The extent of the literature on the essential trace element Se appears to have increased exponentially over the last decade reflecting the tremendous growth of interest in this nutrient since it was shown by Clark et al. to reduce cancer risk in their landmark trial(Reference Clark, Combs and Turnbull1). Though the form of Se used in that trial was high-Se yeast, when large-scale funding was obtained from the National Cancer Institute for a follow-up randomised trial of the effect of supplemental Se on prostate cancer risk (SELECT), the decision was taken to use selenomethionine (SeMet) owing to the perceived importance of being able to define the specific form of Se that might be associated with an important health effect(Reference Lippman, Goodman and Klein2). Thus we are no longer satisfied with knowing simply the amount of Se that may be associated with benefit but seek to know the species of Se to which that alleged benefit may be attributed. Furthermore, we have come to realise that different species of an element (for example, arsenic) can have very different health effects. The present review therefore attempts to pull together what is known about the species of Se in foods and supplements, the pathways by which they are synthesised, their apparent bioavailability as found in different food sources as this has implications for Se requirements, and the health effects that can be ascribed to specific Se species.
Biosynthesis and metabolism of dietary selenium species
A consideration of Se speciation in plant and animal food sources requires some understanding of the biosynthetic pathways involved in Se assimilation by plants and how these species are metabolised in animals. Such knowledge enables us to predict to some extent the Se species likely to be contained in foods. The biosynthetic pathways for Se in plants, some of which are assumed by analogy with S pathways, are shown in Fig. 1 (adapted from Ellis & Salt(Reference Ellis and Salt3), Whanger(Reference Whanger and Cheeke4, Reference Whanger5), Terry et al. (Reference Terry, Zayed, De Souza and Tarun6), Tagmount et al. (Reference Tagmount, Berken and Terry7) and Sors et al. (Reference Sors, Ellis and Salt8)). The relative dominance of the pathways differs for Se-accumulators and non-accumulators.
Biosynthetic pathways elucidated for Se in higher plants (some by analogy with S pathways) (adapted from Ellis & Salt(Reference Ellis and Salt3), Whanger(Reference Whanger and Cheeke4, Reference Whanger5), Terry et al. (Reference Terry, Zayed, De Souza and Tarun6), Tagmount et al. (Reference Tagmount, Berken and Terry7) and Sors et al. (Reference Sors, Ellis and Salt8)). It should be noted that reactions vary from species to species so that compounds formed and their relative quantities differ between species and strains. APSe, adenosine-5′-phosphoselenate; 
, glutathione-S-selenite; GSH, glutathione; DMDSe, dimethyl-diselenide (volatile); GSSeSG, selenodiglutathione; MeSeH, methyl selenol; GSSeH, glutathione-selenopersulfide; MeSeCysSeO, Se-methyl-selenocysteine selenoxide; GSSe− , glutathione-conjugated selenide; γ-GMeSeCys, γ-glutamyl-Se-methyl-selenocysteine; MeSeCys, Se-methyl-selenocysteine; SeCys MeTransferase, selenocysteine methyltransferase; SeCys, selenocysteine; DMSe, dimethyl selenide (volatile); DMSeP, dimethyl-selenonio-propionate (CH3Se+(CH3)2CH2CH2COO− ); MeSeMet+, Se-methyl-selenomethionine; SeCysth, selenocystathionine; γ-GSeCysth, γ-glutamyl-selenocystathionine; SAM, S-adenosyl methionine; Se-ASeMet, Se-adenosyl-selenomethionine; SeMet, selenomethionine; MeTHF, methyl-tetrahydrofolate; SeHomocysteine, selenohomocysteine.
The major species in plant sources of Se are: selenate (translocated directly from the soil and less readily bound to soil components than selenite); SeMet (biosynthesised) and a smaller amount of selenocysteine (SeCys; biosynthesised); Se-containing proteins (where SeMet and SeCys have been incorporated non-specifically in place of methionine and cysteine); Se-methyl-selenocysteine and γ-glutamyl-Se-methyl-selenocysteine (considered as detoxification products, notably formed in Se-accumulators and plants of the Brassica and Allium families). Plants can volatilise significant amounts of Se as dimethylselenide (non-accumulators) and dimethyldiselenide (accumulators)(Reference Terry, Zayed, De Souza and Tarun6). To avoid an over-complicated Fig. 1, the enzymes implicated in these pathways are not shown, with the exception of SeCys methyltransferase, the enzyme notably present in Se-accumulators and responsible for the methylation of SeCys to the characteristic methylated metabolites that are believed to have anti-cancer properties.
