Vitamin deficiencies in humans: can plant science help?

The term vitamin describes a small group of organic compounds that are absolutely required in the human diet. Although for the most part, dependency criteria are met in developed countries through balanced diets, this is not the case for the five billion people in developing countries who depend predominantly on a single staple crop for survival. Thus, providing a more balanced vitamin intake from high-quality food remains one of the grandest challenges for global human nutrition in the coming decade(s). Here, we describe the known importance of vitamins in human health and current knowledge on their metabolism in plants. Deficits in developing countries are a combined consequence of a paucity of specific vitamins in major food staple crops, losses during crop processing, and/or overreliance on a single species as a primary food source. We discuss the role that plant science can play in addressing this problem and review successful engineering of vitamin pathways. We conclude that while considerable advances have been made in understanding vitamin metabolic pathways in plants, more cross-disciplinary approaches must be adopted to provide adequate levels of all vitamins in the major staple crops to eradicate vitamin deficiencies from the global population.

Gene network reconstruction identifies the authentic trans-prenyl diphosphate synthase that makes the solanesyl moiety of ubiquinone-9 in Arabidopsis.

Ubiquinone (coenzyme Q) is the generic name of a class of lipid-soluble electron carriers formed of a redox active benzoquinone ring attached to a prenyl side chain. The length of the latter varies among species, and depends upon the product specificity of a trans-long-chain prenyl diphosphate synthase that elongates an allylic diphosphate precursor. In Arabidopsis, this enzyme is assumed to correspond to an endoplasmic reticulum-located solanesyl diphosphate synthase, although direct genetic evidence was lacking. In this study, the reconstruction of the functional network of Arabidopsis genes linked to ubiquinone biosynthesis singled out an unsuspected solanesyl diphosphate synthase candidate--product of gene At2g34630--that, extraordinarily, had been shown previously to be targeted to plastids and to contribute to the biosynthesis of gibberellins. Green fluorescent protein (GFP) fusion experiments in tobacco and Arabidopsis, and complementation of a yeast coq1 knockout lacking mitochondrial hexaprenyl diphosphate synthase demonstrated that At2g34630 is also targeted to mitochondria. At2g34630 is the main--if not sole--contributor to solanesyl diphosphate synthase activity required for the biosynthesis of ubiquinone, as demonstrated by the dramatic (75-80%) reduction of the ubiquinone pool size in corresponding RNAi lines. Overexpression of At2g34630 gave up to a 40% increase in ubiquinone content compared to wild-type plants. None of the silenced or overexpressing lines, in contrast, displayed altered levels of plastoquinone. Phylogenetic analyses revealed that At2g34630 is the only Arabidopsis trans-long-chain prenyl diphosphate synthase that clusters with the Coq1 orthologs involved in the biosynthesis of ubiquinone in other eukaryotes.

Phylloquinone (vitamin K(1) ) biosynthesis in plants: two peroxisomal thioesterases of Lactobacillales origin hydrolyze 1,4-dihydroxy-2-naphthoyl-CoA.

It is not known how plants cleave the thioester bond of 1,4-dihydroxy-2-naphthoyl-CoA (DHNA-CoA), a necessary step to form the naphthoquinone ring of phylloquinone (vitamin K(1) ). In fact, only recently has the hydrolysis of DHNA-CoA been demonstrated to be enzyme driven in vivo, and the cognate thioesterase characterized in the cyanobacterium Synechocystis. With a few exceptions in certain prokaryotic (Sorangium and Opitutus) and eukaryotic (Cyanidium, Cyanidioschyzon and Paulinella) organisms, orthologs of DHNA-CoA thioesterase are missing outside of the cyanobacterial lineage. In this study, genomic approaches and functional complementation experiments identified two Arabidopsis genes encoding functional DHNA-CoA thioesterases. The deduced plant proteins display low percentages of identity with cyanobacterial DHNA-CoA thioesterases, and do not even share the same catalytic motif. GFP-fusion experiments demonstrated that the Arabidopsis proteins are targeted to peroxisomes, and subcellular fractionations of Arabidopsis leaves confirmed that DHNA-CoA thioesterase activity occurs in this organelle. In vitro assays with various aromatic and aliphatic acyl-CoA thioester substrates showed that the recombinant Arabidopsis enzymes preferentially hydrolyze DHNA-CoA. Cognate T-DNA knock-down lines display reduced DHNA-CoA thioesterase activity and phylloquinone content, establishing in vivo evidence that the Arabidopsis enzymes are involved in phylloquinone biosynthesis. Extraordinarily, structure-based phylogenies coupled to comparative genomics demonstrate that plant DHNA-CoA thioesterases originate from a horizontal gene transfer with a bacterial species of the Lactobacillales order.

