CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

FEBS Letters 579 (2005) 792–796 FEBS 29196

Effects of Se-depletion on and W expression in the colon

Vasileios Pagmantidisa, Giovanna Bermanob, Stephane Villettea, Iain Broomb, John Arthurc, John Hesketha,* a School of and Molecular Biosciences, University of Newcastle upon Tyne, NE1 7RU, UK b School of Life Sciences, The Robert Gordon University, Aberdeen AB25 1HG, UK c Rowett Research Institute, Bucksburn, Aberdeen AB21 9SB, UK

Received 3 September 2004; revised 22 November 2004; accepted 9 December 2004

Available online 27 December 2004

Edited by Lukas Huber

tant in determining the level of anti-oxidant protection and Abstract (Se)-containing have important roles in protecting cells from oxidative damage. This work inves- inflammatory responses in the colon [8]. In addition, supple- tigated the effects of Se-depletion on the expression of the mentation with Se above normal intake reduces mortality from encoding in colonic mucosa from rats fed diets of cancer of the colon [2,3], as well as that of lung and prostate. different Se content and in human intestinal Caco-2 cells grown These observations indicate that selenoproteins such as the in Se-adequate or Se-depleted culture medium. Se-depletion pro- GPXs are critical for colon function. However, little is known duced statistically significant (P < 0.05) falls in glutathione per- of how selenoprotein gene expression in colonic cells, particu- oxidase (GPX) 1 mRNA (60–83%) and selenoprotein W mRNA larly in vivo, is regulated by Se availability. (73%) levels, a small but significant fall in GPX4 mRNA (17– The aim of the present work was to investigate how Se- 25%) but no significant change in GPX2. The data show that deficiency modulates expression of a range of selenoprotein SelW expression in the colon is highly sensitive to Se-depletion. genes in the rat colon and a human gastrointestinal cell line. 2004 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Since patterns of selenoprotein activity are accompanied by altered patterns of mRNA abundances and both are regu- Keywords: Colon; ; Selenium; lated by Se availability [9,10], our approach was to use a no- Selenoprotein; Caco-2; Gene array vel ÔselenoproteinÕ macroarray to screen for genes affected by Se depletion and then to extend the studies using Northern hybridisation.

1. Introduction 2. Materials and methods The micronutrient selenium (Se) is essential for health [1]. 2.1. DNA probes Low intake of Se has been implicated with the increased inci- The GPX1, GPX4 and glyceraldehyde-3-phosphate dehydrogenase dence of disease and Se supplementation lowered both cancer (GAPDH) probes were 880 bp EcoRI, 814 bp EcoRI, and 780 bp incidence and mortality from cancer [2,3]. Se is present in sev- PstI/XbaI fragments, respectively, as used previously [9]. The human eral proteins, called selenoproteins, as the amino-acid seleno- GPX2 probe (from Prof. R. Brigelius-Flohe) was a 663 bp EcoRI frag- (Se-cys) [4]. Approximately 30 selenoproteins and up ment [11]. The human GPX1 probe was a 300 bp BamHI/XbaI frag- ment [12] and the human GPX4 probe (from Prof. K. Yagi, Gifu) to 25 selenoprotein genes have been identified in mammals was an 874 bp HindIII/XbaI fragment [13]. The SelW probe (from and they have a variety of biological roles [4]. They include Dr. I. Kim) was a 420 bp BamHI/XhoI fragment corresponding to the glutathione peroxidase family of , all of which the mouse brain cDNA [14]. The 18S rRNA probe was a 328 bp frag- are involved in the cellÕs systems [5], the thiore- ment corresponding to nucleotides 4852–5179 of the human ribosomal RNA sequence [15] cloned into the SrfI site of PCR-Script AMP vec- doxin reductases, which are involved in the redox regulation tor (Stratagene) and excised using KpnI and SacI. of gene expression [6] and a number of recently identified sele- noproteins (e.g., SelX, SelN, SelW, and 15 kDa selenoprotein) 2.2. RNA extraction and Northern hybridisation that are less well characterised [4]. The glutathione Total RNA was extracted by the guanidinium thiocyanate/phenol/ include cytosolic glutathione peroxidase (GPX1), gastrointesti- chloroform procedure [16]. Northern blotting was carried out using nal glutathione peroxidase (GPX2), plasma glutathione perox- standard procedures [17] and membranes pre-hybridised for 20 min at 68 C in 10 ml of QuickHyb solution (Stratagene). 25 ng of idase (GPX3) and phospholipid hydroperoxide glutathione 32 DNA probes was labelled with [ P]dCTP by random priming, as peroxidase (GPX4). described previously [9], and hybridisation carried out for 1 h at The phenotype of knock-out mice lacking GPX1 and GPX2 68 C. Non-specifically bound probe was removed from membranes includes colon pathology and increased susceptibility to coli- by two washes in 2· SSC (1· SSC = 0.15 M NaCl/0.015 M sodium tis[7], leading to the suggestion that these enzymes are impor- citrate)/0.1% SDS at room temperature for 15 min, followed by one wash in 0.1· SSC/0.1% SDS at 60 C for 30 min. Specific hybridisation was detected using a Canberra Packard Instantimager for quantification (results for each probe were expressed per unit of *Corresponding author. Fax: +44 0 191 222 8684. hybridisation achieved with the 18S rRNA probe to allow correction E-mail address: [email protected] (J. Hesketh). for any variation between loading of RNA on the gel or transfer to

