Content area
Bixin, also known as annatto, is a seed-specific pigment widely used in foods and cosmetics since pre-Columbian times. We show that three genes from Bixa orellana, native to tropical America, govern bixin biosynthesis. These genes code for lycopene cleavage dioxygenase, bixin aldehyde dehydrogenase, and norbixin carboxyl methyltransferase, which catalyze the sequential conversion of lycopene into bixin. Introduction of these three genes in Escherichia coli engineered to produce lycopene induced bixin synthesis, thus expanding the supply of this economically important plant product. [PUBLICATION ABSTRACT]
Full text
Bixin, also known as annatto, is a seed-specific pigment widely used in foods and cosmetics since pre-Columbian times. We show that three genes from Bixa orellana, native to tropical America, govern bixin biosynthesis. These genes code for lycopene cleavage dioxygenase, bixin aldehyde dehydrogenase, and norbixin carboxyl methyltransferase, which catalyze the sequential conversion of lycopene into bixin. Introduction of these three genes in Escherichia coli engineered to produce lycopene induced bixin synthesis, thus expanding the supply of this economically important plant product.
Plants produce a vast repertoire of lowmolecular-mass natural products used in the food, cosmetic, and pharmaceutical industries. The molecular basis of this diversity is largely unknown, as many of the enzymes and the genes regulating their biosynthesis have not been identified. It has been estimated that 15 to 25% of the plant genome specifies pathways of natural product biosynthesis (1). One such pathway leads to bixin, also known as annatto, a pigment synthesized by a single terrestrial plant, Bixa orellana, which is native to tropical America. Bixin is a dicarboxylic monomethyl ester apocarotenoid that confers red color to seeds of B. orellana (Fig. 1A) from which it has been extracted and used in foods and cosmetics as a soluble color additive (2-4) since pre-Colombian times. From an economic point of view, bixin ranks second among natural color additives used in industry (3).
The mechanism of bixin biosynthesis remains unresolved. However, it has been suggested that a C^sub 40^ carotenoid (5), most probably lycopene, may be the precursor of bixin (6, 7). On the basis of structural similarity between bixin and the saffron pigment crocetin, the reaction could implicate a dioxygenase, an aldehyde dehydrogenase, and a methyltransferase in a series of reactions proceeding sequentially from lycopene (Fig. 1B). This putative pathway is supported by analysis of lycopene cleavage derivatives that accumulate in trace amounts in Bixa seeds (8) and also by the recent characterization of plant dioxygenases that cleave carotenoid chromophore at defined sites. Maize VP14, the first characterized of this series, catalyzes the cleavage of 9-cis-violaxanthin and 9'-cisneoxanthin into xanthoxin, the precursor of abscisic acid (9). Along the same lines, Arabidopis thaliana carotenoid cleavage dioxygenase 1 (AtCCDI) (10) and Crocus sativus zeaxanthin cleavage dioxygenase (CsZCD) (11) catalyze the initial steps of mycorradicin and crocetin, which give, respectively, the yellow color of arbuscular mycorrhizal roots and the red color of saffron.
To identify the first gene governing bin biosynthesis, we synthesized a DNA probe using total mRNA isolated from B. orellana developing seeds and a reverse transcriptasepolymerase chain reaction (RT-PCR) based on two peptide sequences of CsZCD (11, 12). The resulting 400-bp DNA revealed a likely carotenoid dioxygenase sequence and was used to screen a complementary DNA (cDNA) library from B. orellana developing seeds. We isolated a full-length 1140-bp cDNA, corresponding to the putative B. orellana lycopene cleavage dioxygenase (BoLCD, GenBank accession number AJ489277). The predicted polypeptide chain of 369 amino acids showed high sequence similarity to plant carotenoid dioxygenases from C. sativus CsZCD (identity, 97%) (11); to an uncharacterized apple flower protein, Md-FS2 (identity, 44%) (13); to Arabidopsis AtCCD1 (identity, 28%) (10); and to maize VP14 (identity, 25%) (9) (fig. S1).
