Carotenoid modifying enzymes in metazoans

Abstract

Carotenoids constitute an essential dietary component of animals and other non-carotenogenic species which use these pigments in both their modified and unmodified forms. Animals utilize uncleaved carotenoids to mitigate light damage and oxidative stress and to signal fitness and health. Carotenoids also serve as precursors of apocarotenoids including retinol, and its retinoid metabolites, which carry out essential functions in animals by forming the visual chromophore 11-cis-retinaldehyde. Retinoids, such as all-trans-retinoic acid, can also act as ligands of nuclear hormone receptors. The fact that enzymes and biochemical pathways responsible for the metabolism of carotenoids in animals bear resemblance to the ones in plants and other carotenogenic species suggests an evolutionary relationship. We will explore some of the modes of transmission of carotenoid genes from carotenogenic species to metazoans. This apparent relationship has been successfully exploited in the past to identify and characterize new carotenoid and retinoid modifying enzymes. We will review approaches used to identify putative animal carotenoid enzymes, and we will describe methods used to functionally validate and analyze the biochemistry of carotenoid modifying enzymes encoded by animals.

1. Introduction

Carotenoids are a large class of compounds produced by photosynthetic algae, plants and bacteria, as well as by some non-photosynthetic fungi, archaea, protists and bacteria. Carotenoids are part of the terpenoid family of compounds which also includes sterols, hopanoids and ubiquinones. The basic building block of carotenoids, the C5 isoprene unit is composed of isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) (reviewed in Moise, Al-Babili, & Wurtzel, 2014). In yeasts, archaea, and animals, IPP and DMAPP are derived exclusively via the mevalonate pathway while in eubacteria, protists, cyanobacteria, alga and plants these compounds can also be generated via the 2C-methyl-Derythritol-4-phosphate (MEP) pathway. The head-to-tail condensation of three IPP and one DMAPP molecule affords the diterpene geranylgeranyl diphosphate (GGPP). Two molecules of GGPP then undergo head-to-head condensation to form prephytoene and then phytoene via phytoene synthase (PSY), which catalyzes the rate limiting and first committed step of carotenogenesis. Therefore, expression of an enzymatically active PSY enzyme by a tissue or an organism largely reflects its ability to carry out carotenoid biosynthesis. Phytoene is converted to the linear carotenoid lycopene through a series of desaturation and isomerization reactions involving a single enzyme (CRTI) in bacteria, or four separate enzymes in plants and cyanobacteria, (PDS, Z-ISO, ZDS and CRTISO). Lycopene is cyclized to generate β- or ε-ionone rings which together with uncyclized ψ-ends can be further substituted via hydroxylation, ketolation, epoxidation, and desaturation to generate a very wide variety (>1100) of carotenoid species (Britton, 2022). Alternatively, the polyene chain of carotenoids can be isomerized and oxidatively cleaved to generate apocarotenoids which are then modified to produce bioactive signaling molecules such as abscisic acid from xanthoxin, or strigolactones via carlactone (Moreno, Mi, Alagoz, & Al-Babili, 2021).

The primary roles of carotenoids in carotenogenic organisms involve photosynthesis and photoprotection. Unmodified carotenoids also play other important roles in preventing oxidative stress and in attracting pollinators of flowering plants. Apocarotenoids such as abscisic acid and strigolactones act as hormones and signaling molecules in regulating plant development and fruit ripening, and the plant response to biotic and abiotic stressors (Moreno et al., 2021). In addition, strigolactones, β-ionone and other apocarotenoids also mediate interspecies interactions of plants with other plants, herbivores and symbiotic fungi. Fungi utilize apocarotenoids such as trisporic acid to induce sexual differentiation, while bacteria, algae and archaea utilize retinaldehyde-coupled-opsins as light-driven membrane transporters to facilitate phototaxis and energy production.

2. Roles of carotenoids in the animal kingdom

Carotenoids and carotenoid-derived compounds are essential micronutrients for animals (Harrison & Quadro, 2018; Toews, Hofmeister, & Taylor, 2017; von Lintig, Moon, & Babino, 2021). The animal kingdom, or Metazoa, is divided into five clades that includes Porifera (sponges), Ctenophora (comb jellies), Cnidaria (corals, jellyfish), Placozoa and Bilateria. Bilateria is further divided into Deuterostomes (including Chordates, Echinoderms) and Protostomes (including nematodes, arthropods, mollusks, annelids, and flatworms) (Dunn, Giribet, Edgecombe, & Hejnol, 2014). The best understood contribution of carotenoids to animal biology is in the form of precursors of vitamin A or all-trans-retinol. Vitamin A can be obtained by animals from a plant-based diet in the form of provitamin A carotenoids, such as β-carotene and β-cryptoxanthin, which are characterized by having at least one unmodified β-ionone ring. In addition, preformed vitamin A in the form of retinol and retinyl esters can be obtained by consumption of other animals. Dietary deficiency of vitamin A (either provitamin A carotenoids or preformed retinol) can lead to a plethora of human diseases such as birth defects, blinding disorders, immune deficiencies, tissue damage and cancer. Importantly, excess of preformed vitamin A in the diet can also lead to birth defects and other disorders.

Retinol, retinaldehyde and their retinoid metabolites play essential roles in metazoan biology. In what is perhaps their earliest role in animal biology, retinaldehyde associated with opsin proteins forms the visual pigment of animals which allows for light sensing (Gehring, 2014). Just as in eubacteria (Jung, Trivedi, & Spudich, 2003), retinaldehyde-coupled opsins mediate light-dependent processes of animals, such as photoreception, vision and phototaxis. Similar to microbial opsins, animal visual opsins are characterized by a seven-transmembrane α-helical motif and are covalently coupled to a cis-retinaldehyde-based chromophore such as 11-cis-retinaldehyde, 11-cis-3,4-didehydroretinaldehyde, or 11-cis-3-hydroxy-retinaldehyde. Animal visual opsins employ a G-protein mediated signaling mechanism that is distinct from the light-driven ion pump bacteriorhodopsin or channelopsin of bacteria and algae (Terakita, 2005). Visual opsins include rod opsin, or rhodopsin, and several cone opsins which are distinguished based on their spectral properties. Upon light exposure, the retinaldehyde moiety undergoes isomerization and induces a conformational shift of opsin which activates phototransduction resulting in the activation of a G-protein cascade. In ciliary-type opsin photoreceptors such as rod and cone cells in the retina of vertebrates, phototransduction culminates in the closure of cGMP-gated channels and hyperpolarization of photoreceptor cells (Yau & Hardie, 2009). Meanwhile, light illumination of rhabdomeric opsins such as those found in arthropods leads to depolarization of photoreceptor cells. In addition to visual opsins, vertebrates encode an opsin localized in retinal ganglion cells which project to the suprachiasmatic nucleus and which contributes to photoentrainment of the circadian clock (Feuda, Hamilton, McInerney, & Pisani, 2012; Provencio, Jiang, De Grip, Hayes, & Rollag, 1998; Terakita, 2005). The pathways involved in the process of biosynthesis of visual chromophore from carotenoids and its regeneration following illumination are conserved in different phyla and have been recently reviewed (von Lintig et al., 2021).

In addition to supporting vision and photoreception, retinoids occupy an important place in the regulation of tissue differentiation and embryonic development of vertebrates. Vertebrates rely on the transcriptional functions imparted by the retinoid metabolite, all-trans-retinoic acid generated by oxidation of retinaldehyde. Retinaldehyde itself is obtained by oxidative cleavage of provitamin A carotenoids or by oxidation of retinol (Shannon, Moise, & Trainor, 2017). All-trans-retinoic acid activates the nuclear hormone receptor, retinoic acid receptor (RAR), which forms a heterodimer with the retinoid X receptor (RXR) and controls the expression of hundreds of genes involved in patterning, organ development, cell proliferation and differentiation (Ghyselinck & Duester, 2019). Thus, all-trans-retinoic acid is essential for embryonic development, immunity, cell plasticity and tissue repair of vertebrates. There is evidence that some non-vertebrate chordate species such as cephalochordates and urochordates also use retinoic acid – RAR/RXR signaling (Gutierrez-Mazariegos, Schubert, & Laudet, 2014). Emerging studies also suggest that retinoid processing genes and receptors are also encoded by other non-chordate species including lophotrochozoans, echinoderms, cnidarians, and ecdysozoans (Fonseca et al., 2019; Fuchs et al., 2014; Yamakawa, Morino, Honda, & Wada, 2018). Retinoic acid can be detected in tissues of non-chordates (several protostomes and cnidarians) and retinoic acid treatment was shown to induce changes in biological functions of these species including neuronal differentiation and morphogenesis (Albalat & Canestro, 2009; Handberg-Thorsager et al., 2018). Though these observations suggest retinoic acid may play a role in signaling and differentiation in non-chordate species, genetic and molecular evidence that retinoic acid is a relevant physiological signaling molecule in these species is still sparse. However, it is likely that a low affinity retinoic acid binding receptor was present early in metazoan evolution and that signaling by retinoic acid via a high affinity nuclear hormone receptor developed later in early chordates. Chordate RAR/RXR was further fine-tuned to convey precise positional information in the chordate lineage (Handberg-Thorsager et al., 2018). The observed effects of exogenous retinoic acid on the function and development of aquatic chordates and non-chordates alike is also an important consideration given evidence of production of significant environmental levels of retinoic acid by cyanobacterial blooms (Pipal et al., 2020; Wu, Jiang, Wan, Giesy, & Hu, 2012; Yeung, Zhou, Hilscherova, Giesy, & Leung, 2020).

