Transcripts of CrCYC1C and CrCYC1D were first detected in the early developmental stage (E) of the floral bud and then the expression of CrCYC1C and CrCYC1D were gradually reduced to almost invisible from middle (M) to late stage (L) of entire floral buds (Fig. 3-5). To further confirm organ expression pattern of CrCYC1C and CrCYC1D, pooled tissues including petals, gynoecium, stamens and sepals dissected from early development stage were used. Surprisingly, transcripts of CrCYC1C and CrCYC1D were both only detected in sepals (Fig. 3-5). This unique expression pattern of
CrCYC1C and CrCYC1D in C. ramondioides seems to correlate with the residual zygomorphy found in sepal whorl (see Fig. 3-2a). As to CrCYC2, its expression was undetectable throughout all flower developmental stages and dissected organs. In summary, our RT-PCR results showed that no CrCYCs transcripts were detectable at petal and stamen whorl.
As to CrRADs, CrRAD1 expression can be detected throughout all flower developmental stages but CrRAD2 could only be detected in the early floral development stage (Fig. 3-5). Their organ
expression pattern revealed that both CrRAD1 and CrRAD2 expressed in stamens, while CrRAD2 also expressed in gynoecium (Fig. 3-5). The CrDIV was continuously expressed throughout all flower developmental stages. Specifically, CrDIV expressed in petals, stamens and gynoecium but not in sepals (Fig. 3-5).
Discussion
Reversal to actinomorphy correlate to diverse CYC expression shifts in Gesneriaceae and other eudicot lineages
Our SEM pictures showed that all five petals of C. ramondioides seems to grow in consistent rate of enlargement ever since petal primordia initiation stages (Fig. 3-2). Similarly, this was also observed in the stamen whorl. Development of actinomorphy in C. ramondioides therefore resembles to that in cyc/dich mutant of A. majus in which all petal primordia maintain equal growth rate along flower developmental stages and all stamens are fully developed (Luo et al. 1996). From RT-PCR results, CrCYCs and CrRADs have no sign of expression in petal whorl (Fig. 3-5). This suggests C.
ramondioides is a ventralized actinomorphy. In Gesneriaceae, ventralized actinomorphy has been reported in the peloric cultivar of Sinningia speciosa where CYC is not expressed in petal whorl because a deletion in CYC coding region (Hsu et al. 2015, 2017; Wang et al. 2015). Loss of CYC expression in late petal developmental stage but not in early stage can also contribute to
actinomorphic reversal in B. leiophylla (Zhou et al. 2008). On the other hand, dorsalized actinomorphy due to ubiquitous CYC expression in entire petal whorl has been reported in T.
scopulorum and occasionally in Petrocosmea hybrids (Pang et al. 2010; Yang et al. 2015). Other than Gesneriaceae, examples of dorsalized actinomorphy have been reported in Cardia purpurea (Leguminosae), in Aquilegia alpina (Ranunculaceae), and in certain Malpighiaceae lineages (Citerne et al. 2006; Zhang et al. 2010, 2012, 2013; Jabbour et al. 2014). While ventralized actinomorphy where the loss of CYC expression has been documented in Tradescantia (Commelinaceae), in Nigella damascene (Ranunculaceae), in Plantago lanceolata (Plantaginaceae), in Malpighiaceae lineages and in Arabidopsis (Cubas et al. 2001; Reardon et al. 2009; Preston et al. 2011; Preston and Hileman 2012; Zhang et al. 2013; Jabbour et al. 2014).
Although we did not detect any CrCYC1C and CrCYC1D expression in petal and stamen whorls, we did find them distinctively expressed in sepals. Sepal specific expression of CYC has been inferred the ancestral state among Ranunculales including both actinomorphic Papaveraceae species and
zygomorphic Fumariaceae species, the basal most core Eudicot lineage (Damerval et al. 2007). Thus, the shifts of CrCYC1C/1D back to ancestral expression in sepals and cease of expression in
petals/stamens seem to associate with the actinomorphic reversal in C. ramondioides. Indeed, the recruitment of CYC expression (ECE clade) in sepal was evolved earlier than expression in dorsal-specific manner (Preston and Hileman 2009). It would be interesting to examine whether all those reversals to actinomorphy species (Wang et al. 2010; Clark et al. 2011) in Geseneriaceae also have their CYC expression return to ancestral state of sepal expression.
