basal eudicots Aquilegia , , . Additionally, regarding the unigenes enriched in the vetch floral organs, most of the homologous transcripts in the counterpart organs in Arabidopsis did not display enriched expression, and vice versa. We identified and compared the evolution and expression patterns of the TCP gene family members in vetch at a whole-genome scale, and we further obtained a number of unigenes that were specific to the three different types of vetch petals. These results might provide important clues with respect to elucidating the molecular mechanisms underlying the floral zygomorphy in the Papilionoideae. Our data revealed, for the first time, the comprehensive characteristics of the evolution and development of zygomorphic flowers.
The length distribution of the unigenes obtained using Illumina paired-end technology in vetch.
COG functional classification of the vetch unigenes.
Validation of the expression profiles in the vetch organs. a) Heatmap of the expression levels obtained using Illumina sequencing data and real-time quantitative PCR. The bar represents the scale of the expression levels of the unigenes (log2). P denotes the expression detected using real-time PCR, and S denotes the expression detected vis Illumina sequencing. Unigene68614 was used as an internal control in the real-time PCR analyses. b) Semi-quantitative PCR analysis of 49 unigenes using Unigene68614 as an internal control. Corresponding detailed information is displayed in Table S1.
Hierarchical clustering based on the expression patterns in vetch. Cluster analyses for the vetch floral organs using 71,553 unigenes. The color scale ranges from 0.0 (dark) to 44.0314 (yellow).
The distribution of the organ-enriched and downregulated unigenes in vetch with functional annotation. The genes are classified into 14 functional groups. Positive and negative numbers denote the percentages of unigenes enriched and downregulated in different organs, respectively. Corresponding detailed information is listed in Table S4a and S4b (enriched expression), and S4c and S4d (downregulated expression).
Gene Ontology term “enrichment status” for targets of the dorsal petal-specific unigenes. (a–c) Targets with GO term “enrichment status” and “hierarchy” for a) biological process, b) molecular function and c) cellular component branches. The classification terms and their serial numbers are represented as boxes. For significant terms, the box includes the GO term, adjusted P-value (in parentheses), item number mapping the GO term in the query list and background, and total number of items in the query list and background. The color scale shows the P-value cutoff levels for each biological process; the more statistically significant, the darker and redder the color.
The expression patterns of some key vetch genes.
Unigenes examined using quantitative PCR and RT-PCR.
Correlation coefficients between pairs of floral organs in vetch.
Statistics for the enriched and the downregulated transcripts in vetch and Arabidopsis .
Enriched and downregulated unigenes in vetch floral organs and their functional categories.
Six groups of unigenes enriched in individual organs and combinations of organs in vetch.
Comparison of enriched transcription factors between vetch and Arabidopsis .
Homologous comparison of the enriched floral transcripts between vetch and Arabidopsis .
We thank Prof. Hongzhi Kong for insightful comments on the manuscript. We also thank Longfen Yan, Liqin Wei, and Zhiguang Zhao for their technical assistance with the sequence analyses, and we thank Lingfang Kong, Lei Zhang and Miaoqing Zhang for their help in dissecting and collecting floral organs. We are also grateful to two anonymous reviewers for their suggestions regarding this paper.
Conceived and designed the experiments: YRW ZPL ZBN. Performed the experiments: ZPL LCM ZBN. Analyzed the data: ZPL LCM. Contributed reagents/materials/analysis tools: ZPL LCM YRW. Wrote the paper: YRW ZPL ZBN.
- 1. Coen ES, Meyerowitz EM (1991) The war of the whorls: genetic interactions controlling flower development. Nature 353: 31–37.
- 2. Pelaz S, Ditta GS, Baumann E, Wisman E, Yanofsky MF (2000) B and C floral organ identity functions require SEPALLATA MADS-box genes. Nature 405: 200–203.
- 3. Schmid M, Davison TS, Henz SR, Pape UJ, Demar M, et al. (2005) A gene expression map of Arabidopsis thaliana development. Nat Genet 37: 501–506.
- 4. Zahn LM, Ma XA, Altman NS, Zhang Q, Wall PK, et al. (2010) Comparative transcriptomics among floral organs of the basal eudicot Eschscholzia californica as reference for floral evolutionary developmental studies. Genome Biol 11: R101.
- 5. Voelckel C, Borevitz JO, Kramer EM, Hodges SA (2010) Within and between whorls: comparative transcriptional profiling of Aquilegia and Arabidopsis. PLoS ONE 5: e9735.
- 6. Chanderbali AS, Albert VA, Leebens-Mack J, Altman NS, Soltis DE, et al. (2009) Transcriptional signatures of ancient floral developmental genetics in avocado (Persea americana; Lauraceae). PNAS 106: 8929–8934.