A dedicated thioesterase of the Hotdog-fold family is required for the biosynthesis of the naphthoquinone ring of vitamin K1.

Phylloquinone (vitamin K(1)) is a bipartite molecule that consists of a naphthoquinone ring attached to a phytyl side chain. The coupling of these 2 moieties depends on the hydrolysis of the CoA thioester of 1,4-dihydroxy-2-naphthoate (DHNA), which forms the naphthalenoid backbone. It is not known whether such a hydrolysis is enzymatic or chemical. In this study, comparative genomic analyses identified orthologous genes of unknown function that in most species of cyanobacteria cluster with predicted phylloquinone biosynthetic genes. The encoded approximately 16-kDa proteins display homology with some Hotdog domain-containing CoA thioesterases that are involved in the catabolism of 4-hydroxybenzoyl-CoA and gentisyl-CoA (2,5-dihydroxybenzoyl-CoA) in certain soil-dwelling bacteria. The Synechocystis ortholog, encoded by gene slr0204, was expressed as a recombinant protein and was found to form DHNA as reaction product. Unlike its homologs in the Hotdog domain family, Slr0204 showed strict substrate specificity. The Synechocystis slr0204 knockout was devoid of DHNA-CoA thioesterease activity and accumulated DHNA-CoA. As a result, knockout cells contained 13-fold less phylloquinone than their wild-type counterparts and displayed the typical photosensitivity to high light associated to phylloquinone deficiency in cyanobacteria.

A bimodular oxidoreductase mediates the specific reduction of phylloquinone (vitamin K1) in chloroplasts.

Plants and certain species of cyanobacteria are the only organisms capable of synthesizing phylloquinone (vitamin K1 for vertebrates), which they use as an electron carrier during photosynthesis. Recent studies, however, have identified a plastidial pool of non-photoactive phylloquinone that could be involved in additional cellular functions. Here, we characterized an Arabidopsis bimodular enzyme--the At4g35760 gene product--comprising an integral domain homologous to the catalytic subunit of mammalian vitamin K1 epoxide reductase (VKORC1, EC 1.1.4.1) that is fused to a soluble thioredoxin-like moiety. GFP-fusion experiments in tobacco mesophyll cells established that the plant protein is targeted to plastids, and analyses of transcript and protein levels showed that expression is maximal in leaf tissues. The fused and individual VKORC1 domains were separately expressed in yeast, removing their chloroplast targeting pre-sequence and adding a C-terminal consensus signal for retention in the endoplasmic reticulum. The corresponding microsomal preparations were equally effective at mediating the dithiotreitol-dependent reduction of phylloquinone and menaquinone into their respective quinol forms. Strikingly, unlike mammalian VKORC1, the Arabidopsis enzyme did not reduce phylloquinone epoxide, and was resistant to inhibition by warfarin. The isoprenoid benzoquinone conjugates plastoquinone and ubiquinone were not substrates, establishing that the plant enzyme evolved strict specificity for the quinone form of naphthalenoid conjugates. In vitro reconstitution experiments established that the soluble thioredoxin-like domain can function as an electron donor for its integral VKORC1 partner.

