Volume 216, Issue 2 p. 361-366
Tansley insight
Free Access

Genetics of flower development in Ranunculales – a new, basal eudicot model order for studying flower evolution

Catherine Damerval

Catherine Damerval

GQE – Le Moulon, INRA, Univ. Paris-Sud, CNRS, AgroParisTech, Université Paris-Saclay, Gif-sur-Yvette, 91190 France

Search for more papers by this author
Annette Becker

Corresponding Author

Annette Becker

Justus-Liebig-Universität Gießen, Institut für Botanik, Heinrich-Buff-Ring 38, Gießen, 35392 Germany

Author for correspondence:

Annette Becker

Tel: +49 (0) 6409 9935200

Email: [email protected]

Search for more papers by this author
First published: 04 January 2017
Citations: 35

Summary

Ranunculales, the sister group to all other eudicots, encompasses species with a remarkable floral diversity, which are currently emerging as new model organisms to address questions relating to the genetic architecture of flower morphology and its evolution. These questions concern either traits only found in members of the Ranunculales or traits that have convergently evolved in other large clades of flowering plants. We present recent results obtained on floral organ identity and number, symmetry evolution and spur formation in Ranunculales species. We discuss benefits and future prospects of evo-devo studies in Ranunculales, which can provide the opportunity to decipher the genetic architecture of novel floral traits and also to appraise the degree of conservation of genetic mechanisms involved in homoplasious traits.

Contents
Summary 361
I. Introduction 361
II. The major regulators of floral organ identity in Ranunculales 362
III. Floral diversity in Ranunculales 363
IV. Pollination modes, reproduction systems and species diversification 364
V. Conclusion: from model organisms to model orders? 364
Acknowledgements 365
References 365

I. Introduction

The order Ranunculales includes > 4500 species distributed among seven families with unequal species numbers (Fig. 1a) and the order as a whole is sister to all other eudicots (APG, 2016). Ranunculales emerged c. 115 million yr ago (Ma), approximately at the same time as Rosidae (123–115 Ma) and Asteridae (119–110 Ma), the largest clades in the core eudicots (Magallon et al., 2015). They exhibit a large diversity in life-history traits, growth habit, leaf shape, flower and fruit forms, etc. The flower, in particular, is extremely diverse. Besides a basic number of organs (merism; see Box 1 Glossary) ranging from two to five, the perianth can be absent (Eupteleaceae) or, when present, differentiated into petals and sepals (bipartite) or composed of a single type of organ (unipartite), either sepaloid or petaloid. Additionally, floral organs vary extensively in form, size and color, in relationships with pollination modes. Because of phylogenetic position and age, and this remarkable morphological diversity in the order, genera or species belonging to the largest families (Papaver, Eschscholzia: Papaveraceae; Aquilegia, Thalictrum, Nigella: Ranunculaceae) are emerging as new models to address questions relating to the genetic architecture of flower morphology and its evolution. These concern specific traits found in members of the Ranunculales that are poorly represented among classical core eudicot or monocot model species, such as changes in merism, changes between whorled vs spiral phyllotaxis, or formation of novel organs. They also concern traits that are found in both Ranunculales and core eudicots or monocots, such as radial vs bilateral symmetry, spur formation, and transition between sexual systems in closely related taxa (Endress, 1995; Soza et al., 2012). Such homoplasious characters provide the opportunity to study the conservation of genetic mechanisms involved in the origin of homologous vs convergent traits.

Details are in the caption following the image
Ranunculales – phylogeny, examples, and morphological traits. (a) A simplified phylogeny of the Ranunculales with photographs of members of this order. The number of species for each family is given in brackets next to the family name (The Angiosperm Phylogeny Group, 2016). Genera comprising model species of each family are listed, for which genetic resources, at least virus-induced gene silencing (VIGS), are established. Photographs are of (top left to bottom right): Anemone hupehensis (Ranunculaceae), Epimedium × rubrum (Berberidaceae), Akebia quinata (Lardizabalaceae) and Papaver somniferum (Papaveraceae). (b) Evolution of specific flower morphological traits within a subgroup of the Ranuculaceae, with transitions to spiral phyllotaxy, zygomorphy (Z), dioecy (♀♂) and multiple spurs (symbol like spur) indicated: S, sepal; P, petal. Photographs (left to right): Aquilegia vulgaris, Thalictrum flavum, Nigella damascena, Aconitum carmichaelii. (c) Evolution of specific flower morphological traits within a subgroup of the Papaveraceae, with transitions to spurred corollas and zygomorphy (Z) indicated: P, petal. Photographs (left to right): Lamprocapnos spectabilis, Cysticapnos vesicaria, Eschscholzia californica.

