Regulation of carotenoid biosynthesis during tomato fruit development: expression of the gene for lycopene epsilon-cyclase is down-regulated during ripening and is elevated in the mutantDelta

Summary

The red colour of tomato (Lycopersicon esculentum) fruits is provided by the carotenoid pigment lycopene whose concentration increases dramatically during the ripening process. A single dominant gene,Del, in the tomato mutantDeltachanges the fruit colour to orange as a result of accumulation of δ-carotene at the expense of lycopene. The cDNA for lycopene ε-cyclase (CrtL-e), which converts lycopene to δ-carotene, was cloned from tomato. The primary structure of CRTL-E is 71% identical to the homologous polypeptide fromArabidopsisand 36% identical to the tomato lycopene β-cyclase, CRTL-B. TheCrtL-egene was mapped to a single locus on chromosome 12 of the tomato linkage map. This locus co-segregated with theDelgene. In the wild-type tomato, the transcript level ofCrtL-edecreases at the ‘breaker' stage of ripening to a non-detectable level in the ripe fruit. In contrast, it increases approximatley 30-fold during fruit ripening in theDeltaplants. TheDeltamutation does not affect carotenoid composition nor the mRNA level ofCrtL-ein leaves and flowers. These results strongly suggest that the mutationDelis an allele of the gene for ε-cyclase. Together with previous data, our results indicate that the primary mechanism that controls lycopene accumulation in tomato fruits is based on the differential regulation of expression of carotenoid biosynthesis genes. During fruit development, the mRNA levels for the lycopene-producing enzymes phytoene synthase (PSY) and phytoene desaturase (PDS) increase, while the mRNA levels of the genes for the lycopene β- and ε-cyclases, which convert lycopene to either β- or δ-carotene, respectively, decline and completely disappear.

Introduction

Carotenoids are C₄₀ isoprenoids which consist of eight isoprene units. Carotenes are hydrocarbons that are either linear or cyclized at one or both ends of the molecule, and xanthophylls are formed by the introduction of various oxygen functions to carotenes ( Goodwin 1980). The most prominent chemical feature of the carotenoids is the polyene chain, consisting of 3–15 conjugated double bonds, which is responsible for the characteristic absorption spectrum and therefore the colour of the carotenoid, and for the photochemical properties of the molecule ( Britton 1995). Due to these photochemical properties, carotenoids are essential components for all photosynthetic organisms, where they participate in a number of processes ( Frank & Cogdell 1996). These include light-harvesting, resulting in energy transfer to the chlorophylls; photoprotection by quenching triplet-state chlorophyll molecules; and scavenging singlet-state chlorophyll (reviewed by Demmig-Adams & Adams 1996). Carotenoids also serve structural functions in the photosynthetic pigment–protein complexes of the reaction centres and the light-harvesting antennae, where they are bound to specific chlorophyll/carotenoid-binding proteins.

In higher plants, carotenoids fulfil an additional important purpose as colorants of flowers and fruits. In these tissues they accumulate in chromoplasts and render bright yellow, orange or red colours that attract animals which facilitate pollination and seed dispersion.

The carotenoid biosynthesis pathway takes place within the plastids (reviewed by Cunningham & Gantt 1998;Harker & Hirschberg 1998) ( Fig. 1). The first committed step is the head-to-head condensation of two geranylgeranyl diphosphate (GGDP) molecules to produce phytoene, catalysed by the enzyme phytoene synthase (PSY). Two enzymes, phytoene desaturase (PDS) and ζ-carotene desaturase (ZDS or CRTQ), introduce four double bonds that convert phytoene to lycopene via phytofluene, ζ-carotene and neurosporene. The cyclization of lycopene is an important branching point in the pathway. One route leads to β-carotene (β,β-carotene) and its derivative xanthophylls: zeaxanthin, violaxanthin and neoxanthin. The latter two are precursors for the synthesis of the hormone abscisic acid (ABA). The alternative pathway leads to carotenoids with one β and one ε end ring, such as α-carotene and lutein which is a major xanthophyll in the light-harvesting system of most higher plants.

Figure 1. Pathway of carotenoid biosynthesis in plants and algae.

Cyclization of lycopene marks a branching point of the pathway to either α- or β-carotene. Enzymes are indicated by their gene assignment symbols: aba2, zeaxanthin epoxidase;CrtL-b, lycopene β-cyclase;CrtL-e, lycopene ε-cyclase;CrtR-b, β-ring hydroxylase;CrtR-e, ε-ring hydroxylase;Pds, phytoene desaturase (crtP in cyanobacteria);Psy, phytoene synthase (crtB in cyanobacteria);Zds, ζ-carotene desaturase (crtQ in cyanobacteria). GGDP, geranylgeranyl diphosphate.

