Novel Carotenoid Oxidase Involved in Biosynthesis of 4,4′-Diapolycopene Dialdehyde

Abstract

Biosynthesis of C₃₀ carotenoids is relatively restricted in nature but has been described in Staphylococcus and in methylotrophic bacteria. We report here identification of a novel gene (crtNb) involved in conversion of 4,4′-diapolycopene to 4,4′-diapolycopene aldehyde. An aldehyde dehydrogenase gene (ald) responsible for the subsequent oxidation of 4,4′-diapolycopene aldehyde to 4,4′-diapolycopene acid was also identified in Methylomonas. CrtNb has significant sequence homology with diapophytoene desaturases (CrtN). However, data from knockout of crtNb and expression of crtNb in Escherichia coli indicated that CrtNb is not a desaturase but rather a novel carotenoid oxidase catalyzing oxidation of the terminal methyl group(s) of 4,4′-diaponeurosporene and 4,4′-diapolycopene to the corresponding terminal aldehyde. It has moderate to low activity on neurosporene and lycopene and no activity on β-carotene or ζ-carotene. Using a combination of C₃₀ carotenoid synthesis genes from Staphylococcus and Methylomonas, 4,4′-diapolycopene dialdehyde was produced in E. coli as the predominant carotenoid. This C₃₀ dialdehyde is a dark-reddish purple pigment that may have potential uses in foods and cosmetics.

More than 600 different carotenoids are synthesized in nature by plants, algae, fungi, and bacteria. Most of these carotenoids contain a 40-carbon backbone (C₄₀ carotenoids). However, C₃₀ carotenoids with 30-carbon backbones have been reported in a small group of bacteria, including Staphylococcus aureus (12), Methylobacterium rhodinum (formerly Pseudomonas rhodos) (10), Streptococcus faecium (23), and Heliobacteria spp. (17). By analysis of pigment mutants, biosynthetic pathways were proposed for C₃₀ carotenoids in Staphylococcus (13, 24) and Methylobacterium (9). Both pathways involve an important intermediate step, which is the functionalization of the C₃₀ carotene backbone with an aldehyde group(s). The C₃₀ carotenoid, staphyloxanthin, synthesized in Staphylococcus is functionalized at one end of the C₃₀ carotene (Fig. 1A). The C₃₀ carotenoic acid glucosyl ester synthesized in Methylobacterium is functionalized at both ends of the C₃₀ carotene (Fig. 1B).

FIG. 1.

Proposed pathways for biosynthesis of C₃₀ carotenoids in Staphylococcus aureus (A) and Methylobacterium rhodinum or Methylomonas sp. strain 16a (B). CrtM and CrtN from S. aureus were previously identified (27), and CrtNb from S. aureus was identified in this study. CrtN, CrtNb, Ald, and Sqs were identified from Methylomonas sp. in this study. The genes from M. rhodinum have not yet been identified.

Genetic data on C₃₀ carotenoid biosynthesis are limited, although two genes, crtM, encoding diapophytoene synthase, and crtN, encoding diapophytoene desaturase, were cloned from Staphylococcus (27). These two genes were confirmed to be responsible for the synthesis of 4,4′-diaponeurosporene. However, it is not known how the end of 4,4′-diaponeurosporene is functionalized with an aldehyde group, which is subsequently oxidized and esterified to synthesize staphyloxanthin in Staphylococcus. Several open reading frames (ORFs) with significant homology to crtM or crtN are evident from microbial genomic sequencing (18, 28). Nevertheless, no gene was identified for functionalization of the C₃₀ carotene backbone.

