The homolog of Arabidopsis TCP4, AqTCP4, was expressed at fairly consistent levels across all stages in the spurred taxa, but showed a substantial increase in expression in A. ecalcarata at DS3 and DS4, while AqTCP5 showed a gradual decrease in expression across all taxa, but a slightly higher level of expression in A. ecalcarata.Aquilegia has two BB homologs, one that is expressed at a low level in all taxa and shows a pattern of decreasing expression over development , and one that has fairly steady expression in the spurred taxa across phase I, but starts with a higher level of expression in A. ecalcarata .Variation in the shape of plant organs is determined by combinatorial differences in cell number and cell shape. The Aquilegia petal has evolved substantial morphological variation, particularly in several aspects of nectar spur shape, including length, width, and curvature. Our current understanding of the development of the Aquilegia nectar spur involves two phases, an early mitotic phase and a later cell expansion phase . Despite differences in adult morphology, detailed ontogenetic studies using SEM have shown that from petal initiation through the earliest stages of Phase I, petals of a spurred species, A. olympica, and those of the spurless species, A. ecalcarata, are quite similar in shape. In other species comparisons,vertical hydroponic garden differences in spur length at maturity have been largely attributed to differential cell elongation during phase II of development, however, the developmental basis of other aspects of spur shape have not been studied in detail.
Based on the sampling conducted here, discernible differences in several axes of shape, particularly width and curvature, are apparent quite early in Phase I of petal development, suggesting that cell division plays a role in determining their differences. By examining gene expression across early petal development in several species with variable morphologies, this study allows us to identify gene expression modules that appear to be conserved across petals with diverse morphologies as well as modules that differ in correlation with variation in spur morphology. Further, this comparative approach may provide insight into the developmental processes that underlie the morphologies.Over the course of petal development, GO enrichment analysis of genes commonly DE across all four taxa detected a pattern of declining expression of genes involved in mitotic activity. This pattern was also supported by correlations between WGCNA modules and developmental stages. Modules with eigengenes highly positively correlated with DS1 are enriched for GO terms related to mitosis. This finding of a decrease in mitotic activity throughout development is consistent both with petal developmental patterns in other model systems and with previous studies in Aquilegia. In contrast to this pattern, an enrichment of genes involved in oxidation reduction processes were found to be up-regulated at later developmental stages in all four taxa.
This is supported by both the DE analyses between DS1 and DS5 and by the WGCNA analyses where several modules with eigengenes highly positively correlated with DS5 are enriched for loci involved in oxidation-reduction processes. Exploring the types of genes that are differentially expressed between the spurred taxa and A. ecalcarata detected up-regulation of loci with GO categories related to mitosis in the spurred taxa, but only at DS5 . As Aquilegia petals develop, they begin to transition from cell division to expansion and differentiation, starting at the distal margin of the blade. In spurred taxa, this transition progresses from the petal margins toward the nascent nectary, with cell division persisting longest in the spur itself. Considering the pattern seen in the developmental comparisons and what is known about cellular processes during spur development, the increased expression of genes related to mitosis in the spurred taxa relative to A. ecalcarata suggests that the entire A. ecalcarata petal shifts into the differentiation phase at an earlier time point than in spurred taxa. Another pattern that supports this assertion is that there is an enrichment of loci involved with oxidation reduction processes expressed more highly late in development when considering all taxa, while when considering the loci DE between the spurred taxa and A. ecalcarata across development, there is an enrichment for loci with these processes at early stages in A. ecalcarata . A closer examination of genes with oxidation-reduction GO categorization in the developmental and spurred/spurlesss comparisons revealed a number of cytochrome P450 monooxygenases that at a molecular level function through heme/iron binding and oxidation. Although CYPs have similar molecular functions, they comprise the largest enzymatic gene family in plants and have evolved to play diverse roles in an array of cellular, developmental, and metabolic functions from hormone synthesis to pigment production. Many of these CYP functions appear to be important in differentiated cell types, rather than undifferentiated mitotically active cells. Thus, the upregulation of CYPs earlier in A. ecalcarata development may be consistent with our hypothesis that these petals are accelerated in their differentiation relative to those in spurred species.
