Many bacteria that have been isolated from marine environments can synthesize a variety of carotenoid pigments [1]. For example, acyclic C₃₀-type carotenoic acids were identified in some marine bacteria such as Planococcus maritimus [2] and Rubritalea squalenifaciens [3]. Algoriphagus sp. KK10202C of the Flexibacteraceae family, which was isolated from a marine sponge, was found to produce flexixanthin ((3S)-3,1′-dihydroxy-3′,4′-didehydro-1′2′-dihydro-β,ψ-caroten-4-one) and deoxyflexixanthin (1′-hydroxy-3′,4′-didehydro-1′2′-dihydro-β,ψ-caroten-4-one) [4], which are C₄₀-type monocyclic carotenoids containing one β-end group (β ring) (called monocyclic carotenoids in this review). Other marine bacteria including strain P99-3, which belong to the Flavobacteriaceae family, were shown to produce monocyclic carotenoids, myxol ((3R,2′S)-3′,4′-didehydro-1′,2′-dihydro-β,ψ-carotene-3,1′,2′-triol) and saproxanthin ((3R)-3′,4′-didehydro-1′,2′-dihydro-β,ψ-carotene-3,1′-diol), and zeaxanthin ((3R,3′R)-β,β-carotene-3,3′-diol) [5,6], which are a C₄₀-type dicyclic carotenoid containing two β-end groups (called dicyclic carotenids in this review). Marine bacteria belonging to genus Paracoccus, Brevundimonas or Erythrobacter in the α-Proteobacteria class have been revealed to synthesize dicyclic carotenoids that are ketolated at the 4(4′)-position(s) (called ketocarotenoids), e.g., astaxanthin ((3S,3′S)-3,3′-dihydroxy-β,β-carotene-4,4′-dione) and adonixanthin ((3S,3′R)-3,3′-dyhydroxy-β,β-caroten-4-one) (Figure 1) [7–9].

Among ketocarotenoids, astaxanthin and canthaxanthin (β,β-carotene-4,4′-dione) (specifically the former), are commercially important pigments as nutraceuticals and cosmetics that have anti-oxidation and anti-aging effects as well as colorants in aquaculture, while other ketocarotenoids are likely to have industrial potentials [12–16]. This review focuses on carotenoid β-ring 4(4′)-ketolase (4-oxygenase), carotenoid β-ring 3(3′)-hydroxylase, and carotenoid β-ring 2(2′)-hydroxylase, derived from the marine bacteria that belong to α-Proteobacteria, and pathway engineering for the production of functional xanthophylls via the incorporation of these β-ring-modifying enzyme genes.

Paracoccus sp. strain N81106 (NBRC 101723), isolated from surface seawater near Aka island, Okinawa, Japan, was first shown to produce astaxanthin in bacteria [7,17]. This bacterium was also found to synthesize adonixanthin, adonixanthin β-d-glucoside, and astaxanthin β-d-glucoside [7,18]. Paracoccus haeundaesis BC74171^T, isolated from the Haeundae Coast, Korea, was shown to produce astaxanthin mainly [19]. Paracoccus marinus KKL-A5^T (NBRC 100637^T), isolated from coastal seawater in Tokyo Bay, Japan, was found to produce adonixanthin diglucoside predominantly [20,21]. On the other hands, a marine bacterium Brevundimonas sp. strain SD212 (NBRC 101024) was revealed to synthesize not only astaxanthin and adonixanthin but also their 2(2′)-hydroxylated metabolites, that is, 2-hydroxyastaxanthin ((2R,3S,3′S)-2,3,3′-trihydroxy-β,β-carotene-4,4′-dione), 2-hydroxyadonixanthin ((2R,3S,3′R)-2,3,3′-trihydroxy-β,β-caroten-4-one), erythroxanthin ((3S,2′R,3′R)-3,2′,3′-trihydroxy-β,β-caroten-4-one), 4-ketonostoxanthin ((2R,3S,2′R,3′R)-2,3,2′,3′-tetrahydroxy-β,β-caroten-4-one) and 2,2′-dyhydroxyastaxanthin ((2R,3S,2′R,3′S)-2,3,2′,3′-tetrahydroxy-β,β-carotene-4,4′-dione) [9]. Figure 1 shows the structures of the ketocarotenoids shown in this section and their feasible biosynthetic pathway. Figure 2 shows phylogenetic tree of the marine bacteria that produce astaxanthin and other ketocarotenoids, which were isolated in Marine biotechnology Institute (Kamaishi, Japan), along with the type strains relative to these bacteria, many of which are not marine bacteria but soil bacteria. Interestingly, the phylogenetically closest strains to the marine bacteria Paracoccus sp. N81106 and Brevundimonas sp. SD212 were soil bacteria Paracoccus marcusii DSM 11574^T [22] and Brevundimonas aurantiaca ATCC 15266^T, respectively (Figure 2).