Carotenoid formation is a highly regulated process. Concentration and composition of leaf xanthophylls are affected by light intensity ( Ruban et al. 1994 ) and the accumulation of specific carotenoids in chromoplasts of fruits and flowers is developmentally regulated ( Fraser et al. 1994 ;Giuliano et al. 1993 ). Carotenoid accumulation during fruit ripening in tomato serves as a model system to investigate the regulation of the process. In tomato, total carotenoid concentration increases between 10- and 15-fold during fruit ripening. This change is due mainly to a 500-fold increase in the concentration of lycopene ( Fraser et al. 1994 ). Accumulation of lycopene begins at the ‘breaker' stage of fruit ripening after the fruit has reached the ‘mature green' stage. Following cloning of the genes for phytoene synthase (Psy) and phytoene desaturase (Pds), it was possible to demonstrate that the mRNA levels of these genes increase significantly during the ‘breaker' stage ( Fraser et al. 1994 ;Giuliano et al. 1993 ;Pecker et al. 1992 ). The changes in the steady-state levels of mRNA of Pds have been attributed to transcriptional regulation ( Corona et al. 1996 and unpublished data). A similar increase in mRNA was found in the genes GGPPS (for GGDP synthase), Psy and Pds during fruit ripening of bell pepper ( Camara et al. 1995 ). In contrast, the mRNA of CrtL-b, which encodes lycopene β-cyclase, decreases at the ‘breaker' stage. Evidence for transcriptional up-regulation of carotenoid genes in flowers has been described for Psy, Pds and CrtL-b ( Corona et al. 1996 ;Giuliano et al. 1993 ;Pecker et al. 1996 and unpublished data).

The cDNA for lycopene ε-cyclase has been cloned from Arabidopsis thaliana ( Cunningham et al. 1996 ). This enzyme in Arabidopsis produces only a single ε-ring in the lycopene molecule to yield δ-carotene. In this paper we report on cloning of the cDNA of CrtL-e from tomato, encoding lycopene ε-cyclase. Measurements of mRNA levels of CrtL-e indicated that down-regulation of gene expression of both types of lycopene cyclase is a major mechanism that is responsible for lycopene accumulation during tomato fruit ripening. Evidence is presented that strongly suggests that the locus Del in the fruit-colour mutation Delta encodes the gene for lycopene ε-cyclase.

Results

Cloning the tomato cDNA of lycopene ε-cyclase

The cDNA for lycopene ε-cyclase (CrtL-e) from Arabidopsis thaliana has been previously cloned and characterized ( Cunningham et al. 1996 ). We used the Arabidopsis cDNA as a molecular probe for screening a cDNA library from tomato (Lycopersicon esculentum) shoots in the vector λ-ZapII. A positive clone was isolated and the insert was excised in pBluescript SK– vector which was designated pCRTLE (see Experimental procedures).

Nucleotide sequence analysis of the 1700 bp DNA insert in pCRTLE revealed an open reading frame of 526 codons, potentially coding for a polypeptide of a calculated molecular mass of 58.8 kDa (GenBank accession no. Y14387) ( Fig. 2). The amino acid sequence of the putative CRTL-E from tomato is 71% identical to its homologous protein from Arabidopsis and 36% identical to the tomato lycopene β-cyclase, CRTL-B. Surprisingly, CRTL-E is 35% identical to the enzyme capsanthin-capsorubin synthase (CCS) from pepper.

Figure 2. Alignment of amino acid sequences of lycopene cyclases from tomato and Arabidopsis.

Lycopene ε-cyclase from wild-type tomato, L. esculentum (CRTL-E L.e.) is compared with the Del allele originated from L. pennellii (CRTL-E L.p.), the homologous enzyme from Arabidopsis (CRTL-E A.t.), lycopene β-cyclase from tomato (CRTL-B L.e.) and capsanthin capsorubin synthase from bell pepper (CCS C.a.). Highly conserved sequences are highlighted with a grey box. A putative FAD (NAD/P)-binding motif, common to all the cyclases, is depicted in position 110–121 of the tomato CRTL-E. Other conserved sequences are underlined. (*, identical residues in CRTL-E of L.e. and L.p.;–, gap).

Expression of CrtL-e in E. coli

The use of E. coli heterologous system for carotenoid biosynthesis is a powerful tool for identifying genes for carotenoid biosynthesis enzymes ( Hirschberg et al. 1997 ). E. coli cells of the strain XLI-Blue carrying the plasmid pACCRT-EIB accumulate lycopene ( Fig. 3 and Table 1). This plasmid contains the genes crtE, crtB and crtI from Erwinia uredovora, that encode geranylgeranyl pyrophosphate synthase, phytoene synthase and phytoene desaturase, respectively, on a pACYC184 vector. Lycopene-accumulating E. coli cells were co-transformed with the plasmid pCRTLE and selected on LB medium containing both ampicillin and chloramphenicol. Carotenoids from cells carrying pACCRT-EIB alone, or pACCRT-EIB and pCRTLE were extracted and analysed by HPLC ( Fig. 3).

