Seed-specific overexpression of phytoene synthase: increase in carotenoids and other metabolic effects


*For correspondence (fax + 530 792 2453;


A bacterial phytoene synthase (crtB) gene was overexpressed in a seed-specific manner and the protein product targeted to the plastid in Brassica napus (canola). The resultant embryos from these transgenic plants were visibly orange and the mature seed contained up to a 50-fold increase in carotenoids. The predominant carotenoids accumulating in the seeds of the transgenic plants were alpha and beta-carotene. Other precursors such as phytoene were also detected. Lutein, the predominant carotenoid in control seeds, was not substantially increased in the transgenics. The total amount of carotenoids in these seeds is now equivalent to or greater than those seen in the mesocarp of oil palm. Other metabolites in the isoprenoid pathway were examined in these seeds. Sterol levels remained essentially the same, while tocopherol levels decreased significantly as compared to non-transgenic controls. Chlorophyll levels were also reduced in developing transgenic seed. Additionally, the fatty acyl composition was altered with the transgenic seeds having a relatively higher percentage of the 18 : 1 (oleic acid) component and a decreased percentage of the 18 : 2 (linoleic acid) and 18 : 3 (linolenic acid) components. This dramatic increase in flux through the carotenoid pathway and the other metabolic effects are discussed.


Carotenoids are a large group of compounds, often highly colored, which are derived from isoprenoid precursors. They are synthesized in plants as well as some bacteria, fungi and algae. In general, carotenoids are thought to have an antioxidant function and in higher plants, where they are synthesized in the plastid, certain carotenoids are integral parts of the photosynthetic apparatus (Bartley & Scolnik 1995; Cunningham & Gantt 1998). Additionally, carotenoids may accumulate to high levels in some non-photosynthetic plant organs such as petals and fruits where they are found in modified plastids termed chromoplasts (Camara et al. 1995). Common examples include flower petals such as marigolds and Adonis aestivalis and fruits such as tomatoes, peppers, pumpkins, etc.

As products of the isoprenoid pathway, carotenoids are derived from the five carbon isoprenoid precursor, IPP (isopentenyl diphosphate). It has recently been shown that in plants there are two major biosynthetic pathways leading to IPP. Cytoplasmic IPP is predominately derived from the mevalonate pathway and gives rise to sesquiterpenes (C15) such as cadinene and triterpenes (C30) which include sterols (Disch et al. 1998). The plastidial route for production of IPP is presently being elucidated but appears to begin with glyceraldhyde-3-phosphate and pyruvate (Arigoni et al. 1997; Lichtenthaler et al. 1997). This pathway produces the majority of the IPP for the monoterpenes (C10) such as limonene and menthol, the diterpenes(C20) such as taxol and casbene, the phytyl (C20) conjugates such as tocopherols and phylloquinones, and the carotenoids (C40).

The carotenoid pathway has been elucidated since the mid-1960s and cDNAs for most of the major enzymes have been cloned from both plant and microbial sources during the past decade (Cunningham & Gantt 1998). The first committed step in carotenoid biosynthesis is the condensation of two geranylgeranyl diphosphate (GGPP) (C20) moieties to give phytoene (a colorless carotenoid). The gene responsible for this reaction, phytoene synthase, has been cloned from a variety of microbes including Erwinia (Misawa et al. 1990; Perry et al. 1986; Sandmann & Misawa 1992) and Synechococcus (Chamovitz et al. 1992) and several plants including tomato (Bartley et al. 1992; Ray et al. 1992) and corn (Buckner et al. 1996). GGPP is also a precursor for tocopherols, phylloquinones and chlorophyll. As the first committed step in carotenoid biosynthesis, phytoene synthase has been considered to be a regulatory point in the pathway. Once phytoene, which has nine double bonds, is formed it is further desaturated to lycopene which contains 13 double bonds (Sandmann 1994). In bacteria these four desaturations are performed by one enzyme, phytoene desaturase, while in plants, two enzymes, phytoene desaturase and zeta-carotene desaturase, each of which adds two symmetrical double bonds are required (Klein et al. 1995). As double bonds are added and the number of conjugated double bonds increases, the carotenoids gain color from zeta-carotene (pale yellow) to lycopene (red). Cyclohexene rings can then be formed at either or both ends of the molecule. If two beta rings are formed, the product is beta-carotene; if one beta and one epsilon ring is formed, the product is alpha-carotene. These reactions are performed by the enzymes beta and epsilon cyclase (Cunningham et al. 1996). Both alpha-carotene and beta-carotene are orange-yellow in color. Up to this point in the pathway, the carotenoids are entirely hydrocarbon in nature. Further steps involve the addition of oxygen moieties and this subgroup is called xanthophylls. The addition of two hydroxyl groups to the 3 and 3′ positions of the rings on beta-carotene gives rise to zeaxanthin, while the addition of two hydroxyls to alpha carotene yields lutein. Lutein plays a central role in the photosynthetic apparatus. The hydroxylations of the beta and epsilon cyclohexene rings are performed by two separate enzymes (Sun et al. 1996). Further addition of 2 keto groups at the 4 and 4′ position of zeaxanthin via a ketolase (Misawa et al. 1995) yields astaxanthin, a prominent carotenoid in marine microalgae and the major pigment in salmonoid fishes. With the exception of epsilon-ring hydroxylase, all the enzymes involved in the pathway described above have been cloned from a variety of plant and/or microbial sources (Armstrong 1997; Cunningham & Gantt 1998).

