Focusing on TERMINAL FLOWER 1 (TFL1) regulation

by Timo Hytönen
University of Helsinki, Finland

FLOWERING LOCUS T (FT) and TERMINAL FLOWER 1 (TFL1) are small closely related proteins that have a great impact on the production of agricultural and horticultural crops and forest trees. FT functions as a mobile signaling molecule that travels from leaves to shoot meristems to mediate information about the suitable season to induce flowers, produce tubers or set a bud (reviewed by Wickland & Hanzawa, 2015). TFL1, in contrast, is a repressor of flowering that is expressed locally in meristems. In species with indeterminate growth habit, one of the main functions of TFL1 is to maintain indeterminate inflorescence meristem and to allow flower bud development only in the flanks of the meristem. This has a direct effect on yield because it affects the number of seeds or fruits. In some other species with closed inflorescence structures, high TFL1 expression can completely prevent floral development (Costes et al., 2014). Both FT and TFL1 are transcriptional cofactors that bind with the same transcription factor FD, and there is increasing evidence that the balance of these antagonistic signals determines the developmental output (Hanano & Goto, 2011; Wickland & Hanzawa, 2015). The molecular control of FT has been studied in detail especially in Arabidopsis, but much less is known about factors controlling spatiotemporal expression pattern of TFL1.


Indeterminate inflorescences at the lava desert of El Teide.

In their recent study, Serrano-Mislata et al. (2016) explored cis-regulatory elements of TFL1 using various experimental approaches including mutant complementation, phylogenetic shadowing and promoter::GUS fusion lines. First, they tried to complement tfl1 mutant using a genomic construct containing full 5’ and 3’ intergenic regions and a similar construct lacking introns. Both constructs similarly complemented the mutant phenotype indicating that introns are not needed for the transcriptional regulation of TFL1. Next, using phylogenetic shadowing of TFL1 orthologues of several Brassicaceae species, they found seven conserved blocks that were tested further using genomic constructs of different lengths as well as similar constructs containing GUS reporter in the place of TFL1. Using these constructs the authors successfully dissected the roles of different promoter blocks in controlling TFL1 expression and shoot architecture. They found that 300 bp upstream and 3.3 kb downstream regions are needed to fully complement the defects of tfl1 mutant and to drive similar expression of GUS reporter than the full-length genomic construct. This short 5’ region is needed to maintain high TFL1 expression level, whereas separate 3’ elements control its spatiotemporal expression in different meristems. A 3’ region +2.8-3.3 kb after the stop codon is needed to maintain TFL1 expression in the inflorescence meristem, and another region at +1.6-2.2 kb controls its expression in axillary meristems. Finally, 3’ region between +1.0 and +1.3 is required to control flowering time by affecting TFL1 expression in vegetative meristems as well as its up-regulation following floral transition.


TFL1 promoter elements controlling its spatiotemporal expression pattern (based on Serrano-Mislata et al. (2016).

The authors suggested that a modular structure of TFL1 promoter might facilitate gene evolution to generate different plant architectures as already observed in Leavenforthia crassa. More importantly, the identification of these modules is instrumental for future research focusing on upstream regulators of TFL1 that may control inflorescence architecture and/or flowering time in different species. One important question is what is causing the up-regulation of TFL1 in the vegetative meristem upon flower induction. One of the suggested regulators is a MADS transcription factor XAANTAL2 that shows sequence similarity with SUPPRESSOR OF THE OVEREXPRESSION OF CONSTANS1 (SOC1) (Pérez-Ruiz et al., 2015). Interestingly, also SOC1 is highly expressed in the inflorescence meristem (Immink et al., 2012), and the strawberry orthologue of SOC1 has been shown to up-regulate TFL1 (Mouhu et al., 2013). In addition, an early study indicated that CO may induce the expression of TFL1 (Simon et al., 1996), perhaps through FT. Using tools produced by Serrano-Mislata et al. (2016), time is now ripen to focus on the molecular control of TFL1 expression that may translate to increased crop yields.


Costes E, Crespel L, Denoyes B, Morel P, Demene M, Lauri P & Wenden B. 2014. Bud structure, position and fate generate various branching patterns along shoots of closely related Rosaceae species: a review. Frontiers in Plant Science 5: 666.

Hanano S & Goto K. 2011. Arabidopsis TERMINAL FLOWER1 is involved in the regulation of flowering time and inflorescence development through transcriptional repression. The Plant Cell 23: 3172–3184.

Immink RGH, Posé D, Ferrario S, Ott F, Kaufmann K, Valentim FL, de Folter S, van der Wal F, van Dijk ADJ, Schmid M & Angenent GC. 2012. Characterization of SOC1’s central role in flowering by the identification of its upstream and downstream regulators. Plant Physiology 160: 433–449.

Mouhu K, Kurokura T, Koskela EA, Albert VA, Elomaa P & Hytönen T. 2013. The Fragaria vesca homolog of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 represses flowering and promotes vegetative growth. The Plant Cell 25: 3296–3310.

Pérez-Ruiz RV, García-Ponce B, Marsch-Martínez N, Ugartechea-Chirino Y, Villajuana-Bonequi M, de Folter S, Azpeitia E, Dávila-Velderrain J, Cruz-Sánchez D, Garay-Arroyo A, de la Paz Sánchez M, Estévez-Palmas JM, Álvarez-Buylla ER. 2015. XAANTAL2 (AGL14) is an important component of the complex gene regulatory network that underlies Arabidopsis shoot apical meristem transitions. Molecular Plant 8: 796–813.

Serrano-Mislata A, Fernández-Nohales P, Doménech MJ, Hanzawa J, Bradley J, Madueño F. 2016. Separate elements of the TERMINAL FLOWER 1 cis-regulatory region integrate pathways to control flowering time and shoot meristem identity. Development 143: 3315–3327.

Simon R, Igeño MI, Coupland G. 1996. Activation of floral meristem identity genes in Arabidopsis. Nature 384: 59–62.

Wickland DP & Hanzawa Y. 2015. The FLOWERING LOCUS T/TERMINAL FLOWER 1 gene family: functional evolution and molecular mechanisms. Molecular Plant 8: 983–997.

Posted in flowering | Tagged , , , , | Leave a comment

A new comparative approach to understand local adaptation

Martin Lascoux
Department of Ecology and Genetics, Uppsala University

This is a flowering highlight that only indirectly concerns flowering. Furthermore the study does not address the molecular basis of flowering as is often the case in these columns but instead is highly relevant to the understanding of its adaptive basis. Indeed the study by Yeaman et al. (2016) may well herald a new era in the analysis of local adaptation.  While there is ample evidence of local adaptation for flowering time and other putatively adaptive traits (Savolainen et al., 2013) it has proven harder to identify the genes associated with it. There are various methods to do so, all of them with strengths and weaknesses. One popular approach is to test for statistical association between flowering time variation and genetic polymorphisms along the genome.

