Lutein and zeaxanthin cannot be synthesized de novo in humans, and although lutein is abundant in fruit and vegetables, good dietary sources of zeaxanthin are scarce. Certain corn varieties provide adequate amounts because the ratio of endosperm β : ε lycopene cyclase activity favours the β-carotene/zeaxanthin branch of the carotenoid pathway. We previously described a transgenic corn line expressing the early enzymes in the pathway (including lycopene β-cyclase) and therefore accumulating extraordinary levels of β-carotene. Here, we demonstrate that introgressing the transgenic mini-pathway into wild-type yellow endosperm varieties gives rise to hybrids in which the β : ε ratio is altered additively. Where the β : ε ratio in the genetic background is high, introgression of the mini-pathway allows zeaxanthin production at an unprecedented 56 μg/g dry weight. This result shows that metabolic synergy between endogenous and heterologous pathways can be used to enhance the levels of nutritionally important metabolites.
The carotenoids are a group of about 800 lipid-soluble organic molecules whose spectral properties are responsible for the yellow, orange and red colours of the tissues in which they accumulate (Britton et al., 2004). Carotenoids are synthesized by all photosynthetic organisms and many nonphotosynthetic bacteria and fungi. In plants, they protect the photosynthetic apparatus from photo-oxidation and act as precursors for the growth regulator abscisic acid, as well as attracting pollinators and seed-distributing herbivores (Creelman and Zeevart, 1984). In animals, they function generally as antioxidants, but specific carotenoids are essential nutrients because they act as precursors for important molecules that cannot be synthesized de novo. In humans, for example, carotenoids that contain at least one unsubstituted β-ring (α-, β- and γ-carotene and β-cryptoxanthin) are needed to synthesize the visual pigment retinal and the morphogenetic regulator retinoic acid (Botella-Pavia and Rodriguez-Concepcion, 2006).
The biosynthesis of all C40 carotenoids involves a series of common steps leading from geranylgeranyl pyrophosphate to lycopene (Cunningham and Gantt, 1998; Sandmann, 2001; Zhu et al., 2009). The pathway then splits into two branches, one producing α-carotene and the other β-carotene, both of which are hydroxylated in two steps to generate lutein and zeaxanthin, respectively (Figure 1). The amount of lutein and zeaxanthin therefore depends on the relative activities of the enzymes lycopene β-cyclase and lycopene ε-cyclase, the former introducing β-ionone rings at both ends of lycopene to generate β-carotene, and the latter introducing an ε-ionone ring at one end, the other end being cyclized by lycopene β-cyclase to generate α-carotene (Cunningham and Gantt, 1998; Sandmann, 2001; Zhu et al., 2009). In Arabidopsis, the introduction of hydroxyl moieties into the cyclic end groups by β-carotene hydroxylase (BCH and CYP97A) and carotene ε-hydroxylase (CYP97C) results in the formation of zeaxanthin from β-carotene and lutein from α -carotene (Kim et al., 2009). Two classes of structurally unrelated enzymes catalyse these ring hydroxylations: a pair of nonhaeme di-iron hydroxylases (BCH1 and BCH2) and two haeme-containing cytochrome P450 hydroxylases (CYP97A and CYP97C) (Kim et al., 2009; Vallabhaneni et al., 2009). CYP97A and CYP97C have been identified from Arabidopsis knockout mutants. CYP97A (LUT5 locus) encodes a β-carotene hydroxylase with predominant activity towards the β-ring of β-carotene and lesser activity towards the β-ring of α-carotene (Kim and DellaPenna, 2006; Kim et al., 2009). CYP97C encoded by the LUT1 locus can hydroxylate both the β- and ε-ring of α-carotene (Tian et al., 2004; Kim et al., 2009).
