- Open Access
The developmental and genetic bases of apetaly in Bocconia frutescens (Chelidonieae: Papaveraceae)
EvoDevo volume 7, Article number: 16 (2016)
Bocconia and Macleaya are the only genera of the poppy family (Papaveraceae) lacking petals; however, the developmental and genetic processes underlying such evolutionary shift have not yet been studied.
We studied floral development in two species of petal-less poppies Bocconia frutescens and Macleaya cordata as well as in the closely related petal-bearing Stylophorum diphyllum. We generated a floral transcriptome of B. frutescens to identify MADS-box ABCE floral organ identity genes expressed during early floral development. We performed phylogenetic analyses of these genes across Ranunculales as well as RT-PCR and qRT-PCR to assess loci-specific expression patterns. We found that petal-to-stamen homeosis in petal-less poppies occurs through distinct developmental pathways. Transcriptomic analyses of B. frutescens floral buds showed that homologs of all MADS-box genes are expressed except for the APETALA3-3 ortholog. Species-specific duplications of other ABCE genes in B. frutescens have resulted in functional copies with expanded expression patterns than those predicted by the model.
Petal loss in B. frutescens is likely associated with the lack of expression of AP3-3 and an expanded expression of AGAMOUS. The genetic basis of petal identity is conserved in Ranunculaceae and Papaveraceae although they have different number of AP3 paralogs and exhibit dissimilar floral groundplans.
Flowers are evolutionary novelties of nearly 300,000 species. A typical flower has one or two whorls of sterile organs, collectively known as the perianth [1–3], which enclose the stamens and the carpels. Frequently, the perianth consists of an outer series of sepals (collectively termed calyx) and an inner series of petals (collectively termed corolla). Sepals often resemble leaves or bracts in that they are persistent, protect the rest of the floral organs in the bud and often are photosynthetic, while petals are colorful, attract pollinators, are rich in pigments, oils and waxes [3, 4]. The number and morphology of the perianth parts are extremely variable and are often associated with changes in the mechanisms of pollination .
The Ranunculales, the largest order in basal eudicots, exhibit an incredible array of floral forms which range from perianth-less flowers (like in Eupteleaceae) to flowers with bipartite perianths (like in Papaveraceae and Ranunculaceae), sepaloid tepals (like in Menispermaceae) and petaloid tepals (like in Lardizabalaceae and Berberidaceae) [6–8]. With 760 species in 41 genera, the poppy family (Papaveraceae) is the second largest of the order Ranunculales. As currently circumscribed the Papaveraceae s. str. comprise the subfamilies Fumarioideae, with the tribes Fumarieae and Hypecoeae, and Papaveroideae, with the tribes Chelidonieae, Eschscholzieae, Fumarioideae, Papavereae and Platystemoneae . All Papaveraceae exhibit a dimerous floral plan with 2 deciduous sepals, 4 petals in two whorls, many stamens (up to 700) and 2(−8) fused carpels (Fig. 1a–d) . The only exceptions to this floral groundplan are the naturally occurring mutants in Sanguinaria canadensis L., which possess eight petals in four whorls as a result of a stamen-to-petal homeosis, and all species in the sister genera Bocconia Plum. ex L., and Macleaya R. Br., which lack petals likely as a result of a petal-to-stamen homeosis (Fig. 1e–m) [10, 11]. This variation provides an excellent comparative morphological platform for studying the genetic basis of petal identity in non-model organisms. The framework for comparison is the ABCE genetic model of flower development established in Arabidopsis thaliana that describes how specific combinations of four classes of MADS-box transcription factors A, B, C and E determine the identity of all floral organs [12–14]. However, all the MADS-box gene lineages have suffered duplications during the course of angiosperm evolution, making the model difficult to extrapolate to non-model flowering plants [15–20]. According to the model, petal identity depends on the combination of A and B and E class genes and the repression of C class genes [12, 14, 21]. Both A and B gene lineages have undergone duplications in the core eudicots, basal eudicots and monocots [16, 22, 23]. In basal eudicots, local duplications in both gene lineages have also been detected [23–25]. A class genes duplicated once, before the diversification of the Ranunculales, resulting in two FUL-like clades named RanFL1 and RanFL2 . Gene copies have been functionally characterized in Eschscholzia californica Cham. and Papaver somniferum L. (Papaveraceae) and in Aquilegia coerulea E. James. (Ranunculaceae). FUL-like genes in Papaveraceae control flowering time, inflorescence architecture, floral meristem, sepal identity and fruit development. Only papsfl1-/fl2-downregulated individuals show atypical green areas in the outer petals, but have normal inner petals . In Ranunculaceae, AqcFL1A and AqcFL1B (RanFL1 clade) play roles in inflorescence architecture and leaf morphogenesis, but their role in sepal and petal development is unclear . On the other hand, AP2-like genes are expressed during carpel, fruit and leaf development and possibly do not control perianth development .
B class genes duplicated two times consecutively in the Ranunculales resulting in the APETALA3I (AP3I), AP3II and AP3III clades [7, 21, 25]. However, the Papaveraceae only have members of the AP3I and AP3III clades, whereas the Ranunculaceae have gene representatives in all clades . Gene copies have been functionally characterized in Papaver somniferum, Nigella damascena and Aquilegia coerulea and in general have shown subfunctionalization [25, 28–30]. In P. somniferum PapsAP3-1 is responsible for the identity of petals, whereas PapsAP3-2 controls the identity of stamens. In A. coerulea, AqAP3-1 specifies the identity of staminodia, whereas AqAP3-2 is responsible for identity of stamens and AqAP3-3 specifically controls petal identity [25, 29, 31]. Finally, in N. damascena NdAP3-3 is responsible for petal and outer stamen identity, while NdAP3-1 and NdAP3-2 control stamen identity [30, 32].
This work aims to identify the developmental and genetic mechanisms responsible for petal loss in Bocconia frutescens and Macleaya cordata, having as references the typical Papaveraceae floral groundplan exhibited by E. californica, P. somniferum and Stylophorum diphyllum. Here we report the floral ontogeny of B. frutescens and M. cordata, and we present data on gene copy number and expression of A, B and C class gene homologs of B. frutescens to explain the genetic underpinning of apetaly. Finally, we also present expression data of the E class genes in the model to test how they change their typical expression patterns in the context of petal loss.
Material for SEM studies was collected in FAA or EtOH 70 % from the following taxa: Bocconia frutescens (voucher: Colombia, Antioquia, Llanos de Cuivá, March 05, 2013, N. Pabón-Mora and C. Arango-Ocampo 292, HUA); Macleaya cordata (voucher: USA, New York, Bronx, Lehman College, Davis Hall, April 09, 2011N. Pabón-Mora 250, NY); and Stylophorum diphyllum (voucher: USA, New York, Bronx, Living collections, New York Botanical Garden, April 16, 2011, N. Pabón-Mora 254, NY).