While a study of these pathways suggests Se species that may be expected in foods from plant sources, it should be noted that compounds formed and their relative quantities differ not only between Se-accumulators and non-accumulators but also between species.
There is much less information on the species of Se in dietary sources of animal origin(Reference Whanger9). When inorganic Se is given to animals, SeCys is the main seleno-compound formed but when animals eat Se-containing foods of plant origin, protein-bound SeMet will also be formed from the non-specific incorporation of plant-derived SeMet in place of methionine. Selenotrisulfide, glutathione selenopersulfide and metallic selenides have also been reported in tissues(Reference Burk10). The presence of some of these compounds can be explained by the metabolic pathway of dietary Se in animals which resembles that in humans as described below.
Most of what we know about the metabolism of dietary (or supplement) Se in humans is inferred from studies in rats and mice. A simplified version of the metabolic pathway is shown in Fig. 2 and clearly illustrates the central role of hydrogen selenide (H2Se) (adapted from Combs(Reference Combs11) and Rayman(Reference Rayman12))(Reference Suzuki, Doi and Suzuki13, Reference Suzuki, Somekawa and Suzuki14). SeMet catabolised from proteins can be trans-selenated to SeCys (by analogy with the trans-sulfuration pathway). SeCys, either from this source or directly from the diet, is then converted to H2Se by SeCys β-lyase. Alternatively, SeMet can undergo α,γ-elimination catalysed by a γ-lyase to yield CH3SeH, though the relative importance of this route in humans is not known(Reference Suzuki, Doi and Suzuki13, Reference Okuno, Kubota, Kuroda, Ueno and Nakamuro15, Reference Okuno, Ueno and Nakamuro16). CH3SeH is also produced by a β-lyase from plant sources containing Se-methyl-selenocysteine and γ-glutamyl-Se-methyl-selenocysteine. Utilisation of selenate or selenite (plant sources or supplements) for selenoprotein synthesis, or excretion of excess, first requires reduction to the central Se metabolite, H2Se, via interaction with the tripeptide, glutathione. The H2Se so formed may be converted to selenophosphate (
) which then reacts with tRNA-bound serinyl residues to give SeCys-bound tRNA from which SeCys is inserted co-translationally, at loci encoded by specific UGA codons, to give selenoproteins(Reference Berry, Banu, Chen, Mandel, Kieffer, Harney and Larsen17, Reference Berry, Banu, Harney and Larsen18). As CH3SeH can be demethylated to H2Se in an equilibrium reaction, both it and its precursors can also act as Se sources for selenoprotein synthesis(Reference Suzuki, Doi and Suzuki13). Oxidation of excess H2Se can lead to the production of superoxide and other reactive oxygen species with associated toxic effects(Reference Combs11).
Metabolic pathway of dietary Se in humans (adapted from Combs(Reference Combs11) and Rayman(Reference Rayman12))(Reference Suzuki, Doi and Suzuki13, Reference Suzuki, Somekawa and Suzuki14). SeMet, selenomethionine; SeCys, selenocysteine; GSSeSG, selenodiglutathione; γ-glutamyl-CH3SeCys, γ-glutamyl-Se-methyl-selenocysteine; H2Se, hydrogen selenide; 
, selenophosphate; CH3SeCys, Se-methyl-selenocysteine; CH3SeH, methyl selenol; (CH3)2Se, dimethyl selenide; SeO2, selenium dioxide; (CH3)3Se+, trimethyl selenonium ion.
Surplus Se is transformed to methylated metabolites mostly for excretion into urine. Excretion of Se is either from H2Se through a methylated selenosugar (1β-methylseleno-N-acetyl-d-galactosamine) in urine or by further methylation of CH3SeH to dimethyl selenide ((CH3)2Se) which is exhaled in breath, and trimethyl selenonium ion ((CH3)3Se+) excreted in urine(Reference Francesconi and Pannier19–Reference Kuehnelt, Kienzl, Traar, Le, Francesconi and Ochi21). Though 1β-methylseleno-N-acetyl-d-galactosamine is the most significant urinary metabolite in most individuals, (CH3)3Se+ is a major product from Se-methyl-selenocysteine(Reference Suzuki, Doi and Suzuki13, Reference Kuehnelt, Kienzl, Traar, Le, Francesconi and Ochi21, Reference Kuehnelt, Juresa, Kienzl and Francesconi22).
