Positive selection on the K domain of the AGAMOUS protein in the Zingiberales suggests a mechanism for the evolution of androecial morphology
© Almeida et al.; licensee BioMed Central. 2015
Received: 17 December 2014
Accepted: 20 February 2015
Published: 8 April 2015
The ABC model of flower development describes the molecular basis for specification of floral organ identity in model eudicots such as Arabidopsis and Antirrhinum. According to this model, expression of C-class genes is linked to stamen and gynoecium organ identity. The Zingiberales is an order of tropical monocots in which the evolution of floral morphology is characterized by a marked increase in petaloidy in the androecium. Petaloidy is a derived characteristic of the ginger families and seems to have arisen in the common ancestor of the ginger clade. We hypothesize that duplication of the C-class AGAMOUS (AG) gene followed by divergence of the duplicated AG copies during the diversification of the ginger clade lineages explains the evolution of petaloidy in the androecium. In order to address this hypothesis, we carried out phylogenetic analyses of the AG gene family across the Zingiberales and investigated patterns of gene expression within the androecium.
Phylogenetic analysis supports a scenario in which Zingiberales-specific AG genes have undergone at least one round of duplication. Gene duplication was immediately followed by divergence of the retained copies. In particular, we detect positive selection in the third alpha-helix of the K domain of Zingiberales AGAMOUS copy 1 (ZinAG-1). A single fixed amino acid change is observed in ZinAG-1 within the ginger clade when compared to the banana grade. Expression analyses of AG and APETALA1/FRUITFULL (AP1/FUL) in Musa basjoo is similar to A- and C-class gene expressions in the Arabidopsis thaliana model, while Costus spicatus exhibits simultaneous expression of AG and AP1/FUL in most floral organs. We propose that this novel expression pattern could be correlated with the evolution of androecial petaloidy within the Zingiberales.
Our results present an intricate story in which duplication of the AG lineage has lead to the retention of at least two diverged Zingiberales-specific copies, ZinAG-1 and Zingiberales AGAMOUS copy 2 (ZinAG-2). Positive selection on ZinAG-1 residues suggests a mechanism by which AG gene divergence may explain observed morphological changes in Zingiberales flowers. Expression data provides preliminary support for the proposed mechanism, although further studies are required to fully test this hypothesis.
KeywordsAGAMOUS Androecial morphogenesis Gene duplication K domain Petaloidy Positive selection Protein divergence Zingiberales
The genetic control of flower morphogenesis has long been studied in Arabidopsis thaliana and Antirrhinum majus . Classically, floral organ specification has been described by combinatorial patterns of gene expression in what is well known as the ABC model of floral organ identity. In this model, specific domains of expression of A-, B-, and C-class MADS-box genes correlate with the position of the developing sepals (A-class genes only), petals (a combination of A- and B-class genes), stamens (a combination of B- and C-class genes), and gynoecium (C-class genes only): thus, gene expression is correlated with organ identity. In A. thaliana, there are two A-class genes (APETALA-1 (AP1), and APETALA-2 (AP2)), two B-class genes (APETALA-3 (AP3), PISTILLATA (PI)), and one C-class gene (AGAMOUS (AG)). With the exception of AP2, all other genetic components of the ABC model are type-II MIKCc MADS-box genes as determined by their arrangement of protein domains. Their proper function as transcriptional regulators is dependent on the protein-protein interactions that occur between the A-, B-, and C-class genes, as well as with the SEPALLATA genes , to form protein dimers and functional tetramers. The protein-level explanation for A-, B-, and C-class functions is known as the quartet model, and asserts that only in tetramers are the A-, B-, and C-class proteins capable of regulating downstream genes (for review, ).
Although the ABC model has proven fruitful in describing organ identity and floral organ development, it lacks a mechanism to explain how such robust gene expression patterns are established during development and how changes in expression, function, or copy number may correlate with evolutionary changes in organ morphology. In order to address this mechanism, the A-, B-, and C-class genes have been integrated into an elegant complex-system approach, capable of explaining the robustness of the ABC gene expression patterns during flower development [4,5]. By mapping the landscape of known gene interactions during floral development in A. thaliana, Mendoza and coworkers (1999) were able to recover the stable states (that is, attractors) that correspond to the gene expression patterns correlated to floral organ identity, as described by the ABC model. In doing so, the authors proposed a set of necessary and sufficient genes and genetic interactions that provide a dynamical and mechanistic explanation for the establishment of the ABC gene expression patterns [6,7].