We report here identification of a gene cluster involved in C₃₀ carotenoid synthesis in Methylomonas sp. strain 16a. We identified a novel gene (crtNb) from Methylomonas and its homologue in Staphylococcus, which catalyze oxidation of the terminal methyl group(s) of the C₃₀ carotene backbone to an aldehyde group(s).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Methylomonas sp. strain 16a (ATCC PTA-2402) was grown in 550-ml serum bottles closed by gas-tight butyl septa for 2 days at 30°C in BTZ3 mineral medium (100 ml per bottle) (15). Methane served as a sole carbon and energy source and was fed via the gaseous phase (25%). Methylobacterium rhodinum ATCC 14821 was grown at 30°C in 1-liter Erlenmeyer flasks, each containing 100 ml of BTZ3 medium with methanol (100 mM) as a sole carbon and energy source. After 2 days, more methanol (100 mM) was fed, and the cells were harvested after 4 days. Escherichia coli cells were grown in Luria broth (16) with appropriate antibiotics at 37°C overnight. Genomic DNA of the following strains was used as a template for amplification of certain carotenoid synthesis genes by PCR: Methylomonas sp. strain 16a (ATCC PTA-2402), Staphylococcus aureus ATCC 35556, Pantoea stewartii ATCC 8199, Synechocystis sp. strain PCC6803, and Rhodobacter sphaeroides ATCC 17023. The plasmids used in this study are listed in Table 1.

TABLE 1.

Plasmids used in this study

Plasmid	Backbone	Expressed gene(s) and organisma	Reference
pBHR-crt1	pBHR1b	crtEXYIB (P. stewartii)	20
pDCQ51	pBHR1	crtEXY::Tn5d-crtIB (P. stewartii)	20
pDCQ52	pBHR1	crtEXYI::Tn5-crtB (P. stewartii)	This study
pDCQ150	pCR2.1-TOPOc	crtN (16a)-ald (16a)-crtNb (16a)	This study
pDCQ151	pBHR1	crtN (16a)-ald (16a)-crtNb (16a)	This study
pDCQ153	pTrcHis2-TOPOc	crtM (S. aureus)	This study
pDCQ155	pTrcHis2-TOPO	crtM (S. aureus)-crtN (16a)-ald (16a)-crtNb (16a)	This study
pDCQ159	pTrcHis2-TOPO	crtM (S. aureus)-crtN (16a)-ald (16a)	This study
pDCQ160	pTrcHis2-TOPO	crtM (S. aureus)-crtN (16a)-ald (16a)-crtNb (S. aureus)	This study
pDCQ166	pBHR1	crtM (S. aureus)-crtN (S. aureus)	20
pDCQ167	pTrcHis2-TOPO	crtNb (16a)	This study
pDCQ168	pTrcHis2-TOPO	crtNb (S. aureus)	This study
pDCQ174	pTrcHis2-TOPO	crtN (16a)	This study
pDCQ177	pTrcHis2-TOPO	crtN (16a)-crtNb (16a)	This study
pDCQ196	pBHR1	crtE (P. stewartii)-crtP (Synechocystis)-crtB (P. stewartii)	20
pDCQ198	pBHR1	crtE (P. stewartii)-crtI (R. sphaeroides)-crtB (P. stewartii)	20

^aOrganism refers to the organism from which the gene was isolated (16a, Methylomonas sp. strain 16a).

^bMoBiTec (Goettingen, Germany).

^cInvitrogen (Carlsbad, CA).

^dTn5 insertion in crtY for pDCQ51; Tn5 insertion in crtI for pDCQ52.

Carotenoid analysis.