Selenium in food sources and dietary supplements: speciation and concentration
Table 1 shows the Se species apparently identified in foods and dietary supplements and their concentrations or relative concentrations in some cases(Reference Whanger and Cheeke4, Reference Whanger5, Reference Whanger9, Reference Burk10, Reference Rayman12, Reference Kotrebai, Birringer, Tyson, Block and Uden23–Reference Diaz Huerta, Fernandez Sanchez and Sanz-Medel72) (H Goenaga Infante, G O'Connor and MP Rayman, unpublished results). In terms of identification, it must be borne in mind that many of these studies were carried out when the available analytical strategies that combined both elemental and molecular MS were less well developed than is currently the case. In the case of most foods, however, they are the only data we have and can help focus the direction of further studies. Column 5 shows the methodology used for Se species identification. Readers should be aware, however, that identification that is only based on retention-time matching with authentic standards by HPLC–inductively coupled plasma MS is tentative and that electrospray ionisation MS data alone do not provide enough evidence of structural confirmation. To obtain this, fragmentation of the molecular ion has to be performed(Reference Goenaga Infante, Hearn and Catterick28). Table 1 contains some speciation data that have been obtained in this way, for example, by inductively coupled plasma MS combined with MS/MS data obtained by matrix-assisted laser desorption/ionisation (MALDI) or electrospray ionisation MS with fragmentation of the precursor/molecular ion (electrospray ionisation MS/MS)(Reference Goenaga Infante, O'Connor, Rayman, Wahlen, Spalholz, Norris, Hearn and Catterick27, Reference Encinar, Ruzik, Buchmann, Tortajada, Lobinski and Szpunar30–Reference Goenaga Infante, O'Connor, Rayman, Hearn and Cook33, Reference Warburton and Goenaga Infante37, Reference Dernovics, Giusti and Lobinski47, Reference Dernovics, García-Barrera, Bierla, Preud'homme and Lobinski48, Reference Ogra, Ishiwata, Ruiz-Encinar, Lobinski and Suzuki56–Reference Shah, Kannamkumarath, Wuilloud, Wuilloud and Caruso58, Reference Dumont, Ogra, Vanhaecke, Suzuki and Cornelis61, Reference McSheehy, Yang, Pannier, Szpunar, Lobinski, Auger and Potin-Gautier62). Those wishing to understand more about speciation-analysis methodology are referred to critical reviews of recent analytical developments for the Se speciation analysis of foods, supplements and biosamples(Reference Goenaga Infante, Hearn and Catterick28, Reference Dumont, Vanhaecke and Cornelis73).
Species and concentrations of selenium in foods and supplements (concentrations of selenium species are expressed in terms of concentration of elemental selenium) and the methodology used for species identification
ESI-MS, electrospray ionisation MS; ESI-MS/MS, electrospray ionisation MS with fragmentation of the precursor/molecular ion.
*For ramps, sample from 2nd year of growth.
Most quantitative data in Table 1 have been calculated from the peak area for a particular Se species expressed as a percentage of the total area of eluted Se peaks. However, accurate measurements by isotope-dilution MS or standard additions are also reported for methylated Se compounds such as SeMet and γ-glutamyl-Se-methyl-selenocysteine(Reference Larsen, Hansen, Paulin, Moesgaard, Reid and Rayman25, Reference Goenaga Infante, O'Connor, Rayman, Wahlen, Spalholz, Norris, Hearn and Catterick27, Reference Mester, Willie and Lu29, Reference Diaz Huerta, Fernandez Sanchez and Sanz-Medel72). Ideally, full mass balance data (i.e. total Se, total extracted Se, Se species, sum of species, extraction efficiency) should be considered together with recovery results from spiking experiments or analysis of ‘speciated’ certified reference materials for validation of speciation methodologies.
The total Se concentration has been reported in Table 1 where possible, as it can affect the distribution of Se between species, as in the case of Se-enriched garlic and yeast(Reference Kotrebai, Birringer, Tyson, Block and Uden23). As the concentration of Se in Se-enriched foods is considerably higher that in the corresponding natural foods, such foods must be treated with caution, though the amounts in which they are eaten (for example, garlic) may reduce the risk of toxicity.