Figure 3. Analysis by HPLC of carotenoids in E. coli cells that express CrtL-e from tomato.

Carotenoids were extracted from suspension cultures of cells with the plasmids pACCRT-EIB, carrying crtB, crtE and crtI from E. uredovora (upper panel), or with pACCRT-EIB + pCRTLE, expressing the cDNA of CrtL-e from tomato (lower panel). Absorbance spectra of the peaks are presented in the boxes: 1, lycopene; 2, δ-carotene; 3, ε-carotene.

Table 1.  Accumulation of carotenoids in E. coli cells expressing CrtL-b and CrtL-e from tomato (percentage of total carotenoids)

Plasmids	Lycopene	δ-carotene	β-carotene	α-carotene	ε-carotene	Others
pACCRT-EIB	100
pCAR-EIB + pCRTLE	2	86			12
pBCAR-T	6		94
BCAR-T + pCRTLE			24	70		6
pDCAR	2	98
DCAR + pCRTLTOM			91	9
pBCAR	2		98
pBCAR + pCRTLE			67	33

Cells carrying the plasmid pACCRT-EIB produced lycopene as expected, while cells carrying both pACCRT-EIB and pCRTLE accumulated mostly δ-carotene ( Fig. 3 and Table 1). This result indicates that the cDNA product of CrtL-e is indeed lycopene ε-cyclase, which catalyses the asymmetric formation of an ε-ionone ring on the linear lycopene molecule. However, a small portion of the two ringed molecule ε-carotene was also produced in E. coli.

Expression of both types of lycopene cyclase, β-cyclase and ε-cyclase, from tomato in E. coli cells that produce lycopene was achieved by co-transformation of E. coli with either plasmids pBCAR-T and pCRTLE or pDCAR and pCRTLTOM. The difference between the two pairs of plasmids in the proportion of α-carotene production is due mainly to the unequal expression of these genes when they are carried on either pACYC184 or pBluescript vectors, which are present in the bacteria at a different copy number. These results establish that two different lycopene cyclase enzymes participate in the formation of α-carotene and therefore they are likely to comprise the enzymatic pathway leading to α-carotene in plants. It is also evident that the same lycopene β-cyclase can function in the biosynthesis of both β-carotene and α-carotene. It is interesting to note that α-carotene was produced in E. coli also when the bacterial-type lycopene β-cyclase, CRTY, carried by pBCAR, was co-expressed with the tomato ε-cyclase ( Table 1).

Inhibition of lycopene ε-cyclase by herbicides

The effect of the tri-alkyl amine compounds, 2-(4-methylphenoxy)triethylamine hydrochloride (MPTA) and 2-(4-chlorophenylthio)-triethylamine hydrochloride (CPTA), on lycopene cyclization was examined by adding the inhibitors at various concentrations to suspension cultures of E. coli which carried plasmids that expressed the different lycopene cyclases. The I₅₀ concentration for inhibition of CRTL-E as measured in cells carrying either pDCAR or pACCRT-EIB + pCRTLE, was 5.0 μm for CPTA and 13.5 μm for MPTA (data not shown). These compounds inhibited the tomato lycopene β-cyclase with an I₅₀ of 13.5 μm for CPTA and 6.0 μm for MPTA.

Analysis of carotenoids in tomato fruits of wild-type and the mutant Delta

Carotenoid composition in the green stages of fruit ripening in tomato is similar to that of green leaves where lutein, violaxanthin, neoxanthin and β-carotene predominate ( Table 2). At the ‘breaker' stage of fruit ripening, lycopene begins to accumulate and reaches up to a 500-fold increase in the red stage ( Fraser et al. 1994 ). Carotenoids were analysed in ripe fruits (‘red' stage) of wild-type and Delta mutants of L. esculentum. In ripe fruits of the wild-type tomato, lycopene is the major carotenoid (approximately 90%), with smaller amounts of β-carotene (5–10%) and lutein (1–5%), and trace amounts (< 1.0%) of other carotenoids, including δ-carotene ( Table 2). In fruits of the Delta mutant, the major carotenoid that begins to accumulate at the ‘breaker' stage is δ-carotene, which makes up 65% of total carotenoids in the ripe stage, while lycopene is only 25%. It is interesting to note that β-carotene appears in these fruits only in trace amounts whereas α-carotene and lutein, both derived from δ-carotene, appeared at significant levels of 4% and 6%, respectively. Altogether, carotenoids containing an ε-ring reached 75% in the Delta fruits, in contrast to 1–5% in the wild-type fruits.