Although animals do not synthesize carotenoids, carotenoids do serve a variety of functions in mammals and humans. Most widely known is beta-carotene, which serves as a dietary precursor of vitamin A in the body. Other carotenoids with at least one beta ring such as beta-cryptoxanthin and alpha-carotene can also serve as precursors to vitamin A (Bauernfeind et al. 1981; Gross 1991). A number of other carotenoids have been proposed to have positive health effects. These include lutein and zeaxanthin which may play a role in the prevention of age-related macular degeneration (Sommerburg et al. 1998) and lycopene which is a very potent antioxidant and may have positive cardiovascular effects as well as possible effects in preventing prostate cancer (Giovannucci et al. 1995; Kohlmeier et al. 1997). Additionally, due to their high degree of coloration, carotenoids are used as colorants in foods and animal feeds (Nelis & De Leenheer 1991). Beta-carotene is widely used as a food colorant in margarine and other foods. Lutein and zeaxanthin are routinely added to poultry feed to optimize yellow coloration in eggs and astaxanthin is added to the feed of farmed salmonoid fish to ensure an appropriate pink color.

The availability of a large number of genes as described above makes it possible to begin studying various aspects of the control of the carotenoid pathway and to explore the possibilities of engineering the carotenoid pathway in plants for enhanced nutrition or other commercial utility. Published results to date in this area have demonstrated a two- to fivefold increase in carotenoids in carrot root with a bacterial phytoene synthase (Hauptmann et al. 1997) and a small, but measurable accumulation of phytoene was detected in rice endosperm using a plant phytoene synthase (Burkhardt et al. 1997). The constitutive expression of a plant phytoene synthase gene in transgenic tomatoes resulted in dwarfism (Fray et al. 1995). This dwarfism was attributed to reduced giberellin levels. GGPP is the substrate both for phytoene synthase and the giberellin pathway and it was proposed that overexpression of the phytoene synthase leads to reduced levels of substrate for giberellin formation. This result underscores the necessity for using promoters of a variety of specificities when overexpressing carotenoid biosynthetic genes in transgenic plants. When a bacterial phytoene desaturase was expressed in transgenic tobacco (Misawa et al. 1993; Misawa et al. 1994), herbicide resistance and an alteration in xanthophyll composition was observed but no increase in carotenoids was obtained. The bacterial family of phytoene desaturases which perform all four desaturations are resistant to the herbicide norflurazon, while the plant phytoene desaturases which perform only two desaturations are a distinct family that is sensitive to norflurazon (Sandmann 1994). Thus, the tobacco plants expressing the bacterial phytoene desaturase were resistant to norflurazon. The work herein describes the results obtained when a bacterial phytoene synthase gene was used in the green embryo of Brassica napus. The increase in carotenoid levels shown in this study is quite dramatic (50-fold) and is one of the more effective increases in flux through a pathway demonstrated to date in transgenic plants.


Seed-specific expression of a bacterial phytoene synthase leads to a dramatically orange embryo

Phytoene synthase is the enzyme that catalyzes the first committed step in carotenoid biosynthesis and is one of the enzymes that has been proposed to regulate flux through the pathway (Fig. 1). To test the role of phytoene synthase in regulating carotenoid levels in the developing embryos, a seed-specific phytoene synthase construct was made. A seed-specific promoter from Brassica, napin (Kridl et al. 1991), was used in conjunction with a plastid targeted bacterial phytoene synthase (crtB) gene. The crtB gene was from Erwinia uredovora (Sandmann & Misawa 1992) and the plastid targeting sequence was from the small-subunit (SSU) of ribulose-bisphosphate carboxylase (Fig. 2). Transgenic canola was generated using this construct and the seeds visually followed throughout development. By 21–25 days post-anthesis (d.p.a.), T2 seeds were visually orange (Fig. 3a) and by 35–40 d.p.a. the embryos were very orange (Fig. 3b). In the siliques in Fig. 3(a), both green and orange seeds are visible. These are T2 (R1) seeds and are thus segregating for the napin-crtB trait. As the trait is easily discerned at 35–40 d.p.a., it is possible to determine segregation ratios and thus loci number on T2 (R1) seeds by determining the orange to green ratio. When a small number of seeds was extracted with hexane, it was clear that hydrophobic colored compounds (later confirmed as carotenoids) were being synthesized (Fig. 3c, lanes 3 and 4).

Figure 1.

Carotenoid biosynthetic pathway.