The putative B. orellana carotenoid dioxygenase protein was overexpressed in Escherichia coli to test its catalytic activity by using lycopene, beta-carotene, and zeaxanthin as potential carotenoid substrates. When the purified recombinant protein was incubated with lycopene, beta-carotene, or zeaxanthin, we observed that, after thin-layer chromatography, only lycopene was transformed into an aldehyde derivative having an R^sub f^ value of 0.75 (retention factor, the ratio of the movement of the solute to that of the solvent front), and no activity was observed with protein extract from E. coli harboring empty vector (Fig. 2A), (12). The product was further purified by high-performance liquid chromatography (HPLC) coupled with a diode array detection (Fig. 213) and subjected to mass spectrometry (MS) (Fig. 2C) (12). The ultraviolet (UV)-visible absorption spectrum and the MS revealed that the product had 11 conjugated double bonds and a molecular ion M^sup +^ at m/z 348, which are consistent with the molecular formula of bixin aldehyde (C^sub 24^H^sub 28^O^sub 2^) (Fig. 2C). This indicated that the cloned cDNA codes for BoLCD and suggests that the initial step of bixin synthesis involves the conversion of lycopene into bixin aldehyde, whereas subsequent reactions involve the action of a bixin aldehyde dehydrogenase and a norbixin methyltransferase (Fig. 1B).
To clone the gene coding for the aldehyde dehydrogenase and the methyltransferase acting downstream of BoLCD, we used primers derived from plant aldehyde dehydrogenases (14) and carboxyl methyltransferases (15, 16) for RT-PCR (12). The resulting fragments were used to screen the cDNA from B. orellana developing seeds (12). DNA sequencing of two isolated clones revealed two partial cDNA clones with a likely aldehyde dehydrogenase sequence and a likely methyltransferase sequence that were used as a probe to isolate a full-length 1548-bp cDNA encoding the putative B. orellana bixin aldehyde dehydrogenase (BoBADH, GenBank accession number AJ548846) coding for 504 amino acids and a full-length 1149-bp cDNA encoding the putative B. orellana norbixin methyltransferase (BonBMT), (GenBank accession number AJ548847) coding for 375 amino acids. The deduced amino acid sequence of the BoBADH cDNA showed 76% and 66% similarity with maize and Arabidopsis aldehyde dehydrogenase [fig. S2 and (14)], respectively. The predicted polypeptide sequence of the BonBMT cDNA showed that 51 % of the amino acid sequence was identical to Clarkia breweri salicylic acid methyltransferase (15) and 38% of the sequence to Antirrhinum majus benzoic acid methyltransferase (16) (fig. S3).
To investigate the catalytic properties of BoBADH and BonBMT proteins, we overexpressed them in E. coli and incubated the purified recombinant proteins with bixin aldehyde as a substrate for aldehyde dehydrogenase activity and norbixin as a substrate for methyltransferase activity. This revealed that recombinant BoBADH efficiently converted bixin aldehyde into a compound identified as trans-norbixin by HPLC and UV-visible spectrophotometry (Fig. 2D). Similarly, recombinant BonBMT converted norbixin mainly into a compound identified as the natural bixin isomer and traces of bixin dimethyl ester (Fig. 2E). Bixin aldehyde and norbixin were not modified in incubation with boiled enzymes or with protein extract from E. coli harboring empty vector. Thus, BoBADH and BonBMT represent bixin aldehyde dehydrogenase and norbixin methyltransferase, respectively, and support the reaction sequence proposed for bixin synthesis (Fig. 1B).
