Biofortification of crops with seven mineral elements often lacking in human diets – iron, zinc, copper, calcium, magnesium, selenium and iodine

Authors


Author for correspondence:
Philip J. White
Tel:+44 (0)1382 560043
Fax:+44 (0)1382 562426
Email: philip.white@scri.ac.uk

Summary

The diets of over two-thirds of the world's population lack one or more essential mineral elements. This can be remedied through dietary diversification, mineral supplementation, food fortification, or increasing the concentrations and/or bioavailability of mineral elements in produce (biofortification). This article reviews aspects of soil science, plant physiology and genetics underpinning crop biofortification strategies, as well as agronomic and genetic approaches currently taken to biofortify food crops with the mineral elements most commonly lacking in human diets: iron (Fe), zinc (Zn), copper (Cu), calcium (Ca), magnesium (Mg), iodine (I) and selenium (Se). Two complementary approaches have been successfully adopted to increase the concentrations of bioavailable mineral elements in food crops. First, agronomic approaches optimizing the application of mineral fertilizers and/or improving the solubilization and mobilization of mineral elements in the soil have been implemented. Secondly, crops have been developed with: increased abilities to acquire mineral elements and accumulate them in edible tissues; increased concentrations of ‘promoter’ substances, such as ascorbate, β-carotene and cysteine-rich polypeptides which stimulate the absorption of essential mineral elements by the gut; and reduced concentrations of ‘antinutrients’, such as oxalate, polyphenolics or phytate, which interfere with their absorption. These approaches are addressing mineral malnutrition in humans globally.

Abbreviations: 
ABC

 

 

ATP-binding cassette

CAX

Ca2+/H+ antiporter

CCC

cation chloride co-transporter

CDF

cation diffusion facilitator

CLC

chloride channel

CNGC

cyclic nucleotide gated channel

CTR

copper transporter

FRO

ferric reductase oxidase

GLR

glutamate receptor

HAST

high-affinity sulphate transporter

HMA

heavy metal P1B-ATPase

lpa

low phytic acid

MATE

multidrug and toxin efflux

MRS2

mitochondrial RNA splicing 2

MSL

MscS-like

MTP

metal tolerance protein

NA

nicotianamine

NAS

nicotianamine synthase

NRAMP

natural resistance-associated macrophage protein

OPHS

O-phosphohomoserine

OPT

oligopeptide transporter

PIC1

permease in chloroplasts 1

QTL

quantitative trait locus

SAM

S-adenosyl methionine

SeCys

selenocysteine

SeMet

selenomethionine

VIT1

vacuolar iron transporter 1

VSP

vegetative storage protein

YSL

yellow stripe like

ZIP

ZRT-, IRT-like protein; ZRT, zinc-regulated transporter

Mineral elements required by humans

Humans require at least 22 mineral elements for their wellbeing (Welch & Graham, 2004; White & Broadley, 2005a; Graham et al., 2007). These can be supplied by an appropriate diet. However, it is estimated that over 60% of the world's 6 billion people are iron (Fe) deficient, over 30% are zinc (Zn) deficient, 30% are iodine (I) deficient and c. 15% are selenium (Se) deficient (see Supporting Information References S1). In addition, calcium (Ca), magnesium (Mg) and copper (Cu) deficiencies are common in many developed and developing countries (Frossard et al., 2000; Welch & Graham, 2002, 2005; Rude & Gruber, 2004; Grusak & Cakmak, 2005; Thacher et al., 2006). This situation is attributed to crop production in areas with low mineral phytoavailability and/or consumption of (staple) crops with inherently low tissue mineral concentrations, compounded by a lack of fish or animal products in the diet (Welch & Graham, 2002, 2005; Poletti et al., 2004; White & Broadley, 2005a; Gibson, 2006; Graham et al., 2007). Currently, mineral malnutrition is considered to be among the most serious global challenges to humankind and is avoidable (Copenhagen Consensus 2004; http://www.copenhagenconsensus.com). Mineral malnutrition can be addressed through dietary diversification, mineral supplementation, food fortification and/or increasing mineral concentrations in edible crops (biofortification). However, strategies to increase dietary diversification, mineral supplementation and food fortification have not always been successful (see References S2). For this reason, biofortification of crops through the application of mineral fertilizers, combined with breeding varieties with an increased ability to acquire mineral elements, is advocated as an immediate strategy not only to increase mineral concentrations in edible crops but also to improve yields on infertile soils (Graham et al., 2001, 2007; Bouis et al., 2003; Genc et al., 2005; White & Broadley, 2005a; Pfeiffer & McClafferty, 2007). In addition, as mineral elements in edible portions of biofortified crops must be bioavailable to humans, parallel attempts are advocated (1) to increase the concentrations of ‘promoter’ substances, such as ascorbate (vitamin C), β-carotene, cysteine-rich polypeptides and certain organic and amino acids, which stimulate the absorption of essential mineral elements by the gut, and (2) to reduce the concentrations of ‘antinutrients’, such as oxalate, polyphenolics (tannins) or phytate (IP6), which interfere with their absorption (References S3). This review will focus on strategies to increase the concentrations and bioavailability of the seven mineral elements most often lacking in human diets: Fe, Zn, Cu, Ca, Mg, Se and I.

Phytoavailability of mineral elements

Plants can only acquire mineral elements supplied to them in specific chemical forms. For a biofortification strategy to prove successful, it is necessary to be aware of the forms of mineral elements acquired by plant roots, and the limitations to the supply and phytoavailability of mineral elements in the rhizosphere solution. The supply and phytoavailability of mineral elements in the rhizosphere solution ultimately limit the accumulation of mineral elements by crops, unless foliar fertilizers are applied.

Roots of all plant species can take up Fe, Zn, Cu, Ca and Mg in their cationic forms and graminaceous species can also take up Fe, Zn and Cu as metal-chelates (see section ‘Uptake, distribution and accumulation of mineral elements by plants’; Marschner, 1995; White, 2003). Selenium can be taken up by plant roots as selenate, selenite or organoselenium compounds (White et al., 2004, 2007b; Li et al., 2008) and iodine can be taken up as either iodide or iodate (Umaly & Poel, 1971; Mackowiak & Grossl, 1999). The occurrence of these chemical forms in the rhizosphere solution is a function of the soil's physicochemical and biological properties, which will ultimately determine the phytoavailability of these elements in the soil.

Mineral elements can be present in the soil as free ions, or ions adsorbed onto mineral or organic surfaces, as dissolved compounds or precipitates, as part of lattice structures or contained within the soil biota. The most important soil properties governing mineral availability are soil pH, redox conditions, cation exchange capacity, activity of microbes, soil structure, organic matter and water content (Shuman, 1998; Frossard et al., 2000). Indeed, although high concentrations of Fe, Zn and Cu occur in many soils, the phytoavailability of these mineral elements is often restricted by soil properties (see References S4), which predetermine both genetic and agricultural strategies for their effective utilization.

Concentrations of cationic Fe, Zn and Cu in the rhizosphere solution are determined by soil-specific precipitation, complexation and adsorption reactions, and high pH is often the major factor limiting the phytoavailability of these elements. Iron deficiency in plants often occurs on well-aerated, calcareous or alkaline soils (Shuman, 1998; Schmidt, 1999; Frossard et al., 2000). These soils cover 25 to 30% of the land surface and are distributed throughout the world (Frossard et al., 2000; Cakmak, 2002, 2008; Alloway, 2004). In such soils, soluble Fe species rarely exceed 10−10 M (Frossard et al., 2000). Both Zn and Cu deficiencies also occur in plants growing on calcareous or alkaline soils, especially in arid and semi-arid environments (see References S5). It is estimated that about half the agricultural soils in India and Turkey, a third of agricultural soils in China, and most soils in Western Australia lack sufficient phytoavailable Zn (Frossard et al., 2000; Cakmak, 2004; Broadley et al., 2007; Ismail et al., 2007). In nonpolluted areas, typical Zn2+ concentrations in the soil solution range from 10−8 to 10−6 M and Cu2+ concentrations range from 10−9 to 10−6 M (Barber, 1995; Welch, 1995; Frossard et al., 2000; Broadley et al., 2007). Because of their low concentrations in the soil solution and small diffusion coefficients, Zn2+ and Cu2+ have limited mobility in the soil (Gupta, 1979; Barber, 1995; Shuman, 1998; Whiting et al., 2003; Broadley et al., 2007; Cakmak, 2008) and plant roots must forage through the soil to acquire sufficient Zn and Cu for plant nutrition (Rengel, 2001; Hacisalihoglu & Kochian, 2003). Processes that increase Fe, Zn and Cu phytoavailability in the rhizosphere, such as the exudation of protons, phytosiderophores and organic acids by roots, generally increase the concentrations of these elements in crops (Welch, 1995; Rengel, 2001; Abadía et al., 2002; Hoffland et al., 2006; Wissuwa et al., 2006; Ismail et al., 2007; Degryse et al., 2008).

Plants rarely lack a Ca supply from the soil solution sufficient for growth, and Ca2+ concentrations in the rhizosphere solution generally lie in the millimolar range (White & Broadley, 2003). However, Ca deficiency can occur in plants growing on highly weathered tropical soils, because of their low total Ca content (Richey et al., 1982), on strongly acidic soils, where Al3+ may inhibit Ca2+ uptake, and on sodic or saline soils, where excessive sodium (Na+) inhibits Ca2+ uptake (Epstein, 1972; Marschner, 1995). Sodic or saline soils occur worldwide, but mostly in the arid subtropics (Frossard et al., 2000). In addition, several costly Ca-deficiency disorders occur in horticulture, which arise when sufficient Ca is temporarily unavailable to developing tissues (Shear, 1975; Ho & White, 2005). The supply of Ca2+ to field crops is determined by various aspects of soil chemistry including cation exchange capacity, representation of Ca in the base cation pool, the rate at which mineralization of soil organic matter releases Ca2+, and the pH of the soil solution (McLaughlin & Wimmer, 1999).

Magnesium is present as a divalent cation in the soil solution, which, because it binds less avidly to soil particles than other cations, is prone to leaching. This is considered to be an important factor influencing Mg phytoavailability in shallow or coarse-textured soils (Wilkinson et al., 1990). Magnesium deficiency in plants occurs worldwide, especially on strongly acidic soils (Metson, 1974; Wilkinson et al., 1990; Hailes et al., 1997; Aitken et al., 1999), and is aggravated by high concentrations of competing cations, particularly Al3+ and Mn2+, in the soil solution. On alkaline soils, carbonate formation and excess Ca, potassium (K) and Na reduce Mg phytoavailability. It is also possible that the incidence of Mg deficiency in crop plants is increasing as a result of intensive crop production without concomitant Mg fertilization. The Mg concentrations in soil solutions extracted at field capacity generally lie between 125 µM and 8.5 mM, which is sufficient for mass flow to supply Mg to plant roots (Wilkinson et al., 1990; Barber, 1995; Hailes et al., 1997).

The concentration and forms of Se in soils are determined primarily by their geochemistry (Gissel-Nielsen, 1998; Combs, 2001; Gupta & Gupta, 2002; Lyons et al., 2003; Broadley et al., 2006b; White et al., 2007b; Rayman, 2008). Although seleniferous soils can have Se concentrations in excess of 1200 mg Se kg−1, the Se concentrations in most soils lie between 0.01 and 2.0 mg Se kg−1. Selenate is the major Se species in alkaline and oxidized soils (pe + pH > 15), whereas selenite predominates in well-drained mineral soils with a neutral to acidic pH (pe + pH = 7.5 to 15). Selenide species are stable only under low redox conditions (pe + pH < 7.5). Selenate is relatively mobile in the soil solution, but selenite is strongly absorbed by Fe and aluminium (Al) oxides/hydroxides and, to a lesser extent, by clays and organic matter. In most cultivated soils, selenate is the form of Se available to plants (Broadley et al., 2006b; White et al., 2007b).

Iodine is present in soils as iodide, iodate and organic I compounds. Little I is present in the soil solution and most soil I is associated with organic matter, clays and oxides of Fe and Al (Fuge & Johnson, 1986). The prevalent form of I in the soil solution is often iodide, but iodate may also occur depending upon pH and redox conditions (Fuge & Johnson, 1986; Yuita, 1992; Mackowiak & Grossl, 1999; Kodama et al., 2006). As atmospheric I deposition is an important source of soil I, mid-continental soils often lack I (Fuge & Johnson, 1986), but soils with low I concentrations or phytoavailability occur worldwide (Lyons et al., 2004).

Many strategies for the biofortification of crops with essential mineral elements rely on increasing the acquisition of these elements from the soil. However, it is obvious that if the soil contains insufficient amounts of these elements then they must be added to the agricultural system as fertilizers. If sufficient amounts of these elements are present in the soil, then the focus turns to increasing the supply and phytoavailability of these elements in the rhizosphere, and their uptake by plant roots and redistribution to edible portions, such that biofortification is effective.

Uptake, distribution and accumulation of mineral elements by plants

To increase mineral concentrations in edible tissues, without loss of yield, there must be increased uptake by roots (of minerals present in the soil solution) or leaves (for foliar applied minerals), effective redistribution within the plant to the edible portion, and accumulation in edible tissues in a nontoxic form (Welch & Graham, 2005). This section provides a brief overview of the molecular and physiological processes involved in the uptake, distribution and accumulation of mineral elements in plants.

Iron

Plants have two strategies for acquiring Fe from the soil (Marschner, 1995; Welch, 1995; Schmidt, 1999; Gross et al., 2003; Grotz & Guerinot, 2006; Puig et al., 2007a). In strategy I, which is employed by nongraminaceous species, roots acidify the rhizosphere and release organic acids and phenolic compounds to increase Fe3+ concentrations in the soil solution. These compounds chelate Fe3+, which is subsequently reduced to Fe2+ by ferric reductases in the plasma membrane of root epidermal cells, which are encoded by members of the ferric reductase oxidase (FRO) gene family (Robinson et al., 1999; Wu et al., 2005; Mukherjee et al., 2006). Members of the zinc-regulated transporter (ZRT)-, iron-regulated transporter (IRT)-like protein (ZIP) family, such as AtIRT1 in Arabidopsis, then mediate Fe2+ influx to root cells (Vert et al., 2002; Ishimaru et al., 2006). In strategy II, which is employed by cereals and grasses, phytosiderophores (structural derivatives of mugineic acid) are released into the rhizosphere to chelate Fe3+ and the Fe3+-phytosiderophore complex is taken up by root cells (von Wirén et al., 1995; Ishimaru et al., 2006). The chemistry of phytosiderophores is species-specific and determines the contrasting abilities of different grasses and cereals to acquire Fe (Marschner, 1995; Bashir et al., 2006). It is thought that homologues of the maize (Zea mays) yellow stripe 1 (YS1) protein belonging to the oligopeptide transporter (OPT) family are responsible for Fe3+-phytosiderophore uptake by strategy II plants (von Wirén et al., 1995; Curie et al., 2001; Roberts et al., 2004; Ishimaru et al., 2006; Haydon & Cobbett, 2007a; Puig et al., 2007a). The YS1 protein is a proton-coupled metal-complex symporter (Schaaf et al., 2004). Intriguingly, the genomes of both Arabidopsis and rice (Oryza sativa) appear to contain the genes necessary for both strategy I and strategy II Fe acquisition, but there are quantitative differences in their number and qualitative differences in their structure and activity (Gross et al., 2003; Puig et al., 2007a).

