**3. Ketocarotenoid Biosynthesis Genes**

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 ·*-Proteobacteria* (the same family to *Escherichia coli*). The *Pantoea* carotenoid biosynthesis genes composed a gene cluster for the synthesis of zeaxanthin Ά-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 (Μ,Μcarotene) Ά-cyclase (CrtY) [23,31], Ά-carotene (Ά,Ά-carotene) 3-hydrocylase (CrtZ) [23], and zeaxanthin glucosyltransferase (CrtX) [23,32] (Figure 3). 

**Figure 2.** Phylogenetic positions of *Paracoccus* sp., *Erythrobacer* sp., and *Brevundimonas* sp. strains deduced from their 16S rRNA sequences. ż represents marine bacteria. Bacterial strains, whose carotenoid biosynthesis genes were elucidated, are shown in boldface, and the second accession numbers in the parentheses shows those of carotenoid biosynthesis genes. *Paracoccus* sp. strain N81106 (MBIC01143 = NBRC 101723) and *Paracoccus* sp. strain PC1 (MBIC03024 = NBRC 101025) were formerly classified as *Agrobacterium aurantiacum* [17] and *Alacaligenes* sp. PC-1 [33], respectively. The phylogenetic tree was constructed as described [10]. The scale bar indicates a genetic distance of 0.02 (*Knuc*). 

**Figure 3.** Pathway engineering for the production of functional xanthophylls using the carotenoid biosynthesis genes, *crtW*, *crtZ*, and/or *crtG*, which were isolated from the marine bacteria, *Paracoccus* sp. strain N81106 or *Brevundimonas* sp. strain SD212, in addition to the *crtE*, *crtB*, *crtI*, and *crtY* genes (and *crtX*) from *P. ananatis*. 

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 Άcarotene 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. 

## **4. Carotenoid 4,4ȝ-Ketolase**

It has been revealed that only two enzymes, carotenoid 4,4ȝ-ketolase (4,4<sup>ȝ</sup>oxygenase) (Ά-ring 4(4ȝ)-ketolase; CrtW) and carotenoid 3,3ȝ-hydroxylase (Ά-ring 3(3<sup>ȝ</sup>)-hydroxylase; CrtZ), are sufficient to biosynthesize astaxanthin from Ά-carotene via eight intermediates including zeaxanthin, canthaxanthin and adonixanthin [35,40,41]. CrtW can convert not only the (un-substituted) Ά ring but also the 3- hydroxylated Ά ring into the respective 4-ketolated groups, and CrtZ can convert not only the (un-substituted) Ά ring but also the 4-ketolated Ά-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 (΅-ketoglutarate)- dependent dioxygenases [42]. 

**Figure 4.** Catalytic functions of carotenoid 4,4ȝ-ketolases (oxygenases) and carotenoid 3<sup>ȝ</sup>3<sup>ȝ</sup>-hydroxylases. BKT means BKT1 or BKT2 from *H. pluvialis*. 

The *crtW* genes were present not only in the above-mentioned ΅*-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 (Ά,Ά-caroten4-one), and 4-ketomyxol 2ȝ-fucoside, a monocyclic carotenoid that includes the 4- ketolated Ά-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, *Brevundimona*s sp. SD212, *Anabaena* sp. PCC7120, and *N. punctiforme* was expressed [44,46]. It was consequently shown that the *Brevundimona*s 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 Ά-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 Ά-ring 4(4ȝ)-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 Ά-ring 4(4ȝ)-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 3ȝ-hydroxyechinenone (3ȝ-hydroxy-Ά,Ά-caroten-4-one), zeaxanthin and myxol 2ȝ-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 · and UV radiation [56], which produced other monocyclic carotenoids, e.g., the latter strain produced deinoxanthin (2,1<sup>ȝ</sup>dihydroxy-3<sup>ȝ</sup>,4<sup>ȝ</sup>-didehydro-1<sup>ȝ</sup>2<sup>ȝ</sup>-dihydro-Ά,Μ-caroten-4-one) [1]. An *in vivo* analysis on *crtO* was performed with recombinant *E. coli* cells that synthesize the carotenoid substrate Ά-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 Ά-ring 4(4ȝ)- ketolases can accept only the (un-substituted) Ά ring(s) in Ά-carotene and probably in monocyclic carotenoids as the substrates (Figure 4). 