local adaptation.jpg

Picea obovata on Mount Iremel in south Urals

A major issue with this approach is that the signal of local adaptation can be confounded by population genetic structure. For example, if one samples plants along a latitudinal gradient in Scandinavia, any difference between individuals from the North and the South could reflect the fact that Scandinavia was recolonized along two main routes after the Last Glacial Maximum (18,000 years ago), one entering Scandinavia from the south and the other from the north. So plants from northern Scandinavia can differ from their southern counterparts simply because they have different origins and differences between them cannot automatically be assigned to adaptation to the local environment. Of course, there are methods to correct for population structure, but if adaptation and demographic history go hand in hand, removing one may well end up removing the other leading to false negatives. How did Yeaman et al. (2016) address this thorny issue? Well, they chose not to address it directly; instead they took a different road. Here is the main thrust of their reasoning. Lodgepole pine (Pinus contorta) and interior spruce (Picea glauca, Picea engelmannii and their hybrids) are two common conifer species in western Canada. Pines and spruces diverged some 140 million years ago, and we therefore do not expect to see shared polymorphism left between the two species. Being a complex species interior spruce has a particularly intricate population structure. So rather than risking a high number of false negatives by attempting to correct for population structure they assessed the relationship between genome-wide SNP variation, on the one hand, and 17 phenotypic traits assessed in growth chamber and 22 environmental variables, on the other hand, in both species. This was done without correcting for population structure. For each species they then obtained a list of top candidate genes. Of course many of them are likely false positives.  But which ones are false positives and which ones true candidate genes? In order to find out which genes could be good candidate genes for local adaptation, and incidentally test for convergent local adaptation, the authors looked for genes that are top candidates in both species. Top candidates were defined as genes with an exceptional proportion of their total SNPs being associated to either phenotype or environment. Depending on how stringent the False Discovery Rate (FDR) is the number of genes with strong signature of convergent local adaptation varied between 6 and 83. For a FDR=0.05 the number was 47. Many of those 47 top candidates had conserved patterns of differential expression in both species and/or were enriched for transcription factors and genes involved in biological regulation.  Several of the convergent genes such as Pseudo-response regulator 5 (PPR5) that regulates the circadian clock or FY that indirectly regulates FLOWERING LOCUS C are part of the pathways controlling flowering time in A. thaliana and are prime candidates for being involved in the control of phenology in trees. So, after all, this study is not that far from the usual topic of this column. But, in my view, its foremost value is its very innovative use of comparative genomics. It is not the first comparative study of local adaptation in trees (see Chen et al., 2012, 2014) but it is certainly the first to be carried out on a genomic scale and on highly diverged species.  I do not doubt that it will have a strong impact on future studies of local adaptation for phenological traits, in general, and on flowering time, in particular. Associated with functional studies the general approach developed by Yeaman et al. (2016) could lead to new insights on the evolution of the basic molecular mechanisms associated with local adaptation.


Yeaman S., Hodgins KA, Lotterhos KE, Suren H, Nadeau S, Degner JC, et al. 2016. Convergent local adaptation to climate in distantly related conifers. Science 353(6306), 1431–1433.

Chen J, Källman, T, Ma X, Gyllenstrand N, Zaina G, Morgante M, et al. 2012. Disentangling the roles of history and local selection in shaping clinal variation of allele frequencies and gene expression in Norway spruce (Picea abies). Genetics 191, 865–881.

Chen J, Tsuda Y, Stocks M, Källman T, Xu N, Kärkkäinen K, et al. 2014. Clinal variation at phenology-related genes in spruce: parallel evolution in FTL2 and Gigantea? Genetics 197(3), 1025–1038.

Savolainen O, Lascoux M, & Merilä J. 2013. Ecological genomics of local adaptation. Nature Reviews Genetics 14(11), 807–820.

Posted in flowering | Tagged , , , , , , , | Leave a comment

New insights into LEAFY structure and function

by Frank Wellmer
Smurfit Institute of Genetics, Trinity College Dublin, Ireland

The plant-specific transcription factor LEAFY (LFY) is a master regulator of early flower development (Weigel et al., 1992). It orchestrates the onset of flowering and is involved in activating the expression of floral organ identity genes (Busch et al., 1999; Parcy et al., 1998), which specify the different types of floral organs. Because of LFY’s central role in reproductive development, its function and molecular evolution have been studied in detail over the past 25 years using a multitude of different experimental approaches. In fact, there are arguably few plant transcriptional regulators that have been studied as extensively as LFY. However, the protein structure of LFY remained unknown for many years, precluding detailed insights in the molecular mechanism underlying its activity. This knowledge gap was partially closed in 2008, when the structure of the highly conserved carboxy-terminal DNA binding domain was resolved through X-ray crystallography (Hames et al., 2008). This work showed that the DNA binding domain is composed of a helix-turn-helix motif and binds DNA as a dimer. In contrast, the function of the amino-terminal half of LFY, which contains another conserved domain, remained unknown.


A LEAFY dimer bound to DNA. The N-terminal SAM and C-terminal DNA binding domains are shown. Image kindly provided by Dr. François Parcy.

Using LFY from the gymnosperm Gingko biloba, Sayou et al., 2016 recently succeeded in crystallizing the N-terminal domain and resolved its structure at 2.3 Å resolution. They found that the domain resembles a Sterile Alpha Motif (SAM) and is composed of five α-helices that are connected by four loops. SAM domains are common structural motifs in eukaryotes mediating the interaction with nucleic acids, lipids and other proteins. Some SAM domains have also been shown to trigger protein oligomerization. The SAM domain in LFY appears to belong to this latter group as indicated by the formation of LFY-SAM domain polymer chains in a crystal, where monomers contact each other in a head-to-tail arrangement.
What is the functional relevance of this oligomerization event and is it required for LFY activity? To investigate this, Sayou et al. first identified the amino acid residues that are essential for oligomerization. To this end, they used the structural information they had obtained as well as data from the analysis of other SAM domain-containing proteins. They then carried out a series of experiments with modified LFY proteins in which these essential amino acid residues had been substituted, leading to the suppression of oligomerization. They found that the expression of the mutated form of LFY in plants resulted in a much reduced activity when compared to wild-type LFY protein, thus implying that oligomerization is required for proper LFY function. Additional tests showed that the SAM domain prevents LFY from binding to DNA as a monomer and hence promotes dimer formation. Furthermore, the results of genome-wide localization studies showed that the mutated form of LFY exhibited a substantial decrease in its overall ability to bind to DNA.
In addition to regulating the binding properties of LFY, the SAM domain also appears to be involved in binding site selection. Specifically, Sayou et al. showed that the SAM domain mediates cooperative binding of LFY protein oligomers to multiple binding sites and facilitates binding to sites with low affinity for LFY. Thus, the SAM domain plays a central role in regulating the DNA binding activity of the LFY transcription factor.
Another interesting observation reported by Sayou et al. is that the SAM domain appears to promote binding of LFY to regions of chromatin with low accessibility. Because it has been shown previously that LFY can interact with chromatin remodelling factors it is possible that LFY acts as a ‘pioneer transcription factor’ and recruits these proteins to regions of closed chromatin to bring about changes in chromatin conformation and gene expression. Notably, it has been suggested that some of the floral organ identity factors, which mediate floral organ specification, also function as pioneer factors (Pajoro et al., 2014). These include the MADS-domain protein APETALA1, which controls early flower development together with LFY. Thus, pioneer factor activity may be at the core of the mechanism that regulates the onset of flower formation, possibly explaining the massive changes in gene expression observed during this developmental stage.

In summary, the detailed insights into LFY structure and function obtained by Sayou et al. represent a major breakthrough in the analysis of this master regulator of flower development. They will undoubtedly open up new avenues for experimentation that will lead to an even deeper mechanistic understanding of the activities of LFY during plant reproduction.

Busch MA, Bomblies K, Weigel D. 1999. Activation of a floral homeotic gene in Arabidopsis. Science 285, 585-587.

Hames C, Ptchelkine D, Grimm C, Thevenon E, Moyroud E, Gerard F, Martiel JL, Benlloch R, Parcy F, Muller CW. 2008. Structural basis for LEAFY floral switch function and similarity with helix-turn-helix proteins. Embo Journal 27, 2628-2637.

Pajoro A, Madrigal P, Muino JM, Matus JT, Jin J, Mecchia MA, Debernardi JM, Palatnik JF, Balazadeh S, Arif M, O’Maoileidigh DS, Wellmer F, Krajewski P, Riechmann JL, Angenent GC, Kaufmann K. 2014. Dynamics of chromatin accessibility and gene regulation by MADS-domain transcription factors in flower development. Genome Biology 15, R41.

Parcy F, Nilsson O, Busch MA, Lee I, Weigel D. 1998. A genetic framework for floral patterning. Nature 395, 561-566.

Sayou C, Nanao MH, Jamin M, Pose D, Thevenon E, Gregoire L, Tichtinsky G, Denay G, Ott F, Peirats Llobet M, Schmid M, Dumas R, Parcy F. 2016. A SAM oligomerization domain shapes the genomic binding landscape of the LEAFY transcription factor. Nature Communications 7, 11222.

Weigel D, Alvarez J, Smyth DR, Yanofsky MF, Meyerowitz EM. 1992. LEAFY controls floral meristem identity in Arabidopsis. Cell 69, 843-859.