The most widely grown corn varieties have either white or yellow endosperm, in which the β : ε ratio (expressed as the ratio of β, β-carotenoids (γ-carotene, β-carotene, β-cryptoxanthin and zeaxanthin) to β, ε-carotenoids (δ-carotene, α-carotene, α-cryptoxanthin and lutein) varies by cultivar and can be altered by metabolic engineering (Zhu et al., 2008) or conventional breeding (Harjes et al., 2008). We have previously introduced genes encoding the early part of the carotenoid pathway into carotenoid-deficient white M37W corn, increasing the endosperm carotenoid content and generating novel carotenoid profiles depending on the combination of expressed genes (Zhu et al., 2008; Naqvi et al., 2009). The white corn inbred line M37W has a β : ε ratio of 1.21 but this was increased to 3.51 in a transgenic line (Ph-4) expressing corn phytoene synthase 1, a bacterial phytoene desaturase and Gentiana lutea lycopene β-cyclase (Zhu et al., 2008). The carotenoid pathway has been dissected by complementation analysis in bacteria, so we set out to establish whether the same principle could be applied to metabolic engineering in corn. We therefore introgressed the transgenic mini-pathway from the Ph-4 line into two yellow corn inbreds, EP42 and A632, which have contrasting β : ε ratios of 0.61 and 1.90, respectively. The difference in β : ε ratio between the hybrids correlates with the levels of endogenous lyce, the higher level in EP42 favouring higher lutein accumulation and the lower level in A632 favouring higher zeaxanthin accumulation. We found that the β : ε ratio of the Ph-4 × EP42 hybrid was reduced to near the mean value of the two parental lines whereas the β : ε ratio of the Ph-4 × A632 line was increased to 6.80, allowing it to accumulate zeaxanthin at an unprecedented 56 μg/g dry weight (DW). This showed that zeaxanthin synthesis in the two parental lines is limited by different bottlenecks that can be overcome by complementation in the hybrid. This synergistic response shows that metabolic attributes in different corn genetic backgrounds can be combined in hybrids to overcome bottlenecks and generate novel plant lines producing extraordinary levels of specific metabolites.
Carotenoid levels in wild-type corn endosperm
Corn endosperm was excised 25 days after pollination (DAP) from EP42, A632 and M37W, which were grown in the greenhouse under controlled conditions (Zhu et al., 2008; Naqvi et al., 2009). The basal carotenoid content was determined by HPLC (Figure 2), which showed that M37W endosperm contained only trace amounts of carotenoids (max. 1.42 μg/g DW), whereas the yellow inbreds EP42 and A632 accumulated up to 29 and 26 μg/g DW, respectively. Lutein and zeaxanthin were the predominant carotenoids in yellow corn endosperm, but the two lines showed contrasting β : ε ratios. Low levels of β-carotene, and α- and β-cryptoxanthin (the immediate precursors of lutein and zeaxanthin) were also detected in EP42 and A632 endosperm (Figure 2).
Transcript levels for carotenogenic genes in wild-type corn endosperm
Real-time RT-PCR was used to compare the transcript levels of enzymes in the carotenoid synthesis pathway, and relative expression levels were determined by adjusting Ct values to standard curves derived from serial cDNA dilutions generated separately for each gene in each experiment, normalized against actin mRNA (Figure 3). The mRNA for psy1 (encoding the endosperm-specific isoform of phytoene synthase) was not detected in the white M37W endosperm but it was present in the yellow endosperm varieties. In contrast, psy2 mRNA (encoding the isoform of phytoene synthase that is preferentially expressed in vegetative tissues) was present at low levels in all three varieties, but was more abundant in M37W endosperm which perhaps accounts for the trace levels of carotenoids detected in this inbred. The levels of pds (phytoene desaturase) and zds (ζ-carotene desaturase) mRNAs were similar in all three lines, but the yellow varieties had a slightly higher level of crtiso (carotenoid isomerase) than M37W. The mRNA for lycb (lycopene β-cyclase) was similar in the two yellow inbreds and lower in M37W, suggesting the absolute activity of lycopene β-cyclase cannot be responsible for the different β : ε ratios in EP42 and A632. In agreement with this, EP42 endosperm accumulates much higher levels of lyce (lycopene ε-cyclase) mRNA than the other varieties, and also higher levels of the bch1 and bch2 (β-carotene hydroxylase) mRNAs. The CYP97A mRNA was more abundant in M37W compared to the two yellow corn varieties. However, the level of CYP97C transcript was similar in the three corn backgrounds.
