Enhancement of folates in plants through metabolic engineering

Abstract

Humans depend on plants as a major source of dietary folates. Inadequate dietary levels of the vitamin folate can lead to megaloblastic anemia, birth defects, impaired cognitive development, and increased risk of cardiovascular disease and cancer. The biofortification of folate levels in food crops is a target for metabolic engineering. Folates are synthesized de novo from pterins and para-amino benzoic acid, which are subsequently combined to form dihydropteroate, the direct precursor to dihydrofolate. We postulated that GTP cyclohydrolase-1, which catalyzes the first committed step in pterin biosynthesis, was a rate-limiting step in pterin synthesis in plants and, therefore, in folate synthesis. On this basis, we proposed that the expression of an unregulated bacterial GTP cyclohydrolase-1 in plants would increase pterin biosynthesis with a concomitant enhancement of folate levels. The folE gene encoding GTP cyclohydrolase-1 was cloned from Escherichia coli and introduced into Arabidopsis thaliana through plant transformation. The expression of bacterial GTP cyclohydrolase-1 in transgenic Arabidopsis resulted in a 1,250-fold and 2- to 4-fold enhancement of pterins and folates, respectively. These results helped to identify other potential factors regulating folate synthesis, suggesting ways to further enhance folate levels in food crops.

Inadequate dietary levels of the vitamin folate can lead to megaloblastic anemia, birth defects, impaired cognitive development, and increased risk of cardiovascular disease and cancer (1-10). The March of Dimes has stated that inadequate intake of folate before pregnancy is the most common cause of birth defects, including neural tube defects (NTDs) such as spina bifida and anencephaly. NTDs affect one of every 1,000 live births in the United States (11), whereas the incidence of NTDs in other countries may be 10-20 times greater (12-14).

Plants are the primary source of folate in human nutrition. Green leafy vegetables, legumes, and certain fruits are the richest sources of dietary folates, whereas folate levels are extremely low in cereals and in root and tuber crops, which are the primary sources of calories in the diets of people in developing nations. Our long-term goal is to develop biofortified foods for the sustainable delivery of adequate levels of folates to reduce the negative health outcomes of folate deficiency.

Folates are synthesized de novo from pterins and para-aminobenzoic acid (PABA) by means of a multistep pathway (15-17), whereas pterins and PABA are synthesized from GTP and chorismate, respectively (Fig. 1). Factors regulating folate biosynthesis, turnover, and accumulation in plants are poorly understood. Based on current information on biosynthetic pathways, we proposed that the reaction catalyzed by plant GTP cyclohydrolase-1 (GCH) is a rate-determining step in de novo pterin and folate biosynthesis in plants. Therefore, we proposed that expression of bacterial GCH, which reportedly is not subject to metabolic regulation, would increase the metabolic flux through the pterin and folate biosynthetic pathways and thereby result in an increase in pterin pools and folate levels. To test this hypothesis, we cloned the GCH (folE) gene from Escherichia coli (EcGCH) and expressed the gene in transgenic Arabidopsis thaliana.

Fig. 1.

The proposed pathway and localization of key enzymes for synthesis of pterins and folates in plants. PABA-glc, glucose ester of PABA.

Materials and Methods

Materials. E. coli K12 MG1655 was a gift from the Laboratory of Genetics, University of Wisconsin, Madison (E. coli Genome Project, Laboratory of Frederick Blattner). Restriction enzymes used were from New England Biolabs. DNA markers were from Gene Choice (PGC Scientific, Gaithersburg, MD), and the BenchMark protein ladder was from Bio-Rad. Neopterin, GTP, isoxanthopterin, and xanthopterin were obtained from Sigma. Finale [glufosinate-ammonium: butanoic acid, 2-amino-4-(hydroxymethylphosphinyl)monoammonium salt] was purchased from a local nursery.