The AG gene was first isolated from A. thaliana over two decades ago , when fully penetrant mutations were shown to cause abnormalities in the development of the floral reproductive organs. AG has since been implicated in proper development of reproductive organ identity across flowering plants and is thought to play an additional role in ovule development and meristem determinacy in some lineages [10-12]. The evolution of the AG subfamily of transcription factors has been extensively studied across angiosperms. Phylogenetic analyses of the AG subfamily demonstrate that a duplication event occurring early during angiosperm diversification resulted in the origin of two major lineages: the AG and the AGAMOUS-like 11 (AGL11) lineages [13-16]. In the Arabidopsis canonical model, the AG and AGL11 lineages correspond functionally to C-class and D-class homeotic genes, respectively, in which C-class homeotic genes are involved in stamen and carpel identity and in floral meristem determinacy and D-class genes are more specifically involved in ovule and fruit development [14,12]. The AG lineage itself has undergone subsequent gene duplications, and parsing of function of duplicated copies is thought to have occurred independently in the major angiosperm lineages. Within the core eudicots, for example, the AG lineage is divided into the euAG and the PLENA (PLE) lineages [14,15] with various subclades within both lineages demonstrating neofunctionalization, subfunctionalization, redundancy, and loss of duplicated copies . In Antirrhinum, the AG lineage genes PLE and FARINELLI (FAR) contribute unequally to specify male and female reproductive organs [18,19], while in Petunia FBP6 and PMADS3 act redundantly as C-function genes . In Zea mays, however, the AG paralogs ZAG1 and ZMM2 appear to be expressed in spatially distinct domains of the developing flower, and may have subfunctionalized into carpel- and stamen-specific paralogs , while in Oryza sativa, AG paralogs OSMADS3 and OSMADS58 are essential for reproductive organ identity and together with AG11 lineage OSMADS13 are important for floral meristem determinacy [10,12,13].
In addition to its role in reproductive organ identity, differential expression of AG in A. thaliana has been shown to be involved in the development of petaloidy in the androecium. For instance, the ag-11 allele, bearing a single point mutation in the regulatory region of Arabidopsis AG, results in the transformation of stamens into petaloid organs . Also, downregulation of AG by anti-sense RNA in A. thaliana leads to a variety of aberrant floral morphologies, including petaloid stamens . Most recently, Tanaka et al. ) demonstrated the role of AG in regulating proper petal and stamen differentiation in cyclamen (Cyclamen persicum), where two paralogs (CpAG1 and CpAG2) are present and repression of CpAG2 leads to the formation of infertile and petaloid structures in the stamen whorl .
The molecular mechanisms underlying the evolution of androecial petaloidy in the angiosperms have not been studied in detail. It is likely that developmental processes underlying androecial petaloidy are homoplasious across flowering plants, as petal-like stamens have evolved independently in a variety of angiosperm lineages . In male flowers of the early diverging Amborella trichopoda, stamen filaments are expanded into petal-like structures. In several other basal angiosperm and magnoliid lineages, flowers display a gradual transition between petal and stamen organs, with multiple degrees of androecial petaloidy present even within the same flower (for example, Nymphaea alba) : A gradient of AG and B-class gene expression has been implicated in this gradual morphological transition between laminar petals and filamentous stamens .
Given the involvement of the AG gene lineage in reproductive organ development and its interconnectivity within the floral organ specification gene regulatory network (FOS-GRN), we hypothesize that gene duplication followed by potential functional divergence of the AG lineage in the ginger clade is correlated with the evolution of petaloidy in the androecium. In order to test our hypothesis, AG lineage genes were amplified from across the Zingiberales, and their expression was assessed during flower development in representative species. Tests of selection were carried out to investigate the role of selection on gene evolution and function. Our results suggest that positive selection has played a role in the evolution of AG across the Zingiberales order, particularly in protein divergence within the K domain. These protein modifications, together with comparative analyses of AG and AP1 expression across the order, suggest a mechanism by which androecial petaloidy may have evolved in the Zingiberales, and support the hypothesis that modifications in AG expression and function are correlated with androecial petaloidy.
Plant material, RNA extraction, and cDNA synthesis
List of species and accession numbers used in this study
NM_001111851.1; NM_001111456.1; X80206.1; X81199.1
L37258.1; NM_001061424.1; AF151693.1
Hordeum vulgare subsp. Vulgare
AY727648.1; AY727647.1; AY727624.1; NM_001203767.1
Amplification of AGAMOUS genes in the Zingiberales
A multiple sequence alignment (MSA) for the AG gene lineage was generated from sequences downloaded from NCBI (Table 1). The MSA was used to design general primers for amplification of Zingiberales AG genes. Multiple primer combinations, with different degrees of degeneracy, were used in order to improve chances of assessing all copies of the AG gene lineage within the Zingiberales. Primer sequences were as follows: (i) forward primers: 5′ ACI AAY MGI CAR GTI ACI TTY TG 3′; 5′ ATG GSI MGI GGI AAR ATI SAR AT 3′; 5′ CAR GTK ACC TTC TGC AAG 3′; 5′ ATC CCA TGG AGC ATA AAG CA 3′; 5′ GRG GRA AGA TCG AGA TCA AG 3′; (ii) reverse primers: 5′ ACC CTA TCA GTC TCG GCG ATC TTG TTC C 3′; 5′ TCA TCG TTC AAC CAA AGT GG 3′; 5′ TTG MAK RAA GTT CCY TGA RTM RT 3′.
PCR reactions were carried out using Phire Hot Start II DNA Polymerase (Thermo Scientific) and 2 μl of 5X Phire buffer; 0.2 μl 10 mM dNTPs; 0.5 μl of each primer; 0.1 μl Phire Polymerase; 1 μl [1:10] cDNA; and ddH2O, for a total volume of 10 μl. Thermocycling conditions followed manufacturer’s recommendations. PCR products were visualized on a 1% agarose gel stained with GelRed™ (Phoenix Research Products) according to the manufacturer’s protocol. PCR products were cloned into Top10 cells and sequenced using Big Dye Terminator Kit v3.1 (Applied Biosystems) on a 3700 sequencer. At least 16 clones were sequenced for each of the species sampled. Over 40 clones were sequenced for Costus spicatus, in order to insure deep sampling of gene copy number in this species.