Table 2.  Carotenoid composition in leaves, flowers and fruits of the wild-type and the Delta mutant of tomato

	Wild-type				Delta
Carotenoid	Leaf ^a	Petals ^a	Green fruit ^a	Ripe fruit	Ripe fruit
Values are percentage of total carotenoids. ^a Similar values were measured in leaves, flowers and green fruits of Delta.
Lycopene	0	0	0	92	25
β-carotene	23	0.5	22	6	traces
Zeaxanthin	1	1.5	4	0	0
Antheraxanthin	1	0	4	0	0
Violaxanthin	14	70	6	0	0
Neoxanthin	11	21	9	0	0
δ-carotene	0	0	0	traces	65
ε-carotene	0	0	0	0	traces
α-carotene	0	0	0	traces	4
Lutein	50	7	55	2	6

CrtL-e maps to Del

All known Delta mutations are derived from green-fruited wild tomato species ( Tomes 1967;Tomes 1969). As a source for the Delta mutation we used introgression line IL 12–2 that has previously been described ( Eshed & Zamir 1995). This line originated from a cross between the green-fruited wild species Lycopersicon pennellii and the cultivated tomato L. esculentum CV M-82. It contains a homozygous substitution of an approximately 35 cM segment on chromosome 12, which is derived from L. pennellii, in a genetic background of L. esculentum. Crosses between known Delta mutants and IL 12–2 demonstrated that they are all allelic (data not shown).

The Del locus has been roughly mapped in the past on chromosome 12. To accurately map Del we have crossed IL12–2 with M-82, and the F1 hybrid was selfed. A population of 287 F2 plants was grown and the Del phenotype of the progeny was scored. DNA was extracted from each plant and RFLP analysis was carried out using ³²P-labelled probes of known RFLP markers of chromosome 12. It was determined that Del is located 2 cM away from CT-79 and 3.2 cM from TG-263-A ( Fig. 4).

Figure 4. Fine genetic mapping of Del on chromosome 12 of the tomato linkage map.

The linkage map was adopted from Eshed & Zamir (1995). The relevant chromosomal segments from L. pennellii that were introgressed to L. esculentum are represented by black bars with the name of the lines that carry them. A high-resolution map around Del is displayed on the right with genetic distances in map units.

When the cDNA of CrtL-e was used as a molecular probe in the DNA blot hybridization of the segregating population, it was mapped to a single locus in the tomato genome that showed 100% co-segregation with Del (in a less than 0.18 cM resolution).

Expression of CrtL-e during fruit ripening in wild-type and Delta

Previously, it has been shown that the steady-state levels of mRNA of the genes for early enzymes in the carotenoid biosynthesis pathway, phytoene synthase and phytoene desaturase, increase during fruit ripening in tomato. In the case of Pds, it was demonstrated that transcriptional up-regulation is responsible for this increase ( Corona et al. 1996 ;Giuliano et al. 1993 ;Pecker et al. 1992 ). Previously, we have observed that the mRNA level of CrtL-b, which encodes lycopene β-cyclase, decreases during tomato fruit ripening ( Pecker et al. 1996 ). To determine the regulation of expression of CrtL-e during fruit development in tomato, we have measured by RT–PCR its mRNA level at different stages of ripening. As can be seen in Fig. 5, a significant level of mRNA of CrtL-e was detected during the green stages of fruit ripening. However, it decreased at around the ‘breaker' stage to an undetectable level at the subsequent ripening stages. This marked drop of CrtL-e mRNA is contrasted by the dramatic increase in mRNA level of Pds, which was measured simultaneously by RT–PCR at the same stages of fruit ripening. In contrast to the wild-type tomato, the mRNA of CrtL-e in fruits of the Delta mutant showed a striking increase after the ‘breaker' stage, that surpassed that of Pds ( Fig. 5).

Figure 5. Expression of CrtL-e, CrtL-b and Pds during tomato fruit development.

Steady-state levels of mRNA of CrtL-e, CrtL-b and Pds were measured concomitantly by RT–PCR from total RNA isolated from different stages of fruit ripening of wild-type (WT) L. esculentum and from the mutant Delta. Autoradiographs of PCR-amplified DNA fragments that had been separated by polyacrylamide gel electrophoresis are exhibited in the top and middle panels, and ethidium bromide-stained RT–PCR products are shown in the bottom panel. The relative amount of PCR products was determined by counting the radioactivity. Ripening stages: IG, immature green; MG, mature green; B, breaker; O, orange; P, pink; R, ripe. L, leaf. L:3 and L×3 are samples which contained one-third or three times the total leaf RNA that was included in the RT–PCR amplification reactions, respectively, and they served as a control for the linearity of the RT–PCR measurement of mRNA. Control, rat pre-albumin amplification (see Experimental procedures).

Expression of CrtL-b and CrtL-e in flowers of wild-type and Delta

The steady-state levels of mRNA of CrtL-b and CrtL-e in petals of young or old flowers of tomato was measured by RT–PCR ( Fig. 6). The results indicated that the mRNA level for CrtL-b increases in petals relative to leaves, while CrtL-e is not expressed at all. This result corresponds to the accumulation of violaxanthin and neoxanthin in tomato petals ( Table 2). The mRNA levels of CrtL-e in Delta flowers were similar to those of the wild-type (data not shown).