PCR Amplification of the folE Gene. The bacterial gene encoding GCH (folE, GenBank accession no. AE000304) was amplified by PCR from E. coli K12 MG1655. E. coli DNA was purified according to the DNeasy tissue kit (Qiagen, Valencia, CA) and used as template for PCR amplification. The sense primer 5′-CCCATCACTCAGTAAAGAAGCGGC-3′ and antisense primer 5′-CCGTTGTGATGACGCACAGCG-3′ were synthesized based on the nucleotide sequence of the folE gene (18). Primers were designed to insert the DNA fragment in the correct reading frame in a pET-Blue 2 Blunt vector purchased from Novagen. PCR was performed according to Sambrook et al. (19), and the product was gel-purified by using the Qiagen QIAquick gel extraction kit. To confirm the identity of the amplified folE gene, DNA sequencing was performed at the Protein and Nucleic Acid Core Laboratory at the Washington University School of Medicine (St. Louis).

Construction of the Bacterial Expression Vector. The PCR insert was converted to a blunt phosphorylated form and ligated with blunt dephosphorylated pET-Blue 2 vector from Novagen. The resulting plasmid, pTK101, was transformed into Nova Blue Singles bacterial cells. Recombinant plasmids were isolated and transformed into pET-Blue-compatible expression host strain Tuner DE3 pLac1.

Bacterial Expression and Purification of His/Herpes Simplex Virus-Tagged Protein. Protein expression was induced with isopropyl β-D-thiogalactopyranoside. The expressed protein was purified by nickel-column chromatography using Novagen nickel-nitrilotriacetic acid-His-binding resin.

Protein Detection and Identification. Protein samples were analyzed by SDS/PAGE (12.5%) to determine purity and estimate molecular mass of the protein (20). Protein was transferred electrophoretically to a Protran pure nitrocellulose membrane (Schleicher & Schuell). The His-tagged EcGCH fusion protein was detected by using mouse anti-His-tag monoclonal antibody as primary antibody and alkaline phosphatase-conjugated goat anti-mouse IgG as secondary antibody. Gel-purified fusion protein was used as an antigen for production of polyclonal antibodies in rabbits by Bethyl Laboratories (Montgomery, TX) using standard protocols. Alkaline phosphatase-conjugated goat anti-rabbit IgG was used to detect rabbit anti-EcGCH.

Vector Construction and Transformation of A. thaliana. The plant transformation vector PC-Gus-Bar was provided by J. Koo (Danforth Plant Science Center). DNA of PC-Gus-Bar and pTK101 were digested with NcoI and EcoR1, and DNA fragments were separated on a 1% agarose gel and purified from the gel. Fragments containing PC-Gus-BAR (without the coding sequence for GUS) and the folE DNA insert (including short N-terminal and C-terminal extensions) were ligated by using the Roche Applied Science rapid ligation kit to obtain the plant transformation vector pTK202. Products of ligation were introduced into DH5α cells and selected on LB media containing kanamycin. Colonies containing the folE coding sequence in the desired orientation relative to the cauliflower mosaic virus 35S promoter were identified by restriction with EcoR1 and NcoI and electrophoresis in a 1% agarose gel. Competent Agrobacterium tumefaciens GV3010 cells were transformed with pTK202 (21). Transformation of Arabidopsis with A. tumefaciens GV3010 carrying the plasmid pTK202 was performed by using a standard Arabidopsis transformation protocol (22, 23). Plants were grown in a Conviron growth chamber (22°C, 50% relative humidity, 200 μmol of light, 10-h photoperiod).

Selection of Transformed Plants. To select for transformed plants, plants at the four-leaf stage were sprayed with Finale [BASTA, 1:400 (vol/vol)]. BASTA selection was repeated every fifth day for a total of five applications. Leaf discs were collected from the BASTA-selected plants for DNA extraction by using the REDExtract-N-Amp plant PCR kit (Sigma). Extracted DNA was used as template for PCR detection by using the folE primers listed above, and plants containing the folE gene were transplanted and maintained. Seeds from the primary (T₁) generation were planted, and resultant T₂ plants were subjected to another round of BASTA selection and characterization by means of PCR and GCH assays. The process was repeated to obtain nonsegregating T₃ transgenic lines.