A multiple sequence alignment was generated using MacClade (4.06 OS X) with all generated Zingiberales sequences aligned to outgroup sequences downloaded from NCBI (Table 1). Model selection for the final alignment was tested in jModeltest 0.1.1  using the Bayesian information criterion (BIC). PartitionFinder  was used to test for the best partition scheme for the dataset and substitution model, also based on the BIC criterion. According to PartitionFinder as well as jModelTest, the best-fit model was K80 + G and no data partition was advised. The best-fit model was implemented in MrBayes 3.2  and PhyML  in order to assess gene tree topology. MrBayes runs were implemented on the CIPRES Science Gateway (www.phylo.org) under the model specified above, as well as under variations of the best-fit model to ensure that topology was not influenced by model selection and ran for 1.5 M generations. Data were further analyzed for convergence with Tracer v1.5 (http://beast.bio.ed.ac.uk/). SumTrees v.3.0.0 using the DendroPy Phylogenetic Computing Library v3.7.1  was used to calculate the burn-in and remove the appropriate trees saved prior to stationarity, and to assemble a 50% majority rule tree from the remaining trees. Maximum likelihood (ML) analyses were performed using PhyML implemented on the ATGC South of France bioinformatics platform (http://atgc.lirmm.fr/phyml/). Bootstrap support from 100 replicates and posterior probabilities (PP) were calculated for ML and MrBayes analyses, respectively, and are used as branch support in the gene tree (bootstrap/PP). In order to test the likelihood of different evolutionary scenarios for the AG gene tree, a Shimodaira-Hasegawa (SH) test  was performed by manually generating a constrained gene tree in which the two first branching lineages (all “ZinAG-1”) were forced to form a monophyletic group. The likelihood of the constrained tree was tested against the unconstrained gene tree obtained in our phylogenetic analysis using the likelihood ratio test (LRT). Likelihood tests for constrained topologies were obtained in PAUP* .
Both branch- and site-specific selection tests were performed in order to assess signals of selection across the AG subfamily, along branches leading to the major clades and at specific amino acid sites. Branch selection was assessed using PAML codeml branch models by setting the model = 2 in order to allow several omega (w) ratios compared to a fixed w, while site selection was evaluated using the site models M1a and M2a . Site selection was also assessed using the fixed-effect likelihood (FEL) model in HyPhy 2.0  under a stringent cut-off of 0.1, as suggested by the HyPhy program. A Bayesian 50% majority rule consensus tree was used for all selection analyses. For each node tested, a two-rate analysis was used to allow adjustment of the ratio of non-synonymous (dN) and synonymous (dS) substitutions across sites, and models determined by jModeltest were specified for the nucleotide model of evolution.
Gene expression of AGAMOUS and APETALA1 in Zingiberales developing flowers
RT-PCRs for ZinAG-1 and ZinAG-2 were carried out in all Musa basjoo floral organs. Primers were designed on intron-exon boundaries whenever possible, and sequences are as follows: MbAGcp1 forward 5′ TTG AAA GGT ATA AGA AAG CAT 3′; MbAGcp1 reverse 5′ TTA TTC TCG AGT TGC TTC ATG TCT 3′; MbAGcp2 forward 5′ TCG AGA GGT GGT ACA AGA AAG CAT GT 3′; MbAGcp2 reverse 5′ CGA GTC TCA AGC TGC TTC AG 3′. Reactions were carried out using Phire Hot Start II DNA Polymerase (Finnzymes, Finland) and the following protocol: 2 μl of 5X PHIRE buffer; 0.5 μl of each of the 10 mM primers; 0.1 μl of PHIRE polymerase; 0.2 μl of 50 mM dNTPs; 1 μl of [1:10] dilution of organ-specific cDNA; and water up to a 10 μl total volume, for 25 cycles. PCR reactions were performed on BioRad Thermocyclers and were visualized on 1% agarose gel, post-stained with Gel-Red™ (Biotium).
Expression of AGAMOUS and APETALA1 in C. spicatus and M. basjoo was assessed by generating organ-specific transcriptomes. cDNA libraries for sequencing on the Illumina platform were prepared using the TruSeq RNA sample prep kit v2. Two cDNA libraries each were prepared with 2.0 μg of RNA extracted from dissected tissue of the filament, theca, and free petal of M. basjoo and the petaloid filament, theca, petal, and labellum (fused petaloid staminodes) of C. spicatus. Libraries were multiplexed using barcoding set A. Samples were run on a HiSeq2000 at IIGB HT Sequencing Facility at the University of California, Riverside. Raw reads were trimmed to remove adapters and regions of poor quality with cutadapt . Costus spicatus sequences were assembled into a reference transcriptome using Trinity  with minimum contig length of 300. All other parameters were used according to Trinity default settings. GMAP/GSNAP  was used to align M. basjoo trimmed reads to annotated CDS from the published Musa acuminata genome, while C. spicatus trimmed reads were aligned to the C. spicatus reference transcriptome. Expression of AG and AP1-like genes was estimated using eXpress  in units of FPKM (frequency per kilobase of exon per million aligned reads). Replicates were independently processed, and gene expression was compared between libraries for consistency. ACTIN1 expression was used to normalize targeted gene expression across transcriptome libraries. Error bars were calculated based on the standard deviation (SD) of the two normalized samples for each organ.