Centrifuged cell pellets of Methylomonas sp. strain 16a and Methylobacterium rhodinum were extracted with methanol. The extracts were saponified (10% KOH; 30°C; 14 h) and analyzed using high-performance liquid chromatography (HPLC) with diode array detection (Beckman Gold Nouveau System, Columbia, MD). Samples (100- to 250-μl injections) were loaded onto a 125- by 4-mm RP8 (5-μm particles) column and corresponding guard column (Hewlett-Packard, San Fernando, CA). The eluents used in the gradient program were methanol (CH₃OH) and deionized water (dH₂O) at 1 ml/min: 0 to 2 min, CH₃OH-dH₂O (60:40 [vol/vol]); 2 to 11.5 min, linear gradient to 100% CH₃OH; 11.5 to 20 min, 100% CH₃OH; and then reequilibration to the starting conditions. For carotenoid analysis in transposon-mutagenized Escherichia coli, cells were harvested by centrifugation and extracted with acetone. After evaporation of the acetone, extracts were redissolved in 100 μl of methanol and injected onto HPLC. The solvent program with a 1-ml/min flow rate was as follows: 0 to 11.5 min, linear gradient from CH₃OH-dH₂O (60:40 [vol/vol]) to 100% CH₃OH; 11.5 to 20 min, 100% CH₃OH; 20 to 30 min, CH₃OH-dH₂O (60:40 [vol/vol]). For carotenoid analysis in the crtNb expression clones and for molecular weight determination, the samples were analyzed using an AgilentSeries 1100 LC/MSD (Agilent, Foster City, CA). Samples (20 μl in acetone) were loaded onto a 150-mm by 4.6-mm ZORBAX C₁₈ (3.5-μm particles) column (Agilent Technologies, Inc.). The column temperature was kept at 40°C. The flow rate was 1 ml/min. The eluents were a mixture of buffer A, which was acetonitrile-deionized water (95:5 [vol/vol]), and buffer B, which was tetrahydrofuran. The gradient was as follows: 0 to 2 min, buffer A-buffer B (95:5 [vol/vol]); 2 to 10 min, linear gradient from buffer A-buffer B (95:5 [vol/vol]) to buffer A-buffer B (60:40 [vol/vol]); 10 to 12 min, linear gradient from buffer A-buffer B (60:40 [vol/vol]) to buffer A-buffer B (50:50 [vol/vol]); 12 to 18 min, buffer A-buffer B (50:50 [vol/vol]); and 18 to 20 min, buffer A-buffer B (95:5 [vol/vol]).

The synthetic standards of β-carotene and lycopene were purchased from Sigma (St. Louis, MO). The synthetic standards of 4,4′-diapolycopene-4,4′-dialdehyde and 4,4′-diapolycopene-4,4′dioic acid were purchased from CaroteNature (Lupsingen, Switzerland).

Construction of C₃₀ carotenoid synthesis gene cluster.

Primers 5′-ATGACAATGATGGATATGAATTTTAAA-3′ and 5′-GGATCCTATATTCTATGATATTTACTATTTATTTC-3′ were used to amplify the 869-bp crtM gene from Staphylococcus. The amplified gene product was cloned into the pTrcHis2-TOPO expression vector (Invitrogen, Carlsbad, CA) in forward orientation as pDCQ153 (Table 1). Primers 5′-GGTCTCAAATTGCATCAACGGATCATCATGGCCAAC-3′ and 5′-GGTCTCTAATTGCTAGCTTATTGCAAATCCGCCACAATCTTGTC-3′ were used to amplify the 4,668-bp crtN ald crtNb gene cluster from Methylomonas. The amplified product was first cloned into the pCR2.1-TOPO vector (Invitrogen). The BsaI fragment of the TOPO construct (pDCQ150) containing the crtN ald crtNb cluster was then cloned into the EcoRI site of the pBHR1 vector (MobiTec, Goettingen, Germany). The pBHR1 construct containing the cluster in forward orientation with respect to the Cm^r gene promoter was designated pDCQ151 (Table 1).

To facilitate mutagenesis, the four required crt genes for producing carboxy-carotenoid in E. coli were strung together by subcloning the 4.7-kb BsaI fragment of pDCQ150 containing the three Methylomonas crtN ald crtNb genes into the EcoRI site downstream of the staphylococcal crtM gene in pDCQ153. The resulting construct, pDCQ155, contained the crtM gene and the crtN ald crtNb genes in the same orientation, under the control of the trc promoter from the pTrcHis2-TOPO vector (Table 1).

In vitro transposon mutagenesis. In vitro transposon mutagenesis was performed on pDCQ155 DNA, using the EZ::TN <TET-1> Insertion kit (Epicentre, Madison, WI). The transposon-treated pDCQ155 DNA was transformed into TOP10 competent cells (Invitrogen). Cells containing transposon insertions were selected on LB plates with 10 μg/ml tetracycline. The transposon mutants were screened by PCR and sequenced to identify the transposon insertion sites within the respective genes.

Construction of E. coli strains producing alternative carotenoid substrates.

E. coli strains producing β-carotene, lycopene, neurosporene, ζ-carotene and 4,4′-diaponeurosporene were described previously (20). A phytoene-accumulating E. coli strain was constructed similarly to the lycopene-accumulating strain by isolation of a transposon insertion in the crtI gene on pDCQ52. The 4,4′-diapophytoene-accumulating strain containing pDCQ153 was constructed as described above.