DMAPP is dimethylallyl diphosphate and IPP is isopentenyl diphosphate. *Phytoene synthase.

Figure 2.

Map of phytoene synthase expression chimera in pCGN3390.

The source of the bacterial phytoene synthase (crtB) is Erwinia uredovora. TP indicates the transit peptide of the small-subunit of ribulose bis-phosphate carboxylase.

Figure 3.

Napin-crtB embryos, plants and oil.

(a) Twenty-eight d.p.a. silque from S3390-1 (T2 seeds)

(b) Thirty-five d.p.a. embryos from S3390-1(T3 homozygoyte) and non-transgenic control.

(c) Lane 1: cold-pressed oil from non-transgenic S control; lane 2: cold-pressed oil from S3390-1 from Brawley field trial (homozygous seed); lane 3: complete extraction of 12 control S seeds in 2 ml hexane; lane 4: complete extraction of 12 S3390-1 seeds (T4 homozygous seed) in 2 ml hexane.

As seeds do not normally contain large amounts of carotenoids, it was important to determine if the seeds would sprout and the plants grow normally. The napin-crtB seeds do germinate. They may emerge from the soil 1–2 days later than controls. Depending on the light conditions, the seedlings may appear pink to orange to yellow for the first day before changing to green. In lower light conditions such as those used in growth chambers, the pink-orange-yellow color is usually observed. However, in the field where light intensity may be higher, the pink-orange-yellow color may not be seen. By 10–12 days after planting, the napin-crtB plants look similar to normal canola plants.

HPLC analysis reveals the major carotenoids produced in napin-crtB embryos are alpha and beta-carotene

To determine the carotenoid content and composition in the transgenic napin-crtB seeds, HPLC analysis was performed. A representative trace at 450 nm is shown in Fig. 4. Two new prominent peaks were observed. These correspond to alpha and beta-carotene. Other new small peaks are seen, including lycopene. The peak corresponding to lutein was basically unaffected in the transgenic sample as compared to the control. At 280 nm (data not shown), a new peak that corresponded to phytoene was also seen in the transgenic sample. Table 1 shows the HPLC analysis of several representative lines in two B. napus culitvars, Quantum (Q) and 212/86 (S) and is representative of that observed in greater than 90% of the transgenic lines. Total carotenoid concentrations increased up to 50-fold in these transgenic seeds. Growth for several generations, both in the greenhouse and the field, demonstrated that the trait is stable and reproducible. The S3390–1 line was grown in the field in California, seed was harvested and oil produced by a cold-press procedure. As can be seen in Fig. 3(c), lanes 1 and 2, the oil from the transgenic line is very dark orange consistent with the presence of high levels of carotenoids. The concentration of carotenoids in the transgenic oil (μg g− 1) was a little over twice that seen in transgenic seeds. As canola seeds routinely contain 35–50% oil (Salunkhe et al. 1992), this indicates that the hydrophobic carotenoids are partitioning with the oil as expected.

Figure 4.

HPLC analysis of mature control and pCGN3390 seed.

(a) S control.

(b) S3390-1 – Homozygous seed. Peaks are 1, solvent front; 2, lutein; 3, internal standard; 4, lycopene; 5, alpha-carotene; and 6, beta-carotene. Extracts were from similar tissue weights.

Table 1. . Carotenoid concentrations of canola seeds from selected lines transformed with phytoene synthase (crtB) gene
Carotenoid concentration (μg gFW− 1)
Sample IDGeneration segregation ratio & production siteLuteinLycopeneα-Caroteneβ-CarotenePhytoeneTotal
  1. Abbreviations: FW, fresh weight; GH, greenhouse; ND, not detected. Seeds were randomly sampled in each generation.

Q controlhomo, GH30NDND3ND33
Q3390–2T2, 15 : 1, GH5063727211921341
Q3390–9T2, >  63 : 1, GH68123949491941617
Q3390–12T2, 3 : 1, GH48104007391711368
Q3390–15T2, null, GH27NDND1ND28
Q3390–18T2, >  63 : 1, GH5084497591281394
Q3390–26T2, 3 : 1, GH3493115841491087
Q3390–37T2, 3 : 1, GH50102916261191096
Q3390–49T2, 15 : 1, GH2823466771461199
Q3390–12T4, homo, GH45103955654431458
Q3390–26T4, homo, GH3092346723931337
Q3390–26T4, homo, field57142793793441073
S controlhomo, GH31NDND5ND36
S3390–1T3, homo, GH5224406694301163
S3390–4T3, homo, GH44172826372391219
S3390–5T3, hetero, GH512191387120751
S3390–5T3, homo, GH4642566332201159
S3390–11T3, homo, GH54104065564271453
S3390–14T3, homo, GH66134316742631447
S3390–35T3, hetero, GH38212531476555
S3390–35T3, homo, GH445234504169956
S3390–1T4, homo, GH44263445991751188
S3390–1T4, homo, field72252254013321055

