Maize phytoene desaturase and ζ‐carotene desaturase catalyse a poly‐Z desaturation pathway: implications for genetic engineering of carotenoid content among cereal crops (2024)

Search for other works by this author on:

Abstract

Received 5 March 2003; Accepted 9 June 2003

Introduction

Carotenoids are a large class of yellow, red, and orange pigments derived from isoprenoids. These pigments are produced in certain bacteria and fungi, and by all plants and cyanobacteria, and are, furthermore, dietary essentials to animals. In plants, carotenoids are both primary and secondary metabolites. As primary metabolites, carotenoids serve as regulators of plant growth and development, as accessory pigments in photosynthesis, as photoprotectors preventing photo‐oxidative damage, and as precursors to the hormone abscisic acid (ABA). In higher plants, primary carotenoids are synthesized in chloroplasts by nuclear‐encoded enzymes (Kirk, 1967). Secondary (taxon‐specific) carotenoids are often synthesized in tissues containing plastids of alternative chemistries and architectures, such as corollar, radical and nectarial chromoplasts, endospermal amyloplasts, or tapetal elaioplasts, where they may serve ecological roles as antioxidants, ornaments and precursors to odorants.

For humans, the presence of carotenoids in the endosperm and fruits of crop plants adds to their nutritional value. Dietary carotenoids are essential precursors to vitamin A and to retinoid compounds needed in animal morphogenesis (Bendich and Olson, 1989; Lee et al., 1981). Specific dietary carotenoids, perhaps serving as antioxidants and free radical scavengers among other roles, promote health. For example, lycopene and carotenes have been shown to prevent prostate, breast and other cancers (Agarwal and Rao, 2000a, b; Nishino, 1998; Rao and Agarwal, 2000; Rock, 1997) as well as cardiovascular disease (Cooper et al., 1999; Dagenais et al., 2000; McDermott, 2000). Lutein and zeaxanthin promote macular longevity (McDermott, 2000). In lesser‐developed countries, avitaminosis A is correlated to increased mortality (Fawzi and Hunter, 1998; Semba et al., 1994; Sommer, 1993; Tomkins, 2001). Since different carotenoids have various roles in human and animal health and dietary deficiencies vary with location, the control of specific content among a variety of geographically distributed cereal staples and fodders is desirable.

Geometric isomer states of carotenoids add to a great diversity of structures and influence the biological activities of carotenoids, including intestinal absorption, tissue localization, and metabolic channelling (Bjerkeng and Berge, 2000; Holloway et al., 2000; Krinsky et al., 1990; Osterlie et al., 1999; Patrick, 2000). Progression of metabolites through the carotenoid biosynthetic pathway depends on the geometric isomer states of substrates and products, as carotene desaturases may be stereo‐selective, stereo‐sensitive, and stereo‐specific in their activities (Bartley et al., 1999; Beyer et al., 1989; Breitenbach et al., 1999). Structural isomer diversity and specific carotenoid accumulation are also modulated by carotene desaturases (Garcia‐Asua et al., 1998).

The accumulation of carotenes depends on a series of isoprene condensations, followed by a series of desaturations of the polyene chain and subsequent cyclization of the end structures; see Hirschberg (2001) for a review. Each of these steps has stereochemical aspects. In maize, there are multiple mutant loci that condition the accumulation of intermediate metabolites for each conversion. These mutants are valuable for dissecting regulation of the pathway and identifying stereochemical requirements (Neill et al., 1986; Robertson et al., 1966; Robertson, 1975; Treharne et al., 1966; Wurtzel, 1992). The substrate for the carotene desaturases, phytoene, is the condensation product of two geranylgeranyl pyrophosphate (GGPP) molecules. Phytoene synthase (PSY) introduces a central 15‐Z double bond during hydrogen elimination (Cunningham and Gantt, 1998). Phytoene then undergoes four sequential desaturation steps, via intermediates, neurosporene and zetacarotene (ζ‐carotene), to lycopene (Fig. 1). In plants and cyanobacteria, these steps are catalysed by two enzymes, phytoene desaturase (PDS) (Bartley et al., 1991) and ζ‐carotene desaturase (ZDS) (Albrecht et al., 1995), each mediating two symmetrically positioned desaturations (Fig. 1). Maize mutants viviparous5 (vp5), vp2 and white3 (w3) are associated with the function of PDS by the accumulation of phytoene and vp9 and y9 are associated with ZDS by the accumulation of ζ‐carotene. While PDS catalyses two E‐hydrogen eliminations at the 11 and 11′ carbons of the polyene hydrocarbon chain, the few characterized plant ZDS enzymes catalyse Z‐eliminations (Bartley et al., 1999; Beyer et al., 1989; Breitenbach et al., 1999) at the 7 and 7′ carbons. These few analyses of the geometric isomers of carotenes accumulating in vitro or in heterologous carotenogenic systems in E. coli using overexpressed plant desaturases have demonstrated that Z‐configured isomers of neurosporene, ζ‐carotene, and lycopene predominate. For example, the Arabidopsis carotene desaturases catalyse the production of poly‐Z‐lycopene as the major product through a poly‐Z‐desaturation pathway in a heterologous E. coli system (Bartley et al., 1999). The final product does not retain the Z‐configuration of its central double bond. Similarly, overexpressed Capsicum annuum ZDS catalyses the formation of 7,9,7′,9′‐tetra‐Z‐lycopene (poly Z‐lycopene), when fed 9,9′‐di‐Z‐ζ‐carotene in E. coli cell‐free extracts (Breitenbach et al., 1999). Surprisingly, Capsicum ZDS, the only plant ZDS activity successfully purified in vitro, exhibits both E‐ and Z‐ hydrogen elimination and is apparently stereosensitive to the geometric form of its substrate. That is, Capsicum ZDS introduces Z‐configured double bonds if the 9 and 9′ positions are Z‐configured, but introduces E‐configured double bonds, if the 9 and 9′ positions are E‐configured. Prior to the above studies and the present study, a poly‐Z‐desaturation pathway was only observed in isolated daffodil chromoplasts (Beyer et al., 1989) in the Tangerine variety of tomato (Clough and Pattenden, 1983; Tomes et al., 1953), and in mutants of Scenedesmus C‐6D (Ernst and Sandmann, 1988). In fungi and non‐photosynthetic bacteria, the four carotene desaturations are catalysed by a single enzyme, designated as the ‘CRTI’ type. CRTI only catalyses E‐eliminations (Linden et al., 1991), producing all‐E products. Thus, the bacterial carotene desaturase CRTI produces different geometric isomers than do the plant carotene desaturases, which, as a consequence, require a companion carotene isomerase (Isaacson et al., 2002; Park et al., 2002). This stereochemical difference in enzyme specificity is critical when considering genetic engineering efforts using CRTI.