Extraction of Leaf Tissue for GCH and Pterin Assays. Leaf tissue from transgenic lines and wild-type Columbia was harvested before the onset of flowering, frozen in liquid nitrogen, and stored at -80°C. Tissue was extracted in 10 mM Tris·HCl buffer (pH 8.0) with 0.2 g of insoluble polyvinyl-polypyrrolidone (Sigma catalog no. P6755) per g of tissue. Extract was filtered through Miracloth and centrifuged at 10,000 × g for 15 min. The supernatant (crude extract) was used for analysis of pterins, protein, and GCH activity. Crude extract was dialyzed extensively before the GCH assay.

GCH Assay. The activity of purified EcGCH was determined by measuring the production of neopterin from GTP (24, 25). The enzymatic product, dihydroneopterin triphosphate, was first oxidized and then dephosphorylated to form neopterin. The latter was analyzed by reverse-phase HPLC on a Beckman Coulter C18 RP column. Neopterin was eluted isocratically with 0.5% acetonitrile and 0.1% tetrahydrofuran in water at a flow rate of 1 ml/min. Neopterin and other pterins were detected by fluorescence (365 nm excitation; 446 nm emission) with a Jasco (Easton, MD) FP 1520 fluorescence detector and by absorbance with a Beckman Coulter 168 diode array detector. Protein was measured by using the Bradford assay (26).

Analysis of Pterins. Unconjugated pterins in leaves were analyzed by reverse-phase chromatography by using the same conditions as used to detect neopterin in the GCH assay. Pterins were detected by fluorescence and absorbance at 280 and 330 nm. Diode array and fluorescence data were compared with authentic standards and literature values (27-29). Pterin concentrations were estimated by integration of the major fluorescent pterin peaks using neopterin as a standard. The elution of pterins was compared with the elution of selected pterin standards, and the identity of these pterins was analyzed further by MS analysis at the Protein and Nucleic Acid Core Laboratory at the Washington University School of Medicine. Acidified samples in 0.5% acetonitrile and 1% tetrahydrofuran were collected after HPLC separation, introduced into a Finnegan LC-DECA ion-trap MS, and analyzed in positive-ion mode.

Folate Analysis. Total folate was analyzed in the Vitamin Metabolism Laboratory (Jean Mayer Human Nutrition Research Center on Aging, Tufts University). Leaf tissue from transgenic A. thaliana (T₂ and T₃ lines) was extracted in extraction buffer (20 g of Bis-Tris, 20 g of Na-ascorbate, and 500 μl of mercaptoethanol per liter of water), and extracts were treated with conjugase enzymes to free folate for bacterial growth. The microbial assay with Lactobacillus casei was used to measure total folate content (30-32). Triplicate samples from each T₃ line were analyzed in replicated analyses (n = 8). Results were expressed as nmol of folate per g fresh weight (gfw) of tissue. Controls for folate recovery and effectiveness of conjugase treatment were included.

Results

The EcGCH (folE) gene was PCR-amplified, cloned, and expressed in E. coli. The expressed His-tagged fusion protein was purified by nickel-affinity chromatography, analyzed by SDS/PAGE (Fig. 2A), and detected on Western blots using anti-His antibody (Fig. 2B). Antiserum raised against affinity and gel-purified fusion protein detected the same protein band (Fig. 2C), whereas preimmune serum did not react with the purified fusion protein. GCH activity of the purified fusion protein was 20-fold higher than the activity of the fusion protein isolated from uninduced transformed cells (data not shown). The identity of the expressed protein was confirmed by peptide sequence analysis of the affinity-purified protein after trypsin digestion and de novo sequencing on an Applied Biosystems quadrupole/time-of-flight mass spectrometer equipped with a Hewlett-Packard capillary liquid chromatography system (data not shown). These results confirm the identity of the expressed protein and indicate that the EcGCH fusion protein was enzymatically active.

Fig. 2.