Genes required for the biosynthesis of dicyclic carotenoids were first isolated from soil bacteria Pantoea ananatis (reclassified from Erwinia uredovora) [23] and Pantoea agglomerans (Erwinia herbicola) [24,25], which cannot produce ketocarotenoids and belong to the Enterobacteriaceae family of class γ-Proteobacteria (the same family to Escherichia coli). The Pantoea carotenoid biosynthesis genes composed a gene cluster for the synthesis of zeaxanthin β-d-diglucoside from farnesyl diphosphate (farnesyl pyrophosphate; FPP) [25–27], and comprised six genes that encode geranygeranyl diphosphate (GGPP) synthase (CrtE) [27,28], phytoene synthase (CrtB) [27,29], phytoene desaturase (CrtI) [23,30], lycopene (ψ,ψ-carotene) β-cyclase (CrtY) [23,31], β-carotene (β,β-carotene) 3-hydrocylase (CrtZ) [23], and zeaxanthin glucosyltransferase (CrtX) [23,32] (Figure 3).

These crt genes have widely used for complementation analysis of carotenoid biosynthesis genes isolated from other organisms, since they are functionally expressed in E. coli with ease [11,34–37]. The P. agglomerans gene cluster contained a gene encoding isopentenyl diphosphate (IPP) isomerase (Idi; type 2) [38] in addition to the six crt genes [39]. These seven carotenogenic (carotenoid-biosynthetic) genes were also found to exist in a carotenoid biosynthesis gene cluster of Paracoccus sp. strain N81106 [10,39]. This cluster included an additional gene, designated CrtW, which was elucidated to code for an enzyme responsible for ketocarotenoid formation, that is, CrtW proved to catalyze the synthesis of canthaxanthin from β-carotene by complementation analysis using recombinant E. coli cells that contains the P. ananatis crtE, crtB, crtI, and crtY genes [33] (Figure 3). The hydropathy and transmembrane prediction analyses indicated that CrtW from Paracoccus sp. N81106 contains four transmembrane domains and two other hydrophobic regions, and its topology model is very similar to those for fatty acid desaturases [40]. It should be noted that it is recalcitrant to purify active CrtW and CrtZ proteins, which both are very likely iron-dependent integral membrane proteins, from the recombinant hosts as well as the native hosts, precluding their close enzymatic characterizations.

It has been revealed that only two enzymes, carotenoid 4,4′-ketolase (4,4′-oxygenase) (β-ring 4(4′)-ketolase; CrtW) and carotenoid 3,3′-hydroxylase (β-ring 3(3′)-hydroxylase; CrtZ), are sufficient to biosynthesize astaxanthin from β-carotene via eight intermediates including zeaxanthin, canthaxanthin and adonixanthin [35,40,41]. CrtW can convert not only the (un-substituted) β ring but also the 3-hydroxylated β ring into the respective 4-ketolated groups, and CrtZ can convert not only the (un-substituted) β ring but also the 4-ketolated β-ring into the respective 3-hydroxylated groups, as shown in Figure 4 [42–46]. An in vitro analysis with the crude enzymes of CrtW and CrtZ from the E. coli cells expressing the corresponding genes indicated that these enzymes are likely 2-oxoglutarate (α-ketoglutarate)-dependent dioxygenases [42].

