- Email: [email protected]
Evolution and development of virtual inflorescences
Update Research Focus
Evolution and development of virtual inflorescences Ronald Koes Institute for Molecular Cell Biology, Graduate School of Experimental Plant Sciences, Vrije Universiteit, de Boelelaan 1085, 1081HV Amsterdam, The Netherlands
The architecture of inflorescences diverged during the evolution of distinct plant families by mechanisms that remain unknown. Using computer modeling, Przemyslaw Prusinkiewicz and colleagues established a single model for the development of distinct inflorescences. Selection restricts inflorescence evolution to high fitness paths that vary with climate and other factors that influence reproductive success – explaining why some evolutionary transitions are more likely than others. This model presents an important framework for future plant ‘evo-devo’ research.
Evolution of body shape An eye-catching manifestation of evolution is the diversity of body architecture among species. One of the most elementary discoveries in developmental biology was that many genes that determine the identity and anatomy of large body parts (e.g. segments) or organs, for example Homeobox (HOX) genes, are conserved between animals as anatomically diverse as nematodes, flies and mammals. If this is the case, what causes anatomical diversity? The prevailing view in evolutionary developmental biology (‘evo-devo’) is that animal body shapes evolved mostly by changes in gene-expression patterns rather than through the emergence of new genes [1]. Whether this occurred primarily via changes in cis-regulatory elements in gene promoters or through alterations in the coding sequences of trans-acting transcription factor genes is difficult to ascertain from the limited data available and is a subject of debate [2], however there is evidence for both mechanisms [1,3]. The diversification of animals is mirrored by that of plants. For example, angiosperms, which contain the vast majority of plant species, evolved in < 300 million years and generated an enormous diversity in shapes, numbers and positions of organs [4], compared with the evolution of vertebrate animals, which is thought to have taken some 500 million years. It was suggested early on that plant shape evolved primarily through alterations in transcriptional control [5] but supporting evidence remains scarce. Because a range of plant species with diverse morphologies are amenable to forward genetic analyses (e.g. maize, rice, Arabidopsis, tomato, petunia etc.) and/or transgenesis (numerous species), they offer excellent possibilities for evo-devo research that have hardly been explored to date.
Corresponding author: Koes, R. ([email protected]).
When and where to make flowers? Plant species exhibit a large diversity in when and where they form flowers. Flowers can arise either as single flowers at the end of a branch, or they can be arranged into complex inflorescences that bear many flowers. The inflorescences found in nature fall into three main classes (Figure 1a). In racemes, flowers appear along the flanks of axes (branches). In cymes, the primary axis terminates by forming a flower and growth continues from a lateral axis that repeats this pattern. In panicles, primary and higher order lateral axes terminate in a flower. Most plant families contain species with distinct inflorescence types, suggesting that they evolved from one another multiple times independently [6]. How this occurred remains unknown; however, Prusinkiewicz et al. [7] have presented an extensive model that merges developmental genetics with population biology to explain how it might have taken place. The development of virtual inflorescences In their paper Prusinkiewicz et al. [7] address two questions. First, why did evolution produce only three main inflorescence types and not others? Second, how did these inflorescence types evolve? They devised a simple computer model to show that distinct inflorescence types might be formed by variations in a common mechanism. In these virtual inflorescences, the identity of floral and shoot meristems is reflected by a quantitative character called ‘vegetativeness’ (veg). Thus, veg is not a compound but a measure of the ‘state’ of a meristem. If veg is high, a meristem will produce a new metamer with a lateral meristem, but if veg drops below a certain threshold, the meristem terminates by forming a flower. Initially, veg levels are high but generally decrease over time as a function of plant and/ or meristem age. If veg decreases with similar kinetics in all meristems, a panicle is produced, in which meristems first proliferate and at some point all turn into a flower (Figure 1b, left). If veg decreases quickly and reaches the minimum threshold during development of the first metamer, a simple panicle with only two flowers will form, but if veg decreases more slowly, a more compound structure with many flowers is generated (Figure 1a). The creation of cymes and racemes requires differential regulation of veg in newly emerging ‘immature’ lateral meristems and the ‘mature’ apical meristems, and thus requires more complexity. If veg is lower in the apical meristems than in the lateral meristems, a cyme is created. By contrast, if veg is lower in lateral meristems, a raceme is formed (Figure 1a). To ensure that the co-florescences develop into bona fide racemes rather than terminating with a flower, the immature state of lateral meristems
Update
TRENDS in Plant Science Vol.13 No.1
Figure 1. Development of distinct virtual inflorescence structures. (a) Structure of inflorescences and position in morphospace. Flowers are indicated by red circles and meristems by green arrows. The inflorescence types are positioned in a 2D morphospace defined by the time required for apical (Tapical) and lateral (Tlateral) meristems to acquire floral fate. (b) Expression of veg in compound panicle (left), raceme (middle) and cyme (right). The black line depicts veg levels in the primary apical meristem. The colored lines depict veg in the first (red), second (green) and third (orange) lateral meristems formed by the primary apex. Adapted from [7].