Expression and characterization of EcGCH. (A) Analysis of purified EcGCH fusion protein by SDS/PAGE. The molecular mass of EcGCH determined by SDS/PAGE (33) and EcGCH fusion protein with N- and C-terminal extensions were 25,500 and 31,461 Da, respectively. The major protein band detected (^*) had an estimated molecular mass of 31,000 Da. Lane 1, supernatant from induced cells transformed with plasmid pTK101; lane 2, nickel-column-purified protein fraction; lane 3, supernatant from cells with empty vector. (B and C) Western blot of EcGCH expressed in E. coli detected by using either mouse anti-His-tag monoclonal antibody (B) or anti-EcGCH (C). Lane 1, supernatant from induced cells transformed with plasmid pTK101; lane 2, affinity-purified EcGCH fusion protein; lane 3, supernatant from cells with empty vector. Protein (2.5 μg) was loaded onto each lane. (D) Western blot of transgenic and nontransgenic plants. Lane 1, extract from wild-type Arabidopsis; lanes 2-4, extracts from T₃ lines 5-21, 6-12, and 2-19, respectively. Protein (15 μg) was loaded in each lane. (E) Extracts from 6-10, 6-13, and nontransgenic control were concentrated ≈16-fold. Lanes 1 and 2, ≈210 μgof protein 6-10 and 6-12, respectively; lane 3, ≈15 μg of 5-21; lane 4, ≈210 μgof nontransgenic control; lane 5, BenchMark prestained molecular mass standards. EcGCH (26.2 kDa) is marked with ^*.

The folE gene sequence was inserted into a modified pCAMBIA plant transformation vector, and the resulting vector was used to transform A. thaliana cv. Columbia. The primary (T₁) transformants were selected for BASTA resistance and screened for the presence of the folE gene sequence by PCR. Extracts from BASTA-resistant, PCR-positive transgenic plants and nontransgenic Columbia were tested for GCH activity. Only lines with GCH activity were retained. Activity was not detected in wild-type Arabidopsis, whereas GCH activity was detected in extracts of transgenic T₁ lines. Transgenic plants from primary T₁ lines 2, 5, and 6 expressing EcGCH (n = 26) were subjected to a second round of BASTA selection and PCR analysis. Three individual T₂ plants were selected from each of three primary transgenic lines (2-14, 2-15, 2-19, 5-14, 5-15, 5-21, 6-10, 6-12, and 6-13) for additional analysis. Plants derived from these nine lines were subjected to another round of BASTA selection and testing to obtain nonsegregating, PCR-positive T₃ transgenic lines. Leaf tissue from each of the nine transgenic lines was collected for additional analysis.

To confirm that the bacterial gene was expressed, leaf tissue from the T₃ plants was extracted and analyzed by SDS/PAGE and immunoblotting. An immunoreactive band corresponding in size to the predicted molecular mass (26.2 kDa) of the folE gene product EcGCH was detected in these extracts but not in wild-type control (Fig. 2D). Although the quantity of EcGCH protein detected in extracts from plants derived from any one of the three primary transgenic lines was quite similar, the differences between lines 2, 5, and 6 were more significant. Plants derived from primary lines 2 and 5 had substantially more immunodetectable protein than plants derived from line 6 (Fig. 2D). When ≈15 μg of total leaf protein was loaded onto a minigel, lines 2 and 5 exhibited a distinct immunoreactive band of the correct molecular mass (note that several smaller bands were detected when the blot was overexposed to detect EcGCH in line 6, suggesting that proteolytic cleavage may have occurred), whereas there was very little, if any, band visible in extracts from line 6 under the same conditions. However, when extract from derivatives of primary line 6 was concentrated and ≈210 μg of protein was loaded onto a larger gel, an immunodetectable band was observed in line 6 (Fig. 2E) with additional smaller bands (data not shown). The results of this analysis suggest that the levels of EcGCH in line 6 are 10- to 20-fold lower than the levels in lines 2 and 5.