Carotenoid levels and mRNA analysis in transgenic and hybrid corn lines
Transgenic line Ph-4 expressing corn psy1, bacterial crtI (encoding a multifunctional phytoene desaturase that catalyses the three steps requiring pds, zds and crtiso in plants) and G. lutea lycb in a M37W background was used to introgress the transgenic mini-pathway into the EP42 and A632 backgrounds. The novel hybrids, designated Ph-4 × EP42 and Ph-4 × A632, were selfed for two generations to obtain homozygous lines expressing all three transgenes. RT-PCR analysis was then carried out on endosperm tissue from wild-type M37W, EP42 and A632 plants along with Ph-4, Ph-4 × EP42 and Ph-4 × A632.
It was clear that all three transgenes were expressed as strongly in the hybrid lines as in the transgenic parent (Figure 4). The hybrid lines presented a bright orange endosperm phenotype indicating the accumulation of more carotenoids, or different carotenoids compared to the transgenic parent (Figure 2b). HPLC analysis showed that the wild-type M37W endosperm contained a maximum of 1.42 μg/g DW total carotenoids, which increased to a maximum of 127 μg/g DW in transgenic line Ph-4 where it comprised up to 10.42 μg/g DW lycopene, 41.20 μg/g DW β-carotene and 29.64 μg/g DW zeaxanthin (Figure 2a). Ph-4 also accumulated phytoene and other intermediates such as α- and β-cryptoxanthin and α-carotene at lower levels (Figure 5a,b). In Ph-4 × EP42, the total carotenoid content increased to 90.32 μg/g DW, including 19.31 μg/g DW β-carotene, 23.41 μg/g DW lutein and 38.07 μg/g DW zeaxanthin. In Ph-4 × A632, the total carotenoid content was 88.53 μg/g DW, including 15.24 μg/g DW β-carotene, 9.72 μg/g DW lutein and 56.49 μg/g DW zeaxanthin (Figure 5a,b). Both lines also accumulated intermediates such as α- and β-cryptoxanthin, but unlike the Ph-4 parent there was no evidence of phytoene or lycopene. The β : ε ratio of both hybrids was significantly higher than the wild-type parents, 2.05 in the case of Ph-4 × EP42 and 6.80 in the case of Ph-4 × A632.
The carotenoids lutein and zeaxanthin accumulate to high levels in the human retinal macula and are thought to play an important role in the prevention of ARMD (Mozaffarieh et al., 2003). These pigments are also thought to protect the macula and photoreceptor outer segments throughout the retina from oxidative stress (Rapp et al., 2000). Insufficient dietary intake reduces plasma lutein and zeaxanthin concentrations resulting in low macular pigment density and an increased risk of pathological changes in the macula, eventually leading to ARMD (Semba and Dagnelie, 2003).
Whereas lutein is abundant in most fruits and vegetables, corn is one of very few dietary sources of zeaxanthin and even so only selected varieties produce adequate amounts of this molecule. The amount of zeaxanthin in inbred corn lines ranges from 0.6 to 27 μg/g DW (Quackenbush et al., 1963), reflecting corn’s diverse gene pool and suggesting that conventional breeding could be used to improve zeaxanthin levels (Kurilich and Juvik, 1999). Several corn mutants have been mapped to carotenoid genes (Matthews et al., 2003) and a number of QTLs affecting carotenoid accumulation have also been described (Wong et al., 2004) but most studies have concentrated on increasing the production of β-carotene and have disregarded the impact of zeaxanthin. Recently, Harjes et al. (2008) described four polymorphisms in the corn lyce locus which were used to improve β-carotene levels in seeds by selecting for low lycopene ε-cyclase levels. As zeaxanthin is derived from β-carotene, these polymorphisms would be a good starting point for breeding high-zeaxanthin corn, although the theoretical maximum levels would still be limited by the wild-type corn gene pool.
