Seeing red to being red: conserved genetic mechanism for red cone oil droplets and co-option for red coloration in birds and turtles

1. Introduction

Ketocarotenoids are red xanthophyll carotenoids whose terminal ketone groups, when on the C4 position on one or both end rings, lead to the absorption of light at longer wavelengths than in other xanthophylls [1]. They have been extensively studied in birds where they are the predominant mechanism of saturated red colours that are frequently sexually or socially selected signals [2–5]. In addition, ketocarotenoids perform another key function in birds: the ketocarotenoid astaxanthin (3,3′-dihydroxy-β-carotene-4,4′-dione) is the red pigment found in the oil droplets in long-wave sensitive (LWS) cones of the avian retina [6], where they act as cut-off filters to enhance colour discrimination [7–10]. Ketocarotenoids in most land birds are not obtained directly from the diet but are instead synthesized by ketolation (addition of double-bonded oxygen at the C4 position of the end ring) of dietary yellow carotenoids such as zeaxanthin [1]. The mechanism was for a long time obscure, but recent studies have strongly implicated a locus encoding a cytochrome P450 monooxygenase (CYP2J19) as the main or only avian ketolase enzyme that catalyses the conversion of dietary xanthophylls to the ketocarotenoids that are deployed in both red integumentary coloration and red retinal oil droplets [11,12]. This raises the intriguing question of the origin of CYP2J19 function in birds, and its ancestry in vertebrates more generally.

Turtles (Testudines) are the only group of tetrapods apart from birds that possess red retinal oil droplets [6,13,14], and these probably contain ketocarotenoids [15,16]. By contrast, other extant reptile lineages either lack retinal oil droplets entirely (crocodiles, snakes) or have coloured oil droplets varying from clear to yellow and green, but not red (many lizards) [14]. Like birds, turtles have excellent tetrachromatic colour vision [17,18].

Red coloration of the integument is uncommon in turtles, but red coloration specifically due to the presence of ketocarotenoids has been reported [19] and shown to be related to sexual selection [20]. Importantly, aquatic turtles, unlike most land birds, have the potential to obtain red ketocarotenoids directly from their diet [21,22], and whether turtles have the capacity for carotenoid ketolation remains unknown.

Several interesting questions arise: is there a shared genetic basis for ketocarotenoid synthesis in birds and turtles? Alternatively, was ketocarotenoid synthesis independently derived in the two groups, or even absent in turtles? If ketocarotenoid synthesis does have a shared genetic basis, was the original function of CYP2J19 for colour vision or coloration?

The phylogenetic position of turtles within the reptiles has long been a source of debate [23] mainly due to the highly derived morphologies of turtles and the somewhat contradictory molecular evidence [24,25]. More recently, a number of large-scale molecular studies using genomic data support a sister relationship between turtles and the archosaurs, forming the turtle–archosaur clade (Archelosauria) independent from the lepidosaurs (lizards, snakes and tuatara) [26–31]. With turtles now placed as an outgroup to archosaurs, the evolution of a locus related to colour vision and coloration in turtles has the potential to inform the occurrence of these traits in the dinosaurs and pterosaurs.

Here, we address the genetic basis of ketocarotenoids in the retina and integument of turtles and birds using two approaches. First, we mine recent genomic data and perform phylogenetic analyses to reconstruct the evolutionary history of CYP2J19 within reptiles and birds. Second, we examine expression of CYP2J-like loci in the western painted turtle (Chrysemys picta bellii), which possesses both red retinal oil droplets and red ketocarotenoid-based integumentary coloration, to obtain evidence for functional involvement of CYP2J19 in ketocarotenoid synthesis in different tissues in Testudines. We find that CYP2J19 is conserved in birds and turtles but has apparently been lost in crocodiles, and that patterns of CYP2J19 expression are consistent with a function in both colour vision and coloration in turtles, as in birds. We interpret these results to suggest that the carotenoid ketolase CYP2J19 originally functioned in colour vision in the Testudines–archosaur clade ancestor, and has since been independently and repeatedly recruited for red integumental coloration by changes in gene regulation in turtles and birds. Finally, our results imply the presence of red retinal oil droplets in lineages related to the birds, including dinosaurs and pterosaurs, which provides new insights into the reconstruction of the evolution of colour vision in these groups.

