Assessing the Flowering Genetic Regulatory Network in Neotropical Orchids

: During the reproductive transition in ﬂowering plants, a vegetative apical meristem (SAM) transforms into an inﬂorescence meristem (IM) that forms bracts and ﬂowers. In grasses such as rice, a genetic regulatory network (GRN) controlling reproductive transitions has been identiﬁed. It includes the integration of promoters and repressors from different gene lineages with active duplication events during angiosperm diversiﬁcation. With the objective to understand the evolution and expression of ﬂowering GRN in Orchidaceae, we performed comprehensive phylogenetic analyses of all genes from the ﬂowering GRN and analyzed by RT-PCR the expression of targeted homologs in key developmental stages. Our ML results indicate that the FT/TFL1 , FD , FLC/FUL , SOC1 and AGL24/SVP gene lineages have been subject to multiple duplications in monocots, as well as in Orchidaceae. Conversely, FLC genes are lost in Orchidaceae, suggesting major changes in the repression of ﬂowering. Our studies also show active expression of many target genes in Elleanthus aurantiacus (Orchidoideae) in the SAM and in IM, indicating important functions in the reproductive transition. We describe how the ﬂowering GRN in orchids has signiﬁcant variations in copy number and expression patterns when compared to the canonical rice ﬂowering GRN.

Although the flowering GRN has been well studied in grasses, little is known about the genetic mechanisms of flowering in non-model monocots, including orchids.The isolation and characterization of some flowering controlling transcription factors have been done in commercial, mostly temperate orchids like Cymbidium, Dendrobium, Oncidium and Phalaenopsis: here, homologs of FT or SOC1 genes play an important role in promoting flowering [19,20].Nevertheless, comprehensive phylogenetic analyses for all gene lineages involved in the flowering GRN are lacking, and as a consequence, few homologs have been studied, sometimes with unclear affiliation to a specific clade.This is particularly problematic considering that whole genome duplication (WGD) events are abundant in monocots.In turn, gene copy number and homology for all copies needs to be established prior to the expression and functional characterization of the flowering GRN.Our goal is to evaluate the evolution of the flowering GRN in the Orchidaceae (ca.25,000 species), one of the most diverse groups of ornamental angiosperms.Here we use reference transcriptomes from 13 neotropical orchid species to find homologs from the transcription factors known to control flowering and perform comprehensive ML phylogenetic analyses to understand the evolution of all gene lineages involved in the reproductive transition.Our ML results indicate that FT/TFL1, FD, FLC/FUL, SOC1 and AGL24/SVP gene lineages have been subject to multiple duplications in monocots, as well as in Orchidaceae.We also show that FLC genes are lost in orchids.Finally, we evaluate the expression of all target genes in Elleanthus aurantiacus, a tropical and terrestrial member of the Epidendroideae (Orchidaceae), and show the active expression of several factors in the SAM and IM, indicating important functions in the reproductive transition.We show that the flowering GRN in orchids has significant variations in copy number and expression patterns when compared to the canonical rice flowering GRN.

Phylogenetic Analyses of Flowering Candidate Genes
In order to analyze the evolution of flowering-related gene lineages FD, FLC/FUL and SOC1 and identify putative duplication events, we performed searches for gene homologs of all candidate genes using tBLASTX tools.Searches were done in our own reference transcriptomes, as well as in the Orchidstra and OrchidBase, which serve as repositories for orchid genomes and transcriptomes [21,22].The queries were FD, FUL and SOC1 homologs from Arabidopsis, orchids and rice.Detailed methodology for phylogenetic analyses can be found in [23][24][25][26].

Morpho-Anatomical Characterization of the Flowering Transition in Orchidaceae
In order to establish changes in size, and the initiation of lateral organs as well as new morphological features occurring during flowering transition in Elleanthus aurantiacus, light and scanning electron microscopy were used.Detailed steps for sample processing follow [24].

RT-PCR Expression Analysis of GRN Candidate Genes
RT-PCR using cDNA from dissected parts in Elleanthus aurantiacus was performed to evaluate the expression patterns of flowering gene homologs.Dissections follow [23].For the amplification of each homolog, specific primers were designed for each copy, avoiding conserved domains and sometimes including either the 3 or 5 UTRs (Appendix A Table A1).Amplification reactions were done following [25].ACTIN was used as a positive control.

