Combinatorial Biosynthesis of Novel Multi-Hydroxy Carotenoids in the Red Yeast Xanthophyllomyces dendrorhous

The red yeast Xanthophyllomyces dendrorhous is an established platform for the synthesis of carotenoids. It was used for the generation of novel multi oxygenated carotenoid structures. This was achieved by a combinatorial approach starting with the selection of a β-carotene accumulating mutant, stepwise pathway engineering by integration of three microbial genes into the genome and finally the chemical reduction of the resulting 4,4’-diketo-nostoxanthin (2,3,2’,3’-tetrahydroxy-4,4’-diketo-β-carotene) and 4-keto-nostoxanthin (2,3,2’,3’-tetrahydroxy-4-monoketo-β-carotene). Both keto carotenoids and the resulting 4,4’-dihydroxy-nostoxanthin (2,3,4,2’,3’,4’-hexahydroxy-β-carotene) and 4-hydroxy-nostoxanthin (2,3,4,2’3’-pentahydroxy-β-carotene) were separated by high-performance liquid chromatography (HPLC) and analyzed by mass spectrometry. Their molecular masses and fragmentation patterns allowed the unequivocal identification of all four carotenoids.


Introduction
Xanthophyllomyces dendrorhous (with Phaffia rhodozyma as its anamorphic state) is a basidiomycetous red yeast that accumulates astaxanthin [1], which is a unique feature among fungi [2]. This carotenoid is formed via the mevalonate pathway starting with a condensation of two molecules of geranylgeranyl pyrophosphate, a 4-step desaturation, cyclization and a final 4-ketolation plus 3-hydroxylation [3,4]. In contrast to other organisms, only three genes are involved in the whole pathway from phytoene to astaxanthin, which facilitates genetic modification of carotenoid biosynthesis in X. dendrorhous. This yeast has the potential to be engineered as a cell-factory for the production of industrially valuable carotenoids [5]. Tools and techniques for genetic manipulations of X. dendrorhous are available [6], including several integrative transformation plasmids based on four different selection markers [7,8]. It is also advantageous for the development of high-yield carotenoid producers since a carotenoid pathway that can be manipulated is already established, a carotenoid storage system exists and a very active acetyl-CoA metabolism can be utilized. The published genomic sequence of X. dendrorhous CBS6938 [9] is also very helpful for genetic engineering of the carotenoid pathway.
The potential of X. dendrorhous for the production of economically interesting carotenoids like astaxanthin for feed and zeaxanthin as a nutraceutical for eye care has been demonstrated. The highest production of astaxanthin was reached by combining classical mutagenesis with genetic pathway engineering [7]. Starting from a high-yield astaxanthin mutant, genes of three limiting enzymes were over-expressed, enhancing metabolite flow toward carotenoid biosynthesis and into the astaxanthin pathway [7].
Zeaxanthin is another carotenoid of interest as a nutraceutical, which is important for protection of our vision. For the engineering of a zeaxanthin producing strain, a mutant with inactive astaxanthin synthase accumulating β-carotene [10,11] was used to extend the carotenoid pathway to zeaxanthin by expression of a bacterial β-carotene hydroxylase gene and engineering of an enhanced metabolite flow into the carotenoid pathway [12].
Hydroxylated carotenoids such as the dihydroxy compound zeaxanthin [13], tetrahydroxy-βcarotene derivative nostoxanthin [14] and various other hydroxylated acyclic carotenoids [15] have been generated in Escherichia coli by combination of carotenogenic genes from different organisms. It has been shown that not only zeaxanthin but also several other hydroxyl derivatives have superior antioxidative activity. In our engineering approach with X. dendrorhous, we utilized β-carotene mutants to integrate hydroxylase and ketolase genes from bacteria and algae and finally chemically reduced the resulting tetrahydroxy-monoketo and tetrahydroxy-diketo products to a pentahydroxy and a hexahydroxy β-carotene derivative, respectively.