needs to be transient. If veg does not reach the threshold during this immature phase the meristem turns into an (mature) apical meristem with corresponding veg levels and continues to grow accordingly (Figure 1b). Thus, inflorescence architectures are defined by only two variables – the times at which apical and lateral meristems attain low levels of veg – and can be plotted in a 2D ‘morphospace’ (Figure 1a). In this morphospace cymes and racemes are at opposite ends and are separated by panicles with variable degrees of complexity. It is noteworthy that other inflorescences with distinct arrangements of flowers and meristems cannot be produced by this simple model or by nature. This is because such structures cannot be made when all apical and all lateral meristems display the same veg decay, and would therefore require additional complexity. A virtual Arabidopsis Two obvious questions are: do these virtual inflorescences have any relation to real ones, and can veg be explained in 2
terms of molecules? The answer to both these questions is ‘yes’. Genetic experiments identified several genes that determine the racemose architecture of the Arabidopsis inflorescence. The most important ones are LEAFY (LFY) and TERMINAL FLOWER 1 (TFL1). In lfy mutants, flowers are replaced by shoot-like structures, whereas tfl1 generates short inflorescences that terminate with a flower [8]. Therefore, TFL1 promotes veg and LFY represses veg. In the wild-type, TFL1 mRNA is expressed in the apical meristem [9] and LFY in the lateral floral meristems, partly because LFY and TFL1 downregulate each other [8]. These, and additional genetic data, have been incorporated in the model to compute veg, which resulted in a virtual Arabidopsis that recreates the wild-type, in addition to single and double mutants containing gainand/or loss-of-function alleles of TFL1 and LFY [7]. A more rigorous test would be to determine the validity of the model for distinct inflorescence types, such as cymes and panicles. Cymes and panicles have been studied less
Update
TRENDS in Plant Science Vol.13 No.1
intensively and the genetic control of veg is only partially understood. In the cymose inflorescences of tomato (Solanum lycopersicon) and petunia (Petunia hybrida), LFY homologs are expressed within the apical meristem where they specify floral identity and, thus, downregulate veg [10,11]. In the emerging lateral meristems of petunia the expression of the LFY homolog is delayed, or transiently repressed [10]. In tobacco (Nicotiana tabacum), which is closely related to petunia and tomato, constitutive expression of LFY results in a solitary terminal flower [12], indicating that the transient inactivity of LFY in the lateral meristem is essential for formation of a cyme, as predicted by the theoretical model. Furthermore, putative TFL homologs from tobacco are not expressed in the apex of inflorescences [13], which might explain the decrease of veg in the apical meristem and the formation of a terminal flower. However, other observations are less easy to fit with the model. For example, ectopic expression of LFY in citrus trees causes early flowering (indicating that LFY can inhibit veg in this species) but does not alter inflorescence structure [14]. Furthermore, SINGLE FLOWER TRUSS (SFT) from tomato is, like its Arabidopsis homolog FLOWERING LOCUS T, a mobile protein that promotes flowering [15]. This indicates that SFT represses veg and would predict that sft produces a complex cyme, rather than the solitary flower that is observed. This does not necessarily imply that the model is incorrect, but that the genetic regulation of veg in cymes involves mechanisms that are not understood and which are in part distinct from those in Arabidopsis. Who is the fittest? Who will survive? Prusinkiewicz et al. took their model one step further by considering how selection might act on inflorescence structures. Panicles produce many more flowers than cymes and racemes, but in a more narrow time window. This ‘allor-nothing’ strategy would make panicles the ‘fittest’ in tropical climates, where the length of the season for pollination and seed-development is constant over the years. However, when this season is variable, it is better to ‘play safe’ and make flowers over a longer period, as cymes and racemes do. This would explain why panicles seem to occur more frequently in the tropics, and why cymes and racemes are more abundant in temperate zones [7]. Risks can also be spread by a perennial lifestyle (i.e. flowering in multiple years). One can easily imagine many more factors influencing fitness (e.g. availability of specific pollinators in relation to flower color, seed dispersal strategy, etc.) indicating that selection of inflorescence types is highly complex and difficult to unravel in detail. However, the