Over the years, multiple reports have suggested that there are other retinoid metabolites besides 11-cis-retinaldehyde and all-trans-retinoic acid, which mediate biological functions. For example, many animals convert vitamin A1 to A2 by desaturating the ionone ring at the C3–4 double bond (Kramlinger et al., 2016). All-trans-3,4-didehydroretinol, or vitamin A2, is the precursor of the red-shifted chromophore 11-cis-3,- 4-didehydroretinaldehyde which allows fish to see under different light conditions (Enright et al., 2015; Kramlinger et al., 2016). Anhydroretinol and retro-retinoids have been detected in various animal species (Bhat et al., 1979; Buck et al., 1993; Tzimas, Collins, & Nau, 1996). Retro-retinoids were proposed to play roles in immune regulation in vertebrates (Derguini, Nakanishi, Hammerling, & Buck, 1994). Anhydroretinol, which is generated by a retinol dehydratase in Fall armyworm (Spodoptera frugiperda), plays a role in tissue morphogenesis (Pakhomova, Kobayashi, Buck, & Newcomer, 2001). In addition to anhydroretinol and retro-retinoids, dihydroretinoids which have their terminal C13–14 double bond saturated, have been detected in animal tissues and have been shown to have agonist activities against both RAR and RXR (Moise et al., 2009; Moise, Kuksa, Blaner, Baehr, & Palczewski, 2005; Moise, Kuksa, Imanishi, & Palczewski, 2004; Ruhl et al., 2015; Schmidt, Volland, Hamscher, & Nau, 2002; Schuchardt et al., 2009; Shirley et al., 1996). Some of these metabolites may be a product of the retinol saturase (RETSAT) enzyme (discussed later), or may be generated via alternate pathways. Oxidized forms of all-trans-retinoic acid have also been proposed to carry out biological functions in animals (Pijnappel et al., 1998; Zhong, Ortiz, Zelter, Nath, & Isoherranen, 2018).

In their uncleaved form, carotenoids are proposed to contribute antioxidant and photoprotective functions in animals. For example, hydroxylated carotenoids (xanthophylls) accumulate in the macula lutea and protect the highly sensitive, cone rich fovea from blue light-associated damage. The light-filtering functions of oil droplets and macular carotenoids also contribute to reduce glare and enhance chromatic contrast (Arunkumar, Gorusupudi, & Bernstein, 1865; Hammond Jr., Fletcher, & Elliott, 2013). Carotenoids also have antioxidant properties by quenching singlet oxygen and scavenge peroxyl radicals (Baswan et al., 2021; Stahl & Sies, 2002). There is, however, evidence that carotenoids can also carry out pro-oxidative effects in animal tissues in certain environments (Edge & Truscott, 2018). Finally, carotenoids accumulating in integument or feathers of fish, amphibians or birds contribute to the pigmentation displayed by animals to signal sex-specific traits, health and/or fitness (Gazda et al., 2020; Koch et al., 2019; Toomey & Corbo, 2017).

Apocarotenoids that are either longer or shorter than retinoids (C₂₀) are generated by eccentric cleavage of carotenoids. We refer the reader to the article by George Britton on the numbering and nomenclature of carotenoids and apocarotenoids (Britton, 2022). This cleavage can occur through either enzymatic or non-enzymatic means. Such apocarotenoids carry out roles in modulating nuclear receptor signaling, regulation of lipid metabolism and lipoprotein secretion, as well as filtering light (Eroglu & Harrison, 2013; Toomey & Corbo, 2017). Another product of BCO2, β-ionone, was also proposed to carry out specific roles in animals (Aloum, Alefishat, Adem, & Petroianu, 2020). In general, however, the metabolic transformations, physiological roles and signaling processes that involve non-retinoid apocarotenoids in animals are still poorly understood and await further clarifications (Harrison & Quadro, 2018).

3. Metabolism of carotenoids in animals

The metabolism of carotenoids to retinoids and other apocarotenoids in animals parallels the pathways of conversion of carotenoids in plants and other carotenogenic species. The metabolism of vitamin A to visual chromophores or to all-trans-retinoic acid in animals have been extensively explored at the genetic and biochemical level. We refer the reader to several reviews on retinoid metabolism (Ghyselinck & Duester, 2019; Sirbu, Chis, & Moise, 1865; von Lintig et al., 2021; Widjaja-Adhi & Golczak, 1865). We will focus here primarily on carotenogenic enzymes and carotenoids modifying enzymes operating in animals.

3.1. Carotenoid cleavage oxygenases

Since animals are known to use β-carotene to produce retinoids, it was suspected for a long time that an enzyme that catalyzes cleavage of the central 15–15′ bond of β-carotene to produce retinal was operating in animals (Goodman & Huang, 1965; Karrer, Helfenstein, Wehrli, & Wettstein, 1930; Olson & Hayaishi, 1965). The first member of the Carotenoid Cleavage Oxygenase (CCO) family to be identified was Viviparous 14 (VP14), a maize encoded 9′-cis-epoxycarotenoid dioxygenase 1 (NCED1) enzyme that cleaves 9-cis-violaxanthin to produce xanthoxin, the precursor of abscisic acid (Schwartz, Tan, Gage, Zeevaart, & McCarty, 1997). All carotenogenic organisms were later found to encode CCO enzymes which use a non-heme iron center to cleave carotenoids at various positions of the polyene chain generating apocarotenoids that play important signaling roles (reviewed in Ahrazem, Gómez-Gómez, Rodrigo, Avalos, & Limón, 2016). Genome surveys led to identification of an expressed sequence tag (EST) encoded by Drosophila melanogaster and which has sequence similarity to maize VP14. Cloning and heterologous expression of the cDNA of the putative Drosophila CCO in an Escherichia coli (E. coli) strain engineered to synthesize β-carotene, led to the identification of the first animal CCO. This enzyme termed β,β-carotene-15,15^’-oxygenase (BCO1) cleaves β-carotene to produce all-trans-retinaldehyde (von Lintig & Vogt, 2000) (Fig. 1). Independently, Wyss and colleagues identified an avian CCO, and similar enzymes were later found in mouse, human and zebrafish (Lampert et al., 2003; Paik et al., 2001; Redmond et al., 2001; Wyss et al., 2000; Yan et al., 2001). Genetic support for the role of BCO1 quickly followed. Mutation of the fly neither inactivation nor after potential B (ninaB), which encodes a CCO, was shown to cause chromophore deficiency and blindness in flies. Meanwhile, deficiency of fish bco1 caused patterning defects (Lampert et al., 2003), and ablation of mouse BCO1 causes impaired conversion of β-carotene to retinal (Hessel et al., 2007). Later small nucleotide polymorphisms (SNPs) were identified which affect the catalytic efficiency or expression of human BCO1 and which alter the ability of individuals that bear these mutations to utilize provitamin A carotenoids (Jlali et al., 2014; Leung et al., 2009; Lietz, Oxley, Leung, & Hesketh, 2012). In addition to carotenoids, BCO1 can also target the 15–15′ bond of longer β-apocarotenals such as β-apo-10′- and β-apo-12′-carotenoids to allow these compounds to be utilized for the biosynthesis of retinol (Amengual et al., 2013; Thomas et al., 2020) (Fig. 1). Cumulatively, these findings firmly establish the biochemical pathway by which provitamin A carotenoids are utilized by the animal kingdom to produce retinoids.

Fig. 1.

Carotenoid cleaving enzymatic activities in vertebrates. (A) BCO1 catalyzes the oxidative cleavage of the 15–15′ double bond of carotenoids and apocarotenoids containing unsubstituted β-ionone rings and is the enzyme responsible for converting provitamin A carotenoids into all-trans-retinal which can be subsequently reduced to all-trans-retinol by a short chain dehydrogenase reductase enzyme (SDR). Alternatively, BCO2 catalyzed the cleavage of the 9′−10′ or the 9–10 double bond of various carotenoids to produce β-ionone and β-apo-10′-carotenal or rosafluene, respectively. (B) Preferred BCO2 substrates include hydroxylated carotenoids (xanthophylls) and their apocarotenoid products.