Material for gene expression studies was collected in liquid nitrogen directly from field collections from Bocconia frutescens (voucher: Colombia, Antioquia, Rionegro, August 1, 2014, N. Pabón-Mora and C. Arango-Ocampo 337, HUA). Material was collected separately from floral buds, anthetic flowers, fruits and leaves. Floral buds were collected in four different developmental stages: The first (S0) is characterized by having stigmas inserted into the green sepals, the second (S1) is characterized by having slightly exerted stigmas through the green sepals, the third (S2) is characterized by having fully exerted, non-rotated stigmas over the green sepals, and the fourth (S3) is characterized by having fully exerted rotated (90°) stigmas over the pink sepals. For stages S1, S2, S3 floral organs were also dissected into sepals, stamens and carpels. Fruits were separated into young green fruits (Fr1) and mature yellow fruits (Fr2) (Figs. 1e–h, 8a–i). Biological replicates of these collections were made to corroborate gene expression via qRT-PCR.
Vegetative and flowering shoots of B. frutescens, M. cordata and S. diphyllum (Figs. 1, 2, 3) were fixed for 48 h in formalin–acetic acid–ethanol (3.7 % formaldehyde, 5 % glacial acetic acid, 50 % ethanol) and then transferred and stored in 70 % ethanol. For SEM, buds were dissected in 90 % ethanol under a Zeiss Stemi DV4 stereoscope and then dehydrated in an absolute ethanol/acetone series. Dehydrated material was then critical point-dried using a Balzer 790 CPD (Balzer Union, Furstentum, Liechtenstein, Rockville, MD), coated with gold and palladium in a 6.2 sputter coater at the New York Botanical Garden and examined using a JEOL (Tokyo, Japan) JSM-5410 LV SEM at 10 kV, onto which digital images were stored. Field photographs were taken with a Nikon FM-2 camera.
Fresh inflorescences carrying floral buds and flowers during organ initiation (Fig. 3), as well as young flowers in S0–S4 (see above in “Plant material”) and flowers in anthesis from three different cultivated plants, were mixed together and were ground using liquid nitrogen. Further total RNA extraction was carried out using TRIzol reagent (Invitrogen). The RNAseq experiment was conducted using the truseq mRNA library construction kit (Illumina) and sequenced in a HiSeq 2000 instrument reading 100 bases paired end reads. A total of 25,185,945 raw read pairs were obtained. The transcriptome was assembled de novo. Read cleaning was performed with prinseq-lite with a quality threshold of Q35, and contig assembly was computed using Trinity package following default settings. Contig metrics are as follows: total assembled bases: 149,710,500; total number of contigs (>101 bases): 211,821; average contig length: 706 b; largest contig: 17,004 b; contig N50: 1877 bp; contig GC%: 40.09; number of Ns: 0. Orthologous gene search was performed using BLASTN  with the reference sequences downloaded from GenBank (see below). Protein domain searches were carried out using HMMER and the PFAM database downloaded from the Danger Institute FTP server. Top hits for each sequence were individually analyzed and confirmed by protein domain searches, prior to phylogenetic analysis.
Cloning and gene characterization
For each of the gene families, searches on the transcriptome were performed by using other Ranunculales sequences as a query to identify a first batch of homologs using Blast tools . Query sequences for FRUITFULL-like genes came from ; for APETALA3 and PISTILLATA genes came from [7, 15, 21, 25]; for AGAMOUS genes came from [18, 20]; and for SEPALLATA genes came from . After a search for homologous genes in the B. frutescens transcriptome we were able to identify 3 FUL-like (FUL), 1 APETALA3 (AP3), 4 PISTILLATA (PI), 1 AGAMOUS (AG) and 5 SEPALLATA (SEP) homologs. We named them BofrFL1, BofrFL2, BofrFL3, BofrAP3, BofrPI1, BofrPI2, BofrPI3, BofrPI4, BofrAG, BofrSEP1-1, BofrSEP1-2, BofrSEP2, BofrSEP3-1 and BofrSEP3-2. Sequences are deposited in GenBank with the accession numbers: KF500160, KF500161, KX574344-KX574356.
Sequences in the transcriptome were added to each dataset consisting of sequences available from NCBI, as well as genes available in Phytozome, specifically in the Aquilegia coerulea genome (http://www.phytozome.net/). Other homologs for all MADS-box gene lineages studied here were isolated from the 1kp transcriptome database (http://188.8.131.52/Blast4OneKP/home.php), Phytometasyn (http://www.phytometasyn.ca) and Plantrans DB (http://lifecenter.sgst.cn/plantransdb) after a search in the available transcriptomes of members of the Ranunculales (Additional file 1: Table S1). A matrix for each gene lineage was generated. All sequences were then compiled using Bioedit (http://www.mbio.ncsu.edu/bioedit/bioedit.html), where they were cleaned to keep exclusively the open reading frame. Nucleotide sequences were then aligned using the online version of MAFFT (http://mafft.cbrc.jp/alignment/server/) , with a gap open penalty of 3.0, an offset value of 0.8 and all other default settings. The alignment was then refined by hand using Bioedit taking into account the protein domains and amino acid motifs that have been reported as conserved for the five gene lineages. Maximum likelihood (ML) phylogenetic analyses using the nucleotide sequences were performed in RaxML-HPC2 BlackBox  on the CIPRES Science Gateway . The best performing evolutionary model was obtained by the Akaike information criterion, using ModelTest incorporated in MEGA6 . Bootstrapping was performed according to the default criteria in RAxML where bootstrapping stopped after 200–600 replicates when the criteria were met. Trees were observed and edited using FigTree v1.4.0.
For RT-PCR total RNA was prepared using TRIzol® reagent (Ambion, Grand Island, NY, USA) from all of the dissected organs and stages mentioned above. Genomic DNA contamination was removed using DNaseI (RNase-free, Austin, TX, USA) following the manufacturer instructions. First-strand cDNA was synthesized from 5 µg of total RNA using the SuperScript® III First-Strand Synthesis System (Invitrogen, Grand Island, NY, USA) with the oligodT20 primers, following the manufacturer instructions. PCRs were carried out using EconoTaq® Plus (Lucigen, Middleton, USA) following the manufacturer protocol using 1 µl of cDNA as template in a total volume of 20 µl. We manually designed primers to specifically amplify each gene copy targeting sequences unique to each gene in the K and C domains ranging between 160 and 360nt (Additional file 2: Table S2). Specific primers were hard to design for the PISTILLATA copies as BofrPI1 and BofrPI3 on the one hand and BofrPI2 and BofrPI4 on the other, shared identical amino acid sequences with few nucleotide changes and small insertions (30nt) in BofrPI1 and BofrPI2. After multiple attempts to separate double bands in agarose gels and confirmation through sequencing we were able to obtain BofrPI1 and BofrPI3 independent products but were not able to distinguish between BofrPI2 and BofrPI4. The following PCR program was used: 2 min at 94 °C; followed by 27–32 cycles of 30 s at 94 °C, 30 s at 55 °C and 1 min at 72 °C; followed by 10 min at 72 °C. A total of 3–4 µl of each PCR product were run on a 1 % agarose gel and digitally photographed using a Transilluminator Biometra Ti5 BioDocAnalyze.