Duplications of CYC may associate to expression shift and flower shape variation
Our phylogenetic analysis revealed that the three CrCYCs we isolated (CrCYC1C, CrCYC1D and CrCYC2) were resulted from at least two duplication events among Gesneriaceae species, congruent to previous findings (Möller et al. 1999; Citerne et al. 2000; Wang et al. 2004; Du and Wang 2008;
Song et al. 2009; Pang et al. 2010; Yang et al. 2015). Although these CrCYCs have lost their expression in petal and stamen whorls, which is correlate to the reversal to actinomorphy in C.
ramondioides, their coding sequences contain no frame shift or nonsense mutations. This implies these CrCYCs may still function yet the regulation controls on cis-elements in their promoter regions could have been mutated to become mis-expression (i.e. sepal only expression). It would be
interesting to test this hypothesis by ectopically express these CrCYCs separately in Arabidopsis and compare their phenotypic effects.
In snapdragon, CYC evolved as major effect copy (higher and broader dorsal-specific expression) and its duplicates DICH as helper function (Luo et al. 1996, 1999). Similarly, CYC tend to duplicate in most angiosperm lineages with either both copies retain similar expression pattern (major/helper) or expressions become diversified thus under different selection pressures (Ree et al. 2004; Chapman et al. 2008; Bello et al. 2017). In C. ramondioides, CrCYC1C and CrCYC1D both have sepal only expression but expression level of CrCYC1C is higher than CrCYC1D. This is similar to the case of
CYC and DICH in snapdragon. Duplication could allow one copy to maintain the essential function but the other to evolve into novel or modified function. CYC duplications may therefore link to the evolution of diverse floral shape in angiosperms, although yet to be determined. There are reports, however, indicating shifts of expression between CYC paralogs correlate to floral symmetry transitions and/or flower shape variations (Bartlett and Specht 2011;
Zhang et al. 2012; Jabbour et al. 2014).
Specific expression patterns of CrRADs and CrDIV may suggest loss of antagonistic expression pattern between CrRADs and CrDIV following loss of CYC expression in C. ramondioides In cyc/dich double mutant A. majus, RAD does not express in dorsal region, thus, allowing ventral region restricted DIV to spread to whole flower (Corley et al. 2005). In B. leiophylla, once dorsal region restricted BlCYC1 and BlRAD are downregulated, BlDIV spread to corolla and stamen whorl (Zhou et al. 2008). To sum up, RAD-like genes and DIV-like genes seem have antagonistic function in A. majus and B. leiophylla. However, expression patterns of CrRADs and CrDIV are different from those gene expression pattern in A. majus and B. leiophylla. In C. ramondioides, both CrRADs and CrDIV expressed in stamens (including CrRAD1, CrRAD2 and CrDIV) and gynoecium
(including CrRAD2 and CrDIV) at the same development stage (Fig. 3-5). Comparing expression patterns of RAD-like and DIV-like genes among these three species, antagonistic function seem have lost in C. ramondioides. Based on expression patterns from these three species, we postulate that loss of expression of upstream gene (e.g. CYC and its homologue) may provide opportunity for its
downstream gene (e.g. RAD and its homologue, DIV and its homologue) releasing from genetic constraint. To conclude, antagonistic expression pattern between RAD-like and DIV-like genes was maintained in B. leiophylla, which resembling to CYC -mediated regulatory pathway in A. majus.
However, this antagonistic expression pattern was lost in C. ramondioides. Since the RAD-like and DIV-like genes were rarely studied in Gesneriaceae, it would be interesting to examine whether the maintaining or loss of antagonistic function between RAD-like and DIV-like genes is a common
pattern or not in Gesneriaceae.
Reversal to actinomorphy may help to attract general pollinator visiting C. ramondioides In C. ramondioides, reversal to actinomorphy coupling with very short corolla tube at anthesis may facilitate generalist (e.g. bees, small beetles) visitation because they can obtain pollen from any direction (Fig. 3-1). Similarly, it has been postulated that reversal to actinomorphy in R. myconii could allow visits from a wider range of pollinators in alpine extreme habitats (Cronk and Möller 1997). In alpine or harsh conditions, plants may suffer low pollinator visiting. If certain zygomorphic species still rely on their specific pollinator visiting in harsh condition, their reproductive success may be low. But if species which can reverse to actinomorphy by opening the corolla, such as C.
ramondioides and R. myconii, they could have pollinator shifts to a variety of pollinators to increase visiting rate. Although C. ramondioides is not distributed in alpine environment like R. myconii, its deep-forest dense shape habitat may discourage insect pollinator visiting (Peat and Goulson, 2005).