To determine whether EcGCH expressed in Arabidopsis was active, leaf extracts from T₃ plants of each of the three primary lines were tested for GCH activity by using the standard fluorometric assay. GCH activity was detected in transgenic plants but not in nontransgenic controls under equivalent conditions (data not shown). Crude extracts from different T₃ plants derived from the same primary line exhibited similar GCH activity, whereas there were substantial differences in GCH activity between plants derived from different primary lines. Plants from primary lines 2 and 5 exhibited measurable activity, whereas activity was not detectable above background in extracts of plants derived from line 6 under standard assay conditions. These differences in GCH activity were positively correlated with the differences in the amount of EcGCH detected by means of immunoblot reactions.

To test the hypothesis that expression of EcGCH increases pterin synthesis, undialyzed crude leaf extracts of transgenic lines were subjected to HPLC analysis by using the same chromatographic conditions as for the analysis of neopterin in the GCH assay. This analysis confirmed the accumulation of high amounts of fluorescent pterin-like metabolites in leaf extracts of T₂ (data not shown) and T₃ plants (Fig. 3A) compared with nontransgenic plants. The fluorescence properties and UV/visible spectra and preliminary analysis by MS indicate that the fluorescent compounds detected in the crude extracts were unconjugated pterins, referred to hereafter as pterins. The compound eluting at 4.5 min (peak 1) was tentatively identified as neopterin based on comparison of elution time (Fig. 3A), UV/visible spectrum (Fig. 3B), and mass spectrum with neopterin standard (Fig. 3C). Two other fluorescent metabolites present in extracts of transgenic plants exhibited similar retention times, spectral properties, and fluorescence characteristics (27-29) to xanthopterin and isoxanthopterin (Fig. 3A Inset), products of pterin degradation.

Fig. 3.

Analysis and characterization of pterins from transgenic and nontransgenic Arabidopsis. (A) Pterins from undialyzed extract from T₃ line 2-19 were separated by reverse-phase chromatography and detected by fluorescence. Pterins were identified by comparison with elution times (A, Inset) and spectral characteristics of peak 1 (B), and neopterin (neo) (B, Inset), xanthopterin (xan), and isoxanthopterin (isoxan) (data not shown) standards were measured by diode array. Individual HPLC peaks were collected, and identification of neopterin was confirmed by MS analysis of peak 1 (C) and neopterin standard (Inset) ([M + H]⁺; m/z = 254).

Examination of fluorescence profiles indicated that the most abundant pterins can be detected in crude extracts of nontransgenic plants, but the levels are very low. The concentration of pterins increased for each line from the T₂ to the T₃ generation with levels in T₃ plants from primary lines 2 and 5 being 10- to 20-fold higher than the corresponding levels in derivatives of line 6, as summarized graphically in Fig. 4A. The levels of neopterin observed in transgenic plants were up to 1,100-fold higher than the corresponding levels in nontransgenic plants, and the quantities of pterins present in transgenic lines were highly correlated with the amount of EcGCH detected (see Fig. 2 D and E), supporting the conclusion that these products accumulated in response to the expression of the transgene. The concentration of pterins in nontransgenic Arabidopsis were comparable with those reported for other plant species (Fig. 4B), whereas the levels in transgenic plants were 750- to 1,250-fold higher than controls (Fig. 4 A and B).

Fig. 4.

(A) Levels of pterins in T₂ and T₃ transgenic lines and nontransgenic Arabidopsis. (B) Values for transgenic T₃ line 5-21 (value × 0.02) and nontransgenic plants (solid bars) were compared with values from the literature (29) for other plant sources.