For quantitative real-time PCR (qRT-PCR) total RNA was prepared from biological replicates with the same protocols as for RT-PCR. cDNA was synthesized from 3 µg of total RNA. Selected organs for qRT-PCR included sepals, stamens and carpels, from early S1 stages, late S3 stages and anthesis as well as late developing fruits, as they represent the entire variation for gene expression detected by RT-PCR. The resulting cDNA was diluted 1:4. PCR product was amplified using locus-specific primers designed by introducing selected paralog-specific regions into online tools available at https://www.genscript.com/ssl-bin/app/primer (Additional file 2: Table S2). For the Bocconia frutescens PISTILLATA1/3 and PI2/4 paralogous copies specific amplification was complicated by extensive sequence similarity, so they were amplified by pairs and not individually. Quantitative PCR was performed as described in . Glyceraldehyde 3-phosphate dehydrogenase (GADPH), ACTIN2 (ACT2) and ELONGATION FACTOR 1α (EF1α) were tested as putative endogenous controls, and GADPH was selected as the control for gene target quantification as it showed the lowest variation of Ct values across organs. The level of the ABC homologs BofrAP3, BofrPI1/3, BofrPI2/4, BofrAG, BofrFL2 and BofrFL3 was analyzed relative to GADPH using the 2−ΔΔCt method , and the relative level across organs was set to be compared to the organ with the highest expression detected per gene arbitrarily set to 1. The expression of BofrFL1 was not examined as RT-PCR results already suggested low expression. Quantitative RT-PCRs were performed using the qTower2.2 qPCR system and the qPCRsoft software (Analytik Jena).
Flower development of Stylophorum diphyllum
Flower development of S. diphyllum begins with the simultaneous initiation of two sepal primordia from a radial flower primordium (Fig. 2a, b). The two outer petal primordia arise simultaneously from flat primordia that alternate with the sepal primordia; then, the two inner petals initiate (Fig. 1c, d, 2c–e). The four petal primordia grow slowly during early stages of floral development (Fig. 2d,e). Flower development continues with the centripetal initiation of four early isodiametric stamen primordia arranged in an outer whorl that gets filled in with 10–12 staminal primordia, then a second whorl of 6–10 primordia that alternate with the outer ones and 5–10 primordia that form the inner whorl; sometimes, the latter ones are roughly arranged in two, instead of one (Figs. 1d; 2c–e). Mature stamens are formed by an upright filament and a short flattened anther, both yellow (Fig. 1c). Flower development ends with the formation of four fused carpels with trichomes externally and four massive commissural placentas internally (Figs. 1c,d, 2e–h).
Flower development of Bocconia frutescens
Each inflorescence of B. frutescens consists of groups of 1–3 pedicellate flowers in which the terminal one develops first and is larger than the lateral ones (Fig. 3a–c). Floral primordia are hemispherical. Flower development begins with the initiation of two synchronous sepals (Fig. 3b–d); rarely, three sepals are formed in the terminal flower (Additional file 3: Fig. S1). Then, two simultaneous stamens initiate alternating with the sepals in the position of the first two petals (Figs. 1i, 3e, f). Thus, we consider these two stamens to be homeotic (see “Discussion”). No primordia develop opposite to the sepals, where the second whorl of petals is expected by comparison with closely related petalous species (Figs. 1i, 3h, j). Next, the initiation of 4 stamens is evident in the fourth whorl, followed by 5 or 6 additional stamens sometimes developed from a common primordium (Fig. 3h, i) opposite to the sepals (Fig. 3i). Then, a second whorl of 3–7 stamens develops alternating with the previous one (Fig. 3j–m). The number and arrangement of the stamens are variable, but they appear to occur in whorls rather than in spiral, although the second whorl of stamens often possesses less stamens in comparison with the first one (Fig. 3j–m). The number of stamens is often higher in the terminal flowers reaching up to 21 (Additional file 3: Fig. S1). The two carpel primordia develop next subtended by a gynophore that grows conspicuously during preanthesis (Fig. 3j–o). Closure of the carpels occurs late, accompanied by the formation of the dehiscent horse-shoe-shaped valve (Fig. 3m–o). In addition, the style and the stigma grow and differentiate conspicuously. Carpels at preanthesis are three times the total length of the stamens (Fig. 3l). Each mature stamen consists of a filiform filament and an elongated anther that occupies the 2/3-s distal-most portion of the stamen (Fig. 1g, h). Final changes during preanthesis include the protrusion of the stigmas through the sepals, the orientation of the stigmatic lobes toward the carpel commissures and the change in the color of the sepals from green to pink (Fig. 1e, f). During anthesis the sepals are caducous, and all stamens (both homeotic and true) are pendulous and produce copious pollen (Fig. 1g, h).
Flower development of Macleaya cordata
The inflorescence of M. cordata consists of groups of 1–3 pedicellate flowers in which the terminal one develops first (Fig. 4a). The floral primordium is hemispherical in shape (Fig. 4a); the primordium of the terminal flower is often larger than that of the lateral flowers; however, unlike B. frutescens, flower development of the terminal and the lateral flowers in M. cordata is identical. Flower development begins with the initiation of two synchronous sepals (Fig. 4b). This is followed by the initiation of the four homeotic stamens in two decussate whorls; the first whorl is formed alternating with the sepals, and the second whorl is formed opposite to the sepals in the positions where petals are expected to occur (Figs. 1m, 4b–e). True stamens (4) initiate next in a single whorl (Fig. 4e–g), and unlike B. frutescens, these are the only stamens formed (Fig. 1m). Thus, M. cordata flowers consistently have a total of eight stamens. Flower development ends with the differentiation of a two carpellate gynoecium, which, unlike that of B. frutescens, lacks a gynophore (Fig. 4g–j). Closure of the two carpels occurs late in development, and no tissue differentiation occurs between valves and commissural tissue (Fig. 4j). At preanthesis, carpels reach less than half of the total length of the stamens (Fig. 4i). Final changes during preanthesis include the change of the color of the sepals from green to yellow (Fig. 1j–m). During anthesis the sepals are caducous, and the stamens are pendulous (Fig. 1l).