Figure 6. Expression of Pds, CrtL-b1 and CrtL-e in tomato flowers.

Steady-state levels of mRNA of Pds, CrtL-b1 and CrtL-e were measured by RT–PCR from total RNA isolated from leaf (L) or petals from young (Fy) or old (Fo) flowers of L. esculentum. Ethidium bromide-stained PCR-amplified DNA fragments that were separated by agarose gel electrophoresis are exhibited.

Discussion

ε-cyclase and the biosynthesis of α-carotene

Expression of the cDNA of CrtL-e from tomato cDNA in E. coli cells confirmed that it encodes lycopene ε-cyclase, which converts lycopene to δ-carotene. This characteristic is essentially similar to the previously described ε-cyclase of Arabidopsis ( Cunningham et al. 1996 ) except that in the case of the tomato enzyme, a small portion of the δ-carotene was converted to ε-carotene (ε,ε-carotene) ( Table 1). Since ε-carotene does not normally accumulate to any detectable level in tomato tissues, it is possible that this result is a fault of the heterologous E. coli system. Furthermore, the accumulation of the single ringed δ-carotene, with only trace amounts of ε-carotene in fruits of the tomato mutant Delta, where ε-cyclase is up-regulated, supports this explanation and establishes that in the plant CRTL-E is indeed a monocyclase enzyme.

Our results with the tomato CrtL-e confirm previous analyses in Arabidopsis ( Cunningham et al. 1996 ) showing that α-carotene is synthesized by two enzymes, lycopene β-cyclase and lycopene ε-cyclase. This is in contrast to β-carotene which is produced in tomato from lycopene by a single enzyme, lycopene β-cyclase ( Pecker et al. 1996 ). There are two CrtL-b genes in tomato ( Pecker et al. 1996 ), which code for polypeptides that are 95% identical in their amino acid sequence. In contrast, genomic DNA blot hybridization identified only a single CrtL-e gene. It is possible that one of the two β-cyclases functions specifically in the synthesis of α-carotene and potentially could be associated with the ε-cyclase. However, the data presented here demonstrate that the same β-cyclase can equally operate in the two pathways. Moreover, synthesis of α-carotene was achieved in E. coli also when the tomato CrtL-e was co-expressed with crtL-b from the cyanobacterium Synechococcus PCC 7942, a species that does not contain ε-cyclase (data not shown) and with the bacterial CRTY-type lycopene β-cyclase ( Table 1). Therefore, we postulate that in E. coliα-carotene is produced from lycopene by the two cyclases operating independently in a sequential way. There is no information on the order of the reactions but, since the β-cyclase is a bifunctional enzyme that interact with both the linear lycopene and the single-ringed γ-carotene (1996;Cunningham et al. 1993 ; Pecker et al. ), it is more likely that ε-cyclization is the first reaction in the biosynthesis of α-carotene.

Cyclization of lycopene denotes a central branch point in the carotenoid biosynthesis pathway, one route leading to the β,β-xanthophylls, zeaxanthin, violaxanthin and neoxanthin, and the other leading to the β,ε-xanthophyll, lutein. The relative activities of ε-cyclase versus β-cyclase may determine the flow of carotenoids from lycopene to either α-carotene or β-carotene. The regulation of the two types of lycopene cyclizations could therefore be a major mechanism that controls carotenoid composition in vivo. In the case that ε-cyclization is the initial process in α-carotene biosynthesis, it could serve as the regulator step in the pathway since its product, δ-carotene, is destined to the α-carotene branch. We propose that the regulation of ε-cyclase activity, possibly through control of CrtL-e expression, determines xanthophyll composition in leaves and other plant tissues.

The lycopene cyclase gene family

There is a high degree of structural resemblance, 36% identity in amino acid sequence, between the β- and ε-cyclases in both tomato and Arabidopsis ( Fig. 2). It is believed that the amino termini of both polypeptides are involved with targeting to the plastids and are cleaved off during this process ( Cunningham et al. 1996 ;Pecker et al. 1996 ). The two enzymes contain a characteristic FAD/NAD(P)-binding sequence motif at the amino termini of the mature polypeptides. The role of dinucleotide co-factors in lycopene cyclization is unclear at the moment since no net redox reaction is required. The overall sequence conservation between β- and ε-cyclases suggests that these enzymes have a similar mechanism of action. Both use the same substrate in reactions that are proposed to proceed through a common intermediate of carbonium ion ( Britton 1988). The primary structures of the two lycopene cyclases, CRTL-B and CRTL-E, are 50–55% identical to that of the pepper enzyme capsanthincapsorubin synthase (CCS), also called κ-cyclase, which converts antheraxanthin and violaxanthin to capsanthin and capsorubin, respectively, ( Bouvier et al. 1994 ). In heterologous and in vitro systems CCS also has some activity as lycopene β-cyclase ( Hugueney et al. 1995 ). It has recently been proposed that κ-ring formation employs a similar mechanism as β-ring formation and that two domains of aromatic and carboxylic amino acid residues in CRTL-B and CCS are involved with these processes ( Bouvier et al. 1997 ). These nucleophilic residues are thought to neutralize the incipient carbocations that are transiently formed during carotenoid cyclization.