Construction of expression clones of crtNb.

The Methylomonas crtNb gene was PCR amplified from its genomic DNA using primers 5′-ATGAACTCAAATGACAACCAACG-3′ and 5′-GAATTCTATTGCAAATCCGCCACAATCT-3′. The Staphylococcus crtNb gene was amplified from the genomic DNA of ATCC 35556, using primers 5′-ATGACTAAACATATCATCGTTATTG-3′ and 5′-GAATTCACTTCCTATTCTTCGCTTCTC-3′. The crtNb genes were cloned into the pTrcHis2-TOPO cloning vector, resulting in plasmids pDCQ167 and pDCQ168 (Table 1). The Methylomonas crtN gene was amplified from its genomic DNA using primers 5′-ATGGCCAACACCAAACACATCA-3′ and 5′-ggatccTCAGGCTTTGGCTTTTTTCAGC-3′ and cloned into the pTrcHis2-TOPO vector (pDCQ174). Methylomonas crtNb with an artificial ribosome binding site (RBS) was amplified using primers 5′-ggatccaagcttAAGGAGGAATAAACCATGAACTCAAATGACAACCAACGC-3′ and 5′-ggatccaagcttATTGCAAATCCGCCACAATCTT-3′. The incorporated restriction sites in the primers are underlined, and boldface indicates the RBS. The Methylomonas crtNb gene with the RBS was then subcloned downstream of crtN in pDCQ174, resulting in plasmid pDCQ177 (Table 1).

Nucleotide sequence accession number.

The sequence of the crtN ald crtNb gene cluster from Methylomonas sp. strain 16a has been deposited in GenBank under accession number AY841893.

RESULTS

Methylomonas sp. strain 16a produces esters of 4,4′-diapolycopene-4-oic acid and 4,4′-diapolycopene-4,4′-dioic acid.

Methylomonas sp. strain 16a (ATCC PTA-2402) is a naturally occurring methanotrophic isolate. It produces pink pigments that can easily be extracted from the cells using polar organic solvents, such as methanol or acetone. Preliminary analysis and a literature search led to the assumption that carotenoids similar to those found in Methylobacterium rhodinum (9, 10) are produced in Methylomonas sp. strain 16a. Analysis of saponified extracts from the two bacteria suggested the presence of 4,4′-diapolycopene-4,4′-dioic acid and 4,4′-diapolycopene-4-oic acid in both extracts, as shown in cochromatographic HPLC with diode array detection. However, the nonsaponified carotenoids from the two bacteria migrate slightly differently, which suggested that the nonchromophoric ester moieties of the Methylomonas carotenoids are different from those found in Methylobacterium rhodinum.

Identification of a carotenoid synthesis gene cluster in Methylomonas sp. strain 16a.

The genome sequence of Methylomonas sp. strain 16a has been determined and fully annotated (J.-F. Tomb, personal communication). A C₃₀ carotenoid biosynthesis pathway is proposed for Methylomonas sp. strain 16a based on genomic data, pigment analysis, and the pathway proposed for Methylobacterium rhodinum (9) (Fig. 1B).

A gene cluster was identified in the Methylomonas genome that appeared to be involved in carotenoid synthesis based on sequence homology with other known carotenoid synthesis genes (Fig. 2A). Two of the genes in this cluster, designated crtN and crtNb, showed sequence homology to carotenoid desaturases by BLAST analysis (1). COG analysis (21, 22) also showed that both crtN and crtNb belong to the phytoene desaturase family. The Methylomonas crtN product had 34% amino acid identity with that of a crtN homologue in Heliobacillus mobilis (28). It had 31% amino acid identity with the crtN product in Staphylococcus aureus (27). The Methylomonas crtNb product had 51% amino acid identity with that of an unknown ORF in Staphylococcus aureus identified from genomic sequencing. It also had 30% amino acid identity with the product of crtN in Staphylococcus aureus. Methylomonas crtN and crtNb products showed 28% amino acid identity to each other. An aldehyde dehydrogenase gene (designated ald) was located between crtN and crtNb in the same gene cluster. The Methylomonas ald product had 36% amino acid identity with the aldehyde dehydrogenase from Nostoc sp. (7).

FIG. 2.