Evolution of the APETALA3 and PISTILLATA gene lineages in Ranunculales
The maximum likelihood (ML) AP3 analysis recovered three deeply conserved, paralogous lineages, termed AP3-I, AP3-II and AP3-III in the Ranunculales. The AP3-III clade is sister to AP3-I and AP3-II and has been functionally assigned to a petal-specific role. The Papaveraceae possesses gene members only in the AP3-I and AP3-III clades; so far, AP3-II genes have not been recovered from any Papaveraceae species [7, 25, 39] (Fig. 5a). Sampling within Papaveraceae has been scarce, especially in the Chelidonieae, and AP3 genes have been found only in Bocconia frutescens, Sanguinaria canadensis and Stylophorum diphyllum (Fig. 5a). Unlike in other closely related petalous species, the AP3-III putative copy was not found in the floral transcriptome of B. frutescens, which suggests that AP3-III is not expressed. The remaining AP3-I ortholog, named BofrAP3-I, is expressed predominantly in stamens, carpels and fruits. In addition, BofrAP3 is turned on in sepals when they switch color from green to pink (Fig. 8j, k).
The PISTILLATA ML analysis shows four PI duplicates unique to B. frutescens, namely, BofrPI-1, BofrPI-2, BofrPI-3 and BofrPI-4 (Fig. 5b). Taxon-specific duplications seem to have occurred also in other species, such as Berberis gilgiana (Berberidaceae), and the Ranunculaceae Xantorrhiza simplicissima, Helleborus hybrida and species of Adonis, Anemone, Ranunculus, Thalictrum and Trollium (Fig. 5b) . Results on gene expression of the four BofrPI orthologs are mainly restricted to stamens and only sometimes extend to sepals (BofrPI1, BofrPI3), carpels (BofrPI3) and leaves (BofrPI2/4) (Fig. 8j, l, m). BofrPI1 and BofrPI2/4 have opposite expression levels, as stamens grow older, while BofrPI1 increases, BofrPI2/4 levels decrease (Fig. 8j).
Evolution of the APETALA1/FRUITFULL gene lineage in Ranunculales
ML analysis of the AP1/FUL genes recovered a single duplication early in the diversification of the Ranunculales resulting in two clades of FUL-like genes, named RanFL1 and RanFL2 . Bootstrap support for the RanFL1 and RanFL2 clades is low (>50); however, gene copies from the same family within each clade are grouped together with strong support, and the relationships among gene clades are mostly consistent with the phylogenetic relationships of the sampled taxa . The Papaveraceae have representative genes in both clades. All members of Chelidonieae have two copies, except for the taxa-specific duplications in M. cordata and B. frutescens, which have four and three copies, respectively. MacoFL1 and MacoFL2 are orthologs of BofrFL3 in the RanFL1 clade, whereas MacoFL3 and MacoFL4 are orthologs of BofrFL1 and BofrFL2 in the RanFL2 clade (Fig. 6a). Expression patterns for FUL-like homologs are variable. BofrFL3 is broadly expressed in sepals, stamens, carpels and fruits, BofruFL2 is mostly restricted to sepals and carpels, and finally BofrFL1 is expressed at very low levels predominantly in sepals, stamens and leaves (Fig. 8j,o,p).
Evolution of the SEPALLATA gene lineage in Ranunculales
The ML analysis of SEP genes shows three duplication events in the diversification of the Ranunculales, resulting in three clades, the SEP3 clade and two LOF-SEP clades (LOF-SEP1 and LOF-SEP1) (Fig. 6b). B. frutescens has the following five paralogs: BofrSEP1-1 and BofrSEP1-2, which belong to the LOF-SEP1 clade; BofrSEP2, which belongs the LOF-SEP2 clade; and BofrSEP3-1 and BofrSEP3-2, from the SEP3 clade. BofrSEP2 has a broad expression in all floral organs and fruits in every developmental stage. BofrSEP1-1 and BofrSEP1-2 have overlapping expression in stamens and carpels, but the expression of BofrSEP1-1 extends to sepals, fruits and leaves (Fig. 8j). On the other hand, BofrSEP3-1 and BofrSEP3-2 are expressed in stamens, carpels, fruits and leaves, but while BofrSEP3-2 is strongly expressed, the levels of BofrSEP3-1 are almost undetected (Fig. 8j).
Evolution of the AGAMOUS gene lineage in Ranunculales
The ML reconstruction for AG genes exhibits a large-scale duplication previous (at least) to the diversification of the Ranunculaceae as well as numerous species-specific duplications in the Papaveraceae (Fig. 7). However, a single copy of AG was found in B. frutescens. BofrAG1 is expressed predominantly in stamens and carpels in all stages of flower development, but an extended expression was detected in sepals during S1, S2 and S3 and in young fruits (Fig. 8j, n).
Early floral development in Bocconia frutescens, Macleaya cordata and Stylophorum diphyllum is strikingly similar to all species of Papaveraceae studied to date (i.e., Chelidonium majus Lour., Eschscholzia californica, Papaver somniferum and Sanguinaria candensis), as they initiate from an elliptic or hemispheric floral apex, from which sepals form first (Figs. 2a,b, 3a–e, 4a,b) [10, 28, 40–42]. Most Papaveraceae develop an outer and an inner whorl of two petals; the two outer petals develop first alternating with the two sepals, followed by the two inner petal primordia (Figs. 1d, 2c–e) [10, 28, 40–42]. Although both B. frutescens and M. cordata exhibit a petal-to-stamen homeosis, as stamens develop in the place and time of petals compared to petalous closely related species resulting in apetalous flowers, they differ in the number of homeotic staminal whorls that actually develop. In B. frutescens only the outer whorl of petals is clearly replaced by homeotic stamens, and the second one, opposite to the sepals, does not develop (Figs. 1i, 3g–k), whereas in M. cordata both whorls of petals are replaced by homeotic stamens (Figs. 1m, 4d–g). Homeotic stamens in B. frutescens and M. cordata are bulged and rounded from early stages of development, suggesting complete homeosis (Figs. 3f, 4d,e). These results contrast with the stamen-to-petal homeosis observed in S. canadensis, which exhibits flowers with eight petals. In S. canadensis both the stamen primordia and the homeotic petal primordia are bulged and rounded, contrasting with the true petals that are broad compared to their length . In turn, early ontogeny of the homeotic petals differs from that of true petals, and homeosis occurs late in development. This shows that homeosis can vary in terms of placement and timing in different Papaveraceae species.