The compounds CPTA and MPTA were previously described as non-competitive specific inhibitors of lycopene β-cyclase in plants, algae and cyanobacteria ( Cunningham et al. 1993 ;Yokoyama et al. 1982 ) and of capsanthin-capsorubin synthase in pepper ( Bouvier et al. 1997 ). The proposed mechanism of inhibition involves a protonated nitrogen in these inhibitors that mimics the carotenoid carbocation and interacts with the nucleophilic residues in the cyclases. The fact that the ε-cyclase from tomato is inhibited by these chemicals with similar I₅₀ values as the β-cyclase, suggests that CPTA and MPTA interact with CRTL-E similarly as with CRTL-B, thus confirming the resemblance in the mode of activity of the two enzymes.

Conservation of amino acid sequence as well as similar mechanism of catalysis suggest that all plant cyclases evolved from a common ancestor enzyme. To postulate a possible phylogenetic relationship among this group of enzymes, we have used the maximum likelihood method of sequence analysis and the results are depicted as an unrooted phylogenetic tree ( Fig. 7). As illustrated in Fig. 7, it is likely that CrtL-e in higher plants evolved from the plant-type CrtL-b, which appeared first in cyanobacteria. Since none of the cyanobacterial species contain any ε-cyclase activity ( Hirschberg & Chamovitz 1994), it is presumed that CrtL-e evolved from CrtL-b in green algae, where light harvesting antenna complexes of protein-bound chlorophylls and xanthophylls first developed, and where lutein and other ε-ring xanthophylls are the predominant carotenoid species. It is interesting to note that Ccs is closer to CrtL-b, suggesting that it was derived from the latter after the divergence of CrtL-e.

Figure 7. Phylogeny of lycopene cyclase genes from various plants, algae and cyanobacteria (blue-green algae).

The phylogenetic tree of the cyclase genes was calculated from nucleotide sequences by the maximum likelihood DNAMLK program of the PHYLIP (Phylogeny Inference Package) version 3.5c. The sequence of 1230 nucleotides from each cDNA (or gene), starting 18 nucleotides upstream of the conserved FAD-binding motif, was included in the comparison. Sequences were derived from GenBank.

Regulation of lycopene accumulation in tomato fruits

Carotenoid biosynthesis in tomato fruits is a model case for other chromoplast-containing tissues, where the pathway is extremely active and results in the accumulation of high concentration of specific carotenoids ( Gillaspy et al. 1993 ;Grierson & Schuch 1993). The tomato fruit is a unique case because the carotenoid species that accumulates in the mature chromoplasts is an intermediate in the pathway that is constitutively operating during the earlier developmental stages. Our data indicate that the relative abundance of mRNA of genes for the carotenoid biosynthesis enzymes changes dramatically during fruit development. While mRNA levels for the enzymes that are responsible for lycopene biosynthesis, PSY and PDS, increase at the breaker stage of ripening ( Giuliano et al. 1993 ;Pecker et al. 1992 ), the mRNAs for the two enzymes that utilize lycopene for further biosynthetic steps, CRTL-B and CRTL-E, disappear at this stage ( Fig. 8). In the case of Psy and Pds, it was established that the increase is due to transcriptional regulation ( Corona et al. 1996 and unpublished results). The mechanism by which the mRNAs of CrtL-b and CrtL-e diminish to practically undetectable levels is as yet unknown. It is likely that transcription of these genes stops at the breaker stage but this will have to be established. The data indicate that differential gene expression plays a major role in the accumulation of lycopene in tomato fruits by elevating the concentration of its biosynthetic enzymes and blocking the synthesis of enzymes that convert it to cyclic carotenoids. This hypothesis is further supported by the accumulation of δ-carotene in the fruits of the Delta mutant. In this mutant the mRNA level of CrtL-e increases at the breaker stage and remains elevated until the fruit fully ripens. Consequently, most of the lycopene is converted to δ-carotene.

Figure 8. Regulation of expression of carotenoid biosynthesis genes during tomato fruit ripening.

Relative levels of mRNA of Psy, Pds, CrtL-b and CrtL-e in tomato fruits during different stages of development. Data compiled from this work and from Pecker et al. (1992 ), Giuliano et al. (1993 ), Fraser et al. (1994 ) and Pecker et al. (1996 ).

The molecular basis of the Del phenotype

Several lines of evidence strongly suggest that the locus Del is the gene CrtL-e which encodes lycopene ε-cyclase.

(1) Genetic analysis showed that CrtL-e co-segregated with the locus Del.