- 7. Yoo MJ, Chanderbali AS, Altman NS, Soltis PS, Soltis DE (2010) Evolutionary trends in the floral transcriptome: insights from one of the basalmost angiosperms, the water lily Nuphar advena (Nymphaeaceae). Plant J 64: 687–698.
- 8. Doyle JJ, Luckow MA (2003) The rest of the iceberg. Legume diversity and evolution in a phylogenetic context. Plant Physiol 131: 900–910.
- 9. Tucker SC (2003) Floral development in legumes. Plant Physiol 131: 911–926.
- 10. Darwin CR (1857) Bees and fertilisation of kidney beans. Gard Chron Agric Gaz 43: 725.
- 11. Hofer J, Turner L, Hellens R, Ambrose M, Matthews P, et al. (1997) UNIFOLIATA regulates leaf and flower morphogenesis in pea. Curr Biol 7: 581–587.
- 12. Ferrandiz C, Navarro C, Gomez MD, Canas LA, Beltran JP (1999) Flower development in Pisum sativum: from the war of the whorls to the battle of the common primordia. Dev Genet 25: 280–290.
- 13. Benloch R, Navarro C, Beltran JP, Canas LA (2003) Floral development of the model legume Medicago truncatula: ontogeny studies as a tool to better characterize homeotic mutations. Sex Plant Reprod 15: 231–241.
- 14. Benlloch R, d'Erfurth I, Ferrandiz C, Cosson V, Beltran JP, et al. (2006) Isolation of mtpim proves Tnt1 a useful reverse genetics tool in Medicago truncatula and uncovers new aspects of AP1-like functions in legumes. Plant Physiol 142: 972–983.
- 15. Foucher F, Morin J, Courtiade J, Cadioux S, Ellis N, et al. (2003) DETERMINATE and LATE FLOWERING are two TERMINAL FLOWER1/CENTRORADIALIS homologs that control two distinct phases of flowering initiation and development in pea. Plant Cell 15: 2742–2754.
- 16. Dong ZC, Zhao Z, Liu CW, Luo JH, Yang J, et al. (2005) Floral patterning in Lotus japonicus. Plant Physiol 137: 1272–1282.
- 17. Citerne HL, Pennington RT, Cronk QCB (2006) An apparent reversal in floral symmetry in the legume Cadia is a homeotic transformation. PNAS 103: 12017–12020.
- 18. Domoney C, Duc G, Ellis THN, Ferrandiz C, Firnhaber C, et al. (2006) Genetic and genomic analysis of legume flowers and seeds. Curr Opin Plant Biol 9: 133–141.
- 19. Feng XZ, Zhao Z, Tian ZX, Xu SL, Luo YH, et al. (2006) Control of petal shape and floral zygomorphy in Lotus japonicus. PNAS 103: 4970–4975.
- 20. Wang HL, Chen JH, Wen JQ, Tadege M, Li GM, et al. (2008) Control of compound leaf development by FLORICAULA/LEAFY ortholog SINGLE LEAFLET1 in Medicago truncatula. Plant Physiol 146: 1759–1772.
- 21. Wang Z, Luo YH, Li X, Wang LP, Xu SL, et al. (2008) Genetic control of floral zygomorphy in pea (Pisum sativum L.). PNAS 105: 10414–10419.
- 22. Yan J, Cai XF, Luo JH, Sato S, Jiang QY, et al. (2010) The REDUCED LEAFLET genes encode key components of the trans-acting small interfering RNA pathway and regulate compound leaf and flower development in Lotus japonicus. Plant Physiol 152: 797–807.
- 23. Yang X, Pang HB, Liu BL, Qiu ZJ, Gao Q, et al. (2012) Evolution of double positive autoregulatory feedback loops in CYCLOIDEA2 clade genes is associated with the origin of floral zygomorphy. Plant Cell 24: 1834–1847.
- 24. Hogslund N, Radutoiu S, Krusell L, Voroshilova V, Hannah MA, et al. (2009) Dissection of symbiosis and organ development by integrated transcriptome analysis of Lotus japonicus mutant and wild-type plants. PLoS ONE 4: e6556.
- 25. Benedito VA, Torres-Jerez I, Murray JD, Andriankaja A, Allen S, et al. (2008) A gene expression atlas of the model legume Medicago truncatula. Plant J 55: 504–513.
- 26. Libault M, Farmer A, Joshi T, Takahashi K, Langley RJ, et al. (2010) An integrated transcriptome atlas of the crop model Glycine max, and its use in comparative analyses in plants. Plant J 63: 86–99.
- 27. Hoen PAC, Ariyurek Y, Thygesen HH, Vreugdenhil E, Vossen R, et al. (2008) Deep sequencing-based expression analysis shows major advances in robustness, resolution and inter-lab portability over five microarray platforms. Nucleic Acids Res 36: e141.