Molecular analysis was undertaken to correlate the carotenoid trait with the foreign gene expression. Northern analysis on RNA from developing seed from a napin-crtB line demonstrated that the crtB RNA was absent in control plants (Fig. 5a). In the transgenic plants, crtB RNA was first detected at 18 d.p.a., increased to 24 d.p.a., and then began to decrease at 30 d.p.a. This is a pattern of expression that would be expected from the napin promoter. Similarly, a Western blot for crtB protein (Fig. 5b) demonstrated that crtB protein was absent at 15 d.p.a., and accumulated as the seed aged. An immunoreactive polypeptide of the expected size for the processed protein, 33 kd, was present in addition to a smaller band that may come from either incorrect processing or degradation. Samples beyond 24 d.p.a. were also analyzed by Western blot, however, as the highly abundant seed storage protein cruciferin (Crouch & Sussex 1981; Sjödahl et al. 1991) migrates at the same molecular weight as crtB and interferes with the signal from crtB, these data are not shown. Carotenoid analysis on this same time course indicated that carotenoid production in the napin-crtB transgenic line began to rise above the control between 18 and 21 d.p.a. and continued to rise throughout development (Fig. 5c). The beta-carotene level continued to rise until the seed was mature, while the alpha-carotene concentration leveled off around 35–40 d.p.a. The increase in carotenoid level correlates both with the production of crtB mRNA as well as the presence of crtB protein. However, the fact that the carotenoid levels continue to rise while the mRNA levels drop, indicates that the crtB protein is probably quite stable in the transgenic seed or that little protein is needed to maintain the increase in carotenoid production.

Figure 5.

Developmental analysis of napin-crtB seeds

(a) Northern analysis of developing seed RNA. d =  days post anthesis.

(b) Western analysis of developing seed. d =  days post anthesis.

(c) Carotenoid analysis of developing seed.

Analysis of other isoprenoids and fatty acids reveals other metabolic changes

The substrate for phytoene synthase, GGPP, is also a precursor for tocopherols and chlorophylls in the plastids. To ascertain the effect of the overexpression of crtB on the levels of these compounds, HPLC analysis was performed on the developing seed of several pCGN3390 lines (Table 2). In normal canola seeds, chlorophyll levels are high during development and decline somewhat with age. In mature seeds, the chlorophyll levels have decreased significantly from the green, developing seed. In the transgenics, however, the chlorophyll levels were low throughout development. Strikingly, the chlorophyll levels were reduced by as much as 80% in some developing transgenic seeds. By the time the seeds were mature, this difference in chlorophyll levels had disappeared. In contrast to chlorophyll, tocopherol levels rise during development of normal seeds, with the majority of tocopherol accumulation coming late in development and with full maturity. In the transgenics, the levels of tocopherols failed to rise in the later developmental stages. This led to an overall average decrease in tocopherol levels of 50% in mature seeds. The decrease in tocopherols occurred mainly at the expense of gamma tocopherol. The levels of sterols, which are derived from non-plastidal isoprene units, were not changed in the mature transgenics as compared to the controls (data not shown).

Table 2. . Changes in chlorophyll and tocopherol concentrations during development of seeds from selected canola lines transformed with phytoene synthase (crtB) gene
Pigment concentration (μg gFW− 1)
sample IDDevelopmental stageTotal chlorophyllsδ-tocopherolγ-tocopherolα-tocopherolTotal tocopherols
  1. Abbreviations: d.p.a., days post-anthesis; FW, fresh weight. Total chlorophylls included chlorophyll a and b. Seeds were randomly sampled in each generation.

Q control35 d.p.a.5130302656
T2 Q3390–235 d.p.a.510493988
T4 Q3390–1235 d.p.a.1380332760
T2 Q3390–1835 d.p.a.9507943122
T2 Q3390–2635 d.p.a.1030493887
Q control45 d.p.a.5210492877
T4 Q3390–1245 d.p.a.940382967
Q control55 d.p.a.35409438132
T4 Q3390–1255 d.p.a.620394180
Q controlMature16926285356
T4 Q3390–12Mature2306499163
T4 Q3390–26Mature100434689
S control27 d.p.a.5570322557
T4 S3390–127 d.p.a.1790473178
S control40 d.p.a.31208646132
T4 S3390–140 d.p.a.4405943102
S control50 d.p.a.71415674234
T4 S3390–150 d.p.a.2506242104
S controlMature267210118335
T4 S3390–1Mature23011973192

As the change in carotenoid concentration was fairly dramatic, levels of other seed constituents were analyzed to determine if they had also been altered. Analysis of fatty acyl composition of a number of pCGN3390 lines is shown in Table 3. This revealed that the relative percentage of oleic acid (18 : 1) increased in the transgenic seeds coupled with a concomitant decrease in percentage of linoleic acid (18 : 2) and linolenic acid (18 : 3). This difference in fatty acyl composition held in several generations and under both greenhouse and field conditions, although the magnitude of the change varied with growth conditions.