In order to confirm transcriptome expression data for C. spicatus AG and AP1-like genes, quantitative PCR was performed on each floral organ of C. spicatus flowers (sepals, petals, labellum, stamen, and gynoecium). A general C. spicatus AP1/FUL-like primer was designed, as at this point, we were not able to produce copy-specific primer pairs. qPCR was performed on C. spicatus floral organs using iQ™ SYBR® Green Supermix (Life Science Research), and following the manufacturer’s protocol. Clade-specific qPCR primers were designed based on the nucleotide differences between C. spicatus sequences from different clades on the AGAMOUS gene tree, and primer sequences are as follows: CsAGcp1 forward 5′ AAC AGC AGT GTG AGA GCG ACT 3′; CsAGcp1 reverse 5′ GGT CTC TAA GGC TCA TAG AAC CGA GA 3′; CsAGcp2 forward 5′ CCA ACA GTG TGA GAG CAA CAA 3′; CsAGcp2 reverse 5′ CTT CAT ATC GCG TAG GCT CA 3′; CsAP1/FUL-like forward 5′ ATA TCA GGT CAA GAA AGA ACC AAA TC 3′; CsAP1/FUL-like reverse 5′ GGG CTT GTT TGG ATT CGT T 3′. Two reactions were performed, one for ZinAG-1 and ZinAG-2, and another for AP1/FUL and ZinAG-1 with annealing temperatures of 60°C and 58°C, respectively. In both cases, three replicates per gene were performed and ACTIN1 was used as an internal control.
Amplification and phylogenetic analyses
AG sequences were obtained for all families within the order, with the exception of Lowiaceae (Orchidantha), a monogeneric lineage. A multiple sequence alignment (MSA) of 558 bp was generated and encompasses all protein domains, with the exception of the first nine codons of the MADS domain and the end of the C-terminal domain, for which alignment to outgroup sequences became increasingly challenging. The final MSA comprises a total of 37 ingroup and 13 outgroup sequences. This MSA was used as the input to jModelTest, and determined the best-fit model as the K80 + G model. The best-fit model, as well as other more parameterized models (GTR, GTR + I, and GTR + I + G), was implemented in both MrBayes and PhyML.
The SH test  was performed using a constrained gene tree in which the two first branching lineages (all “ZinAG-1”) were forced to form a monophyletic group (Figure 4B), and compared for likelihood score against the unconstrained gene tree obtained in our analyses (Figure 3). Our results indicate that the constrained gene tree has a likelihood score that is not significantly different from the unconstrained analysis, suggesting that scenario 1 (Figure 4A) is equally as likely as scenario 2 in describing the evolutionary history for ZinAG. The most parsimonious explanation, then, is described by scenario 1, in which a single duplication event happened at the base of the Zingiberales order prior to lineage diversification.
In the organ-specific transcriptome data, C. spicatus AG expression is dominated by ZinAG-1, and extremely low levels of ZinAG-2 are only observed in the petal (Figure 8A), although ZinAG-2 can be amplified by RT-PCR in these organs (data not shown). In M. basjoo, AP1/FUL-like expression largely agrees with that anticipated based on a hypothesis of mutual exclusion : APETALA1/FRUITFULL-like (AP1/FUL-like) genes are mostly expressed in petals where there is very low expression of AG, while in stamens, AP1/FUL-like expression is almost abolished and AG is highly expressed (Figure 8B). In C. spicatus, AP1/FUL-like and AG gene expressions show a different pattern to that observed in Musa. Although AP1/FUL-like show very low expression values across Costus floral organs in comparison to AG expression, AG and AP1/FUL-like are simultaneously expressed in the androecial organs (labellum, stamen filament, and stamen theca), suggesting that ZinAG-1 is not capable of fully suppressing AP1/FUL-like expression in these organs.
The AG gene subfamily has been extensively implicated in the development of reproductive organs (carpels and stamens) and meristem determinacy in angiosperms. In both monocots and eudicots, the conservation of these functions by AG lineage genes is remarkable considering multiple gene duplication and subfunctionalization events [12,16,21], even though AGL11 lineage genes might act redundantly in some lineages [10,20]. In the Zingiberales, at least one lineage-specific duplication event is observed within the AG lineage. Sequence divergence between the two copies (ZinAG-1 and ZinAG-2), as well as their expression patterns, suggests the involvement of Zingiberales AG genes in the evolution of reproductive organ development and the evolution of petaloidy in the order.
Based on the branch selection patterns observed in the Zingiberales AG gene tree, it is likely that functional divergence between lineages happened early after the duplication event, at the base of the ginger clade. As expected for functionally important and highly interconnected genes, most of the observed site selection is due to balancing selection, suggesting functional conservation. However, three residues in the Zingiberales show signs of positive selection, and fixed differences among members of the ginger clade indicate that these modifications might be implicated in the morphological changes observed in the androecium of the Zingiberales.