The AAE14 gene encodes the Arabidopsis o-succinylbenzoyl-CoA ligase that is essential for phylloquinone synthesis and photosystem-I function.

Phylloquinone is the one-electron carrier at the A(1) site of photosystem I, and is essential for photosynthesis. Arabidopsis mutants deficient in early steps of phylloquinone synthesis do not become autotrophic and are seedling lethals, even when grown on sucrose-supplemented media. Here, we identify acyl-activating enzyme 14 (AAE14, At1g30520) as the o-succinylbenzoyl-coenzyme A (OSB-CoA) ligase acting in phylloquinone synthesis. Three aae14 mutant alleles, identified by reverse genetics, were found to be seedling lethal, to contain no detectable phylloquinone (< 0.1 pmol mg(-1) fresh weight) compared with 10 pmol mg(-1) fresh weight in wild-type leaves, and to accumulate OSB. AAE14 was able to restore menaquinone biosynthesis when expressed in an Escherichia coli mutant disrupted in the menE gene that encodes the bacterial OSB-CoA ligase. Weak expression of an AAE14 transgene in mutant plants (controlled by the uninduced XVE promoter) resulted in chlorotic, slow-growing plants that accumulated an average of 4.7 pmol mg(-1) fresh weight of phylloquinone. Inducing the XVE promoter in these plants, or expressing an AAE14 transgene under the control of the CaMV 35S promoter, led to full complementation of the mutant phenotype. aae14-mutant plants were also able to synthesize phylloquinone when provided with 1,4-dihydroxy-2-naphthoate, an intermediate in phylloquinone synthesis downstream of the OSB-CoA ligase reaction. Expression of an AAE14:GFP reporter construct indicated that the protein accumulated in discrete foci within the chloroplasts. This and other evidence suggests that the enzymes of phylloquinone synthesis from isochorismate may form a complex in the chloroplast stroma to facilitate the efficient channeling of intermediates through the pathway.

Detection and quantification of vitamin K(1) quinol in leaf tissues.

Phylloquinone (2-methyl-3-phytyl-1,4-naphthoquinone; vitamin K(1)) is vital to plants. It is responsible for the one-electron transfer at the A(1) site of photosystem I, a process that involves turnover between the quinone and semi-quinone forms of phylloquinone. Using HPLC coupled with fluorometric detection to analyze Arabidopsis leaf extracts, we detected a third redox form of phylloquinone corresponding to its fully reduced - quinol-naphthoquinone ring (PhQH(2)). A method was developed to quantify PhQH(2) and its corresponding oxidized quinone (PhQ) counterpart in a single HPLC run. PhQH(2) was found in leaves of all dicotyledonous and monocotyledonous species tested, but not in fruits or in tubers. Its level correlated with that of PhQ, and represented 5-10% of total leaf phylloquinone. Analysis of purified pea chloroplasts showed that these organelles accounted for the bulk of PhQH(2). The respective pool sizes of PhQH(2) and PhQ were remarkably stable throughout the development of Arabidopsis green leaves. On the other hand, in Arabidopsis and tomato senescing leaves, PhQH(2) was found to increase at the expense of PhQ, and represented 25-35% of the total pool of phylloquinone. Arabidopsis leaves exposed to light contained lower level of PhQH(2) than those kept in the dark. These data indicate that PhQH(2) does not originate from the photochemical reduction of PhQ, and point to a hitherto unsuspected function of phylloquinone in plants. The putative origin of PhQH(2) and its recycling into PhQ are discussed.

Characterization of the folate salvage enzyme p-aminobenzoylglutamate hydrolase in plants.