2. Results

(a) Evolution of CYP2J genes in amniotes

BLASTn searches revealed the presence of two CYP2J family loci in the three most complete turtle genomes out of the four available (western painted turtle, Chinese soft-shell turtle and green sea turtle) [28,29]. By contrast, we found at most a single CYP2J sequence in crocodilians (American alligator, Chinese alligator) and lepidosaurs (green anole, Schlegel's Japanese gecko). Strikingly, phylogenetic analyses revealed that one of the CYP2J loci in turtles is orthologous to avian CYP2J19, because turtle sequences form a well-supported (100% bootstrap support) monophyletic clade with CYP2J19 sequences from birds (figure 1). No CYP2J19 orthologues were identified in either crocodilians or lepidosaurs. The remaining reptile and avian CYP2J loci group together by order but support for other relationships is variable. Reconstructions suggest that the crocodilian and second turtle CYP2J locus are related to avian CYP2J40, but the bootstrap support is weak. The single CYP2J locus in lepidosaurs is placed as an outgroup to the Testudines and archosaur (crocodile + bird) sequences, with moderate bootstrap support (60%). The single human CYP2J locus (CYP2J2) is clearly an outgroup to all of the reptile and avian sequences (99% bootstrap support).

• Download figure
• Open in new tab
• Download PowerPoint

Overall therefore, we find strong evidence for the conservation of the CYP2J19 locus in avian and turtle (Testudines) lineages, but no evidence for the presence of CYP2J19 in crocodilians or lepidosaurs. Furthermore, our phylogenetic analysis indicates that CYP2J19 arose from a duplication event that occurred after the divergence of reptiles and mammals and before the split between Testudines and archosaurs and CYP2J19 was subsequently lost in crocodilians.

(b) Specific expression of CYP2J19 in tissues containing red ketocarotenoids in the western painted turtle

Preliminary experiments revealed high CYP2J19 expression in the retina and red plastron of western painted turtles of both sexes, and lower, variable expression in the liver, black plastron, yellow tail, black tail, yellow neck and black neck regions (not shown). By contrast, CYP2J40 expression was detected in all tissue types (not shown).

Quantification of expression via qRT-PCR confirmed the presence of high CYP2J19 expression in the ketocarotenoid-containing tissues (retina, red plastron) of both sexes, and low expression in other tissues (figure 2). There was significant heterogeneity of CYP2J19 expression level among different tissue types (F_4,25 = 35.83, p < 0.001) but not among sexes (F_1,28 = 0.4524, p = 0.51) (table 1). As no sex-biased expression was detected, a comparison between all tissue types pooled for both sexes was carried out. The red plastron had the highest CYP2J19 expression relative to all other tissues (p < 0.05), while the retina had higher expression when compared with all tissues except for the red plastron (p < 0.001) (table 1).

• Download figure
• Open in new tab
• Download PowerPoint

Table 1.Tukey's pairwise tests of CYP2J19 expression.

Collapse

pairwise tissue comparisons	p adj
liver, retina	0.0000420***
yellow neck, retina	0.0000284***
black plastron, retina	0.0000188***
red plastron, retina	0.0434566*
yellow neck, liver	0.9998532
black plastron, liver	0.9975022
red plastron, liver	0.0000000***
black plastron, yellow neck	0.9998199
red plastron, yellow neck	0.0000000***
red plastron, black plastron	0.0000000***

3. Discussion

We provide compelling evidence for the presence of a CYP2J19 orthologue in three divergent Testudines lineages that suggests conservation of CYP2J19 function in turtles and birds, despite approximately 250 million year divergence between these lineages [27]. Expression patterns of CYP2J19 in painted turtles are consistent with this gene playing a role in ketocarotenoid synthesis in both retina and red-coloured integument, as recently described in birds [11,12]. This raises the question of whether the original function of CYP2J19 was for colour vision or coloration. Consideration of the likely ancestral traits, given the extant distributions, strongly favours colour vision as the original function, as we now discuss (figure 3).