Using a strategy of homology search followed by heterologous expression in carotenogenic hosts, Kiefer et al. identified a human and mouse CCO that catalyzes the eccentric cleavage of β-carotene at the 9′−10′ double bond and which is referred to as BCO2 (Kiefer et al., 2001) (Fig. 1). Proteins with similarity to BCO2 can be found in all vertebrates examined and even in non-vertebrate chordates, nematodes, and mollusks (Poliakov, Soucy, Gentleman, Rogozin, & Redmond, 2017). In comparison with cytoplasmic BCO1, BCO2 localizes to mitochondria and readily accepts carotenoid sub-strates with substituted ionone rings (Bandara et al., 2021; Lindqvist & Andersson, 2002; Thomas et al., 2020). If acting on provitamin A carotenoids that contain one substituted ionone ring, BCO2 generates β-ionone and β-apo-10′-carotenal, which can in turn be cleaved by BCO1 to produce all-trans-retinaldehyde (Kelly et al., 2018) (Fig. 1). Therefore, BCO2 by acting before BCO1 can help extend the range of potential provitamin A carotenoid substrates available from the animal diet. If acting on non-provitamin A carotenoids, such as xanthophylls, BCO2 can cleave these carotenoids at 9–10 and 9′−10′ bonds to produce hydroxy-β-ionone and diapocarotenals such as the C14-dialdehyde rosafluene (Fig. 1).

Another potential role of BCO1 and BCO2 in animals is in regulating the levels of carotenoids and apocarotenoids in animals. Deficiency of BCO2 can lead to accumulation of carotenoids in tissues and milk (Berry et al., 2009; Eriksson et al., 2008; Vage & Boman, 2010). Moreover, mice deficient in BCO2 accumulate large levels of xanthophylls and show signs of mitochondrial dysfunction and oxidative stress suggesting that BCO2 protects tissues from cytotoxic effects of xanthophylls (Amengual et al., 2011; Lobo, Isken, Hoff, Babino, & von Lintig, 2012). BCO2 is even involved in controlling the sex-specific coloration (sexual dichromatism) of birds, where sex-determined gene expression leads to changes in the levels of carotenoids displayed by male or female plumage (Gazda et al., 2020). As a possible explanation for the unusually high accumulation of lutein and zeaxanthin in the primate retina (macular pigment), it was proposed that BCO2 enzyme encoded in the primate lineage is in fact enzymatically inactive (Li, Vachali, et al., 2014). However, other studies revealed that BCO2 enzymes from human and non-human primates are enzymatically active as CCOs and optimization strategies for expression of human BCO2 in a functional form in bacteria have now been established (Babino et al., 2015; Palczewski, Amengual, Hoppel, & von Lintig, 2014; Thomas et al., 2020). Apocarotenoids are known to be potent signaling mediators in all organisms (Harrison & Quadro, 2018), so it is likely that their effects need to be tightly controlled. Therefore, another potentially important function of BCO1 and BCO2 could be to ensure that apocarotenoids obtained through diet or produced endogenously are either processed via BCO2-BCO1 to enter the well-regulated retinoid metabolic pathway (Kelly et al., 2018), or, alternatively, that excess apocarotenoids are cleaved by BCO2 at the 9–10 and 9′−10′ positions to produce β-ionone or 3-hydroxy-β-ionone and diapocarotenal (Bandara et al., 2021) (Fig. 1).

In addition to BCO1 and 2, vertebrates code for a third CCO enzyme that bears sequence homology to plant and animal CCOs including the four histidine and three glutamic acid residues required to coordinate ferrous iron. The retinal pigment epithelium-specific 65 kDa protein (RPE65) is an enzyme best known as a retinoid isomerohydrolase, a key player in the vertebrate visual cycle which is responsible for the recycling of visual chromophore by isomerizing and hydrolyzing all-trans-retinyl esters to 11-cis-retinol (Jin, Li, Moghrabi, Sun, & Travis, 2005; Kiser et al., 2015; Moiseyev, Chen, Takahashi, Wu, & Ma, 2005) (Fig. 2). Despite evidence that mutations in RPE65 in both humans and mice result in blindness and impaired chromophore production, the biochemical characterization of RPE65 was hampered by the labile nature of this protein. As in the case of BCO1 and 2, demonstration of robust isomerohydrolase activity by RPE65 required its expression in cells that also coexpressed enzymes required to generate its putative substrate, i.e., retinyl esters produced by lecithin:retinol acyltransferase (LRAT) (Jin et al., 2005; Moiseyev et al., 2005; Redmond et al., 2005). In addition to its better-known role in recycling chromophore, RPE65 was also proposed to carry out an additional function in the conversion of lutein to meso-zeaxanthin isomerase in the vertebrate eye (Shyam, Gorusupudi, Nelson, Horvath, & Bernstein, 2017). Among macular pigments, lutein and zeaxanthin are both commonly found in plants and, hence, in the animal diet, while meso-zeaxanthin is proposed to be derived from dietary lutein (Arunkumar et al., 1865; Bhosale, Serban, Zhao, & Bernstein, 2007; Johnson, Neuringer, Russell, Schalch, & Snodderly, 2005) (Fig. 2). Using both primary chicken RPE cells and human cells which heterologously express chicken RPE65, Shyam et al. found that RPE65 carries out the conversion of lutein into meso-zeaxanthin (Shyam et al., 2017) (Fig. 2). Moreover, pharmacological inhibition of RPE65 with a specific RPE65 inhibitor abrogated production of mesozeaxanthin in chick embryos which suggests that a catalytically active RPE65 enzyme is required for this biotransformation in vivo. (Shyam et al., 2017). Therefore, RPE65 carries out the isomerization of retinoid and carotenoid substrates alike. Meanwhile, insect CCO NinaB was found to combine carotenoid oxygenase and retinoid isomerase activity in a single polypeptide by directly converting carotenoid substrates into cis-retinoids (Babino et al., 2016; Chai et al., 2019; Oberhauser, Voolstra, Bangert, von Lintig, & Vogt, 2008) (Fig. 2).

Fig. 2.

Roles of CCO enzymes in the visual process. (A) CCO enzymes are responsible for the generation of the visual chromophore moiety of opsin proteins. In mammals, the CCO-related enzyme, RPE65, catalyzes the hydrolysis and isomerization of all-trans-retinyl esters to 11-cis-retinaldehyde. Meanwhile, in insects, the CCO enzyme NinaB catalyzes the combined reaction of central cleavage of the 15–15′ and isomerization of the 11–12 double bonds of zeaxanthin to produce 11-cis-3-hydroxy-retinal and all-trans-3-hydroxy-retinal. (B) RPE65 plays additional roles in vision in the metabolism of macular pigments of the primate retina. Here, RPE65 carries out the isomerization of the dietary lutein (3R, 3′R, 6′R) to produce meso-zeaxanthin (3R, 3′S), which is an enantiomer of the diet-derived zeaxanthin (3R, 3′R).

Remaining questions:

While the biochemistry and genetics of carotenoid conversion by CCOs are increasingly well-defined, several important questions remain to be answered to fully understand the pivotal role these enzymes play in metazoan carotenoid biology. Rodents such as mice express only one isoform of BCO2 that efficiently converts a large set of carotenoids and apocarotenoids as seen in Bco2-knockout mice (Amengual et al., 2011). This adaptation prevents carotenoid and apocarotenoid accumulation in tissues of this nocturnal animal. However, many other vertebrates display tissue and cell-specific accumulation of these pigments and often display them as ornaments. The level of expression of BCO2 is inversely correlated with the distribution of carotenoids. For instance, peripheral cone photoreceptors in humans express high levels of BCO2, whereas cone photoreceptors of the central retina display low levels of BCO2 gene expression (Voigt et al., 2019). Differential expression of BCO2 is also seen in tissues of birds, lizards (Andrade et al., 2019), and frogs (Rodriguez, Mundy, Ibanez, & Prohl, 2020). Therefore, the differential expression of BCO2 controls carotenoid distribution in tissues and defines the coloration and function of these pigments in the skin, feathers, and retina. Additionally, genetic analyses revealed that many vertebrates express several BCO2 isoforms that may arise from alternative splicing and/or differential expression of the Bco2 gene. It is evident then that the function of BCO2 variants and the mechanism of tissue-specific regulation of Bco2 gene activity are important areas for further research. The elucidation of factors that shape the expression pattern of BCO2 and its isoforms will greatly contribute to our understanding of carotenoid levels and patterns in different cells and tissues.