Folates break down in vivo to give pterin and p-aminobenzoylglutamate (pABAGlu) fragments, the latter usually having a polyglutamyl tail. Pilot studies have shown that plants can hydrolyze pABAGlu and its polyglutamates to p-aminobenzoate, a folate biosynthesis precursor. The enzymatic basis of this hydrolysis was further investigated. pABAGlu hydrolase activity was found in all species and organs tested; activity levels implied that the proteins responsible are very rare. The activity was located in cytosol/vacuole and mitochondrial fractions of pea (Pisum sativum L.) leaves, and column chromatography of the activity from Arabidopsis tissues indicated at least three peaks. A major activity peak from Arabidopsis roots was purified 86-fold by a three-column procedure; activity loss during purification exceeded 95%. Size exclusion chromatography gave a molecular mass of approximately 200 kDa. Partially purified preparations showed a pH optimum near 7.5, a Km value for pABAGlu of 370 microM, and activity against folic acid. Activity was relatively insensitive to thiol and serine reagents, but was strongly inhibited by 8-hydroxyquinoline-5-sulfonic acid and stimulated by Mn2+, pointing to a metalloenzyme. The Arabidopsis genome was searched for proteins similar to Pseudomonas carboxypeptidase G, which contains zinc and is the only enzyme yet confirmed to attack pABAGlu. The sole significant matches were auxin conjugate hydrolase family members and the At4g17830 protein. None was found to have significant pABAGlu hydrolase activity, suggesting that this activity resides in hitherto unrecognized enzymes. The finding that Arabidopsis has folate-hydrolyzing activity points to an enzymatic component of folate degradation in plants.

Higher plant plastids and cyanobacteria have folate carriers related to those of trypanosomatids.

Cyanobacterial and plant genomes encode proteins with some similarity to the folate and biopterin transporters of the trypanosomatid parasite Leishmania. The Synechocystis slr0642 gene product and its closest Arabidopsis homolog, the At2g32040 gene product, are representative examples. Both have 12 probable transmembrane domains, and the At2g32040 protein has a predicted chloroplast transit peptide. When expressed in Escherichia coli pabA pabB or folE, mutants, which are unable to produce or take up folates, the slr0642 protein and a modified At2g32040 protein (truncated and fused to the N terminus of slr0642) enabled growth on 5-formyltetrahydrofolate or folic acid but not on 5-formyltetrahydrofolate triglutamate, demonstrating that both proteins mediate folate monoglutamate transport. Both proteins also mediate transport of the antifolate analogs methotrexate and aminopterin, as evidenced by their ability to greatly increase the sensitivity of E. coli to these inhibitors. The full-length At2g32040 polypeptide was translocated into isolated pea chloroplasts and, when fused to green fluorescent protein, directed the passenger protein to the envelope of Arabidopsis chloroplasts in transient expression experiments. At2g32040 transcripts were present at similar levels in roots and aerial organs, indicating that the protein occurs in non-green plastids as well as chloroplasts. Insertional inactivation of At2g32040 significantly raised the total folate content of chloroplasts and lowered the proportion of 5-methyltetrahydrofolate but did not discernibly affect growth. These findings establish conservation of function among folate and biopterin transporter family proteins from three kingdoms of life.

Folate biofortification in tomatoes by engineering the pteridine branch of folate synthesis.

Plants are the main source of folate in human diets, but many fruits, tubers, and seeds are poor in this vitamin, and folate deficiency is a worldwide problem. Plants synthesize folate from pteridine, p-aminobenzoate (PABA), and glutamate moieties. Pteridine synthesis capacity is known to drop in ripening tomato fruit; therefore, we countered this decline by fruit-specific overexpression of GTP cyclohydrolase I, the first enzyme of pteridine synthesis. We used a synthetic gene based on mammalian GTP cyclohydrolase I, because this enzyme is predicted to escape feedback control in planta. This engineering maneuver raised fruit pteridine content by 3- to 140-fold and fruit folate content by an average of 2-fold among 12 independent transformants, relative to vector-alone controls. Most of the folate increase was contributed by 5-methyltetrahydrofolate polyglutamates and 5,10-methenyltetrahydrofolate polyglutamates, which were also major forms of folate in control fruit. The accumulated pteridines included neopterin, monapterin, and hydroxymethylpterin; their reduced forms, which are folate biosynthesis intermediates; and pteridine glycosides not previously found in plants. Engineered fruit with intermediate levels of pteridine overproduction attained the highest folate levels. PABA pools were severely depleted in engineered fruit that were high in folate, and supplying such fruit with PABA by means of the fruit stalk increased their folate content by up to 10-fold. These results demonstrate that engineering a moderate increase in pteridine production can significantly enhance the folate content in food plants and that boosting the PABA supply can produce further gains.