Box 1. Glossary

  • Androecium – the sum of all stamens
  • Gynoecium – the sum of all carpels (often fused into a single organ)
  • Floral homeotic gene – gene required for floral organ identity specification
  • Homeosis – the transformation of one organ type into another organ type (e.g. in place of the stamens petals are found)
  • Homoplasy – similarity in a trait that is not a result of common ancestry but of convergent evolution
  • Merism – basic number of each type of organ
  • NGS – Next-generation sequencing
  • Orthologs – a gene pair that originated as a result of a speciation event
  • Paralogs – a gene pair that originated as a result of a gene duplication event
  • Phyllotaxis – insertion pattern of organs along the shoot or on the floral receptacle
  • Spur – nectar-filled, hollow, more or less elongated pouch forming on petals or sepals and extending behind the flower
  • Staminodium – novel Aquilegia-specific organ type of unknown function, formed between stamens and gynoecium
  • VIGS – virus-induced gene silencing, a technique to induce systemic transcriptional gene silencing by double-stranded RNA utilizing viral vector systems – the resulting phenotypes are transient and, in most cases, not fully penetrant

In the following, we present the major determinants of floral organ identity in Ranunculales, and we discuss several traits participating in floral diversity, their genetic bases and possible role in species diversification. Keeping the exemplary nature of our approach in mind, we aim to broaden our understanding of evolutionary pathways to floral diversity.

II. The major regulators of floral organ identity in Ranunculales

The principal genetic blueprint for floral organ specification in Arabidopsis thaliana is integrated in the ABCE model of floral organ identity: floral homeotic genes of A function together with E function specify sepals; A, B and E functions specify petals; B, C and E functions specify stamens; and C and E functions specify carpel/gynoecium organ identity. With the notable exception of one A function gene, all the floral homeotic genes belong to the MADS-box family of transcription factor genes (reviewed in Becker & Ehlers, 2016). They are thereafter named in reference to the A. thaliana genes: PISTILLATA (PI)- and APETALA3 (AP3)-like genes constitute the B function; AGAMOUS (AG)-like genes constitute the C function; and SEPALLATA (SEP)-like genes constitute the E function genes. The A function as defined in A. thaliana appears generally not conserved in other angiosperms, leading to the idea of a simplified conserved (A)B(C) model of floral organ identity based on functional data and phylogeny reconstructions (Theißen et al., 2016). The A function here specifies floral meristem and floral organ identity and requires a group of related genes such as homologs of APETALA1 (AP1), SEP and AGAMOUS-Like6 (AGL6) genes. The C and D (ovule identity) function genes trace back to a common ancestor and may be more freely interchangeable in function, leaving only the B function highly conserved (Theißen et al., 2016).

Floral homeotic genes have been characterized in many Ranunculales species; however, substantial expression data and functional validation have been obtained in only a few species described in Fig. 2.

Details are in the caption following the image
Comparison of expression of floral homeotic genes in Ranunculales species for which comprehensive data are available at the late stages of flower development, except for Thalictrum thalictroides B and C function genes, where only younger developmental stages were analyzed. (a) Thalictrum thalictroides, (b) Aquilegia coerulea, (c) Nigella damascena and (d) Eschscholzia californica. Yellow, putative homeotic B class genes; red, putative homeotic C class genes; purple, putative homeotic E function genes. Multiple gene copies are symbolized by multiple bars; for a given class of genes or copy, expression strength is indicated by the thickness of the bars. G, gynoecium; iSt, inner stamens; oSt, outer stamens; P, petals; Pa, perianth; Se, sepals; Std, staminodia; St, stamens.