3.2. Carotenoid ketolases

Many animals make use of the vibrant color of carotenoids to communicate with each other. One of the liveliest displays presented by many birds, fish and amphibians is imparted by β-carotenoids ketolated at position 4 and/or 4′ (ketocarotenoids) such as canthaxanthin, echinenone or astaxanthin, and which are red because of an extended pi-electron system. Ketocarotenoids are also present in the oil droplets found in the retina of many birds to enhance spectral tuning (Toomey et al., 2015; Toomey, Smith, Gonzales, & McGraw, 2022). Ketocarotenoids are produced by bacteria, red yeast (Phaffia rhodozyma), and a few algae (e.g., Haematococcus pluvialis), and by plants which are not available in the diet of most animals (Cunningham Jr & Gantt, 2005). This means that with a few exceptions, animals that accumulate ketocarotenoids must transform existing dietary carotenoids to produce ketocarotenoids.

The first enzyme responsible for ketocarotenoid biosynthesis in birds was identified separately by Lopes et al. and Mundy et al. in 2016 (reviewed in Toews et al., 2017) (Fig. 3). Linkage analysis of the naturally occurring autosomal recessive yellowbeak mutation found in captive zebra finches led to the identification of enzyme CYP2J19 which converts β-carotene into canthaxanthin (Mundy et al., 2016). The same gene was also identified through analysis of introgressed regions from red siskins hybridized with common (yellow) canaries to derive “red factor” canaries (Lopes et al., 2016). Birds which operate in dim light presumably do not need oil droplets, and also lack ketocarotenoid-based plumage coloration. Interestingly, many of such bird species, including owl, kiwis and penguins have accumulated inactivating mutations in CYP2J19 (Emerling, 2018). Ketocarotenoids also accumulate in the spider mite Tetranychus kanzawai which produces them from β-carotene (Wybouw et al., 2019). Mapping of the locus responsible for the spontaneously occurring pigment mutant, Jp-lemon present in T. kanzawai, led to the identification of yet another P450 enzyme, subclass CYP384A1, which could potentially ketolate carotenoids. All Trombidiformes mites encode a putative CYP384A1 suggesting they could generate ketocarotenoids (Fig. 3).

Fig. 3.

Role of cytochrome P450 enzymes in carotenoid-based coloration displays. Ketocarotenoids play important roles in visual communication. Diagram of the conversion of β-carotene to canthaxanthin by CYP2J19 from birds (R¹=H). Alternatively, if R¹=OH, this depicts the diagram of the conversion of zeaxanthin to astaxanthin by CYP384A1 from spider mites.

Different animal and plant enzymes achieve the same goal of conversion of dietary carotenoids into ketocarotenoids. Enzymes responsible for ketocarotenoid biosynthesis in carotenogenic organisms belong to two classes of enzymes, namely, ferrodoxin-dependent nonheme diiron enzymes (HYD) (such as CrtW enzymes found in marine bacteria and green alga) and CrtO enzymes encoded by cyanobacteria such as Synechocystis and which have similarity to the bacterial phytoene desaturase CRTI (Moise et al., 2014). In contrast, animal β-carotene ketolases belong to the cytochrome P450 class of oxygenases. In addition, there are other unrelated P450 enzymes in animals which also carry out oxidation of the ionone ring. For example, CYP26A1, B1 and C1 oxidize retinoic acid to 4-oxo-, 4-hydroxy or 18-hydroxy-retinoic to block RAR signaling (MacLean et al., 2001; Tahayato, Dolle, & Petkovich, 2003; Topletz et al., 2015; White et al., 1997; Zhong et al., 2018). Meanwhile CYP27C1 catalyzes 3,4-desaturation of retinol to generate vitamin A2 (Enright et al., 2015; Kramlinger et al., 2016). P450 enzymes are involved in a multitude of biochemical transformations in animals, including xenobiotic metabolism, hormone biosynthesis and breakdown. There are also cytochrome P450 enzymes which participate in carotenoid metabolism and strigolactone bio-synthesis in plants (Moreno et al., 2021). However, ionone ring oxidation does not seem restricted to any one particular subfamily of P450 enzymes. Therefore, animal cytochrome P450 enzymes involved in oxidation of β-carotene (ketolation, hydroxylation, desaturation) appear to have arisen recurrently during evolution by having been repurposed from the existing collection of P450 enzymes present in animals (Wybouw et al., 2019).

Remaining questions:

Though genetic evidence supports a role for CYP2J19 and CYP384A1 in β-carotene ketolation, further studies need to evaluate the biochemical activity of these enzymes in utilizing various carotenoid substrates. Enzymatic oxidation of the 3-hydroxy-group in the ionone ring systems of carotenoids such as zeaxanthin and lutein has also been described in rodents, birds and humans (for review, see von Lintig, Moon, Lee, & Ramkumar, 2020). It would also be important to identify enzymes responsible for this activity in such species.

3.3. Apocarotenoid saturases

A novel metazoan enzyme was identified in 2004 through sequence similarity to tomato and cyanobacterial carotene isomerases (CRTISO). Tomato CRTISO is encoded by the tangerine locus which is associated with accumulation of prolycopene, or 7,9,9′,7′-tetra-cis-lycopene, in this strain (Clough & Pattenden, 1983; Zechmeister, Lerosen, Went, & Pauling, 1941). CRTISO was shown to carry out the isomerization of 7,9,9′,7′-tetra-cis-lycopene to all-trans-lycopene (Breitenbach, Vioque, & Sandmann, 2001; Isaacson, Ohad, Beyer, & Hirschberg, 2004; Isaacson, Ronen, Zamir, & Hirschberg, 2002; Moise et al., 2004; Park, Kreunen, Cuttriss, DellaPenna, & Pogson, 2002; Yu, Ghisla, Hirschberg, Mann, & Beyer, 2011). CRTISO is similar in sequence to CRTI from non-photosynthetic bacteria, an enzyme which catalyzes sequential desaturation of phytoene to all-trans-lycopene (Moise, von Lintig, & Palczewski, 2005). Side by side comparisons of tomato CRTISO and its mouse homologue expressed in mammalian cells and incubated with potential substrates showed that the animal homologue of CRTISO does not act on 7,9,9′,7′-tetra-cis-lycopene but instead was found to saturate double bonds in the polyene chain of all-trans-retinol (Moise et al., 2004) (Fig. 4). As a result, animal homologues of CRTISO were named retinol saturase (RETSAT). Further analysis revealed that mouse RETSAT carries out a stereospecific saturation of all-trans-retinol to produce (13R)-all-trans-13,14-dihydroretinol (Moise et al., 2008) (Fig. 4). Zebrafish RETSAT can also saturate either the C13–14 or C7–8 double bond of all-trans-retinol to produce either all-trans-13,14-dihydroretinol, or all-trans-7,8-dihydroretinol, (Moise et al., 2007) (Fig. 4). Zebrafish RETSAT also saturates vitamin A2, all-trans-3,4-didehydroretinol, to produce all-trans-13,14-dihydro-3,4-didehydroretinol, or all-trans-7,8-dihydro-3,4-didehydroretinol, respectively. Dihydroretinoids produced by RETSAT are metabolized by the same set of enzymes that operate on canonical retinoid metabolites to produce all-trans-13,14-dihydroretinoic acid and its oxidated metabolites (Moise, Kuksa, Blaner, et al., 2005). Like all-trans-retinoic acid, all-trans-13,14-dihydroretinoic acid can also act as a ligand of nuclear receptors by stimulating RAR (Moise et al., 2008, 2009; Moise, Kuksa, Blaner, et al., 2005). All-trans-13,14-dihydroretinoic acid has similar potency as all-trans-retinoic acid in stimulating cofactor recruitment by RAR in vitro using purified RARand cofactor proteins. However, all-trans-13,14-dihydroretinoic acid is 100-fold less effective than all-trans-retinoic acid in stimulating RAR in treated cells (Moise et al., 2009). The less efficient RAR agonist activity of all-trans-13,14-dihydroretinoic acid in treated cells could be due to reduced binding to the accessory protein CRABP, but there could be other factors involved.

Fig. 4.

Apocarotenoid reductases. (A) Vertebrates encode retinol saturase (RETSAT) which carries out the saturation of all-trans-retinol at the 13–14 bond of all-trans-retinol, in the case of mouse RETSAT, or at the 13–14 or 7–8 double bond of all-trans-retinol, in the case of zebrafish RETSAT. (B) Avian RETSAT also shows activity against apocarotenals, such as 10′-β-apocarotene-3,10′-diol (or galloxanthin) derived from zeaxanthin, which is saturated by avian RETSAT to 11′,12′-dihydro-10′-β-apocarotene-3,10′-diol (or dihydrogalloxanthin). This transformation plays a role in the light filtering functions of UV-sensitive cones.