(2) The property of the CrtL-e-encoded enzyme as lycopene ε-cyclase, which introduces a single epsilon ring to lycopene, conforms with the phenotype of δ-carotene accumulation in fruits of Delta.

(3) The increase of CrtL-e mRNA during fruit ripening in Delta, which is in contrast to the decrease in the wild-type, explains the phenotype of δ-carotene accumulation.

(4) The above molecular data are consistent with the dominant nature of the Delta mutation.

The amino acid sequence of CRTL-E is 98.5% identical between Delta and wild-type and both are equally functional in E. coli (data not shown). Therefore, we assume that the two CrtL-e alleles could differ in their promoter sequence. This hypothesis is supported by the fact that the origin of Del in L. esculentum is from the wild species L. pennellii, whose fruits remain green during ripening and continue to synthesize all major xanthophylls. There are no data available on the expression of carotenoid biosynthesis genes during fruit ripening in L. pennellii. It is probable that it is unchanged and in all stages of development resembles the expression in the green stages of L. esculentum. It is evident that in L. esculentum the expression of CrtL-e from L. pennellii responds to the same developmental regulation of gene expression that boosts expression of Psy and Pds.

In petals of tomato flowers, the major carotenoid that accumulates is violaxanthin. In this tissue, the mRNA levels of Psy, Pds ( Giuliano et al. 1993 ;Pecker et al. 1992 ) and CrtL-b ( Pecker et al. 1996 ) increase while the mRNA of CrtL-e is undetectable. Chromoplasts in petals develop without going through green stages and no CrtL-e mRNA could be detected even in petals of young flower buds. The carotenoid composition in flowers of the mutant Delta is similar to that of the wild-type (data not shown). It is apparent that the unusual regulation of CrtL-e expression in Delta is fruit-specific and does not relate to chromoplast development.

Experimental procedures

Bacteria and plants

E. coli strain XL1-Blue was used in all experiments that are described in this work. Tomato (Lycopersicon esculentum) CV M82 served as the ‘wild-type' strain in the fruit-ripening measurements. The introgression line IL12–2 ( Eshed & Zamir 1995) was used as the source for the Delta mutation and employed for fine mapping of Del.

Cloning and sequence analysis

A cDNA library from shoots of tomato (Lycopersicon esculentum cv 93–137), in the vector λ-ZapII was screened using a 1860 bp DNA fragment of the cDNA of the epsilon cyclase gene from Arabidopsis thaliana (kindly provided by Dr Francis X. Cunningham Jr). DNA hybridization was done at low stringency in a buffer containing 35% formamide, 5× SSC, 5× Denhardt's solution and 0.2% SDS at 37°C. Filters were washed with 4× SSC at 37°C. Positive clones were excised using the λ-zap protocol (Stratagene).

DNA sequence analysis was performed by the ABI Prism 377 DNA (Perkin Elmer) sequencer and processed with the ABI sequence analysis software. Nucleotide and amino acid sequence analysis and comparisons were done using the UWGCG software package. Phylogeny of the cyclase genes was calculated from nucleotide sequences by the maximum likelihood DNAMLK program of the phylip (Phylogeny Inference Package) version 3.5c, distributed by J. Felsenstein, Department of Genetics, University of Washington, Seattle.

Plasmids

Plasmids pACCRT-EIB and pBCAR for functional expression in E. coli of the carotenoid biosynthesis genes crtE, crtB, crtI and crtY from Erwinia herbicola, have been previously described ( Cunningham et al. 1993 ;Lotan & Hirschberg 1995). Plasmid pCRTLTOM, expressing the tomato cDNA of CrtL-b in E. coli was described in Pecker et al. (1996 ). Plasmid pCRTLE was constructed for the current work by subcloning a 1700 bp DNA fragment containing the cDNA of the tomato CrtL-e in the EcoRI site of the vector pBluescript SK–. Plasmid pDCAR was constructed for this work by introducing the EcoRI insert of pCRTLE into pACCRT-EIB. E. coli cells that carry this plasmid produce and accumulate δ-carotene. Plasmid pBCAR-T is similar to pBCAR, except that the gene crtY, encoding lycopene β-cyclase, was replaced by the cDNA of CrtL-b from tomato.

Pigment extraction and analysis

For extraction of pigments from E. coli, aliquots of 2 ml were taken from bacterial suspension cultures. The cells were harvested by centrifugation at 4000 g, washed once with water, resuspended in 2 ml of acetone and incubated at 65°C for 10 min in the dark.

The samples were centrifuged again at 13 000 g for 5 min and the acetone supernatant containing the pigments was placed in a clean tube. More than 99% of the caroteniods were extracted by this procedure as determined by re-extraction after breaking and grinding the samples. The pigment extract was blown to dryness under a stream of N₂ and stored at –20°C until required for analysis.