It is noteworthy that while wheat, other grains and soya contain predominantly SeMet with lesser amounts of SeCys and selenate, the major seleno-amino acids found in Allium and Broccoli species are Se-methyl-selenocysteine and γ-glutamyl-Se-methyl-selenocysteine. The latter two compounds are characteristic of the Se species produced by Se-accumulator plants which avoid the toxic effects of incorporation of excessive amounts of SeCys and SeMet into their proteins by accumulating non-protein seleno-amino acids or their γ-glutamyl derivatives(Reference Terry, Zayed, De Souza and Tarun6). Other non-protein seleno-amino acids that have been identified in Se-accumulator plants are selenocystathionine, Se-methyl-selenomethionine, γ-glutamyl-selenocystathionine, selenopeptides and selenohomocysteine(Reference Whanger9), though, of these, only selenocystathionine has been fully identified in foods (Table 1).
Given that Brazil nuts are potentially the richest food source of Se, and the tree that produces them, Bertholletia excelsa, is regarded as an Se-accumulator, it might be expected that the major Se species would be Se-methyl-selenocysteine or γ-glutamyl-Se-methyl-selenocysteine, as described above. Instead the major species in Brazil nuts appears to be SeMet(Reference Palmer, Herr and Nelson44–Reference Kannamkumarath, Wrobel, Wrobel, Vonderheide and Caruso46). This may to some extent be an illustration of the differences in concentration and speciation found between different plant tissues, Brazil nuts being seeds rather than fleshy leaves or florets as in the case of garlic or broccoli(Reference Ellis and Salt3, Reference Terry, Zayed, De Souza and Tarun6). However, it may also be due to more general differences in Se metabolism between plant species (Dr Martin Broadley (2007), personal communication).
Considerably less information is available on Se species in animal foods than is available for plant foods. Although the Se content of fish and other seafoods has been reviewed by Reilly(Reference Reilly74), normally ranging from 0·1 to 1·0 μg/g fresh weight, there is little information on specific Se species in fish. Several studies have found that seafood Se appears to be less bioavailable than that from other dietary sources, the implication being that the molecular form of at least some of the fish Se is such that it is not utilisable for selenoprotein synthesis(Reference Barceloux40, Reference Huang, Akesson, Svensson, Schutz, Burk and Skerfving75, Reference Meltzer, Bibow, Paulsen, Mundal, Norheim and Holm76). Though it has been suggested that an explanation for this lower bioavailability may be interaction with Hg in seafood, the molar concentration of Se exceeds that of Hg by one or two orders of magnitude except in the case of sea mammals (cetaceans), suggesting that this is an unlikely explanation(Reference Endo, Haraguchi, Hotta, Hisamichi, Lavery, Dalebout and Baker77–Reference Bates, Prentice, Birch and Delves79). While Se and Hg undoubtedly have very high affinity for one another(Reference Dyrssen and Wedborg80), there are as yet no published data identifying Se–Hg species in seafood. However, according to Dr Nick Ralston (2007, personal communication) it appears that inorganic HgSe is present in the muscle meat of blue marlin as has already been shown in organs of mammals(Reference Wageman, Trebacz, Boila and Lockhart81). SeMet was the only compound identified in fish samples of high Se content in a speciation study(Reference Cabanero, Carvalho, Madrid, Batoreu and Camara67) though other studies found from 4 to 47 % of total fish Se in the form of selenate(Reference Cappon and Smith68–Reference Cappon and Smith70). This is an area ripe for further speciation studies.
Recently, new Se-containing glutathione species, S-selenomethyl-glutathione and glutathione-S-selenoglutathione have been identified in aqueous extracts of Se-yeast(Reference Goenaga Infante, O'Connor, Rayman, Hearn and Cook33). As shown in Fig. 1, bonding of Se to glutathione via a non-enzymic reaction occurs in metabolism at the point where selenite enters the pathway to SeCys(Reference Terry, Zayed, De Souza and Tarun6). Alternatively, as glutathione is a tripeptide of γ-glutamine, cysteine and glycine, it seems possible that the formation of these Se-containing glutathione species could result from the incorporation of SeCys (or methylated SeCys) in place of cysteine in the biosynthetic pathway to glutathione.