central message remains: inflorescences cannot evolve freely through the available morphospace, but are constrained by selection to narrow paths of high fitness, or evolutionary ‘wormholes’. Thus selection limits the ability of inflorescences to evolve freely, but it might also drive the diversification of inflorescence structure in response to changing conditions and explain why this occurred so frequently. Acknowledgements I thank Alexandra Rebocho for help with preparation of the artwork. Research in my laboratory is supported by a Vici grant of the Netherlands Organization for Scientific Research.
References 1 Carroll, S.B. (2005) Evolution at two levels: on genes and form. PLoS Biol.3, e245 DOI: 10.1371/journal.pbio.0030245 (http://biology. plosjournals.org) 2 Hoekstra, H.E. and Coyne, J.A. (2007) The locus of evolution: evo devo and the genetics of adaptation. Evolution Int. J. Org. Evolution 61, 995– 1016 3 Prud’homme, B. et al. (2007) Emerging principles of regulatory evolution. Proc. Natl. Acad. Sci. U. S. A. 104 (Suppl 1), 8605–8612 4 Bowman, J.L. et al. (2007) Green genes – comparative genomics of the green branch of life. Cell 129, 229–234 5 Doebley, J. and Lukens, L. (1998) Transcriptional regulators and the evolution of plant form. Plant Cell 10, 1075–1082 6 Watson, L. and Dalwitz, M.J. (2007) The families of flowering plants: descriptions, illustrations, identification, and information retrieval. Version: 1st June 2007 (1992 onwards) http://delta-intkey.com 7 Prusinkiewicz, P. et al. (2007) Evolution and development of inflorescence architectures. Science 316, 1452–1456 8 Jack, T. (2004) Molecular and genetic mechanisms of floral control. Plant Cell 16, S1–S17 9 Conti, L. and Bradley, D. (2007) TERMINAL FLOWER1 is a mobile signal controlling Arabidopsis architecture. Plant Cell 19, 767–778 10 Souer, E. et al. (1998) Genetic control of branching pattern and floral identity during Petunia inflorescence development. Development 125, 733–742 11 Molinero-Rosales, N. et al. (1999) FALSIFLORA, the tomato orthologue of FLORICAULA and LEAFY, controls flowering time and floral meristem identity. Plant J. 20, 685–693 12 Ahearn, K.P. et al. (2001) NFL1, a Nicotiana tabacum LEAFY-like gene, controls meristem initiation and floral structure. Plant Cell Physiol. 42, 1130–1139 13 Amighta, I. et al. (1999) Expression of CENTRORADIALIS (CEN) and CEN-like genes in tobacco reveals a conserved mechanism controlling phase change in diverse species. Plant Cell 11, 1405–1418 14 Pena, L. et al. (2001) Constitutive expression of Arabidopsis LEAFY or APETALA1 genes in citrus reduces their generation time. Nat. Biotechnol. 19, 263–267 15 Lifschitz, E. and Eshed, Y. (2006) Universal florigenic signals triggered by FT homologues regulate growth and flowering cycles in perennial day-neutral tomato. J. Exp. Bot. 57, 3405–3414
1360-1385/$ – see front matter ß 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.tplants.2007.11.004
Reproduction of material from Elsevier articles Interested in reproducing part or all of an article published by Elsevier, or one of our article figures? If so, please contact our Global Rights Department with details of how and where the requested material will be used. To submit a permission request online, please visit:
www.elsevier.com/locate/permissions 3
Evolution and development of virtual inflorescences Ronald Koes Institute for Molecular Cell Biology, Graduate School of Experimental Plant Sciences, Vrije Universiteit, de Boelelaan 1085, 1081HV Amsterdam, The Netherlands
The architecture of inflorescences diverged during the evolution of distinct plant families by mechanisms that remain unknown. Using computer modeling, Przemyslaw Prusinkiewicz and colleagues established a single model for the development of distinct inflorescences. Selection restricts inflorescence evolution to high fitness paths that vary with climate and other factors that influence reproductive success – explaining why some evolutionary transitions are more likely than others. This model presents an important framework for future plant ‘evo-devo’ research.