RETSAT represents a highly conserved enzyme family with members found in all chordates and even more basal phyla. Closely related mouse and zebrafish RETSAT proteins exhibit an altered preference in terms of the double bond being reduced, C7–8 versus C13–14, which suggests the possibility that RETSAT accepts a wider range of apocarotenoid substrates (Moise et al., 2007). Importantly, mice lacking Retsat are not able to produce all-trans-13,–14-dihydroretinol which suggests that the activity of RETSAT in saturating all-trans-retinol is nonredundant (Moise et al., 2010). However, it is very likely that RETSAT has additional, hitherto, uncharacterized substrates. Downregulation or ablation of the Retsat genein cell culture or in in vivo models showed that RETSAT affects lipid metabolism, adipogenesis and the cellular response to peroxyl radicals (Chidawanyika, Mark, & Supattapone, 2020; Heidenreich et al., 2017; Moise et al., 2010; Nagaoka-Yasuda, Matsuo, Perkins, Limbaeck-Stokin, & Mayford, 2007; Pang, Wang, Jurczak, Shulman,& Moise, 2017; Schupp et al., 2009). Supplementation of the product of RETSAT, namely, all-trans-13,14-dihydroretinol, does not rescue the effect of ablation of Retsat in relation to inhibition of adipogenesis or oxidative stress (Pang et al., 2017; Schupp et al., 2009). Therefore, RETSAT might have additional products other than all-trans-13,14-dihydroretinol which may be involved in facilitating adipogenesis or promoting oxidative stress pathways (Pang et al., 2017; Schupp et al., 2009; Weber, Flores, Kiefer, & Schupp, 2020).

Using functional assays, Toomey et al. identified a novel substrate for avian RETSAT which saturates the C11′−12′ bond of the apocarotenoid galloxanthin (10′-apo-β-carotene-3,10′-diol) to produce 11′,12′-dihydrogalloxanthin (Toomey et al., 2016) (Fig. 4). Galloxanthin is derived from dietary zeaxanthin and is found in the C-type cone oil droplets which tune the spectrum of short-wavelength sensitive 2 (SWS2) cones of the avian retina. The spectrum of short-wavelength sensitive 1 (SWS1) cones found in ultraviolet (UV)-sensitive species is determined by a simple change in the sequence of SWS1 opsin compared to that seen in SWS1 opsin of violet-sensitive species. However, blue SWS2 cones need to accommodate the violet-UV spectral shift of SWS1 to maintain spectral separation from UV-SWS1 cones. This tuning is achieved through filtering of the incident light through C-type oil droplets. Toomey et al. found that galloxanthin with a peak absorbance at 402 nm predominates in oil droplets associated with cones of violet-sensitive species. Meanwhile, its saturated derivative 11′12′-dihydrogalloxanthin has a shorter wavelength because of one less double bond (380 nm) and predominates in oil droplets associated with SWS2 cones found in UV-sensitive species. In conclusion, the activity of avian RETSAT causes a blue shift in the spectrum of the C-type oil droplet by modulating the ratio of galloxanthin:dihydrogalloxanthin in the C-type oil droplet of SWS2 cones. Importantly, heterologous expression of avian BCO2, RDH12, and RETSAT enzymes recapitulated the 11′12′-dihydrogalloxanthin biosynthetic pathway in vitro.

Remaining questions:

Biochemical and genetic evidence demonstrates that RETSAT carries out the saturation of double bonds of retinoids and other apocarotenoids. RetSat ablation affects lipid metabolism, adipogenesis and oxidative stress but it remains to be established if any of its current dihydroapocarotenoid products or other yet-to-be discovered products play a role in these processes. The most pressing questions in relation to mammalian RETSAT involve reconciling its biochemical and physiological roles. What is the nature of the biochemical transformation carried out by RetSat which allows it to influence lipid metabolism/adipogenesis or oxidative stress? In relation to avian RETSAT, it is not clear how the expression or activity of avian RETSAT is regulated to account for the higher activity of RETSAT seen in the SWS2 cones from ultraviolet sensitive species than that seen in violet-sensitive ones. Does RetSat act on other apocarotenoids besides retinol or galloxanthin, and what are the physiological roles of such dihydroapocarotenoids in various species?

3.4. Lateral transmission of carotenogenic enzymes to animals

Carotenoids are critical for all life on Earth both through their role in photosynthesis, photoprotection in photosynthetic organisms, and as precursors for apocarotenoids which mediate vision and/or various signaling functions in most kingdoms of life. A better understanding of the selective pressures that led to the acquisition of carotenoid and apocarotenoid processing capacity by animals is important to understanding the role of carotenoids in the diversification and adaptation of animals.

There is strong evidence of vertical transmission from a common early ancestor in the acquisition of carotenoid processing enzymes by animals. CCO enzymes are found in all taxa, and animals have extended the function of CCOs to isomerization of retinoid and carotenoid substrates. Similarly, there is good evidence that chordate encoded RETSAT was also inherited from a phytoene desaturase CRTI-like bacterial ancestor. RETSAT was then presumably lost shortly before the protostome divergence, as it does not appear in extant protostomes, but it is well conserved in all chordates examined. Beta-carotene ketolases from animals do not appear to be related to enzymes carrying out the same function in algae, plants, or fungi and seem to be a result of repurposing the oxidative function of xenobiotic enzymes to generate the red pigmented ketocarotenoids.

In addition to vertical transmission and convergent evolution, there are also several fascinating examples of lateral transfer of carotenoid genes amongst fungi and animals. For example, in 2010, Moran et al. presented evidence that the red-green polymorphism of pea aphids (Acyrthosiphon pisum) is determined by the accumulation of torulene by red morphs (Moran & Jarvik, 2010). Bioinformatic analysis of the genome of other species of aphids and sequencing of several mutants of A. pisum, led to the identification of potential aphid carotenogenic genes, many present in multiple copies. The newly discovered aphid genes are related to a bifunctional phytoene synthase/carotene cyclase enzyme CarRA found in Phycomyces fungi. A potential aphid geranylgeranyl diphosphate synthase (GGPPS) that could provide substrates for phytoene biosynthesis was also identified (Ding et al., 2019). Silencing of aphid GGPPS or a putative aphid phytoene desaturase via siRNA led to reduction in aphid carotenoid levels, providing genetic support for the endogenous biosynthesis of carotenoids by aphids themselves and not by a bacterial endosymbiont (Ding et al., 2019, 2020). Final functional validation and elucidation of the carotenogenic pathway of aphids was obtained via heterologous expression of aphid cDNA in E. coli engineered to produce GGPP or phytoene. During this study, Takemura et al. also documented the production by aphids of unusual γ-ring carotenoids species derived from torulene (Takemura et al., 2021).

Genes related to the aphid carotenogenic genes were identified in the closely related adelgid and phylloxerids species suggesting a one-time horizontal gene transfer of carotenogenic genes from a fungus to a common ancestor of Sternorrhyncha followed by duplication, subfunctionalization and positive selection (Altincicek, Kovacs, & Gerardo, 2012; Cobbs, Heath, Stireman 3rd, & Abbot, 2013; Novakova & Moran, 2012; Wang, Dong, Zhang, Roberts, & Park, 2021; Zhao & Nabity, 2017). The phenomenon of horizontal gene transfer of genes encoding carotene biosynthetic enzymes is also detected in genomes of more distant gall midges (Cecidomyiidae) and in the genome of spider mites (Tetranychus urticae) (Bryon et al., 2017; Cobbs et al., 2013; Grbic et al., 2011). Inactivating mutations in spider mite phytoene desaturase cause loss of pigmentation and capacity for carotenoid biosynthesis (Bryon et al., 2017). Interestingly, the spider mite phytoene desaturase is required for diapause induction, suggesting one possible role of endogenously synthesized carotenoids in this species. Albino spider mite mutants resulting from mutations in phytoene desaturase remain white and over half cannot enter diapause despite a diet rich in plant-supplied carotenoids. This observation suggests that dietary carotenoid cannot rescue loss of endogenous carotenoid biosynthesis, and perhaps that carotenoid absorption pathways of spider mites were lost or reduced whilst having acquired endogenous capacity for carotenoid biosynthesis. Moreover, vitamin A supplementation supported diapause induction in the albino spider mite mutants (Bosse & Veerman, 1996; Veerman, 1980), which implies that perhaps one of the functions of endogenous derived carotenoids in spider mites is the generation of a retinoid species necessary for diapause induction via retinoid-mediated signaling or light perception. Interestingly, spider mites are a real powerhouse of animal carotenoid metabolism since their genome also codes for β-carotene ketolase CYP384A1, which could convert endogenous carotenoids into ketocarotenoids (Wybouw et al., 2019).

It is difficult to establish the source and timeframe of an evolutionarily singular event such as the horizontal gene transfer that allowed for carotenoid biosynthesis in some animals. There are currently numerous examples of horizontal gene transfer which allow recipient organisms to acquire important new abilities. As 20% of insect species harbor bacterial or fungal endosymbionts, such as Wolbachia pipientis (Hotopp et al., 2007), it is not surprising that horizontal gene transfer is more commonly seen in the arthropod phylum, but there are also examples of horizontal gene transfer in cnidarians, sponges and nematodes (reviewed in Boto, 2014). Examples of functional horizontal gene transfer include improved assimilation of nutrients and resistance to toxins and anti-herbivore defenses in plant feeding insects, as well increased resistance to infections (Wybouw, Pauchet, Heckel, & Leeuwen, 2016). There are important considerations in obtaining proof of functional horizontal gene transfer, which are described elsewhere (Husnik & McCutcheon, 2018), including rigorous sample preparation, valid genomic and phylogenetic evidence and foremost genetic and bio-chemical evidence of function of the acquired genes in their host. The molecular mechanisms responsible for horizontal gene transfer are also not known, but there are examples of horizontal gene transfer for which the mechanism of transfer is apparent such as in the case of an horizontal gene transfer mediated by a virus infection.