Fruit pigments were extracted from 1.0 g of fresh tissue. The tissue was ground in 2 ml of acetone and incubated at room temperature in the dark for 10 min. Then 2 ml of dichloromethane were added and the samples were agitated until all pigments were transferred to the supernatant which was then filtered. Four millilitres of ether and 0.4 ml of 12% w/v NaCl/H2O were added and the mixture was shaken gently until all pigment was transferred to the upper (ether) phase. The ether was collected, dried under a stream of N₂ and stored at –20°C until required for analysis.

Carotenoids were separated by reverse-phase HPLC using a Spherisorb ODS-2 column (silica 5 μm, 3.2 mm × 250 mm) (Phenomenex^®). Samples of 50 μl of acetone-dissolved pigments were injected to a Waters 600 pump. The mobile phase consisted of acetonitrile:H₂O (9:1) (solvent A) and 100% ethyl acetate (solvent B), which were used in a linear gradient between A and B for 30 min, at a flow of 1 ml min^–1. Light absorption peaks were detected in the range of 200–600 nm using a Waters 996 photo diode-array detector. All spectra were recorded in the eluting HPLC solvent as was the fine absorbance spectral structure. Carotenoids were identified by their characteristic absorption spectra and their typical retention time, which corresponded to standard compounds of lycopene and β-carotene. Peak areas were integrated by the Millenium chromatography software (Waters).

Preparation of nucleic acids from tomato tissues and RFLP mapping

Genomic DNA was prepared from 5 g of leaf tissue as described by Eshed & Zamir 1995. Restriction fragment length polymorphism (RFLP) in the tomato genomic DNA was analysed as previously described ( Eshed & Zamir 1995) using the markers TG-68, TG-263A, CT-79 and TG-263B.

Total RNA was extracted from 3 g of fruit or leaf tissues as previously described ( Pecker et al. 1996 ) with the following modifications. Frozen samples were ground in liquid nitrogen and RNA was extracted using the TRI-REAGENT protocol, according to the manufacturer's recommended protocol (Molecular Research Center, Inc., Cincinnati, Ohio, USA). RNA concentration and purity were determined by spectrophotometry and its integrity was examined by electrophoresis in formaldehyde gels stained with ethidium bromide. The visual inspection of the rRNA bands also verified the spectrophotometrical results regarding relative concentrations of RNA in each sample.

Measurement of mRNA by RT–PCR

Reverse transcription of total RNA isolated from fruit tissues was carried out using oligo-dT as a primer. Each sample contained 250 ng of total RNA. The reaction mixture included 1 m m dNTPs, 0.5 μm oligo-dT, 20 units of RNAasin (Promega), 10 ng of rat liver RNA, 10 m m DTT, RT buffer (BRL) and SuperScript II reverse transcriptase (BRL) in a total reaction volume of 20 μl. The reaction mixture was incubated at room temperature for 10 min and then at 42°C for 45 min. The polymerase chain reaction (PCR) mixture contained 0.625 μm of each oligonucleotide primer, 1× Taq polymerase buffer, 2 units of Taq polymerase (DynaZyme II DNA polymerase, recombinant, FinnZyme OY) and 10 μl of the RT reaction mixture in a total volume of 40 μl. The reverse transcription products were incubated for 5 min at 95°C prior to addition to the PCR mixture. The amplification procedure by PCR consisted of 22 cycles of 1 min at 95°C, 1 min at 56°C and 1 min at 72°C. Various initial concentrations of mRNA, ranging over 9 fold difference, were used in order to demonstrate a linear ratio between concentration of the template cDNA (corresponding to the mRNA) and the final PCR products.

The following primers were used for amplification. For Pds-1: 5′-TTGTGTTTGCCGCTCCAGTGGATAT-3′ (forward) and 5′-GCGCCTTCCATTGAAGCCAAGTAT-3′ (reverse); for CrtL-b: 5′-GGCTTCTCTAGATCTCTTGTTCAG-3′ and 5′-GTTCAGGTAGAAACAATCGAGACG-3′; for CrtL-e: 5′-GGCAGCCTCGGGGAAATTC-3′ (forward) and 5′-CACACGGAAGAATGCGCGC-3′ (reverse); for rat prealbumin: 5′-AGTCCTGGATGCTGTCCGAG-3′ (forward) and 5′-TTCCTGAGCTGCTAACACGG-3′ (reverse).

Measurement of PCR products was done by including 5 μCi of ³²P-dCTP (specific activity 3000 Ci mmole^–1) in the PCR mixture. Products of the PCR were separated by electrophoresis on 7% polyacrylamide gels or 1% agarose gels. After electrophoresis, the gels were dried, monitored by phosphoimager (MacBAS V2.2 FUJIX) and exposed to an X-ray film. The results were corrected according to the relative quantity of the control PCR product of rat pre-albumin mRNA.

Acknowledgements

This work was supported by Grant 578/97 from the Israel Science Foundation and by the Ministry of Science, under the auspices of the Plant Genome Center. We thank Dr Francis X. Cunningham Jr for the gift of the cDNA of CrtL-e from Arabidopsis.