Proteins that load Fe into the xylem have not been identified yet, but they are believed to transport Fe2+. Within the xylem, Fe is transported as Fe3+-citrate (Brown & Chaney, 1971; Welch, 1995; von Wirén et al., 1999; Abadía et al., 2002; Mukherjee et al., 2006). FRD3, a member of the multidrug and toxin efflux (MATE) family present in the root pericycle, is important for Fe transport from root to shoot and appears to be involved in loading citrate into the xylem in Arabidopsis (Durrett et al., 2007; Haydon & Cobbett, 2007a; Puig et al., 2007a). Presumably, members of the ZIP family are responsible for Fe2+ uptake by shoot cells. Members of the natural resistance-associated macrophage protein (NRAMP) family are not thought to be responsible for Fe uptake from the soil, but have been implicated in Fe homeostasis within plant cells. In particular, NRAMP3 and NRAMP4 are thought to facilitate Fe2+ release from the vacuole (Thomine et al., 2003; Gross et al., 2003; Hall & Williams, 2003; Lanquar et al., 2005; Grotz & Guerinot, 2006; Puig et al., 2007a), opposing the activity of the vacuolar iron transporter 1 (VIT1) protein which catalyses Fe influx to the vacuole (Kim SA et al., 2006b). In leaves of plants overloaded with Fe, and in some seeds, Fe can accumulate as Fe-chelates in the vacuole (Pich et al., 2001; Lanquar et al., 2005; Kim SA et al., 2006b). However, under most environmental conditions, the majority of cellular Fe is located in the plastid, where it is associated with the Fe-storage protein ferritin (Briat et al., 1999; Petit et al., 2001). The permease in chloroplasts 1 (PIC1) protein is thought to transport Fe from the cytoplasm into the plastid (Duy et al., 2007). It is thought that yellow stripe like (YSL) proteins, and related OPTs, load and unload Fe2+-nicotianamine (Fe2+-NA) complexes into and out of the phloem for Fe redistribution within the plant (see References S6).

The expression of genes encoding many of the proteins responsible for Fe uptake and redistribution within the plant are up-regulated during Fe deficiency. These include genes encoding FROs (Robinson et al., 1999; Wintz et al., 2003; Li et al., 2004; Ishimaru et al., 2006; Mukherjee et al., 2006), ZIPs (Curie et al., 2000; Bereczky et al., 2003; Wintz et al., 2003; Li et al., 2004; Ishimaru et al., 2005; Grotz & Guerinot, 2006), NRAMPs (Curie et al., 2000; Bereczky et al., 2003; Lanquar et al., 2005; Grotz & Guerinot, 2006; Krämer et al., 2007), YS1 (Curie et al., 2001), YSLs (Wintz et al., 2003; Koike et al., 2004; Roberts et al., 2004; Suzuki et al., 2006; Ogo et al., 2007; Stacey et al., 2008) and enzymes involved in the biosynthesis of NA, such as nicotianamine synthase (NAS), and phytosiderophores (Wintz et al., 2003; Bashir et al., 2006; Suzuki et al., 2006; Ogo et al., 2007). The expression of strategy I responses is regulated by the basic helix-loop-helix (bHLH) transcription factor LeFER in tomato (Solanum lycopersicum) (Ling et al., 2002; Bereczky et al., 2003; Li et al., 2004) or its orthologue AtFIT1 in Arabidopsis (Colangelo & Guerinot, 2004; Jakoby et al., 2004; Yuan et al., 2005). However, overexpression of neither LeFER nor AtFIT1 leads to constitutive Fe uptake (Colangelo & Guerinot, 2004; Yuan et al., 2005), suggesting the existence of additional regulatory cascades. It is possible that one of these is initiated by a shoot-derived signal (Enomoto et al., 2007). In rice, the ABI3/VP1 transcription factor OsIDEF1 appears to regulate the expression of strategy II responses through the bHLH transcription factor OsIRO2 (Kobayashi et al., 2007; Ogo et al., 2007).

Zinc

It is commonly assumed that most Zn is transported symplastically across the root to the xylem, although a substantial fraction may traverse the root and reach the xylem via the apoplast (White et al., 2002b; Broadley et al., 2007). Zinc can be taken up across the plasma membrane of root cells as Zn2+ or as a Zn-phytosiderophore complex (von Wirén et al., 1996; Grotz & Guerinot, 2006; Suzuki et al., 2006; Broadley et al., 2007; Ismail et al., 2007). Although some plasma membrane Ca2+ channels are permeable to Zn2+ (Demidchik et al., 2002; White et al., 2002a), it is thought that most Zn2+ influx to the cytoplasm is mediated by ZIPs (ZIP1, ZIP3 and ZIP4; Pence et al., 2000; Assunção et al., 2001; López-Millán et al., 2004; Colangelo & Guerinot, 2006; Broadley et al., 2007; Palmgren et al., 2008), and that YSL proteins catalyse the uptake of Zn-phytosiderophore complexes in strategy II plants (von Wirén et al., 1996; Suzuki et al., 2006; Haydon & Cobbett, 2007a). As the cytoplasm of plant cells contains an abundance of proteins that bind Zn2+, cytoplasmic Zn2+ concentrations are likely to be vanishingly small (Broadley et al., 2007).

Members of the cation diffusion facilitator (CDF) family, such as the metal tolerance proteins MTP1 and MTP3 (see References S7), the Mg2+/H+ antiporter, MHX (Shaul et al., 1999; Elbaz et al., 2006), and the zinc-induced facilitator 1 (ZIF1) transporter (Haydon & Cobbett, 2007b) appear to transport Zn into the vacuole via a Zn2+/H+ antiport mechanism, whilst NRAMPs have been implicated in Zn mobilization from the vacuole (Thomine et al., 2003). Members of the heavy metal P1B-ATPase family (HMA1, HMA2, HMA3 and HMA4) appear to have Zn-transport functions throughout the cell, as well as in loading Zn into the xylem (see References S8). It is thought that Zn is sequestered in the vacuole as an organic acid complex (Broadley et al., 2007). Within the xylem, Zn may be transported as Zn2+ or complexed with organic acids, histidine or nicotianamine (Welch, 1995; von Wirén et al., 1999; White et al., 2002b; Broadley et al., 2007; Palmgren et al., 2008). Members of the ZIP family are thought to mediate Zn2+ influx to leaf cells and to the phloem (Ishimaru et al., 2005). In addition, YSL proteins may load Zn into the phloem, where Zn is transported as a Zn-NA complex, or as a complex with small proteins, to sink tissues (Krüger et al., 2002; Gross et al., 2003; Haydon & Cobbett, 2007a; Puig et al., 2007a; Waters & Grusak, 2008a). Although Zn mobility in the phloem is generally considered to be low, this may not always be the case (Welch, 1995; Haslett et al., 2001).

The genes encoding many of the proteins responsible for Zn uptake, sequestration and redistribution within the plant show up-regulated expression during Zn deficiency. These include ZIPs (Pence et al., 2000; Wintz et al., 2003; Ishimaru et al., 2005; Filatov et al., 2006; Hammond et al., 2006; Talke et al., 2006; van de Mortel et al., 2006; Krämer et al., 2007; Milner & Kochian, 2008), HMAs (Wintz et al., 2003; Papoyan & Kochian, 2004; van de Mortel et al., 2006), YSLs (Wintz et al., 2003; Suzuki et al., 2006; van de Mortel et al., 2006), MTPs (Arrivault et al., 2006; van de Mortel et al., 2006), ZIF1 (Haydon & Cobbett, 2007b), FRD3 (van de Mortel et al., 2006), NASs and genes encoding enzymes involved in the biosynthesis of phytosiderophores (Wintz et al., 2003; Suzuki et al., 2006; Talke et al., 2006; van de Mortel et al., 2006). Interestingly, genes encoding ZIPs, MTPs, NRAMPs, HMAs, FRD3 and YSL proteins and NAS exhibit constitutively high expression in plants that hyperaccumulate Zn (see References S9).

Copper

Copper is taken up as Cu+ by high-affinity transporters belonging to the copper transporter (CTR) family (COPT1, COPT2, COPT3 and COPT4; Sancenón et al., 2003, 2004; Grotz & Guerinot, 2006; Puig et al., 2007a) and/or as Cu2+ by ZIPs (ZIP2 and ZIP4) whose expression in roots is up-regulated by Cu deficiency (Grotz et al., 1998; Wintz et al., 2003). Once inside plant cells, Cu+ is bound by metallotheioneins (Guo et al., 2003, 2008) and Cu chaperone proteins, such as the Arabidopsis AtCCH, AtCCS, AtATX1 and AtCOX17 metallochaperones, which deliver it to specific apoproteins to form biologically active Cu-proteins (Hall & Williams, 2003; Grotz & Guerinot, 2006; Krämer et al., 2007; Puig et al., 2007a,b). The expression of genes encoding metallotheioneins is strongly induced by Cu in the environment and has been shown to correlate with Cu tolerance among Arabidopsis ecotypes and populations of Silene vulgaris and Silene paradoxa (Murphy & Taiz, 1995; van Hoof et al., 2001; Mengoni et al., 2003). Members of the heavy metal P1B-ATPase family remove Cu from the cytoplasm. In Arabidopsis, AtHMA1, AtHMA2, AtHMA3 and AtHMA4 transport divalent Cu2+, whereas AtHMA5, AtHMA6, AtHMA7 and AtHMA8 transport monovalent Cu+ (Baxter et al., 2003; Hall & Williams, 2003; Grotz & Guerinot, 2006; Puig et al., 2007a). These transporters appear to function in Cu detoxification (AtHMA5) or in delivering Cu for the formation of Cu-proteins in the secretory pathway or plastids (AtHMA1; AtHMA6 = AtPAA1; AtHMA7 = AtRAN1; AtHMA8 = AtPAA2; Shikanai et al., 2003; Abdel-Ghany et al., 2005; Andrés-Colás et al., 2006; Seigneurin-Berny et al., 2006; Puig et al., 2007b). Similarly, the rice ATPase OsHMA9 has been implicated in Cu efflux from the cytoplasm (Lee et al., 2007). The NRAMPs have been implicated in Cu2+ transport to the vacuole (Hall & Williams, 2003) and Cu-binding vegetative storage proteins (VSPs) appear to have a role in Cu homeostasis and Cu detoxification (Mira et al., 2002; Kung et al., 2006). It is not known how Cu is loaded into the xylem, but it is transported in a complexed form, probably as Cu2+-NA (von Wirén et al., 1999). Estimates of Cu mobility in the phloem vary widely (Welch, 1995). Copper is probably loaded into the phloem by a YSL protein and transported complexed with small proteins or as Cu-NA (Mira et al., 2001; Guo et al., 2003; DiDonato et al., 2004; Puig et al., 2007a; Waters & Grusak, 2008a). Interestingly, the YSL proteins may transport both Cu-NA complexes and the free Cu2+ and Fe2+ cations (Wintz et al., 2003).

Calcium

It is thought that Ca can reach the xylem solely via the root apoplast in regions where Casparian bands are absent or via the cytoplasm of unsuberized endodermal cells where Casparian bands are present (White, 2001; Moore et al., 2002). However, the relative contributions of apoplastic and symplastic pathways to the delivery of Ca to the xylem are unknown. Calcium influx to root cells is mediated by a variety of Ca2+-permeable cation channels (White, 2000; White et al., 2002a; White & Broadley, 2003; Demidchik & Maathuis, 2007; Haswell, 2007; Roux & Steinebrunner, 2007; Miedema et al., 2008; Wheeler & Brownlee, 2008). These include hyperpolarization-activated Ca2+ channels, thought to be formed by plant annexins, voltage-independent cation channels, thought to be formed by members of the cyclic nucleotide gated channel (CNGC) and/or glutamate receptor (GLR) protein families, and depolarization-activated Ca2+ channels, one of which may be encoded by homologues of the AtTPC1 gene (see References S10). The plasma membrane may also contain mechanosensitive ion channels permeable to Ca2+ encoded by members of the mechanosensitive channel of small conductance (MscS)-like (MSL) gene family (Haswell, 2007). The activity of all these ion channels is exquisitely regulated, because explicit perturbations in cytosolic Ca2+ concentrations co-ordinate specific responses to many developmental and environmental stimuli (White & Broadley, 2003). Homologues of the low-affinity cation transporter of wheat (TaLCT1) can also facilitate Ca2+ influx to root cells (Clemens et al., 1998). Cytosolic Ca2+ is maintained at submicromolar concentrations by Ca2+-ATPases, encoded by members of the P2B-ATPase (ECA/ACA) gene family, and Ca2+/H+ antiporters, such as those encoded by the Ca2+/H+ antiporter (CAX) genes, which export Ca2+ to the apoplast, endoplasmic reticulum, plastids or vacuoles (Hirschi, 2001; Baxter et al., 2003; White & Broadley, 2003; Shigaki & Hirschi, 2006; George et al., 2008). CAX-mediated Ca2+ influx to vacuoles is energized by the H+ gradient generated by vacuolar ATPases and/or PPiases (Shigaki & Hirschi, 2006). Within the cytosol Ca2+ is complexed to diverse proteins including calmodulin, calmodulin-related proteins, calcineurin-B-like proteins, Ca2+-dependent protein kinases and annexins (White & Broadley, 2003). The Ca2+-binding proteins calreticulin, calsequestrin, calnexin and the lumenal binding protein (BiP) are present in the endoplasmic reticulum (White & Broadley, 2003) and vacuoles may also contain Ca2+-binding proteins, such as the radish (Raphanus sativus) vacuolar calcium-binding (RVCaB) protein (Yuasa & Maeshima, 2000). Calcium is released from the vacuole through Ca2+-permeable cation channels (White, 2000), which again may include homologues of AtTPC1 and/or annexins in some plant species (Peiter et al., 2005; Pottosin & Schönknecht, 2007; Mortimer et al., 2008; Wheeler & Brownlee, 2008). No gene encoding any ligand-gated, vacuolar Ca2+ channel has been identified to date (Nagata et al., 2004; Krinke et al., 2007; Wheeler & Brownlee, 2008).

Calcium transport to the shoot occurs largely from the root apex and/or regions of lateral root initiation (White, 2001) and, within the xylem, Ca appears to be transported either as Ca2+ or complexed with organic acids (Welch, 1995). When grown in the same environment, differences between taxa in their shoot Ca concentrations correlate well with cell wall chemistry and cation-binding capacity (White & Broadley, 2003; White, 2005), but the absolute shoot Ca concentration found in leaves depends greatly upon the phytoavailability of Ca in the rhizosphere and the transpirational water flux (White, 2001; White & Broadley, 2003). As Ca is almost immobile in the phloem, fruits, seeds and tubers rely on its delivery via the xylem and, consequently, contain low Ca concentrations (Welch, 1999; White & Broadley, 2003; Ho & White, 2005; White et al., in press). In vacuoles, Ca is present as Ca2+, as soluble complexes with proteins and/or organic acids, and in insoluble forms, such as Ca-oxalate and Ca-phytate, depending upon the plant species and the phytoavailability of Ca in the environment (Kinzel, 1982; Kinzel & Lechner, 1992; White & Broadley, 2003; Franceschi & Nakata, 2005; White, 2005). Some plants accumulate soluble Ca in specific cell types, such as leaf trichomes (White & Broadley, 2003), and the formation Ca-oxalate crystals also occurs in specific cell types, according to genetic predisposition (Franceschi & Nakata, 2005).