Genetic modification for enhanced carotenoid accumulation or for specific or novel structures has been accomplished in bacteria (Rohlin et al., 2001), yeast (Shimada et al., 1998), and with mixed success in several crops (reviewed in Giuliano et al., 2000; Broun and Somerville, 2001; Hirschberg, 2001; Sandmann and Mitchell, 2001) including canola (Shewmaker et al., 1999), tomato (Rosati et al., 2000), and rice (Burkhardt et al., 1997; Ye et al., 2000). Carotene accumulation in transgenic rice (Ye et al., 2000) and tomato (Rosati et al., 2000) employed ectopic expression of CRTI. Rational metabolic engineering of the specific carotenoid content requires not only isolation and ectopic expression of genes for structural enzymes within the pathway, but also an understanding of many factors affecting carotenogenesis. Recently, much progress has been made towards elucidating these factors, which include: (1) rate‐control of flow through isoprenoid precursors pools (Gallagher et al., 2003; Matthews and Wurtzel, 2000), (2) ancillary co‐factors and ancillary enzyme activities controlling carotenoid accumulation (Li et al., 2001; Norris et al., 1998), (3) the topology and assembly of the membrane‐bound carotenoid complex (Cunningham and Gantt, 1998), (4) the substrate stereospecificities of enzymes (Breitenbach et al., 1999), and (5) requirements for the accumulation and sequestration of products (Cervantes‐Cervantes et al., 1990; Li et al., 2001). These factors may be specific to species and plastid types.

The isolation of a maize cDNA by homology to a partial rice cDNA and its functional assignment as ZDS is reported here. In additional, the chemical complementation among maize mutants affecting desaturation associated another locus (vp2) with the biosynthesis of quinones essential to electron transport from the desaturations. While these studies have associated some known morphological loci with functional cDNAs, other desaturation‐associated loci, such as y8 and y9, have yet to be cloned or assigned specific functions. This functional demonstration of poly‐Z‐desaturations catalysed by the concerted action of maize PDS and ZDS allow the possibility of an isomerase activity for these as yet functionally‐unassigned loci. The implications for the metabolic engineering of specific carotenoid accumulation in cereals are discussed.

Materials and methods

Plant material

Rice (Oryza sativa) IR36 and maize (Zea mays) inbred line B73 were grown in the greenhouse (20–25 °C). Green shoots were collected after 7–10 d, immediately frozen in liquid nitrogen, and stored at –80 °C until used. Maize mutants viviparous9‐Mutator (vp9‐Mum), vp2, vp5, vp7, white3 (w3), and yellow9 (y9) were grown in the experimental field at Pelham Bay Park, Bronx, New York, while y8 was grown in the greenhouse. Mutant and normal endosperms of vp9‐Mum,y9 and y8 were dissected from maize harvested at 20 DAP (days after pollination), frozen in liquid nitrogen and stored at –80 °C until used.

Cloning and sequencing

A rice EST clone (GenBank D48291) showing 75% nucleotide identity to ArabidopsisZds was fully sequenced (GenBank AF054629) and found to contain a 1632 bp cDNA insert having 74% nucleotide identity and 81% deduced amino acid identity (94% similarity) to the ArabidopsisZds cDNA, indicating that the EST encoded a putative rice ZDS. Two primers were designed based on a rice EST sequence: primer 5′‐GCATTCCTACGAACTAACCAACTCAAGG‐3′, and 5′‐GGTGCTTACATAGAGACCTCCCTACACC‐3′. A 417 bp RT‐PCR product was amplified from rice IR36 leaf RNA (Logemann et al., 1987) as follows: 1 µg total RNA was used as template for cDNA first strand synthesis (Gibco‐BRL SuperScript™ preamplification system), then, 1 µl of cDNA product (c. 0.2–0.8 ng), was utilized in a 25 µl PCR reaction (20 mM TRIS‐HCl, pH 8.4, 50 mM KCl, 2.5 mM MgCl₂, 0.2 mM of each dNTP, 0.5 µM of each primer, 0.04 U µl^–1Taq polymerase (Gibco‐BRL, Gaithersburg, MD). The PCR conditions were as follows: 1 cycle at 94 °C (3 min); 35 cycles at 94 °C (30 s), 68 °C (30 s), 72 °C (30 s); and one cycle at 72 °C (10 min). After DNA sequence confirmation, the 417 bp RT‐PCR product was used as a hybridization probe to screen 1 000 000 clones of a λgt11 maize endosperm cDNA library (Fontes et al., 1991) according to Sambrook et al. (1989). Hybridization was performed in 40% (v/v) formamide, 6× SSC, 1× Denhardt’s solution, and 1 mM EDTA, at 42 °C overnight; filters were washed twice with 2× SSC at room temperature and twice with 0.1× SSC and 0.1% (w/v) SDS at 50 °C, 15 min. Eight positive λ‐phage clones, whose insert sizes varied from 1.4 kb to 2.2 kb, were obtained. The 2.2 kb EcoRI insert was subcloned into the EcoRI site of the pBluescript II SK (–) vector and the resulting plasmid was named pMzds48. A plasmid insert deletion series was produced by double digesting pMzds48 with SalI and KpnI followed by 5′‐deletion using the Exonuclease III and Mung Bean Nuclease Deletion Kit (Stratagene, La Jolla, CA). Sequencing of one strand was carried out manually by using the Sequenase Version 2.0 DNA Sequencing Kit (United States Biochemical Corporation); the second strand was sequenced in an automated sequencer (Molecular Biology Core, Oregon Regional Primate Research Center, Beaverton, OR 97006). The resulting maize Zds sequence was deposited as GenBank AF047490 (Luo and Wurtzel, 1999). Other Zds sequences used for comparisons were plant‐type, from Capsicum (GenBank 2129927); Narcissus (GenBank AJ224683); and bacterial‐type, from Anabaena (GenBank D26095). The maize Pds sequence used for comparison was GenBank U37285. Sequence comparisons were carried out using BLAST programs (Altschul et al., 1997; Tatusova and Madden, 1999) and the Align program of VectorNTi (InforMax Inc., Bethesda, MD).