Folate synthesis in plants: the last step of the p-aminobenzoate branch is catalyzed by a plastidial aminodeoxychorismate lyase.

In plants, the last step in the synthesis of p-aminobenzoate (PABA) moiety of folate remains to be elucidated. In Escherichia coli, this step is catalyzed by the PabC protein, a beta-lyase that converts 4-amino-4-deoxychorismate (ADC)--the reaction product of the PabA and PabB enzymes--to PABA and pyruvate. So far, the only known plant enzyme involved in PABA synthesis is ADC synthase, which has fused domains homologous to E. coli PabA and PabB and is located in plastids. ADC synthase has no lyase activity, implying that plants have a separate ADC lyase. No such lyase is known in any eukaryote. Genomic and phylogenetic approaches identified Arabidopsis and tomato cDNAs encoding PabC homologs with putative chloroplast-targeting peptides. These cDNAs were shown to encode functional enzymes by complementation of an E. coli pabC mutant, and by demonstrating that the partially purified recombinant proteins convert ADC to PABA. Plant ADC lyase is active as dimer and is not feedback inhibited by physiologic concentrations of PABA, its glucose ester, or folates. The full-length Arabidopsis ADC lyase polypeptide was translocated into isolated pea chloroplasts and, when fused to green fluorescent protein, directed the passenger protein to Arabidopsis chloroplasts in transient expression experiments. These data indicate that ADC lyase, like ADC synthase, is present in plastids. As shown previously for the ADC synthase transcript, the level of ADC lyase mRNA in the pericarp of tomato fruit falls sharply as ripening advances, suggesting that the expression of these two enzymes is coregulated.

Folate synthesis in plants: the p-aminobenzoate branch is initiated by a bifunctional PabA-PabB protein that is targeted to plastids.

It is not known how plants synthesize the p-aminobenzoate (PABA) moiety of folates. In Escherichia coli, PABA is made from chorismate in two steps. First, the PabA and PabB proteins interact to catalyze transfer of the amide nitrogen of glutamine to chorismate, forming 4-amino-4-deoxychorismate (ADC). The PabC protein then mediates elimination of pyruvate and aromatization to give PABA. Fungi, actinomycetes, and Plasmodium spp. also synthesize PABA but have proteins comprising fused domains homologous to PabA and PabB. These bipartite proteins are commonly called "PABA synthases," although it is unclear whether they produce PABA or ADC. Genomic approaches identified Arabidopsis and tomato cDNAs encoding bipartite proteins containing fused PabA and PabB domains, plus a putative chloroplast targeting peptide. These cDNAs encode functional enzymes, as demonstrated by complementation of an E. coli pabA pabB double mutant and a yeast PABA-synthase deletant. The partially purified recombinant Arabidopsis protein did not produce PABA unless the E. coli PabC enzyme was added, indicating that it forms ADC, not PABA. The enzyme behaved as a monomer in size-exclusion chromatography and was not inhibited by physiological concentrations of PABA, its glucose ester, or folates. When the putative targeting peptide was fused to GFP and expressed in protoplasts, the fusion protein appeared only in chloroplasts, indicating that PABA synthesis is plastidial. In the pericarp of tomato fruit, the PabA-PabB mRNA level fell drastically as ripening advanced, but there was no fall in total PABA content, which stayed between 0.7 and 2.3 nmol.g(-1) fresh weight.