Posted in flowering, LEAFY | Tagged , , , | Leave a comment

The interactions they are a changin’

New study of the complex role of protein interactions in the evolution of flower development

by Aalt-Jan Van Dijk

Among the various transcription factors involved in regulating flowering, no doubt MADS domain proteins involve some of the best studied. B-class proteins are a specific type of MADS domain proteins, named after their role in the classic ABC model for floral development. According to the ABC model, B-class proteins are involved in petal and stamen formation; for more background information on this model, you might want to read the recent post by Theissen and Parcy  (which focusses on other components of the ABC model, namely A-class proteins). B-class proteins can either form heterodimers, involving two different B-class proteins, or homodimers. Variation in B-class proteins has been suggested to be relevant for variation in floral organ development. However, a lot is still unclear about how variation in these proteins and their interactions influences phenotypic differences related to flowering. Better understanding of the evolution of B-class dimerization is clearly needed. A recent paper in Molecular Biology and Evolution  (Bartlett et al., 2016) uncovers various layers of complexity of this evolution.

First, Bartlett et al., 2016 characterized obligate heterodimerization versus homodimerization in taxa spanning the Poales (the order that contains the grass family), and found multiple transitions between factultative homodimerization and obligate heterodimerization. Such evolutionary changes were also present specifically within the grasses.

The next layer of complexity is, that there is a clear context-dependence of the effect of specific amino acids on the dimerization landscape. This is demonstrated by results from the experiments presented by Bartlett et al. to find sequence regions influencing homodimerization versus heterodimerization. These experiments involved three B-class proteins, two of which formed homodimers: J-PI (from the grass relative Joinvillea) and BdSTS1 (from Brachypodium distachyon) and a third one forming heterodimers (STS1, a maize protein). See figure for a schematic overview of the main findings of these experiments.


Context-dependence of the effect of specific domains on dimerization.
Ovals represent B-class proteins: green, B. distachyon BdSTS1; blue, maize STS1; red, Joinvillea J-PI. Dashed line indicates homodimerization. The STS1 I-domain (small blue oval) disrupts homodimerization in the context of J-PI, but not so in BdSTS1. Similarly, the J-PI I-domain (small red oval) enables homodimerization of STS1, but the BdSTS1 I-domain (small green oval) does not.

It was found that the I-domain of J-PI   is sufficient for homodimerization of STS1; here the I-domain refers to a specific domain in MADS proteins. This domain is in fact known to be relevant for MADS domain interaction specificity in general. However the I-domain of BdSTS1 was not sufficient for homodimerization of STS1. Reciprocally, the STS1 I-domain was sufficient to abolish J-PI homodimerization but did not affect BdSTS1 homodimerization.  A comparison of I-domain sequence of STS1 and J-PI showed four amino acid differences. A single amino acid change (Gly81 to Asp) was sufficient to confer homodimerization ability on STS1, and the reciprocal change of Asp to Gly prevented J-PI from homodimerization. Intriguingly, however, the homodimerizing BdSTS1 I-domain also contains Asp at position 81, but this domain was not sufficient to confer homodimerization ability to STS1. BdSTS1 contains Ile at position 73, and introducing this Ile abolished the homodimerization capacity that the Gly81Asp mutation conferred on STS1. In brief, this indicates that a specific amino acid (Asp) can be sufficient to allow homodimerization in one sequence context but not so in another sequence context. This context-dependence of the role of specific amino acids provides insight to the multiple evolutionary routes via which B-class heterodimerization versus homodimerization evolved.

It is interesting to see that two computational methods contributed to the identification of relevant amino acids (Bartlett et al., 2016). One involves the analysis of positive selection. The other involves the prediction of interaction motifs. Both approaches were used for further identification of amino acids that contribute to dimerization landscape.

A final level of complexity in their analysis is the interconnectedness of coding and non-coding changes. Note that Theissen and Parcy on their post on the floral A-function gave an example of the importance of non-coding changes. Bartlett et al. found that the (homodimerizing) J-PI and the (non-homodimerizing) STS1 both showed similar rescue of an Arabidopsis pi mutant, and both could not rescue an ap3 mutant. (PI and AP3 are the two Arabidopsis B-class proteins). However, they found a distinction between J-PI and STS1 when they were expressed at a high level in developing sepals. Only J-PI expression resulted in transformation of the first whorl organs into petals. This shows that J-PI, when expressed at a high level in a novel domain, differs from STS1 in its effect on floral development. The J-PI homodimer on its own is not sufficient to confer B-class function in Arabidopsis, but as a novel protein complex in a novel domain it can effect phenotypic change. The interactions they are a changin – but the effect of changing protein interactions clearly depends on the expression pattern of the protein in question, and of its interactors.

Bartlett M, Thompson B, Brabazon H, Del Gizzi R, Zhang T and Whipple C. 2016. Evolutionary Dynamics of Floral Homeotic Transcription Factor Protein–Protein Interactions. Molecular Biology and Evolution 33(6):1486–1501. doi:10.1093/molbev/msw031

Posted in flowering | Leave a comment

What a difference a CArG-box makes

New insights into the Development and Evolution of Floral Meristem and Organ Identity in Arabidopsis

Günter Theißen1 & Francois Parcy2

1Friedrich Schiller University Jena, Department of Genetics, Philosophenweg 12, D-07743 Jena, Germany;
2 Laboratory of Plant & Cell Physiology, CNRS, CEA, Univ. Grenoble Alpes, INRA, 38000 Grenoble, France

A report by Ye et al. (2016) about the relatively recent acquisition of a cis-regulatory element during the evolution of a MADS-box gene involved in flower development may help to answer some long-standing questions about flower development and evolution.

The first set of questions concerns the origin and conservation of the floral homeotic A function during evolution. In addition to the B and C functions, the A function is part of the iconic ABC model of flower development (Haughn and Somerville 1988; Coen and Meyerowitz 1991). The ABC model postulates three classes of homeotic functions (A, B, and C) that specify floral organ identity in a combinatorial way: the A function alone specifies sepals in the first floral whorl, A + B defines petals in the second whorl, B + C specifies stamens in the third whorl, and C defines carpels in the fourth whorl (Coen and Meyerowitz, 1991). According to the ABC model an antagonism between the A and C function is responsible for the spatial restriction of the A function in the outer two and the C function to the inner two whorls of the flower (Coen and Meyerowitz, 1991). In consequence, if all three homeotic functions work properly in an developing flower of the model plant Arabidopsis thaliana (Arabidopsis, for short), sepals, petals, stamens and carpels develop in whorls 1-4, respectively, as shown in the picture.

Gunther image.jpg

Photo kindly provided by Grégoire Denay

Even though the ABC model was originally largely based on mutant phenotypes of Arabidopsis (Coen and Meyerowitz 1991), mutational and gene expression analyses during the last 25 years revealed that the B and C functions are highly conserved throughout flowering plants and even in gymnosperms (Becker and Theißen 2003; Melzer et al. 2010).

In contrast, the concept of an A function has always been discussed controversially. For example, almost simultaneously to the ABC model an alternative model was proposed that focused on snapdragon (Antirrhinum majus) rather than Arabidopsis. This model postulated two developmental pathways in addition to a floral ground state (Schwarz-Sommer et al., 1990). These two pathways were historically (and confusingly) named A and B (Schwarz-Sommer et al., 1990), but they are equivalent to the B and C functions, respectively, of the ABC model (Theißen et al., 2000). According to the snapdragon-based model sepal development represents the developmental ‘default state’ of floral organ development and thus does not require an extra A function (Schwarz-Sommer et al., 1990). Indeed, except Arabidopsis, in almost all plants that have been investigated so far one does not find recessive mutants in which the identity of both types of perianth organs (sepals and petals) is affected (discussed by Theißen et al., 2000; Litt 2007).