Viruses can transfer genetic information to the organisms they infect. In some cases, these genes help the virus subvert the immune response of the host they infect, but there are also instances of virus conferring special metabolic adaptations to the host to maximize viral replication. These viral-transferred genes can include those encoding photoreceptors and carotenoid biosynthetic enzymes. Choanoflagellates are Opisthokont (flagellated eukaryote) marine predators that are considered to be the closest living relatives of metazoans, having diverged from a common ancestor 600 My ago (King et al., 2008). Through single cell sorting and genome sequencing of samples surveyed in the North Pacific, Needham et al., reported the discovery of a new virus infecting choanoflagellate cells (Needham et al., 2019). The new virus named Choanovirus, belongs to the Mimiviridae clade which are nucleocytoplasmic large DNA viruses (NCLDV) infecting eukaryotic cells (Schulz et al., 2020). The 875-Kb genome of the Choanovirus also included 3 putative rhodopsin genes belonging to type-1 rhodopsin family and which exhibited light-dependent proton pump activity when expressed heterologously. In addition, Choanovirus also carry genes related to phytoene synthase, phytoene desaturase and lycopene cyclase required for biosynthesis of β-carotene. Another Choanovirus protein, which bears similarity to the microbial carotene cleavage enzyme bacteriorhodopsin-related protein-like homolog (BLH), could potentially be involved in cleavage of β-carotene to produce the retinal chromophore required for rhodopsin activity (Kim, Kim, Yeom, Kim, & Oh, 2009; Peck et al., 2000). While genes involved in carotenoid biosynthesis and cleavage from Choanavirus await functional validation, it is remarkable that Choanavirus infection could potentially allow a heterotrophic predator to produce carotenoids, convert carotenoids to chromophore, as well as sense light. Interestingly, the genomes of other extant choanoflagellates Monosiga brevicollis and Salpingoeca rosetta encode genes with similarity to animal CCOs which presents evidence of the vertical transmission of CCOs from an early eukaryotic ancestor to metazoans (Needham et al., 2019).

A more extreme example of animals taking advantage of the carotenoid biosynthetic capacity of other organisms is seen in the case of kleptoplasty. In this case, sacoglossan sea slug species, such as Elysia viridis, incorporate chloroplasts from their algal diet into their digestive tract. The chloroplasts retain their integrity and function including formation of antenna pigment carotenoids and continue to perform photosynthesis up to several months after being ingested and through this mechanism contribute to their host’s nutrition. There is also a report of an algal gene encoding plastid manganesestabilizing protein, PsbO, having been transferred to the slug genome and incorporated in the germline (Rumpho et al., 2008), but other studies have failed to corroborate this transfer (Wägele et al., 2010). A related strategy for producing carotenoids is seen in sponges which use photosynthetic autotrophic symbionts (Scott, Wetherbee, & Kraft, 1984).

4. Functional identification of carotenoid processing enzymes

4.1. Identification of novel animal carotenoid processing enzymes

The metabolic pathways that govern carotenoid metabolism in plants, fungi and bacteria are far better understood than the ones that operate in animals. Carotenogenic species are inherently known to rely on carotenoids for their basic biology and disruptions in carotenoid metabolism in such species lead to phenotypic changes that can be readily explored. Forward genetic approaches based on colorless, carotenoid devoid plants or bacteria have provided us with all the building blocks of carotenoid biosynthesis. Current developments in understanding apocarotenoid metabolism and signaling in plants have also been successful (Moreno et al., 2021).

Knowledge of the carotenoids and apocarotenoid metabolic pathways which operate in animals are still in their infancy but have benefited greatly from advances made in plant and bacterial carotenoid biology. Many animal carotenoid processing enzymes, such as, BCO1, BCO2, RETSAT and carotenogenic enzymes of aphids, midges and spider mites have been identified through sequence homology to known plant carotenoid enzymes. In these cases, plant and animal carotenoid enzymes share a common ancestor but in some cases they do not. However, even when we cannot establish a direct evolutionary relationship between animal and plant carotenoid enzymes, much can still be learned from plant carotenoid metabolism. Known modifications of (apo)carotenoids in plants and animals alike, appear to be stereotypical. We discussed enzymes which perform oxidative cleavage of the polyene chain, ring modifications (hydroxylation, ketolation), and isomerization, desaturation, or saturation of double bonds. Since there are a limited number of possible transformations of carotenoids in biology, there is a good chance that enzymes from evolutionarily distant species which perform the same reactions, still use the same chemistry of reaction and may even appear to be related through convergent evolution. There are therefore many lessons to be learned from plant and bacterial carotenoid biology which can guide our understanding of animal carotenoid metabolism.

The current post-genomic era offers unprecedented public access to hundreds of thousands of fully sequenced genomes of organisms from all kingdoms of life (Richards, 2015). Homologues of carotenoid enzymes can be readily identified in many species through simple sequence homology searches, however, even for some model organisms, genome annotation is still poor and automatic pathway annotations are often misleading. Use of the Basic Local Alignment Search Tool (BLAST) available through the National Center for Biotechnology Information (NCBI) has become routine (Altschul, Gish, Miller, Myers, & Lipman, 1990). BLAST can be used in combination with functional prediction tools such as Pfam which relies on multiple alignments and hidden Markov model-based profiles (HMM-profiles) of protein domains (Sonnhammer, Eddy, & Durbin, 1997). Conserved domains of many carotenoid enzymes are also represented in the National Library of Medicine (NLM) Conserved Domain Database (CDD) as protein families, e.g., COG3670 for Carotenoid cleavage dioxygenase and related enzymes (Lu et al., 2020). These conserved domains can be used to design a successful sequence-based homology search for gene annotation and an example of this approach in the case of the CCO encoded by N. devanaterra (NdCCD) has been described (Daruwalla et al., 2020).

In other cases, such as carotenoid enzymes of the cytochrome P450 enzymes, the sequence alone is not very informative with regards to potential substrates. Therefore, we list several ancillary strategies which could be employed to increase the likelihood of success in carotenoid gene functional annotation. Some of these strategies have been reviewed (Dunn & Munro, 2016).

(1). Genome Wide Association Studies (GWAS).

GWAS studies seek to identify genetic polymorphisms associated with a certain disease or phenotype and can be performed in both human and animal model populations that harbor sufficient genetic diversity. For example, GWAS studies have provided evidence of the role of BCO1 and BCO2 in carotenoid metabolism and health, and of the role of CYP2J19 in ketocarotenoid biosynthesis in birds (Blay et al., 2021; Ferrucci et al., 2009; Kirschel et al., 2020; Lehnert et al., 2019; Leung et al., 2009 and reviewed in Borel & Desmarchelier, 2017). There are also many examples of GWAS revealing the factors that influence carotenoid accumulation in plants (and reviewed in Borel & Desmarchelier, 2017).

(2). Genome-wide Mutagenesis, Gene Expression or RNAi Interference Screens.

Manipulation of gene expression in genome-wide studies at both whole organism or cellular models is a very powerful tool to uncover gene function by identifying genes whose loss (or gain) is associated with a certain phenotype. Gene expression can be manipulated using shRNA libraries or viral mediated overexpression. Mutagenesis screens can use chemical mutagens, transposons, or newer genome editing approaches such CRISPR-CAS9. CRISPR-CAS9 approaches can also be used to transcriptionally repress or overexpress a gene. The other advantage of CRISPR-CAS9 approaches is that evaluation of the abundance of a single-guide RNA (sgRNA) allows for rapid identification of gene perturbations which are either positively selected or negatively selected in relation to the particular phenotype (Doench, 2018; Li, Xu, et al., 2014; Shalem et al., 2014). There are examples of such studies in the carotenoid field. A forward screen based on N-ethyl-N-nitrosourea (ENU) mutagenesis of mouse embryos revealed the essential role of retinol dehydrogenase 10 (RDH10) in retinol oxidation (Sandell et al., 2007). A genome-wide RNAi screen revealed the role of RETSAT in oxidative stress (Nagaoka-Yasuda et al., 2007). The main requirement for a genome-wide mutagenesis screen is to have a robust readout consisting of a clearly distinguishable phenotype.

(3). Protein Co-evolution.