We recently described a combinatorial nuclear transformation strategy that can be used to modulate the carotenoid biosynthetic pathway in the endosperm of white corn (Zhu et al., 2008). We introduced five transgenes encoding corn phytoene synthase 1 (psy1), Pantoea ananatis phytoene desaturase (crtI), G. lutea lycopene β-cyclase (lycb) and β-carotene hydroxylase (bch), and Paracoccus sp. β-carotene ketolase (crtW) and recovered transgenic plants expressing different combinations of transgenes and generating corresponding profiles of novel carotenoids. The Ph-4 line, expressing psy1, crtI and lycb, recapitulated the entire pathway through to β-carotene and thus accumulated β-carotene as the major carotenoid, but there were also higher levels of lycopene, lutein and zeaxanthin, the latter suggesting that endogenous BCH activity was converting some of the additional β-carotene into zeaxanthin and demonstrating there was spare capacity in the pathway. The β : ε ratio in this transgenic line was 3.51 (compared to 1.21 in wild-type M37W endosperm) with a 134-fold increase in total carotenoids and a zeaxanthin level of up to 29 μg/g DW. Although many other crops have been engineered to improve carotenoid levels, previous studies have concentrated mainly on β-carotene and few have looked at the possibility of zeaxanthin production (DellaPenna and Pogson, 2006; Sandmann et al., 2006). In tomato fruits, the lycopene pool was successfully converted into β-carotene by the expression of Arabidopsis lycopene β-cyclase under the control of a fruit-specific promoter (Rosati et al., 2000), and the addition of β-carotene hydroxylase (bch) under the control of the same promoter increased zeaxanthin levels up to 24 μg/g FW (Dharmapuri et al., 2002). Antisense and co-suppression strategies against zeaxanthin epoxidase in potato elevated zeaxanthin up to 40 μg/g DW, which is a 130-fold increase over wild-type (Romer et al., 2002). Our research is, however, the first to demonstrate zeaxanthin metabolic engineering in a staple cereal crop, an important breakthrough considering that much of the developing world subsists on cereals and does not have access to fresh fruits and vegetables (Farre et al., 2010).
The existence of both wild-type corn varieties and transgenic lines with high zeaxanthin levels suggested the tantalizing possibility of combining the productive transgene combination of Ph-4 with the permissive genetic background of zeaxanthin-dense natural varieties to enhance zeaxanthin levels even further. Complementation analysis is often used to characterize the carotenoid pathway, with bacterial strains containing partial pathways or deliberately introduced bottlenecks used to test the function of novel carotenoid genes. We envisaged applying this principle to deliberately eliminate nutritional bottlenecks in corn hybrids, allowing any tendency towards the production of zeaxanthin to be augmented by the unrestricted flux through to β-carotene. We tested this hypothesis by introgressing the carotenoid mini-pathway from the Ph-4 transgenic line into two yellow inbreds with similar carotenogenic potential but opposite characteristics in terms of the β : ε ratio. Both EP42 and A632 accumulate carotenoids at levels up to 29 μg/g DW, with lutein and zeaxanthin as major components of the carotenoid profile. However, the low β : ε ratio of EP42 corn provides a much higher potential for lutein synthesis because ε-cyclase activity exceeds β-cyclase activity, whereas the much higher β : ε ratio of A632 corn favours the synthesis of zeaxanthin because the balance of enzyme activities is reversed. Our RT-PCR experiments showed that the two inbred lines had similar levels of lycb mRNA but differed with respect to the abundance of lyce mRNA, suggesting that the higher β : ε ratio of A632 corn reflected lower lycopene ε-cyclase activity. The increased β : ε ratio in the Ph-4 × A632 hybrid shows that a combination of reduced ε-cyclase activity and increased β-cyclase activity (provided by the transgene) can help skew the ratio even further, leading to the production of 56 μg/g DW zeaxanthin, more than twice the level produced in the best-performing natural varieties or the best-performing transgenic lines reported thus far. The Ph-4 × EP42 hybrid also showed an increased β : ε ratio of 2.05 compared to the wild-type parent, but in this case the higher lycopene ε-cyclase activity provided by the parental genotype prevented the balance tipping too far towards the zeaxanthin pathway.