0014-5793/$30.00 2004 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2004.12.042 V. Pagmantidis et al. / FEBS Letters 579 (2005) 792–796 793 the nylon membrane) and by autoradiography using Fujifilm Super Table 1 FX at 80 C. After analysis, membranes were stripped by washing Effects of Se-deficiency on expression of a range of genes detected by in 0.1% SDS for 10 min at 95 C before rehybridisation to other macroarray probes. Gene name Ratio Se/Se+

2.3. cDNA expression arrays GPX1 0.46 32P-labelled cDNA probe synthesis from 3.5 lg total RNA and GPX2 1.63 hybridisation to the Atlas Array membranes were performed according GPX3 0.67 to the Atlas cDNA expression Arrays User Manual (BD Biosciences GPX4 0.76 Clontech, Palo Alto, USA). Detection was performed with a phos- SelP 0.90 phorimager (Molecular Dynamics, Phosphorimager Storm 860) using SBP1 Up 24 h and 3 days exposure. Analysis of the results was performed using DbpB 0.94 the AtlasImage v2.7 software (BD Biosciences Clontech) with a specific SPS (SelD) 1.03 grid corresponding to our custom made ‘‘selenoprotein’’ array: the ar- Phospholipase A2 0.95 ray was spotted with all available human selenoprotein cDNAs, Cu/Zn SOD 1.29 cDNAs corresponding to genes whose products are associated with Mn SOD 1.04 Se metabolism or arachidonate metabolism, and a series of standard SOD3 (extracellular) 1.58 housekeeping genes (Table 1 and associated legend). To analyse differ- Glutathione-S- A2 1.50 ential gene expression using the Atlas Array hybridisation, intensity Glutathione-S-transferase A4 2.76 differences were normalised by taking into account changes in the Interleukin 1; b 3.89 housekeeping genes. SelN 0.76 SelW 0.27 SelB (elongation factor) Down 2.4. Cell culture tRNA associated 1.20 Stocks of human colon adenocarcinoma cells (Caco-2) were grown SelZ f1 Down at 37 C in DulbeccoÕs modified EagleÕs medium (with 4.5 g/l glucose TR2 0.73 and Glutamax) supplemented with 10% (v/v) foetal calf serum (Sigma, TR3 1.33 Poole, UK), 60 lg/ml gentamycin, 1% (v/v) non-essential amino acids MHC; class1; C Up and penicillin/streptomycin (100 units/ml and 100 lg/ml, respectively). b-Actin 1.08 Cells were transferred to serum-free medium to which was added either GAPDH 1.11 insulin (5 lg/ml) and transferrin (5 lg/ml) (selenium-deficient cells), or a-Tubulin 1.20 insulin, transferrin and selenium as sodium selenite (7 ng/ml) (sele- Ubiquitin 1.09 nium-replete cells) [18]. Medium was changed every 2 days. Ribosomal protein L13a 0.90 Ribosomal protein S9 0.96 2.5. and diets HPRT 1.67 Weanling Male Hooded Lister Rats of the Rowett strain were ran- Genomic DNA 1.19 domly allocated to groups of six and fed ad libitum on a semi-synthetic mRNA abundances were measured in Caco-2 cells after 2 days in Se- diet containing different amounts of Se for 6 weeks. The basal diet [19] deficient or Se-replete conditions. Results are the mean values obtained contained 8 ng of Se/g (severely Se-deficient) and the other diets con- from two experiments each analysed using duplicate arrays. Results tained 33, 63 (Se-deficient) or 111 (Se-adequate) ng of Se/g as sodium are expressed as a ratio of the Se-depleted cells to Se-adequate cells. selenite. At the end of the 6 weeks, the animals were anaesthetised with GPX1 and SelW mRNA abundances; showed