While on the subject of Se-yeast, we should make it clear that it is not a defined form of Se. There is considerable variability in products described as Se-yeast which is reflected in the species composition. Se-yeast is produced by fermenting yeast in an Se-enriched medium when the Se becomes organically bound to yeast components. With reputable manufacturers, the percentage of Se that is organically bound should be greater than 90 % and more than 80 % should be bound to yeast proteins, including cell-wall proteins(Reference Rayman12). However, in some products, the percentage of sodium selenite is such that most of the Se is clearly not bound to the yeast; at worst, there may merely be a mixture of sodium selenite and yeast, the Se not being bound to the yeast(Reference Uden, Totoe Boakye, Kahakachchi, Hafezi, Nolibos, Block, Johnson and Tyson24). Such products dupe the consumer, as they do not conform to the normal understanding of Se-yeast as containing Se in an organic form. While they may be capable of increasing the production of selenoproteins, they will be less good at increasing plasma Se and acting as a storage form of Se in the body (see below), thereby maintaining Se status(Reference Burk, Norsworthy, Hill, Motley and Byrne82).
Selenium in food sources and dietary supplements: bioavailability
Bioavailability of a nutrient is conventionally defined as that fraction of ingested nutrient that is utilised for normal physiological functions(Reference Fox, Van den Heuvel and Atherton83); absorption and retention of the nutrient are taken as indirect measures of bioavailability as these are measurable(Reference Fox, Van den Heuvel and Atherton83) though they cannot address functional bioavailability which is that most likely to be relevant to health.
Absorption of Se is not homeostatically regulated and is not believed to be affected by nutritional status. Absorption of dietary Se is generally believed to be good – about 80 % from food(Reference Reilly74). Guar gum is thought to reduce its absorption in humans(Reference Fairweather-Tait84), as is high dietary sulfur, probably because of competition between chemically similar sulfur and Se species(Reference Reilly74, Reference Combs, Chichester and Schweiger85). Absorption of SeMet is active and uses the same enzyme transport system as does methionine(Reference Reilly74). Absorption and retention of a commercially produced Se-yeast, in which 66 % of the Se present was in the form of SeMet (SelenoPrecise™), were measured as 90 and 75 % respectively (see Rayman(Reference Rayman12)(Reference Sloth, Larsen, Bugel and Moesgaard86).
A number of supplementation studies have compared the bioavailability of different forms of Se to humans, i.e. Se-rich wheat, Se-enriched yeast, SeMet, sodium selenate and sodium selenite (for a review, see Rayman(Reference Rayman12)). Organic forms of Se (wheat Se, SeMet and high-Se-yeast) were found to be more bioavailable than selenate and selenite in that they were more effective in raising blood Se concentrations (suggesting better absorption and retention), though all forms were able to increase selenoenzyme (glutathione peroxidase) activity. This difference is undoubtedly due to the ability of SeMet from digested organic Se sources to be incorporated in place of methionine into tissue proteins such as skeletal muscle, erythrocytes and plasma albumin where it can act as a Se store though it becomes available to the body only upon turnover of tissue proteins(Reference Schrauzer87). Organic Se (Se-yeast) was also more effective than inorganic forms in its ability to transfer Se to breast-fed infants or suckling animals, thereby reducing the risk of deficiency in the offspring(Reference Rayman12). Foods that contain high proportions of SeMet, such as Brazil nuts and wheat, are good bioavailable sources of the element(Reference Ip and Lisk88, 89). Though the Se content of mushrooms is higher than that of most other vegetables(Reference Reilly74), its bioavailability is said to be very low(Reference Fordyce, Selinus, Alloway, Centeno, Finkelman, Fuge, Lindh and Smedley90). However, our own recent work on Se-enriched mushrooms shows SeMet to be the major Se species and bioavailability to be good(Reference Rayman, Angus and Goenaga-Infante57). A speciation effect may be responsible for the bioavailability of Se from fish being inconsistent(Reference Rayman91); one study has shown a daily intake of 115 μg Se from fish to be unable to increase Se status(Reference Meltzer, Bibow, Paulsen, Mundal, Norheim and Holm76).
There is good evidence that the increased Se status attained after supplementation with organic forms of Se is retained for a longer period after supplementation has ceased than is the case with selenite or selenate(Reference Rayman12). Reported whole-body half-lives of SeMet and selenite in humans were 252 and 102 d respectively, implying that Se administered as SeMet is retained 2·5 times longer in the body than is selenite(Reference Combs, Chichester and Schweiger85). Accordingly, foods or supplements containing SeMet can maintain the activities of selenoenzymes during Se depletion for longer periods of time than those containing inorganic Se owing to the recycling of SeMet catabolised from protein stores(Reference Combs, Chichester and Schweiger85).
No bioavailability data exist for Se-methyl-selenocysteine or γ-glutamyl-Se-methyl-selenocysteine.