Because the seeds of B. orellana are the only natural source of industrially produced bixin, we investigated whether bixin could be produced in E. coli pACCRT-EB engineered to accumulate lycopene (17). We transformed E. coli-producing lycopene with the plasmid pUC19-LCD-BADH-nBMT coding for B. orellana lycopene cleavage dioxygenase, bixin aldehyde dehydrogenase, and norbixin methyltransferase (12). HPLC analysis of the lipid extract of transformed E. coli harboring pUC19-LCD-BADH-nBMT revealed the accumulation of a new derivative corresponding to bixin (Fig. 3A). No such change was observed in E. coli cells transformed with an empty vector control (Fig. 3B). The average production level of bixin in E. coli was 5 mg/g dry weight.
Bixin is one of the oldest pigments used by humans and is increasingly in demand because of the consumer ban on the chemically synthesized azo dye (2, 4). Bixin synthesis involves an unprecedented oxidative remodeling of lycopene, a common intermediate which is the precursor of beta-carotene, provitamin A. Given the feasibility of engineering bixin in a heterologous host such as E. coli, we assume that coexpressing the three cloned genes in sink organs such as tomato fruit, which accumulates massive amounts of the necessary precursor lycopene, should lead to an alternative and competitive source for natural bixin production.
References and Notes
1. E. Pichersky, D. R. Gang Trends Plant Sci. 5, 439 (2000).
2. P. Col[ins, Foods Ingred. Process. Int. 13, 23 (1992). 3. G. J. Lauro, Cereal Foods World 36, 949 (1991).
4. C. F. Timberlake, B. S. Henry, Endeavour 10, 31 (1986).
5. L Zechmeister, Cis-Trans Isomeric Carotenoids, Vitamins A, and Arylpolyenes (Springer-Veria, Vienna, 1962).
6. G. Britton, in Carotenoids, G. Britton, S. Laaen-jensen, H. Pfander, Eds. (Birkhauser Verlag, Basel, 1998), voL 3, pp. 13-147.
7. A. Z Mercadante, A. Steck, H. Pfander, Phytochemistry 46, 1379 (1997).
8. A. Z. Mercadante, A. Steck, H. Pfander, J. Agric. Food Chem. 45,1050 (1997).
9. S. H. Schwartz, B. C. Tan, D. A. Gage, J. A. D. Zeevaart, D. R. McCarty, Science 276, 1872 (1997).
10. S. H. Schwartz, X. Qin, J. A. D. Zeevaart, J. Not. Chem. 276, 25208 (2001).
11. F. Bouvier, C. Suire, J. Mutterer, B. Camara, Plant Cell 15, 47 (2003).
12. Materials and methods are available as supporting material on Science Online.
13. B. Watil[lon et al., Plant Mol. Biol. 36, 909 (1998). 14. D. S. Skibbe et al., Plant Mol. Biol. 88, 751 (2002).
15. J. R. Ross, K. H. Nam, J. C. D'Auria, E. Pichersky, Arch. Biochem. Biophys. 367, 9 (1999).
16. N. Dudareva et al., Plant Cell 12, 949 (2000). 17. N. Misawa et al., J. Bacteriol. 177, 6575 (1995).
18. We thank P. Harriman and A. Malek for DNA sequencing and M. Kumagai for helpful comments on the manuscript This work was supported by European Community grant QLK3-CT-2000-00809.
FLorence Bouvier,1 Odette Dogbo,2 Bilal Camara1*
1Institut de Biologie Molecuiaire des Plantes, CNRS, University Louis Pasteur, 67084 Strasbourg, Cedex, France. 2Unites de Formation et de Recherche Sciences de la Nature, (UFR SN), University d'AboboAdjame 02 Boite Postale 801, Abidjan 02, Me d'Ivoire.
*To whom correspondence should be addressed. Email: [email protected]
Supporting OnLine Material
www.sciencemag.org/cgi/content/full/300/5628/2089/
DC1
Materials and Methods
Figs. S1 to S3
References
1 April 2003; accepted 2 June 2003
Copyright American Association for the Advancement of Science Jun 27, 2003