greater than twofold isofluorane and blood samples were taken by cardiac puncture. The change; which is considered to be significant. When expression levels animals were killed by cervical dislocation and the entire colon dis- were close to the limit of detection the ratio of expression is given as sected out and washed with cold sterile PBS; the colon was then cut up/down and not as a precise value. The following genes were also open, the mucosa scraped off and then rapidly frozen in liquid nitrogen present on the array but expression was not detected: thioredoxin and stored at 80 C. reductase 1, type I, II and III, 15 kDa selenoprotein, sele- noprotein M, T, and X, mitochondrial capsule selenoprotein, SECIS- binding protein 2, 5-lipoxygenase, arachidonate 15-lipoxygenase types 3. Results and discussion 1 and 2, arachidonate 12-lipoxygenase 5-lipoxygenase activating pro- tein, cyclooxygenases 1 and 2, tumor necrosis factor a and b, LTB4 receptor, thromboxane A2 receptor, prostaglandin F and D2 recep- Caco-2 cells were grown for 2–6 days in either a Se-sup- tors, glutathione-S-transferase theta 1 and 2, thyroid hormone recep- plemented or Se-depleted medium and RNA extracted and tor a and b2, interferon gamma receptor and precursor, analysed by Northern hybridisation. Cells grown in the Se- metallothionein isoform 1rR, interleukin 1 a and b, interleukin 2, depleted medium showed a 60% decrease in the level of tyrosine 3-monooxygenase. Abbreviations: Sel, selenoprotein; SOD, superoxide dismutase; TR, GPX1 mRNA by 2 days (P < 0.05) compared with cells thioredoxin reductase; MHC, major histocompatibility complex. grown in Se-supplemented medium (Fig. 1). This low expres- sion remained after 4 or 6 days in the Se-depleted medium. Addition of selenite to the medium at day 6 led to an in- mented medium, it was found that Caco-2 cells expressed ten crease in GPX1 mRNA abundance such that it returned to selenoprotein mRNAs (GPX1, GPX2, GPX3, GPX4, TR2, control levels by day 8 (Fig. 1). The level of GPX1 mRNA TR3, selenoproteins N, P, W and Z) at a level that was detect- is a consistent marker of Se status in cell and exper- able using this technology (Table 1). Expression of other sele- iments [9,10,18,20] and these changes indicate that after 2 noproteins genes on the array (see legend to Table 1) was days in Se-depleted medium the Caco-2 cells become severely below the level of detection. After 2 days in Se-depleted med- Se deficient, that the effect is not increased significantly over ium, only two selenoproteins showed a major change in a further 4 days and that the effect can be reversed by re- mRNA abundance compared with Se-supplemented cells addition of selenite; this reflects closely the changes in (Table 1): GPX1 and Selenoprotein W (SelW), for which the GPX1 mRNA in H4 hepatoma cell and FRTL-5 thyroid ratio of relative expression was 0.46 and 0.27, respectively. cells grown in similar media [18,21]. In addition, Se depletion led to increased expression of the Using a custom-made macroarray and RNA extracted from non-selenoprotein mRNAs interleukin 1b and glutathione-S- Caco-2 cells grown for 2 days in Se-deficient or Se-supple- transferase A4, for which the ratio of relative expression was 794 V. Pagmantidis et al. / FEBS Letters 579 (2005) 792–796