Plasmid Construction and Transformantion of X. dendrorhous, Combinatorial Biosynthesis of Carotenoid Standards in E. coli
The integrative plasmids pPR2TNo-crtZo [12] and pPRcDNA1bkt830 [11] were previously described. Details on the origin of genes and the primers for the amplification of the X. dendrorhous transformation plasmids are shown in Table 1. For the construction of plasmid pPR2TNH-crtG, the crtG gene from plasmid pUC-Bre-O11 [16] was amplified, which also generated an EcoRI and XhoI restriction site (Table 1), and was cloned into the XcmI site of the E. coli plasmid pMon38201 by its a-overhang [17]. From there, the crtG gene was cut out with EcoRI and XhoI and ligated into pUC8∆EcoRI-HNNH [7] for fusion with the promoter and terminator of the glyceraldehyde phosphate dehydrogenase gene from X. dendrorhous. Then, the whole cassette was cut out by restriction with HindIII and ligated into the HindIII site of pPR2TNH [7], yielding pPR2TNH-crtG. Transformation of X. dendrorhous was performed by electroporation as described in Visser et al. [6] with 10 µg plasmid DNA. After phenol-chloroform purification, plasmid DNA was linearized by digestion with SfiI.

Carotenoid Extraction, Purification and Chemical Derivatization
For carotenoid extraction, 20 mg of freeze-dried X. dendrorhous cells were mixed with 500 µL glass beads (0.25-0.5 mm), 675 µL methanol and 75 µL of a 60% KOH solution for saponification and broken in a cell mill (Retsch MM 400) for 8 min at a frequency of 30 per second and heated for 20 min at 60 • C. Samples containing keto carotenoid were extracted without addition of KOH. After partitioning of the carotenoid extract into 50% diethyl ether in petrol (bp 40-60 • C), the upper phase was collected and dried in a stream of nitrogen. Carotenoids were quantified from three independently grown cultures.
For the purification of multi-hydroxy keto carotenoids, the extract of PR1-104-ZGbkt was fractionated by TLC on activated silica plates developed with toluene/ethyl acetate/methanol (65:35:5, by volume). The bands with an R f value of 0.20 and 0.25 were collected and extracted with acetone.
The reduction of carotenoid ketones dissolved in ethanol to the corresponding alcohol was performed according to Eugster [22] with NaBH 4 . The reaction mixture was transferred to 65% diethyl ether in petrol (bp 40-60 • C) and the upper phase collected and dried in stream of nitrogen. The same procedure was applied to canthaxanthin, yielding 4,4'-diketo-β-carotene together with 4-HO-4'-keto-β-carotene and to echinenone yielding 4-HO-β-carotene.

HPLC and HR-ESI-MS Analysis
HPLC analysis was performed on three different HPLC systems. For separation of the non-ketolated multi-hydroxy carotenoids, system I with a 15 × 0.4 cm Nucleosil 100 C18, 3 µm column and acetonitrile (ACN)/methanol/2-propanol (85:10:5, by volume) plus 3% H 2 O [23] was used as mobile phase with a flow rate of 0.8 mL/min at 10 • C. System II was employed for the separation of the hydroxy-keto carotenoids and their reduction products on a 25 cm C30 RP, 3 µm column (YMC, Wilmington, NC, USA) with a mobile phase of 3% methyl tertiary-butyl ether in methanol for 48 min followed by an increase to 20% with a flow rate of 0.8 mL/min at 20 • C. This system was also used for the quantification of the keto derivatives. HR-ESI-MS analysis was carried out in system III on a 2.1 mm × 50 mm ACQUITY UPLC BEH C18, 1.7 µm column. A binary gradient was applied with ACN (+0.1% formic acid) and H 2 O (+0.1% formic acid) for 12 min in the following steps: 0-2 min 5% ACN , 2-2.5 min 40% ACN, 2.5-4 min 40% ACN, 4-14 min 40%-95% at a flow rate of 0.4 mL/min at 40 • C. This HPLC system was coupled to an Impact II QTOF (Bruker) mass spectrometer using Na-formiate as an internal calibration standard [24]. Carotenoids were detected in a positive ion mode with scanning range from 100-1200 m/z. Optical spectra were recorded online with a photodiode array detector 994 (Waters, Milford, CT, USA). Carotenoid standards for identification were generated in E. coli by the combination of different crt genes as previously described [25].