Reversal to actinomorphy to attract more general pollinators in pollinator scarce habitats may actually compensate for maintaining the reproductive success in C. ramondioides. To support this idea, detailed pollination experiment and pollinator observation in the field are necessary in the future.
Figures
Fig. 3-1 The flower of C. ramondioides from early to late stage.
Early, middle and late stage corresponding to Stage 10, 13, 15 in supporting file 1: Table S1.
Actinomorphy was observed in late stage C. ramondioides flower. Scale bar represented 1 cm
Fig. 3-2 The SEM photos of morphological development process of Conandron ramondioides flowers.
Zygomorphy was observed at Stage 3 of sepal whorl. Definition of developmental stages of C.
ramondioides based on (Harrison et al. 1999). a Stage 3, the sepals form as bulges at the points of the pentagon. b Stage 4, the sepals grow, while the floral meristem remains undifferentiated. c Stage 6, the corolla and androecium grow, and the gynoecium initiates. d Stage 7A, petal growth. e Stage 7B, androecium and gynoecium development. f Floral diagram of C. ramondioides. Se Sepals,
Pe petals, S stamen, G gynoecium. Scale bar represents 50 μm.
Fig. 3-3 Alignments of protein sequences of CrCYC, CrRAD and CrDIV genes with homologs from Antirrhinum majus (CYC, RAD and DIV) and Bournea leiophylla (BlCYC1, BlCYC2, BlRAD, BlDIV1 and BlDIV2).
a Alignmentes of CYC homologs. TCP, ECE and R domain are outlined. Identical amino acids are in black and similar amino acids are in gray. b Alignments of RAD homologs. MYB domain are
outlined. c Alignments of DIV homologs. Two MYB domains (R2 and R3) are outlined. A highly conserved SHAQKY motif in R3 MYB domain is identified and labelled in white box. Arrows indicate sequence region used in phylogeny analysis. Sequence region used in phylogenetic analysis coving almost all important domains in all three gene dataset except for PlDIV in DIV dataset.
Fig. 3-4 Neighbour‑joining trees of CYC‑like, RAD‑like and DIV‑like genes.
Trees from a to c are reconstructed based on amino acid sequences. a CYC is from A. majus, others
are CYC‑like genes from Gesneriaceae species (see supporting file 2: Table S2). Trees show CrCYCs cluster into three groups with high support. b The tree shows CrRADs cluster into two distinct clade with high support. c Bootstrap values from NJ and ML are listed above and below branches respectively. Bootstrap values > 70 are shown. Detailed sequences information is listed in supporting file 2: Table S2.
Fig. 3-5 Gene‑specific reverse transcriptase polymerase chain reaction (RT‑PCR) analysis of CrCYC, CrRAD and CrDIV genes from C. ramondioides buds and dissected flower tissues.
E, M, L represent three flower development stage defined in this study. E stands for early flower development stage; M stands for middle flower development stage; L stands for anthesis stage. P1 to P5 represent dissected petal from flower bud at early (E) flower developmental stage. Sep, Pe, Sta and gyn denote pooled sepals, pooled petals, pooled stamens and gynoecium dissected from early flower developmental stage. CrCYC1C, CrCYC1D and CrCYC2 indicate expression of CrCYC1C, CrCYC1D and CrCYC2. CrRAD1 and CrRAD2 indicate expression of CrRAD1 and CrRAD2. CrDIV indicates expression of CrDIV. 18S is included as a positive control. CrCYC1C, CrCYC1D, CrRAD1, CrRAD2 and CrDIV are detected through flower development stages, only CrCYC2 is restricted through flower development stages. CrDIV is detected in all petals, whereas CrCYC1C, CrCYC1D, CrCYC2, CrRAD1 and CrRAD2 are restricted in petals. CrCYC1C and CrCYC2 were detected in pooled sepal tissue.