Expression and functional complementation of maize Zds

To test for functional complementation of the maize Zds cDNA, the plasmid pACCRT‐EBP, conferring ζ‐carotene accumulation, was first constructed. A 2.0 kb PvuII‐PvuII fragment containing a maize Pds cDNA was removed from pMPDS3 (Li et al., 1996), both ends were filled in with Klenow and inserted into the BamHI site (filled in with Klenow) of pACCRT‐EB (Linden et al., 1991). The resulting plasmid pACCRT‐EBP carries Erwinia uredovora crtE encoding GGPP synthase, and Erwinia uredovora crtB encoding phytoene synthase, and maize Pds encoding phytoene desaturase. pMzds48 was digested with NotI, the 5′‐end filled in with Klenow and digested with XbaI. A 5′ deletion from XbaI to the 5′ end of maize Zds was performed with the Exonuclease III and Mung Bean Nuclease Deletion Kit (Stratagene, La Jolla, CA). One of the pMzds48‐deleted plasmids, pMzds‐107 from which 106 bp had been deleted, was confirmed to be in frame with lacZ. E. coli XL‐1 Blue (Stratagene, La Jolle, CA) cells carrying pACCRT‐EBP were transformed with pMzds‐107. Ampicillin (50 µg ml^–1) and chloramphenicol (70 µg ml^–1) were used to select transformants containing both pMzds‐107 and pACCRT‐EBP, since pMzds‐107 carries only an Amp^R marker and pACCRT‐EBP possesses only a Chl^R marker. Co‐transformed cells were grown in 1000 ml rotary shaker (260 rpm) cultures in Luria–Bertani broth pH 7.2 for 8 h at 37 °C, induced with 10 mM IPTG at late‐log growth phase, and then further incubated for 46 h at 26 °C with mild agitation (100 rpm). Since carotenoids have been shown to continue to accumulate in stationary‐phase cultures (Kim and Keasling, 2001; Matthews and Wurtzel, 2000), stationary‐phase cultures were held in the dark for 2 d at room temperature before extraction of total carotenoids. Control cells with either pACCRT‐EBP, producing ζ‐carotene, or pACCRT‐EIB/pTAC‐ORF2 (Matthews and Wurtzel, 2000), producing lycopene or pAC‐Neurosporene (Cunningham et al., 1996), producing neurosporene, were grown similarly with appropriate antibiotics. Pigmented cells were harvested by centrifugation. Cell pellets were exhaustively extracted with three changes of acetone and homogenization in a Brinkman Polytron (Kinematica, Switzerland) at 30 000 g. Acetone extracts were made 2.5% NaCl and carotenoids were partitioned into an equal volume of diethyl ether. The diethyl ether was dried over anhydrous NaSO₄, and evaporated in a SpeedVac (Savant, Farmingdale, NY) evaporator. Samples were subjected to HPLC immediately, or when necessary, stored under an inert atmosphere at –80 °C. All manipulations of cells and pigments were carried out in the dark or very dim light. Maize ZDS did not function efficiently in this heterologous complementation system as shown by a large accumulation of ζ‐carotene and a small accumulation of poly‐Z‐lycopene and Z‐neurosporene intermediates. Low activity may be due to problems intrinsic to the heterologous system, such as choice of bacterial strain, the availability of appropriate substrate, metabolic burden of high copy‐number plasmids or timing of induction of transgene products. Indeed, numerous laboratories have attempted to define and improve the intrinsic components of the heterologous carotenogenic systems (Harker and Bramley, 1999; Jones et al., 2000; Kajiwara et al., 1997; Kim and Keasling, 2001; Matthews and Wurtzel, 2000; Ruther et al., 1997; Wang et al., 1999; Wurtzel et al., 1997). Based on these leads, attempts to increase poly‐Z‐lycopene accumulation included late‐growth‐phase promoter induction for ZDS and extended stationary‐phase incubation, both of which were necessary to improve overall total carotenoid production sufficiently to allow definitive determination of the final product, poly‐Z‐lycopene. Expression of ZDS from a low‐copy number plasmid (Bartley et al., 1999; Jones et al., 2000; Wurtzel et al., 1997), as a general strategy to increase final end‐product accumulation, was not attempted in this study.

A Waters (Millipore, Franklin, MA) HPLC system with 2690 separation module, 996 photodiode array detector, and a 717 autosampler was used to identify carotenoids by reverse phase chromatography on a Nucleosil 5 C₁₈ (5 µ, 250×4.6mm) column (Phenomenex, Torrance, CA). Extracts were resuspended in methanol and injected in a mobile (0.8 ml min^–1) phase of 100% acetonitrile (HPLC System I). Elution profiles were monitored at 400 and 450 nm and selected peaks collected, evaporated and resuspended in methanol for re‐injection into a mobile (1 ml min^–1) phase of 100% methanol (HPLC System II).

RT‐PCR analysis of Zds expression in maize endosperm

RNA extraction and concentration estimates were accomplished according to Li et al. (1996). One µg total RNA was used as template for first strand cDNA synthesis (Gibco‐BRL Superscript™ preamplification system) in a 20 µl reaction. Afterwards, 1 µl (c. 0.2– 0.8 ng) cDNA product was used in a 25 µl RT‐PCR reaction containing 20 mM TRIS‐HCl pH 8.4, 50 mM KCl, 2.5 mM MgCl₂, 0.2 mM of each dNTP, 0.5 µM of each primer; and 0.04 U µl^–1Taq polymerase (Gibco‐BRL). RT‐PCR primers for amplifying maize Zds (GenBank AF047490) were primer number 279 (bp 1408–1426), 5′‐GGTTGGGTTGGATAACCTT‐3′ and primer number 262 (bp 1712–1695), 5′‐TCTGATCAGGCCTGAATG‐3′; and for amplification of maize Pds (GenBank U37285), were primer number 110 (bp 1377–1396), 5′‐GGAACTGTGAAACACTTCGC‐3′ and primer number 111 (bp 1904–1885), 5′‐GAAACCTTCGATAGGTGACC‐3′. The PCR conditions for Zds were 1 cycle at 94 °C (3 min); 35 cycles at 94 °C (30 s), 68 °C (30 s), 72 °C (1 min); and one cycle at 72 °C (10 min). The PCR conditions for Pds were 1 cycle at 94 °C (3 min); 40 cycles at 94 °C (30 s), 52 °C (30 s), 72 °C (30s); and one cycle at 72 °C (10 min). RT‐PCR products were applied to a 1.6% (w/v) agarose gel, at 100 V for about 2 h. For amplification of genomic DNA, 0.1 µg of maize B73 genomic DNA was used as a template in 25 µl reactions; PCR conditions were as described above.

