Analysis of DFR family members
Dihydroflavonol 4-reductase appears to be one of the most conserved enzymes in the evolution of terrestrial plants. Our DNA cladogram (Figure 2A) shows evidence that this enzyme has representatives with high similarity in all land plants going back to the Bryophytes. The cladogram is rooted with the DFR of the moss Physcomitrella, since this is the most ancient ancestor in the analysis. Fern (Adiantum capillus-veneris) and spike moss (Salaginella moellendorfii) also clade in an outgroup on their own branch. The gymnosperm branch (spruce and pine) lies just below the fern (Figure 2A) with another high bootstrap value (1000). After these outgroups at the base of the tree, the angiosperms branch off into their separate clades of monocotyledons and eudicotyledons, clearly separated from each other.
The DFR protein phylogenetic analysis (Figure 2B) generally follows the same pattern of the DNA cladogram with some anomalies. Again moss roots the protein tree with fern on a neighboring outgroup branch. And again, angiosperms are clearly delineated into eudicots and monocots. The major anomaly arises from the gymnosperm branch clading down near the eudicots with a high bootstrap value (707).
Despite the important phylogenetic relationships that our data suggest, we should be sanguine, but cautious with all these orthologue analyses. Our assumptions throughout this paper are that sequence orthology is the equivalent of functional orthology. Although the homology between sequences is present without doubt, many of these enzymes have not been characterized for functionality. In short, they may be highly similar sequences, but not necessarily functional analogues.
Analysis of F3GT and ANS family members
Both F3GT and ANS demonstrate great conservation, being detectable in the most ancient of plant species (Figures 3, 4). The F3GT DNA cladogram (Figure 3A) demonstrates a straight-forward chronological adherence to proposed phylogeny, while its protein tree shows a minor anomaly (Figure 3B). The anomaly shifts the putative Pinus taeda F3GT from an outgroup position and clades it with barley and wheat.
The ANS analysis shows the greatest anomalies of any of the members of the anthocyanin synthesis pathway. Although, the DNA and protein cladograms (Figure 4) show general agreement with known plant phylogeny--with monocots/eudicots and gymnosperms being in separate clades-- we still see the gymnosperm clade being displaced into the center of the clade and not acting as a general outgroup to all angiosperms. Additionally, Salaginella ANS becomes part of the monocot clade in both its DNA and protein form. We were not able to locate an ANS orthologue for a representative “true” fern. We attribute this result to the present lack of comprehensive sequencing of DNA in the Pteridaceae genomes.
Analysis of F3′H and F3′5′H family members
The flavonol 3′hydroxylase DNA/protein cladograms follow the same phylogenetic model, and the pattern agrees with our current understanding of plant evolution (Figure 5). Again, moss acts as the outgroup followed by spike moss, gymnosperms, and angiosperms. This result supports the hypothesis that production of cyanidin has been evolutionarily conserved from the bryophytes until modern flowering plants. The resultant branch separation also suggests that there have been negligible structure/function modifications in the F3′H during the major evolutionary shifts from mosses to angiosperms. Again, we could discover no F3′H family members in the limited fern genome database found in GenBank.
Note that flavonol 3′hydroxylase (Figure 5) is one of the loci in our study to show a) no changes between the DNA and protein phylogenies and b) “proper” phylogeny in the protein cladogram. This result suggests a strong positive selection pressure to ensure that F3′H DNA and protein structure drifted little over evolutionary time.
The flavonol 3′5′hydroxylase is the most recent addition to the anthocyanin synthesis pathway. We found that F3′5′H cannot be traced to earlier antecedents than gymnosperms (Figure 6). Spike mosses, ferns and mosses show no evidence of this enzyme, which is necessary for the production of the purple/blue anthocyanin pigment delphinidin. As with the F3′H, the F3′5′H family demonstrates no changes between the DNA and protein phylogenies.