However, compared to the B and C functions, even in Arabidopsis thaliana the A function appears not well defined (reviewed by Theißen et al., 2000; Litt, 2007; Causier et al., 2010). Originally, only the gene APETALA2 (AP2) was thought to provide the A-function (Coen and Meyerowitz, 1991), but later on also the gene APETALA1 (AP1) was established in the literature as an A-function gene (for a review, see Litt, 2007). According to the ABC model, the loss-of-function of these genes should lead to the development of carpels and stamens in the whorl 1 and 2, respectively, due to the ectopic expression of the C function gene. However, depending on environmental conditions and genetic background, the organs in the outer whorls of ap1 or ap2 loss-of-function mutants may also develop into leaf- or bract-like structures, or can even be missing; in case of ap1 mutants secondary flowers may eventually develop in the axils of leaves that replace sepals (reviewed by Litt, 2007; Causier et al., 2010). Thus, the A function that specifies sepal and petal organ identity and that antagonizes the C function is difficult to separate genetically from a more basic function of AP1 and AP2 in specifying floral meristem identity.

This raises the question as to why then does it seem that A function in Arabidopsis is an exception rather than the rule, and why is it less-well defined than the other floral homeotic functions? The recent publication by Ye et al. (2016) may have brought us much closer to reasonable answers to these questions.

In contrast to AP2, AP1 represents a MADS-box gene and thus encodes a MADS-domain transcription factor (Mandel et al., 1992). MADS-domain transcription factors bind to cis-regulatory elements termed CArG-boxes (for ‘C-A-rich-G’). Ye et al. (2016) have studied the AP1 gene in comparison to its closely related paralog CAULIFLOWER (CAL). While mutants in which CAL has lost its function (cal) have no obvious mutant phenotype, double mutants of AP1 and CAL (ap1 cal) have a much stronger mutant phenotype than ap1 mutants alone, in that cauliflower-like structures composed of proliferating meristems develop (Bowman et al. 1993; Kempin et al., 1995). AP1 and CAL may have originated from an ancestral gene by the a whole genome duplication close to the origin of Brassicaceae roughly about 40 – 50 million years ago (MYA) (Cheng et al., 2013; Franzke et al., 2016).

Ye et al. (2016) wanted to determine the molecular changes that are responsible for the differences of AP1 and CAL in gene expression and function. They identified a number of putative transcription factor binding sites (TFBS) that were gained or lost during evolution, the most striking one being a CArG-box in the promoter region of AP1 but not of CAL. The authors provide several lines of evidence suggesting that this ‘functional’ CArG-box brought AP1 under auto-regulatory and cross-regulatory control by the AP1 protein itself and the CAL protein, respectively. It was probably the gain of this CArG-box and the resulting regulatory control by AP1 and CAL that brought about an up-regulation of AP1 in sepals and petals during relatively late stages of flower development. This change in gene expression was obviously crucial for the functional importance of AP1 in the specification of the identity of these organs, and hence the contribution of AP1 to the A function.

It is then interesting to note that Ye et al. (2016) identified the crucial CArG-box only in two Arabidopsis species, but not in the closely related Capsella rubella or any other Brassicaceae species they investigated. This finding suggests that the CArG-box originated in the lineage that led to extant Arabidopsis, but after the lineage that led to Capsella had already branched-off. This would imply a relatively recent origin of this CArG-box roughly about 10 – 20 MYA. In mutational terms the origin of this cis-regulatory element seems not to have been a big deal; sequence comparisons comprising orthologous sequences from diverse Brassicaceae species suggest that it may have required only very few (possibly even only one or two) point mutations to transform a 10 base pair long precursor sequence into a sequence that is able to bind the MADS-domain proteins AP1 and CAL (Ye et al., 2016).

It is tempting to speculate that the involvement of AP1 in the floral homeotic A function, or even the whole A function as we know it from Arabidopsis thaliana, is thus a relatively recent evolutionary acquisition of the Arabidopsis lineage rather than a conserved feature of angiosperm flower development. This would be very much in line with the difficulties to identify an A function proper outside of Arabidopsis. So if any other species may turn up with AP1-like genes being involved in the specification of perianth organ identity, and hence a floral homeotic A function, they very likely acquired that function independently from the A function in Arabidopsis.

If our hypothesis is right, in Brassicaceae species other than Arabidopsis, an ap1 mutant may have no mutant phenotype, because the AP1 gene should have only a function in specifying floral meristem identity redundant with CAL. However, since in any lineage neo-functionalization cannot be excluded, a less strong prediction would be that the ap1 mutant in these species should at least do not show a class A loss-of-function phenotype. Fortunately, given that genome editing with CRISPR/Cas9 works very well in plants, testing such hypotheses in species such as Capsella rubella is not rocket science anymore.

The recent acquisition of a CArG-box in AP1 may also explain why the loss-of-function phenotype of the floral meristem identity gene LEAFY (LFY) is not as drastic as that of its orthologue in snapdragon, termed FLORICAULA (FLO) (Coen et al., 1990). In Arabidopsis lfy mutant, flowers are converted into leaves plus shoots at lower positions of the inflorescence, but at higher ones, LFY-independent activation of AP1 triggers the development of flower-like structures (albeit without class B gene activity) (Weigel and Meyerowitz, 1993). When both LFY and AP1 are mutated (lfy ap1), these pseudo-flowers disappear, showing that CAL cannot compensate for the loss of AP1 in this case (Huala and Sussex, 1992; Weigel et al., 1992). This strongly suggests that CAL, as opposed to AP1, cannot be induced independently of LFY (Parcy, 2005). The LFY-independent AP1 induction was shown to involve SVP and AGL24 (Grandi et al., 2012), two MADS-domain transcription factors that could also act through the very same CArG-box, although this remains to be directly established (Grandi et al., 2012; Gregis et al., 2013). The presence of this CArG-box could have thus blurred the central role that LFY plays in most species being the prime inductor of AP1. It is also interesting to note that AP1 and CAL, despite born by duplication from an ancestral single LFY target, appear to be induced through different mechanisms. AP1 is induced directly through identified LFY binding sites (Benlloch et al., 2011; Winter et al., 2011; Moyroud et al., 2011) whereas CAL, as opposed to what was originally proposed (William et al., 2004), appear to have no LFY binding site (Minguet et al., 2015; Winter et al., 2011; Moyroud et al., 2011) and is induced by LFY indirectly, through the action of LMI1 (Saddic et al., 2006). The two paralogs thus appear to have evolved by loss and gain of different TFBS.


GT thanks Rainer Melzer and Elliot M. Meyerowitz for helpful discussions about the history of the A function.


Becker A and Theißen G. 2003. The major clades of MADS-box genes and their role in the development and evolution of flowering plants. Molecular Phylogenetics and Evolution 29, 464-489.

Benlloch R, Kim MC, Sayou C, Thévenon E, Parcy F, and Nilsson O. 2011. Integrating long-day flowering signals: a LEAFY binding site is essential for proper photoperiodic activation of APETALA1. Plant Journal 67, 1094–1102.

Bowman JL, Alvarez J, Weigel D, Meyerowitz EM, and Smyth DR. 1993. Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development 119, 721-743.

Causier B, Schwarz-Sommer Z, and Davies B. 2010. Floral organ identity: 20 years of ABCs. Seminars in Cell & Developmental Biology 21, 73-79.

Cheng S, van den Bergh E, Zeng P, Zhong X, Xu J, Liu X, Hofberger J, de Bruijn S, Bhide AS, Kuelahoglu C, Bian C, Chen J, Fan G, Kaufmann K, Hall JC, Becker A, Bräutigam A, Weber AP, Shi C, Zheng Z, Li W, Lv M, Tao Y, Wang J, Zou H, Quan Z, Hibberd JM, Zhang G, Zhu XG, Xu X, and Schranz ME. 2013. The Tarenaya hassleriana genome provides insight into reproductive trait and genome evolution of crucifers. Plant Cell 25, 2813-2830.

Coen ES and Meyerowitz EM. 1991. The war of the whorls – Genetic interactions controlling flower development. Nature 353, 31-37.

Coen ES, Romero JM, Doyle S, Elliott R, Murphy G, and Carpenter R. 1990. Floricaula: a homeotic gene required for flower development in Antirrhinum majus. Cell. 63, 1311-22.