Magnesium

It is thought that Mg enters root cells through Mg2+-permeable cation channels (White, 2000; White & Broadley, 2003) and/or that members of the mitochondrial RNA splicing 2 (MRS2) family of transport proteins (AtMGT1 and AtMGT10) catalyse Mg2+ influx across the plasma membrane (Schock et al., 2000; Li et al., 2001b; Shaul, 2002; Gardner, 2003). Most cellular Mg is associated with proteins, and cytosolic Mg2+ concentrations approximate 0.4 mM (White et al., 1990). The vacuole is the main storage compartment for Mg in plants and the MHX transporter, which is encoded by a single gene in Arabidopsis, is thought to dominate Mg2+ transport into the vacuole (Shaul et al., 1999; David-Assael et al., 2006). Interestingly, AtMHX co-localizes with a major chromosomal quantitative trait locus (QTL) affecting seed Mg concentration in Arabidopsis (Vreugdenhil et al., 2004). Magnesium is released from the vacuole through Mg2+-permeable cation channels, including the slow vacuolar (SV) channel (White, 2000; Pottosin & Schönknecht, 2007). It is possible that ATPases catalyse Mg2+ efflux from root cells into the xylem, where it is transported either as Mg2+ or complexed with organic acids (Welch, 1995).

Shoot Mg concentrations in plants supplied with adequate Mg generally approximate 1 to 10 mg g−1 dry matter (DM) (Wilkinson et al., 1990; Broadley et al., 2004). Approximately 75% of leaf Mg appears to be associated with protein synthesis via its roles in ribosomal structure and function (Wilkinson et al., 1990). In leaves, the MRS2-11 transporter is thought to facilitate Mg entry to the chloroplast (Drummond et al., 2006), where between 15 and 20% of the Mg in leaves is associated with chlorophyll (Wilkinson et al., 1990). Because Mg is a phloem-mobile element, it is readily translocated to fruit, seed and tubers (Wilkinson et al., 1990; White et al., in press).

Selenium

Plant roots can take up Se as selenate, selenite or organoselenium compounds, such as selenocysteine (SeCys) and selenomethionine (SeMet), but cannot take up colloidal elemental Se or metal selenides (White et al., 2004, 2007b). Selenate is transported across the plasma membrane of root cells by high-affinity sulphate transporters (HASTs; Terry et al., 2000; White et al., 2004, 2007b; Sors et al., 2005b; Broadley et al., 2006b; Hawkesford & Zhao, 2007), whilst selenite is thought to be transported by phosphate transporters (Li et al., 2008). Selenite is rapidly converted to organoselenium compounds in the root, whereas selenate is delivered to the xylem and transported to the shoot, where it is assimilated into organoselenium compounds and redistributed within the plant in a manner analogous to S (Terry et al., 2000; Broadley et al., 2006b; Sors et al., 2005b; White et al., 2007b; Hawkesford & Zhao, 2007; Li et al., 2008).

The enzymes of the Se/S assimilation pathway are generally encoded by extensive gene families whose products are directed to different intracellular compartments (Hawkesford & De Kok, 2006). Selenate is thought to be activated by adenosine triphosphate sulphurylase to form adenosine 5′-phosphoselenate (APSe), which then is reduced to selenite by adenosine 5′-phosphosulphate (APS) reductase. The activity of APS reductase appears to exert a major effect on the flux through the Se/S assimilation pathway (Vauclare et al., 2002). Selenite is then reduced to selenide by a sulphite reductase located in the chloroplast. SeCys is synthesized from serine and selenide by cysteine synthase, an enzyme complex containing both serine acetyl transferase and O-acetylserine (thiol) lyase subunits. The activity of cysteine synthase is regulated dynamically by the concentration of O-acetylserine. Selenomethionine is synthesized from SeCys and O-phosphohomoserine (OPHS) through the sequential actions of cystathionine γ-synthase, which produces selenocystathionine, cystathionine β-lyase, which produces homoselenocysteine, and methionine synthase. It is noteworthy that OPHS is also the precursor of threonine and an increase in the concentration of S-adenosyl methionine (SAM), which is synthesized from methionine by SAM synthase, accelerates the conversion of OPHS to threonine, rather than methionine, by allosteric activation of threonine synthase (Curien et al., 1998). The number of cystathionine γ-synthase transcripts is also reduced by increasing SAM or methionine concentrations (Chiba et al., 1999; Hesse & Hoefgen, 2003). Both SeCys and SeMet can either be incorporated into proteins or methylated. For example, Se-methylselenocysteine (SeMSeCys), γ-glutamyl-SeMSeCys and Se-methylselenomethionine are characteristic Se assimilation products of species in the genera Allium and Brassica (Broadley et al., 2006b; White et al., 2007b). Unfortunately, the nonspecific replacement of cysteine by SeCys and/or methionine by SeMet in proteins can alter their stability and activity. The methylation of SeCys and SeMet is thought to reduce the incorporation of these amino acids into proteins and may account for the ability of some plants to accumulate high tissue Se concentrations (Brown & Shrift, 1982). Shoot Se concentrations tend to increase to a maximum during seedling growth, then decline before, or upon, flowering (Rosenfeld & Beath, 1964; Xue et al., 2001; Turakainen et al., 2004; White et al., 2007b), which is consistent with transcriptional analyses suggesting that Se/S assimilation occurs predominantly in the first leaves that a plant produces (White et al., 2007b). Selenium is thought to be redistributed within the plant as selenate and/or organoselenium compounds via the phloem (White et al., 2007b).

Iodine

It is thought that plant root cells take up I as the iodide anion (Umaly & Poel, 1971; Mackowiak & Grossl, 1999; Zhu et al., 2003; Blasco et al., 2008) and that I follows the chloride (Cl) transport pathway with H+/anion symporters catalysing I uptake and anion channels releasing I into the xylem (White & Broadley, 2001; Roberts, 2006). However, the molecular identities of transporters catalysing these fluxes are not firmly established. Putative H+/halide transporters belong to the chloride channel (CLC) transporter family (De Angeli et al., 2006; Marmagne et al., 2007) and a subset of the ATP-binding cassette (ABC) protein superfamily (Verrier et al., 2008), whilst Na:K/Cl symporters belong to the cation chloride co-transporter (CCC) gene family (Colmenero-Flores et al., 2007). Homologues of the band 3 protein are also suspected to transport I (Frommer & von Wirén, 2002; Bruce et al., 2004). Plant Cl channels are readily permeable to I (Barbier-Brygoo et al., 2000; White & Broadley, 2001; Roberts, 2006). These are thought to be encoded by CLC genes (Barbier-Brygoo et al., 2000; White & Broadley, 2001; Nakamura et al., 2006), although other members of this family are I-permeable H+/anion antiporters (De Angeli et al., 2006). Halide fluxes may also be facilitated by organic acid transporters (White & Broadley, 2001). Iodide-permeable H+/anion antiporters and anion channels in the tonoplast are likely to mediate I fluxes into and out of the vacuole (White & Broadley, 2001; De Angeli et al., 2006; Nakamura et al., 2006). Although fertilizer I is readily accumulated in roots and leaves (Mackowiak & Grossl, 1999; Zhu et al., 2003; Dai et al., 2004, 2006; Kashparov et al., 2005; Mackowiak et al., 2005; Blasco et al., 2008), little I is redistributed via the phloem to fruits or seeds (Muramatsu et al., 1993, 1995).

Variation in tissue concentrations of mineral elements among plant species

Tissue concentrations of mineral elements can differ markedly between plant species growing in the same environment. Systematic variation in shoot concentrations of Fe, Zn, Cu, Ca, Mg and Se has been documented (Thompson et al., 1997; Broadley et al., 2001, 2003, 2004, 2007; White et al., 2004, 2007a; Watanabe et al., 2007). There is also variation in leaf I concentration among angiosperm species, although this has not been quantified explicitly (Fuge & Johnson, 1986; Dai et al., 2004).

Much of the genetic variation in shoot Ca and Mg concentrations occurs at the ordinal level or above (Thompson et al., 1997; Broadley et al., 2003, 2004; Watanabe et al., 2007). This implies that concentrations of these elements in shoot tissues are constrained by an ancient evolutionary heritage. Commelinoid monocots, such as cereals and grasses, have lower shoot Ca concentrations than eudicot species (Sleper et al., 1989; Thompson et al., 1997; Broadley et al., 2003, 2004). This has been attributed to differences in their cell wall chemistry and cation exchange capacity (White & Broadley, 2003; White, 2005). Among the eudicots, members of orders within the rosid (Brassicales, Cucurbitales, Malvales and Rosales) and asterid (Apiales, Asterales, Lamiales and Solanales) clades typically have higher shoot Ca concentrations (Broadley et al., 2003, 2004). Within the magnoliids, members of noncommelinoid orders, such as the Asparagales, have higher shoot Ca concentrations than members of the commelinoid orders (Broadley et al., 2003, 2004). There is a strong correlation between the ability of a plant to accumulate Ca and its ability to accumulate Mg (White, 2001, 2005; Broadley et al., 2004). Thus, phylogenetic variation in shoot Mg concentrations generally resembles that for shoot Ca concentrations. However, species from families within the Caryophyllales, including the Amaranthaceae and Caryophyllaceae, have a tendency towards uncommonly high leaf Mg concentrations and therefore lower leaf Ca/Mg quotients than most other angiosperms (Broadley et al., 2004, 2008; Watanabe et al., 2007). For example, Ca/Mg quotients for 322 spinach (Spinacia oleracea, Amaranthaceae) accessions grown hydroponically were typically 5-fold lower than the Ca/Mg quotients for over 400 accessions of Brassica oleracea (Brassicaceae) grown under a variety of conditions (Broadley et al., 2008). At the subordinal level, species and ecotypes adapted to Ca-rich environments generally have lower tissue Ca concentrations than those adapted to low-Ca soils when they are grown together (Zohlen & Tyler, 2004) and nonCaryophyllales species and ecotypes adapted to Mg-rich, serpentine environments generally have lower shoot Mg concentrations than congenerics or conspecifics adapted to other soils, when grown in the same environment (Bradshaw, 2005; O'Dell et al., 2006).

In contrast to Ca and Mg, relatively little variation in shoot Fe, Zn, Cu and Se concentrations occurs at the ordinal level or above (Broadley et al., 2001, 2007; White et al., 2004, 2007a; Watanabe et al., 2007). This implies that closely related species will often differ substantially in their tissue Fe, Zn, Cu and/or Se concentrations. Indeed, some plant species ‘hyperaccumulate’ Zn, Cu and Se, and can contain leaf concentrations of these elements several orders of magnitude greater than those in closely-related species growing on the same substrate (Rosenfeld & Beath, 1964; Reeves & Baker, 2000; Broadley et al., 2001, 2007; White et al., 2007a). Traits determining the accumulation of extraordinarily high concentrations of these elements appear to have evolved by convergent evolution of appropriate transport and metabolic pathways in several distinct angiosperm clades (Brown & Shrift, 1982; Reeves & Baker, 2000; Broadley et al., 2001, 2007; White et al., 2004, 2007a,b). Members of the Brassicaceae (e.g. Arabidopsis halleri and Noccaea spp.), Caryophyllaceae (e.g. Minuartia verna), Polygonaceae (e.g. Rumex acetosa) and Dichapetalaceae (e.g. Dichapetalum gelonioides) have been observed to hyperaccumulate Zn (Broadley et al., 2007). Members of several angiosperm orders, including the Fabales, Malvales, Asterales, Solanales, Lamiales, Caryophyllales and Poales, appear to hyperaccumulate Cu (Reeves & Baker, 2000). Members of the Fabaceae (e.g. Astragalus bisulcatus and Astragalus racemosus), Asteraceae (e.g. Aster occidentalis and Machaeranthera ramosa) and Brassicaceae (e.g. Stanleya pinnata) hyperaccumulate Se in their leaves (Rosenfeld & Beath, 1964; White et al., 2004, 2007a), whilst some members of the Lecythidaceae family accumulate large Se concentrations in their fruits and seeds (Broadley et al., 2006b; White et al., 2007b).

In addition to the phenomenon of hyperaccumulation, there are general differences among angiosperm orders in their shoot Zn, Cu and Se concentrations (Broadley et al., 2001, 2007; White et al., 2004, 2007a). Shoot Cu concentrations appear to be lower in members of the Brassicales and Poales, and higher in members of the Malvales and Malphigiales (Broadley et al., 2001). Shoot Zn concentrations are lower in the Ericales and commelinoid monocotyledons, and higher in the Caryophyllales and noncommelinoid monocotyledons. In a meta-analysis of 1108 studies comparing shoot Zn concentrations among 365 species from 48 plant families and 12 key angiosperm clades, Broadley et al. (2007) reported that the lowest shoot Zn concentrations occurred in the Linaceae, Poaceae and Solanaceae, and the highest shoot Zn concentrations occurred in the Amaranthaceae and Salicaceae. Similarly, differences in Se metabolism occur among angiosperm orders that affect tissue Se concentrations and bioavailability to animals. For example, brassicas and alliums accumulate unique organo-Se compounds in their tissues and, consequently, have higher tissue Se concentrations than many other plants grown under the same conditions (Broadley et al., 2006b; White et al., 2007a,b).

One consequence of the strong phylogenetic control of tissue Ca and Mg concentrations, and above all the lower tissue Ca and Mg concentrations in commelinoid monocotyledon species than in species from other angiosperm families, is an increased risk of Ca- and Mg-related deficiency disorders in populations changing from bean-rich to cereal-rich diets (Graham et al., 2001; Welch & Graham, 2004; White & Broadley, 2005a). Similarly, there has been an increase in Fe-deficiency anaemia and Zn-deficiency disorders as cereals have replaced traditional, more mineral-rich dicotyledonous crops such as pulses, vegetables and fruits (Welch & Graham, 2004). The seeds of cereals generally have far lower concentrations of Fe and Zn than seeds of legumes (Table 1; Rengel et al., 1999; White & Broadley, 2005a). This highlights the importance of the choice of plant species in strategies designed to increase the delivery of mineral elements to vulnerable populations.