Carotenoid synthesis gene cluster in Methylomonas sp. strain 16a (A) and Staphylococcus aureus (B). The crtM homologue was not identified in Methylomonas sp. strain 16a.

Functional determination of the C₃₀ carotenoid synthesis genes in E. coli.

A crtM homologue of the diapophytoene synthase of Staphylococcus was not found in Methylomonas. The crtN ald crtNb gene cluster from Methylomonas was coexpressed with Staphylococcus crtM in E. coli, either on two compatible plasmids (pDCQ151 and pDCQ153) or as a single crtM crtN ald crtNb cluster on pDCQ155. In both cases, a pink pigment was produced in E. coli with absorption spectra (λ_max, 465, 490, and 519) similar to those of the Methylomonas native pigment. However, the retention time of the carotenoid from E. coli (11.4 min) was different from that of the Methylomonas native carotenoid (12.7 min), which suggested that the C₃₀ carotenoid produced in E. coli either was not esterified or had a different nonchromophoric ester moiety from the one in the native Methylomonas host. LC-mass spectrometry analysis determined that the molecular mass of the carotenoid produced in E. coli was 460 Da. The absorption spectra, retention time, and molecular weight of the carotenoid produced in E. coli all matched well with those of the synthetic standard of 4,4′-diapocarotenoic diacid.

Synthesis of 4,4′-diapocarotenoic diacid in E. coli confirmed the involvement of the crtN ald crtNb gene cluster in C₃₀ carotenoid synthesis. To determine the function of each gene and to identify the molecular mechanism for carboxylation, in vitro transposon mutagenesis was performed with plasmid pDCQ155 containing the coexpressed gene cluster crtM crtN ald crtNb. This transposon was used previously to assign functions to gene clusters and showed no polar effect (5). Table 2 summarizes the analyses of the transposon mutants. Mutant w13, with a transposon insertion in the crtN gene, produced only the colorless C₃₀ carotenoid precursor 4,4′-diapophytoene. This is consistent with the function of Methylomonas crtN, to encode a diapophytoene desaturase. Mutant p33, with a transposon insertion in the ald gene, produced the 4,4′-diapolycopene dialdehyde (molecular weight m/e, 428) in addition to the 4,4′-diapophytoene precursor. The red pigment (λ_max, 507) produced in mutant p33 was rapidly reduced by NaBH₄ to a yellow pigment (λ_max, 444, 470, and 501). The reduction experiment suggested that the carotenoid had an aldehyde group(s) (3). This also confirmed the function of Methylomonas ald as encoding an aldehyde dehydrogenase. Mutant w18, with a transposon insertion in the crtNb gene, produced the fully unsaturated C₃₀ carotenoid backbone 4,4′-diapolycopene with some additional less unsaturated intermediates. The fact that only the C₃₀ carotenoid backbones were produced in the w18 mutant suggested that functionalization of the 3,4-didehydro-psi end group of 4,4′-diapolycopene was blocked in this mutant. crtNb is not a diapophytoene desaturase gene; instead, it likely encodes an enzyme that oxidizes the terminal methyl group of the 4,4′-diapolycopene to produce 4,4′-diapolycopene dialdehyde, which is further oxidized by the aldehyde dehydrogenase to produce the C₃₀-carboxy carotenoid, 4,4′-diapolycopene diacid.

TABLE 2.

Summary of E. coli strains containing pDCQ155 with Tn5 insertions

Strain	Color	Insertion sitea	Mutated geneb	Elution time (min)	UV/visible maxima (nm)	Suggested carotenoid(s)
wtc	Pink	NAd	NA	11.4	465, 490, 519	4,4′-Diapolycopene-4,4′-diacid
w32	White	869	crtM (422-1285)	NDd	ND	ND
w13	White	1748	crtN (1318-2853)	15.1	286, 298	4,4′-Diapophytoene
p33	Deep red	3764	ald (2856-4448)	12.4	507	4,4′-Diapolycopene-4-4′-dial
				15.1	286, 298	4,4′-Diapophytoene
w18	Light orange	5817	crtNb (4448-5941)	14.2	442, 468, 498	4,4′-Diapolycopene
				14.8	329, 346, 366	4,4′-Diapophytofluene
				15.1	286, 298	4,4′-Diapophytoene

^aThe numbers refer to the base pairs on pDCQ155 after which the transposon was inserted.