In particular, the positive selected amino acid change observed at position 142 of the K domain is of particular relevance. The role of subdomains of the K domain in MADS-box protein-protein interactions has been studied, especially in the formation of dimers between B-class genes and SEP genes [47,48]. The K domain of the MADS proteins are involved in the formation of protein complexes for DNA binding. In particular, K1 and K2 helices are involved in dimer formation, while K3 is involved in the formation of tetramers [48-51]. Also, in Antirrhinum, a single amino acid change has been implicated in differences in the establishment of male and female identity between AG lineage genes PLE and FAR. A single glutamine insert in the K3 domain of FAR leads to a limited protein-protein interaction between AG and SEPALLATA (SEP) proteins, underlying the functional differences observed between FAR and PLE genes in determining reproductive organ identity .
In Zingiberales, it is possible that the amino acid change at the K3 domain observed in ZinAG-1 between the banana grade and the ginger group might change AG protein ability to form higher level complexes while maintaining the capacity to form protein dimers. This suggests an interesting mechanism in which ZinAG-1 from the ginger group could act as a negative regulator of tetramer formation: while binding to AG interacting proteins to form dimers, this complex would be less likely involved in the formation of quartets, resulting in a post-transcriptional downregulation of AG downstream targets.
If one assumes that ZinAG-1 in the ginger clade (exemplified by Costus) inhibits quartet formation, and thus its expression leads to the downregulation of downstream targets in C. spicatus (as suggested by the amino acid change; Figure 6), we expect that high levels of Costus ZinAG-1 would lead to a stronger suppression of downstream genes, and a more petal-like phenotype in the stamen whorl. The correlation between higher levels of ZinAG-1 in Costus labellum and filament and a petaloid phenotype of these organs is consistent with increased levels of ZinAG-1 in the labellum and filament and decreasing levels of ZinAG-1 expression towards the fertile theca.
This interpretation is also supported by changes in the expression profile of AP1/FUL-like genes across floral organs of the Zingiberales. In A. thaliana, relatively high levels of AP1/FUL were detected in petaloid stamens and sepaloid carpels of flowers with reduced levels of AG due to anti-sense (RNAi) knockdown . Accordingly, petaloid organs in the androecium such as those observed in C. spicatus are characterized by simultaneous expression of AG and AP1/FUL-like, indicating a lack of negative interaction between these two gene families. Morphologically, this expression profile corresponds to a ‘hybrid’ organ (petaloid staminode) and could potentially represent a ‘mix-attractor’ between stamen and petal in the A. thaliana FOS-GRN (, modified in Figure 1). However, this ‘hybrid’ attractor has not been observed as a stable state of the A. thaliana FOS-GRN, potentially due to the presumed fixed mutual negative regulation between AG and AP1/FUL. It is possible that duplication of many of the FOS-GRN genes observed in the Zingiberales could lead to stable states that are not observed in Arabidopsis, as different lineage-specific duplication events and subsequent differential retention/loss of duplicated copies as well as sequence divergence would provide the opportunity for novel protein interactions leading to novel stable states.
Such novel interactions are suggested by data from other monocot lineages, where an expanded AP1/FUL-like expression pattern has been observed in various grass lineages [52,53]. Interestingly, in Z. mays, constitutive expression of one of the AP1/FUL-like copies (ZmFUL2a) leads to the development of undifferentiated floral organs in the male spikelet. The authors propose an ‘interference hypothesis’ where interference of AP1/FUL-like proteins in the formation of proper protein-protein interactions during particular stages of development could result in the observed phenotypes . Although the precise function of A. thaliana AP1/FUL gene might be specific to Arabidopsis and closely related species, studies in grasses support the idea that AP1/FUL-like genes do play a role in transition to flowering, meristem and perianth identity, or even in determining the identity of all floral organs ( and references therein).
It is important to note that the results on AP1/FUL-like genes presented here are preliminary. Although there is an unexpected expansion of AP1/FUL-like gene expression towards the inner flower whorls, a more in-depth analysis of this gene family within the Zingiberales, as well as a comprehensive survey of the expression patterns of AG downstream genes, is required to fully test our hypothesized scenario. Also, protein-protein interaction studies are critical to test the functions of the described AG protein modifications observed across the Zingiberales.
In transgenic Arabidopsis plants carrying AG anti-sense RNA, a range of floral organ phenotypes is observed including the occurrence of petaloid stamens . Likewise, mutations in the regulatory site of AG in Arabidopsis can lead to the development of petaloidy in the androecium . Here, we show that androecial petaloidy in the Zingiberales is likewise associated with evolution of the AG lineage, and may result from a single amino acid change in the K domain of ZinAG-1 after the divergence of the banana lineages and the ginger clade.