TLC and HPLC analyses of normal and mutant maize endosperms

Carotenoids for TLC and HPLC analyses were extracted by grinding 0.5–1 g mutant endosperm, dissected at 20 DAP, to a powder form in liquid nitrogen, followed by the addition of 10 ml of acetone, incubation at 65 °C for 30 min, then centrifugation at 10 000 g for 15 min. The supernatant was collected and reduced to 1–2 ml by blowing nitrogen. For TLC analysis, the mobile phase was ethyl ether/hexane, 60:40 (v/v), and the stationary phase was silica Gel 60 F₂₅₄ (Kieselgel 60 F₂₅₄, Merck, Darmstadt, Germany). One ml of the final extract was applied in multiple applications on the TLC plates, air‐drying between each application. For HPLC analysis, 50 µl of the extracts was applied to a C₁₈ column (25 cm × 4.6 mm SphereClone ODS (1) (Phenomenex, Torrance, CA) using acetonitrile/H₂O (9:1, v/v) and 100% ethyl acetate in a 25 min linear gradient at 1 ml min^–1 flow rate. The HPLC system was the same as described above.

Chemical complementation of maize mutants

Developing seeds, harvested at 20 DAP, were surface‐sterilized for 5 min in a solution of 10% (v/v) household bleach and 10% ethanol (v/v), and rinsed several times with sterilized water. Embryos were dissected under a sterile hood and planted in Murashige and Skoog medium (cat. 11117–066, Gibco‐BRL), supplemented with sucrose at 20 g l^–1 and varying concentrations (0, 0.001, 0.01, 0.1 or 1 mM) of homogentisic acid (HGA) or 4‐hydroxyphenylpyruvate (OHPP). Twenty seeds of each maize mutant at each condition were grown for 1 week in a 26 °C growth chamber, with a cycle of 12/12 h light/dark.

Results

Isolation and characterization of maize Zds cDNAs

A rice EST clone (GenBank D48291) was fully sequenced (GenBank AF054629) and found to contain a 1632 bp cDNA insert having 74% nucleotide identity and 81% deduced amino acid identity (94% similarity) to the ArabidopsisZds cDNA (GenBank U38550). Primers were designed according to the rice EST sequence, a 417 bp RT‐PCR product from rice leaf mRNA was amplified and was used to probe a maize endosperm cDNA library. Eight positive clones were identified. Each clone hybridized to maize genomic DNA digested with HindIII, EcoRI or BamHI with the same restriction fragment pattern (data not shown). The longest, 2265 bp, clone was subcloned into pBluescript II SK (–), designated as pMzds48, and used for further sequencing and functional analysis. This maize Zds cDNA (GenBank AF047490) contains untranslated 5′‐ and 3′‐ regions of 205 and 350 bp, respectively, and an open reading frame predicted to encode a 570 amino acid, 63.1 kDa protein based on the position of the first ATG codon at bp 206 and a stop codon at bp 1916. There are no stop codons upstream of the first ATG, and based on sequence comparisons with other ZDS sequences (Fig. 2), it is unlikely that there is another upstream methionine codon. Using the maize ZDS sequence as a guide, the gene and deduced amino acid sequence of rice ZDS was deduced from Rice Genome Project chromosome 3 nucleotide sequence data (GenBank NC_003074) and submitted as TPA BK000417. Similarly, the gene and protein of rice PDS was manually deduced from GenBank AC079633.

The deduced amino acid sequences of maize and rice ZDS proteins were compared with other selected carotene desaturase amino acid sequences (Fig. 2). Based on amino acid and nucleotide primary structure homology, the maize ZDS shows higher homology to the sequence of the plant‐type rather than the bacterial (CRTI)‐type gene (nucleotide comparison not shown). PDS and ZDS are also distantly related to amino oxidases and to the carotenoid isomerase CRTISO. A phenogram of the alignment in Fig. 2 shows that maize and rice PDS proteins and maize and rice ZDS proteins are highly homologous to each other, while the ZDS for the monocotyledonous plant daffodil (Narcissus) is no more closely related to the grass carotenoid desaturases than are ZDS proteins from the selected dicotyledonous plants Arabidopsis and Citrus. Interest ingly, CRTISO is no more related to PDS or ZDS than CRTI: CRTISO is an anciently diverged member of the gene family. The deduced amino acid sequence of maize ZDS revealed a typical dinucleotide binding domain, represented by GenBank PfamA01494, that was present in maize PDS (Li et al., 1996). The similar enzymatic function of ZDS and PDS was reflected in the sequence similarity of the mature proteins, which showed an amino acid identity of about 32% and a similarity of 50%. A putative transit sequence for plastid targeting was predicted to be minimally the N‐terminal residues 1 to 30 by comparing the maize ZDS sequence with that of the cyanobacterium Synechocystis (GenBank D90914) and by concordance to the predicted transit peptides of the monocotyledonous daffodil, and bell pepper (Albrecht et al., 1995) and Arabidopsis (Scolnik and Bartley, 1995), which are 61, 49 and 59 residues, respectively. The computer algorithm ChoroP 1.1 (Emanuelsson et al., 1999) predicts a transit peptide cleavage site between residue 30V and 31R (see arrow, Fig. 2). The predicted molecular mass of the mature maize ZDS polypeptide found in plastids is about 56 kDa.

Gene structure predictions for rice PDS and ZDS

Completion of the rice and Arabidopsis genome sequences allows multiple comparisons of Pds and Zds gene structures for the first time. Figure 3 shows the deduced introns and exons of rice Pds and Zds compared to those of Arabidopsis. Intron/exon boundaries are perfectly conserved for both genes with the exception of ArabidopsisZds exon 9, which occurs as two exons, 9 and 10, in rice. The distantly related Pds and Zds genes show no exon/intron boundary conservation among rice and Arabidopsis.