Franzke A, Koch MA, and Mummenhoff K. 2016. Turnip time travels: age estimates in Brassicaceae. Trends in Plant Science. 21, 554-531.

Grandi, V., Gregis, V., and Kater, M.M. 2012. Uncovering genetic and molecular interactions among floral meristem identity genes in Arabidopsis thaliana. Plant Journal. 69: 881–893.

Gregis, V., F. Andrés, A. Sessa, R. F. Guerra, S. Simonini, J. L. Mateos, S. Torti, F. Zambelli, G. M. Prazzoli, K. N. Bjerkan, P. E. Grini, G. Pavesi, L. Colombo, G. Coupland, and M. M. Kater. 2013. Identification of pathways directly regulated by SHORT VEGETATIVE PHASE during vegetative and reproductive development in Arabidopsis. Genome Biology. 14: R56.

Haughn GW and Somerville CR. 1988. Genetic control of morphogenesis in Arabidopsis. Developmental Genetics 9, 73-89.

Huala E and Sussex IM. 1992. LEAFY interacts with floral homeotic genes to regulate Arabidopsis floral development. Plant Cell 4, 901–913.

Kempin SA, Savidge B, and Yanofsky MF. 1995. Molecular basis of the cauliflower phenotype in Arabidopsis. Science 267, 522-525.

Litt A. 2007. An evaluation of A-function: Evidence from the APETALA1 and APETALA2 gene lineages. International Journal of Plant Sciences. 168, 73-91.

Mandel MA, Gustafson-Brown C, Savidge B, and Yanofsky MF. 1992. Molecular characterization of the Arabidopsis floral homeotic gene APETALA1. Nature 360, 273-277.

Melzer, R., Wang, Y.-Q., and Theißen, G. 2010. The naked and the dead: the ABCs of gymnosperm reproduction and the origin of the angiosperm flower. Sem Cell Dev Biol. 21, 118-128.

Minguet EG, Segard S, Charavay C, and Parcy F. 2015. MORPHEUS, a webtool for transcription factor binding analysis using position weight matrices with dependency. PLoS One 10: e0135586.

Moyroud E, Minguet EG, Ott F, Yant L, Posé D, Monniaux M, Blanchet S, Bastien O, Thévenon E, Weigel D, Schmid M, and Parcy F. 2011. Prediction of regulatory interactions from genome sequences using a biophysical model for the Arabidopsis LEAFY transcription factor. Plant Cell 23, 1293–1306.

Parcy F. 2005. Flowering: a time for integration. International Journal of Developmental Biology 49, 585–593.

Saddic L, Huvermann B, Bezhani S, Su Y, Winter CM, Kwon CS, Collum RP, and Wagner D. 2006. The LEAFY target LMI1 is a meristem identity regulator and acts together with LEAFY to regulate expression of CAULIFLOWER. Development 133: 1673–1682.

Schwarz-Sommer Z, Huijser P, Nacken W, Saedler H, and Sommer H. 1990. Genetic control of flower development by homeotic genes in Antirrhinum majus. Science 250, 931-936.

Theißen G, Becker A, Di Rosa A, Kanno A, Kim JT, Münster T, Winter K-U, and Saedler H. 2000. A short history of MADS-box genes in plants. Plant Molecular Biology 42, 115-149.

Weigel D, Alvarez J, Smyth DR, Yanofsky MF, and Meyerowitz EM. 1992. LEAFY controls floral meristem identity in Arabidopsis. Cell 69, 843–859.

Weigel D, Meyerowitz EM. 1993. Activation of floral homeotic genes in Arabidopsis. Science 261, 1723–1726.

William D, Su Y, Smith MR, Lu M, Baldwin D, and Wagner D. 2004. Genomic identification of direct target genes of LEAFY. Proccedings of the National Academy of Science U. S. A. 101, 1775–1780.

Winter, C.M., R. S. Austin, S. Blanvillain-Baufumé, M. a Reback, M. Monniaux, M.-F. Wu, Y. Sang, A. Yamaguchi, N. Yamaguchi, J. E. Parker, F. Parcy, S. T. Jensen, H. Li, and D. Wagner. 2011. LEAFY target genes reveal floral regulatory logic, cis motifs, and a link to biotic stimulus response. Developmental Cell 20, 430–443.

Ye L, Wang B, Zhang W, Shan H, and Kong H. 2016. Gains and losses of cis-regulatory elements led to divergence of the Arabidopsis APETALA1 and CAULIFLOWER duplicate genes in the time, space and level of expression and regulation of one paralog by the other. Plant Physiology 171, 1055-1069.


Posted in floral meristem, flowering | Tagged , , , , , , , , , , , , | Leave a comment

Is the cold gone? – Many roads to flowering after winter

Lars Hennig
Swedish University of Agricultural Sciences, Department of Plant Biology & Linnean Center for Plant Biology, PO-Box 7080, SE-75007 Uppsala, Sweden

Strolling through gardens and nature in March, one sees signs of spring everywhere. It is less the fresh green of young leaves but rather the colour of first flowers that attract the attention of the eye. Naturally – plants flower in spring. Although this sounds like a trivial statement, do we know how plants tell time? Admittedly, we know a lot: day length (photoperiod) is a mighty signal of spring and coming summer, often promoting floral activators such as the florigen FT. Rising temperatures are similarly effective to stimulate the earliest plants of the year to flower. Gardeners observed that climate change with milder early springs brings typical bulbous spring flowers to blossom earlier than they used to. Many plants, however, flower only later in the year and possibly not at all within their first year. Requiring weeks or even months of winter cold to permit subsequent flowering is common to vernalization-dependent plants (Chouard, 1960). Much research has focused on molecular mechanisms of vernalization, and in the model Brassicaceae Arabidopsis it is well established that the main responsibility to prevent or delay flowering before the winter lies with the repressor FLC, which keeps FT inactive (van Dijk, 2015). Winter cold in turn stably switches off expression of the repressor clearing the stage for flowering. Chromatin-based (epigenetic) mechanisms ensure that FLC remains switched off long after the cold of winter has passed. A protein complex that contains the Polycomb group proteins VRN2 converts chromatin at the FLC locus into a repressive local wall (Zografos et al., 2012).

Many molecular mechanisms and pathways are widely conserved in nature – FT, for instance, is an activator of flowering in all flowering plants. In contrast, vernalization pathways are not conserved. It is thought that as plant families spread from warmer into winter-cold regions, vernalization evolved multiple times independently. Many plants do not even carry an FLC gene. Not surprisingly, temperate grasses establish flowering after cold very differently to Arabidopsis. Cold regulates grass FT but homologs of Arabidopsis FLC or VRN2 are not involved (Fjellheim et al., 2014). Also sugar beet, where vernalization is of great economic relevance, has its unique way to detect winter. It uses a pair of FT paralogs differentially regulated by cold but as far as we know no FLC or VRN2 homologs (Pin et al., 2010). Recent work from the Putterill group now shows that the legume Medicago has yet another way to respond to winter (Jaudal et al., 2016).

In Arabidopsis, the main function of VRN2 is to repress FLC after winter. Like some other plants, Medicago has a VRN2 gene but no FLC. medicago detail_IMG_0002_1 crop But does MtVRN2 affect flowering? Yes it does! Jaudal and colleagues show that plants lacking MtVRN2 loose the requirement for vernalization. They flower early under inductive long days even without a cold treatment. This early flowering is accompanied by increased expression of the FT homolog FTa1. Loss of FTa1 abolishes the early flowering of Mtvrn2 mutants. Thus, the function of VRN2 differs between Arabidopsis and Medicago: Arabidopsis VRN2 represses a floral repressor after vernalization and vrn2 mutants flower late. Medicago VRN2 represses a floral activator before vernalization and Mtvrn2 mutants flower early. It is not known repression of FTa1 by MtVRN2 is direct or mediated by another protein. Initial chromatin immunoprecipitation experiments failed to strongly support a direct role (Jaudal et al., 2016). Given that FTa1 may be expressed only in a few cells, direct regulation of FTa1 by MtVRN2-mediated chromatin modifications remains, however, fully consistent with the reported data.