- 28. Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y (2008) RNA-seq: An assessment of technical reproducibility and comparison with gene expression arrays. Genome Res 18: 1509–1517.
- 29. Asmann YW, Klee EW, Thompson EA, Perez EA, Middha S, et al. (2009) 3 ' tag digital gene expression profiling of human brain and universal reference RNA using Illumina Genome Analyzer. BMC Genomics 10: 531.
- 30. Sanders PM, Bui AQ, Weterings K, McIntire KN, Hsu YC, et al. (1999) Anther developmental defects in Arabidopsis thaliana male-sterile mutants. Sex Plant Reprod 11: 297–322.
- 31. Li RQ, Zhu HM, Ruan J, Qian WB, Fang XD, et al. (2010) De novo assembly of human genomes with massively parallel short read sequencing. Genome Res 20: 265–272.
- 32. Pertea G, Huang XQ, Liang F, Antonescu V, Sultana R, et al. (2003) TIGR Gene Indices clustering tools (TGICL): a software system for fast clustering of large EST datasets. Bioinformatics 19: 651–652.
- 33. Conesa A, Gotz S, Garcia-Gomez JM, Terol J, Talon M, et al. (2005) Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics 21: 3674–3676.
- 34. Wei LQ, Yan LF, Wang T (2011) Deep sequencing on genome-wide scale reveals the unique composition and expression patterns of microRNAs in developing pollen of Oryza sativa. Genome Biol 12: R53.
- 35. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B (2008) Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 5: 621–628.
- 36. Wei LQ, Xu WY, Deng ZY, Su Z, Xue YB, et al. (2010) Genome-scale analysis and comparison of gene expression profiles in developing and germinated pollen in Oryza sativa. BMC Genomics 11: 338.
- 37. Li MN, Xu WY, Yang WQ, Kong ZS, Xue YB (2007) Genome-wide gene expression profiling reveals conserved and novel molecular functions of the stigma in rice. Plant Physiol 144: 1797–1812.
- 38. Zhou D, Zhou X, Ling Y, Zhang ZH, Zhen S (2010) agriGO: a GO analysis toolkit for the agricultural community. Nucleic Acids Res 38: 64–70.
- 39. Higo K, Ugawa Y, Iwamoto M, Korenaga T (1999) Plant cis-acting regulatory DNA elements (PLACE) database:1999. Nucleic Acids Res 27: 297–300.
- 40. Chanderbali AS, Yoo MJ, Zahn LM, Brockington SF, Wall PK, et al. (2010) Conservation and canalization of gene expression during angiosperm diversification accompany the origin and evolution of the flower. PNAS 107: 22570–22575.
- 41. Martin-Trillo M, Cubas P (2010) TCP genes: a family snapshot ten years later. Trends in Plant Science 15: 31–39.
- 42. Zik M, Irish VF (2003) Global identification of target genes regulated by APETALA3 and PISTILLATA floral homeotic gene action. Plant Cell 15: 207–222.
- 43. Wellmer F, Riechmann JL, Alves-Ferreira M, Meyerowitz EM (2004) Genome-wide analysis of spatial gene expression in Arabidopsis flowers. Plant Cell 16: 1314–1326.
- 44. Bowman JL, Smyth DR, Meyerowitz EM (1991) Genetic interactions among floral homeotic genes of Arabidopsis. Development 112: 1–20.
- 45. Sommer H, Nacken W, Beltran P, Huijser P, Pape H, et al. (1991) Properties of deficiens, a homeotic gene involved in the control of flower morphogenesis in Antirrhinum majus. Development 9: 169–175.
- 46. Zachgo S, Silva ED, Motte P, Trobner W, Saedler H, et al. (1995) Functional-analysis of the Antirrhinum floral homeotic deficiens gene in-vivo and in-vitro by using a temperature-sensitive mutant. Development 121: 2861–2875.
- 47. Angenent GC, Franken J, Busscher M, Colombo L, Vantunen AJ (1993) Petal and stamen formation in petunia is regulated by the homeotic gene fbp1. Plant J 4: 101–112.
- 48. Kang HG, Jeon JS, Lee S, An GH (1998) Identification of class B and class C floral organ identity genes from rice plants. Plant Mol Biol 38: 1021–1029.
- 49. Fornara F, Parenicova L, Falasca G, Pelucchi N, Masiero S, et al. (2004) Functional characterization of OsMADS18, a member of the AP1/SQUA subfamily of MADS box genes. Plant Physiol 135: 2207–2219.
- 50. Yamaguchi T, Lee DY, Miyao A, Hirochika H, An GH, et al. (2006) Functional diversification of the two C-class MADS box genes OSMADS3 and OSMADS58 in Oryza sativa. Plant Cell 18: 15–28.