Examination of proteins that co-evolve could help identify pairs of enzymes, signaling proteins, ligand-receptor pairs and transporters that operate in the same pathway. An important caveat is that incomplete genome coverage could lead to falsely inferred gene absences (Deutekom, Vosseberg, Dam, & Snel, 2019).

(4). Gene Expression Profiling.

As in (3), proteins that interact with one another are often co-expressed. Comparisons of known tissue expression patterns of proteins could be used to identify proteins that are co-expressed. Several databases such as STRING allow for comparison of expression profiles of various genes (reviewed in Huang et al., 2018; Szklarczyk et al., 2020)

(5). Phylogenetic Comparisons.

Phylogenetic approaches as described by Nagy, Merényi, Hegedüs, and Bálint (2020) can uncover gene-phenotype association studies through surveys of gene absence/presence combined with knowledge of the carotenoid content and/or carotenoid transformations that occur in various species. This approach would require complete genomes and knowledge of the carotenoid biochemistry of species involved. Analysis of syntenic relationships between the genomes of related species could help uncover orthologs amongst various homologue genes (Vallabhaneni et al., 2009). To validate if homologous proteins have the same function, biochemical studies using purified or heterologously expressed proteins are also very important.

Phylogenetic comparisons may reveal factors that control the metabolism of carotenoids or retinoids in various animal species. Such an example can be seen in a recent study focused on carnivore and herbivore placental mam-mals. Comparisons of the genomes of 16 obligate herbivores and 15 obligate carnivores revealed genetic differences which manifested as genes consistently lost in herbivores and genes consistently lost in carnivores (Hecker, Sharma, & Hiller, 2019). Convergent gene losses in animals that share a dietary specialization may be related to selection pressures (or lack thereof ) related to their diet. Repeatedly, genes involved in pancreatic enzyme secretion or function were observed to accumulate inactivating mutations in herbivore species, since herbivores have less need for fat-digestive functions (Hecker et al., 2019). Conversely, carnivore genomes were marked by the recurrent loss of genes involved in carbohydrate metabolism and glucose regulation, as well as genes involved in the innate immune response and the metabolism and detoxification of xenobiotics. These losses also seem to tailor to diet since xenobiotic metabolism would be more useful in herbivores than carnivores. However, carnivores and herbivores also differ in relation to the dietary source of vitamin A. Herbivores rely on provitamin A carotenoids from plants, while carnivores have access to dietary preformed retinol and its esters. Examining the same data set from Hecker et al. for genes involved in carotenoid and retinoid metabolism reveals that Rbp7, which encodes a cellular retinol binding protein (CRBP-III) (Piantedosi, Ghyselinck, Blaner, & Vogel, 2005; Vogel et al., 2001), is repeatedly lost in herbivores (Dataset S2, Hecker et al., 2019). CRBP-III is required for retinoid incorporation into milk, but also plays a role in lipid metabolism (Zizola, Schwartz, & Vogel, 2008). Therefore, it is possible that the recurrent loss of CRBP-III in herbivores could be related to differences in lipid metabolism between herbivores and carnivores species, as opposed to differences in the types of dietary vitamin A precursors utilized. Another carotenoid relevant gene lost in herbivores is Stra8, a retinoic acid-inducible gene that plays a role in spermatogenesis (Ma et al., 2018). Interestingly, there were no genes with a known role in carotenoid metabolism found to be lost in carnivores. This could mean that despite carnivore reliance on preformed vitamin A, carotenoid metabolism continues to occupy an important place in carnivore metabolism. This observation is in contrast to early reports of inefficient conversion of provitamin A carotenoids by ferrets (ferrets are also included in the genomic comparison and shown to encode functional carotene cleavage enzymes, see Dataset S3 (Hecker et al., 2019).

4.2. Heterologous expression systems

Heterologous expression assays are critical for screening candidate carotenoid processing enzymes (Leonelli, 2022; Moreno & Stange, 2022). While protein purification and in vitro enzymatic studies of isolated carotenoid enzymes continues to be notoriously challenging, heterologous expression has been instrumental in defining the enzymatic activities of various carotenoid enzymes, The difficulty of analyzing carotenoid enzymes in vitro is due to both uncertainties of the exact conditions, substrate conformation and cofactors required for protein activity. In addition, carotenoid enzymes often become insoluble upon purification due to membrane domains and posttranslational processing that does not occur correctly in the bacterial host. Other than BCO1 (Wyss et al., 2000), initial assays of purified carotenoid proteins have been, at least at first, unsuccessful. A heterologous expression approach was employed in the characterization of the activity of BCO1, BCO2, RPE65, RetSat, as well as CYP27C1. Heterologous expression in bacterial hosts engineered to produce carotenoids, or in mammalian cells that synthesize all-trans-retinyl ester substrate or are exposed to potential retinoid or carotenoid substrates in the case of RPE65 was a critical step in helping perform a complete characterization of putative carotenoid enzymes. More recently, advances in performing in vitro assays using purified protein fractions of carotenoid enzymes, BCO2, BCO1 and RPE65 have provided conclusive evidence of their enzymatic activity, cofactors and catalytic mechanisms (Bandara et al., 2021; Nikolaeva, Takahashi, Moiseyev, & Ma, 2009; Thomas et al., 2020). In a few cases, crystal structures of several CCOs were solved including those of bovine RPE65 purified from native RPE membranes, and that of an archeal CCO encoded by Candidatus Nitrosotalea devanaterra (NdCCD), and which is similar to metazoan CCOs (Daruwalla et al., 2020; Kiser, Golczak, Lodowski, Chance, & Palczewski, 2009). Interesting to note, evidence presented by Daruwalla et al. suggests that NdCCD was acquired by the archaeal host through horizontal gene transfer from a bacterium (Daruwalla et al., 2020). As we discussed, there are other examples of horizontal gene transfer or viral-mediated gene transfer involved in the acquisition of carotenoid biosynthetic or processing ability by various organisms, and these instances reveal a unique perspective on functional adaptation and carotenoid metabolism in animals.

4.2.1. Heterologous expression of carotenoid cleavage enzymes

A major challenge with heterologous expression of carotenoid cleavage enzyme is to acquire the soluble and enzymatically active form of the protein. The challenging biochemistry of CCOs has previously led to conflicting results. For instance, RPE65 was proposed to be a retinoid binding protein and the existence of a CCO required for conversion of β-carotene to vitamin A was disputed in some studies. Major drawbacks in this research field were the lack of methodology in the expression of recombinant CCOs in active form. Recently, assays optimized for heterologous carotenoid enzyme expression, have overcome these obstacles and recombinant expression is critical component for comparative enzymatic and structural analyses of carotenoid enzymes. To bridge this knowledge gap, we established novel protocols to express mammalian CCOs and to test their activities in in vitro enzyme assays. We optimized the assays and recent experiments show that fusion proteins of CCOs with maltose-binding protein (MBP) result in soluble and active enzymes. The expression of CCOs starts with the cloning of cDNAs of interest into the pMAL expression vector, in the appropriate cloning sites. The coding sequence can be obtained by RT-PCR and/or through customized gene synthesis, which allows its codon usage to be optimized for heterologous expression in bacteria. Following verification of the coding sequence by Sanger sequencing, the expression plasmid is transfected into bacterial cells. The protein expression and purification protocol for maltose-binding protein fused to CCOs and heterologously expressed in E. coli is described below.

Equipment

1. A 37 °C incubator

2. A 16 °C incubator

3. Avanti JXN-26 centrifuge (Beckman) and appropriate rotor

4. A French Press

5. Ultracentrifuge and appropriate rotor

6. Vortex

7. pH meter

8. SDS-PAGE electrophoresis equipment

9. Disposable gravity columns

10. Culture flasks

11. Culture tubes

12. Centrifuge tubes

13. 50 mL Falcon tubes (Corning)

14. Microfuge tubes (1.5 mL)

15. 50 kDa molecular weight cut off centrifugal units (Amicon Ultra-Millipore).

Reagents, cells and others

1. BL21(DE3) Chemically Competent Cells (Sigma, CMC0014)

2. CCO cloned in the pMAL expression vector (New England Biolabs) as described (Bandara et al., 2021; Thomas et al., 2020)

3. Sterile Lysogeny Broth (LB) media (Fisher Bioreagents, BP1426-2) prepared in deionized water

4. Glucose (Sigma-Aldrich, G7021-1KG)

5. 100 mg/mL ampicillin (Sigma-Aldrich, A0166-25g) stock solution prepared in ultrapure water

6. 1M Isopropyl β-D-1-thiogalactopyranoside (IPTG) (Promega, V395A) stock solution prepared in ultrapure water

7. Ferrous sulfate (Fisher scientific, S93248)

8. Tricine (Sigma-Aldrich, T0377-250G)

9. 100 mM Tris(2-carboxyethyl) phosphine hydrochloride (TCEP-HCl) (Thermoscientific, 20491) stock solution prepared in ultrapure water

10. Sodium Chloride (DOT Scientific Inc., DSS23020-2500)

11. Ethylene Diamine Tetra Acetic acid (EDTA) -free protease inhibitor tablet (Roche,04 693 159 001)

12. RNase-free DNase (QIAGEN,79254)

13. Lysozyme (Roche, 10 837 059 001)

14. Maltose monohydrate (Sigma-Aldrich, M5895-500G)

15. Amylose resin (New England BioLabs, E8021S)

16. 10% Sodium dodecyl sulfate Polyacrylamide gels

Protocol CCO enzyme purification

• Preculture: 10 mL of freshly prepared LB with ampicillin (100 μg/mL) is inoculated with the CCO-pMAL plasmid which was transformed into BL21 E. coli cells. The culture is grown overnight in a 37 °C shaker (200 rpm).