Flowering GRN Genes Have Undergone Multiple Duplication Events
The BLAST search resulted in the recovery of FT, FD, FLC/FUL, SOC1 and AGL24/SVP homologs in all orchid repositories, including our own reference transcriptomes from neotropical orchids (Table 1), as well as other publicly available angiosperm databases used.All sequences were evaluated using ML phylogenetic analyses and resulted in a comprehensive assessment of the flowering GRN evolution in Orchidaceae.A total of 349 PEBP homologs were included to assess the evolution of the FT/TFL1 genes in Orchidaceae.The Amborella trichopoda TFL1 (AmtrTFL1) homolog was used as an outgroup.The topology shows a duplication event prior to angiosperm diversification, resulting in the FT and TFL1 clades [23].TFL1 genes are either lacking or found scarcely in monocots when compared to eudicots [23].Conversely, more copies of FT are found when compared to TFL1.FT genes show a duplication prior to angiosperm diversification, which generates clades FT1 and FT2.In monocots, the MonFT1 genes form a monophyletic group and have undergone at least two rounds of duplication, resulting in the MonFT1A, MonF1B and MonFT1C clades, respectively.On the other hand, the FT2 genes appear to be exclusive to monocots, being absent in the other angiosperm lineages.These genes were duplicated at least twice in monocots, resulting in the MonFT2A, MonFT2B, and MonFT2C (Figure 1a) [23].
genes were duplicated at least twice in monocots, resulting in the MonFT2A, MonFT2B, and MonFT2C (Figure 1a) [23].The FD genes (belonging to bZIP family) were analyzed in a matrix of 156 sequences including diverse angiosperm taxa (Figure 2b).The Amborella trichopoda FD homolog The FD genes (belonging to bZIP family) were analyzed in a matrix of 156 sequences including diverse angiosperm taxa (Figure 2b).The Amborella trichopoda FD homolog (AmtrFD) was used as an outgroup.These genes have undergone specific duplication in Brassicales and Solanales inside core eudicots.In monocots, these genes have undergone at least three duplication events prior to the diversification of the Orchidaceae, forming the OrchFD1, OrchFD2a and OrchFD2b clades.Finally, local duplications have also occurred in Poales.
(AmtrFD) was used as an outgroup.These genes have undergone specific duplication in Brassicales and Solanales inside core eudicots.In monocots, these genes have undergone at least three duplication events prior to the diversification of the Orchidaceae, forming the OrchFD1, OrchFD2a and OrchFD2b clades.Finally, local duplications have also occurred in Poales.ML analyses for FLC/FUL (belonging to MADS-box family) were also performed to understand the evolution and the homology of FLC genes in orchids (Figure 1c).An exhaustive search was done across angiosperms resulting in a matrix with 273 putative homologs.The Amborella trichopoda AGL6 homolog (AmtrAGL6) was used as an outgroup.The resulting phylogenetic tree shows that FLC genes are lacking in orchids, while they are still present in Poales.FLC homologs however have extensively diversified in eudicots.In addition, FUL genes have undergone at least two duplication events in monocots, resulting in the MonFUL1 (also called VRN1 clade), MonFUL2 and MonFUL3 clades.Interestingly, orchids lack homologs in the VRN1 clade and only have FUL2 and FUL3 homologs.
SOC1 gene evolution (belonging to MADS-box family) was also analyzed.The complete matrix comprised 268 angiosperm sequences (Figure 1d).The Amborella trichopoda SOC1 homolog (AmtrSOC1) was used as an outgroup.The ML resulting topology shows at least three duplications prior to the diversification of eudicots resulting in the Eu-diAGL42/71/72, EudiAGL14/19, and EudiSOC1 clades.In monocots, there are three independent duplications prior to the diversification of the Orchidaceae, resulting in the OrchSOC1-1a, OrchSOC1-1b and OrchSOC1-2 clades.
Finally, the AGL24/SVP genes (belonging to the MADS-box family) were analyzed using a matrix of 363 sequences (Figure 1e) [26].The Amborella trichopoda SVP homolog (AmtrSVP) was used as an outgroup.The topology shows a duplication prior to the diversification of eudicots, resulting in the AGL24 and SVP clades.Additional duplications have occurred for AGL24 in eudicots, resulting in the Core-eudi_AGL24a/b clades.Early diverging angiosperms and monocots only have pre-duplication copies.However, at least one independent duplication has occurred in monocots, resulting in the MonSVPLa and MonSVPLb clades, and two additional duplications have occurred in MonSVPLa, generating the orchid-specific OrchSVPLa and OrchSVPLb clades.