Table 3. . Fatty acid composition of napin-crtB linesa
LineLocationGenerationSegregation ratio16 : 018 : 018 : 118 : 218 : 320 : 0
  • a

    All values were determined on random pools of 50 seeds. Each value represents the relative fatty acid percentage (w/w) of total fatty acids.

S ControlGHN/AN/A5.11.759.917.112.00.6
S ControlGHN/AN/A4.71.662.715.311.30.6
S3390–1GHT23 : 14.62.770.
S3390–25GHT2 >  63 :
S3390–15GHT23 : 14.41.569.314.97.71.4
S3390–4GHT23 : 15.02.467.414.58.61.2
S3390–21GHT215 :
S ControlFieldN/AN/A5.41.656.720.713.30.5
S ControlFieldN/AN/A5.31.656.420.913.50.5
Q ControlGHN/AN/A3.81.859.821.210.10.7
Q3390–2GHT215 :
Q3390–12GHT23 : 14.02.462.918.88.70.9
Q3390–9GHT2 >  63 : 13.72.665.318.26.81.1

Electron microscopy of napin-crtB embryos reveals ultrastructural changes

To ascertain if the accumulation of the high levels of carotenoids had altered the cell structure, electron microcroscopy was performed on embryos from 35 d.p.a. developing seed from transgenic and control plants (Fig. 6). The basic overall cell structure appeared similar with a predominance of lipid and protein bodies (Fig. 6a,c). The major difference was noted in the plastids. In the control seed, normal chloroplasts with thylakoid membranes and starch bodies are seen (Fig. 6b). In some plastids from the transgenic line (Fig. 6d), starch granules were present but the internal membrane structure no longer resembled a classic thylakoid structure. Instead, a thread-like inclusion body was seen that was surrounded by a membrane. The composition of the inclusion body or the membrane surrounding it was not determined. The majority of cells had at least one altered plastid. As we targeted the crtB gene product to the plastid and as carotenoid biosynthesis takes place in the plastid, one could speculate that the thread-like inclusion body contains some of the new carotenoids synthesized in these transgenic seeds.

Figure 6.

Electron micrographs of control S and S3390-1 seeds.

(a,b) Control S seed, 35 d.p.a.

(c,d) S3390-1 seed, 35 d.p.a. PB, protein body; L, lipid body; P, plastid; S, starch; arrow, thread-like inclusion body.


The increase in the carotenoids seen in the seeds of the napin-crtB plants was dramatic. The levels of carotenoids that accumulated in these seeds (1000–1500 μg gFW− 1) were as high or higher than many other plant sources, such as certain flowers and fruits which accumulate high amounts of carotenoids in chromoplasts. The mesocarp of oil palm seed contains total carotenoids in the range of 300–1000 μg gFW− 1 with the carotenoids in the pressed oil ranging from 500 to 2000 μg g− 1 (Clegg 1973; May 1994). Marigold petals, another source of high amounts of carotenoids, show levels of total carotenoids in the range of 400–5000 μg gFW− 1 (Quackenbush & Miller 1972). Other vegetable sources such as tomatoes and carrots have carotenoid concentrations in the range of 50–200 μg gFW− 1 (Edwards & Reuter 1967; Gross 1991; Johjima & Matsuzoe 1995). The fact that it was possible to increase carotenoids in a tissue that does not normally accumulate them to levels as high or higher than those found in tissues which do, indicates the flexibility of the pathway and accumulation mechanism, at least in Brassica seed.

Overexpression of phytoene synthase in other plant species did not result in as dramatic an increase in carotenoid levels as we obtained in the Brassica seeds. In the only published seed experiment, a plant phytoene synthase was expressed in the endosperm of rice leading to the accumulation of a small amount (about 1 μg gFW− 1) of phytoene (Burkhardt et al. 1997). Experiments in vegetative tissues using a bacterial phytoene synthase gene in the roots of carrots (Hauptmann et al. 1997) demonstrated a two- to fivefold increase in carotenoids but no significant change in composition. As carrots range from 50 to 100 μg gFW− 1 total carotenoids, this represents a final concentration of 250–500 μg gFW− 1 in the transgenics. Additionally, the constitutive expression of a plant phytoene synthase led to a 1.5–2-fold increase in carotenoids in the vegetative tissue in tomatoes (Fray et al. 1995). The differences between these results and those observed in Brassica seed may be due to the source of the phytoene synthase gene (plant versus bacterial); the tissue where expression is targeted (embryo versus endosperm versus vegetative); and/or the photosynthetic capability of the tissue, i.e. green versus white. The capacity of various tissues to make carotenoids may differ widely. It may be that in photosynthetic tissues, which make carotenoids and chlorophylls for the photosynthetic apparatus, there is a larger pool of substrates (IPP and GGPP) for carotenoid biosynthesis. It is also possible that the complement of downstream enzymes varies with tissues and plants. Thus, the endosperm of rice may not convert the phytoene to downstream products because it lacks a significant level of phytoene desaturase. Furthermore various tissues may have different capabilities to store unintended carotenoids.