Together, vernalization responses have often evolved to repress FT homologs in the absence of vernalization but the molecular details differ greatly between species. It is difficult to extrapolate from current knowledge to novel species and further work in additional models is needed. Now that spring brings new energy to nature, researchers should become inspired and decipher more ways how to feel that the cold is gone.

Image courtesy of Joanna J Putterill

Chouard P. 1960. Vernalization and its relations to dormancy. Annual Review of Plant Physiology 11, 191-238.
Fjellheim S, Boden S, and Trevaskis B. 2014. The role of seasonal flowering responses in adaptation of grasses to temperate climates. Frontiers in Plant Sciences 5, 431.
Jaudal M, Zhang L, Che C, Hurley DG, Thomson G, Wen J, Mysore KS, and Putterill J. 2016. MtVRN2 is a Polycomb VRN2-like gene which represses the transition to flowering in the model legume Medicago truncatula. Plant Journal. DOI: 10.1111/tpj.13156
Pin PA, Benlloch R, Bonnet D, Wremerth-Weich E, Kraft T, Gielen, JJ, and Nilsson O. 2010. An antagonistic pair of FT homologs mediates the control of flowering time in sugar beet. Science 330, 1397-1400.
van Dijk A-J. 2015. Cold kick-start for flowering. Flowering Highlights.
Zografos BR and Sung S. 2012. Vernalization-mediated chromatin changes. Journal of Experimental Botany 63, 4343-4348.

Posted in flowering, FLOWERING LOCUS T, flowering time, vernalization | Tagged , , , , , , , | Leave a comment

More than the (human) eye can see: the regulation of floral UV absorbance

Beverley Glover

Department of Plant Sciences, University of Cambridge, Cambridge CB2 3EA, UK.

Flowers come in many different colours, and those colours cans be arranged in a wide variety of patterns. Colour itself arises from a combination of differential reflection of light of different wavelengths, combined with the ability of an animal to perceive that differential reflection. If light across the human visible spectrum (blue to red) is all reflected equally, we perceive the reflecting object as white (no colour). If light of a shorter wavelength, in the ultraviolet (UV) range, is differentially reflected then we will still perceive the object as white, but an animal with UV-sensitive photoreceptors in its eyes will perceive the object as coloured. Most insects and many other animals have UV-sensitive vision. I was once asked in a popular science talk whether it’s possible that there is a planet out there on which colour doesn’t exist, and everything is black or white. To which the answer is, there are many planets out there on which no organism has the ability to perceive differential light reflection, but I’m sure that on all planets on which light falls, there are materials which reflect different wavelengths to different extents. Whether that means those planets have colour, if nothing can see it, is a question for a philosopher to tackle!

Bev blog on UV light

Arctotis fastuosa photographed under visible (left) and UV (right) light. The human visible patterning resulting from differential anthocyanin pigmentation is very different from the insect visible patterning that results from UV absorbance by the inner ring of ray florets but UV reflectance by the outer ring.

The colour of flowers is widely considered to be part of the signalling between plants and animal pollinators, with the brightly coloured petals acting as part of a ‘sensory billboard’ (Raguso, 2004) to attract animals to the floral rewards (usually nectar or pollen). It is clear that different colours are visible to different animals, and so particular colours or colour patterns can be interpreted as part of an evolutionary trend in response to attracting particular groups of pollinators. The extent to which specific flower colours do attract specific pollinators is a source of great debate in the literature, with some authors arguing that, like a human-attracting billboard, any colour that stands out against the green vegetation will be as attractive as any other.
While there are clearly some very specific cases of colour (and scent) patterns combining to attract very specific pollinators, such as in the sexually deceptive orchids like Ophrys speculum , it is likely that most flower colours are signalling more broadly.

Ophrys speculum has a highly zygomorphic flower, with inner tepals and both whorls of reproductive organs affected

One good example of specific colours signalling to specific pollinators is found in the genus Petunia, where bee-pollinated flowers such as P. inflata are purple and reflect UV, moth-pollinated P. axillaris flowers are white and absorb UV, and hummingbird-pollinated P. exserta flowers are red and reflect UV. The evolutionary transitions between these different pollination syndromes must have involved switches in regulation of pigment production, to cause both human-visible and human-invisible colour changes. Previous studies have uncovered some of the mechanisms underpinning the human-visible shifts between purple and white flowers, which are the result of changes in the regulation of anthocyanin production. In Petunia this evolutionary transition is caused by changes to the activity of a MYB transcription factor which regulates the genes encoding the enzymes of anthocyanin biosynthesis (Hoballah et al., 2007; Quattrocchio et al., 1999). Interestingly, in various other phylogenetically unrelated examples of evolutionary transitions in anthocyanin content, MYB genes have also been implicated (eg. Yuan et al., 2013).

Recent work by Sheehan et al. (2016) has tackled the much more difficult problem of how the human-invisible UV absorbance has changed between different Petunia species. UV absorbance is the result of flavonol production – these compounds are produced by a biosynthetic pathway which branches off the main anthocyanin pathway using dihydroflavonols as substrate. Flavonols appear white to the human eye, but they absorb UV, and so to a moth with UV-sensitive vision flavonol-pigmented flowers will reflect longer wavelengths more strongly that shorter ones, and appear blue-green. Sheehan and colleagues first used accessions of P. axillaris differing in flavonol production to confirm that hawkmoths preferentially visit UV-absorbing flowers over UV-reflecting ones. They then used QTL mapping with an F2 population derived from a cross between P. exserta and P. axillaris to identify a major locus explaining 79% of the variation in flavonol levels between the 2 species. In combination with previous reports of a flavonol QTL on chromosome II when comparing P. integrifolia (UV reflecting) and P. axillaris (UV absorbing), these data led the authors to investigate the gene encoding flavonol synthase (FLS), the enzyme that catalyses the production of flavonols from dihydroflavonols. FLS had been previously mapped to chromosome II. They found that the FLS locus itself could not explain the differences between species – there were no differences in coding sequence, and although the gene was differentially expressed in different parent species, when alleles from different parents were combined in a single heterozygous plant the different alleles did not show differential expression. Sheehan and colleagues concluded that changes to a regulator of FLS was the likely explanation for the evolutionary transitions. They therefore used a transposon mutagenesis screen to search for candidates, crossing a UV-reflective line of P. hybrida with active transposons to UV-absorbing P. axillaris. The F1 generation is heterozygous for UV absorption, which is a dominant trait. Flowers were screened for UV reflecting sectors where the transposon had disrupted the target gene in the P. axillaris chromosome. This work was made much easier by the surprising realisation that UV reflecting sectors were also a deeper pink in the human-visible spectrum. Identification of the gene disrupted in these sectors revealed the key regulator to be another MYB transcription factor, named MYB-FL, which is linked to and regulates the expression of FLS. Following this molecular genetic detective work Sheehan and colleagues were able to trace the evolution of UV absorbance between their 3 Petunia species, finding that MYB-FL is not expressed in bee-pollinated P. inflata. Expression is activated in moth-pollinated P. axillaris, inducing UV absorbance, by mutation of the MYB-FL promoter region. This expression is maintained in hummingbird-pollinated P. exserta but UV absorbance is lost due to a single base pair deletion in the coding region, causing a frameshift and an early stop codon.
This dissection of the molecular basis for shifts in UV absorption is a powerful example of the use of molecular genetic tools to tackle evolutionary questions even outside the range of what is usually visible to humans. However, the interesting twist in the tale is the discovery that UV reflecting sectors were also brighter pink to the human eye. We cannot see flavonols, so the explanation proposed by Sheehan et al. is that the FLS enzyme is in competition with dihydroflavonol reductase (DFR) for dihydroflavonol substrates. As FLS activity reduced with the inactivation of MYB-FL, then so DFR could channel more dihydroflavonol into the anthocyanin pathway. This hypothesis remains to be tested, but it is certainly logical and, if proven correct, will generate a need to re-think our ideas about the independence of different pigment classes and the ability of different aspects of flower colour to evolve independently.