Although the WGCNA identified a module that is highly correlated with the presence or absence of spurs , this module showed no GO enrichment. A curious result that emerged when examining the genes that are commonly up-regulated across development in only spurred taxa was the enrichment of genes involved in photosynthetic processes. This was also seen in WGCNA module 20, which contains genes expressed more highly in spurred taxa at DS5 and is enriched for GO terms related to photosynthesis. While this enrichment of photosynthetic genes seems perplexing given that petals are generally not considered photosynthetic organs, we hypothesize that this result is likely an indirect consequence of spur development, rather than a cause. During phase I of A. ecalcarata development, the entire petal is shaded from light by the enclosing sepals. In the spurred taxa, however, elongation of the petal spurs causes them to emerge from the bud and become exposed to direct light, likely inducing baseline expression of photosynthetic loci . Thus, this appears to be a background temporal component of spur development rather than a controlling factor.Although one might predict that more genes would be required for the development of the nectar spur, given its more complex three-dimensional structure compared to laminar petals and blades, a number of lines of evidence indicate that the number of loci required for early spur development may actually be relatively small. Considering the PCA conducted across the set of genes DE between DS1 and DS5 in any taxon , one might have expected that the presence or absence of spurs would be a major PC resulting in the grouping of samples from the spurred taxa apart from A. ecalcarata. However, none of the first 10 principle components examined capture such variation. The first principal component clustered samples by developmental stage,vertical home farming which is not surprising given how genes were selected for inclusion in the analysis. The second PC in this data set groups samples by geographic origin, with the Eurasian taxa clustering together and the North American taxa clustering together. This suggests that phylogenetic relatedness explains more shared developmental differences in gene expression in our data set than whether or not a nectar spur is produced. Several scenarios may contribute to this phenomenon. This pattern may indicate that relatively few genes are necessary to make a nectar spur and, thus, loci that consistently vary between the spurred taxa and A. ecalcarata do not explain a significant proportion of variation in this data set. Not mutually exclusive to this possibility, it may be that the genes important for spur production do not have variable expression levels across phase I of development and therefore were not captured in this set of analyses. Regardless, this result underscores the importance of sampling more than two species across divergent lineages for this type of study since, for instance, a pairwise comparison of A. ecalcarata and any single spurred species might have primarily identified loci that differ due to phylogenetic divergence rather than their morphological differences.
Given that crucial loci in nectar spur development may not be differentially expressed between DS1 and DS5, we also compared expression differences between the spurred taxa and A. ecalcarata at each of our developmental stages. These comparisons showed that during the earliest stages of spur development , there are fewer genes up-regulated in the spurred species than in A. ecalcarata. One possible explanation may be related to the proportion of differentiating cells in the petals of A. ecalcarata versus the spurred species. Given that the A. ecalcarata petal does not produce a spur and is entirely composed of blade tissue, a greater proportion of the A. ecalcarata petal may be in the differentiation phase relative to an equivalent sized petal that is producing a spur. Although a large set of genes are necessary for mitosis, the differentiation of cells into many specialized types may require the up-regulation of an even greater number of loci. In support of this hypothesis, several comparisons from the A. coerulea ‘Origami’ expression data set show a similar pattern. Comparing expression between the blade and the spur showed more loci are up-regulated in the blade and developmental comparisons between the 1mm and 3mm blade and spur cup tissue samples demonstrated that a greater number of genes are up-regulated in both tissues at the later developmental stage . Another consistent data point is seen in the expression of AqTCP4 . Previous studies have found that AqTCP4 is critical for the cell division to cell expansion/differentiation transition in Aquilegia and is expressed in a wave that starts at the blade margin and progresses towards the spur tip. However, in the comparison between the four sampled taxa, AqTCP4 is observed to peak during DS3-4 in A. ecalcarata while it is present at lower levels in the three spurred taxa. This is likely due to the absence of the prolonged proliferation needed for spur development, but may also reflect a developmental acceleration in A. ecalcarata relative to the spurred taxa, representing a heterochronic shift. Therefore, we infer that the greater up-regulation of genes in A. ecalcarata petals suggests that, across the organ, a larger proportion of cells have blade identity and have transitioned to expansion and differentiation. This pattern is likely to change over development as a greater proportion of the spurred petal begins to differentiate, including into cell types that are not present in the blade, such as trichomes and cells associated with the complex nectary. Along these lines, it was surprising to find that STY homologs were not strongly differentially expressed between A. ecalcarata and the spurred taxa, given the absence of nectaries in A. ecalcarata. This may reflect the fact that in situ expression studies have revealed early expression of STY homologs at the distal tip of developing petals, which is consistent with the more deeply conserved role for STY homologs in controlling auxin homeostasis in lateral organs. It is likely that this distal expression domain is also present in A. ecalcarata during the early stages sampled here, making the detection of differential expression in whole petals more difficult than in the previous study where dissected blade and spur tissues were compared at later stages. Even so, there is a discernible trend of increasing AqSTY expression across time in the spurred species relative to A. ecalcarata. While a relatively small number of genes were found to be consistently differentially expressed in entire early petals between the three spurred species and A. ecalcarata, an even smaller number of these genes were also found to be differentially expressed between the blade and the spur tissue of A. coerulea ‘Origami’. Only 35 genes showed consistent differential expression in what we term the ‘blade’ and ‘spur’ comparison classes between these studies, with 27 genes falling into the ‘blade’ class and 8 genes in the ‘spur’ class. Key loci in the development of nectar spurs are likely to act by prolonging mitosis in the spur cup, but none of these 35 loci are homologs of genes known to regulate the transition from cell division to differentiation in Arabidopsis by promoting or repressing mitosis. This list of 35 loci also does not contain several genes that are known to be necessary for proper spur formation. For example, the Aquilegia homolog of JAG has been shown to promote cell proliferation in petals as well as other organs. In the current dataset, AqJAG is expressed at somewhat lower levels in A. ecalcarata versus the spurred taxa, but the temporal expression dynamics across the five stages are quite similar .