Evolution of body shape An eye-catching manifestation of evolution is the diversity of body architecture among species. One of the most elementary discoveries in developmental biology was that many genes that determine the identity and anatomy of large body parts (e.g. segments) or organs, for example Homeobox (HOX) genes, are conserved between animals as anatomically diverse as nematodes, flies and mammals. If this is the case, what causes anatomical diversity? The prevailing view in evolutionary developmental biology (‘evo-devo’) is that animal body shapes evolved mostly by changes in gene-expression patterns rather than through the emergence of new genes [1]. Whether this occurred primarily via changes in cis-regulatory elements in gene promoters or through alterations in the coding sequences of trans-acting transcription factor genes is difficult to ascertain from the limited data available and is a subject of debate [2], however there is evidence for both mechanisms [1,3]. The diversification of animals is mirrored by that of plants. For example, angiosperms, which contain the vast majority of plant species, evolved in < 300 million years and generated an enormous diversity in shapes, numbers and positions of organs [4], compared with the evolution of vertebrate animals, which is thought to have taken some 500 million years. It was suggested early on that plant shape evolved primarily through alterations in transcriptional control [5] but supporting evidence remains scarce. Because a range of plant species with diverse morphologies are amenable to forward genetic analyses (e.g. maize, rice, Arabidopsis, tomato, petunia etc.) and/or transgenesis (numerous species), they offer excellent possibilities for evo-devo research that have hardly been explored to date.
Corresponding author: Koes, R. ([email protected]).
When and where to make flowers? Plant species exhibit a large diversity in when and where they form flowers. Flowers can arise either as single flowers at the end of a branch, or they can be arranged into complex inflorescences that bear many flowers. The inflorescences found in nature fall into three main classes (Figure 1a). In racemes, flowers appear along the flanks of axes (branches). In cymes, the primary axis terminates by forming a flower and growth continues from a lateral axis that repeats this pattern. In panicles, primary and higher order lateral axes terminate in a flower. Most plant families contain species with distinct inflorescence types, suggesting that they evolved from one another multiple times independently [6]. How this occurred remains unknown; however, Prusinkiewicz et al. [7] have presented an extensive model that merges developmental genetics with population biology to explain how it might have taken place. The development of virtual inflorescences In their paper Prusinkiewicz et al. [7] address two questions. First, why did evolution produce only three main inflorescence types and not others? Second, how did these inflorescence types evolve? They devised a simple computer model to show that distinct inflorescence types might be formed by variations in a common mechanism. In these virtual inflorescences, the identity of floral and shoot meristems is reflected by a quantitative character called ‘vegetativeness’ (veg). Thus, veg is not a compound but a measure of the ‘state’ of a meristem. If veg is high, a meristem will produce a new metamer with a lateral meristem, but if veg drops below a certain threshold, the meristem terminates by forming a flower. Initially, veg levels are high but generally decrease over time as a function of plant and/ or meristem age. If veg decreases with similar kinetics in all meristems, a panicle is produced, in which meristems first proliferate and at some point all turn into a flower (Figure 1b, left). If veg decreases quickly and reaches the minimum threshold during development of the first metamer, a simple panicle with only two flowers will form, but if veg decreases more slowly, a more compound structure with many flowers is generated (Figure 1a). The creation of cymes and racemes requires differential regulation of veg in newly emerging ‘immature’ lateral meristems and the ‘mature’ apical meristems, and thus requires more complexity. If veg is lower in the apical meristems than in the lateral meristems, a cyme is created. By contrast, if veg is lower in lateral meristems, a raceme is formed (Figure 1a). To ensure that the co-florescences develop into bona fide racemes rather than terminating with a flower, the immature state of lateral meristems
Update
TRENDS in Plant Science Vol.13 No.1
Figure 1. Development of distinct virtual inflorescence structures. (a) Structure of inflorescences and position in morphospace. Flowers are indicated by red circles and meristems by green arrows. The inflorescence types are positioned in a 2D morphospace defined by the time required for apical (Tapical) and lateral (Tlateral) meristems to acquire floral fate. (b) Expression of veg in compound panicle (left), raceme (middle) and cyme (right). The black line depicts veg levels in the primary apical meristem. The colored lines depict veg in the first (red), second (green) and third (orange) lateral meristems formed by the primary apex. Adapted from [7].