Hoballah M,  Gubitz T, Stuurman J, Broger L,  Barone M,  Mandel T, Dell’Olivo A, Arnold M, and Kuhlemeier C. 2007. Single gene-mediated shift in pollinator attraction in Petunia. Plant Cell 19, 779-790.

Quattrocchio F,  Wing J, van der Woude K, Souer E, de Vetten N, Mol J, and  Koes R. 1999. Molecular analysis of the anthocyanin2 gene of petunia and its role in the evolution of flower colour. Plant Cell 11, 1433–1444.

Raguso RA. 2004. Flowers as sensory billboards: progress towards an integrated understanding of floral advertisement. Current Opinion in Plant Biology 7, 434-440.

Sheehan H, Moser M, Klahre U, Esfeld K, Dell’Olivo A, Mandel T, Metzger S, Vandenbussche M, Freitas L, and Kuhlemeier C. 2016. MYB-FL controls gain and loss of floral UV absorbance, a key trait affecting pollinator preference and reproductive isolation. Nature Genetics 48, 159-166.

Yuan Y, Sagawa J, Young R, Christensen B, and Bradshaw HD Jr. 2013. Genetic dissection of a major anthocyanin QTL contributing to pollinator-mediated reproductive isolation between sister species of Mimulus. Genetics 194, 255-263.

Posted in flowering | Tagged , , , , , , , , , , | Leave a comment

A database for all things flowering

by Frank Wellmer
Smurfit Institute of Genetics, Trinity College Dublin, Ireland

Three decades of research on the genetic and molecular control of flowering has led to a staggering amount of data from a multitude of different experimental approaches (summarized for example in Fornara et al., 2010; Prunet and Jack, 2014). This work has been described in several thousand publications, and even noted experts in this area will hardly be in full command of the wealth of available information. What’s more, young researchers entering the field have an almost unscalable mountain of literature to climb when they want to familiarize themselves with what is already known and what still remains to be investigated. A key question is thus how this large amount of information can be unified, simplified, displayed and brought into a format that readily allows it to be used for approaches such as data mining, meta-analysis or even mathematical modeling. Arguably one of the biggest issues here is that the available data are often of markedly different quality: while some observations are being support by evidence from multiple independent experimental approaches generated by different research groups, others are based on the results of a single study (which, in some cases, might have been poorly designed). And there are examples where the available data are, at least in part, contradictory. Also, the advent of genomics approaches, while highly influential and impactful in plant biology, has led to the generation of datasets with inherent errors, but error rates are difficult to estimate and thus are unknown in most cases. The results from these approaches in particular depend on the experimental set-up used. For example, has the function of a given floral regulator been studied through transcriptomics after its constitutive and ectopic over-expression, or have more sophisticated transgenic lines been established for its analysis resulting in conditions that are closer to the situation found in the wild type?

The points outlined above (and probably many others as well) make the indexing, the processing and the representation of the available data a truly daunting challenge. Recently, Bouché and colleagues have taken on this herculean task and set up a free Flowering-Interactive Database, FLOR-ID (Bouché et al., 2016). While primarily containing information for gene networks involved in the control of flowering time, FLOR-ID also represents knowledge on early events during flower development (Fig. 1).

Fig. 1

Fig. 1. Screenshot of one of the interactive schemes available in FLOR-ID. Floral organ identity factors (in red) and their known regulatory functions in the different types of floral organs are shown. Created by Bouché.

In total, data from almost 1,600 publications have been used to assemble the database, which covers around 300 regulatory genes. Importantly, FLOR-ID is hand-curated, using a simple but effective search strategy (UniProt and PubMed database searches combined with information taken from reviews and primary research papers). The information in FLOR-ID can be accessed either by downloading tabulated data or through a user interface, which provides beautifully designed and interactive schemes for the different processes known to be involved in flowering-time control and early flower development (Fig. 1). Notably, the results from papers using -omics approaches have been largely omitted, especially if they are only predictive and not backed by results from independent experiments. Some may find this approach too exclusive, but the authors had the good sense to design FLOR-ID to allow user input and to suggest modifications, which, importantly, will be curated and only then incorporated into the active database. Thus, FLOR-ID could become a real community effort and will certainly be a major boon in the field of flowering for years to come.


Fornara F, de Montaigu A, Coupland G. 2010. SnapShot: Control of flowering in Arabidopsis. Cell 141(3):550, 550.e1-2.

Prunet N, Jack TP. 2014. Flower development in Arabidopsis: there is more to it than learning your ABCs. Methods in Molecular Biology 1110, 3-33.

Bouché F, Lobet G, Tocquin P, Perilleux C. 2016. FLOR-ID: an interactive database of flowering-time gene networks in Arabidopsis thaliana. Nucleic Acids Research 44, D1167-1171.


Posted in flowering, genetic resources | Tagged , , , , | Leave a comment

Distorted flower development and rainforest loss

How basic research in epigenetics helps to improve somatic embryogenesis technologies and contributes to protect rain forests

by Lars Hennig
Swedish University of Agricultural Sciences, Department of Plant Biology and Linnean Center for Plant Biology, Uppsala, Sweden

Not often is the relevance of basic research immediately obvious. Although history is full of examples of how human curiosity led to discoveries that changed the world, basic research finds itself very often in a defensive position. Recent work from the Martienssen lab (Ong-Abdullah et al., 2015) is a brilliant example how basic research findings from two fields – plant floral development and epigenetics – pave the way to improve crop yield and eventually alleviate the deforestation pressure on tropical rain forests. This work very clearly illustrates the great importance of knowledge about genes steering flower development and the mechanisms by which plants defend their genomes.
The paper by Ong-Abdullah and colleagues contributes to overcome one of the great disappointments of plant biotechnology – somaclonal variation. Some decades ago, in vitro regeneration and, in particular, somatic embryogenesis methods were developed to rapidly multiply elite genotypes. Such techniques can maintain favourable allele combinations, eg. in F1 hybrids, and are particularly valuable in perennial species with long juvenile phases. However, the great hopes that were connected to in vitro techniques were soon confronted by unwanted phenotypic variation in the in vitro offspring – so-called somaclonal variation.

One-year-old oil palm plant generated by somatic embryogenesis grown in vitro. (Image courtesy of Marcelinus Rocky Hatorangan, Uppsala, Sweden)

One-year-old oil palm plant generated by somatic embryogenesis grown in vitro. (Image courtesy of Marcelinus Rocky Hatorangan, Uppsala, Sweden)

The African oil palm (Elaeis guineensis) is one of the major sources for plant oil and estimates suggest that somatic embryogenesis may contribute to 20-30% yield increase in this species. This is highly relevant because rain forest is cleared to increase the land available for oil palm cultivation to meet the growing demand for plant oil. Unfortunately, oil palm suffers from severe somaclonal variation with many plants derived from somatic embryos developing abnormal flowers that result in great yield losses (Corley, 1986). These flowers have carpelloid stamens and staminodes suggesting reduced B-class floral homeotic gene function. The phenotype was called ‘mantled’ and has escaped molecular understanding for several decades. Like many somaclonal variation effects, the mantled phenotype occurs in a stochastic manner and can sometimes revert. The non-Mendelian genetics have suggested an epigenetic nature of the effect. Mantled oil palms arise stochastically after tissue culture but appear symptom-free during the long juvenile phase of several years before first flowering.
The work by Ong-Abdullah et al. (2015) shows that mantled plants are characterized by reduced DNA methylation on a transposable element (TE) related to the rice Karma element. In oil palm, the critical TE is located in a large intron of EgDEF1 – a homolog of Arabidopsis APETALA3 and snap dragon DEFICIENS, and a likely B-class floral homeotic gene. Loss of KARMA methylation affects EgDEF1 pre-RNA splicing and thereby creates an altered EgDEF1 protein that possibly could act in a dominant negative way to interfere with normal floral development. The mantled phenotype is fully consistent with reduced B-class function and the EgDEF1 gene has been studied as a candidate gene in mantled oil palms before (Adam et al., 2007; Jaligot et al., 2014).