Results
Although fungi are in general unable to synthesize zeaxanthin or other hydroxy-carotenoids, formation of the 3,3'-dihydroxy-β-carotene can be engineered into X. dendrorhous [12]. This potential was extended for the synthesis of other derivatives with up to six hydroxyl groups. As outlined in Figure 1, this was achieved by a strategy involving the use of a β-carotene accumulating mutant of X. dendrorhous, its consecutive transformation with three microbial transgenes and finally the reduction of the 4-and 4'-keto groups. Figure 2 shows the HPLC analysis to identify some of the β-carotene-derived oxo compounds from the transgenic lines. Transformation with a bacterial 3-hydroxylase gene changed the PR1-104 from a β-carotene accumulator (trace A) to a transformant synthesizing zeaxanthin together with small amounts of the intermediate β-cryptoxanthin (trace B). In a second transformation step, with a bacterial 2-hydroxylase gene, the line PR1-104-ZG was obtained in which β-carotene was converted to nostoxanthin (trace C) as indicated by the nostoxanthin standard in trace D. Oxygenation of nostoxanthin was increased by a third transformation step with an algal 4-ketolase gene. In the resulting transformant PR1-104-ZGbkt, several carotenoids were detected. For base-line HPLC separation of these very polar oxo carotenoids, system I had to be changed to system II ( Figure 2, trace E). In addition to the expected nostoxanthin-derived keto carotenoids, other carotenoids such as β-carotene (peak 7) and its keto derivatives echinenone (peak 6) and canthaxanthin (peak 5) as well as 4-keto-zeaxanthin (peak 4) and nostoxanthin (peak 3) were generated and could be identified with standard carotenoids (Figure 2, traces F and G). Compounds 1 and 2 of trace E are highly polar diketo and mono keto derivatives, respectively, as indicated by their optical absorbance spectra exhibiting a typical shape for keto carotenoids and an absorbance maximum of 475 or 468 nm ( Figure 3).

Results
Although fungi are in general unable to synthesize zeaxanthin or other hydroxy-carotenoids, formation of the 3,3'-dihydroxy-β-carotene can be engineered into X. dendrorhous [12]. This potential was extended for the synthesis of other derivatives with up to six hydroxyl groups. As outlined in Figure 1, this was achieved by a strategy involving the use of a β-carotene accumulating mutant of X. dendrorhous, its consecutive transformation with three microbial transgenes and finally the reduction of the 4-and 4'-keto groups. Figure 2 shows the HPLC analysis to identify some of the βcarotene-derived oxo compounds from the transgenic lines. Transformation with a bacterial 3hydroxylase gene changed the PR1-104 from a β-carotene accumulator (trace A) to a transformant synthesizing zeaxanthin together with small amounts of the intermediate β-cryptoxanthin (trace B). In a second transformation step, with a bacterial 2-hydroxylase gene, the line PR1-104-ZG was obtained in which β-carotene was converted to nostoxanthin (trace C) as indicated by the nostoxanthin standard in trace D. Oxygenation of nostoxanthin was increased by a third transformation step with an algal 4-ketolase gene. In the resulting transformant PR1-104-ZGbkt, several carotenoids were detected. For base-line HPLC separation of these very polar oxo carotenoids, system I had to be changed to system II ( Figure 2, trace E). In addition to the expected nostoxanthin-derived keto carotenoids, other carotenoids such as β-carotene (peak 7) and its keto derivatives echinenone (peak 6) and canthaxanthin (peak 5) as well as 4-keto-zeaxanthin (peak 4) and nostoxanthin (peak 3) were generated and could be identified with standard carotenoids (Figure 2, traces F and G). Compounds 1 and 2 of trace E are highly polar diketo and mono keto derivatives, respectively, as indicated by their optical absorbance spectra exhibiting a typical shape for keto carotenoids and an absorbance maximum of 475 or 468 nm (Figure 3).    The carotenoid extract from transformant PR1-104-ZGbkt was reduced and separated by HPLC ( Figure 4A). New peaks 8 to 12 emerged, which were not present in the non-reduced sample. Peaks   The carotenoid extract from transformant PR1-104-ZGbkt was reduced and separated by HPLC ( Figure 4A). New peaks 8 to 12 emerged, which were not present in the non-reduced sample. Peaks Figure 3. Absorbance spectra of hydroxy-keto carotenoids before (compounds 1 and 2 with cis isomers 1', 1" and 2') and after reduction (compounds 8 and 9 with cis isomer 9').