The number of staminal whorls varies across Papaveraceae from one in M. cordata to two in B. frutescens, three in E. californica (and the terminal flowers of B. frutescens; Additional file 3: Fig. S1), four in S. canadensis, six in C. majus and S. diphyllum or many in P. somniferum (Figs. 2c,d, 3i–k, 4e–g) [10, 40–42]. Such variation suggests a high degree of developmental flexibility in the construction of staminal whorls, as it was noticed by Sattler , who also pointed out that irregularities are rather common and the notion of defining whorls is an over simplification in Papaveraceae. The number of stamens per whorl also varies dramatically in different species and even intraspecifically in flowers occupying different positions in the inflorescence (Figs. 2c–e, 3j, k, m; Additional file 3: Fig. S1). However, homeotic stamens of the first whorl are exceptional in that they are always paired and occupy the same position as petals (Figs. 1d, i, m, 2c, 3i, 4e, g) [40, 42]. Unlike the erect and thickened stamens in most Papaveraceae (e.g., Figs. 1a–c, 2f, g), B. frutescens and M. cordata possess pendulous and versatile stamens during anthesis due to the long and slender filaments (Figs. 1g, h, l, 3o, 4j). Typically, all Papaveraceae species have two carpels with the exception of S. diphyllum that has four (Fig. 2g)  and some Papaver species that have up to 12 [28, 44]. Features unique for B. frutescens include the gynophore that pushes the gynoecium outside of the sepals during preanthesis and the papillose stigma (Figs. 3n, o, 8c–h). These traits are linked to wind pollination in other Ranunculales like Euptelea pleiosperma Hook. f. & Thomson and Thalictrum spp. and are thought to have occurred as a result of convergent evolution [45–47]. Lastly, in B. frutescens and M. cordata a single ovule develops, an uncommon feature in the family likely resulting from an evolutionary reduction (Fig. 3k) [10, 48]; the persistent red aril on the single seed of B. frutescens (Fig. 8c–h), a trait likely related to bird dispersal, is also unique to this genus.
Loss of petals is likely due to APETALA3-III loss of function concomitant with the expansion of AGAMOUS expression domains in B. frutescens
The study of gene lineage evolution linked to expression study analyses is a plausible way to understand the acquisition and loss of floral organs; this is particularly true for genes that control petal identity [17, 21, 49]. It is known that the AP3 gene lineage has undergone functional diversification associated with several duplications occurring before the diversification of the Ranunculales. As a result of two consecutive duplications, all members of the order have gene representatives from three clades, named AP3-I, AP3-II and AP3-III (Fig. 5a) [7, 21, 25]. The Papaveraceae seem to have lost the AP3-II clade paralogs and have genes only belonging to the AP3-I and the AP3-III clades . Functional characterization of gene copies in A. coerulea (Ranunculaceae) shows that AqAP3-1 specifies the identity of the staminodia, AqAP3-2 controls the identity of stamens, and AqAP3-3 controls petal identity. Functional data in other members of the Ranunculaceae, like Nigella damascena, show that while NdAP3-3 has specialized in providing petal identity, NdAP3-1 and NdAP3-2 are functioning mostly in stamen identity and only in a dosage-dependent manner in petal shape [30, 32]. Functional data are also similar in Papaveraceae. In Papaver somniferum it has been shown that PapsAP3-1 (belonging to the AP3-III clade) specifies petal identity, whereas PapsAP3-2 (belonging to the AP3-I clade) controls stamen identity (Fig. 5a) [25, 28–30]. In general, these studies demonstrate that the three gene clades have undergone subfunctionalization, in particular the AP3-I and AP3-III clades, which have specialized in regulating stamen and petal identity, respectively. Subfunctionalization likely occurred first; thus, petal loss can occur without affecting stamen identity, as it occurs in B. frutescens. Our results show that the single copy expressed, BofrAP3, belonging to the AP3-I clade is predominantly found in stamens, consistent with functional analysis of AP3-I orthologs in Papaveraceae and Ranunculaceae (Fig. 5a, 8j) [25, 28, 29]. It was not possible to isolate any AP3-III paralog from the B. frutescens floral transcriptome; thus, we propose that this copy is no longer expressed, and it is linked to petal loss in B. frutescens. Without this AP3-III paralog, it is likely that the petal developmental program, which included an obligate PI heterodimer and the AP1/FUL-like or the AGL6 homolog, is no longer active.
A relation between loss of gene expression and petal loss has also been observed in a number of Ranunculaceae species, where gene loss, deletion of exonic sequences, deletion of regulatory sequences and pseudogenization of the AP3-III ortholog in different apetalous species have been shown to occur [32, 39]. Our analysis shows that the same is likely true for Papaveraceae as the AP3-III paralog is found in all Papaveraceae with petals with available transcriptomes, including the closely petalous relatives like Sanguinaria Canadensis and Stylophorum diphyllum, but the ortholog is not expressed in B. frutescens. One additional point to consider is that the phenotype associated with the absence of AP3 as predicted by the ABCE model, and as it has been shown to occur after PsomAP3-1 (in the AP3-3 clade) downregulation in poppies is the acquisition of sepal identity in the second whorl organs [12, 28]; however, the floral bauplan in B. frutescens, like those exhibited by apetalous Ranunculaceae species, results in the formation of stamens, and not sepals, in the second whorl, which is indicative of an accompanying ectopic expression of C class AGAMOUS genes. AG genes have retained their roles in stamen and carpel identity, despite having undergone duplications in core, basal eudicots and monocots . PaleoAG preduplication genes in basal eudicots (sensu Pabón.Mora et al. ) have been characterized in E. californica and P. somniferum (Papaveraceae) and in Aquilegia coerulea, Nigella damascena and Thalictrum thalictroides (Ranunculaceae), where they are responsible for stamen and carpel identity and at least in one case (ThtAG2 from T. thalictroides) they play key roles in ovule identity [30, 50–54]. Interestingly, different AG copies in Ranunculales do not share a common origin. AG paralogs in Ranunculaceae are the result of a family-level duplication, while AG paralogs in E. californica are the result of a species-specific duplication and AG transcripts in P. somniferum are the result of alternative splicing producing two proteins with different lengths of C-terminus [20, 50–54]. Moreover, it has been shown that E. californica AG copies can negatively regulate AP3-3 paralogs, but not AP3-1 paralogs, suggesting specific repressive interactions of C over B class genes . However, it is not clear whether they are mutually repressive, as functional analysis of AP3 homologs has not been done in E. californica and AG expression was not tested in P. somniferum plants showing AP3 downregulation [28, 54]. We found that B. frutescens has a single copy of AG that is expressed predominantly in stamens and carpels in mature flowers, but also has an extended expression in sepals; therefore, we propose that concomitant with the absence of BofrAP3-3, BofrAG has expanded into the outer whorls (Fig. 9). In the second whorl BofrAG together with BofrAP3 is likely to provide stamen identity (Figs. 7, 8j, k, n). Whether ectopic expression of BofrAG occurred prior to or after loss of expression of BofrAP3-3 is unclear at this point and requires a deep understanding of the regulatory interactions among these key regulatory transcription factors.