Table 1.  Examples of variation in the concentrations of essential mineral elements in edible tissues among genotypes of common crops grown under the same conditions
CropGenotypes, trial site and reference[Fe] (mg kg−1 DW)[Zn] (mg kg−1 DW)[Ca] (g kg−1 DW)[Mg] (g kg−1 DW)[Cu] (mg kg−1 DW)
Rice (Oryza sativa); brown grainCore collection6–2414–58
Field trialQ = 3.87Q = 4.34   
Gregorio et al. (2000)(n = 1138)(n = 1138)   
Rice (Oryza sativa); polished grainCore collection4–308–952.5–143.5
Field trialQ = 7.38Q = 11.6  Q = 57.4
Yang et al. (1998)(n = 285)(n = 285)  (n = 285)
Wheat (Triticum species); grainSelected genotypes25–7325–92
Field trial (Obregon)Q = 2.92Q = 3.68   
Monasterio & Graham (2000)(n = 324)(n = 324)   
Wheat (Triticum aestivum); grainBread wheat genotypes29–5725–530.25–0.730.92–1.43
Field trial (El Batan)Q = 1.96Q = 2.12Q = 2.92Q = 1.56 
Graham et al. (1999)(n = 132)(n = 132)(n = 132)(n = 132) 
Wheat (Triticum aestivum); grainCommercial varieties24–4316–261.7–2.9
Field trial (Hutchinson)Q = 1.75Q = 1.64  Q = 1.62
Garvin et al. (2006)(n = 14)(n = 14)  (n = 14)
Wheat (Triticum aestivum); grainElite genotypes26–4514–200.80–1.39
Field trialQ = 1.71Q = 1.38 Q = 1.74 
Oury et al. (2006)(n = 51)(n = 51) (n = 51) 
Triticale (Triticosecale); grainSelected cultivars28–3621–310.35–0.501.20–1.464.7–7.5
Six field trialsQ = 1.27Q = 1.50Q = 1.42Q = 1.22Q = 1.60
Feil & Fossati (1995)(n = 10)(n = 10)(n = 10)(n = 10)(n = 10)
Maize (Zea mays); grainCore collection16–6313–58
Field trial (Harare)Q = 3.85Q = 4.46   
Bänziger & Long (2000)(n = 1417)(n = 1417)   
Maize (Zea mays); kernelElite varieties17–2416–25
Field trials (mean 3 sites)Q = 1.45Q = 1.49   
Oikeh et al. (2003a)(n = 49)(n = 49)   
Pearl millet (Pennisetum glaucum); grainDiverse germplasm30–7625–65
Field trial (2 seasons)Q = 2.51Q = 2.64   
Velu et al. (2007)(n = 120)(n = 120)   
Pearl millet (Pennisetum glaucum); grainCommercial germplasm70–18053–700.10–0.801.80–2.7010.0–18.0
Field trialQ = 2.57Q = 1.32Q = 8.00Q = 1.50Q = 1.80
Abdalla et al. (1998)(n = 10)(n = 10)(n = 10)(n = 10)(n = 10)
Barley (Hordeum vulgare); grainCore collection21–83
Field trailQ = 3.95    
Ma et al. (2004)(n = 409)    
Barley (Hordeum vulgare); grainCommercial cultivars104–18930–380.82–1.182.09–2.35
Field trialQ = 1.83Q = 1.26Q = 1.44Q = 1.12 
P. J. White & I. J. Bingham, unpublished(n = 16)(n = 16)(n = 16)(n = 16) 
Sorghum (Sorghum bicolor); grainDiverse germplasm20–3713–31
Field trialQ = 1.84Q = 2.31   
Reddy et al. (2005)(n = 84)(n = 84)   
Bean (Phaseolus vulgaris); seedCore collection35–9221–590.5–3.1
Field trialQ = 2.65Q = 2.86Q = 6.20  
Islam et al. (2002)(n = 1072)(n = 1072)(n = 1072)  
Bean (Phaseolus vulgaris); seedSelected genotypes48–7417–281.39–2.041.47–1.965.0–10.0
Field trialsQ = 1.54Q = 1.65Q = 1.46Q = 1.33Q = 2.00
Ariza-Nieto et al. (2007)(n = 8)(n = 8)(n = 8)(n = 8)(n = 8)
Pea (Pisum sativum); seedCore collection23–10516–1070.28–2.561.06–2.471.4–13.8
Glasshouse trialQ = 4.5Q = 6.6Q = 9.1Q = 2.3Q = 10.1
Grusak & Cakmak (2005)(n = 481)(n = 481)(n = 481)(n = 481)(n = 481)
Soybean (Glycine max); seedCommercial lines38–671.8–3.42.2–3.4
Field trial Q = 1.76Q = 1.89Q = 1.55 
Raboy et al. (1984) (n = 38)(n = 38)(n = 38) 
Soybean (Glycine soja); seedCommercial lines59–832.3–4.82.2–3.1
Field trial Q = 1.41Q = 2.09Q = 1.41 
Raboy et al. (1984) (n = 20)(n = 20)(n = 20) 
Peanut (Arachis hypogaea); seedDiverse germplasm24–4125–410.4–0.71.7–2.38.0–17.0
Field trialQ = 1.70Q = 1.64Q = 2.00Q = 1.35Q = 2.13
Branch & Gaines (1983)(n = 26)(n = 27)(n = 26)(n = 26)(n = 26)
Chickpea (Cicer arietinum); seedCore collection42–13345–1230.81–3.021.09–2.590.9–9.3
TrialQ = 3.17Q = 2.72Q = 3.73Q = 2.38Q = 10.6
Grusak (2006)**(n = 239)(n = 239)(n = 239)(n = 239)(n = 239)
Chickpea (Cicer arietinum); seedCommercial cultivars24–4135–601.9–2.20.04–0.05107–122
Field trialQ = 1.71Q = 1.71Q = 1.18Q = 1.16Q = 1.14
Zia-Ul-Haq et al. (2007)(n = 4)(n = 4)(n = 4)(n = 4)(n = 4)
Chickpea (Cicer arietinum); edible leafDiverse accessions79–12056–13714.4–22.63.2–4.61.3–3.0
Glasshouse trialQ = 1.51Q = 2.46Q = 1.57Q = 1.44Q = 2.31
Ibrikci et al. (2003)(n = 19)(n = 19)(n = 19)(n = 19)(n = 19)
Brassica oleracea; leavesCore collection23–10459–22114.3–39.24.0–9.61.4–9.3
Glasshouse trial (P4)Q = 45.2Q = 25.9Q = 2.75Q = 2.43Q = 6.76
Broadley et al. (2008)(n = 345)(n = 339)(n = 348)(n = 348)(n = 346)
Brassica oleracea; leavesCommercial varieties71–33834–12219.1–35.04.2–8.32.9–6.4
Glasshouse trial (P4)Q = 4.79Q = 3.56Q = 1.83Q = 1.98Q = 2.24
Broadley et al. (2008)(n = 74)(n = 74)(n = 74)(n = 74)(n = 74)
Kale and collards (B. oleracea var. acephala); leavesCommercial varieties69–10933–6014.6–26.12.9–4.8
Glasshouse trialQ = 1.57Q = 1.78Q = 1.79Q = 1.68 
Kopsell et al. (2004b)(n = 22)(n = 22)(n = 22)(n = 22) 
Broccoli (B. oleracea var. italica); inflorescencesCommercial cultivars3.40–5.131.34–1.68
Field trial (SS-P)  Q = 1.51Q = 1.25 
Rosa et al. (2002)  (n = 11)(n = 11) 
Brassica rapa; leavesCore collection60–35023–156
Hydroponic trialQ = 5.80Q = 6.72   
Wu et al. (2007)(n = 111)(n = 111)   
Spinach (Spinacia oleracea); leavesCore collection50–13931–3873.0–11.55.2–14.72.2–11.0
Growth chamber trialQ = 2.7Q = 12.3Q = 3.8Q = 2.9Q = 4.9
Grusak & Cakmak (2005)(n = 327)(n = 327)(n = 327)(n = 327)(n = 327)
Carrot (Daucus carota); rootsCommercial cultivars32–19818–392.7–4.50.8–2.3
Field trialQ = 6.19Q = 2.17Q = 1.65Q = 2.89 
Nicolle et al. (2004b)(n = 20)(n = 20)(n = 20)(n = 20) 
Cassava (Manihot esculenta); rootsSelected genotypes6–2303–380.31–2.500.52–2.400.8–40.3
Field trialQ = 38.3Q = 14.3Q = 8.1Q = 4.6Q = 51.0
Chávez et al. (2005)(n = 600)(n = 600)(n = 600)(n = 600)(n = 599)
Potato (Solanum tuberosum); tubersCommercial varieties32–3747–170.27–0.670.87–1.232.3–4.9
Field trialQ = 11.6Q = 2.40Q = 2.48Q = 1.41Q = 2.12
White et al. (in press)(n = 26)(n = 26)(n = 26)(n = 26)(n = 26)
Yam (Dioscorea alata); tubersCore collection9–1768–250.14–0.490.20–0.365.0–13.0
Field trialQ = 19.6Q = 3.13Q = 3.5Q = 1.82Q = 2.6
Agbor-Egbe & Trèche (1995)(n = 23)(n = 23)(n = 23)(n = 23)(n = 23)
Data show minimum-to-maximum concentrations (Fe, Zn and Cu, mg kg−1 dry matter; Mg and Ca, g kg−1 dry matter), the maximum/minimum quotient (Q) and the number of genotypes surveyed (n). *For carrot roots, a fresh weight (FW):dry matter (DM) ratio of 10 was assumed. **Data for chickpea accessions obtained from the Germplasm Resources Information Network (GRIN) website (http://www.ars-grin.gov/cgi-bin/npgs/html/eval.pl?492937). DW, dry weight.

In addition to the phylogenetic heritage of different plant species affecting their ability to accumulate essential mineral elements, the concentrations of mineral elements in edible tissues are also influenced by their mobility within the plant. For example, although Se and Mg are transported readily in the phloem, Fe, Zn, Cu and I are not, and Ca has little phloem mobility (Epstein, 1972; Mackowiak & Grossl, 1999; Welch, 1999; White & Broadley, 2003). Thus, phloem-fed tissues such as fruits, seeds and tubers are often poor sources of Fe, Zn, Cu, I and Ca, whilst leafy vegetables are rich sources of these elements (Table 1; Welch, 1999; White & Broadley, 2005a). The bioavailability of Ca also depends on whether tissues also contain oxalate, as Ca oxalate is not readily absorbed in the human gut. For example, edible portions from species within the Oxalidales (e.g. carambola (Averrhoa carambola) and oca (Oxalis tuberose)), Caryophyllales (e.g. beet/chard (Beta vulgaris), amaranth (Amaranthus spp.), rhubarb (Rheum rhabarbarum) and spinach (S. oleracea)) and Malpighiales (e.g. castor bean (Ricinus communis) and linseed (Linum usitatissimum)) often contain high Ca concentrations, but the Ca bioavailability is low because of high oxalate concentrations (White & Broadley, 2003; White, 2005; Kim et al., 2007; Titchenal & Dobbs, 2007).

Historical trends in the concentrations of mineral elements in edible tissues

Analyses of historical data have suggested that the concentrations of certain mineral elements in edible produce from developed countries have declined over the last half century (Davis, in press). It appears that the mean concentrations of Cu, and possibly also Mg, in the dry matter of vegetables available in the UK declined significantly between the 1930s and the 1980s (White & Broadley, 2005b; Broadley et al., 2006a; Davis, 2006), that the mean concentrations of Fe, Cu and Ca in the dry matter of horticultural produce available in the USA has declined significantly since the mid twentieth century (Davis et al., 2004, White & Broadley, 2005b; Davis, 2006), and that the Zn and Cu concentrations in the dry matter of cereal products, vegetables and fruits from Finland have declined over the last 25 yr (Ekholm et al., 2007). Holden et al. (in press) have cautioned that historical data for the mineral composition of foods can be compromised by differences in crop genotype, crop husbandry, environmental factors, geographical sampling strategy, portion analysed and analytical methods, and by inter-laboratory variability. However, because similar changes in the concentrations of mineral elements in produce have occurred in different countries, which share similar historical farming practices, it has been suggested that this phenomenon might be a consequence of the adoption of modern varieties and/or agronomic practices (White & Broadley, 2005b). The best evidence of this phenomenon, to date, shows that a decline in the concentrations of Fe, Zn, Cu and Mg in wheat (Triticum aestivum) grain coincided with the introduction of semi-dwarf high-yielding cultivars in the Broadbalk Wheat Experiment at Rothamsted, UK (Fan et al., 2008).

Recent research has focused on the effects of increased yield, whether achieved by agronomic or genetic improvement, on the concentrations of mineral elements in produce. It has long been appreciated that environmental factors accelerating plant growth rates, such as higher temperatures, light intensity, CO2 concentrations and irrigation, often result in reduced concentrations of mineral elements in plant tissues (Jarrell & Beverly, 1981; Loladze, 2002), and a number of recent studies have shown that the concentrations of various mineral elements are lower in higher yielding genotypes. For example, weak negative relationships have been found between grain Fe and Cu concentrations and grain yield among triticale cultivars (Feil & Fossati, 1995), and between seed Fe or Zn concentrations and seed yield among sorghum (Sorghum bicolor) genotypes (Reddy et al., 2005). Similarly, negative relationships between Fe, Zn, Mg, Se and phosphorus (P) concentrations in grain and grain yield have been observed among cultivars of bread (T. aestivum) and durum (Triticum durum) wheat (Bänziger & Long, 2000; Monasterio & Graham, 2000, Garvin et al., 2006; Oury et al., 2006; Distelfeld et al., 2007; Ortiz-Monasterio et al., 2007; McDonald et al., 2008), although the strength of these relationship is influenced greatly by the environment. In leafy vegetables, Farnham et al. (2000) found a strong negative relationship between Ca and Mg concentrations and head weight among 27 broccoli (Brassica oleracea var. italica) genotypes, and Broadley et al. (2008) observed weak negative relationships between biomass yield and shoot Ca concentrations among genotypes of the gongylodes (kohlrabi) and sabauda (Savoy cabbage) subtaxa of B. oleracea and between biomass yield and shoot Mg concentrations among genotypes of the gongylodes and acephala (kale) subtaxa (Broadley et al., 2008). However, negative relationships between the concentrations of mineral elements and the yield of edible produce are not always observed in crop genotypes (White et al., in press). For example, some studies report no significant relationships between the concentrations of particular mineral elements in grain and the yield of cereals (Feil & Fossati, 1995; Graham et al., 1999; Ortiz-Monasterio et al., 2007), between seed Fe and Zn concentrations and yield in common bean (Phaseolus vulgaris; Graham et al., 2001), between shoot mineral concentrations and biomass production within most subtaxa of B. oleracea (Broadley et al., 2008), or between the concentrations of mineral elements in tubers and tuber yield among potato (Solanum tuberosum) varieties (White et al., in press). These observations suggest that the biofortification of crops with mineral elements can be achieved without compromising yield.

Agronomic biofortification strategies

Agronomic strategies to increase the concentrations of mineral elements in edible tissues generally rely on the application of mineral fertilizers and/or improvement of the solubilization and mobilization of mineral elements in the soil. When crops are grown where mineral elements become immediately unavailable in the soil, targeted application of soluble inorganic fertilizers to roots or to leaves is practised. In situations where mineral elements are not readily translocated to edible tissues, foliar applications of soluble inorganic fertilizers are made. It has been observed that the human population of the world has exceeded the carrying capacity of low-input agriculture, and modern inorganic fertilizers are necessary to obtain the crop yields required to prevent starvation (Graham et al., 2007). It is argued, therefore, that the use of inorganic fertilizers must be included in any future strategy for food security. If the widespread use of inorganic fertilizers is facilitated, it might be possible to incorporate mineral elements essential for human nutrition before their distribution, as is practised for Se in Finland and Zn in Turkey.