The carotenoid composition of developing non-transgenic Brassica seed consists mainly of lutein, some beta-carotene and a small amount of alpha-carotene. This is a carotenoid composition that is observed in green, photosynthetic tissues in addition to the higher xanthophylls violaxanthin, antheraxanthin and neoxanthin (Gross 1991). In the transgenic napin-crtB plants, the main carotenoids produced were alpha and beta-carotene with a significant amount of phytoene. This indicates that although phytoene synthase may be a primary rate limiting step in these tissues, other enzymes may be limiting as well. The lack of a substantial increase in lutein and the other xanthophylls demonstrates that the hydroxylase enzymes may be limiting in the Brassica seed tissue. Furthermore, the accumulation of phytoene indicates that phytoene desaturase may also be somewhat limiting in these tissues. The fact that these seeds do germinate is also noteworthy. No known seeds have levels of carotenoids that high and none accumulate phytoene, alpha-carotene and beta-carotene. This perhaps indicates that the germinating embryo can break down the carotenoids rapidly. It remains to be determined what causes the 1 or 2 day lag in germination in these seeds.

In the developing transgenic Brassica seeds, the flux through the carotenoid pathway has increased and the amounts of the other plastidal isoprenoids that are derived from GGPP, tocopherols and chlorophyll, have decreased. If we compare total flux or carbon units of GGPP to carotenoids, chlorophylls and tocopherols that are present in these tissues, we find that overall the total number of GGPP units has increased by about 25–100% in the transgenic seeds. These calculations were made from 35 d.p.a. and 45 d.p.a. seeds as the carotenoids are accumulating and the chlorophyll levels have not begun to drop. If the comparison is made on mature seeds, a 400% increase in GGPP units has accumulated in the seeds. This indicates that there is some flexibility in the overall flux through the isoprenoid pathway to GGPP; however, the overall amount of isoprenoid units available to the plastid may be limited by steps prior to phytoene synthase, most likely in the non-mevalonate pathway.

The alteration in fatty acid composition was unexpected. The change cannot be accounted for simply by the stoichiometric increase in double bonds needed to generate the increased carotenoids. Although one can postulate on possible mechanisms, further experiments are needed before any theories can be put forth.

In tissues that normally produce large amounts of carotenoids, the carotenoids are stored in modified plastids termed chromoplasts (Camara et al. 1995). In these chromoplasts, the carotenoids are complexed with lipids and specific carotenoid binding proteins. It is postulated that these lipoprotein complexes are formed to sequester the carotenoids away from other structures to avoid possible harmful affects (Deruere et al. 1994). The electron micrographs of the developing napin-crtB seeds show that the structure of some plastids has been altered. The classic thylakoid membrane structure associated with chloroplasts is no longer evident in these altered plastids. A new inclusion body surrounded by a membrane is noted. Although this structure does not look like known chromoplasts, it is possible that the new structure is the cell's mechanism to sequester the carotenoids away from the rest of the plant cell. As carotenoid binding proteins are not normally present in developing seed, this inclusion structure may be an alternate approach to dealing with excess carotenoids. Although we have targeted the phytoene synthase gene to the plastid, it is possible that some of the excess carotenoids may be sequestered outside of the plastids in these cells. Another approach to dealing with large amounts of carotenoids is found in the green alga, Haematococcus pluvialis, which accumulates large amounts of astaxanthin in separate subcellular structures from the chloroplast (Boussiba & Vonshak 1991; Santos & Mesquita 1984). The alga Dunaliella bardawil, which accumulates large amounts of beta-carotene, deposits it in oily globules in the interthylakoid space of the plastid (Ben-Amotz et al. 1989). This indicates that cells have evolved a variety of ways of sequestering excess carotenoids in addition to the formation of chromoplasts.

The magnitude of increase in the carotenoid levels seen in the transgenic Brassica seeds is noteworthy. Previous studies on raising levels of secondary metabolites in plants have not led to such dramatic increases (Goddijn & Pen 1995). Two of the more recent successes in this area include the two- to fourfold increase in alkaloids produced in transgenic Atropa belladonna (Yun et al. 1992) and the sixfold increase in sterols produced in transgenic tobacco (Schaller et al. 1995). Another dramatic example of the engineering of secondary metabolites was the demonstration of the induction of anthocyanin production in transgenic maize cell culture (Grotewold et al. 1998; Taylor 1998)

In addition to shedding light on the regulation of carotenoid biosynthesis, these results also indicate that Brassica, and perhaps other oil seed crops might be used as commercial sources of carotenoids. The initial trait described in this paper can be a source of beta-carotene, both as a precursor to vitamin A in the diet and as a colorant. The addition of other genes on this original trait may allow the production of other useful carotenoids for supplements, colorants or animal feeds.