Functional analysis of ZDS and of the concerted activity of maize PDS and ZDS

Function of the maize Zds cDNA gene product was confirmed by heterologous expression in E. coli and analysis of accumulated carotenoids by HPLC separation and photo diode array (PDA) detection. First, E. coli cells were transformed with pACCRT‐EBP, which carries genes coding for GGPPS (GGPP synthase), PSY from Erwinia uredovora, named crtE and crtB, respectively; and the maize PDS gene. Together these three genes conferred accumulation of ζ‐carotene in the E. coli host cells under chloramphenicol selection (Li et al., 1996). Next, bp 1–106 was deleted from the 5′ end of the Zds cDNA insert of pMzsd48 to produce an in‐frame translational fusion with lacZ. The expression clone was designated as pMzds‐107. To confirm the enzyme activity coded by the truncated Zds cDNA, cells carrying pACCRT‐EBP were transformed with pMzds‐107 and the accumulation of specific carotenoids was determined. This system was designed not only to demonstrate the function of the novel maize Zds cDNA, but also to examine the products of the maize desaturase pair, PDS and ZDS, acting together.

The function of maize ZDS in E. coli already accumulating ζ‐carotene from the function of maize PDS (Li et al., 1996) was expected to confer the further accumulation of neurosporene and lycopene, the intermediate and final products of ζ‐carotene desaturation. Figure 4 shows the HPLC chromatogram of pigments separated with HPLC System I. In addition to a large accumulation of the product of PDS, all‐E‐ζ‐carotene and 15‐Z‐ζ‐carotene (Fig. 4, peaks 1a, b), new peaks corresponding to neurosporenes (Fig. 4, peaks 2a, b, c) and poly‐Z‐lycopene (Fig. 4, peak 3) were identified based on characteristic λ_Max from chromatographic spectra and shown in Table 1A along with standards. The major product peak (peak 3) was collected from HPLC System I and re‐injected into HPLC System II to be purified further and to confirm the identification of poly‐Z‐lycopene as the major product. Figure 5 shows the elution profile of the major products in System II and the separation of neurosporene isomers (peak 1) from poly‐Z‐lycopene (peak 2). While recovery of neurosporene isomers was too little to determine spectral properties accurately, the amount of poly‐Z‐lycopene, Fig. 5A, was small but sufficient to determine a PDA spectrum definitively (Fig. 5B). Characteristic λ_Max from chromatographic spectra are shown in Table 1B along with available standards separated in the same system. The unmistakable spectrum of poly‐Z‐lycopene (Fig. 5B) is identical to previously published spectral profiles (Clough and Pattenden, 1983; Beyer et al., 1989; Sandmann, 1993; Bartley et al., 1999; Breitenbach et al., 1999). Taken together, these results prove the function of the newly isolated Zds cDNA and show that maize PDS and ZDS together catalyse a poly‐Z‐desaturation pathway in E. coli.

Among higher plants, fungi and bacteria, homologous desaturases catalyse from two to five desaturations (Hausmann and Sandmann, 2000). Therefore, it was tested whether maize ZDS might use phytoene as a substrate. Cells accumulating phytoene were generated by transforming E. coli with pACCRT‐EB, a plasmid carrying the Erwinia genes crtE, coding for GGPPS; and crtB, coding for phytoene synthase. Cells containing pACCRT‐EB were transformed with pMzds‐107, but only phytoene was detected by HPLC, whether or not the pMzds‐107 plasmid was present (data not shown). Therefore, maize ZDS does not act on the substrate phytoene. Since the activity of maize ZDS does not overlap with that of PDS, and a system containing both PDS and ZDS produces lycopene, maize ZDS must mediate two desaturations only.

Although the appearance of the expected metabolites indicates that the maize Zds cDNA encodes a functional enzyme, little (OD₄₄₁=0.05) poly‐Z‐lycopene product accumulated, and much (OD₄₀₀=2.9) ζ‐carotene substrate remained, suggesting that maize ZDS, unlike maize PDS (Li et al., 1996), functioned inefficiently in this E. coli system. The effect of the transit peptide and the timing of transgene induction on ZDS efficiency were both investigated. In order to test whether the presence of a plastid transit signal attenuated enzyme activity, a construct was made by deleting the first 392 bp of the cDNA insert of pMzds48. This deletion removes the first 62 of 64 codons of the putative transit peptide. Co‐transformation of E. coli cells with this construct and pACRRT‐EBP did not increase neurosporene or lycopene accumulation (data not shown). This result indicates that the inefficient production of lycopene by maize ZDS expressed in E. coli is not due to the presence of the transit peptide, but some other factor. The timing of transgene induction has been suggested to be critical to the accumulation of carotenoids in heterologous complementation systems (Matthews and Wurtzel, 2000; Ruther et al., 1997). Induction of PDS and ZDS accumulation, both of which were transcribed under the control of the lacZ promoter, was attempted at various stages of bacterial culture growth, namely: inoculation, early log, midlog, and late‐log phase. Late‐log phase induction led to the greatest accumulation of total carotenoids in liquid culture, but not an increase in the relative amount of final product, lycopene (data not shown).

Southern hybridization and RFLP mapping of maize Zds

To determine how many copies of the Zds gene are present in maize, pMzds48 was hybridized to maize B73 genomic DNA digested with several restriction enzymes (Wurtzel et al., 1987). Although two or three fragments were found in each genomic DNA digested with either XhoI, HindIII, EcoRI or BamHI, KpnI digestion resulted in only one fragment, suggesting only one copy of Zds in maize. Hybridization of either the maize Zds cDNA or the rice EST (GenBank D48291) against rice genomic DNA also revealed single bands, suggesting one gene (data not shown), which was subsequently corroborated by rice genome sequence analysis. Together these results suggest that Zds may be a single copy gene in maize as it is in rice.

Maize Zds was associated with chromosome markers by RFLP analysis using a Tx303× CO159 population (UMC, University of Missouri‐Columbia Maize RFLP laboratory, MO) and a T232×CM37 population (B Burr, Brookhaven National Laboratory, Upton, NY) (Burr et al., 1988). According to RFLP analysis of the Tx303×CO159 population, Zds mapped to the short arm of chromosome 7 (7S) in bin 7.02 between asg34 and asg49, which flank vp9, a locus associated with deficiency in ζ‐carotene desaturation. Linkage mapping using the T232×CM37 population gave similar results, positioning Zds on 7S closely linked to o2 and in1. While o2 and in1 flank y8, vp9 is about five map units proximal to o2 and in1 (http://www.agron.missouri.edu).