needs to be transient. If veg does not reach the threshold during this immature phase the meristem turns into an (mature) apical meristem with corresponding veg levels and continues to grow accordingly (Figure 1b). Thus, inflorescence architectures are defined by only two variables – the times at which apical and lateral meristems attain low levels of veg – and can be plotted in a 2D ‘morphospace’ (Figure 1a). In this morphospace cymes and racemes are at opposite ends and are separated by panicles with variable degrees of complexity. It is noteworthy that other inflorescences with distinct arrangements of flowers and meristems cannot be produced by this simple model or by nature. This is because such structures cannot be made when all apical and all lateral meristems display the same veg decay, and would therefore require additional complexity. A virtual Arabidopsis Two obvious questions are: do these virtual inflorescences have any relation to real ones, and can veg be explained in 2
terms of molecules? The answer to both these questions is ‘yes’. Genetic experiments identified several genes that determine the racemose architecture of the Arabidopsis inflorescence. The most important ones are LEAFY (LFY) and TERMINAL FLOWER 1 (TFL1). In lfy mutants, flowers are replaced by shoot-like structures, whereas tfl1 generates short inflorescences that terminate with a flower [8]. Therefore, TFL1 promotes veg and LFY represses veg. In the wild-type, TFL1 mRNA is expressed in the apical meristem [9] and LFY in the lateral floral meristems, partly because LFY and TFL1 downregulate each other [8]. These, and additional genetic data, have been incorporated in the model to compute veg, which resulted in a virtual Arabidopsis that recreates the wild-type, in addition to single and double mutants containing gainand/or loss-of-function alleles of TFL1 and LFY [7]. A more rigorous test would be to determine the validity of the model for distinct inflorescence types, such as cymes and panicles. Cymes and panicles have been studied less
Update
TRENDS in Plant Science Vol.13 No.1
intensively and the genetic control of veg is only partially understood. In the cymose inflorescences of tomato (Solanum lycopersicon) and petunia (Petunia hybrida), LFY homologs are expressed within the apical meristem where they specify floral identity and, thus, downregulate veg [10,11]. In the emerging lateral meristems of petunia the expression of the LFY homolog is delayed, or transiently repressed [10]. In tobacco (Nicotiana tabacum), which is closely related to petunia and tomato, constitutive expression of LFY results in a solitary terminal flower [12], indicating that the transient inactivity of LFY in the lateral meristem is essential for formation of a cyme, as predicted by the theoretical model. Furthermore, putative TFL homologs from tobacco are not expressed in the apex of inflorescences [13], which might explain the decrease of veg in the apical meristem and the formation of a terminal flower. However, other observations are less easy to fit with the model. For example, ectopic expression of LFY in citrus trees causes early flowering (indicating that LFY can inhibit veg in this species) but does not alter inflorescence structure [14]. Furthermore, SINGLE FLOWER TRUSS (SFT) from tomato is, like its Arabidopsis homolog FLOWERING LOCUS T, a mobile protein that promotes flowering [15]. This indicates that SFT represses veg and would predict that sft produces a complex cyme, rather than the solitary flower that is observed. This does not necessarily imply that the model is incorrect, but that the genetic regulation of veg in cymes involves mechanisms that are not understood and which are in part distinct from those in Arabidopsis. Who is the fittest? Who will survive? Prusinkiewicz et al. took their model one step further by considering how selection might act on inflorescence structures. Panicles produce many more flowers than cymes and racemes, but in a more narrow time window. This ‘allor-nothing’ strategy would make panicles the ‘fittest’ in tropical climates, where the length of the season for pollination and seed-development is constant over the years. However, when this season is variable, it is better to ‘play safe’ and make flowers over a longer period, as cymes and racemes do. This would explain why panicles seem to occur more frequently in the tropics, and why cymes and racemes are more abundant in temperate zones [7]. Risks can also be spread by a perennial lifestyle (i.e. flowering in multiple years). One can easily imagine many more factors influencing fitness (e.g. availability of specific pollinators in relation to flower color, seed dispersal strategy, etc.) indicating that selection of inflorescence types is highly complex and difficult to unravel in detail. However, the
central message remains: inflorescences cannot evolve freely through the available morphospace, but are constrained by selection to narrow paths of high fitness, or evolutionary ‘wormholes’. Thus selection limits the ability of inflorescences to evolve freely, but it might also drive the diversification of inflorescence structure in response to changing conditions and explain why this occurred so frequently. Acknowledgements I thank Alexandra Rebocho for help with preparation of the artwork. Research in my laboratory is supported by a Vici grant of the Netherlands Organization for Scientific Research.
