Analysis of the floral transcriptome of H. coronarium yields insights into its flower development
RNA-Seq has become an important tool to unravel the molecular players underlying floral diversification in non-model plants that lack a reference genome, such as the Zingiberales. In recent years, whole flower transcriptome libraries of exemplar species from the six families (Musaceae, Lowiaceae, Zingiberaceae, Costaceae, Marantaceae and Cannaceae) comprising the Zingiberales have been generated and analyzed in previous studies [19, 30]. Our transcriptome data is consistent in terms of unigene numbers, unigene length distributions, and quality scores with the aforementioned studies. Moreover, given that we sequenced two different developmental floral stages in H. coronarium, we were able to assess differential gene expression, finding that 215 unigenes showed differences between the Hp and Hd stages, with 187 significantly up-regulated and 28 significantly down-regulated. The GO analysis for these DEGs annotated many of them as related to cellular processes and metabolism. Another notable finding is that while the annotated 2142 transcription factor sequences were a small fraction of the overall unigenes recovered (2.65%), we found 56 gene families represented. Interestingly, we also found some of these genes exclusively expressed in the Hp (29) or Hd (25) stage. We also detected 57 transcription factors that were differentially expressed between stages (Fig. 4b), and many of these are MADS-box genes (discussed further in next section).
Comparing our unigenes with previous studies where lineage-specific genes had been retrieved only in certain lineages within the Zingiberales, some of the most interesting genes were homologs to LOB-domain genes, a plant-specific transcription factor family which has been implicated in defining organ boundaries between Arabidopsis flower organs through the negative regulation of brassinosteroid accumulation [23, 42], as well as root development, auxin signaling and stress response [43]. The shared floral transcription factors recovered for all species across the exemplar Zingiberales species analyzed, such as LOB40, 41 and 6-like homologs [30], were also identified in the H. coronarium floral transcriptome, suggesting that these genes could be conserved regulators of floral development across the Zingiberales. In contrast with previous findings, a LOB36-like gene which had been formerly recovered only in the banana clade [30], was recovered in our H. coronarium. We also recovered a LOB18-like sequence, which was previously recovered only in the Canna-Calathea lineage (Cannaceae-Marantaceae lineage) and in Zingiber officinale (Zingiberaceae) [30]. As expected given its close phylogenetic relationship, the LOB4-like gene, which had been previously recovered only in Z. officinale [30] and Z. zerumbet [19], was also retrieved in the H. coronarium floral transcriptome. A CUC2-like gene, previously recovered only in the Canna-Calathea and Zingiber lineages [30], was also identified in our data for H. coronarium, while PTL, previously only recovered in Zingiber, was also retrieved in our Hedychium floral transcriptome, consistent with their close phylogenetic relationship. Most of the genes highlighted above, have been implicated in floral organ boundary formation in A. thaliana [22, 25, 42], but knowledge of their roles outside of core eudicots is still scarce.
Although we didn’t sequence more developmental stages on account of the difficulty in obtaining plant materials, we believe the transcriptome analyses presented here add to the body of data that can provide a crucial bioinformatic basis for future functional studies on the molecular mechanisms of flower development and diversification in the Zingiberaceae.
MADS-box genes involved in floral organ identity specification recovered in H. coronarium
In this study we recovered 51 MADS-box genes, including those involved in floral organ specification according to the ABC model [7]. Previous comparative developmental studies addressing the functional conservation of the orthologs of these MADS-box genes have been conducted in different non-model angiosperm lineages, including the Zingiberales, unraveling their role in floral meristem determinacy and floral organ identity specification in both eudicots and monocots [13, 15, 17, 44–46]. In the Zingiberales, the families comprising the ginger group, show a trend towards stamen modification into petaloid structures, as in all lineages petaloid staminodes replace several fertile stamens in the two concentric stamen whorls present in developing flowers [4, 5]. This kind of homeotic conversion was first considered to be possibly caused by the differential expression of B- and C-class genes in the petaloid staminodes, thereby allowing a shift in organ identity [47]. However, evidence provided by later studies in Alpinia hainanensis (Zingiberaceae) and Canna indica (Cannaceae) demonstrated that the C-class gene is expressed in both the petaloid staminode and the fertile stamen [13, 15, 17, 46], indicating that differential gene expression may not be the main reason for the shift in organ identity. In our work, the expression patterns of the homologs of ABC-class genes analyzed in H. coronarium is in line with previous observations. Based on our quantitative transcriptomic data, the A-class gene HcAP1 displays a higher expression in the undifferentiated floral primordium, while the B and C class genes HcAP3 and HcAG, respectively, are significantly up-regulated in flowers with differentiated floral organs. The relatively high expression of HcAP1 in the floral primordium, might be due to the observation in other taxa that this gene plays an important role in maintaining floral meristem identity and promoting sepal differentiation, functioning at earlier stages before sepal initiation, or, alternatively, that it plays a cadastral role, limiting the domain of expression of the C-class gene [11]. In contrast, B- and C-class genes exert their function in the inner three whorls (petal, stamen and carpel whorls), which was evidenced by a higher expression in organ-differentiated flowers. qRT-PCR detection in different floral organs showed that the expression of the A-class gene HcAP1 was detected not only in sepals, but also in all four floral whorls, with a higher expression in petals (Fig. 5a). The concerted expression of HcAP3 (B-class) and HcAG (C-class) genes was detected in floral organs corresponding to whorl 3, both in the fertile stamen and the sterile stamen-derived petaloid structures (petaloid staminode and the resulting labellum), with no significant differences in their expression levels across these structures (Fig. 5b and c); HcAG, as expected, had a higher expression in the developing carpel (Fig. 5c), where HcAP3 expression was absent.