Inorganic fertilizers

Soils often contain large amounts of Fe, but little of this is phytoavailable. The application of inorganic Fe fertilizers to such soils is usually ineffective as it rapidly becomes unavailable to plant roots through adsorption, precipitation and oxidation reactions. For this reason, Fe-chelates are often used as soil Fe fertilizers (Shuman, 1998; Rengel et al., 1999). In addition, the availability of Fe in the rhizosphere can be increased by soil acidification with elemental S (Shuman, 1998). This has the added benefit of crop S fertilization. Foliar applications of Fe fertilizers are often made to crops growing in Fe-deficient soils, but, because Fe is not readily translocated within plants, these must be repeated throughout the growing season (Loneragan, 1997; Cakmak, 2002). Nevertheless, by appropriate Fe fertilization, Fe concentrations in edible portions of cereals, vegetables and fruits can be increased (Shuman, 1998; Rengel et al., 1999).

Zinc is commonly applied to crops as ZnSO4 or as synthetic chelates (Shuman, 1998; Broadley et al., 2007; Cakmak, 2008). The application of Zn fertilizers to the soil is effective in increasing grain Zn concentrations in cereals growing on most, but not all, soils and foliar applications of either ZnSO4 or Zn-chelates can increase grain Zn concentrations in plants with adequate Zn mobility in the phloem (Rengel et al., 1999; Cakmak, 2002, 2004, 2008; Genc et al., 2005; Oury et al., 2006; Harris et al., 2007; Fang et al., 2008). Similarly, soil and/or foliar applications of Zn fertilizers can increase leaf, tuber and fruit Zn concentrations (Shuman, 1998; Rengel et al., 1999; Broadley et al., 2007). In some soils, the residual effects of a single application of Zn fertilizer can be appreciated over several years.

The phytoavailability of Cu in many agricultural soils is low, and Cu applied to the soil often becomes rapidly unavailable to plants (Gupta, 1979). Nevertheless, Cu concentrations in cereals, vegetables and fruits can be increased by Cu fertilization (Gupta, 1979; Sterrett et al., 1983; Shuman, 1998; Rengel et al., 1999; Bunzl et al., 2001; Tamoutsidis et al., 2002). Crops are generally supplied with Cu as a soil application of CuSO4 or as sewage sludges and manures (Gupta, 1979; Shuman, 1998). These amendments improve plant growth on soils with low Cu phytoavailability and increase Cu concentrations in edible tissues. However, the combination of crop variety and Cu fertilization must be managed appropriately to ensure that Cu fertilization is adequate but not excessive, as too much Cu can be toxic to both plants and humans (Gupta, 1979; Shuman, 1998; White & Broadley, 2005a; Puig et al., 2007a). Foliar applications of Cu fertilizers are occasionally recommended under specific circumstances (Gupta, 1979).

Common Ca fertilizers include lime (CaO and CaCO3), gypsum (CaSO4), calcium phosphate and calcium nitrate. The application of Ca fertilizers to the soil generally increases Ca concentrations in tubers and leaves and sometimes, but not always, in fruits and seeds (Shear, 1975; McLaughlin & Wimmer, 1999; Welch, 1999; White, 2001; White & Broadley, 2003; Ho & White, 2005; White et al., in press). It is thought that the application of Ca fertilizers to soils increases fruit and seed Ca concentrations markedly only when these can be supplied with Ca via the xylem, as Ca transport in the phloem is severely restricted. Foliar applications of soluble Ca fertilizers are commonly made to horticultural crops to prevent Ca-deficiency disorders (Shear, 1975; Ho & White, 2005). The liming of soil increases the pH of the soil solution and provides a Ca source in the topsoil, whilst water-soluble gypsum provides Ca throughout the soil profile.

Magnesium is generally supplied to crops as its sulphate (Epsom salts or kieserite), carbonate or, most commonly, oxide (Metson, 1974; Draycott & Allison, 1998). In addition, the use of magnesium ammonium phosphate (struvite) has recently received attention, as it has potential as a sustainable P source for agriculture (Parsons & Smith, 2008). Magnesium fertilizers are frequently applied to the soil surface or, when less soluble, incorporated into the subsoil (Metson, 1974; Draycott & Allison, 1998). Magnesium sulphate provides readily available Mg2+, whereas MgO behaves as a slow-release fertilizer (Metson, 1974; Draycott & Allison, 1998). Foliar applications of MgSO4 are also common on some crops (Metson, 1974; Draycott & Allison, 1998). The application of Mg fertilizers increases Mg concentrations in plant tissues and there is a strong positive relationship between Mg2+ in the soil solution and Mg concentrations in produce (Metson, 1974; Wilkinson et al., 1990; Draycott & Allison, 1998; Oury et al., 2006).

Tissue Se concentrations in plants can be increased by soil or foliar applications of Se fertilizers and this has been shown to have beneficial effects on animal health and deliver Se to the human diet (Gissel-Nielsen, 1998; Combs, 2001; Gupta & Gupta, 2002; Lyons et al., 2003, 2005; Hartikainen, 2005; Broadley et al., 2006b; Hawkesford & Zhao, 2007; Rayman, 2008). The use of inorganic Se fertilizers to increase crop Se concentrations has been particularly successful in both Finland and New Zealand (Eurola et al., 1989, 1991; Lyons et al., 2003; Hartikainen, 2005). For example, since the incorporation of Se into all multielement fertilizers used in Finnish agriculture became mandatory in July 1984, Se concentrations in many indigenous food items have increased over 10-fold (Eurola et al., 1989, 1991, 2004; Ekholm et al., 2007). Both Na2SeO4 and K2SeO4 provide phytoavailable Se for immediate uptake by crops, but the application of selenite or less soluble forms of selenate, such as BaSeO4, provides longer lasting effects (Gissel-Nielsen, 1998; Gupta & Gupta, 2002; Broadley et al., 2006b). Soil applications are generally recommended, especially for crops subject to late-season moisture and heat stress (Lyons et al., 2005), but foliar applications have also been deployed (Fang et al., 2008). One intriguing proposal is to use the Se-rich straw of plants grown on naturally seleniferous soils as a ‘green manure’ in areas with inadequate soil Se concentrations (Terry et al., 2000).

In most soils, I is present in solution as iodide, although iodate can also be present under strongly oxidizing conditions (Fuge & Johnson, 1986). Fertilization with soluble iodide and/or iodate salts has been practised in agriculture, and the iodinization of irrigation water has successfully increased the delivery of I to humans through edible crops (Jiang et al., 1997; Lyons et al., 2004). The I concentrations in root crops and leafy vegetables can be increased greatly by the application of I fertilizers, and, although I is not readily mobile in the phloem, I concentrations in tubers, fruits and seeds can also be increased by I fertilization to nutritionally significant concentrations (Jiang et al., 1997; Rengel et al., 1999; Dai et al., 2004). It has been suggested that, because human dietary I requirements are quite low, I fertilizers might be added to large areas of agricultural production from aeroplanes (Graham et al., 2007).

From the foregoing discussion, it is clear that the application of inorganic fertilizers can undoubtedly increase the concentrations of mineral elements commonly lacking in human diets in edible produce. However, these fertilizers must be applied regularly and can be costly to manufacture, distribute and apply. Furthermore, the manufacture and use of inorganic fertilizers can incur environmental costs, such as those caused by the production of greenhouse gasses and mineral enrichment of the environment. The supply of certain mineral elements may also become limiting in the future. For example, reserves of Zn and Cu are estimated to be 480 and 940 million tonnes (US Geological Survey, 2007), respectively, and it has been estimated that, at their current rate of consumption, the supply of both these elements may be exhausted within 60 yr (Cohen, 2007; Kesler, 2007). Similarly, the world's Se reserves amount to only 170 000 tonnes (US Geological Survey, 2007). If all 210 Mha of the world's wheat was fertilized at 20 g Se ha−1 (Broadley et al., 2006b), this would consume 4322 t Se yr−1, and the Se supply would be exhausted in less than 40 yr. By contrast, the supplies of Fe, Ca, Mg and I are considered to be secure for over 100 yr at their current rate of consumption (Kesler, 2007). The amounts of mineral elements removed by crops, and, therefore, the minimum required in fertilizer applications to maintain soil mineral concentrations can be estimated from their concentrations in harvested portions (USDA-ARS, 2007; http://www.ars.usda.gov/ba/bhnrc/ndl) and global production statistics (FAO, 2006; http://faostat.fao.org/site/291/default.aspx). These data suggest that, in contrast to Se, the amounts of Fe, Zn, Cu, Ca, Mg and I required for the biofortification of edible crops are negligible compared with their global reserves. Hence, the supply of inorganic fertilizers for agriculture could be secure for many centuries, if it were prioritized.

Increasing the acquisition of mineral elements from unfertilized soils

The total mineral concentrations of Fe, Zn and Cu in most soils would be sufficient to support mineral-dense crops, if these elements were phytoavailable (Loneragan, 1997; Shuman, 1998; Schmidt, 1999; Graham et al., 1999; Frossard et al., 2000; Rengel, 2001). Hence, there is considerable interest in developing management systems that exploit soil and fertilizer sources of mineral elements more effectively and in breeding mineral-efficient crops that produce high yields and accumulate minerals from previously infertile soils. In developing countries, breeding for increased yields on infertile soils is a major objective (Lynch, 2007). This work aims to improve both the acquisition of mineral elements and their physiological utilization in the plant for improved yields (Lynch, 2007).

Crop yields in developing countries are restricted principally by drought, low phytoavailability of P and/or nitrogen (N), and soil acidity, which is often associated with Al toxicity and low phytoavailability of Ca, Mg and K (Lynch, 2007; Kirkby & Johnston, 2008). In addition, the phytoavailability of mineral elements, such as Fe, Zn and Cu, limits crop yields on many calcareous soils of the world (see section ‘Phytoavailability of mineral elements’). The acquisition of mineral elements with restricted mobility in the soil, such as P, K, Fe, Zn and Cu, can be improved by investing more biomass in the root system, by producing a greater number and more even spread of roots, by developing a more extensive root system, with longer, thinner roots with more root hairs, and by proliferating lateral roots in mineral-rich patches (White et al., 2005; Lynch, 2007; Kirkby & Johnston, 2008; White & Hammond, 2008). In addition, the efflux of organic acids, which displace cations from their binding sites in the soil, and the secretion of enzymes capable of degrading organic compounds, such as phytate, that chelate cations can also improve the acquisition of Fe, Zn and Cu (Morgan et al., 2005; Lynch, 2007). There is considerable intraspecific genetic variation in root architecture and root exudation that might improve the acquisition of all these elements from unfertilized soils (White et al., 2005; Lynch, 2007; White & Hammond, 2008). Similarly, rotations and intercropping of plants that are better able to access and mobilize mineral elements with low solubility and/or movement in the soil solution can be utilized to increase their tissue concentrations and crop yield (Rengel et al., 1999; Jolley et al., 2004; Graham et al., 2007; Inal et al., 2007). It has been suggested that such plants might include micronutrient-rich, indigenous food crops that acquire mineral elements more effectively from unfertilized soils (Welch & Graham, 2005). The diversification of rotations to include species with greater concentrations of essential mineral elements for human nutrition in their edible tissue also has the potential to increase the delivery of these elements to the human diet independently (Graham et al., 2007).

Soil micro-organisms can also be exploited to increase the volume of soil explored by crop plants and the phytoavailability of mineral elements (Rengel et al., 1999; Barea et al., 2005; Morgan et al., 2005; Lynch, 2007; Kirkby & Johnston, 2008). Many crops are associated with mycorrhizal fungi, which have the potential to increase the volume of soil exploited for the acquisition of immobile mineral elements, and release organic acids, siderophores and enzymes capable of degrading organic compounds (Rengel et al., 1999; Barea et al., 2005; Morgan et al., 2005; Smith & Read, 2007). Recently, He & Nara (2007) suggested that the agricultural management of mycorrhizal fungi could be used to increase mineral concentrations in edible produce, and several studies have found that mycorrhizal associations increase Se, Fe, Zn and Cu concentrations in crop plants (Kothari et al., 1991; Caris et al., 1998; Rengel et al., 1999; Harrier & Watson, 2003; Larsen et al., 2006; Cavagnaro, 2008). However, because the symbiotic relationship between plants and mycorrhizal fungi is fuelled by photosynthate from plants, such associations can reduce yields in well-fertilized soils (Morgan et al., 2005; Lynch, 2007). Relationships with N2-fixing bacteria, whether symbiotic or associative, are especially important in N-limited environments (Rengel et al., 1999; Hardarson & Broughton, 2003). Thus, the deployment of legumes in N-limited environments is essential, but is often, although not always (Houlton et al., 2008), compromised by their high demand for P and other mineral elements for growth (White & Hammond, 2008). Again, this might be addressed by including plants that are better able to access and mobilize mineral elements in rotations and intercropping schemes (Kirkby & Johnston, 2008). Finally, exudates from plant roots and mycorrhizal fungi can provide carbon for other soil microbes that affect the phytoavailability of mineral elements. Hence, inoculants of growth-promoting bacteria can increase the acquisition of Fe, Zn and Cu by plant roots, tissue mineral concentrations, plant growth and yield (Rengel, 2001; Whiting et al., 2001; Barea et al., 2005).

Genetic biofortification strategies

Genetic variation in the concentrations of mineral elements in edible portions of crop plants

Increasing the concentrations of essential mineral elements in produce through the application of mineral fertilizers can be complemented by breeding crops with an increased ability to acquire and accumulate these minerals in their edible portions. However, it must be recognized that genotypic enhancement can only improve the acquisition, utilization or accumulation of mineral elements available to the crop. Considerable genetic variation appears to exist in the concentrations of the mineral elements most frequently lacking in human diets in the edible portions of most crop species.

Data are being amassed on within-species genetic variation in Fe, Zn, Cu, Ca, Mg and Se concentrations in the edible tissues of crop plants (Table 1). However, published data on genetic variation in I concentrations in plant tissues are scarce. Nevertheless, within-species variation in leaf I concentration has been reported for both perennial ryegrass (Lolium perenne) and white clover (Trifolium repens) and shown to be heritable (Butler & Glenday, 1961; Alderman & Jones, 1967). These studies provide strong evidence that leaf I concentration is under genetic control, which provides the rationale to establish the extent of within-species genetic variation in the I concentration of edible portions of crop plants (Lyons et al., 2004).

Most surveys of genetic variation in mineral concentrations have been performed on the carbohydrate-rich portions of staple crops. These include rice (O. sativa), wheat (Triticum spp.) and maize (Z. mays) grains, common bean (P. vulgaris) seeds, sweet potato (Ipomoea batatas) tubers, and cassava (Manihot esculenta) and yam (Dioscorea alata) roots (Grusak & Cakmak, 2005; White & Broadley, 2005a; Pfeiffer & McClafferty, 2007). In addition, genetic variation in the mineral concentrations of the edible portions of other commonly consumed cereals (such as barley (Hordeum vulgare), millet (Panicum miliaceum) and sorghum), legumes (such as pea (Pisum sativum), soybean (Glycine max and Glycine soja), lentil (Lens culinaris), pigeonpea (Cajanus spp.), cowpea (Vigna unguiculata), groundnut (Vigna subterranea), chickpea (Cicer arietinum) and peanut (Arachis hypogaea)), vegetables (such as brassicas (Brassica spp.), spinach, carrot (Daucus carota) and potato (S. tuberosum and Solanum phureja)) and fruits (such as tomato (S. lycopersicum) and bananas/plantains (Musa spp.)) that contribute significantly to human diets has also been surveyed (Grusak & Cakmak, 2005; White & Broadley, 2005a; Johns & Eyzaguirre, 2007; Pfeiffer & McClafferty, 2007). These data suggest that there is sufficient genetic potential within cultivated species to manipulate the tissue concentrations of all the mineral elements lacking in human diets through breeding programmes.