Experimental procedures

Binary vector construction

The phytoene synthase gene (crtB) from Erwinia uredovora in pCRT-B (pCAR-E) (Misawa et al. 1990) was obtained from N. Misawa (Kirin Brewery, Tokyo, Japan). To target the crtB gene to the plastid, a pea SSU transit peptide sequence was attached creating pCGN3373. The transit peptide sequence used is that previously used to target crtI to plastids (Misawa et al. 1993). The resulting fusion had the following amino acid sequence surrounding the transit peptide cleavage site (-Val-Lys-Cys-Met-Asn-Asn-Pro-Ser-Leu-Leu-Asn-His-Ala-Val-Glu-Thr-Met-Ala-Val-Gly-). To obtain expression of the SSU-crtB fusion in the seeds, it was inserted into a napin expression cassette, pCGN3223 (Kridl et al. 1991; Voelker et al. 1992) creating pCGN3389. This was then inserted into a plant transformation vector containing a 35S-nptII selectable marker, pCGN1559PASS (a derivative of pCGN1559, McBride & Summerfelt 1990) giving rise to the vector, pCGN 3390.

Plant transformation and materials

Agrobacterium-mediated transformation of hypocotyl explants of Brassica napus L. Cultivar 212/86 (S) or Quantum (Q) were carried out as described previously (Radke et al. 1992).

Generation notation is as follows. T1(R0) refers to the original plants regenerated off the hypocotyls. T2 seeds refers to the seeds from T1 plants. T2 plants are second generation plants grown from T2 seeds. Further generations are numbered accordingly.

To determine loci number, 10 35 d.p.a. siliques from T1 transgenic plants (T2 seeds) were collected and the number of orange and green seeds counted. To obtain homozygote plants T2 plants from events with one loci (3 : 1) were planted and the T3 siliques scored at approximately 35 d.p.a. Homozygotes were those in which all of the seeds were orange.

Western analysis

Protein was extracted from control and transgenic Brassica seeds at various days post-anthesis according to the protein extraction method of Bruneau et al. (1991). Twenty-five to 30 frozen seeds per line were ground in a mortar and pestle with liquid nitrogen. Ground tissue was added to Falcon tubes containing 2 ml of boiling DB +  buffer (2% SDS, 40 mm Tris pH 8, 20% glycerol, 1% beta-mercaptoethanol) and boiled for 10 min. Samples were centrifuged at 6000 g for 10 min. The supernatant was collected and the protein quantified by the BCA assay (Pierce). Five μg of proteins per sample were run on a pre-cast 12% acrylamide gel (Novex). The proteins were electroblotted to Immobilon (Millipore) and blocked in 1% BSA in Tris buffered saline [TBST: 50 mm Tris–HCl (pH 7.5), in 200 mm NaCl, 0.05% Tween 20]. The membranes were incubated either overnight at 4°C or at room temperature for 2 h with a 1 : 1000 dilution of primary antibody. A rabbit polyclonal antibody was kindly provided by Dr N. Misawa (Kirin Brewery). The Western blot was developed using an alkaline phosphatase system according to the manufacturer's instructions (Promega).

Northern analysis

RNA was isolated from control and transgenic Brassica seed (cv. 212/86) at various days post-anthesis. The method of RNA extraction was adapted from a CTAB (hexadecyltrimethylammoniumbromide) DNA isolation protocol (Webb & Knapp 1990). Approximately 30 seeds were ground in liquid nitrogen with a mortar and pestle. The powder was homogenized at room temperature with the mortar and pestle in 10 ml of extraction buffer (50 mm Tris–HCl, pH 9, 0.8 m NaCl, 10 mm EDTA, 0.5% [w/v] CTAB, 1%BME, 2% [w/v] polyvinylpyrrolidone). The insoluble fraction was removed by centrifugation for 5 min at 12 900 g. The supernatant was filtered through Miracloth (Calbiochem) and deproteinized by the addition of 3 ml of chloroform. After centrifugation for 5 min at 12 900 g, the upper aqueous phase was removed and mixed with 1 volume of ethanol and centrifuged again for 20 min at 12 900 g. The resultant drained RNA pellet was immediately resuspended in 25 μl of 5 mm EDTA pH 8.0. Twenty μg of RNA from each sample was run on a 1.4% agarose formaldehyde gel and transferred overnight to Nytran (Schleicher & Schull) in 20 ×  SSC. The RNA blot was baked at 80°C for 2 h then pre-hybridized in a 42°C water bath in pre-hybridization buffer of 50% formamide, 10 ×  Denhardts, 0.025 m NaPO4, 5 ×  SSC, 5 mm EDTA, 0.1% SDS, 100 mg ml− 1 salmon sperm DNA. A 1.1 kb fragment containing crtB was obtained by digesting pCRT-B (Misawa et al. 1990) DNA with restriction enzymes HindIII and EcoRI and subsequent gel purification via phenol:chloroform extraction. The crtB fragment was radiolabelled with 32P by Stratagene Prime-It II random priming kit. The crtB probe was added to fresh hybridization buffer containing pre-hybridization buffer and 10% dextran sulfate and incubated with the RNA blot at 42°C overnight. The blot was rinsed once at 50°C for 30 min in 1 ×  SSC, 0.2% SDS and twice for 15 min in 0.1 ×  SSC, 0.2% SDS. The Northern blot was exposed to X-Omat XAR-2 film (Kodak) for 5 days at − 70°C.