^bThe numbers refer to the spans of the individual genes on pDCQ155.

^cwt, E. coli containing wild-type pDCQ155 without a Tn5 insertion.

^dNA, not applicable; ND, no carotenoid detected.

Identification of the crtNb homologue in Staphylococcus.

The crtNb gene from Methylomonas sp. strain 16a was used to search the S. aureus genome database (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). The crtNb homologues were identified in S. aureus genomes, and they are over 99% identical to each other and highly similar to the Methylomonas crtNb gene (51% identity). The crtNb homologues in the two finished genomes of S. aureus strain N315 (ORF identifier, SA2351) and strain Mu50 (ORF identifier, SAV2564) were both annotated as hypothetical proteins. The crtNb homologue in the genome of S. aureus strain NCTC 8325 (at positions 410887 to 409394) was not annotated. This gene was upstream of the crtMN genes with an unknown ORF in between (Fig. 2B). crtNb from S. aureus strain NCTC 8325 was amplified from genomic DNA of ATCC 35556D, and its function was tested by replacing Methylomonas crtNb on pDCQ155.

The E. coli transformants containing this pDCQ160 construct (Table 1) were pink, compared to the light-orange color of the E. coli transformants containing pDCQ159 (Table 1) without crtNb. This indicated that the staphylococcal crtNb homologue on pDCQ160 was actively expressed and contributed to the color difference of the transformants. HPLC analysis of the carotenoids showed that a carotenoid corresponding to 4,4′-diapolycopene diacid (λ_max, 465, 491, and 520) was present in pDCQ160 transformants, whereas only the nonfunctionalized carotenoid backbone 4,4′-diapolycopene (λ_max, 442, 465, and 495) was present in pDCQ159 transformants. The fact that the same carboxy-carotenoid was observed with pDCQ160 and pDCQ155 suggested that the staphylococcal crtNb homologue performed the same function in E. coli as Methylomonas crtNb to introduce two aldehyde groups to the 4,4′-diapolycopene backbone, which were further oxidized to carboxylic acid groups by Methylomonas ald. In the Staphylococcus native host, only the monoaldehyde was formed from the 4,4′-diaponeurosporene substrate.

Substrate specificities of CrtNb enzymes.

The CrtNb proteins from Methylomonas and Staphylococcus function similarly on C₃₀ carotenoid substrates. It appeared that Methylomonas CrtNb had higher activity, and thus it was chosen to test on more alternative carotenoid substrates (Fig. 3). It had high activity on 4,4′-diaponeurosporene. Almost all 4,4′-diaponeurosporene (λ_max, 416, 440, and 468; m/e, 402) was converted to 4,4′-diaponeurosporene-4-aldehyde (λ_max, 468; m/e, 416). It had moderate to low activity on the C₄₀ substrates neurosporene and lycopene. The new products showed the same absorption spectra as that of neurosporene (λ_max, 420, 440, and 472) or lycopene (λ_max, 448, 476, and 508). The molecular masses of the new products (552 and 550 Da) were 14 Da larger than the respective masses of neurosporene and lycopene. It is likely that the monoaldehyde form of neurosporene or lycopene was produced by oxidation of the terminal methyl group (CH₃) of neurosporene or lycopene. The lack of significant change of the absorption spectra was consistent with the nonconjugated nature of the aldehyde group with the conjugated part of the C₄₀ carotenoid backbone. CrtNb had no detectable activity on ζ-carotene or β-carotene.

FIG. 3.

HPLC analysis of carotenoids from E. coli expressing CrtNb. (A) 4,4′-Diaponeurosporene substrate; (B) neurosporene substrate; (C) lycopene substrate; (D) β-carotene substrate. In each panel, the reporter strain alone synthesizing the substrate is shown on top, and the reporter strain expressing Methylomonas CrtNb is shown on the bottom. Monoaldehyde is represented as “-al.” The carotenoids were identified based on HPLC separation, absorption spectra, and molecular weight.

CrtNb was also tested on 4,4′-diapophytoene and phytoene substrates. No desaturated products were observed (data not shown). This confirmed that CrtNb is not a carotenoid desaturase, in spite of its apparent sequence homology.