There is genetic variation in the concentrations of mineral elements in the grains of most cereal species. Concentrations of Fe and Zn in cereal grain vary 1.5- to 4-fold among genotypes depending on the genetic diversity of the material tested (Table 1; see References S11). Some of this variation can be attributed to differences in grain yield (see section ‘Historical trends in the concentrations of mineral elements in edible tissues’). Although the cultivated germplasm of some cereals, such as wheat, may have a limited genetic variation in grain Fe and Zn concentrations, wild relatives often possess considerable variation (Monasterio & Graham, 2000; Cakmak et al., 2004; Welch et al., 2005; Chhuneja et al., 2006; Pfeiffer & McClafferty, 2007; Cakmak, 2008; Tiwari et al., 2008) and accessions with grain Fe and Zn concentrations at least 2-fold higher than the most widely grown varieties are available for many cereal species (White & Broadley, 2005a). Differences in grain Zn and Fe concentrations between species of wild and cultivated wheat have been attributed, in part, to allelic variation at a chromosomal locus that promotes early senescence and remobilization of protein, Fe, Zn and Mn from senescing leaves to seeds, and the introgression of the high grain protein content (Gpc-B1) locus from a wild tetraploid wheat (Triticum turgidum ssp. dicoccoides) to a cultivated wheat (Triticum durum) resulted in higher concentrations of Fe and Zn in its grain (Distelfeld et al., 2007). In addition, highly significant positive correlations between grain Fe and Zn concentrations have been observed in maize (Maziya-Dixon et al., 2000; Long et al., 2004), wheat (Graham et al., 1999; Garvin et al., 2006), sorghum (Reddy et al., 2005) and pearl millet (Pennisetum glaucum; Velu et al., 2007), which increase the possibilities of breeding for increased concentrations of both Fe and Zn simultaneously. Interestingly, studies of related Triticum species showed strong positive correlations between grain protein and Zn concentrations (Ozturk et al., 2006; Distelfeld et al., 2007). Chromosomal loci affecting grain Fe and Zn have been mapped in rice (Gregorio et al., 2000) and wheat (cited by Ortiz-Monasterio et al., 2007). Grain Ca has been found to vary almost 3-fold and grain Mg c. 1.5-fold among wheat accessions (Peterson et al., 1986; Graham et al., 1999; Bálint et al., 2001; Oury et al., 2006) and genetic variation in grain Cu concentrations has been reported for wheat (McGrath, 1985; Peterson et al., 1986; Bálint et al., 2001; Garvin et al., 2006) and pearl millet (Abdalla et al., 1998). Genetic variation in grain Se has also been reported for wheat (Lyons et al., 2005; Garvin et al., 2006) and oat (Avena sativa; Eurola et al., 2004), although the expression of this trait is strongly dependent upon weather conditions, crop husbandry and selenium fertilization (Eurola et al., 2004; Lyons et al., 2005; Garvin et al., 2006). Although it is often suggested that little variation exists in grain Se concentrations amongst modern bread and durum wheats, the wild wheats (Triticum dicoccum and Triticum spelta) and their relatives (Aegilops tauschii) have significantly higher Se concentrations than cultivated wheat and can be integrated into breeding programmes (Graham et al., 2001; Lyons et al., 2003, 2005; White & Broadley, 2005a; White et al., 2007b).

Because the seeds of many cereals are often consumed after milling or polishing, it is pertinent to consider whether genetic variation in the distribution of mineral elements within the seed can be utilized in biofortification strategies. Mineral elements are nonhomogenously distributed within the seed and the concentrations of many mineral elements are highest in the husk and/or aleurone layers (see References S12). Milling or polishing cereal seeds can, therefore, remove large quantities of mineral elements from the diet and the extent of these losses is genotype dependent (Gregorio et al., 2000; Vasconcelos et al., 2003; Ma et al., 2004; Lyons et al., 2005; Prom-u-thai et al., 2007). The partitioning of mineral elements such as Fe and Zn within cereal grains is affected by aspects of grain morphology, such as grain size, embryo size, and the number and thickness of the tissue layers (Welch et al., 1993; Oikeh et al., 2003a; Cakmak et al., 2004). For example, grain Fe, Zn, Ca, Mg, and Cu concentrations in maize, rice and barley are related to the number of aleurone cell layers, which is cultivar dependent (Welch et al., 1993; Prom-u-thai et al., 2007).

Seeds of legumes, such as common bean, pea, lentil, soybean, mungbean (Vigna radiata), chickpea, peanut and groundnut, generally have higher concentrations of Fe, Zn, Ca and Mg than cereal grains, and significant within-species genetic variation in the concentrations of all these elements has been observed (Table 1; White & Broadley, 2005a; Pfeiffer & McClafferty, 2007). Seed Fe and Zn concentrations have been found to vary from 1.4- to 6.6-fold among genotypes of legume species grown together in the same environment (Table 1). Some of this variation is associated with seed morphology, as the tissue distribution of Zn and Fe in legume seeds is under genetic control (Moraghan et al., 2002; Ariza-Nieto et al., 2007). Domestication does not appear to have affected the mean concentration or range of Fe or Zn in bean seeds (Beebe et al., 2000). Seed Ca and Mg concentrations also vary considerably among genotypes of legume species grown in the same environment. For example, seed Ca concentrations varied 9-fold among 120 segregating F2,3s from a wide cross between a wild and a cultivated P. vulgaris genotype (Guzmán-Maldonado et al., 2003), and seed Ca and Mg concentrations ranged from 0.03 to 0.26% dry weight (DW) and from 0.11 to 0.25% DW, respectively, among 481 P. sativum accessions (Grusak & Cakmak, 2005). Seed Ca concentrations varied up to 2.6-fold among 10 chickpea genotypes (Abbo et al., 2000) and 2-fold among soybean cultivars (Raboy et al., 1984; Horner et al., 2005). Similarly, seed Cu concentrations show appreciable within-species genetic variation (Deosthale, 1981; Branch & Gaines, 1983; Grusak & Cakmak, 2005; Ariza-Nieto et al., 2007; Zia-Ul-Haq et al., 2007). Little information is available on within-species variation in seed Se or I concentrations in legumes, although variation in Se accumulation among soybean genotypes has been documented (Yang et al., 2003; Zhang Y et al., 2003b). Seed Fe, Zn and Ca concentrations all behave as quantitative traits in legume species and genetic loci influencing them can be mapped using QTL analysis (Beebe et al., 2000; Guzmán-Maldonado et al., 2003; Cichy et al., 2005; Gelin et al., 2007) and introgressed into commercial germplasm (Islam et al., 2004). It is thought that most genetic variation in Zn concentration in common bean is controlled by a single locus (Guzmán-Maldonado et al., 2003; Cichy et al., 2005; Gelin et al., 2007). The positive correlations among the concentrations of Zn and Fe, and Ca and Mg in beans (Beebe et al., 2000; House et al., 2002; Guzmán-Maldonado et al., 2003; Gelin et al., 2007) suggest the possibility of breeding for increased concentrations of these elements simultaneously.

The concentrations of some mineral elements, such as Ca, Mg, Fe and Zn, are often greater in leafy vegetables than in grain, seed, fruit or tuber crops (Table 1). In a core collection of B. oleracea thought to incorporate approx. 99% of the common allelic polymorphisms in this species, leaf Fe varied 45-fold and leaf Zn varied 26-fold between accessions in one glasshouse trial (M. R. Broadley, J. P. Hammond and P. J. White, unpublished observations). In the same glasshouse trial, leaf Fe varied 4.8-fold and leaf Zn varied 3.6-fold among 74 commercial varieties representing all subtaxa of B. oleracea. Among 22 accessions of B. oleracea var. acephala (kale/collards), leaf Fe varied 1.6-fold and leaf Zn varied 1.8-fold, and the rankings of cultivars were consistent between years despite substantial year-to-year variation in leaf Fe and Zn concentrations (Kopsell et al., 2004b). Among 327 accessions of spinach, leaf Fe varied 2.7-fold, and leaf Zn varied 12-fold (Grusak & Cakmak, 2005). Genetic variation in shoot Fe and Zn concentrations has also been observed in Brassica rapa (Wu et al., 2007, 2008) and onion (Allium cepa; Alvarez et al., 2003). Shoot Ca and Mg concentrations varied up to 3-fold among accessions of B. oleracea (Farnham et al., 2000; Rosa et al., 2002; Kopsell et al., 2004b; Broadley et al., 2008), B. rapa (Wu et al., 2008), spinach (Grusak & Cakmak, 2005), onion (Alvarez et al., 2003) and chickpea (Ibrikci et al., 2003), and the concentrations of these elements in shoot tissues appear to be correlated (White & Broadley, 2005a; Broadley et al., 2008; Wu et al., 2008). Several QTLs affecting shoot Ca and Mg concentrations have been found in a cross between a rapid-cycling B. oleracea var. alboglabra and an F1 B. oleracea var. italica (Broadley et al., 2008), and a QTL affecting shoot Mg concentration has been found in B. rapa (Wu et al., 2008), but no candidate genes have been identified yet. Within-species genetic variation in shoot Mg concentrations has also been reported for forage of Jerusalem artichoke (Helianthus tuberosus; Seiler & Campbell, 2006) and herbage of forage grasses, such as crested wheatgrass (Agropyron cristatum), Italian ryegrass (Lolium multiflorum), orchard grass, perennial ryegrass (Lolium perenne), reed canary grass (Phalaris arundinacea), Russian wildrye (Psathyrostachys juncea) and tall fescue (Festuca arundinacea), which have been studied because of the negative economic consequences of grass tetany in ruminant animals when their diets contain insufficient Mg (Hides & Thomas, 1981; Sleper et al., 1989; Smith et al., 1999; Jefferson et al., 2001). Leaf Cu concentrations vary many fold among genotypes of B. oleracea (M. R. Broadley, P. J. White and J. P. Hammond, unpublished observations), spinach (Grusak & Cakmak, 2005), chickpea (Ibrikci et al., 2003) and onion (Alvarez et al., 2003). Genetic variation in shoot Se concentrations has been observed in B. oleracea (Kopsell & Randle, 2001; Farnham et al., 2007) and onions (Kopsell & Randle, 1997).

The concentrations of some elements, such as Fe, Ca and Mg, are also generally higher in root crops than in cereal grains, legume seeds, fruits or tubers (Table 1). There is considerable within-species genetic variation in the concentrations of these elements in root crops. For example, there is 51-fold variation in root Cu, 38-fold variation in root Fe, 14-fold variation in root Zn, 8-fold variation in root Ca, and 4.6-fold variation in root Mg among 600 cassava genotypes (Chávez et al., 2005). Considerable genetic variation in root Fe and Zn concentrations has also been observed in sweet potato (Pfeiffer & McClafferty, 2007; Courtney et al., 2008) and carrot (Nicolle et al., 2004b), and in root Ca and Mg concentrations in carrot (Nicolle et al., 2004b) and sugar beet (Beta vulgaris) (Baierová & Baier, 1993).

Fruits generally have low concentrations of mineral elements that are less mobile in the phloem, and the concentrations of mineral elements essential to human nutrition do not appear to vary greatly among cultivars. For example, fruit Fe, Zn, Ca and Mg concentrations differed less than 2-fold between six strawberry (Fragaria spp.) cultivars (Hakala et al., 2003), as did Fe and Zn concentrations in fruits of three apple (Malus domestica) varieties (Iwane, 1991) and Fe, Zn, Mg and Cu concentrations in representative fingers of three plantain varieties (Davey et al., 2007). However, mean fruit Ca concentrations differed 2.5-fold among these plantain varieties (Davey et al., 2007) and fruit Cu concentrations varied 2.5-fold among the apple varieties (Iwane, 1991). Similarly, there was 2.3-fold variation in Ca concentrations and 8-fold variation in Fe concentrations among 11 plum (Prunus domestica) varieties (Nergiz & Yildiz, 1997). Some genetic variation in the Se concentrations in tomato has also been reported (Shennan et al., 1990, Pezzarossa et al., 1999).

Limited within-species genetic variation in the concentrations of mineral elements has also been observed in tubers (Table 1; Pfeiffer & McClafferty, 2007; White et al., in press). In general, tuber concentrations of mineral elements differ 2- to 3-fold among genotypes, although greater variation in tuber Fe concentrations is occasionally observed (Agbor-Egbe & Trèche, 1995; White et al., in press). Genetic variation in tuber Fe, Zn, Ca, Mg and Cu concentrations has been observed in yams (Agbor-Egbe & Trèche, 1995; Pfeiffer & McClafferty, 2007), oca (Sangketkit et al., 2001) and cultivated potato species (Pfeiffer & McClafferty, 2007; White et al., in press and references therein). As tubers acquire most of their mineral elements via the phloem (Westermann, 2005), the concentrations of Ca, Mg and Zn in tubers are often lower than in other edible tissues (Table 1).

Genetic variation in the concentrations of promoters and antinutrients in the edible portions of crop plants

The bioavailability of essential mineral elements depends greatly on the presence in a meal of substances that promote or inhibit their absorption by the gut (Frossard et al., 2000; Reddy, 2002; Hotz & Brown, 2004; Welch & Graham, 2004; White & Broadley, 2005a; Slingerland et al., 2006; Bohn et al., 2008). The main substances known to inhibit the absorption of Fe, Zn, Ca and Mg are phytate from cereal grains and legume seeds and polyphenolics from beverages such as tea and coffee, beans and sorghum. The absorption of Ca is inhibited additionally by oxalate present in certain fruit and vegetables (Franceschi & Nakata, 2005; Titchenal & Dobbs, 2007). It is estimated that only c. 5% of the Fe and 25% of the Zn present in legume and cereal seeds is bioavailable (Pfeiffer & McClafferty, 2007). The bioavailability of Fe is reduced at dietary phytate/Fe molar quotients greater than 1 and the bioavailability of Zn is reduced when the phytate/Zn molar quotient exceeds about 6 (Lönnerdal, 2002; Hurrell, 2003). Substances enhancing the absorption of Fe and Zn include ascorbic acid and β-carotene from fruits and vegetables, whilst cysteine-rich polypeptides from plant and animal sources promote the absorption of Fe, Zn and Cu. Inulin, a polysaccharide of fructose often with a terminal glucose unit that is present in significant amounts in a wide range of edible crops, promotes the absorption of both Ca and Mg (Roberfroid, 2005, 2007).