Ancient evolution in the anthocyanin pathway
The most ancient terrestrial plants, the bryophytes, arose 450–425 mya when the ozone layer started to form over the earth (Duff and Nickrent 1998; Shear 2000). Before that time, only algae were extant. Algae developed a range of UV-absorbing compounds, since they still needed to be shielded from ultraviolet light even in the water (Rozema et al. 2002; Xue et al. 2005), but they did not evolve anthocyanins. Since the UV-B wavelengths of light were attenuated by the water column for algae, Rozema et al. (2002) suggests that it is likely that phenolic pigments evolved in terrestrial plants to protect them from increased levels of UV-B found on land. Additionally, the Siluro-Devonian colonization of the land by plants was possible in part because the new shielding properties of the ozone layer arose concurrently with the evolution of endogenous color pigments to protect plants from UV light not filtered by the upper atmosphere (Duff and Nickrent 1998; Shear 2000; Rozema et al. 2002).
We have found that mosses, whose ancestors were the first land plants, may have made anthocyanins as early as 450 mya. Although we do not know if the enzymes were functionally orthologous to those homologues from more recently evolved plant species, the anthocyanin pathways were apparently available as long ago as the late Ordovician period to make both pelargonidins and cyanidins (Figures 2, 3, 4, 5). Our result is supported by known evolutionary biochemistry data (Mues 2002; Rausher 2006). Since flowering plants would not evolve for hundreds of millions of years after mosses, we can only assume that a primary function of these ancient pigments in bryophytes was UV protective and antioxidant in nature. This conclusion is supported in the moss literature (Bendz 1961; Post 1990; Mues 2002; Dunn and Robinson 2006). Although Bendz (1961) found the flavone Luteolin in some moss species, other species such as Sphagnum capillifolium have been shown to produce brick-red cyanidin (Mues 2002), especially under environmentally stressed conditions (Steyn et al. 2002; Bonnett et al. 2010).
Wolf et al. (2010) could not detect the presence of anthocyanin pigments in stressed Physcomitrella patens. However, other moss species readily produce detectable quantities of anthocyanin pigment after stress. Steyn et al. (2002) found that anthocyanins generally accumulate in S. capillifolium peripheral tissues exposed to high irradiance or on occasion in the shade due to a disparity between light capture, carbon dioxide incorporation and carbohydrate consumption. Pigment analysis of red arctic moss under UV stress, due to ozone depletion, has shown increases in anthocyanin pigment and decreases in chlorophyll concentrations, largely accounting for the visible alteration in these mosses from green to red (Post 1990). Post (1990) proposed that these changes in pigmentation are consistent with photo-protection, and they are linked to light dependent variations in plastid structure. Dunn and Robinson (2006) observed similarly that “cosmopolitan” mosses found in more temperate regions consistently had reduced levels of anthocyanins in comparison to their arctic cognates. Although it appears that production of these photo-protective pigments is a new, useful adaptation for the bright, UV-rich arctic environment, it is more likely we are looking at a primordial atavistic reaction to the increase in damaging UV fluence.
There is evidence that non-anthocyanin flavonoids, such as the quercetin-glycosides (Herrmann 2006), arose before anthocyanins as a safeguard against UV (Markham 1988; Stafford 1991; Rozema et al. 2002), but that does not make anthocyanins any less important in their own evolutionary pathway and purpose. Stafford (1991) suggested that initially flavonoids, including anthocyanins, may not have been UV protective because they were probably being produced at low concentrations, and further that flavonoids may have initially been primarily phytohormonal in their function. If this is the case, then we may assume that evolutionary selection, beginning in moss, may have eventually allowed enough accumulation of flavonoids to allow them to assist in the UV defensive role (Stafford 1991). Although phenylpropanoid phenolics and sinapic acid esters (Li et al. 1993) could also serve as UV filters and may have been the initial ones, their absorption coefficients are lower than flavonoids. Plants producing large quantities of anthocyanins for UV protection would not have evolved if pigments such as sinapic esters were a sufficient defense.