Petal loss in B. frutescens occurs independently of duplications and functional evolution of PISTILLATA genes
Contrary to the AP3 gene lineage, the PI gene lineage exhibits numerous recent taxon-specific duplications resulting in exclusive paralogs in different species of Papaveraceae, Berberidaceae and Ranunculaceae (Fig. 5b). Functional studies of PI genes have been done in E. californica and P. somniferum (Papaveraceae) and A. coerulea and Thalictrum thalictroides (Ranunculaceae). In E. californica the seirena-1 (sei-1) mutant showing homeotic conversions of petals to sepals and stamens to carpels was mapped to a mutation in EcGLO, the only PI ortholog in this species (Fig. 5b) . The same phenotype was found after downregulation of AqvPI and ThtPI in Aquilegia and Thalictrum, respectively [31, 56]. In P. somniferum, a local duplication has resulted in two copies PapsPI-1 and PapsPI-2. Functional characterization shows that both copies redundantly control petal and stamen identity . Altogether, these data suggest that, unlike AP3 genes, the role of PI genes has remained conserved over evolutionary time, and that after local duplications, paralogs retain redundancy . Our sampling in B. frutescens detected four PI copies with different expression patterns, three of them (BofrPI1, BofrPI2 and BofrPI4) expressed mainly in stamens, as expected; the fourth copy, BofrPI3, is expressed in stamens, sepals and carpels. Expression detected in the carpel can be the result of the characteristic expression of PI in the growing ovules [56, 57]. These data suggest that BofrPI3 is differentially regulated compared to its paralogs and could be playing additional roles besides stamen identity in B. frutescens (Fig. 8j, l, m).
FUL-like genes: pleiotropic genes with an uncertain contribution to petal identity
In addition to B class genes, the A class genes (APETALA1 an APETALA2) are also involved in petal identity in Arabidopsis thaliana . Numerous duplications have also occurred in these gene lineages [16, 23]. FUL-like genes in Papaveraceae are involved in floral meristem and perianth identity, transition to the reproductive meristem, cauline leaf development, branching and fruit development . Most Chelidonieae have two FUL-like copies, one in each Ranunculid clade (RanFL1 and RanfL2); exceptions are Sanguinaria canadensis, which has only one gene, and B. frutescens and M. cordata, which have three (BofrFL1, BofrFL2 and BofrFL3) and four copies (MacoFL1, MacoFL2, MacoFL3 and MacoFL4), respectively, as a result of species-specific duplications (Fig. 6a). In B. frutescens, BofrFL2 and BofrFL3 are broadly expressed in sepals, stamens, carpels and fruits, which suggests that they both have conserved the pleiotropic roles of other FUL-like genes in Papaveraceae . Conversely, BofrFL1 has a low expression, restricted to stamens, organs in which FUL-like genes are not known to play any developmental role. This suggests a rather different functional evolutionary path for BofrFL1 toward loss of function through pseudogenization [58, 59].
On the other hand, APETALA2/TOE3 genes have also diversified in angiosperms with most duplications occurring in the Brassicales, the Monocots and a local duplication in the Ranunculales . B. frutescens possesses two euAP2 genes named BofrAP2 and BofrAP2-2 mostly expressed in carpels, fruits and leaves, suggesting they are not involved in specifying perianth identity .
E-class genes have broad expression patterns suggestive of the maintenance of their ancestral role in B. frutescens
The E class genes SEPALLATA (SEP) are unique to angiosperms where they have duplicated resulting in the SEP3 and the SEP1/2/4 (also named LOFSEP) clades . SEP genes in Arabidopsis have been shown to play redundant roles on meristem and floral organ identity as the quadruple mutant shows a homeotic transformation from floral organs to leaves [14, 60]. SEP3 orthologs, like SEP3, have remained as single copy genes, are central in protein–protein interactions (PPIs) and are evolving under negative selection [61–64]. On the other hand, LOFSEP orthologs have duplicated in eudicots and monocots, are more labile in their protein interactions and can evolve under positive selection [64, 65]. B. frutescens has two SEP3 copies (BofrSEP3-1 and BofrSEP3-2) and three LOFSEP copies (BofrSEP1-1, BofrSEP1-2 and BofrSEP2). BofrSEP3-2 is expressed at higher levels than BofrSEP3-1, and they are mostly restricted to stamens and carpels (Fig. 8j). This suggests that BofrSEP3-2 may be the key factor mediating PPIs with AP3 and AG homologs, acting similarly to SEP3 in Arabidopsis, but is likely not involved in sepal identity [66–68]. From the LOFSEP paralogs, BofrSEP1-1 is expressed in all floral organs, fruits and leaves, which suggests that this paralog could be performing roles attributed to other LOFSEP homologs, including floral meristem and floral organ identity as well as fruit development. Similar expression patterns have been observed in monocots, where SEP copies have overlapping, although variable, expression patterns that are suggestive of both function redundancy and subfunctionalization among copies [62, 64, 69].
Our study integrating developmental /morphological data as well as gene expression and evolution data points to a number of conclusions regarding the natural homeosis occurring in petal-less poppies: (1) Petal loss in poppies may be acquired via distinct developmental pathways with complete homeosis in the second and third whorls (like in M. cordata) or complete homeosis in the second floral whorl, accompanied by abortion in the third floral whorl (like in B. frutescens). Interestingly, naturally occurring B. frutescens natural mutants never show the petal-to-sepal homeosis (predicted by the model), but a petal-to-stamen homeosis, suggesting that other negative regulatory loops between B and C class genes have likely not been identified. (2) The lack of AP3-3 gene expression in the apetalous B. frutescens is consistent with the reported role in petal identity of AP3-3 orthologs in Papaveraceae. This has been previously well established for Ranunculaceae and supports the idea that subfunctionalization in the AP3 paralogs occurred early in the Ranunculales, prior to the diversification of Papaveraceae and Ranunculaceae.
Critical point dryer
ELONGATION FACTOR 1α
Glyceraldehyde 3-phosphate dehydrogenase
Quantitative reverse transcriptase polymerase chain reaction
Reverse transcriptase polymerase chain reaction
Scanning electron microcopy
Endress PK. Diversity and evolutionary biology of tropical flowers. Cambridge: Cambridge University Press; 1994.
Endress PK. Floral structure and evolution in Ranunculanae. Plant Syst Evol. 1995;9:47–61.
Ronse De Craene LP. Are petals sterile stamens or bracts? The origin and evolution of petals in the core eudicots. Ann Bot. 2007;100:621–30.
Ronse De Craene LP, Brockington SF. Origin and evolution of petals in angiosperms. Plant Ecol Evol. 2013;146:5–25.
Glover BJ. Understanding flowers and flowering: an integrated approach. New York: Oxford University Press; 2007.
Kadereit JW. Papaveraceae. In: Kubitzki K, editor. The families and genera of vascular plants. Berlin: Springer; 1993. p. 494–506.
Rasmussen DA, Kramer EK, Zimmer EA. One size fits all? molecular evidence for a commonly inherited petal identity program in Ranunculales. Am J Bot. 2009;96:96–109.