The results presented here suggest a scenario in which positive selection acting upon AG genes in the Zingiberales has resulted in a fixed change in the K3 domain that can potentially explain the evolution of androecial petaloidy and infertility observed in the order. Selected amino acid changes in the K3 domain might result in differential abilities to form higher level protein-protein complexes between ZinAG-1 and its interaction partners, resulting in a post-translational downregulation of downstream genes. While further studies are needed to fully test this hypothesis, our expression data are consistent with this model. If it is the case that the changes in the AG genes are responsible for the observed changes in floral morphology across Zingiberales, a clear trade-off between production of fertile stamens and increased petaloidy has been fixed by positive selection in this group. Although androecial petaloidy is a remarkable feature of Zingiberales floral evolution, no changes have been observed in meristem determinacy. This might be explained, at least in part, by the potential functional redundancy between AG and AGL11 lineage genes, as already reported for in rice and petunia [10,20]. Further studies of the AG subfamily genes in the Zingiberales will help understand the complete role of the AG subfamily in floral development and evolution across the Zingiberales.
- AG :
- AGL11 :
- AP1/FUL :
- AP2 :
- AP3 :
- CsAGcp :
Costus spicatus AGAMOUS copy 1
- CsAGcp2 :
Costus spicatus AGAMOUS copy 2
- CsAP1/FUL-like :
Costus spicatus APETALA1/FRUITFULL-like gene
- CpAG1 :
Cyclamen persicum AGAMOUS copy 1
- CpAG2 :
Cyclamen persicum AGAMOUS copy 2
ratio of non-synonymous substitution rates
ratio of synonymous substitution rates
- euAG :
- FAR :
- FBP6 :
Floral binding protein gene 6
fixed effect likelihood model
floral organ specification gene regulatory network
frequency per kilobase of exon per million aligned reads
- K domain:
- MbAGcp1 :
Musa basjoo AGAMOUS copy 1
- MbAGcp2 :
Musa basjoo AGAMOUS copy 2
multiple sequence alignment
- OSMADS3 :
Oryza sativa MADS-box gene 3
- OSMADS13 :
Oryza sativa MADS-box gene 13
- OSMADS58 :
Oryza sativa MADS-box gene 58
- PI :
- PLE :
- PMADS3 :
Petunia MADS-box gene 3
posterior probability support
- SE :
- ZAG1 :
Zea mays AGAMOUS 1
- ZinAG-1 :
Zingiberales AGAMOUS copy 1
- ZinAG-2 :
Zingiberales AGAMOUS copy 2
- ZMM2 :
Zea mays MADS-box gene 2
Research contributing to this manuscript was supported by NSF Doctoral Dissertation Improvement Grant (DEB 1110461; AMRA and CDS), an NSF CAREER award to CDS (IOS 0845641), and research support from UC Berkeley Committee on Research (COR) and AES. AMRA was supported by a CAPES/Fulbright Fellowship. The authors thank M. E. Maldaner, A. Brown, and C. Sangster for their help in gathering Zingiberales AGAMOUS sequences, and E. Alvarez-Buylla for insightful discussion during the preparation of this manuscript. The authors thank H. Forbes from the UC Botanical Garden for support with fresh plant material. Publication made possible in part by support from the Berkeley Research Impact Initative (BRII) sponsored by the UC Berkeley Library.
- Coen ES, Meyerowitz EM. The war of the whorls: genetic interactions controlling flower development. Nature. 1991;353:31–7.View ArticlePubMedGoogle Scholar
- Pelaz S, Ditta GS, Baumann E, Wisman E, Yanofsky MF. B and C floral organ identity functions require SEPALLATA MADS-box genes. Nature. 2000;405:200–3.View ArticlePubMedGoogle Scholar
- Melzer R, Theissen G. Reconstitution of ‘floral quartets’ in vitro involving class B and class E floral homeotic proteins. Nucleic Acids Res. 2009;37:2723–36.View ArticlePubMed CentralPubMedGoogle Scholar
- Mendoza L, Alvarez-Buylla E. Dynamics of the genetic regulatory network for Arabidopsis flower morphogenesis. J Theor Biol. 1998;193:307–19.View ArticlePubMedGoogle Scholar
- Mendoza L, Thieffry D, Alvarez-Buylla E. Genetic control of flower morphogenesis in Arabidopsis thaliana: a logical analysis. Bioinformatics. 1999;15:593–606.View ArticlePubMedGoogle Scholar
- Alvarez-Buylla ER, Benítez M, Corvera-Poiré A, Chaos Cador A, de Folter S, Gamboa de Buen A, et al. Flower development. In: American Society of Plant Biologists, editor. The Arabidopsis book. 2010. doi: 10.1199/tab.0127.Google Scholar
- Álvarez-Buylla ER, Azpeitia E, Barrio R, Benítez M, Padilla-Longoria P. From ABC genes to regulatory networks, epigenetic landscapes and flower morphogenesis: making biological sense of theoretical approaches. Semin Cell Dev Biol. 2010;21:108–17.View ArticlePubMedGoogle Scholar