Health effects associated with specific selenium species in foods and supplements
While the nutritionally essential functions of Se are understood to be fulfilled by the selenoproteins, dietary Se can be metabolised to small-molecular-weight species that have more recently generated interest because of putative anti-cancer effects. In contrast to such beneficial effects, at a sufficiently high dose level, Se metabolites can also cause toxicity.
Species-related beneficial effects
Though supplementation with Se or a good Se intake or status has been associated with health benefits, there is little or no evidence to connect such benefits with particular Se species. We know from studies in transgenic mice that selenoproteins are important for the cancer-protective effects of Se(Reference Irons, Carlson, Hatfield and Davis92) and it seems likely that antioxidant selenoproteins may be of benefit in counteracting diseases of oxidative stress. However, selenoproteins can be synthesised more or less efficiently from many different Se species, though if consumed in foods, they are digested and must be resynthesised as shown in Fig. 2.
In mice with genetically impaired selenoprotein expression, the presence of low-molecular-weight selenocompounds has been shown to reduce colon cancer risk(Reference Irons, Carlson, Hatfield and Davis92). Such low-molecular-weight selenocompounds may be an in vivo source of the methylated metabolite, CH3SeH, which is believed to be responsible for the potent anti-carcinogenic and anti-angiogenic effects of Se shown in the rat mammary tumour model and in cells in culture(Reference Whanger5, Reference Ip, Birringer, Block, Kotrebai, Tyson, Uden and Lisk60, Reference Ip93–Reference Spallholz, Palace and Reid97). As shown in Fig. 2 and explained above, CH3SeH can be formed directly from the low-molecular-weight selenocompounds Se-methyl-selenocysteine, by the action of a β-lyase(Reference Combs11), and SeMet by the action of a γ-lyase, also known as methioninase(Reference Suzuki, Doi and Suzuki13, Reference Okuno, Kubota, Kuroda, Ueno and Nakamuro15, Reference Okuno, Ueno and Nakamuro16, Reference Spallholz, Palace and Reid97–Reference Wang, Jiang and Lu99).
Se-methyl-selenocysteine and its γ-glutamyl-derivative are found in a number of edible plants, including garlic, onions and broccoli and others of the Allium and Brassica families, particularly when grown in Se-enriched conditions(Reference Whanger5, Reference Kotrebai, Birringer, Tyson, Block and Uden23, Reference Ip, Birringer, Block, Kotrebai, Tyson, Uden and Lisk60). Se-enriched plants such as broccoli and garlic have been shown to have potent anti-tumour effects in animals that are attributed to the presence of these species(Reference Ip, Birringer, Block, Kotrebai, Tyson, Uden and Lisk60, Reference Davis, Finley and Watson96). Though these species have not yet been tested in human interventions, a number of groups are planning pharmacokinetic studies as a prelude to human trials (Dr C Ip (2006), personal communication). Small amounts of both Se-methyl-selenocysteine and γ-glutamyl-Se-methyl-selenocysteine have also been identified in Se-yeast which may possibly be relevant to the anti-cancer effects seen in human trials with Se-yeast(Reference Goenaga Infante, O'Connor, Rayman, Wahlen, Entwisle, Norris, Hearn and Catterick26, Reference Goenaga Infante, O'Connor, Rayman, Wahlen, Spalholz, Norris, Hearn and Catterick27). Se-methyl-selenocysteine has been commercially available for some time and can be bought over the counter as a supplement.
Though there is as yet no evidence of it, it appears possible that Se analogues of anti-cancer sulfur compounds such as diallyldisulfide and ajoene may also be isolable from Se-enriched garlic or onions. As diallylselenide was found to be more than 300 times more effective than diallylsulfide in protecting against carcinogen-induced mammary adenocarcinoma in rats(Reference Spallholz, Palace and Reid97), attempts to find such species may be worthwhile.
Species-related toxic effects
More is known about species-related toxic effects of Se than about species-related beneficial effects. The toxicity of Se and the mechanisms by which it exerts its toxic effects depend on its form, though there are few species-specific data on the toxicity of Se in humans and none relating to dose nor safe upper limits of particular species.