As shown in Figure 5a, the Ph-4 line not only accumulated high levels of zeaxanthin and lutein, but also carotenoid intermediates such as phytoene, lycopene, α- and β-cryptoxanthin, and α- and β-carotene (Zhu et al., 2008). In the hybrid lines, however, these intermediates were not detected, which suggests that additional, more subtle bottlenecks present in the M37W genetic background were also alleviated by metabolic complementation. M37W endosperm accumulates only traces of carotenoids because of the lack of psy1 expression, which catalyses the first committed step in the pathway, the synthesis of phytoene. In contrast, psy1 mRNA is abundant in both yellow corn inbreds (EP42 and A632). The traces of lutein and zeaxanthin in M37W endosperm probably reflect the presence of the psy2 transcript, which is mainly responsible for carotenoid biosynthesis in green tissues but may have some residual activity in the endosperm. Interestingly, the transgenic line Ph-4 contains a significant amount of phytoene (7.36 μg/g), which suggests that the next step in the pathway (the conversion of phytoene into lycopene by the bacterial enzyme phytoene desaturase) is limiting. In contrast, no phytoene was detected in yellow corn, nor in the hybrids, which shows that the three enzymes carrying out the corresponding endogenous reactions in yellow corn are not limiting and alleviate the bottleneck in the M37W background when the induced and endogenous pathways are combined in the hybrid. Similarly, the transgenic endosperm also contained significant amounts of lycopene (10.42 μg/g) whereas no lycopene was detected in either hybrid. Phytoene and lycopene accounted for ca: 14% of total carotenoids in Ph4. Again, this suggests that the lycopene β-cyclase provided by the transgene introduces a bottleneck in the M37W background which is overcome by the additional lycopene β-cyclase activity in yellow corn. The disappearance of carotene intermediates such as phytoene and lycopene in the two hybrids might also be due, at least in part, to lower plastidial methylerythritol 4-phosphate pathway derived isoprenoid precursor availability in the yellow endosperm lines and/or the higher catabolic activities of zeaxanthin epoxidase and carotenoid cleavage dioxygenase (CCDs) in the yellow endosperm backgrounds. Both hybrid lines contained significant amounts of β-carotene (19 μg/g DW in Ph-4 × EP42 and 15 μg/g DW in Ph-4 × A632) but this was much lower than the 41 μg/g DW we measured in the transgenic parent. A plausible reason for this difference might be the higher levels of bch2 (β-carotene hydroxylase 3 in Vallabhaneni et al., 2009) accumulation in the two yellow lines compared to M37W. Bch2 transcript levels at 25 DAP have been reported to be negatively correlated with β-carotene accumulation and positively correlated with zeaxanthin levels in corn (Vallabhaneni et al., 2009). It is unlikely that differences in lutein accumulation between EP42 and A632 (and the corresponding hybrids with Ph-4) are attributed to cytochrome p450-type hydroxylases, as expression levels of these genes, at least at the mRNA level, were very similar. These data suggest that the yellow corn backgrounds also alleviated a bottleneck in β-carotene hydroxylase activity, allowing the efficient flow of intermediates towards zeaxanthin synthesis. Our collective data indicate that the yellow corn background compensated for inefficient activity at every step of the pathway conferred by the transgenes, but the combination of reduced lycopene ε-cyclase activity and the pooled lycopene β-cyclase activity in the hybrid conferred its highly skewed β : ε ratio and its extraordinary potential to accumulate zeaxanthin.
This study is the first to show that significant increases in zeaxanthin levels in a food crop can be achieved by combining conventional breeding with limited genetic engineering. Whereas genetic engineering provides advantages such as speed and access beyond the species gene pool, it can be difficult and/or time-consuming to transform locally adapted varieties directly and therefore make a practical impact on nutrition and health, particularly in developing countries where staples such as corn represent the predominant food source for many people. Conventional breeding for improved nutrition is slow and laborious, particularly where the intent is to modify several different metabolic pathways simultaneously (Naqvi et al., 2009), and is limited to the gene pools of compatible species. Our combined approach cherry-picks the advantages of both systems—the speed, power and accessibility of genetic engineering and the diversity and practicality of conventional breeding, to generate nutritionally enhanced crops with unprecedented levels of a key nutrient in the human diet.