Discussion
Production of carotenoids by genetically engineered yeasts proved to be a promising alternative to chemical synthesis or extraction from plants [26]. The red yeast X. dendrorhous is the most versatile host with the highest carotenoid yield among fungi [5]. It is possible to construct and implement pathways to different carotenoid structures into this yeast. As an example, the synthesis to multi-

Discussion
Production of carotenoids by genetically engineered yeasts proved to be a promising alternative to chemical synthesis or extraction from plants [26]. The red yeast X. dendrorhous is the most versatile host with the highest carotenoid yield among fungi [5]. It is possible to construct and implement pathways to different carotenoid structures into this yeast. As an example, the synthesis to multi-oxygenated carotenoids was chosen in this publication (Figure 1) to demonstrate the potential of X. dendrorhous as a production platform for complex carotenoid structures. This was possible by extension of the pathway from accumulating β-carotene. The step-by-step transformation resulted in intermediary lines accumulating zeaxanthin or nostoxanthin as major carotenoids (Figure 2). Nostoxanthin is a carotenoid found in cyanobacteria [27] and was accumulated in recombinant Escherichia coli [14]. A final X. dendrorhous line PR1-104-ZGbkt transformed with three microbial genes produced 4-keto-nostoxanthin and 4,4'-diketo-nostoxanthin ( Figure 2). Both carotenoids are extremely rare in nature and have been identified before only from two bacteria Brevundimonas SD212 and Rhizobium lupine [28,29]. By reduction of both keto carotenoids isolated from our line PR1-104-ZGbkt, the novel carotenoids 4-hydroxy-nostoxanthin and 4,4'-dihydroxy-nostoxanthin were obtained (Figure 4). Some of the isolated oxo carotenoids separated into several geometrical isomers on the C30 column ( Figure 3). In each case, the all-trans isomer dominated. The isomers 1' and 2' showed a cis peak at 370 nm and 9' at 330 nm. According to their height in relation to the dominating absorbance maximum, their position in front of the all-trans isomer on a C30 column and in comparison to astaxanthin for the keto derivatives [6] and to zeaxanthin for 4-hydroxy-nostoxanthin [30], these isomers are most likely 13-cis. In contrast, 1" without a cis peak may be a 9-cis 4,4'-diketo-nostoxanthin isomer.
The keto carotenoids and their reduction products were identified by high resolution mass spectrometry ( Figure 5). For all of them, the correct molecular masses and the typical prominent fragments were obtained ( Table 2). As indicated in the right part of Figure 5A, fragment M-92 originates from an in chain elimination of a toluene unit and is an indication of the central polyene chain [31]. All analyzed carotenoids show the elimination of hydroxyl groups either as water (M-18) or as neutral loss of 17 Da. In addition, an intensified peak was found at 147.12 Da in the spectra of the mono and diketo derivaties ( Figure 5A,C). This is typical for a dehydrated 4-keto-β-ionone ring with cleavage of the C7,8 bond [31].
Our combined approach of mutant selection, genetic engineering and chemical modification is set as a general example of how novel carotenoids can be generated in X. dendrorhous. For the production of 4-keto-nostoxanthin and 4,4'-diketo-nostoxanthin, it is a proof of concept, which also indicates how to improve their yields in future studies. Judging from the relative low conversion of β -carotene to zeaxanthin ( Figure 2B) and complete conversion of zeaxanthin to nostoxanthin ( Figure 2C), the 3-hydroxylation step is regarded limiting in the formation of ketolated nostoxanthin. Formation of ketolated β-carotene derivatives echinenone and canthaxanthin demonstrate that the ketolation step is not limited ( Figure 2E). It has previously been shown that conversion rates of transgenic reactions in X. dendrorhous are dependent on the number of trans gene copies integrated into the genome [32]. Therefore, either transformation with a plasmid carrying two copies of the 3-hydroxylase gene crtZ as demonstrated by Pollmann et al. [12] or repeated transformation with the crtZ gene is a promising way to improve intermediate conversion to the end product. In addition, overall carotenoid synthesis can be enhanced up to 3-fold by improvement of precursor supply in combination with increased flux into the carotenoid pathway [12] and in combination with a high carotenoids producing X. dendrorhous mutant, a total increase of carotenoid formation of up to 90-fold can be achieved [7]. Acknowledgments: This work was supported as the ERA-IB project PROCAR through the German Federal Ministry of Education and Research (BMBF) (FKZ 031A569) and in part by COST Action CA15136. The high resolution ESI-MS system was partially funded by the DFG.
Author Contributions: Hendrik Pollmann and Gerhard Sandmann designed the experiments and carried out the HPLC analysis, Jürgen Breitenbach constructed the transformation plasmids, Hendrik Pollmann carried out transformation and cultivation of X. dendrorhous including sample preparation, Hendrik Wolff and Helge B. Bode performed high resolution MS-MS measurement and data analysis. All authors commented on the completed manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.