It is likely that a number of different mechanisms are involved in Se toxicity. According to Spallholz et al. (Reference Spallholz, Palace and Reid97, Reference Nakamuro, Nakanishi, Okuno, Hasegawa and Sayato98), Se compounds that can easily form the anion, RSe− , generate superoxide in the presence of thiols such as glutathione, resulting in redox cycling, cell-cycle arrest and apoptosis. Spallholz ascribes the toxic (and indeed the carcinostatic) effects of Se to this oxidative-stress mechanism. Superoxide has been shown to be generated from selenite and diselenides such as selenocystamine in the presence of reduced glutathione in vitro, though not from selenate, SeMet or Se-methyl-selenocysteine(Reference Spallholz, Palace and Reid97). Neither SeMet nor Se-methyl-selenocysteine is very toxic to cells in culture nor to animals or humans in line with their inability to generate superoxide, although both are capable of conversion to CH3SeH by enzymic systems either in vitro or in vivo (Reference Spallholz, Palace and Reid97).
Selenodiglutathione, an intermediate in the formation of superoxide from selenite and glutathione, has been found to be even more toxic than selenite itself(Reference Nakamuro, Nakanishi, Okuno, Hasegawa and Sayato98, Reference Wang, Jiang and Lu99). However, in contradiction to Spallholz's belief, Harrison et al. (Reference Harrison, Lanfear, Wu, Fleming, McGarry and Blower100) did not find that the growth inhibition observed with this compound resulted from induction of an oxidative-stress mechanism, at least not of the type observed with oxidants such as H2O2. Supporting an oxidative-stress mechanism, selenite-induced redox cycles have been suggested to be responsible for oxygen-dependent DNA fragmentation in Se toxicity to hepatocyte model systems(Reference Garberg, Stahl, Warholm and Hogberg101) and high levels of selenite have been shown to induce the formation of 8-hydroxy-2-deoxyguanosine in rat liver DNA(Reference Wycherly, Moak and Christensen102).
Other suggested mechanisms of Se toxicity include inhibition of Se methylation, the major detoxification pathway for Se, allowing the accumulation of hepato-toxic selenides, notably H2Se. For instance, in mice, high doses of SeCys have been shown to cause hepatic toxicity by depressing Se methylation through the inactivation of methionine adenosyltransferase, the enzyme responsible for S-adenosyl methionine synthesis(Reference Hasegawa, Mihara, Nakamuro and Sayato103).
Although it has been suggested that organic forms of Se may be more toxic than inorganic forms during long-term consumption as they can be incorporated into tissue proteins rather than be excreted rapidly(Reference Patterson and Levander104), there is no evidence that this is the case(Reference Barceloux40). Long-term supplementation studies with Se-yeast (60–80 % of which is SeMet) at doses of 200, 300, 400 and even 800 μg Se/d for lengthy periods (up to 12 years in the case of the 200 μg dose) have been carried out by a number of research groups without any indication of toxic effects (for references, see Rayman(Reference Rayman12)). Furthermore men with prostate cancer tolerated doses of 1600 and 3200 μg Se/d, as Se-yeast, for almost 12 months ‘without any obvious Se-related serious toxicity’(Reference Reid, Stratton, Lillico, Fakih, Natarajan, Clark and Marshall105). Thus these results imply that uncontrolled accumulation of tissue Se does not occur.
Though there is no direct evidence in humans, it is generally accepted on the basis of animal studies that inorganic forms of Se are more acutely toxic than organic forms, selenite being slightly more toxic than selenate(Reference Barceloux40). Though of equivalent toxicity to SeCys in animals, sodium selenite is considerably more acutely toxic than SeMet, dimethyl selenide, trimethyl selenonium ion, selenoethers, selenobetaine or Se-yeast, the major Se component of which is SeMet(Reference Barceloux40). From lethal dose 50 % (LD50) determinations, selenite was found to be four-fold more toxic than SeMet when administered to mice intravenously(Reference Ammar and Couri106) and three-fold more toxic than Se-yeast when given orally to rats(Reference Vinson, Bose, Combs, Spallholz, Levande and Oldfield107).
Chronic toxicity of SeCys is equivalent to that of selenite and both are more toxic than SeMet (the l-isomer of which is more toxic than the d-isomer) and other organic Se compounds in animal studies(Reference Barceloux40). Comparison of selenite and Se-yeast diets in rats showed that Se-yeast was much less toxic than selenite; although the livers of animals fed Se-yeast showed up to 50 % greater deposition of Se, there was no corresponding toxicity, as evidenced by histological examination(Reference Spallholz, Raftery, Combs, Spallholz, Levander and Oldfield108). Se-yeast also seems to be less toxic than l-SeMet; after 2 weeks of feeding 30 μg Se/g diet, survival in mallard ducklings was 36 % for l-SeMet and 88 % for Se-yeast(Reference Heinz, Hoffman and LeCaptain109). Human studies have also shown a lower chronic toxicity of organically bound Se, though there are limited data on the toxicity of individual compounds(Reference Barceloux40). However, SeMet is known to be the main Se species present in the diet of Chinese who developed chronic selenosis from consumption of maize and rice grown in the Enshi area of China(Reference Beilstein, Whanger and Yang39).