Phytate concentrations vary considerably among cereal, legume and vegetable crops (Holland et al., 1991, 1992; Frossard et al., 2000; Reddy, 2002; Hotz & Brown, 2004; Vreugdenhil et al., 2005). Significant within-species genetic variation has been found for grain phytate concentration in rice (Welch et al., 2000, Glahn et al., 2002), wheat (Lolas et al., 1976; Raboy et al., 1991; Erdal et al., 2002; Welch et al., 2005), barley (Dai et al., 2007), pearl millet (Abdalla et al., 1998), oat (Lolas et al., 1976), triticale (Feil & Fossati, 1997) and sorghum (Reddy et al., 2005; Slingerland et al., 2006). Similarly, seed phytate concentrations vary among genotypes of common bean (Lolas & Markakis, 1975; Coelho et al., 2002; House et al., 2002; Cichy et al., 2005; Ariza-Nieto et al., 2007) and soybean (Lolas et al., 1976; Raboy et al., 1984; Horner et al., 2005) and among accessions of B. rapa (Zhao et al., 2007, 2008). In addition, natural and induced mutants with similar total P concentrations to conventional varieties, but reduced seed phytate concentrations, named low phytic acid (lpa) mutants, have been described in wheat, barley, maize, rice and soybean (reviewed by Raboy, 2003, 2007; Bohn et al., 2008). Although many of these mutants have reduced rates of germination and agronomic yield, this is not always the case (Ertl et al., 1998, Pilu et al., 2003; Hulke et al., 2004; Oltmans et al., 2005; Bregitzer & Raboy, 2006; Guttieri et al., 2006a; Raboy, 2007). Two phenotypes of lpa mutants are commonly observed. Seeds of one phenotype do not accumulate greater concentrations of inositol phosphates (IPs), whereas seeds of the other phenotype accumulate IP3, IP4 and IP5 (Raboy, 2003, 2007). The lpa mutations do not appear to have much effect on the concentrations or distributions of mineral elements within the seed (Hatzack et al., 2000; Guttieri et al., 2004, 2006b; Liu et al., 2004a, 2007; Ockenden et al., 2004; Bryant et al., 2005; Joyce et al., 2005; Lin et al., 2005), but replacing conventional varieties with lpa mutants in diets improves the mineral nutrition of monogastric animals, including humans (Adams et al., 2002; Mendoza et al., 1998, 2001; Hambidge et al., 2004, 2005; Mazariegos et al., 2006; see reviews by Mendoza, 2002; Raboy, 2007). In humans, this beneficial effect is most apparent when the dietary consumption of minerals is low. Several QTLs affecting the phytate concentration in seeds of rice (Stangoulis et al., 2007) and common bean (Cichy et al., 2005) and in seeds and leaves of B. rapa (Zhao et al., 2007, 2008) have been identified for use as molecular markers in breeding programmes. Weak correlations between phytate and Fe, Zn, Ca or Mg concentrations in sorghum grain (Reddy et al., 2005), soybean seeds (Raboy et al., 1984) and common bean seeds (Cichy et al., 2005), and a lack of coincidence of QTLs affecting seed phytate and those affecting Fe or Zn concentrations (Vreugdenhil et al., 2004; Cichy et al., 2005; Stangoulis et al., 2007; Waters & Grusak, 2008b) suggest that these traits are controlled by different genes and that it would be possible to breed for Fe-, Zn-, Ca- and/or Mg-dense edible portions with low phytate concentrations.

It has long been known that P and phytate concentrations in cereal grains and legume seeds are increased greatly by the application of P fertilizers and reduced by the application of Zn fertilizers (Reddy et al., 1989; Ryan et al., 2008). Thus, judicious applications of inorganic fertilizers could complement any genetic approaches to reduce dietary phytate intake and increase dietary Zn/phytate quotients.

Genetic variation in the concentrations of polyphenolics in seeds of sorghum (Dicko et al., 2002) and common bean (Guzmán-Maldonado et al., 2000, 2003; House et al., 2002) has also been reported. However, because polyphenolic compounds differ in their abilities to bind Fe (Brune et al., 1989; Hurrell et al., 1999), and many polyphenols have been shown to be beneficial to human health (Scalbert et al., 2005), it will be necessary to reduce the concentrations of only those polyphenols that bind Fe most avidly to increase Fe bioavailability.

Cereal grains, vegetables and fruits vary widely in their oxalate concentrations (USDA, 1984; Libert & Franceschi, 1987; Holland et al., 1992; Kim et al., 2007; Massey, 2007). Most angiosperms deposit Ca-oxalate crystals in cell vacuoles or, occasionally, the cell wall, although members of some commelinoid and noncommelinoid monocot families, including the Poaceae, Liliaceae and Zingiberaceae, appear to lack Ca-oxalate crystals (Prychid & Rudall, 1999; Franceschi & Nakata, 2005). The shapes and tissue distributions of oxalate crystals differ among plant species. Plants that can accumulate large quantities of oxalate (> 5% DM) in their edible tissues include members of the Caryophyllales (e.g. amaranth, beet/chard, purslane (Portulaca oleracea), spinach, tetragona (Tetragonia tetragonioides) and rhubarb), Araceae (e.g. taro (Colocasia esculenta)), and Oxalidaceae (e.g. carambola and oca). Seeds of some legumes, and roots of carrot and cassava can also accumulate high concentrations of oxalate on occasion (USDA, 1984; Massey, 2007). In general, oxalate concentrations are far greater in leaves than in roots or fruits (Libert & Franceschi, 1987). In some species, such as amaranth, beet/chard and spinach, much of the oxalate is insoluble, whereas in other species, such as oca, sweet pepper (Capsicum annuum), eggplant (Solanum melongena) and carrot, the oxalate is soluble (Libert & Franceschi, 1987; Albihn & Savage, 2001; White & Broadley, 2003; Franceschi & Nakata, 2005; White, 2005; Kim et al., 2007). In plants that precipitate Ca-oxalate, the accumulation of oxalate is directly proportional to tissue Ca concentration and is, therefore, strongly dependent upon Ca phytoavailability and plant growth rate (Libert & Franceschi, 1987; Kinzel & Lechner, 1992; Franceschi & Nakata, 2005). Some within-species genetic variation in oxalate concentrations has been observed in beet (Libert & Franceschi, 1987), spinach (Kitchen et al., 1964; Libert & Franceschi, 1987; Kawazu et al., 2003; Mou, 2008), rhubarb (Libert & Creed, 1985; Libert, 1987; Libert & Franceschi, 1987), carambola (Wilson et al., 1982), oca (Ross et al., 1999; Albihn & Savage, 2001; Sangketkit et al., 2001), taro (Tanaka et al., 2003) and soybean (Massey et al., 2001; Horner et al., 2005), and several mutants have been identified in the forage legume Medicago truncatula with reduced or altered accumulation of Ca-oxalate in their leaves (Nakata & McConn, 2000; McConn & Nakata, 2002). For two of the M. truncatula calcium oxalate deficient mutants (cod5 and cod6), reduced accumulation of Ca-oxalate has been correlated with increased Ca bioavailability in herbage (Nakata & McConn, 2006, 2007; Morris et al., 2007).

The concentrations of β-carotene and ascorbic acid vary among plant tissues and among plant species (Bhaskarachary et al., 1995, 2007; Frossard et al., 2000; FSA, 2002; USDA-ARS, 2007). Fruits and vegetables are important dietary sources of ascorbic acid and β-carotene. Significant within-species genetic variation in β-carotene concentration has been found in edible portions of a number of crop species, including cassava (Chávez et al., 2000, 2005; Maziya-Dixon et al., 2000; Thakkar et al., 2007), carrot (Nicolle et al., 2004b; Baranska et al., 2006), lettuce (Lactuca sativa; Nicolle et al., 2004a; Mou, 2005), B. oleracea (Schonhof & Krumbeln, 1996; Kurilich et al., 1999; Kopsell et al., 2004a), sweet potato (Nestel et al., 2006), potato (Nesterenko & Sink, 2003; Morris et al., 2004), sweet pepper (Simonne et al., 1997; Howard et al., 2000), tomato (Fraser & Bramley, 2004), chickpea (Abbo et al., 2005), immature soybean (Simonne et al., 2000), common bean, rice, wheat (Welch & Graham, 2005; Howitt & Pogson, 2006), sorghum (Kapoor & Naik, 1970; Reddy et al., 2005), pearl millet (Kapoor & Naik, 1970), maize (Kurilich & Juvik, 1999; Maziya-Dixon et al., 2000; Hulshof et al., 2007; Ortiz-Monasterio et al., 2007; Harjes et al., 2008) and plantains (Davey et al., 2007). In addition, natural mutations in either LeDET1 or LeDDB1 have been found to lead to increased concentrations of β-carotene and ascorbic acid in tomato fruit (Liu et al., 2004b; Bino et al., 2005), a spontaneous, semidominant mutation of the orange (Or) gene encoding a plastid-associated DnaJ-related protein (Lu et al., 2006) has been found to lead to high concentrations of β-carotene in cauliflower (Brassica oleracea var. botrytis) curd (Li et al., 2001a), and natural mutations in the promoter of the yellow1 (Y1) gene encoding a phytoene synthase have been found to be associated with greater concentrations of β-carotene in the endosperm of maize (Palaisa et al., 2003). Both cauliflower Or hybrids and maize with yellow endosperm are available commercially.

There is also significant genetic variation in the concentration of ascorbic acid in edible tissues of many crop species including carrot (Nicolle et al., 2004b), cassava (Chávez et al., 2000), lettuce (Nicolle et al., 2004a), B. oleracea (Kurilich et al., 1999), sweet pepper (Simonne et al., 1997; Howard et al., 2000), strawberry (Hakala et al., 2003) and banana (Wall, 2006). In addition to genotypic effects, ascorbate concentrations in plant tissues are affected by other factors such as developmental stage, time of day, and growth environment (Smirnoff et al., 2001; Ishikawa et al., 2006; Smith et al., 2007). Ascorbate concentrations increase in response to diverse environmental stresses and, in particular, high light intensities. This is thought to be related to the role of ascorbate as a key antioxidant in plants (Smirnoff et al., 2001; Noctor, 2006).

Transgenic approaches to biofortification

Transgenic approaches to biofortification rely on improving the phytoavailability of mineral elements in the soil, their uptake from the rhizosphere, translocation to the shoot and accumulation in edible tissues (White & Broadley, 2005a; Davies, 2007; Puig et al., 2007a; Zhu et al., 2007). In addition, transgenic approaches may be used to reduce the concentrations of antinutrients and increase the concentrations of promoter substances.

Increasing concentrations of mineral elements

Iron acquisition from the soil can be improved in strategy I plants by overexpressing genes encoding Fe(III) reductases (Samuelsen et al., 1998; Grusak, 2000; Rogers & Guerinot, 2002; Vasconcelos et al., 2006) and Fe2+ transporters of the root plasma membrane (Ramesh et al., 2004; Grotz & Guerinot, 2006), and in strategy II plants by increasing the synthesis and exudation of phytosiderophores (Takahashi et al., 2001; Douchkov et al., 2005), together with increased expression of genes encoding YSL proteins. However, it has been observed that, although the overexpression of FRO genes is sufficient to increase Fe accumulation in leaves, increased biosynthesis of Fe-chelates in the shoot is often required to increase Fe concentrations in seeds, fruits and tubers, which rely on Fe supplied via the phloem (Grusak, 2000; Rogers & Guerinot, 2002; White & Broadley, 2005a). Nevertheless, increasing the capacity of edible tissues to sequester Fe can promote their accumulation of Fe, possibly through feedback mechanisms impacting on plant Fe homeostasis. Thus, altering the activity of vacuolar transporters, such as NRAMPs and VIT1, in seeds can increase their Fe concentrations (Lanquar et al., 2005; Kim et al., 2006b; Puig et al., 2007a), and expressing plant ferritin or human lactoferrin genes has increased Fe, Zn and Cu concentrations in seeds of rice (Goto et al., 1999; Lucca et al., 2001, 2002; Nandi et al., 2002; Krishnan et al., 2003; Vasconcelos et al., 2003; Qu et al., 2005; Sivaprakash et al., 2006), and Fe concentrations in maize seeds (Drakakaki et al., 2005), tomato fruits and potato tubers (Chong & Langridge, 2000). The overexpression of plant ferritin genes has also been reported to increase Fe concentrations in lettuce leaves (Goto et al., 2000).

The misexpression of genes affecting Zn and Cu uptake and movement within the plant can increase the concentration of these elements in edible portions. For example, pea bronze (brz) and degenerated leaflet (dgl) and Arabidopsis ferric reductase defective 3/manganese accumulator 1 (frd3=man1) mutants constitutively expressing rhizosphere Fe(III) reductase activity have greater shoot Zn and Cu concentrations than wild-type plants (Delhaize, 1996; Grusak, 2000; Rogers & Guerinot, 2002), and transgenic barley plants expressing AtZIP1 produce smaller seeds with higher Zn concentrations than wild-type plants (Ramesh et al., 2004). It has also been suggested that, because AtCAX1 can be modified to transport Zn2+ into the vacuole, modified CAX transporters could also be used to increase Zn concentrations in the edible tissues of transgenic plants (Shigaki et al., 2005). Zinc concentrations in Arabidopsis leaves can be increased by overexpressing genes encoding AtHMA4 (Verret et al., 2004) or AtMTP3 (Arrivault et al., 2006), and by reducing the expression of AtHMA2 (Eren & Argüello, 2004) or AtOPT3 (Stacey et al., 2008). Seeds of the opt3-2 mutant have higher Zn and Cu concentrations than seeds of wild-type plants (Stacey et al., 2008). Seed Fe and Zn concentrations can be increased in wheat by expressing RNA interference (RNAi) constructs of an NAC transcription factor (NAM-B1) that accelerates senescence and increases remobilization of mineral elements from leaves to developing grain (Uauy et al., 2006). The accumulation of Cu in plant tissues can be increased by the overexpression of genes encoding Cu transporters of the HMA family and/or Cu-binding metallotheioneins (Puig et al., 2007a). Similarly, Indian mustard (B. juncea) genetically engineered to produce more glutathione, phytochelatins and total thiols had greater concentrations of Zn and Cu than nontransformed controls, when grown on soils containing high concentrations of these elements (Bennett et al., 2003). It has been observed that manipulation of the concentrations of plant growth regulators can also increase the accumulation of Zn and Cu. For example, transgenic tomato plants expressing bacterial 1-aminocyclopropane-1-carboxylate (ACC) deaminase, which produce less ethylene, have greater concentrations of Zn and Cu in their leaves than nontransgenic plants (Grichko et al., 2000).

Interestingly, the misexpression of neither genes encoding Ca2+-permeable channels of the plasma membrane nor genes encoding any Ca2+-ATPases appears to increase shoot Ca concentrations (Kim et al., 2001; Hampton et al., 2005; White, 2005; Kaplan et al., 2007). However, plants overexpressing genes encoding vacuolar Ca2+/H+ antiporters have greater shoot Ca concentrations than wild-type plants (Hirschi, 2001; Hirschi et al., 2001) and the expression of genes encoding AtCAX1 lacking its autoinhibitory domain (sCAX1), a modified AtCAX2 (sCAX2) or AtCAX4 increases Ca concentrations and, potentially, dietary Ca delivery, in edible portions of transgenic carrot (Park et al., 2004; Morris et al., 2008), lettuce (Park et al., 2009), tomato (Park et al., 2005a) and potato (Park et al., 2005b; Kim et al., 2006a). Remarkably, Arabidopsis mutants lacking AtSKOR, a Ca2+-permeable outward-rectifying K+ channel expressed in the pericycle and xylem parenchyma of the root, also have greater shoot Ca concentrations than wild-type plants, which is consistent with AtSKOR removing Ca2+ from the xylem sap (Gaymard et al., 1998). Shoots of transgenic plants overexpressing calreticulin, the main Ca2+-binding protein in the endoplasmic reticulum, have higher Ca concentrations than wild-type plants (Wyatt et al., 2002) and it has been speculated that the overexpression of vacuolar Ca2+-binding proteins might also increase Ca concentrations in edible produce (White, 2005).