Transcript accumulation of Zds in the maize mutants vp9‐Mum, y8, and y9 and their normal endosperm counterparts

Based on the RI‐RFLP mapping results, it was suspected that either phenotypic mutant, y8 or vp9, might represent a mutation of the ZDS structural gene. To test whether either y8 or vp9 might code or regulate Zds, steady‐state transcript levels were compared in single normal (yellow) and mutant (white or light yellow) endosperms segregating on a single ear for either y8 or vp9‐Mum (a Mutator‐induced allele of vp9) self‐pollinated plants. Normal and mutant endosperms of y9 were also included in the analysis, since homozygous y9 endosperms were reported to accumulate ζ‐carotene (Robertson, 1975; Janick‐Buckner et al., 2001). The locus y9 was not a candidate for the ZDS structural gene, since y9 mapped to chromosome 10S and Zds mapped to 7S, but could encode a trans‐acting regulator of ζ‐carotene desaturation or Zds transcript accumulation. RNA was extracted from normal and mutant endosperms dissected from individuals segregating for one of the three mutant alleles and used as a template for amplification of the Zds transcript by RT‐PCR. Primers were designed to flank an intron, such that amplification from genomic DNA (540 bp product) was distinctly different than amplification from cDNA (303 bp product) (Fig. 6A, compare lanes 1 and 2). No products were obtained when reverse transcriptase or either primer were omitted from the RT‐PCR reaction (Fig. 6A, lanes 3–5). Zds transcripts were amplified from each of the three mutant endosperms and compared with amplification of cDNA produced from their normal, segregating counterparts (Fig. 6B). A reduction in amplification products occurred only in the vp9‐Mum material when compared to the transcript levels in the normal endosperm. No difference in Pds amplification products were seen in any of the three mutant genotypes tested (Fig. 6B, bottom), confirming that the integrity of the amplification was as previously reported (Li et al., 1996). Among the carotenoid‐deficient mutants compatible with linkage mapping results or reported to accumulate ζ‐carotene, only vp9‐Mum is associated with Zds by altered transcript accumulation.

Carotenoid content of maize mutants vp9‐Mum, y8, and y9 and their normal endosperm counterparts

Both TLC and HPLC were used to confirm the accumulation of ζ‐carotene in each of the above mutants. The TLC results shown in Table 2 indicate that both y9 and vp9‐Mum, but not y8, endosperms had a pigment matching the R_F value for ζ‐carotene isolated from E. coli cells carrying pACCRT‐EBP, the plasmid conferring ζ‐carotene accumulation. The normal segregating endosperms did not accumulate ζ‐carotene, as expected, but instead accumulated a compound identified as a xanthophyll. Further confirmation of these results was obtained from the HPLC analysis shown in Fig. 7. In extracts from the vp9‐Mum and y9 mutants, ζ‐carotene (peak 1) was detected at 13.0 min, the same retention time as the standard, but no corresponding peak was found for the extract from y8 mutant endosperms. The xanthophyll (peak 2), with a retention time of 6.5 min, was present only in normal endosperms of vp9‐Mum, y9 and y8 endosperms. Therefore, the HPLC results confirm the TLC results and together indicate that y8 shows no relation to ζ‐carotene desaturation, whereas vp9 and y9 are associated with the ζ‐carotene desaturation steps.

Chemical complementation of the maize mutants vp2, w3, vp5, vp7, and vp9

Since vp9 is likely to be the ZDS structural gene, it is unclear what the role of y9 is in ζ‐carotene desaturation. Similarly, for desaturation of phytoene to ζ‐carotene, there are also loci that can be attributed to the PDS structural gene (e.g. vp5), and others (e.g. vp2, w3) which do not regulate transcript accumulation, but for which recessive alleles may confer phytoene accumulation (Li et al., 1996). It is possible that some of these loci might regulate or control the biosynthesis of plastoquinones, known participants in the electron transfer reactions associated with carotene desaturation steps. For example, in Arabidospsis, pds1 conditions phytoene accumulation due to a block in the biosynthesis of plastoquinones, specifically at the step mediated by 4‐hydroxyphenylpyruvate dioxygenase, an enzyme catalysing the synthesis of homogentisic acid (HGA) from 4‐hydroxyphenylpyruvate (OHPP) (Norris et al., 1995, 1998). A chemical complementation experiment was carried out to determine whether any of the maize desaturation mutants were blocked at this step of the plastoquinone biosynthetic pathway. As negative controls, vp7 (that confers lycopene accumulation) and vp5 were included. Unfortunately, the y9 mutant could not be tested as it affects only the endosperm phenotype. Mutant embryos homozygous for vp2, w3, vp5, vp7, or vp9 (20 each) were placed in medium containing various amounts of HGA or OHPP. In medium without either chemical, all of the mutant embryos germinated into albino seedlings or into pink seedlings in the case of vp7 (Table 3). However, in medium supplemented with, 0.1 or 1 mM HGA, the vp2 embryos germinated into pale green and green seedlings, respectively. No other mutant tested showed this effect; regardless of the amount of OHPP added (0–1 mM), all of the seedlings retained their mutant phenotype. Ten vp2 mutant plants germinated with 0.1 or 1 mM HGA that had become pale green or green were transferred to medium lacking HGA. After 3–4 d, the seedlings gradually reverted to an albino phenotype. Therefore, the HGA deficiency causes vp2 mutant plants to be albino. The results indicate that the vp2 locus, but none of the other loci tested, is involved in the biosynthesis of plastoquinones at the step leading to biosynthesis of HGA.

Discussion

In order to analyse the maize carotene desaturase pair, PDS and ZDS, functionally and completely, a cDNA encoding a ZDS has been isolated from a maize endosperm library. The maize cDNA and the corresponding partial Zds cDNA from rice were sequenced. The first desaturase in the series, PDS, had been previously isolated from maize endosperm and characterized as a functional cDNA (Li et al., 1996) and associated with the genetic locus vp5 (Hable and Oishi, 1995; Li et al., 1996; Hable et al., 1998) in maize. Southern blot analyses suggested both PDS and ZDS are single loci in both rice and maize.