When is Karma methylation in EgDEF1 lost? Tissue culture often reduces DNA methylation but most of it is restored during plant regeneration. In some plants, however, DNA methylation at Karma was lost completely and not restored. Because DNA methylation loss appears to be stochastic, lost Karma methylation does not correlate with other methylation changes and phenotypes. This has made it impossible so far to predict the flower morphology at a young age.

One-year-old oil palm plant generated from seed grown on soil. (Image courtesy of Marcelinus Rocky Hatorangan, Uppsala, Sweden)

One-year-old oil palm plant generated from seed grown on soil. (Image courtesy of Marcelinus Rocky Hatorangan, Uppsala, Sweden)

As a result of the work by Ong-Abdullah and colleagues, a diagnostic assay is now available to predict the flower phenotype at early age. Instead of growing affected individuals for many years until the flower defect is manifested by reduced yield, plants can now be screened at a young age. Land for oil palm cultivation can thus be used more efficiently, hopefully alleviating the pressure to increase the arable area.
Although the evidence is compelling and the diagnostic fulfils the practical need, the formal proof that the defective EgDEF1 splicing and the altered EgDEF1 protein cause the phenotype is still missing. Likewise the mechanism by which altered DNA methylation affects EgDEF1 splicing has not been identified. It will be interesting to learn whether other cases of somaclonal variation after tissue culture also alter splicing of genes for developmental regulators.
So far, EgDEF1 with Karma is the only well-documented locus connected to somaclonal variation but it is likely that locally reduced DNA methylation is involved in other cases, too. Karma was found using an association study design in which methylation states were profiled across the genome for populations of mantled and wild-type plants. The analysis landed right on EgDEF1, which was the only locus with DNA methylation consistently reduced in all mantled individuals. This approach can easily be used for other species and may in the future lead to the development of novel diagnostics, and reduce the detrimental effects of somaclonal variation in plant biotechnology. Finally, knowing the mechanisms behind DNA methylation homeostasis may allow developing cures for susceptible loci such as EgDEF1. As Ong-Abdullah and colleagues speculate in their paper, an engineered introduction of 24nt small RNAs targeted at the critical Karma region may re-establish DNA-methylation to a degree that mantled phenotypes do not develop and costly screening of plants is eliminated altogether.
Implications of research are impossible to foresee – they may be wider and more exciting than anticipated!

Corley, RHV (1986). Oil palm. In CRC Handbook of Fruit Set and Development. Monselise, S.P.Ed.; pp. 253-259. CRC Press: Boca Raton, Fl.
Adam H, Jouannic S, Orieux Y, Morcillo F, Richaud F, Duval Y and Tregear JW. (2007). Functional characterization of MADS box genes involved in the determination of oil palm flower structure. Journal of Experimental Botany 58: 1245-1259.
Jaligot E, Hooi WY, Debladis E, Richaud F, Beule T, Collin M, Agbessi MD, Sabot F, Garsmeur O, D’Hont A, Alwee SS and Rival A. (2014). DNA methylation and expression of the EgDEF1 gene and neighboring retrotransposons in mantled somaclonal variants of oil palm. PLoS ONE 9(3): e91896. doi:10.1371/journal.pone.0091896
Ong-Abdullah M, Ordway JM, Jiang N, Ooi SE, Kok SY, Sarpan N, Azimi N, Hashim AT, Ishak Z, Rosli SK, Malike FA, Bakar NA, Marjuni M, Abdullah N, Yaakub Z, Amiruddin MD, Nookiah R, Singh R and Low ET et al.. (2015). Loss of Karma transposon methylation underlies the mantled somaclonal variant of oil palm. Nature 525: 533-537.

Posted in flowering | Tagged , , , , , , , , , , , | Leave a comment

A tale of many variants

by Alt-Jan van Dijk

Plant Research International, Wageningen

In the quest to unravel genetic aspects of flowering time regulation, quantitative trait loci (QTLs) are important data. QTLs indicate genome regions in which variation influences a trait of interest. Typically, a relatively large region of the genome is identified, and for practical purposes in breeding it might not always be needed to zoom in to the nucleotide level. However, for a better understanding of the molecular mechanisms involved in flowering time regulation, and for example, its role in domestication, knowledge of causal genes or nucleotides is clearly required.

In a recent paper in Molecular Biology and Evolution, Liu et al. (2015) aimed to identify a causal gene underlying a flowering time QTL in sorghum. To get as close as possible to the causal variant underlying the QTL, fine-mapping was applied, using fine-spaced markers to delineate a small genome region within the original QTL. In the resulting 37-kb genome region, the authors observed that the gene HD1 (Heading Date 1) in one of the two parents of their population contains a frame-shift inducing deletion. HD1 is a very relevant gene, because homologs in rice and in Arabidopsis (CONSTANS, CO) are known to regulate flowering time. HD1/CO is a transcription factor, and transcription factors are known to be important players in domestication (see for example Meyer and Purugganan, 2013). The frameshift observed by Liu et al. lead to truncation of HD1. Such variant-of-large-effect is quite likely to lead to a dysfunctional protein. Although other types of functional changes are relevant during evolution, in particular cis-regulatory changes, mutations leading to null function appear to be the predominant type of causative change during domestication (Meyer and Purugganan, 2013).

In addition to analyzing the genetic variation in HD1 in the context of their original QTL mapping population, Liu et al. added two additional types of evolutionary comparisons (Fig. 1). The first additional analysis focused on a set of sorghum varieties. In this population, the same truncating variation in HD1 as in the QTL study was found in some individuals, and an alternative deletion in HD1 was also identified. As a second step, additional crops were analyzed, in particular, rice and foxtail millet. In both species, a syntenic region containing HD1 can be identified. This region had previously been mapped as flowering time QTL in both species. In rice, variants with a premature stop codon and with frameshifts are observed, which is somewhat reminiscent of the observed variation in HD1 in sorghum. In foxtail millet, however, the variation influences splicing leading to a deletion of 11 amino acids and again potential loss of HD1 function.

As a general note, the selection of HD1 as a putative causal gene for the QTL by Liu et al. was possible because of available knowledge on HD1/CO function. For many genes, such knowledge is not available – and in fact, genes with yet unknown functions arguably would be the most valuable novel findings. We recently published a method aimed at prioritizing candidate genes underlying QTLs (Bargsten et al., 2014), which integrates sequence and expression information in order to predict gene functions. Subsequently, these predicted gene functions are used to rank genes for their likelihood to be causal genes involved in the trait-of-interest. Using this approach on a rice QTL compendium, a set of 79 genes with so far unknown function was identified as most likely candidates underlying variation in rice flowering time (Bargsten et al., 2014).

The results presented by Liu et al. seem to indicate that variation in HD1 has been repeatedly selected for its effect on flowering time. In fact, an additional example is available: previously, QTL mapping suggested a variation in sorghum HD1 that consisted of a His to Tyr substitution (Yang et al., 2014). The same His to Tyr mutation inactivates the Arabidopsis HD1 homolog CO. Clearly, experimental proof of the hypothesized effects of the various genetic changes is needed. Nevertheless, the results presented by Liu et al. illustrate that for crop domestication, different roads lead to Rome: even though the genetic variation itself is not the same in the different species (deletion, splicing variation or coding change) in each case the effect presumably is that flowering time is affected.


 Bargsten JW, Nap JP, Sanchez-Perez GF, van Dijk ADJ. 2014. Prioritization of candidate genes in QTL regions based on associations between traits and biological processes. BMC Plant Biology 14: 330.

Liu H, Liu H, Zhou L, Zhang Z, Zhang X, Wang M, Li H, Lin Z. 2015. Parallel domestication of the Heading Date 1 gene in cereals. Molecular Biology & Evolution 32(10): 2726-37.

Meyer RS and Purugganan MD. 2013. Evolution of crop species: genetics of domestication and diversification. Nature Reviews Genetics 14: 840–852.

Yang S, Weers BD, Morishige DT and Mullet JE. 2014. CONSTANS is a photoperiod regulated activator of flowering in sorghum. BMC Plant Biology, 14: 148.

Posted in flowering, Uncategorized | Tagged , , , , , , , , | Leave a comment