- 51. Yu DY, Kotilainen M, Pollanen E, Mehto M, Elomaa P, et al. (1999) Organ identity genes and modified patterns of flower development in Gerbera hybrida (Asteraceae). Plant J 17: 51–62.
- 52. Ambrose BA, Lerner DR, Ciceri P, Padilla CM, Yanofsky MF, et al. (2000) Molecular and genetic analyses of the silky1 gene reveal conservation in floral organ specification between eudicots and monocots. Mol Cell 5: 569–579.
- 53. Taylor SA, Hofer JMI, Murfet IC, Sollinger JD, Singer SR, et al. (2002) PROLIFERATING INFLORESCENCE MERISTEM, a MADS-box gene that regulates floral meristem identity in pea. Plant Physiol 129: 1150–1159.
- 54. Yu H, Ito T, Wellmer F, Meyerowitz EM (2004) Repression of AGAMOUS-LIKE 24 is a crucial step in promoting flower development. Nat Genet 36: 157–161.
- 55. Kramer EM, Hall JC (2005) Evolutionary dynamics of genes controlling floral development. Curr Opin Plant Biol 8: 13–18.
- 56. Causier B, Schwarz-Sommer Z, Davies B (2010) Floral organ identity: 20 years of ABCs. Semin Cell Dev Biol 21: 73–79.
- 57. Becker A, Kaufmann K, Freialdenhoven A, Vincent C, Li MA, et al. (2002) A novel MADS-box gene subfamily with a sister-group relationship to class B floral homeotic genes. Mol Genet Genomics 266: 942–950.
- 58. Endress PK (2001) The flowers in extant basal angiosperms and inferences on ancestral flowers. Int J Plant Sci 162: 1111–1140.
- 59. Ronse DCLP (2007) Are petals sterile stamens or bracts? the origin and evolution of petals in the core eudicots. Ann Bot-London 100: 621–630.
- 60. Derstine KS, Tucker SC (1991) Organ initiation and development of inflorescences and flowers of Acacia baileyana. Am J Bot 78: 816–832.
- 61. Liao BY, Zhang JZ (2006) Evolutionary conservation of expression profiles between human and mouse orthologous genes. Mol Biol Evol 23: 530–540.
- 62. Singer GAC, Lloyd AT, Huminiecki LB, Wolfe KH (2005) Clusters of co-expressed genes in mammalian genomes are conserved by natural selection. Mol Biol Evol 22: 767–775.
- 63. Zhang JZ, He XL (2005) Significant impact of protein dispensability on the instantaneous rate of protein evolution. Mol Biol Evol 22: 1147–1155.
- 64. Yanai I, Hunter CP (2009) Comparison of diverse developmental transcriptomes reveals that coexpression of gene neighbors is not evolutionarily conserved. Genome Res 19: 2214–2220.
- 65. Parikh A, Miranda ER, Katoh-Kurasawa M, Fuller D, Rot G, et al. (2010) Conserved developmental transcriptomes in evolutionarily divergent species. Genome Biol 11: R35.
- 66. Luo D, Carpenter R, Vincent C, Copsey L, Coen E (1996) Origin of floral asymmetry in Antirrhinum. Nature 383: 794–799.
- 67. Luo D, Carpenter R, Copsey L, Vincent C, Clark J, et al. (1999) Control of organ asymmetry in flowers of Antirrhinum. Cell 99: 367–376.
- 68. Hileman LC, Kramer EM, Baum DA (2003) Differential regulation of symmetry genes and the evolution of floral morphologies. PNAS 100: 12814–12819.
- 69. Song CF, Lin QB, Liang RH, Wang YZ (2009) Expressions of ECE-CYC2 clade genes relating to abortion of both dorsal and ventral stamens in Opithandra (Gesneriaceae). BMC Evol Biol 9: 244.
- 70. Zhang WH, Kramer EM, Davis CC (2010) Floral symmetry genes and the origin and maintenance of zygomorphy in a plant-pollinator mutualism. PNAS 107: 6388–6393.
- 71. Howarth DG, Martins T, Chimney E, Donoghue MJ (2011) Diversification of CYCLOIDEA expression in the evolution of bilateral flower symmetry in Caprifoliaceae and Lonicera (Dipsacales). Ann Bot-London 107: 1521–1532.
- 72. Citerne HL, Luo D, Pennington RT, Coen E, Cronk QCB (2003) A phylogenomic investigation of CYCLOIDEA-like TCP genes in the Leguminosae. Plant Physiol 131: 1042–1053.
- 73. Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: Molecular evolutionary genetics analysis (MEGA) software version 4.0. Mol Biol Evol 24: 1596–1599.