The crtW genes were present not only in the above-mentioned α-Proteobacteria (Figure 2) but also in the marine bacterium Algoriphagus sp. KK10202C [4] and cyanobacterial strains such as Anabaena (Nostoc) sp. PCC 7120 and N. punctiforme [47,48]. These cyanobacteria produced not astaxanthin but echinenone (β,β-caroten-4-one), and 4-ketomyxol 2′-fucoside, a monocyclic carotenoid that includes the 4-ketolated β-ring [49]. Conversion efficiency to astaxanthin in several CrtWs was compared with recombinant E. coli cells that synthesize the carotenoid substrate zeaxanthin due to the presence of the P. ananatis crtE, crtB, crtI, crtY, and crtZ genes, in which each crtW gene from Paracoccus sp. N81106, Paracoccus sp. PC1, Brevundimonas sp. SD212, Anabaena sp. PCC7120, and N. punctiforme was expressed [44,46]. It was consequently shown that the Brevundimonas sp. SD212 CrtW, which exhibited the highest amino acid identity (96.3%) with that of the B. aurantiaca ATCC 15266 CrtW (accession no. AY166610), converted β-carotene to astaxanthin with the highest efficiency, along with the P. ananatis CrtZ [44,46]. In the case of the Paracoccus CrtWs, not only astaxanthin but also adonixanthin tended to accumulate, and this intermediate was difficult to be converted to astaxanthin [43,44]. The cyanobacterial CrtWs poorly converted zeaxanthin to astaxanthin via adonixanthin [46].

Two paralogous genes exhibiting significant homology to crtW were isolated from H. pluvialis, and designated bkt [50] or crtO [51]. These genes were renamed bkt1 from crtO and bkt2 from bkt, since “crtO” has been used for the other type of cyanobacterial β-ring 4(4′)-ketolase genes, as shown later [52]. The BKT1 and BKT2 enzymes are very likely to have catalytic function same to the Paracoccus (or Brevundimonas) CrtWs, considering results from the in vitro study on BKT2 with E. coli [42] and pathway engineering researches in higher plants as well as E. coli as the hosts [16,50,51,53].

A gene encoding a new type of β-ring 4(4′)-ketolase (named CrtO) that showed apparent homology not to CrtW-type ketolase but to CrtI-type phytoene desaturase was first found in cyanobacterium Synechocystis sp. strain PCC 6803 [54], which produced 3′-hydroxyechinenone (3′-hydroxy-β,β-caroten-4-one), zeaxanthin and myxol 2′-dimethyl-fucoside [55]. The crtO genes were also present in Anabaena sp. PCC 7120 [48], and an actinomycete Rhodococcus erythropolis and Deinococcus radiodurans R1 highly resistant to γ and UV radiation [56], which produced other monocyclic carotenoids, e.g., the latter strain produced deinoxanthin (2,1′-dihydroxy-3′,4′-didehydro-1′2′-dihydro-β,ψ-caroten-4-one) [1]. An in vivo analysis on crtO was performed with recombinant E. coli cells that synthesize the carotenoid substrate β-carotene or zeaxanthin, into which each crtO gene from Synechocystis sp. PCC 6803 and R. erythropolis was introduced and expressed there [57]. This result along with previous finding [48] suggested that the CrtO-type of β-ring 4(4′)-ketolases can accept only the (un-substituted) β ring(s) in β-carotene and probably in monocyclic carotenoids as the substrates (Figure 4).

The crtZ genes have been found not only in carotenogenic bacteria belonging to genera Pantoea, Paracoccus and Brevundimonas, but also in those belonging to the Flavobacteriaceae family [6,39]. Conversion efficiency to astaxanthin in several CrtZs was compared with recombinant E. coli cells that synthesize the carotenoid substrate canthaxanthin due to the presence of the P. ananatis crtE, crtB, crtI and crtY gens, and the Paracoccus N81106 crtW gene, into which each crtZ gene from P. ananatis, Paracoccus sp. N81106, Paracoccus sp. PC1, Brevundimonas sp. SD212, and marine bacterium strain P99-3 of the Flavobacteriacea family was introduced and expressed there [45]. It was consequently shown that the CrtZ enzymes from Brevundimonas sp. SD212 and the bacterial strain P99-3 converted β-carotene to astaxanthin with the highest and lowest efficiency, respectively, along with the Paracoccus N81106 CrtW [45].