FRUITFUL (FUL)- or AGL6-like MADS-box genes seem to partially carry out the A class function in Ranunculales, as the split between FUL- and AP1-like genes is core eudicot-specific (Pabon-Mora et al., 2013; Wang et al., 2015; Becker, 2016).

PI homologs, even when duplicated, are generally equally strongly expressed in petals and stamens, sometimes extending into sepals. Their role in petal and stamen identity have been demonstrated by virus-induced gene silencing (VIGS) or mutant analysis in Aquilegia, Nigella, Papaver and Eschscholzia (Drea et al., 2007; Kramer et al., 2007; Lange et al., 2013; Wang et al., 2015). In contrast to the rather fixed expression of PI-like genes, AP3 homologs show a high degree of organ-specific expression (Fig. 2). The AP3 lineage underwent two successive duplications at the base of the Ranunculales, probably after the divergence of Eupteleaceae, resulting in three paralogous subfamilies (AP3-1, -2, -3), of which only two are retained in the Papaveraceae (e.g. Sharma et al., 2011). Down-regulation of the two paralogs by VIGS in Papaver somniferum showed that the AP3-1 and AP3-3 genes are essential to stamen and petal identity, respectively (Drea et al., 2007). In the Ranunculaceae Aquilegia coerulea, AP3-1exerts a function in the novel staminodia, AP3-2 in stamens, and AP3-2 in petal formation (Sharma et al., 2011; Sharma & Kramer, 2013). Indeed, the expression of AP3-3 genes generally correlates with the presence of petals in Ranunculaceae, Berberidaceae, Papaveraceae and Lardizabalaceae (Rasmussen et al., 2009; Hu et al., 2012). Genes of this AP3 subfamily are nonfunctional in apetalous species of Ranunculaceae and Papaveraceae, arguing against the hypothesis of multiple independent origins of petals in the Ranunculales genera (Gonçalves et al., 2013; Zhang et al., 2013; Arango-Ocampo et al., 2016).

A duplication at the base of the Ranunculales gave rise to two AG paralogs, with one copy typically expressed in both stamens and carpels, while expression of the other copy is restricted to carpels (Fig. 2). Silencing of individual AG homologs by VIGS has not been reported or does not show a phenotype (Yellina et al., 2010; Wang et al., 2015) except in T. thalictroides, where ThtAG2 is involved in ovule identity, whereas ThtAG1 is the C-function gene (Galimba et al., 2012; Galimba & Di Stilio, 2015). Interestingly, the failure of a mutant AG protein in T. thalictroides to form heterodimers with a SEP3 homolog causes a transformation of reproductive organs into perianth organs, a phenotype found in the ornamental cv ‘Double White’ (Galimba et al., 2012).

SEP-like genes are expressed rather uniformly throughout the flower in Ranunculales. Silencing all three SEP-like genes in Nigella damascena resulted in floral meristem termination defects and all floral organs converted to bracts (Wang et al., 2015). VIGS of SEP-like genes in T. thalictroides resulted in sepal abnormalities, organ identity shifts and failure to define whorl boundaries (Soza et al., 2016), suggesting that SEP function may be poorly conserved within Ranunculales.

In summary, floral organ identity in Ranunculales is determined by class B, C, E genes phylogenetically close to those in core eudicots. The variation in floral morphology is largely a result of subfunctionalization and most probably neofunctionalization enabled by Ranunculales-specific duplications, especially in B class genes where AP3-3 function is indispensable for petal formation and the AP3-1 gene plays a major role in the formation of staminodia in Aquilegia.

III. Floral diversity in Ranunculales

The ancestral state of the perianth cannot be reconstructed for the Ranunculales as a whole, because the early diverging Eupteleaceae have none (Damerval & Nadot, 2007). However, the ancestral position appears to be bipartite in both the Papaveraceae and in the core Ranunculales, while a unipartite perianth evolved repeatedly in the Lardizabalaceae, the Berberidaceae and, particularly, the Ranunculaceae (Rasmussen et al., 2009). Homeotic conversion of organs promoted by organ identity gene inactivation or expression shifts accounts for transitions between bipartite and unipartite perianths in some species (Gonçalves et al., 2013; Arango-Ocampo et al., 2016), but whether this is a general process deserves further investigations.