• Download figure
• Open in new tab
• Download PowerPoint

Colour vision involving ketocarotenoid-containing oil droplets in LWS cones appears to be a pervasive feature in almost all birds and turtles [6,15,16], implying that red oil droplets were present in both ancestral birds and ancestral turtles. By contrast, red ketocarotenoid coloration has a patchy distribution in both groups. In turtles, red coloration is sparsely distributed among both the pleurodirans and cryptodirans. Redness appears to be rare or absent in most turtle families except for two cryptodiran families, Emydidae and Geomydidae, where red coloration is more common, as well as several taxa within the pleurodiran family Chelidae. In addition, the pigmentary basis for redness in turtles has not been widely characterized. In birds, reconstruction of carotenoid-based plumage coloration has shown that this trait is not ancestral but was acquired multiple times during avian evolution [32]. Evolutionary reconstructions of carotenoid coloration in bare body parts (such as bills) in birds are still awaited. An independent line of evidence suggesting that carotenoid coloration in the integument was not ancestral in birds concerns their lack of specific pigment cells containing carotenoids. Reptiles, like amphibians and teleosts, express carotenoids (along with pteridines) in specialized xanthophores/erythrophores [33], whereas birds express integumentary carotenoids in keratinocytes. Although studies have found pigment cells in the avian iris that resemble reptilian xanthophores and iridophores [34,35], the absence of specialized pigment cells for carotenoids in the integument of birds seems to imply a period in their ancestry when carotenoids were not used for coloration.

As tetrachromatic colour vision involving red retinal oil droplets was probably present in the ancestral turtle and avian lineages, while red carotenoid-based coloration was probably absent in both, we conclude that the ancestral function of CYP2J19 was for production of ketocarotenoids in red retinal oil droplets and that the role of CYP2J19 in red carotenoid-based coloration evolved independently in the two lineages (figure 3). Many examples of phenotypic convergence via similar genetic changes have been observed across the vertebrates [36–38]. Several examples involve melanin pigmentation, and include the MC1R locus, which has been involved in convergent evolution of pale/dark coloration in birds and reptiles [39–41]. Here, we highlight a remarkable example of convergent evolution whereby CYP2J19 would have been independently recruited in the bird and turtle lineages, through changes in gene expression, to function in ketocarotenoid-based coloration from an ancestral retina-specific function. The ecological driver for this convergence in many instances may have been sexual selection [20,42].

Our results imply that red retinal oil droplets themselves were present in the common ancestor of the archelosaurs (archosaurs and turtles) [31], so that red oil droplets had a single origin in tetrapods (figure 3). An alternative would be if CYP2J19 originally arose for a different function than red oil droplets, but this is unlikely because no other functions for ketocarotenoids have been described. The adaptive function of red retinal oil droplets in these early reptiles is an interesting and open question. Their acquisition presumably led to improved colour discrimination at long wavelengths, particularly in bright light [7,8], but the ecological context for this is obscure.

The inferred timing of the gene duplication that produced CYP2J19 after the split with lepidosaurs is concordant with the absence of red oil droplets in this clade. The apparent absence of CYP2J19 in crocodilian genomes could be explained by the derived loss of all oil droplets from the retina that has occurred in crocodilians, possibly due to nocturnality [43], as in snakes where an ancestral period of nocturnality may have resulted in loss of retinal oil droplets [44,45]. Although we cannot completely rule out the presence of a divergent CYP2J19 locus in crocodiles and/or lepidosaurs, which was not identified in our genome searches, it is notable that our methods did successfully identify one CYP2J locus, as well as CYP2D6 and CYP2R1 in two crocodilian and two lepidosaur genomes (electronic supplementary material, figure S1). The function of CYP2J40 is unknown, but the observed expression among all the tissues studied (in both turtles and birds [11]) does not suggest any direct involvement of this locus in red coloration.