Our results support the suggestion of A-class gene function as discussed in the (A)BC model put forth by Causier and colleagues (2010) [48], and further indicate that the formation of petaloid structures and stamen fertility are not necessarily correlated with the differential expression of B- and C-class genes. Instead, independent gene duplication events in the B and C-class AG lineage, which have been documented in several eudicot and monocot taxa, including species within the Zingiberales [16, 49, 50], may potentially lead to functional divergence among paralogs or dissociation among genetic modules thereby increasing the genetic substrate that can lead to morphological diversity [49]. For instance, the duplication of the AG gene followed by functional divergence of the duplicated AG copies during the diversification of the ginger families, may explain some of the observed morphological changes in Zingiberales flowers, indicating possible mechanisms for the evolution of androecial petaloidy within this order [16, 18]. Future research should address if paralogs of the B-class genes exist in H. coronarium as well as investigate the function of a potential AG paralog detected in the transcriptome analyses (Fig. 4a).
Additionally, while MADS-box transcription factors expressed in developing flowers have been studied in multiple species, these genes have been ignored for their potential roles in floral organ pigmentation. A recent study in orchids found that the expression levels of AP3 and AGL6 displayed more differences between lip segments (the yellow hypochile v. the purple-red epichile), than between the sepal/petal segments which have a similar coloration during floral development, suggesting that AP3 and AGL6 might participate in color differentiation in the perianth and lip segments during flower development [51]. Similar to orchids, the labellum of different genera in the Zingiberaceae displays a wide variation in color and shape, as well as ontogenetic identity [19]. In this context, the potential involvement of MADS-box genes in flower color differentiation across different floral organs has yet to be studied. The MADS-box transcription factors retrieved in this study provide a global candidate gene list for future studies that can address not only the molecular mechanisms underlying floral organ identity specification, but also assess if they play a role in pigmentation in the different petaloid organs present in Zingiberaceae species.
HcPTL may promote growth in the developing floral meristem and floral organs
PTL is a plant-specific trihelix transcription factor involved in regulating perianth architecture in the Arabidopsis flower [52]. Mutations in the PTL gene have a phenotype of petal defects and sometimes fused sepals, indicating that it plays a role in promoting petal initiation and boundary delimitation between adjacent sepal primordia [52, 53]. In Arabidopsis, the PTL transcripts were detected in four inter-sepal zones in the early-developing flower, where they repressed organ growth in regions between newly initiating sepals, ensuring their separation [52]. Although PTL loss of function could influence the number of petals, PTL expression could not be detected in early-developing petals, suggesting that the observed petal defects associated with PTL loss of function may be indirect, perhaps influenced by the overgrowth of nearby inter-sepal zones [52]. Thus, while PTL has been regarded as a boundary gene modulating connation between adjacent organs within the same floral whorl, its activity has not been widely investigated in monocots. If PTL functions as a conserved boundary-specification regulator, it should not be expressed in inter-sepal boundaries during the formation of the synsepalous calyx in the Zingiberaceae. However, the expression pattern of a PTL-like homolog in developing Hedychium flowers contrasts with what has been described in Arabidopsis. The transcripts of HcPTL were still detected in inter-sepal boundaries. In addition, the expression of HcPTL was detected in other structures, such as in developing meristems, including cincinnus primordia, floral primordia, common primordia and almost all new initiating floral organ primordia. This indicates HcPTL may function not only in modifying calyx architecture but throughout early floral organ development in other whorls. Furthermore, while PTL in Arabidopsis typically represses cellular growth within its domain of expression (boundaries), HcPTL could have acquired a novel role, as it seems to be promoting growth in developing meristems and floral organs in H. coronarium, as well as organ differentiation in common primordia. Notably, the floral organ-specific qRT-PCRs for HcPTL showed a pattern of expression more consistent with data from A. thaliana, namely, a higher expression in sepals and petals at late developmental stages (See Additional file 2: Fig. S3). This divergence in function of a boundary gene with respect to its documented role in A. thaliana floral development, where HcPTL seems to promote growth in floral organs rather than limiting it, was documented in another boundary gene related with the CUC1-3 genes, CUPULIFORMIS (CUP) in the Antirrhinum flower [54]. In this context, it might be possible that the slight growth rate difference observed between petal and stamen/inner androecium members in H. coronarium (Fig. 6e-6 h) might be explained, at least in part, by the differential expression of HcPTL in the dorsal/ventral part of the common primordia, so as to ensure that the petals envelop the stamen/inner androecium. Thus, a distinct mechanism of HcPTL boundary delimitation could have evolved in the morphogenesis of the Hedychium flower.