Merism and phyllotaxis are highly diverse within Ranunculales (Endress, 1995; Damerval & Nadot, 2007). High merism such as in the androecium (polyandry) and/or spiral phyllotaxis may favor organ number variations within species, suggesting that Ranunculales development is flexible, in contrast to most core eudicots and monocots. Polyandry is ancestral in the Papaveraceae and most probably in the Ranunculales (Sauquet et al., 2015). In Eschscholzia californica as in other members of the Papaveraceae, a ring primordium is formed on which the numerous stamen primordia progressively emerge. Stochastic variation in the amount of B and C class genes has been suggested to account for the variation in the final stamen number (Becker, 2016). Variation in perianth organ number has been described in many species of Ranunculaceae, owing to homeotic transformation and/or stochastic fluctuation in the number of initiated primordia (Jabbour et al., 2015 and references therein). In N. damascena, all floral organs are spirally inserted and number variation has been observed for each type of organ (Wang et al., 2015). Down-regulating genes of the (A)B(C)model using VIGS revealed a network of interactions that could define the domains of gene expression determining organ identity and number (Wang et al., 2015; Fig. 2c). Indeed, a mathematical model based on the distribution of the expression domain boundaries of floral organ identity genes could account for the variation of perianth organ numbers in about half the number of Ranunculaceae species studied (Kitazawa & Fujimoto, 2014).

Bilateral symmetry (also known as zygomorphy) evolved three times in the Ranunculales (Damerval & Nadot, 2007). The developmental genetic analysis of zygomorphy in Antirrhinum majus has pointed to the key role of the CYCLOIDEA gene (Luo et al., 1996 Luo et al.,1999), encoding a transcription factor of the TCP family. Summarizing more than 15 years of studies of CYC-like genes shows that asymmetric expression patterns (mostly along the dorsoventral axis) are associated with bilateral symmetry in various core eudicots and monocots (Hileman, 2014). In zygomorphic species of Papaveraceae and Ranunculaceae (Fig. 1b,c), an asymmetric expression of CYC-like paralogs has been observed in late-stage flower buds, already exhibiting organ differentiation (Damerval et al., 2013; Jabbour et al., 2014). In these cases, it is not clear whether CYC-like genes direct asymmetric development or are instead targeted by an initial asymmetric clue of unknown nature, possibly hormonal.

In the Ranunculales, spurs evolved twice in relation to zygomorphy but also five times in a number equal to merism, which is quite uncommon in angiosperms (Damerval & Nadot, 2007). In Aquilegia, it has been found that after an initial phase of cell proliferation, the five petal spurs reach their final form by anisotropic cell elongation rather than by cell proliferation, and that this process accounts for c. 99% of the spur length variation in the genus (Puzey et al., 2012). In A. coerulea, transcriptomic and VIGS analyses point to the relocalization of an auxin-related pathway typically expressed at organ margin in Arabidopsis to the cup region of the growing spur, and to a role of the cell division repressor AqTCP4 in spur sculpting. Interestingly, no significant expression of the KNOXI SHOOTMERISTEMLESS (STM) or KNAT homologs has been observed during spur formation (Yant et al., 2015). Similarly, spur development in species of Papaveraceae was not associated with expression of STM homologs (Damerval et al., 2013). These results contrast with the suggested role of STM-like genes in directing spur development in a few core eudicot species (Box et al., 2011), revealing at least partly independent genetic pathways producing a convergent phenotype.

IV. Pollination modes, reproduction systems and species diversification

Floral traits affecting the mode of reproduction and/or the interactions with pollinators are good candidates for promoting speciation. Insect pollination is probably ancestral in the Ranunculales. The genus Thalictrum has been considered a model for deciphering evolutionary transitions between different reproduction modes, but also between insect and wind pollination (Soza et al., 2012, 2013). When the two PI paralogs in the dioecious species Thalictrum dioicum are silenced, male flowers are turned into female flowers, suggesting an important role for PI-like genes in sexual system determination. Evolutionary transition between these systems may be linked to the functionality and/or expression domain shifts of the PI homologs (Larue et al., 2013).