As birds are phylogenetically nested within the saurischian dinosaurs [46], the continuous presence of red retinal oil droplets in the avian lineage since the split with Testudines implies the presence of red oil droplets in the retina of dinosaurs and other lineages (including pterosaurs) that are more closely related to the living birds than crocodiles. Building on the previous inference of colour vision in dinosaurs based on four cone opsin types [47], the presence of red retinal oil droplets in dinosaurs suggests that they possessed similar tetrachromacy to the birds and turtles, with good colour discrimination at long wavelengths. This is concordant with the proposed use of coloration in intraspecific signalling in dinosaurs, where putative melanosomes have been used to infer melanin-based coloration and iridescence [48–51]. The co-option of CYP2J19 for red ketocarotenoid coloration in dinosaurs, as has occurred in birds and turtles, may also have been possible.

The implied presence of a functional carotenoid ketolase in turtles, which are primarily aquatic, is interesting in view of the availability of ketocarotenoids in potential prey items in aquatic environments, e.g. crustaceans, which means that in principle turtles might have lost ketolase activity. If true, this may be a legacy of their terrestrial ancestry in basal archelosaurs. Another possibility is that tight control of carotenoid concentration in retinal oil droplets is required for good colour vision, and that there is too much variation in dietary ketocarotenoids to rely on this source. In this context, it is interesting to note that whereas, as expected, severe dietary carotenoid deprivation is harmful to colour vision in birds, dietary supplementation of carotenoids does not affect the carotenoid content in oil droplets used for colour vision [52–54].

In birds, the anatomical site of ketolation for red coloration has long been contentious but the pattern of CYP2J19 expression has now shown that it is variable among taxa. Some species, e.g. zebra finch, perform ketolation in the peripheral tissues such as the beak where ketocarotenoids are deposited [11,55], whereas in other species such as ploceids, ketolation occurs centrally in the liver, and ketocarotenoids are transported in blood to sites of deposition (H Twyman, M Prager, NI Mundy, S Andersson 2016, unpublished data). In turtles, the presence of CYP2J19 expression in red integument, but not in the liver, is indicative of the peripheral conversion model in this group.

Apart from birds and turtles, the only other vertebrate in which red retinal oil droplets have been reported is the lungfish Neoceratodus [56]. Microspectrophotometry measurements on these red droplets showed that they had similar spectral properties to the red droplets of birds [57]. However, acquisition of red oil droplets in this lineage was probably independent of that of archelosaurs because CYP2J19 arose a long time after the split between lungfish and tetrapods, and this is supported by our failure to find a CYP2J19-like sequence in the Neoceratodus genome using BLAST searches.

In summary, we have uncovered a remarkable case where a gene with a strongly conserved function in colour vision has been independently co-opted for red ketocarotenoid-based integumentary coloration in turtles and birds via changes in patterns of gene expression. As CYP2J19 arose within reptile evolution, the genetic basis for red ketocarotenoid coloration in amphibians and ray-finned fish warrants further research.

4. Material and methods

Ethics

All procedures were approved by the IACUC of Iowa State University.

Data accessibility

DNA sequences: GenBank accessions KX553993-4.

Authors' contributions

H.T. designed the experiments, carried out molecular laboratory work, analysed the data and drafted the manuscript; N.V. and R.L. supplied the samples and helped edit the manuscript; S.A. helped to conceive the study and helped edit the manuscript; N.I.M. conceived the study, designed the experiments and drafted the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

We thank the BBSRC Doctoral Training Partnerships (to H.T.) for funding.