Analysis of carotenoids, tocopherols and chlorophylls

The method employed was modified from Craft (1992). Approximately 100 mg of randomly sampled canola seeds (∼25 seeds) was extracted in 3 ml hexane/acetone/ethanol (50/25/25, v/v) using a mortar and pestle. The residue was extracted two or more times with 2 ml of the extraction solvent until no color was found in the extracted tissue. The extracts were combined and centrifuged at medium speed for 3 min using a tabletop centrifuge. The supernatant was carefully removed and transferred to a glass tube and evaporated under nitrogen gas. Next, 3 ml hexane and 1 ml methanol was used to dissolve the residue, and 1 ml saturated NaCl was added to the tube and mixed. The sample was centrifuged for 3 min for better phase separation. The top layer (with hexane and isoprenoids) was removed and transferred to another glass tube. Another 2 ml hexane was added to the bottom phase, mixed, centrifuged, and the top layer was removed and combined with the previous one. This phase extraction was repeated for one or more times until no color was found in the top layer. The hexane extracts were dried under nitrogen gas. The residue was dissolved in 2 ml acetonitrile/methylene chloride/methanol (50/40/10, v/v). The sample was centrifuged for 3 min. About 1 ml supernatant was filtered through a 0.45 μ filter and collected in a brown autosampler vial and capped immediately. Recovery of isoprenoids during extraction was usually >  95% by using β-apo-8′-carotenal as internal standard for quantification. Carotenoids, tocopherols and chloro-phylls were analyzed with a Hewlett Packard 1100 high-performance liquid chromatograph using a Spherisorb ODS2 reverse phase C-18 column (5 μ, 4.6 mm ×  25 cm) for separation and two detectors for quantification. Mobile phase was acetonitrile/dioxane/methanol containing 150 mm ammonium acetate/triethylamine (82/10/8/0.1, v/v), flow rate was 1 ml min− 1, and the injection volume was 20 μl. A photodiode array detector was used for routine detection of colored carotenoids at 450 nm, phytoene at 280 nm, chlorophyll a at 663 nm, and chlorophyll b at 645 nm. Spectral scans of peak purity were made between 260 and 700 nm. Spectra of peaks at upslope, apex and downslope were normalized and overlaid. Superimposing spectra are taken as evidence of peak purity. Peak identities were assigned by comparing retention times and spectra of sample peaks to those of known standards. Tocopherols were measured by a fluorescence detector using 290 nm as excitation wavelength and 330 nm as emission wavelength.

Determination of fatty acyl composition

Fatty acyl composition was determined by transmethylation and GLC of the methyl esters according to published methods (Browse et al. 1986).

Electron microscopy

Control and transgenic Brassica seeds were collected at 35 days post-anthesis. The seed coats were removed prior to fixing the seeds in 4% glutaraldehyde in 0.1 m sodium cacodylate buffer, pH 7.2, for 4 h at room temperature. The embryos were then washed in cacodylate buffer, cut into 1 mm3 pieces and post-fixed with 1% osmium tetroxide in 0.1 m cacodylate buffer for 2 h. The tissue was dehydrated using ethanol followed by infiltration with propylene oxide and a 1 : 1 mixture of Spurr’s:EMbed 812 resin. The resin was polymerized at 60°C for 48 h. The resulting blocks were sectioned on a Leica Ultracut E microtome using a Diatome diamond knife. Sections 80 nm thick were collected on commercially prepared formvar/carbon coated copper slot grids. The sections were stained with uranyl acetate and lead citrate in an LKB Ultrastainer. Micrographs were obtained on a JEOL 1200 transmission electron microscope (all reagents were obtained from Electron Microscopy Sciences, Fort Washington, PA, USA).


We wish to thank Dr N. Misawa (Kirin Brewery) for both the Erwinia herbicola crtB gene and the gift of crtB antibody. Co-cultivations of pCGN3390 in 212/86 and Quantum were kindly done by Joann Turner, Jean Shepard and Jason Thomas. The excellent care given to the plants in the greenhouse was provided by Kendra Williams, Brenda Reed and Ruthellen Davis. The plants in the field were maintained by Tony Ballman, Jim Byrne and Sharon Radke. We thank Tom Hayes and colleagues in our oils analytical department for performing the analysis of fatty acid composition. For encouragement and advise throughout various stages of the project we thank Vic Knauf, Rick Stonard and Ganesh Kishore. We also wish to thank Dr Toni Voelker for the photographs of the 28 d.p.a. siliques and Vic Knauf, Dave Stalker and Tom Savage for careful reading of the manuscript.