Bilateral symmetry and spurs are considered as adaptations to animal pollination and are highly convergent traits in angiosperms. Both have been considered as key innovations promoting species radiation, e.g. Fumaria and Corydalis, the two zygomorphic spurred genera in Papaveraceae, account for c. 70% of the species. Zygomorphy in Delphinieae, the most species-rich Ranunculaceae tribe, involves the formation of a spur or a hood by the dorsal sepal enclosing nectariferous petal spurs. Most species are pollinated by bumblebees, but hawkmoth pollination has also been reported. No correlation between spur length and pollinator's tongue length was found, possibly because of the particular architecture of the nectar device (two petal spurs, fused or not; Jabbour & Renner, 2012). By contrast, pollinator shifts (bee to hummingbirds, hummingbirds to hawkmoths) are a major driver of spur length and species radiation in North American Aquilegia species (Whittall & Hodges, 2007).

V. Conclusion: from model organisms to model orders?

In the past, knowledge on the developmental networks came mainly from analyses of model species within the core eudicots and grasses, mainly A. thaliana and Oryza sativa, but also A. majus (snapdragon), Petunia hybrida (petunia) or Zea mays (corn), enabling large-scale comparisons of gene regulatory network components and their conservation. However, the limitation to a few model systems hampers our understanding of important evolutionary processes such as the origin of novelties, developmental flexibility vs robustness, and the impact of developmental traits for speciation/coevolution. The classical candidate gene approach has limited power to test fine-grained hypotheses on the conservation of genetic networks controlling similar developmental processes. There is now a need to enlarge the sampling across the angiosperm phylogenetic tree and to expand our knowledge of gene functions outside the model systems.

The order Ranunculales includes morphologically and biochemically very diverse taxa. It comprises many species with a short generation time valuable to obtain large genomic/transcriptomic data sets using NGS, and several species of diverse morphology amenable to gene function analysis by VIGS using the Tobacco rattle virus (TRV) as vector system. Because of their phylogenetic position as early branching eudicots, Ranunculales may enable a reliable reconstruction of ancestral states of eudicot gene networks controlling a variety of floral traits, which could help researchers to understand the large morphological gap between the grass and core eudicot model organisms. For example, deciphering the genetic bases of zygomorphy or nectar spurs that evolved repeatedly in both the Ranunculales and other clades will enable us to disentangle genetic specificities and reuse or coopt genetic circuitries directing convergent traits at various phylogenetic levels. Moreover, Ranunculales species exhibit specific floral features among the eudicots, such as development of spurs in the actinomorphic context and formation of novel organs (staminodia). Hypotheses on the evolution of such specific traits and a gene's potential for neofunctionalization can be tested, as close relatives with opposing states are available. Uncovering genetic networks involved in convergent traits as well as unraveling processes underlying the emergence of specific traits will both be of the utmost importance in allowing better appraisal of the complexity of floral evolution and how it may support species diversification.

Acknowledgements

Work in A.B.'s laboratory is mainly funded by the German Research Foundation (DFG), grants BE 2547/3-1, 7-2, 6-2, 14-1. Work in C.D.'s laboratory is mainly funded by Agence National de la Recherche (ANR-07-BLAN-0112) and grants from the Institut Diversité Ecologie et Evolution du Vivant (IDEEV) and IFR 87: The Plant and its Environment. We are grateful to Hans Bahmer und Evgenia Diel (Gießen) and Yafei Zhao (Helsinki) for generously contributing photographs of A. vulgaris, Epimedium × rubrum, Lamprocapnos spectabilis, Akebia quintata and Cysticapnos vesicaria, and to Sophie Nadot for helpful discussions about floral trait evolution. We apologize sincerely to those who have published important results on Ranunculales floral development and whose work could not be cited because of space limitations.