On the other hands, no crtZ sequences have not been found in cyanobacteria, instead genes encoding a new type of β-ring 3(3′)-hydroxylases (named CrtR) that exhibited moderate homology to CrtW have been found there [58,59]. The crtR genes were isolated from Synechocystis sp. strain PCC 6803, Anabaena sp. PCC 7120, Anabaena variabilis, and N. punctiforme [46,58]. An in vivo analysis on crtR was performed with recombinant E. coli cells that synthesize the carotenoid substrate β-carotene or canthaxanthin, into which each crtR gene from Synechocystis sp. PCC 6803, Anabaena sp. PCC 7120, and A. variabilis was introduced and expressed there [46]. This result along with another result [60] indicated that the CrtR-type enzymes can hydroxylate the (un-substituted) β ring of monocyclic carotenoids such as deoxymyxol and deoxymyxol 2′-fucoside at the 3 position (Figure 4). Among them, only the Synechocystis sp. PCC 6803CrtR was able to convert β-carotene to zeaxanthin [46,58,60]. A thermophilic bacterium Thermus thermophilus HB27, which grows at temperatures above 75 °C, was found to possess another new type of β-ring 3(3′)-hydroxylase of the cytochrome P450 superfamily, named CYP175A1 [61]. The in vivo analysis with the gene strongly suggested that this thermostable P450 accepts only the (un-substituted) β ring of β-carotene as the substrate to form zeaxanthin [45,61].

Carotenoid 2,2′-hydroxylase (β-ring 2(2′)-hydroxylase) was first found in the marine bacterium Brevundimonas sp. strain SD212, and named CrtG [11]. An in vivo analysis on crtG was performed with recombinant E. coli cells that synthesize each carotenoid substrate (β-carotene, zeaxanthin, canthaxanthin, or astaxanthin), into which the crtG gene was introduced and expressed there [11]. The result indicated that the CrtG can hydroxylate the β rings substituted with 3-hydroxy and/or 4-keto groups in dicyclic carotenoids at the 2(2′)-positions (Figures 1 and 3) [11]. The crtG genes were also isolated from soil bacteria Brevundimonas vesicularis DC263 and B. aurantiaca ATCC 15266 [62]. The in vivo analysis with these genes indicated that the B. aurantiaca CrtG enzyme (accession no. DQ497427), which exhibited the highest amino acid identity (98.8%) to that of the Brevundimonas SD212 CrtG, accepted the (un-substituted) β rings of β-carotene in addition to the substituted β rings as the substrates [62]. A crtG gene sequence, whose encoded amino acid sequence was 41% identical to the Brevundimonas sp. SD212 CrtG, was found in a thermophilic cyanobacterium Thermosynechococcus elongatus, which synthesized 2-hydroxylated carotenoids such as caloxanthin ((2R,3R,3′R)-β,β-carotene-2,3,3′-triol), nostoxanthin ((2R,3R,2′R,3′R)-β,β-carotene-2,3,2′,3′-tetrol) (Figure 3), and 2-hydroxymyxol 2′-fucoside [63].

Figure 3 shows xanthophylls that were produced in recombinant E. coli cells via the incorporation of the marine bacterial crtW, crtZ, and/or crtG genes along with the Pantoea crtE, crtB, crtI, and crtY genes. The recombinant E. coli strain that expresses the four Pantoea crt genes can produce β-carotene predominantly (approximately 0.2–1 mg·g⁻¹ dry cell weight). The coexpression of the crtW, crtZ, and/or crtG genes in the β-carotene-synthesizing E. coli cells confer the ability to produce not only commercially important xanthophylls such as astaxanthin but also xanthophylls minor in nature (e.g., β-ring(s)-2(2′)-hydroxylated carotenoids), which are difficult to synthesize chemically. Particularly, the chemical synthesis of 2(2′)-hydroxycarotenoids are likely to be recalcitrant, due to high-density around the 1,2-positions of the β ring in these xanthophylls. We showed that the coexpression of the Brevundimonas sp. SD212 crtW gene and the P. ananatis crtZ gene in the β-carotene-synthesizing E. coli due to the presence of the four crt genes of P. ananatis resulted in predominant production of astaxanthin [44,46]. The Paracoccus sp. N81106 crtW gene was evolved by random mutagenesis to have improved activity [40]. It is also demonstrated that the coexpression of the crtW gene and the crtG gene from Brevundimonas sp. SD212 or from B. aurantiaca ATCC 15266 in the β-carotene-synthesizing E. coli resulted in dominant production of 2,2′-dihydroxycanthaxanthin and 2-hydroxycanthaxanthin, while the substrate canthaxanthin accumulated [11,62]. The coexpression of the crtZ gene and the crtG gene in the β-carotene-synthesizing E. coli resulted in predominant production of nostoxanthin along with small amounts of caloxanthin [11,62]. The coexpression of all the three genes (crtW, crtZ, and crtG) in the β-carotene-synthesizing E. coli resulted in dominant production of 2,2′-dihydroxyastaxanthin and 2-hydroxyastaxanthin [11]. When the P. ananatis crtX gene was coexpressed in addition to appropriate combinations of the above crt genes in E. coli, resultant E. coli cells were able to synthesize carotenoid-glycosides such as caloxanthin β-d-glucoside [64] and astaxanthin β-d-diglucoside [65], as shown in Figure 3.