References 1 Carroll, S.B. (2005) Evolution at two levels: on genes and form. PLoS Biol.3, e245 DOI: 10.1371/journal.pbio.0030245 (http://biology. plosjournals.org) 2 Hoekstra, H.E. and Coyne, J.A. (2007) The locus of evolution: evo devo and the genetics of adaptation. Evolution Int. J. Org. Evolution 61, 995– 1016 3 Prud’homme, B. et al. (2007) Emerging principles of regulatory evolution. Proc. Natl. Acad. Sci. U. S. A. 104 (Suppl 1), 8605–8612 4 Bowman, J.L. et al. (2007) Green genes – comparative genomics of the green branch of life. Cell 129, 229–234 5 Doebley, J. and Lukens, L. (1998) Transcriptional regulators and the evolution of plant form. Plant Cell 10, 1075–1082 6 Watson, L. and Dalwitz, M.J. (2007) The families of flowering plants: descriptions, illustrations, identification, and information retrieval. Version: 1st June 2007 (1992 onwards) http://delta-intkey.com 7 Prusinkiewicz, P. et al. (2007) Evolution and development of inflorescence architectures. Science 316, 1452–1456 8 Jack, T. (2004) Molecular and genetic mechanisms of floral control. Plant Cell 16, S1–S17 9 Conti, L. and Bradley, D. (2007) TERMINAL FLOWER1 is a mobile signal controlling Arabidopsis architecture. Plant Cell 19, 767–778 10 Souer, E. et al. (1998) Genetic control of branching pattern and floral identity during Petunia inflorescence development. Development 125, 733–742 11 Molinero-Rosales, N. et al. (1999) FALSIFLORA, the tomato orthologue of FLORICAULA and LEAFY, controls flowering time and floral meristem identity. Plant J. 20, 685–693 12 Ahearn, K.P. et al. (2001) NFL1, a Nicotiana tabacum LEAFY-like gene, controls meristem initiation and floral structure. Plant Cell Physiol. 42, 1130–1139 13 Amighta, I. et al. (1999) Expression of CENTRORADIALIS (CEN) and CEN-like genes in tobacco reveals a conserved mechanism controlling phase change in diverse species. Plant Cell 11, 1405–1418 14 Pena, L. et al. (2001) Constitutive expression of Arabidopsis LEAFY or APETALA1 genes in citrus reduces their generation time. Nat. Biotechnol. 19, 263–267 15 Lifschitz, E. and Eshed, Y. (2006) Universal florigenic signals triggered by FT homologues regulate growth and flowering cycles in perennial day-neutral tomato. J. Exp. Bot. 57, 3405–3414
1360-1385/$ – see front matter ß 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.tplants.2007.11.004
Reproduction of material from Elsevier articles Interested in reproducing part or all of an article published by Elsevier, or one of our article figures? If so, please contact our Global Rights Department with details of how and where the requested material will be used. To submit a permission request online, please visit:
www.elsevier.com/locate/permissions 3