- Barrio RÁ, Hernández-Machado A, Varea C, Romero-Arias JR, Álvarez-Buylla ER. Flower development as an interplay between dynamical physical fields and genetic networks. PLoS One. 2010;5:e13523.View ArticlePubMed CentralPubMedGoogle Scholar
- Yanofsky MF, Ma H, Bowman JL, Drews GN, Feldman KA, Meyerowitz EM. The protein encoded by the Arabidopsis homeotic gene agamous resembles transcription factors. Nature. 1990;346:35–9.View ArticlePubMedGoogle Scholar
- Dreni L, Pilatone A, Yun D, Erreni S, Pajoro A, Caporali I, et al. Functional analysis of all AGAMOUS subfamily members in rice reveals their roles in reproductive organ identity determination and meristem determinacy. Plant Cell. 2011;23:2850–63.View ArticlePubMed CentralPubMedGoogle Scholar
- Fourquin C, Ferrándiz C. Functional analyses of AGAMOUS subfamily members in Nicotiana benthamiana clarify the evolution of early and late roles of C-function genes in eudicots. Plant J. 2012;71:990–1001.View ArticlePubMedGoogle Scholar
- Yamaguchi T, Lee DY, Miyao A, Hirochika H, An G, Hirano HY. Functional diversification of the two C-class MADS box genes OSMADS3 and OSMADS58 in Oryza sativa. Plant Cell. 2006;18:15–28.View ArticlePubMed CentralPubMedGoogle Scholar
- Dreni L, Osnato M, Kater MM. The ins and outs of the rice AGAMOUS subfamily. Mol Plant. 2013;6:650–64.View ArticlePubMedGoogle Scholar
- Dreni L, Kater MM. MADS reloaded: evolution of AGAMOUS subfamily genes. New Phytol. 2014;201:717–32.View ArticlePubMedGoogle Scholar
- Kramer EM, Jaramillo MA, Di Stilio VS. Patterns of gene duplication and functional evolution during the diversification of the AGAMOUS subfamily of MADS box genes in Angiosperms. Genetics. 2004;166:1011–23.View ArticlePubMed CentralPubMedGoogle Scholar
- Zahn LM, Leebens-Mack JH, Arrington JM, Hu Y, Landherr LL, DePamphilis CW, et al. Conservation and divergence in the AGAMOUS subfamily of MADS-box genes: evidence of independent sub- and neofunctionalization events. Evol Dev. 2006;8:30–45.View ArticlePubMedGoogle Scholar
- Airoldi CA, Davies B. Gene duplication and the evolution of plant MADS-box transcription factors. J Genet Genomics. 2012;13:157–65.View ArticleGoogle Scholar
- Airoldi CA, Bergonzi S, Davies B. Single amino acid change alters the ability to specify male or female organ identity. Proc Natl Acad Sci U S A. 2010;107:18898–902.View ArticlePubMed CentralPubMedGoogle Scholar
- Davies B, Motte P, Keck E, Saedler H, Sommer H, Schwarz-Sommer Z. PLENA and FARINELLI: redundancy and regulatory interactions between two Antirrhinum MADS-box factors controlling flower development. EMBO J. 1999;18:4023–34.View ArticlePubMed CentralPubMedGoogle Scholar
- Heijmans K, Ament K, Rijpkema AS, Zethof J, Wolters-Arts M, Gerats T, et al. Redefining C and D in petunia ABC. Plant Cell. 2012;24:2305–17.View ArticlePubMed CentralPubMedGoogle Scholar
- Mena M, Ambrose BA, Meeley RB, Briggs SP, Yanofsky MF, Schmidt RJ. Diversification of C-function activity in maize flower development. Science. 1996;274:1537–40.View ArticlePubMedGoogle Scholar
- Hong RL, Hamaguchi L, Busch MA, Weigel D. Regulatory elements of the floral homeotic gene AGAMOUS identified by phylogenetic footprinting and shadowing. Plant Cell. 2003;15:1296–309.View ArticlePubMed CentralPubMedGoogle Scholar
- Mizukami Y, Ma H. Separation of AG function in floral meristem determinacy from that in reproductive organ identity by expressing antisense AG RNA. Plant Mol Biol. 1995;28:767–84.View ArticlePubMedGoogle Scholar
- Tanaka Y, Oshima Y, Yamamura T, Sugiyama M, Mitsuda N, Ohtsubo N, et al. Multi-petal cyclamen flowers produced by AGAMOUS chimeric repressor expression. Scientific Rep. 2013; 3: doi: 10.1038/srep02641
- Walker-Larsen J, Harder LD. The evolution of staminodes in angiosperms: patterns of stamen reduction, loss, and functional re-invention. Am J Bot. 2000;87:1367–84.View ArticlePubMedGoogle Scholar
- Volkova PA, Choob VV, Shipunov AB. The flower organ transition in water lily (Nymphaea alba s.l., Nymphaeaceae) under cross-examination with different morphological approaches. Belg J Bot. 2007;140:60–72.Google Scholar
- Soltis D, Chanderbali A, Kim S, Buzgo M, Soltis PS. The ABC model and its applicability to basal angiosperms. Ann Bot. 2007;100:155–63.View ArticlePubMed CentralPubMedGoogle Scholar
- Kress WJ, Prince LM, Hahn WJ, Zimmer EA. Unraveling the evolutionary radiation of the families of the Zingiberales using morphological and molecular evidence. Syst Biol. 2001;50:926–44.View ArticlePubMedGoogle Scholar
- Almeida AMR, Brown A, Specht CD. Tracking the development of the petaloid fertile stamen in Canna indica: insights into the origin of androecial petaloidy in the Zingiberales. AoB PLANTS. 2013;5:plt009.View ArticlePubMed CentralPubMedGoogle Scholar