• Protein expression: 1L of LB medium with 0.2% w/v glucose and ampicillin is inoculated with the preculture and grown in a 37 °C shaker (200 rpm) until an OD₆₀₀ of 0.6–0.7 is reached. Next, protein expression is induced with addition of 0.3 mM IPTG and 100 μM of ferrous sulfate (FeSO₄). The addition of ferrous sulfate supplies the iron cofactor during the protein production. Note, that the amount of inducer and ferrous sulfate may vary between different CCOs. Thus, the IPTG and FeSO₄ needs to be optimized for each recombinant enzyme. The cell culture is then incubated for 24 h at 16 °C with vigorous shaking (200 rpm). The cells are harvested by centrifugation at 4500 rpm (6000 × g) for 15 min at 4 °C. The cell pellet can be used immediately or can be stored at –80 °C until further use.

• Protein purification: The cell pellet is resuspended in 40 mL of column buffer (20 mM Tricine, 150 mM NaCl, 0.5 mM Tris(2-carboxyethyl)phosphine hydrochloride at pH 7.4) on ice. Then 20 μL of lysozyme (10 mg/mL), 4 μL of DNAse (3 Kunitz units per μL) and one EDTA-free protease inhibitor tablet are added and the solution is incubated for 30 min on ice. Next, a French press (Avestin emulsiflex) is used for cell lysis according to the manufacturer’s protocol. Alternatively, sonication can be used for cell lysis. After cell lysis, the homogenate is centrifuged at 125,440 × g at 4 °C for 1 h to separate soluble proteins from insoluble debris and membranes. This step is critical because it separates the soluble, enzymatically active protein from fractions that contain insoluble and improperly folded protein. The soluble protein extract is then added to the disposable gravity column contains 2 mL of amylose resin which was previously equilibrated with 5 column volumes of column buffer. Then, the bound fraction is washed with 10 column volumes of column buffer to remove unbound proteins from the resin. Finally, MBP fused carotenoid enzyme is eluted with the elution buffer (20 mM Tricine, 150 mM NaCl, 0.5 mM TCEP-HCl, 10 mM maltose at pH 7.4). SDS-PAGE is used to analyze the purity of the protein and pure protein fractions are combined and concentrated using 50 kDa molecular weight cut-off Amicon Ultra centrifugal units.

Expected results

Though this protocol has been successfully used in recent studies of BCO2 (Bandara et al., 2021; Thomas et al., 2020), it is difficult to predict whether a novel CCO of interest will express in soluble form as a fusion protein under the same conditions. In terms of protein purification, we use affinity chromatography as an initial purification step for a given CCO. If the purity is not sufficient, anion exchange or size exclusion chromatography methods can be used to further purify the proteins (Duong-Ly & Gabelli, 2014a, 2014b).

4.2.2. Enzyme activity assay

The delivery of lipophilic carotenoid substrate to the purified recombinant protein is a major challenge in the in vitro CCO enzyme activity assays. Therefore, carotenoid detergent micelles are utilized to supply carotenoid substrate to the enzyme. We tested many different kinds of detergents, and the proper choice of detergent is critical for the success of the assays. In our current protocols, we use n-octyl-β-D-thioglucopyranoside (OTG), n-dodecyl-β-D-maltopyranoside (DDM) and 3-[(3-cholamidopropyl)dim-ethylammonio]-1-propanesulfonate (CHAPS) for enzyme assays with β-carotene and BCO1. We observed that long linear detergents such as tetraethylene glycol mono-octyl ether (C8E4) and hexaethylene glycol mono-octyl ether (C8E6) tend to inhibit the enzymatic activity of CCOs. For delivery of xanthophyll substrates to BCO2, we use n-dodecyl-β-D-maltopyranoside and decyl maltose neopentyl glycol (DMN). The detergents are used at concentrations equivalent to their respective critical micellar concentrations. Note that each detergent has a specific capacity to solubilize carotenoids and “overloading” of the micelles with carotenoid substrate impedes substrate availability. We conducted enzyme tests with apocarotenoids delivered as an ethanolic solution to a 0.1% (w/v) Triton-X-100 assay buffer. The final concentration of ethanol in the assay should not exceed 5% v/volume. Note that carotenoids are not converted by the BCO1 and BCO2 enzymes when Triton-X-100 is used to incorporate carotenoids in micelles (Bandara et al., 2021). Finally, great effort should be made to use chromatographically purified carotenoid and apocarotenoid substrates. Commercially available carotenoid and apocarotenoid preparations contain impurities that can confound assays of enzymatic activity leading to misinter-pretation of resulting products and misattribution of enzymatic activities. Therefore, we purify substrates by thin layer chromatography (TLC) and/or HPLC columns (Anwar, Nayak, Alagoz, Wojtalewicz, & Cazzonelli, 2022). A step-by-step protocol for assaying recombinant BCO2 and a carotenoid substrate is outlined next.

Equipment

1. Thermomixer (Eppendorf )

2. Vortex

3. Vacuum centrifuge (Eppendorf )

4. HPLC-Agilent 1260 Infinity Quaternary HPLC system (Santa Clara, CA, USA) equipped with a pump (G1312C) with an integrated degasser (G1322A), a thermostatic column compartment (G1316A), an auto sampler (G1329B), a diode-array detector (G1315D), and online analysis software (Chemstation)

5. Eppendorf tabletop centrifuge

6. Speedvac such as Eppendorf Concentrator 5301 equipped with a cold trap for solvents.

Reagents

1. Carotenoid

2. Acetone (Fisher chemical, A18-4)

3. Decyl maltose neopentyl glycol (DMN) (Anatrace, NG322 1 GM)

4. Ethanol (200 proof ) (KOPTEC, V1001)

5. Diethyl ether (Fisher chemical, E138-4)

6. Petroleum ether (Fisher scientific, E139-4)

7. Hexane (Fisher chemical, HPLC grade, H302-4)

8. Ethyl acetate (Fisher chemical, HPLC grade, E195-4)

Protocol for in vitro CCO assay

3.1. Carotenoid detergent mixture preparation:

Carotenoids (2000 pmol) are dissolved in acetone and mixed with 3% w/v DMN in 100% ethanol. The solvent is evaporated by using a refrigerated vacuum concentrator (Speedvac). The carotenoid detergent mixture is dissolved into 50 μL of reaction buffer (20 mM Tricine, 150 mM NaCl, 0.5 mM TCEP at pH 7.4) through vigorous vortexing.

3.2. Enzyme activity assay:

50 μL of concentrated enzyme solution (approx. 50 μg of recombinant BCO2 enzyme with more than 80% purity) is added to the carotenoid detergent mixture and mixed by vortexing. The assay is incubated at 37 °C under shaking at 600 rpm in a thermomixer for 10 min. Reactions are stopped by the addition of 100 μL of water, 400 μL of acetone, 400 μL of diethyl ether, and 100 μL of petroleum ether. The top organic layer is collected and evaporated under vacuum (Speedvac) or under Nitrogen stream.

3.3. HPLC analysis:

The dried reaction (step 3.2) is dissolved in HPLC solvent (hexane:ethyl acetate, 70:30 or 90:10 (v/v)) and subjected to normal phase isocratic conditions for HPLC analysis using a silica HPLC column (Zorbax Sil (5 μm, 4.6 × 150 mm) (Agilent Technologies, Santa Clara, CA). The column is protected with a guard column with the same stationary phase and the sample is eluted using a flow rate of 1.4 mL per minute. Elution of products and substrates can be monitored at the expected maxima using a diode-array detector. Peak identification can be performed by comparison to elution profiles and spectra of authentic standards and/or by LC/MS-MS analysis.

Expected results

An effective enzyme assay should lead to a turnover of the substrate in a few minutes. The turnover of the carotenoid can be followed by the naked eyes through a color shift of the assay solution from yellow to colorless. As shown in the case of BCO1 (Kowatz, Babino, Kiser, Palczewski, & von Lintig, 2013) and BCO2 (Kelly et al., 2018), mammalian CCOs display Km values in the lower μM range and turnover rates of a few molecules per second.