Corn (Zea mays) varieties M37W (white endosperm), EP42 and A632 (yellow endosperm), and transgenic line Ph-4 expressing corn psy1, Pantoea annatis crtI and G. lutea lycb were grown in the greenhouse at 28/20 °C day/night temperature with a 10-h photoperiod and 60%–90% humidity during the first 50 days, followed by maintenance at 21/18 °C day/night temperature with a 16-h photoperiod thereafter. Ph-4 was out-crossed with EP42 and A632, and homozygous T2 and T3 generations were derived by selfing. Endosperm samples from immature seeds were frozen in liquid nitrogen and stored at −80 °C prior to use. M37W was obtained from CSIR, Pretoria, South Africa, whereas EP42 and A632 were supplied by CSIC, Pontevedra, Spain.
Quantitative real-time RT-PCR
Real-time RT-PCR was performed on a BioRad CFX96™ system using 25-μL mixtures containing 10 ng of synthesized cDNA, 1× iQ SYBR green supermix (BioRad, Hercules, CA, USA) and 0.2 μm forward and reverse primers (Supplemental Table S1). CYP97A and CYP97C primer information was obtained from Vallabhaneni et al., 2009. Relative expression levels were calculated on the basis of serial dilutions of cDNA (125–0.2 ng) which were used to generate standard curves for each gene. PCR was performed in triplicate using 96-well optical reaction plates. Cycling conditions consisted of a single incubation step at 95 °C for 5 min followed by 44 cycles of 95 °C for 10 s, 58 °C for 35 s and 72 °C for 15 s. Specificity was confirmed by product melt curve analysis over the temperature range 50–90 °C with fluorescence acquired after every 0.5 °C increase, and the fluorescence threshold value and gene expression data were calculated with BioRad CFX96™ software. Values represent the mean of three RT-PCR replicates ± SD. Amplification efficiencies were compared by plotting the ΔCt values of different primer combinations of serial dilutions against the log of starting template concentrations using the CFX96™ software.
Total RNA isolation and RT-PCR
Total RNA was isolated from corn endosperm (25 DAP) using the RNeasy® Plant Mini kit (QIAGEN, Hilden, Germany). RNA was treated on-column for DNA contamination using RNase-free DNase (QIAGEN, Hilden, Germany). RNA concentrations were determined with a spectrophotometer using a NanoDrop® ND-1000 (Thermo Scientific, Wilmington, DE, USA). For RT-PCR analysis, 1 μg RNA was used as a template for first-strand cDNA synthesis with Omniscript® Reverse Transcription kit (QIAGEN). Corn actin was used as control to monitor cDNA quality and DNA contamination. PCRs were carried out under standard conditions using 0.625 units of GoTaq® DNA Polymerase (Promega, Madison, WI). The PCR program used was 95 °C for 3 min, followed by 30 cycles of 94 °C for 45 s, 60 °C for 45 s, 72 °C for 90 s, and a final extension at 72 °C for 10 min. The primers are listed in Supplemental Table S2.
Total carotenoids were extracted from freeze-dried endosperm in 20 mL 50/50 (v/v) tetrahydrofuran and methanol at 60 °C for 15–20 min and quantified by measuring absorbance at 465 nm. For HPLC separation, the solvent was evaporated under a stream of nitrogen gas at 37 °C, re-dissolved in 50 μL acetone, and a 20-μL aliquot was injected immediately. Samples were separated on a Nucleosil C18 3-μ column with acetonitrile/methanol/2-propanol (85:15:5, v/v/v as mobile phase at 25 °C. In each case, zeaxanthin and lutein were separated in parallel runs on a C18 Vydac 218TP54 column, with methanol containing 2% water as the mobile phase (Romer et al., 2002). Samples were monitored with a Kontron DAD 440 photodiode array detector with online registration of the spectra. All carotenoids were identified by co-chromatography with authentic reference compounds, which were biosynthesized in E. coli as described in Sandmann G, 2002 and comparison of their spectra. Those standards were also used for quantitation in combination with the extinction coefficients from Davies (1976).
We thank Dr. Ana Butrón and Amando Ordás in Misión Biológica de Galicia, Consejo Superior de Investigaciones Cientificas, Apartado 28, 36080 Pontevedra, Spain, for supplying corn seeds. This research was supported by the Ministry of Science and Innovation, Spain (BFU2007-61413 and BIO2007-30738-E) and European Research Council Advanced Grant (BIOFORCE).