The maize Zds cDNA was functionally tested in E. coli and demonstrated by the profiling of carotene intermediates and products to encode a desaturase that catalyses two sequential di‐dehydrogenations at symmetrically located positions on the substrate ζ‐carotene. ZDS produced poly‐Z‐lycopene via Z‐neurosporene. Employing an alternate complementation system, it was demonstrated that maize ZDS uses only ζ‐carotene as a substrate, and will not catalyse the desaturation of phytoene. Thus, maize ZDS, like ZDS in other higher plants, can only catalyse two desaturation reactions from ζ‐carotene, not four or five desaturations as does the bacterial or fungal CRTI. The maize carotene desaturase pair, PDS and ZDS, acting together in a heterologous system also differs from CRTI in that the PDS/ZDS pair catalyses a ‘poly‐Z‐desaturation pathway’, rather than processing and producing all‐E geometric isomers, as does CRTI. Cloning and functional expression of maize PDS (Li et al., 1996) with maize ZDS has allowed a poly‐Z‐desaturation pathway in an E. coli system to be demonstrated. This finding extends the possibility of a general distribution among species of the poly‐Z‐desaturation pathway operating in planta to maize. A di‐Z didehydrogenation activity for maize ZDS combined with the observation that most plant lycopene cyclases are stereoselective for all‐E lycopene implies an obligate isomerase associated with maize carotene desaturations. The geometric isomer states of the desaturation products as well as the stereo‐specificity of the desaturases themselves may also be influenced by other co‐factors, architecture, and interactions of the carotenoid biosynthetic enzymes in a plastid membrane environment.

Association of the ZDS structural gene with a chromosomally mapped genetic lesion, vp9, narrowed the candidate loci that define other functions and cofactors in maize carotenogenesis. Using a combination of RI‐RFLP analysis, RT‐PCR semi‐quantification of transcript accumulation, and HPLC analysis of maize carotenoid mutants, it was possible to characterize the potential function of several of the maize genetic loci that are linked to the carotenoid desaturation steps further. Analysis of gene associations in each of two RI‐RFLP mapping populations suggested two different loci, Y8 and Vp9, as candidates for the ZDS structural locus. It was confirmed that vp9‐Mum and y9 mutants accumulate ζ‐carotene, while y8 is completely deficient in the carotenoid metabolites and thus does not accumulate any intermediates related to carotene desaturation. Since mapping data indict either Y8 or Vp9 as the ZDS structural gene, but y8 endosperm does not accumulate carotenoid precursors or intermediates of the desaturation series, it is likely that Y8 is not the ZDS structural locus. Since the Zds cDNA maps onto or near vp9, a recessive allele that conditions ζ‐carotene accumulation and no other candidate loci map to this chromosomal region, Vp9 is a good candidate locus for ZDS. A decrease in Zds transcript accumulation was found in vp9‐Mum but not y8 endosperm as measured by RT‐PCR. While this result corroborates the assertion of Vp9 as the structural locus of ZDS, it does not rule out the possibility that Vp9 is a modulator of Zds transcript accumulation, acting on another hitherto undetected structural locus within the chromosomal mapping range. Given the obvious, unpigmented phenotype of desaturation deficiencies, it is unlikely that another such locus would not have been detected and mapped. It was also shown that y9 is not a candidate for the ZDS structural locus based on mapping results with the functional Zds cDNA. Moreover, y9 does not affect Zds transcript accumulation, but the recessive allele does condition accumulation of ζ‐carotene. From these results, it appears that Y9 may have some other role influencing the desaturation process.

ZDS did not function efficiently in a heterologous complementation system where PDS did function efficiently. Suggested reasons for dim activity of carotenoid desaturases in similar systems include (1) endproduct inhibition of ZDS (Breitenbach et al., 1999; Sandmann and Kowalczyk, 1989), (2) failure to reconstitute the interactions of an assembly complex responsible for substrate channelling (Breitenbach et al., 1999), and (3) the absence of a carotene isomerase (Bartley et al., 1999), among others. Recently, carotene isomerases associated with the desaturations have been isolated from algae (Breitenbach et al., 2001; Masamoto et al., 2001) and from higher plants (Isaacson et al., 2002; Park et al., 2002). For example, the deficiency of an obligate isomerase, coded by the tangerine locus in tomato, conditions the accumulation of poly‐Z‐lycopene in fruits (Hirschberg, 2001), given that lycopene beta cyclase (LCY) is stereoselective for all‐E lycopene in most plants (Cunningham et al., 1993, 1996; Pecker et al., 1996; Ronen et al., 1999), with one exception, daffodil LCY Beyer et al. (1994). Bartley et al. (1999) have suggested that an isomerase is missing from the heterologous complementation system used to examine Arabidopsis ZDS, and this deficiency, coupled with putative stereoselectivity of ZDS, accounts for the inefficiency of ZDS observed in this system. 15‐Z ζ‐carotene is the major product in the heterologous maize PDS and ZDS system reported here as well as in the Arabidopsis system (Bartley et al., 1999). Hence, the demonstration of a poly‐Z‐desaturation pathway for maize enzymes considered together with the inefficiency of ZDS in this study’s system adds the likelihood of (1) an isomerase and (2) a putative stereoselectivity of maize ZDS to the litany of many factors that affect the progression of substrates through the carotenoid biosynthetic pathway.