The γ-ray-tolerant bacterium D. radiodurans R1 produces the monocyclic carotenoid including the 2-hydroxy-4-keto-β-ring, deinoxanthin [1]. 2,2′-Dihydroxycanthaxanthin was shown to have strong inhibitory effect against lipid peroxidation in a rat brain homogenate [11]. Such minor ketocarotenoids, which include the 2-hydroxy-4-keto-β-ring, may have beneficial effects on human health as well as anti-oxidation function, while few works are present examining their biological functions.

When carotenoid biosynthesis genes starting from the utilization of FPP are introduced in E. coli, as above-mentioned, amounts of carotenoids produced with the recombinant E. coli cells are far from the practical use, which was difficult to exceed 1 mg·g⁻¹ dry weight. In order to overcome this problem, many pathway engineering researches in E. coli have been performed for increasing intracellular concentration of FPP (e.g., recently reviewed [66,67]). For example, the coexpression of the idi (type 1) gene from H. pluvialis, Xanthophyllomyces dendrorhous (renamed from Phaffia rhodozyma), or Saccharomyces cerevisiae, as well as the idi (type 2) from Streptomyces sp. strain CL190, was shown to be effective to increase FPP content [68,69]. The introduction of heterologous mevalonate pathway genes in E. coli along with an idi (type 2) gene has been described to efficiently improve the productivity of carotenoids or sesquiterpenes that are synthesized from FPP [69–73]. For example, Yoon et al. [73] produced 22 mg·g⁻¹ dry cell weight of lycopene in 72 h using such mevalonate-pathway-engineered E. coli cells. On the other hand, production of lycopene reached high levels (near to 20 mg·g⁻¹ dry cell weight) in 24-h batch flask culture in pathway-engineered E. coli, which reflected results of multi-dimensional gene target search or gene-knockout analysis [74]. These finding should be applied to efficient production of the above-mentioned functional xanthophylls with E. coli cells.

Pathway engineering researches in higher plants have also been performed for efficient production of astaxanthin, which utilized the marine bacterial crtW genes from Paracoccus sp. N81106 or Brevundimonas sp. SD212, or the H. pluvialis bkt1 or bkt2 genes, as reviewed [16,39,53]. For example, the Brevundimonas sp. SD212 crtW and crtZ genes, whose nucleotide sequence is modified to codon usage of higher plants, were successfully overexpressed in the chloroplasts of tobacco plants (Nicotiana tabacum), and astaxanthin level produce there reached 5.44 mg·g⁻¹ dry weight (74% of total carotenoids) [75].

This review has focused on the carotenoid β-ring-modifying enzymes, CrtW, CrtZ and CrtG, derived from the marine bacteria of the α-Proteobacteria class, and pathway engineering for the production of xanthophylls in E. coli, using these enzyme genes. Such pathway engineering researches are performed towards efficient production not only of commercially important xanthophylls such as astaxanthin, but also of xanthophylls minor in nature, which are difficult to synthesize chemically, and expected to have beneficial effects on human health as well as anti-oxidation function.