Selenium concentrations in produce depend upon the ability of crop plants to take up Se, redistribute it within the plant, and accumulate it in nontoxic forms in edible tissues. Differences between Se-accumulator and nonaccumulator plants in their Se uptake capacity and tissue Se/S quotients appear to be related to the selectivity of Se/S transporters in the plasma membrane of root cells. It has, therefore, been speculated that appropriate allelic variation in the domain(s) conferring selenate/sulphate selectivity in HASTs, combined with the constitutive expression of Se-selective HASTs, could be used to produce crops with increased Se concentrations and tissue Se/S quotients (Terry et al., 2000; White et al., 2004, 2007a,b; Sors et al., 2005b; Broadley et al., 2006b). Differences in the ability of plant species to tolerate high tissue Se concentrations are thought to be a consequence of differences in their Se metabolism, and, in particular, the production of nontoxic, sink metabolites (Sors et al., 2005a,b). Consistent with this hypothesis, the leaves of transgenic plants overexpressing genes encoding SeCys methyltransferase and/or ATP sulphurylase have greater Se tolerance, and higher concentrations of methylcysteine, selenomethylcysteine, and total Se, than wild-type plants (Pilon-Smits et al., 1999; Montes-Bayón et al., 2002; Ellis et al., 2004; LeDuc et al., 2004, 2006; Van Huysen et al., 2004; Bañuelos et al., 2005; Sors et al., 2005a). Similarly, transgenic plants expressing a mouse SeCys lyase or the chloroplast protein CpNifS, which both catalyse the conversion of selenocysteine to alanine and elemental Se, also have greater Se tolerance and higher leaf Se concentrations than wild-type plants (Garifullina et al., 2003; Pilon et al., 2003; Van Hoewyk et al., 2005).

The bioavailability and benefits to human health of dietary Se depend upon not only the amounts but also the chemical forms of Se supplied (Combs, 2001; Rayman, 2004; Finlay, 2007). The range between Se deficiency and Se toxicity in human diets is narrow (Combs, 2001; Rayman, 2004; White & Broadley, 2005a), but supplying Se in forms with greater bioavailability, such as SeMet and SeCys, will most improve the Se status of populations with low dietary Se intakes (Rayman, 2004; Thomson, 2004). The dominant organic form of Se in cereals is SeMet (Broadley et al., 2006b; Hawkesford & Zhao, 2007). By contrast, members of the Brassica and Allium genera contain high concentrations of SeMSeCys (Broadley et al., 2006b; Finlay, 2007). This compound may be effective at reducing the incidence of various cancers and has a low potential for Se toxicity (Whanger, 2004; Finlay, 2007). Thus, increasing SeMet, SeCys and/or SeMSeCys are attractive targets for transgenic strategies to improve Se bioavailability from edible crops.

Manipulating concentrations of antinutrients and promoter substances

Transgenic approaches to increase the bioavailability of Fe, Zn and Ca in food have focused on reducing the concentrations of antinutrients, such as oxalate, polyphenolics and phytate, and increasing the concentrations of promoter substances, such as ascorbate, β-carotene and cysteine-rich polypeptides, in edible produce (Lönnerdal, 2003; White & Broadley, 2005a; Kopsell & Kopsell, 2006; Lucca et al., 2006; Davies, 2007; Zhu et al., 2007).

Two transgenic strategies have been adopted to reduce phytate concentrations in edible produce (Lönnerdal, 2003; Raboy, 2003, 2007). The first strategy is to reduce the expression of genes encoding enzymes involved in the synthesis or sequestration of IP6, which has successfully reduced phytate concentrations in seeds of maize, soybean and rice (Kuwano et al., 2006; Nunes et al., 2006; Shi et al., 2005, 2007). The second strategy is to overexpress phytases in edible tissues. This strategy has proved successful in reducing endogenous phytate concentrations in seeds and/or increasing phytase activity in feed from rice (Lucca et al., 2001, 2002; Hong et al., 2004), wheat (Brinch-Pedersen et al., 2006), maize (Drakakaki et al., 2005; Chen et al., 2008), soybean (Chiera et al., 2004; Bilyeu et al., 2008), alfalfa (Medicago sativa; Flachowski et al., 2005) and canola (Brassica napus; Ponstein et al., 2002).

Several crop plants have been genetically modified to contain greater concentrations of β-carotene, the precursor of vitamin A, in their edible tissues. These include golden rice (Al-Babili & Beyer, 2005; Paine et al., 2005) and maize (Aluru et al., 2008), golden canola (Ravanello et al., 2003), orange cauliflower (Lu et al., 2006), tomato (Solanum lycopersicum; Fraser & Bramley, 2004; Davuluri et al., 2005; Taylor & Ramsay, 2005; Botella-Pavía & Rodríguez-Concepción, 2006; Kopsell & Kopsell, 2006; Davies, 2007; Wurbs et al., 2007; Zhu et al., 2007) and yellow potatoes (Solanum tuberosum and S. phureja; Ducreux et al., 2005; Lu et al., 2006; Morris et al., 2006; Diretto et al., 2007; Lopez et al., 2008). The strategy has been either to overexpress carotenogenic transgenes from nuclear or plastid genomes or to alter the expression of genes controlling plastid development. Similarly, genetic modification (GM) approaches have been used to increase ascorbate concentrations in lettuce leaves (Jain & Nessler, 2000) and maize kernels (Chen et al., 2003). Attempts to increase protein cysteine concentrations in edible tissues have also proved successful on occasion (Chakraborty et al., 2000; Lucca et al., 2001, 2002; Sun & Liu, 2004; Welch & Graham, 2004), but not always (Tabe & Droux, 2002; Hagan et al., 2003; Zhang et al., 2003a; Chiaiese et al., 2004). Although genes encoding proteins with abundant methionine and cysteine residues are readily expressed in transgenic plants, and protein methionine concentrations can be increased greatly, a concomitant decrease in the expression of genes encoding endogenous cysteine-rich proteins results in an apparent inability to raise protein cysteine concentrations (Tabe & Droux, 2002; Chiaiese et al., 2004; Ufaz & Galili, 2008). It has been suggested that this might be addressed by the application of S fertilizers, as the down-regulation of the endogenous genes is a characteristic response to S starvation.

Is biofortification of edible produce a solution to the ‘hidden hunger’?

The term ‘hidden hunger’ has been used to describe the micronutrient malnutrition inherent in human diets that are adequate in calories but lack vitamins and/or mineral elements. The diets of a large proportion of the world's population are deficient in Fe, Zn, Ca, Mg, Cu, Se or I, which affects human health and longevity, and therefore national economies. Mineral malnutrition can be addressed by increasing the amount of fish and animal products in diets, mineral supplementation, food fortification and/or increasing the bioavailability of mineral elements in edible crops. However, as observed in the introduction, strategies to increase dietary diversification, mineral supplementation and food fortification have not always proved successful. For this reason, the biofortification of crops through the application of mineral fertilizers, combined with breeding varieties with an increased ability to acquire mineral elements, has been advocated. To determine whether biofortification strategies can address mineral malnutrition of humans, decision-makers have posed six key questions (Bouis, 2000; Bouis et al., 2000; Nestel et al., 2006).

  • • Is breeding (combined with appropriate agronomy) for high nutrient content scientifically feasible?
  • • Will farmers adopt the new genotypes?
  • • What is the target nutrient content for breeding?
  • • What is the impact on nutritional status?
  • • Is it cost-effective?
  • • Will consumers accept the biofortified foods?

It should be clear from the foregoing sections that breeding for edible produce containing higher concentrations of all the mineral elements most often lacking in human diets is possible without affecting yields. However, it is also apparent that, although many soils contain ample mineral elements, plant breeding must be combined with appropriate agronomy and the application of mineral fertilizers when the phytoavailability of mineral elements restricts plant growth or concentrations of mineral elements in edible portions. Although the use of mineral fertilizers is evidently feasible in the developed world, as exemplified by the success of Se fertilization of crops in Finland (Lyons et al., 2003; Hartikainen, 2005; Broadley et al., 2006b), Zn fertilization in Turkey (Cakmak, 2004) and I fertilization in China (Jiang et al., 1997), the distribution of mineral fertilizers requires appropriate social infrastructures, stable political policies and continued investment, which has stymied previous attempts at mineral supplementation and food fortification in developing countries. It is likely that farmers in both developed and developing countries will adopt new genotypes that acquire mineral elements more efficiently, particularly if biofortified produce demands a premium price and crops can be grown on soils with low phytoavailability of mineral elements with reduced fertilizer inputs, better germination, seedling vigour and resistance to pathogens (Rengel & Graham, 1995; Yilmaz et al., 1998; Welch, 1999; Bouis et al., 2000; Genc et al., 2000; Cakmak, 2004, 2008; Graham et al., 2007).

The target concentration for a specific mineral element in the edible portion of a biofortified crop will be determined by the amount of that element required in the human diet, the deficit of the mineral element in the diet of an affected population, the number of crops that will be biofortified, the bioavailability of the mineral element following processing and cooking, and the contributions of each biofortified crop to the diet of the affected population. Thus, strategies for addressing mineral malnutrition through biofortification and, therefore, target concentrations of mineral elements in edible produce will depend greatly upon local diet and culinary customs. When more than one mineral element is lacking in the diet, biofortification strategies must deliver all of them to the affected population. However, when a mineral or vitamin deficiency is induced by the lack of another mineral or vitamin, as occurs among Fe, Zn and provitamin A carotenoid deficiencies (Hess et al., 2005) and between Se and I deficiencies (Lyons et al., 2004), it can be corrected by the biofortification of edible crops with the appropriate mineral and/or vitamin that is lacking in the diet.

In developing countries, it has been suggested that biofortification strategies should focus on the staple foods that dominate people's diets (Bouis, 2000; Pfeiffer & McClafferty, 2007). The argument is simple: if the concentrations of mineral elements in staple foods can be increased, then the delivery of mineral elements to vulnerable populations can be increased pro rata to their contribution to the diet, without a change in behaviour (Bouis, 1999; Bouis et al., 2000; Graham et al., 2007). Target staples include rice and wheat, which are staple foods for over half the world's population, maize, which is the staple food in much of sub-Saharan Africa and in Mesoamerica, common bean, which supplies significant amounts of minerals to populations in Africa and Latin America, and cassava, which is the main source of dietary carbohydrates in sub-Saharan Africa. Theoretical studies indicate that a strategy based on the biofortification of staple crops should increase the delivery of mineral elements to human diets and dramatically improve the nutritional status of vulnerable populations in developing countries (Bouis et al., 2000). In this context, a doubling of the Fe and Zn concentrations in cereal grains and legume seeds would be an appropriate and achievable target. The HarvestPlus consortium has suggested an absolute target for the additional Zn and Fe in biofortified crops of between 30 and 40% of the estimated average dietary requirement for humans (Holtz, 2007). Low-phytate crops could also be used to increase the bioavailability of Fe and Zn in plant foods to achieve these targets (Pfeiffer & McClafferty, 2007), but there is much debate about this strategy because high dietary phytate has been linked to various health benefits (Vucenik & Shamsuddin, 2006). Similarly, although reducing the concentrations of specific polyphenolic compounds could increase the bioavailability of Fe in the diet, it could also compromise the beneficial effects of these compounds on human health (Scalbert et al., 2005).

The impact of biofortified produce on the nutritional status of humans has rarely been tested. However, it is evident that the application of mineral fertilizers containing Se, I or Zn can have a significant impact on the nutritional status of a vulnerable population (Jiang et al., 1997; Cakmak, 2004; Rayman, 2008). In addition, it was found that the consumption of Fe-biofortified rice improved the Fe status of nonanaemic Filipino women (Haas et al., 2005) and that replacing conventional varieties with lpa mutants in people's diets improved their Fe, Zn and Ca status, especially when consumption of dietary minerals was low (Mendoza et al., 1998, 2001; Adams et al., 2002; Hambidge et al., 2004, 2005; Mazariegos et al., 2006). These data suggest that the biofortification of edible produce can improve the nutritional status of humans.

Biofortification of edible produce through genetic strategies is potentially cost effective and will deliver most benefits to the 40% of the world's population who rely primarily on their own food for sustenance. It has been suggested that a one-time financial investment in seeds of cereal staples that acquire mineral elements more effectively from the soil could support the Fe, Zn, Ca and Mg requirements of rural populations in remote areas. Similarly, a one-time financial investment would suffice for vegetatively propagated crops, such as cassava, potato and banana (Johns & Eyzaguirre, 2007). Most economic analyses suggest that genetic strategies towards biofortification are more cost effective than dietary diversification, supplementation or food fortification programmes (Bouis, 1999; Bouis et al., 2000; Horton, 2006; Stein et al., 2007; Ma et al., 2008). Early economic analyses for Zn biofortification of wheat in Turkey suggested a cost-to-benefit quotient of greater than 20 over two decades (Bouis, 1999), and cost-to-benefit quotients of between 20 and 30 for Fe biofortification of rice in South Asia and for Fe biofortification of rice and wheat in Bangladesh and India over the same period (Bouis et al., 2003). Informal estimates of cost-to-benefit quotients for fertilization with Se and/or I also suggest high returns on financial investments (Lyons et al., 2005; Horton, 2006). More recently, the potential impact of biofortification has been quantified as the saving of disability-adjusted life years (DALYs; Stein et al., 2005). It has been estimated that the annual burden of Fe-deficiency anaemia in India is 4 million lost DALYs and that Fe biofortification may reduce this burden significantly. Similarly, it is estimated that the annual burden of Zn deficiency in India is 2.8 million lost DALYs and Zn biofortification of rice and wheat may reduce this burden by 20–51% (Stein et al., 2007). The cost of saving 1 DALY from Zn biofortification of rice and wheat in India was estimated as $US 0.73–7.31 (Stein et al., 2007). The cost of saving 1 DALY from the biofortification of beans/pearl millet/potatoes with Fe and Zn in Africa has been estimated as $US 2–20.

It is thought that consumers in both developed and developing countries will accept foods prepared from biofortified crops provided that they are not appreciably more expensive than the alternatives and that biofortification does not alter the appearance, taste, texture or cooking quality of foods (Bouis et al., 2003). It is thought unlikely that small quantities of mineral elements will alter these properties of foods, but manipulating the concentrations of promoters and antinutrients might affect both taste and colour. If it can be demonstrated that foods prepared using biofortified produce are more beneficial to human health, this will, of course, influence consumer choice in both developed and developing countries.

In conclusion, biofortification strategies based on crop breeding, targeted genetic manipulation and/or the application of mineral fertilizers hold great potential for addressing mineral malnutrition in humans. The questions posed by decision-makers have been answered positively, and international initiatives, such as the HarvestPlus programme, have begun to deliver crops with the potential to increase both the amounts and bioavailability of essential mineral elements in human diets.

Acknowledgements

Work at SCRI was funded by the Scottish Government Rural and Environment Research and Analysis Directorate (RERAD). We thank all our colleagues and friends who commented on the manuscript, especially Dr Tim George and Dr Gavin Ramsay, and we apologize to all the authors whose work has not been cited because of space constraint or oversight. This paper is dedicated to the memory of Dr Mike J. Earnshaw, a valued friend and mentor.