- Yockteng R, Almeida AMR, Yee S, Andre T, Hill C, Specht CD. A method for extracting high-quality RNA from diverse plants for next-generation sequencing and gene expression analyses. Appl Plant Sci. 2013;1:1300070.Google Scholar
- Bartlett ME, Specht CD. Evidence for the involvement of GLOBOSA-like gene duplications and expression divergence in the evolution of floral morphology in the Zingiberales. New Phytol. 2010;187:521–41.View ArticlePubMedGoogle Scholar
- Posada D. jModelTest: phylogenetic model averaging. Mol Biol Evol. 2008;25:1253–6.View ArticlePubMedGoogle Scholar
- Lanfear R, Calcott B, Ho SYW, Guindon S. Partitionfinder: combined selection of partitioning schemes and substitution models for phylogenetic analyses. Mol Biol Evol. 2012;29:1695–701.View ArticlePubMedGoogle Scholar
- Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Hohna S, et al. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol. 2012;61:539–42.View ArticlePubMed CentralPubMedGoogle Scholar
- Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010;59:307–21.View ArticlePubMedGoogle Scholar
- Sukumaran J, Holder MT. DendroPy: a Python library for phylogenetic computing. Bioinformatics. 2010;26:1569–71.View ArticlePubMedGoogle Scholar
- Shimodaira H, Hasewaga M. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol Biol Evol. 1999;16:1114–6.View ArticleGoogle Scholar
- Swofford DL. PAUP*: Phylogenetic analysis using parsimony (and other methods) 4.0 Beta. Sinauer Associates. 2002.Google Scholar
- Yang Z. PAML 4: a program package for phylogenetic analysis by maximum likelihood. Mol Biol Evol. 2007;24:1586–91.View ArticlePubMedGoogle Scholar
- Pond SLK, Frost SDW, Muse SV. HyPhy: hypothesis testing using phylogenies. Bioinformatics. 2005;21:676–9.View ArticlePubMedGoogle Scholar
- Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet Journal. 2011;17:10–2.View ArticleGoogle Scholar
- Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full-length transcriptome assembly from RNA-seq data without a reference genome. Nat Biotechnol. 2011;29:644–52.View ArticlePubMed CentralPubMedGoogle Scholar
- Wu TD, Watanabe CK. GMAP: a genomic mapping and alignment program for mRNA and EST sequences. Bioinformatics. 2005;21:1859–75.View ArticlePubMedGoogle Scholar
- Roberts A, Trapnell C, Donaghey J, Rinn JL, Pachter L. Improving RNA-Seq expression estimates by correcting for fragment bias. Genome Biol. 2011;12:R22.View ArticlePubMed CentralPubMedGoogle Scholar
- D’Hont A, Denoeud F, Aury J-M, Baurens F-C, Carreel F, Garsmeur O, et al. The banana (Musa acuminata) genome and the evolution of monocotyledonous plants. Nature. 2012;488:213–9.View ArticlePubMedGoogle Scholar
- Gustafson-Brown C, Savidge B, Yanofsky MF. Regulation of the Arabidopsis floral homeotic gene APETALA1. Cell. 1994;76:131–43.View ArticlePubMedGoogle Scholar
- Yang Y, Fanning L, Jack T. The K domain mediates heterodimerization of the Arabidopsis floral organ identity proteins, APETALA3 and PISTILLATA. Plant J. 2003;33:47–59.View ArticlePubMedGoogle Scholar
- Yang Y, Jack T. Defining subdomains of the K domain important for protein–protein interactions of plant MADS proteins. Plant Mol Biol. 2004;55:45–59.View ArticlePubMedGoogle Scholar
- Gramzow L, Theissen G. A hitchhicker’s guide to the MADS world of plants. Genome Biol. 2010;11:214. http://genomebiology.com/2010/11/6/214.View ArticlePubMed CentralPubMedGoogle Scholar
- Immink RGH, Tonaco IAN, de Folter S, Shchennikova A, van Dijk ADJ, Busscher-Lange J, et al. SEPALLATA3: the ‘glue’ for MADS box transcription factor complex formation. Genome Biol. 2009;10:R24.View ArticlePubMed CentralPubMedGoogle Scholar
- Kaufmann K, Melzer R, Theissen G. MIKC-type MADS-domain proteins: structural modularity, protein interactions and network evolution in land plants. Gene. 2005;347:183–98.View ArticlePubMedGoogle Scholar
- Gocal GF, King RW, Blundell CA, Schwartz OM, Andersen CH, Weigel D. Evolution of floral meristem identity genes. Analyses of Lolium temulentum genes related to APETALA1 and LEAFY of Arabidopsis. Plant Physiol. 2001;125:1788–801.View ArticlePubMed CentralPubMedGoogle Scholar
- Preston JC, Kellogg EA. Conservation and divergence of APETALA1/FRUITFULL-like gene function in grasses: evidence from gene expression analyses. Plant J. 2007;52:69–81.View ArticlePubMedGoogle Scholar
- Heuer S, Hansen S, Bantin J, Brettschneider R, Kranz E, Lörz H, et al. The maize MADS box gene ZmMADS3 affects node number and spikelet development and is co-expressed with ZmMADS1 during flower development in egg cells, and early embryogenesis. Plant Physiol. 2001;127:33–45.View ArticlePubMed CentralPubMedGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.