Given that the anthocyanin pathway appears so conserved, we examined databases of “lower” plants to determine how far back in evolutionary time these pigments can be traced. However, a search of the brown (Ectocarpus), red (Porphyra), and green (Volvox) algae databases in GenBank and the Gene Index manifested no homologous sequences to any enzymes in the anthocyanin production pathway (data not shown). Again, the work of Caputi et al. (2012) does indicate a UGT being present in Chlamydomonas, but this transferase is not part of the UGT clade of anthocyanins. These results agree with known algal biochemistry (Rausher 2006). Although our lack of success in finding homologous sequences may be due to limitations in the sequence databases of these lower photosynthetic organisms, it seems equally likely that the final pathway components simply did not exist until land plants evolved.
We were able to demonstrate that ferns apparently conserved at least two of the major enzymes in the anthocyanin pathways in DFR and F3GT (Figures 2, 3), while spike mosses retained all the major enzymes (Figures 2, 3, 4, 5). We do not believe that ferns are lacking in the ANS and F3′H loci, since more ancient plant species have these genes, however more comprehensive genome databases are required to probe into which loci are actually present in the fern genome.
When ferns evolved ~400 mya (Pryer et al. 2004), they faced the same photo-stresses as their more ancient counterparts, so selection pressure to keep anthocyanins was still present to ensure continued protection. Additionally, anthocyanins can act as chemical deterrents of herbivory, which would have become an important survival issue in ferns as larger animals evolved (Rausher 2006).
More “recent” evolution in the anthocyanin pathway
Gymnosperms appear to be the first land plants (300–325 mya) [19] in which we can observe the presence of the F3′5′H enzyme, which is required to make delphinidin. We cannot detect this hydroxylase expressed in moss, spike moss, or fern (data not shown). It is not entirely clear why this purple blue pigment arose in gymnosperms, but we can hypothesize that it may have been related to the greater spread of plants to higher elevations during the Carboniferous Era. Delphinidin has greater absorption in the red part of the spectrum (maximum 557 nm) and a greater reflection in the UVB portion of the spectrum than either pelargonidin or cyanidin (Harborne 1958). This observation would suggest that as gymnosperms grew taller and migrated to higher elevations, there would be selection for a pigment, i.e. delphinidin, with greater protection against higher fluence of UV light. Alternatively, there is some evidence that gymnosperms may have been the first plants to use blue pigment to attract foraging animals to distribute seeds. Even today some gymnosperms such as the juniper still produce blue, “berry-like” structures that are actually the female seed cone of the juniper tree (Salomonson 1978).
Angiosperms arose during the late Triassic Era (170–245 mya) (Moore et al. 2007), and the five major genes in the anthocyanin pathways continued to be conserved, with the added ability to produce delphinidin being passed down to flowering plants with the F3′5′H enzyme.
Monocots diverged from eudicots 75–100 million years ago (Kellogg 2001). As the monocots diverged, our data suggest that the components of the anthocyanin production pathway, although conserved functionally, began to show greater evidence of amino acid divergence. Except for slight variations, this conclusion seems to be particularly accurate with analyses of DFR, F3′H, F3′5′H, and F3GT (Figures 2, 3, 5, 6). These gene products demonstrate clear phylogenetic divisions between monocots and eudicots. Despite the structural variations in these enzymes, there were no chemical changes in the major anthocyanin pigments (pelargonidin, cyanidin, and delphinidin) after the monocot/eudicot split, even though we eventually see more anthocyanin variants evolve beyond the major ones (Grotewold 2006).
The ANS family of orthologues seems to demonstrate a different divergence pattern from the rest of the anthocyanin synthesis pathway components (Figure 4). The biggest anomaly is that the gymnosperm ANS appear as an outgroup to the eudicots and not closer to the root of the tree. This ANS clading seems to reflect a general lack of divergence.
As another example of this lack of divergence, we see a grouping of monocot grasses clading with the fern ANS (Figure 4A,B). Since those grasses evolved long after the ferns and gymnosperms, we are likely seeing either a) independent evolution leading those monocots back to a higher orthology with an ancient ancestor (Stafford 1991), or b) evidence that in certain plant lineages there were few changes in ANS structure from fern through these grasses. This selection may have arisen in response to grasses being exposed in open fields to higher levels of UV than other plants, which evolved in potentially more shaded environments.