Production of 4,4′-diapolycopene dialdehyde in E. coli.

4,4′-Diapolycopene dialdehyde is an intermediate that is usually not accumulated in the native C₃₀ carotenoid synthesis bacteria. Synthesis of 4,4′-diapolycopene dialdehyde was observed initially in the transposon mutant p33 when the ald gene was disrupted (Table 2). 4,4′-Diapophytoene was also observed as an accumulated intermediate in p33, suggesting insufficient desaturase activity in this strain. The crtN genes from Staphylococcus and Methylomonas show only moderate sequence similarity, and both crtN genes were used to increase the efficiency of diapophytoene desaturation. When pDCQ174 containing Methylomonas crtN was coexpressed with pDCQ166 containing the Staphylococcus crtM crtN genes, predominantly 4,4′-diapolycopene (λ_max, 441, 464, and 495) was produced, indicating improved desaturation of the C₃₀ backbone. When pDCQ177 containing the Methylomonas crtN crtNb genes was coexpressed with pDCQ166, predominantly 4,4′-diapolycopene dialdehyde (λ_max, 504 and 536; m/e, 428) was produced. This carotenoid peak comigrated with the peak of the synthetic standard of 4,4′-diapolycopene dialdehyde. The absorption spectra and the molecular weight of the carotenoid were identical to those of the synthetic standard. The data indicated that CrtNb was responsible for functionalization with aldehyde groups, whereas CrtN was responsible for diapophytoene desaturation. E. coli coexpressing pDCQ166 and pDCQ177 showed an approximately 20-fold increase in 4,4′-diapolycopene dialdehyde production compared to that in the transposon mutant p33. 4,4′-Diapolycopene dialdehyde could be produced as the predominant carotenoid in E. coli at approximately 500 ppm by fermentation (Fig. 4).

FIG. 4.

HPLC analysis of the carotenoids produced from E. coli coexpressing pDCQ166 and pDCQ177. The predominant peak (λ_max, 504 and 536; m/e, 428) eluted at 4.3 min was 4,4′-diapolycopene dialdehyde. The inset shows the dark-reddish purple color of the lyophilized E. coli cells producing 4,4′-diapolycopene dialdehyde.

DISCUSSION

Carotenoid aldehydes have been reported in the literature to be produced by either chemical synthesis or oxidative cleavage. They are used commercially as chemical synthesis precursors or food colorants or for their biological activity. Chemically synthesized C₁₀-dialdehyde is used as a precursor in the Wittig reaction (4) for synthesis of C₄₀-carotenoids, such as astaxanthin or zeaxanthin. Retinal, the chromophore of the various visual pigments, is formed by enzymatic cleavage of β-carotene by a β-carotene dioxygenase (26). Bixin aldehyde (C₂₄-dialdehyde), the intermediate in the biosynthesis of the plant pigment annatto, is formed by enzymatic cleavage of lycopene by a lycopene dioxygenase (2). Autooxidation of lycopene can also produce aldehyde products of different chain lengths by oxdative cleavage of lycopene (8). Here, we report the biosynthesis of carotenoid aldehydes by enzymatic oxidation of the linear end(s) of the carotenoid backbone. The gene crtNb was shown to be responsible for this novel function.

The crtNb gene appeared to have homology with genes encoding diapophytoene desaturases and phytoene desaturases. However, both the transposon insertion data and the heterologous expression data indicated that the crtNb product was not a diapophytoene desaturase. Instead, it encodes a novel carotenoid oxidase to synthesize aldehyde-functionalized carotenoids. The crtNb product showed no homology with carotenoid dioxygenases that synthesize aldehyde-functionalized carotenoids by oxidative cleavage. The role of crtNb in the C₃₀ carotenoid pathway is strikingly similar to the role of crtO in the C₄₀ carotenoid pathway. CrtO was shown to be a novel carotenoid ketolase to introduce a keto group(s) to the β-ionone ring(s) of the C₄₀ cyclic carotenoids, in spite of its apparent homology to phytoene desaturases (6, 19). CrtNb is another example of an enzyme in the broad phytoene desaturase family but evolved to perform a different function.