More than one isomerase activity may be implied by the finding of a poly‐Z‐pathway for maize PDS and ZDS. Isomerase activity associated with the carotene desaturases is implied by two lines of evidence: (1) photostimulation of ZDS activity and (2) stereoselectivity of LCY in the presence of apparent default poly‐Z eliminations. Because geometric isomers of carotenes may be photoconverted by light, with lower energy E‐conformations being more probable, light may substitute for an isomerase. Conversely, an isomerase may be necessary for geometric isomer transitions in the dark. Photostimulation of carotene desaturation was originally found for the Scenedesmus mutant C‐6D (Sandmann, 1991). Recently, the apparent photostimulation of Arabidopsis ZDS was posited as compelling evidence for the stereoselectivity of Arabidopsis ZDS for 15‐E ζ‐carotene, given that 15‐Z ζ‐carotene accumulated in dark‐grown cultures (Bartley et al., 1999). Treatment of dark‐grown cells with light unblocked the progression of the pathway through ZDS and poly‐Z‐lycopene accumulated. In addition, 15‐Z ζ‐carotene was photoconverted in vitro to all‐E ζ‐carotene by light. No such photostimulation was seen for Capsicum ZDS, while 15‐Z ζ‐carotene accumulated (Breitenbach et al., 1999). In these experiments, no photostimulation of maize ZDS was observed, while 15‐Z ζ‐carotene accumulated. The lack of photostimulation of maize ZDS in this system does not, however, preclude a stereoselectivity of maize ZDS for its substrate. Given two lines of evidence for an isomerase associated with the carotene desaturations, one associated with the accumulation of 15‐Z ζ‐carotene and the other with the accumulation of poly‐Z‐lycopene, it is possible that at least two distinct isomerase activities exist. The distribution of these activities potentially varies among species or plastid types. Since only one carotene isomerase has been identified within algae (Breitenbach et al., 2001; Masamoto et al., 2001) and within plants (Isaacson et al., 2002; Park et al., 2002), a most parsimonious model posits a single enzyme with multiple substrates. Indeed, one such hypothetical scheme for multiple isomerase activities has been proposed for the recently isolated algal carotene isomerase, CrtH (Masamoto et al., 2001). Further functional analyses of the putative maize carotene isomerase in the paired desaturase complementation system used here are merited, once the maize isomerase(s) have been isolated.

Mutants of maize offer candidates for factors affecting the desaturations. Substrates for an isomerase along with candidate loci coding for structural genes and ancillary factors, which condition the accumulation of desaturation intermediates, are shown on the poly‐Z pathway in Fig. 1. In this and a previous study (Li et al., 1996), three out of five of these loci have been assigned to structural genes and ancillary factors. Since the desaturases require oxygen and are coupled to an electron transport chain, with oxygen being the final acceptor, the electron withdrawal from the desaturations involves a membrane‐bound plastoquinone (Beyer et al., 1989; Mayer et al., 1990, 1992; Norris et al., 1995) and a plastidial terminal oxidase (PTOX) (Carol et al., 1999). Among the studies of various organisms, additional identified factors required for enzyme activity include chaperonins (Al‐Babili et al., 1996; Bonk et al., 1996) and galactolipids (Schledz et al., 1996), while the accumulation or sequestration of the carotenoid biosynthetic endproducts, has been found to be regulated through lipid composition (Rabbani et al., 1998) and/or specific carotenoid binding proteins (Cervantes‐Cervantes et al., 1990). Yet other factors are likely to be involved (Li et al., 2001).

By attempted chemical complementation of the maize mutants vp2, w3, vp5, vp7, and vp9, the candidate loci for ancillary functions, including an isomerase, were further narrowed by identifying a locus associated with electron transport away from the desaturations at the level of PDS. Of the other desaturation mutants tested, only vp2 was blocked in the biosynthesis of HGA, an intermediate of plastoquinone biosynthesis, while w3 was not, despite its conditioning of phytoene accumulation. From these results, it was proposed that maize vp2, similar to pds1 in green tissues of Arabidopsis, encodes or regulates the expression of 4‐hydroxyphenylpyruvate dioxygenase. Given the endosperm phenotype of vp2, these results are the first reported extension of the role of plastoquinones in carotene desaturation to amyloplasts. As a corollary, since w3 accumulates phytoene, and thus is a general block in carotene desaturation, and vp5 and vp2 have been functionally identified, w3 becomes a candidate for PTOX. w3 has both an endosperm and plant phenotype, but does not show a variegated plant phenotype as do the Arabidopsis PTOX deficiency, immutans (Wetzel et al., 1994) and the tomato deficiency, ghost (Giuliano et al., 1993). With the elimination of vp5, vp2, and vp9, only w3, y9 and y8 remain as candidates for deficiency of an isomerase associated with carotene desaturation, given the likely mutational saturation for carotenoid deficiency. Only y9 is associated with ZDS, by the accumulation of ζ‐carotene. Therefore, Y9 is the best candidate for an isomerase acting on or before ζ‐carotene accumulation.

Recent manipulation of carotenoid content by the introduction of ectopically‐expressed, transpecific transgenes including the bacterial four‐step desaturase, CRTI, has met with mixed success and unexpected results (Beyer et al., 2002). CRTI mediates an all‐E rather than poly‐Z‐pathway of the paired plant desaturases PDS and ZDS. Perhaps the bacterial and plant enzymes interact differently with the plant metabolon or alter the stereochemical dynamics of the metabolon. Despite the laudable success of ‘golden rice’, unexpected obstacles thwarted the use of the then available plant desaturases. Specifically, daffodil desaturases were ineffective in conferring carotene accumulation upon rice endosperm (Burkhardt et al., 1997), which necessitated the use of bacterial CRTI (Ye et al., 2000). The finding of a poly‐Z‐desaturation pathway for maize carotene desaturation, suggests that improvement of carotenoid content in cereal crops, including rice, will require use of functionally characterized maize desaturases. Although these desaturases are likely to require an additional isomerase function, the maize desaturases are more likely to interact normally with an endogenous metabolon than is CRTI. Furthermore, the functional maize cDNAs are excellent candidate transgenes for the engineering of rice endosperm, since the active proteins are similar to the endogenous rice enzymes, yet the genes are likely to be dissimilar enough in primary sequence to evade transcriptional silencing.

Acknowledgements

We thank Dr B Burr (Brookhaven National Laboratory, Upton, NY) and The University of Missouri‐Columbia RFLP Laboratory (MO) for maize mapping, D Cain (Lehman College) and The New York City Parks Department and The International Garden Club for greenhouse and field support. Drs M Cervantes‐Cervantes, C Gallagher, and P Olhoft are gratefully acknowledged for critical reading of the manuscript. This research was funded in part by The Rockefeller Foundation International Rice Biotechnology Program, The National Institutes of Health‐SCORE/MBRS program (Grant 2 S06 GM08225), and The City University of New York Center for Applied Biomedicine and Biotechnology (CABB) and PSC‐CUNY grants program.

References

Issue Section:

Cell and Molecular Biology, Biochemistry and Molecular Physiology

Comments

0 Comments

I agree to the terms and conditions. You must accept the terms and conditions.

Email alerts

Citing articles via

More from Oxford Academic