Wang W, Lu AM, Ren Y, Endress ME, Chen ZD. Phylogeny and classification of Ranunculales: evidence from four molecular loci and morphological data. Perspect Plant Ecol Evol Syst. 2009;11:81–110.
Stevens PF. Angiosperm phylogeny website. Version 12, July 2012. 2001. (continuously updated since, Accessed 21 Jun 2015).
Lehmann NL, Sattler R. Homeosis in floral development of Sanguinaria canadensis and S. canadensis “Multiplex” (Papaveraceae). Am J Bot. 1993;80:1323–35.
Ronse De Craene LP. The evolutionary significance of homeosis in flowers: a morphological perspective. Int J Plant Sci. 2003;164:225–35.
Coen E, Meyerowitz EM. The war of the whorls: genetic interactions controlling flower development. Nature. 1991;353:31–7.
Bowman J, Alvarez J, Weigel D, Meyerowitz EM, Smyth DR. Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development. 1993;119:721–43.
Pelaz S, Ditta GS, Baumann E, Wisman E, Yanofsky MF. B and C floral organ identity functions require SEPALLATA MADS-box genes. Nature. 2000;405:200–3.
Kramer EM, Dorit RL, Irish VF. Molecular evolution of genes controlling petal and stamen development: duplication and divergence within the APETALA3 and PISTILLATA MADS-Box gene lineages. Genetics. 1998;149:765–83.
Litt A, Irish VF. Duplication and diversification in the APETALA1/FRUITFULL floral homeotic gene lineage: implications for the evolution of floral development. Genetics. 2003;165:821–33.
Litt A, Kramer EM. The ABC model and the diversification of floral organ identity. Semin Cell Dev Biol. 2010;21:129–37.
Kramer EM, Jaramillo MA, DiStilio VS. Patterns of gene duplication and functional evolution during the diversification of the AGAMOUS subfamily of MADS box genes in angiosperms. Genetics. 2004;166:1011–23.
Zahn LM, Kong JH, Leebens-Mack J, Kim S, Soltis PS, Landherr LL, Soltis DE, De Pamphilis CW. The evolution of the SEPALLATA subfamily of MADS-Box genes: a preangiosperm origin with multiple duplications throughout angiosperm history. Genetics. 2005;169:2209–23.
Pabón-Mora N, Wong GK, Ambrose BA. 2014. Evolution of fruit development genes in flowering plants. Front Plant Sci 5. doi:10.3389/fpls.2014.00300.
Kramer EM, DiStilio VS, Schlüter PM. Complex patterns of gene duplication in the APETALA3 and PISTILLATA lineages of the Ranunculaceae. Int J Plant Sci. 2003;164:1–11.
De Martino G, Pan I, Emmanuel E, Levy A, Irish VF. Functional analyses of two tomato APETALA3 genes demonstrate diversification in their roles in regulating floral development. Plant J. 2006;18:1833–45.
Zumajo-Cardona C, Pabón-Mora N. Evolution of the APETALA2 gene lineage in seed plants. Mol Biol Evol. 2016;33:1818–932. doi:10.1093/molbev/msw059.
Pabón-Mora N, Hidalgo O, Gleissberg S, Litt A. 2013. Assessing duplication and loss of APETALA1/FRUITFULL homologs in Ranunculales. Front Plant Sci 4. doi:10.3389/fpls.2013.00358.
Sharma B, Guo C, Kong H, Kramer EM. Petal-specific subfunctionalization of an APETALA3 paralog in the Ranunculales and its implications for petal evolution. N Phytol. 2011;191:870–83.
Pabón-Mora N, Ambrose BA, Litt A. Poppy APETALA1/FRUITFULL orthologs control flowering time, branching, perianth identity and fruit development. Plant Physiol. 2012;158:1685–704.
Pabón-Mora N, Sharma B, Holappa LD, Kramer EM, Litt A. The Aquilegia FRUITFULL-like genes play roles in leaf morphogenesis and inflorescence development. Plant J. 2013;74:197–212.
Drea S, Hileman LC, deMartino G, Irish VF. Functional analyses of genetic pathways controlling petal specification in poppy. Development. 2007;134:4157–66.
Sharma B, Kramer EM. Sub- and neo-functionalization of APETALA3 paralogs have contributed to the evolution of novel floral organ identity in Aquilegia (columbine, Ranunculaceae). N Phytol. 2012;197:949–57.
Wang P, Liao H, Zhang W, Yu X, Zhang R, Shan H, Duan X, Yao X, Kong H. Flexibility in the structure of spiral flowers and its underlying mechanisms. Nature Plants. 2016;2:15188. doi:10.1038/NPLANTS.2015.188.
Kramer EM, Holappa L, Gould B, Jaramillo MA, Setnikov D, Santiago PM. Elaboration of B gene function to include the identity of novel floral organs in the lower eudicot Aquilegia. Plant Cell. 2007;19:750–66.
Gonçalves B, Nougué O, Jabbour F, RidelC Morin H, Laufs P, Manicacci D, Damerval C. An APETALA3 homolog controls both petal identity and floral meristem patterning in Nigella damascena L. (Ranunculaceae). Plant J. 2013;76:223–35.
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215:403–10.
Katoh K, Misawa K, Kuma K, Miyata T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 2002;30:3059–66.
Stamatakis A, Hoover P, Rougemont J. A rapid bootstrap algorithm for the RAxML web servers. Syst Biol. 2008;57:758–71.
Miller MA, Holder HT, Vos R, Midford PE, Liebowitz T, Chan L, et al. The CIPRES portals. 2009. Available online at: http://www.phylo.org.
Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: molecular evolutionary genetics analysis version 6.0. Mol Biol Evol. 2013;30:2725–9.
Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔCt method. Methods. 2001;52:402–8.
Zhang R, Guo C, Zhang W, Wang P, Li L, Duan X, Du Q, Zhao L, Shan H, Hodges SA, Kramer EM, Ren Y, Kong H. Disruption of the petal identity gene APETALA3-3 is highly correlated with loss of petals within the buttercup family (Ranunculaceae). Proc Natl Acad Sci. 2013;13:5074–9.
Becker A, Gleissberg S, Smyth DR. Floral and vegetative morphogenesis in California poppy (Eschscholzia californica Cham.). Int J Plant Sci. 2005;166:537–55.
Ronse De Craene LP. The systematic relationship between Begoniaceae and Papaveraceae: a comparative study of their floral development. Bull Jardin Bot Natl Belgique. 1990;60:229–73.
Sattler R. Organogenesis of flowers, a photographic text-atlas. Toronto: University of Toronto Press; 1973.
Kadereit JW, Blattner FR, Jork KB, Schwarzbach A. Phylogenetic analysis of the Papaveraceae s.l. (including Fumariaceae, Hypecoaceae and Pterydophyllum) based on morphological characters. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie. 1994;116:361–90.