To determine whether the expression of EcGCH increased synthesis and accumulation of folates, the total folate levels of transgenic and nontransgenic plants were measured by microbial bioassay and statistically analyzed by using a one-way ANOVA followed by pairwise comparison of means by using both the Student's t test and the Tukey-Kramer method (Fig. 5A). The mean value for total folate for nontransgenic Arabidopsis was 1.41 nmol of folate/gfw of leaf tissue, which is within the range of values reported for leaves and seeds from other plant species (Fig. 5B) (34). The folate values for primary transgenic lines were up to 3.3-fold higher than nontransgenic Arabidopsis, with values for replicate samples ranging from 2.27 to 4.70 nmol/gfw, with averages of 3.40, 3.40, and 2.55 nmol/gfw for plants derived from line 2, 5, and 6, respectively. Based on the statistical analysis, the folate contents of all transgenic lines were significantly higher (P = 0.05) than control. The folate content of line 6, which had 10- to 20-fold lower levels of EcGCH and total pterins, was ≈50% lower than the folate values for lines 2 and 5. The average value for the ratio of pterin to folate levels for nontransgenic Arabidopsis (this study) and other plant species (values reported in the literature) was 1.40, whereas the ratio for transgenic T₃ lines ranged between 55 and 644 with an average of 350. The observed increase in total folate levels as a function of pterin levels (Fig. 5C) appeared to saturate. The maximum value observed for T3 line 2-19 (4.7 nmol) was ≈1.9-fold higher than those reported for spinach, a plant considered to be rich in folates (Fig. 5B). These results demonstrate the enhancement of folate levels in plants through bioengineering.

Fig. 5.

(A) Total folate levels of T₃ transgenic lines and nontransgenic (NT) Arabidopsis [total folate (nmol per gfw tissue ± the standard deviation) for three replicates]. Folate values for all transgenic plant lines were significantly different from the nontransgenic plants at the 95% confidence level. (B) Comparison of folate levels for transgenic and nontransgenic Arabidopsis plants (solid bars) and values reported in the literature [ref. 48 and U.S. Department of Agriculture Agricultural Research Service Nutrient Data Laboratory (www.nal.usda.gov)] for other plant sources. (C) Plot of folate values as a function of total pterins in transgenic and nontransgenic lines.

Discussion

Although plants are a primary source of dietary folates, the folate contents of many staple crops including rice, maize, wheat, potato, and cassava are quite low. Metabolic engineering provides an opportunity to enhance the folate content of these and other crops. The reaction catalyzed by GCH is a potentially rate-limiting step in folate biosynthesis in plants. In this study, we introduced the gene for an unregulated GCH from E. coli (EcGCH) into Arabidopsis to overcome this limitation, resulting in an increase in pterin synthesis and a corresponding enhancement of folate levels in this plant.

These results support our hypothesis that GCH is a rate-limiting step in pterin biosynthesis in plants. The apparent saturation of total folate levels as a function of pterin levels (see Fig. 5C) suggests that additional factors are limiting the conversion of pterins into folates. The nature and relationship of these pterins to folate biosynthesis is not clear and will require additional investigation. There are several possible factors to explain the surprising accumulation of pterins and the correspondingly lower-than-anticipated increase in total folates in these transgenic plants. One explanation is that these plants may contain suboptimal amounts of available PABA and thereby limit the conversion of pterins into folates, resulting in the accumulation of pterins. Consistent with this interpretation, Quinlivan et al. (46) reported that the levels of free PABA in Arabidopsis are low, whereas levels of total PABA (free and glucose ester of PABA) were approximately the same as the levels of folates (46). In preliminary studies, we observed a decrease in the accumulation of pterins in transgenic plants to which exogenous PABA was added, providing support for the suggestion that PABA levels may be an important factor limiting folate synthesis.

Other factors also may contribute to the accumulation of pterins and the reduced efficiency of conversion of pterins into folates. Competing fates for pterin intermediates and/or differences in reaction rates for pathway enzymes may account for the accumulation or diversion of pterins into alternative pathways, e.g., biopterin or molybdopterin synthesis, both of which share intermediates with the folate pathway (47, 48). Two of the pterins that accumulated in transgenic leaf tissue exhibited similar retention times to xanthopterin and isoxanthopterin (Fig. 4A), products of (tetrahydrobio)pterin degradation/catabolism. Although biopterin has not been reported in plants, various pterins, e.g., carboxypterin, neopterin, xanthopterin, and isoxanthopterin, have been detected in a variety of plant tissues (28, 29). The role of these pterins in plants, if any, is not known.