CrtNb appears to prefer the fully unsaturated linear end of the C₃₀ carotenoid backbone. CrtNb proteins from Methylomonas and Staphylococcus both produced 4,4′-diaponeurosporene monoaldehyde from 4,4′-diaponeurosporene, and 4,4′-diapolycopene dialdehyde from 4,4′-diapolycopene. It appeared that functionalization with monoaldehyde or dialdehyde was dependent on the substrate rather than the host strain source of the CrtNb enzyme. The asymmetric carotenoids produced in Staphylococcus were possibly due to the asymmetric function of CrtN, whereas Staphylococcus CrtNb could oxidize both ends when 4,4′-diapolycopene was provided. CrtNb also exhibits low reactivity with the linear ends of the C₄₀ carotenoids lycopene and neurosporene. CrtNb adds the oxygen group to the linear end (C-1) of the molecule and produces aldehyde-ended carotenoids of the same chain length. This activity is different from the activity of carotenoid dioxygenases, which cleave the polyene chain and produce aldehyde-ended carotenoids of shorter chain lengths. The spheroidene monooxygenase CrtA was also reported to oxygenate linear C₃₀ and C₄₀ carotenoid backbones (11). It produced a 4,4′-diaponeurosporene derivative of unknown structure. This enzyme likely introduced a keto group to the C-2 of the C₄₀ linear carotenoids to produce neurosporene-2-one and lycopene-2-one, which had molecular masses 2 Da larger than neurosporene-al and lycopene-al produced by CrtNb.

An aldehyde dehydrogenase gene (ald), which was responsible for further oxidation of the C₃₀ aldehyde to C₃₀ acid, was located between crtN and crtNb in Methylomonas. A glycine betaine aldehyde dehydrogenase gene (gbsA) in Staphylococcus showed moderate homology (29% identity; 51% similarity) with Methylomonas ald, even though gbsA is not linked to the crtMN or crtNb gene in Staphylococcus. It is not yet known whether gbsA is involved in staphyloxanthin synthesis. A knockout mutant of gbsA in Staphylococcus will address the question. It appears that the C₃₀ aldehyde could not be oxidized to the C₃₀ acid by cross-activity of some aldehyde dehydrogenases with broad substrate specificities. The fact that no significant amount of acid was observed in E. coli strains producing 4,4′-diapolycopene dialdehyde suggested that the native aldehyde dehydrogenases in E. coli could not efficiently carry out the oxidation of C₃₀ aldehyde to C₃₀ acid. Our observation was consistent with what Bouvier et al. reported (2). They did not observe modification of bixin aldehyde when it was incubated with an E. coli protein extract.

The first gene committed in the C₃₀ carotenoid synthesis pathway in Staphylococcus is crtM, which converts farnesyl pyrophosphate to 4,4′-diapophytoene. No crtM homologue was identified in Methylomonas. Instead, a squalene synthase gene (sqs) and a squalene-hopene cyclase (shc) gene were found in Methylomonas, presumably responsible for synthesis of squalene and hopanoids (25) present in Methylomonas (data not shown). Our preliminary evidence suggested that the Methylomonas Sqs protein might have low CrtM activity to convert farnesyl pyrophosphate to 4,4′-diapophytoene, also called dehydrosqualene (data not shown). This secondary activity was also reported for squalene synthase from Arabidopsis (14). It is not known whether this secondary activity of Sqs contributes to the C₃₀ carotenoid synthesis in Methylomonas. Possibly another gene(s) is involved in the synthesis of 4,4′-diapophytoene, the first intermediate of the C₃₀ carotenoids in Methylomonas.

Carotenoid aldehydes have many potential applications. The introduction of aldehyde groups provides a handle and adds flexibility to the uses of the carotenoid molecules. Carotenoids with longer conjugated backbones may be synthesized with longer dialdehyde precursors, which might increase their antioxidant effects. Carotenoids with increased solubility, which is desired in many colorant applications, might be achieved by chemical modification via the aldehyde groups. The aldehyde groups can also be cross-linked with proteins, which might provide long-lasting effects in cosmetic applications. Production of the carotenoid aldehydes in E. coli provides a fermentation source for the materials used for testing the feasibility of the potential applications.