The Flowering Transition in Orchidaceae Recruits Several Flowering GRN Genes, Actively
Expressed in the SAM and the IM Morpho-anatomical analyses in Elleanthus aurantiacus (Epidendroideae, Orchidaceae) show that vegetative growth can occur until plants reach ca.1.5 m tall (Figure 2a).The IM starts to differentiate during the rainy seasons (Figure 2e,f), blooming two times per year and yielding inflorescences of 4 to 10 cm long.Light and scanning electron microscopy show that the SAM is ca.150 µm in diameter, forming in its flanks alternate enveloping leaves (Figure 2b-d).During the floral transition, the IM narrows down to ca. 100 µm in diameter and shifts to forming bracts in its flanks with axillary floral meristems (FM) (Figure 2g-j).Each racemose inflorescence forms up to 22-24 flowers.
Expression analyses were performed in dissected organs to understand the possible contribution of the flowering GRN homologs in E. aurantiacus.RT-PCR analyses show a homogeneous expression of the SOC1 genes in vegetative and inflorescence meristems and greater expression of FD in SAM (Figure 2k).It is noteworthy that copies of SOC1 are also expressed in leaves.None of these genes are expressed in fully differentiated floral buds.Additionally, FT1 genes are expressed in the IM, while FT2 genes have wide expression patterns in all tissues analyzed [23].Finally, from the 7 AGL24/SVP copies, only two are expressed; specifically, MonSVPLa is active in the SAM, and OrchSVPLa is expressed in leaves, SAM and IM [26].

Discussion
Most expression and functional analyses of selected flowering genes have been done in model orchids like Cymbidium, Dendrobium and Phalaenopsis [19,20,27].However, little is known about the evolution of each gene lineage across angiosperms in general and Orchidaceae in particular, as well as about their contribution to flowering in neotropical orchids.Our exhaustive phylogenetic analyses of all flowering genes, taking advantage of private and public databases (Figure 1), highlight that the FT, FD, FUL, SOC1 and AGL24/SVP gene lineages have been subject to multiple duplication events in monocots, contrary to what is established in eudicot model species [28][29][30][31].Also, although the Orchidaceae share some duplications with other monocots [32][33][34][35], there are additional family exclusive duplications, and, in turn, orchids have a greater number of gene copies than grasses.It is possible that the increase in copy number is linked to changes in protein structure and, as a consequence, to functional diversification across homologs [23].One of the major differences we were able to find is the absence of canonical flowering repressors.Contrary to the other lineages, FLC genes have only been found in eudicots and Poales [36][37][38][39] and are lost in orchids (Figure 1c).The lack of FLC indicates a profound shift in the vernalization pathway for all orchids, temperate and tropical.It is possible that other genes are being recruited to fill that repressive function when needed.
The observations in E. aurantiacus allow us to conclude that: 1. Rainy seasons control flowering for this terrestrial orchid species in native environments.2. The transition from the SAM to the IM triggers the reduction in size of the meristems concomitant with a shift in gene expression.3.There is overlapping expression in the SAM and in the IM for the following copies: ElauSOC1-1-3, ElauFD1-2, ElauSVP2, ElauFT1A, ElauFT1C2, ElauFT2A2 and ElauMFT.Our results suggest important functions for these transcription factors in the reproductive transition in orchids.Endogenous functional analysis have only been standardized in Dendrobium, wherein the overexpression of DOFT (one of many FT homologs) [40] and DOSOC1 (one of three SOC1 homologs) [41] exhibits earlier flowering than wild-type orchids.These results suggest that both FT and SOC1 genes play an important role in promoting flowering in the Orchidaceae.However, the increase in the gene copy number and our findings about their expression in SAM and IM imply that functional studies from GRN are necessary to find the floral integrator genes with determining functions in flowering transition in Orchidaceae.
Based on our data, we propose two important assessments about the flowering GRN in Orchidaceae: (1) the genes of interest in orchids have undergone different evolution pathways in comparison with grass model species, due to independent duplication events in each group; (2) the increased number of homologs in orchids makes it difficult to assign a promoter or repressor function, and, for that, directed RNA-seq, as well as functional analyses, are a clue to understand the flowering mechanisms employed by the Orchidaceae.

Conclusions
Due to several independent WGD that have occurred inside both Orchidaceae and grasses, the flowering GRN has a remarkable increase in the gene copy number with unknown functions in orchids.Functional and comparative analyses are necessary to understand the role of the different homologs in flowering.It is probable that some of the GRN genes would be conserved in orchids, but the other ones have probably changed in function related to flowering repression.

Table 1 .
Neotropical orchid species with available reference transcriptome s 1 and their number of GRN homologs included in ML phylogenetic analyses.
1Contig statistics for reference transcriptomes available in [23],2Species selected for expression analysis in this study.