The toxicity of the Se-accumulators to livestock has been linked to the high levels of Se-methyl-selenocysteine found in these species(Reference Spallholz110). Se-accumulator plants are able to circumvent the toxicity that would result from the non-specific integration of the seleno-amino acids SeCys and SeMet into proteins by converting the precursor, SeCys, into the non-protein amino acids Se-methyl-selenocysteine, γ-glutamyl-Se-methyl-selenocysteine and selenocystathionine(Reference Sors, Ellis and Salt8). The potent toxicity of Se-accumulator plants to grazing animals is probably more a reflection of the extremely high concentrations of Se that can build up in these plants – up to 10–15 mg Se/g dry weight even on non-seleniferous soils(Reference Sors, Ellis and Salt8) – rather than the toxicity of Se-methyl-selenocysteine per se. According to Dr C Ip (2006, personal communication) who has worked with Se-methyl-selenocysteine for many years, it should be a safer compound than SeMet based on its biochemistry; though both compounds are equally well absorbed, Se-methyl-selenocysteine is converted to excretable metabolites more rapidly resulting in lower tissue retention of Se. Comparison of the no observable adverse effect level (NOAEL) in male and female rats for Se-methyl-selenocysteine (1·0 and 0·5 mg/kg per d, respectively) with that for selenite (0·14 and 0·2 mg/kg per d, respectively) suggests that Se-methyl-selenocysteine is less toxic at least than selenite(Reference Jia, Li and Chen111) (C Ip (2006), personal communication). Results from Hasegawa et al. (Reference Hasegawa, Mihara, Nakamuro and Sayato103) similarly suggest that methylated forms of Se are generally less toxic than non-methylated compounds. This postulated lower toxicity may be highly relevant to the potential for use of Se-methyl-selenocysteine in human cancer prevention studies.
Conclusion
The development of state-of-the-art analytical methods that combine elemental and molecular mass spectrometric detection to investigate different chemical forms of Se in food has made possible the identification of a variety of Se species in foods and supplements. However, this is such a difficult and exacting area of research that, to date, we have only scratched the surface. It is difficult to maintain the integrity of species through the extraction process. Though we may know the identity of some Se species present in foods, there is no case where we know the identity of all the Se species; only where we have mass balance can we ensure that all species have been captured. We need to take food processing and preparation into account so that we are actually investigating the species that will be consumed (for example, Japanese soup stock made from shiitake mushrooms(Reference Ogra, Ishiwata, Ruiz-Encinar, Lobinski and Suzuki56)).
There remain considerable gaps in our knowledge of the forms of Se that naturally occur in foods. For instance, we know little about species of Se, other than SeMet, in fish, normally considered a good source of the element, or indeed what Se–Hg species may be present; we need to know full speciation of Se in Se-yeast because of its frequent use in human intervention studies; and perhaps most importantly, there is a need to know to which Se species beneficial or detrimental health effects can be attributed.
We need to continue to develop speciation methodology, and to further investigate biosynthetic and metabolic pathways in order to have a steer on what species we should be searching for. Where we do suspect we know the identity of an active species (for example, Se-methyl-selenocysteine), we need single-species trials to prove efficacy or relative efficacy to help us towards a better understanding of how dietary Se should be supplemented.
Finally, there is a clear need for analytical chemists to present the data in a form that is understandable to and usable by consumers, nutritionists and legislators. Without adequate knowledge of Se speciation, false conclusions may be drawn when assessing Se requirements for optimal health. Furthermore, the ability to identify and accurately quantify Se species with powerful anti-cancer or other valuable effects will be essential for the development of plant-breeding programmes to optimise the biosynthesis of such species if clear proof of their health effects should be forthcoming.
Acknowledgements
M. P. R. wrote the manuscript, prepared the figures and compiled the first draft of the table. H. G. I. contributed to sections of the manuscript and to the table. M. S. initiated the writing of the review and advised on its content. The authors have no conflict of interest to declare.