Metabolic channeling either through formation of multienzyme complexes and/or compartmentalization may play important roles in the synthesis and accumulation of folates from pterins and PABA. Although little is known about the nature and role of metabolic channeling in folate biosynthesis in plants, there is evidence of compartmentalization and the existence of multifunctional enzymes in the tetrahydrofolate biosynthetic pathway in plants and other species (15-17). In microorganisms, GCH is part of a multienzyme complex with 2-amino-4-hydroxy-6-hydroxymethyldihydropteridinephosphokinase (49). In plants, GCH is presumably localized in the ground cytoplasm based on the prediction that the protein lacks a targeting signal, whereas synthesis of PABA from chorismate takes place in chloroplasts, and the last five steps of the folate pathway are localized in mitochondria (15-17). The intermediates from both the pteridine and PABA branches of the pathway must be transported into the mitochondria for folate synthesis to occur. To date, there are no reports of organelle-specific transporters for pterins or PABA in plants. Furthermore, the possibility that GCH is localized to mitochondria cannot be ruled out. In fact, the last two enzymes of the pteridine branch of the pathway are located in the mitochondria. Because intermediates in the pterin and folate biosynthetic pathways are unstable and subject to oxidation, synthesis in the reducing environment within the mitochondria could minimize the loss of intermediates. The localization of EcGCH in the cell may interfere with normal metabolic channeling. As a consequence, pathway intermediates may be oxidized or diverted into competing fates and/or accumulate as metabolically inactive products.

Tetrahydrofolate biosynthesis, like all biosynthetic pathways, must be regulated tightly to maintain sufficient levels of essential products. The existence of mechanisms to maintain homeostasis in the metabolically active pools of essential cofactors such as folates are not known. Increases in the pools of folates above certain levels could result in down-regulation of earlier steps in the pathway causing a backlog of intermediates. The levels of provitamin A, vitamin C, vitamin E, and phosphocholine in plants have been enhanced 2- to 15-fold through pathway engineering by insertion of one or two genes (Table 1). The enhancement of folate levels reported herein are comparable with those reported for increases in vitamins observed in genetically modified plants, whereas the observed increase in pterins seems to far exceed any increase reported to date in response to the insertion of a single transgene (Table 1).

Table 1. Enhancement of vitamins through metabolic engineering.

Vitamin	Enzyme/gene	Enhancement	Ref.
Folate	EcGCH	Up to 3.3-fold increase in total folate; 1,250-fold increase total pterins	This study
C	d-Galaturonate reductase	2- to 3-fold increase in vitamin C	35
	Dehydroascorbate reductase	2- to 4-fold increase in vitamin C	36
	l-Gulono-γ-lactone oxidase	Up to 7-fold increase in vitamin C	37
A	Phytoene synthase	Synthesis β-carotene in rice endosperm; slight increase in total carotenoids	38
	Phytoene desaturase
	Phytoene synthase	50-fold increase total carotenoids	39
	Phytoene desaturase	0.5-fold decrease in carotenoids; ≈3-fold increase in β-carotene	40
E	γ-Tocopherol methytransferase	Shift from γ-tocopherol to α-tocopherol; no change in vitamin E	41
	Chorismate mutase/prephenate dehydrogenase	3-fold increase in vitamin E	42
	Tyrosine aminotransferase	4-fold increase in vitamin E	43
	Homogentistic acid geranylgeranyl transferase	10-to 15-fold increase in Arabidopsis; up to 6-fold increase in maize	44
Choline	Phosphoethanolamine N-methyltransferase	5-fold increase in phosphocholine; 50-fold increase in free choline	45

In summary, the observed increase in pterin synthesis and total folate levels in transgenic plants expressing EcGCH are encouraging in terms of the potential to enhance the amount of folates in food crops. Improved efficiency of conversion of dihydroneopterin produced as a result of expression of EcGCH could lead to additional increases in folates. Thus, a better understanding of regulation of folate synthesis and the factors contributing to the accumulation of pterins will be essential to efforts for folate biofortification.