Kadereit JW, Erbar C. Evolution of gynoecium morphology in Old World Papaveroideae: a combined phylogenetic/ontogenetic approach. Am J Bot. 2011;98:1243–51.
Ren Y, Li HF, Zhao L, Endress PK. Floral morphogenesis in Euptelea (Eupteleaceae, Ranunculales). Ann Bot. 2007;100:185–93.
Di Stilio VS, Martin C, Shulfer AF, Connelly CF. An ortholog of MIXTA-like 2 controls epidermal cell shape in flowers of Thalictrum. N Phytol. 2009;183:718–28.
Soza VL, Brunet J, Liston A, Salles Smith P, Di Stilio VS. Phylogenetic insights into the correlates of dioecy in meadow-rues (Thalictrum, Ranunculaceae). Mol Phylogenet Evol. 2012;63:180–92.
Jernstedt JA, Clark C. Stomata on the fruits and seeds of Eschscholzia (Papaveraceae). Am J Bot. 1979;66:586–90.
Specht CD, Howarth DG. Adaptation in flower form: a comparative evodevo approach. N Phylol. 2014;206:74–90.
Di Stilio VS, Kumar RA, Oddone AM, Tolkin TR, Salles P, McCarthy K. Virus-induced gene silencing as a tool for comparative functional studies in Thalictrum. PlosOne. 2010;5:e12064. doi:10.1371/journal.pone.0012064.
Galimba KD, Tolkin TR, Sullivan AM, Melzer R, Theißen G, Di Stilio VS. Loss of deeply conserved C-class floral homeotic gene function and C and E-class protein interaction in a double flowered ranunculid mutant. Proc Natl Acad Sci. 2012;109:2267–75.
Galimba KD, DiStilio VS. Subfunctionalization to ovule development following duplication of a floral organ identity gene. Dev Biol. 2015;405:158–72.
Hands P, Vosnakis N, Betts D, Irish VF, Drea S. Alternate transcripts of a floral developmental regulator have both distinct and redundant functions in opium poppy. Ann Bot. 2011;107:1557–66. doi:10.1093/aob/mcr045.
Yellina AL, Orashakova S, Lange S, Erdmann R, Leebens-Mack J, Becker A. Floral homeotic C function genes repress specific B function genes in the carpel whorl of the basal eudicot California poppy (Eschscholzia californica). Evodevo. 2010;1:13. doi:10.1186/2041-9139-1-13.
Lange M, Orashakova S, Lange S, Melzer R, Theißen G, Smyth DR, Becker A. The seirena B class floral homeotic mutant of California poppy (Eschscholzia californica) reveals a function of the enigmatic PI motif in the formation of specific multimeric MADS domain protein complexes. Plant Cell. 2013;25:438–53.
La Rue N, Sullivan AM, Di Stilio VS. Functional recapitulation of transitions in sexual systems by homeosis during the evolution of dioecy in Thalictrum. Front Plant Sci. 2013;4:487.
Kramer EM, Irish VF. Evolution of genetic mechanisms controlling petal development. Nature. 1999;399:144–8.
Duarte JM, Cui LY, Wall PK, Zhang Q, Zhang XH, Leebens-Mack J, Ma H, Altman N, dePamphilis CW. Expression pattern shifts following duplication indicative of subfunctionalization and neofunctionalization in regulatory genes of Arabidopsis. Mol Biol Evol. 2006;23:469–78.
Yang L, Takuno S, Waters ER, Gaut BS. Lowly expressed genes in Arabidopsis thaliana bear the signature of possible pseudogenization by promoter degradation. Mol Biol Evol. 2011;28:1193–203.
Ditta G, Pinyopich A, Robles P, Pelaz S, Yanofsky MF. The SEP4 gene of Arabidopsis thaliana functions in floral organ and meristem identity. Curr Biol. 2004;14:1935–40.
Immink RGH, Kaufmann K, Angenent GC. The ‘ABC’ of MADS domain protein behaviour and interactions. Semin Cell Dev Biol. 2010;21:87–93.
Malcomber ST, Kellogg EA. Heterogeneous expression patterns and separate roles of the SEPALLATA gene LEAFY HULL STERILE1 in grasses. Plant Cell. 2005;16:1692–706.
Shan HY, Zahn L, Guindon S, Wall PK, Kong HZ, Ma H, dePamphilis CW, Leebens-Mack J. Evolution of plant MADS box transcription factors: evidence for shifts in selection associated with early angiosperm diversification and concerted gene duplications. Mol Biol Evol. 2009;26:2229–44.
Yockteng R, Almeida AMR, Morioka K, Alvarez Buylla ER, Specht CD. Molecular evolution and patterns of duplication in the SEP/AGL6-like lineage of the Zingiberales: a proposed mechanism for floral diversification. Mol Biol Evol. 2013;30:2401–22.
Liu C, Zhang J, Zhang N, Shan H, Su K, Zhang J, et al. Interactions among proteins of floral MADS-box genes in basal eudicots: implications for evolution of the regulatory network for flower development. Mol Biol Evol. 2010;27:1598–611. doi:10.1093/molbev/msq044.
Castillejo C, Romera-Branchat M, Pelaz S. A new role of the Arabidopsis SEPALLATA3 gene revealed by its constitutive expression. Plant J. 2005;43:586–96.
Goto K, Kyozuka J, Bowman JL. Turning floral organs into leaves, leaves into floral organs. Curr Opin Genet Dev. 2001;11:449–56.
Honma T, Goto K. Complexes of MADS-box proteins are sufficient to convert leaves into floral organs. Nature. 2001;409:525–9.
Pelucchi N, Fornara F, Favalli C, Masiero S, Lago C, Pe ME, Colombo L, Kater MM. Comparative analysis of rice MADS box genes expressed during flower development. Sex Plant Reprod. 2002;15:113–22.
NP-M, FG and JFA planned and designed the research. CA-O, NP-M and FG conducted fieldwork, CA-O, NP-M, FG and JFA performed experiments, CA-O, NP-M, FG and JFA analyzed the data and wrote and approved the final version of the manuscript. All authors read and approved the final manuscript.
We thank D.W. Stevenson and L.M. Campbell (Structural Botany Lab, NYBG), M. Baxter (Lehman College, CUNY) and R. Callejas Posada and L. Atehortúa (Universidad de Antioquia) for allowing us to use their laboratory facilities. We thank Barbara Ambrose and Elena Kramer for helpful discussions and Cecilia Zumajo-Cardona for laboratory assistance.
The authors declare that they have no competing interests.
Ethical approval and consent to participate
The work was done with plants, so animal ethical approval and consent to participate are not applicable. As the plants are cultivated, they are exempt of legal access to genetic resources as stated by the Andean and Colombian law.
This research was funded by the Fundación para la Promoción de la Investigación y la Tecnología, Banco de la República, Colombia (Project Number 3374).