Fig. 1. Effect of Se-depletion on GPX1 mRNA level in Caco-2 cells. Cells were grown for 2–6 days in Se-supplemented or Se-deficient Fig. 2. Effect of Se-deficiency on the abundance of glutathione serum-free medium. After 6 days, selenite was given back to the Se- peroxidase 1, 2 and 4 mRNAs in Caco-2 cells. All lanes were loaded depleted cells. GPX1mRNA and 18S rRNA levels were determined by with 10 lg of total RNA and filters were successively hybridised with Northern hybridisation using a Canberra Packard Instantimager for GPX1, GPX2 and GPX4 probes. The ratio of hybridisation with each quantification. Data were expressed as percentages of the mean ratio of probe to that with 18S rRNA probe was calculated using Canberra GPX1 mRNA abundance/18S rRNA abundance obtained for Se- Packard Instantimager. Results are shown as means ± S.E.M. from replete cells (day 0). Results are shown as means ± S.E.M. from two two independent experiments. Statistical significance was assessed by a independent experiments. Statistical significance was assessed by a Mann–Whitney U test, \P < 0.05. Mann–Whitney U test, \P < 0.05.

accepted sensitive marker of Se-depletion, were consistent with 3.89 and 2.76, respectively. The relative expression of GPX3, data from earlier studies (results not shown). RNA from GPX4 and SelN in Se-depleted cells (0.67, 0.76, and 0.76, mucosal scrapings was subjected to Northern hybridisation respectively) was less than the twofold change considered to with probes to GPX1, 2 and 4. Quantification of the hybridisa- be significant but suggested that there may be a small fall in tion showed that mean GPX1 mRNA abundance was expression. There was no evidence of changes in relative decreased by 83% (P < 0.05) in the colon of severely Se-defi- expression of GPX2, SelP, TR2 and TR3 or of non-selenopro- cient rats compared with tissue from the Se-adequate rats tein antioxidant enzymes such as Mn and Cu/Zn superoxide and decreased by 60% in the colon of the deficient rats dismutases. These array data suggest that in Caco-2 cells (Fig. 3). In contrast, mean GPX4 mRNA abundance showed expression of GPX1, 2 and 4 are differentially affected by Se- depletion and that the major effect of Se-depletion on seleno- protein expression in colonic epithelial cells, as assessed by mRNA abundance, is a dramatic fall in the expression of SelW and GPX1. To confirm these findings, two experiments were carried out. First, mRNA abundances were measured in Se-supplemented and deficient Caco-2 cells by Northern hybridisation. Second, and more importantly, since little was known about the effects of Se-depletion on the colon in vivo, selenoprotein gene expres- sion was studied in rats that were fed diets that provided differ- ent levels of dietary Se. Quantification of Northern hybridisation data from cell cul- ture experiments showed that there was no statistical signifi- cant difference in GPX2 mRNA levels compared to cells grown in medium supplemented with Se (Fig. 2). However, there was a small (17%) but statistically significant decrease (P < 0.05) in the GPX4 mRNA level in the Se-depleted cells (Fig. 2). This was in contrast with the marked decline in the le- vel of GPX1 mRNA. These data are consistent with earlier observations that Se-supplementation of cells grown in stan- dard medium increases GPX1 mRNA abundance but not that of GPX2 [10]. Fig. 3. Change of glutathione peroxidase mRNA abundances in the Rats were fed diets containing either an adequate supply of rat colon in response to Se supply. Samples of total RNA were Se (111 ng Se/g; group D) or three levels of low Se supply rang- analysed by Northern hybridisation with probes for GPX1, GPX2, ing from a severely deficient Se supply (8 ng Se/g; group A), to GPX4 and GAPDH mRNAs, and for 18S rRNA. The ratio of deficient (33 ng Se/g; group B) or marginal supply (63 ng Se/g; hybridisation with each probe to that with the 18S probe was calculated and data expressed as percentages of the mean mRNA group C). Previously, such diets have been found to consis- abundances obtained for animals on the Se-adequate diet (group D). tently produce Se-depletion in rats [9,20]. In the present exper- Results are shown as means ± S.E.M. Results were compared by a iments measurements of liver GPX1 mRNA abundance, an Mann–Whitney U test, \P < 0.05. V. Pagmantidis et al. / FEBS Letters 579 (2005) 792–796 795 only a small decrease (25%; P < 0.05) in the colon of the se- Acknowledgements: This research was funded by a grant from World verely deficient rats and GPX2 mRNA levels showed no statis- Cancer Research Fund to J.E.H., I.B. and G.B. J.R.A. is supported tically significant change in rats fed any of the Se-deficient by Scottish Executive Environment and Rural Affairs Department. S.V. was supported by Food Standards Agency, which also provided diets. These results confirm and extend the array data by indi- funding for the macroarrays. We thank Dr. S. Dillon (Liverpool) for cating that the effect of Se in cell culture is mirrored by that in help with the macroarray experiments. the colon in vivo and by showing that there is a differential ef- fect of Se-depletion on GPX1, 2 and 4 in the colon. In the liver, Se-depletion has no effect on GPX4 mRNA References expression although GPX1 decreases dramatically [9]. The present data show that, in the colon, GPX4 mRNA abundance [1] Rayman, M.P. (2000) The importance of selenium to human health. Lancet 356, 233–241. decreases as well as that of GPX1, indicating that the response [2] Clark, L.C., Combs Jr., G.F., Turnbull, B.W., Slate, E., Alberts, of glutathione peroxidase expression to Se-depletion is differ- D., Abele, D., Allison, R., Bradshaw, J., Chalker, D., Chow, J., ent in the colon from that observed previously in liver or heart Curtis, D., Dalen, J., Davis, L., Deal, R. and Dellasega, M. (1997) [9]. This supports the contention that regulation of selenopro- The nutritional prevention of cancer with selenium 1983–1993; a randomised controlled trial. JAMA 276, 1957–1963. tein expression by Se supply is affected differentially in different [3] Whanger, P.D. (2004) Selenium and its relationship to cancer: an tissues [9,10]. update. Brit. J. Nutr. 91, 11–28. Using a mouse SelW cDNA as probe, Northern hybridisa- [4] Kryukov, G.V., Castellano, S., Novoselov, S.V., Lobanov, A.V., tion showed a band corresponding to SelW in RNA samples Zehtab, O., Guigo, R. and Gladyshev, V.N. (2003) Characteriza- from rats fed the Se-adequate diet (Fig. 4). However, the tion of mammalian selenoproteomes. Science 300, 1439–1443. [5] Brigelius-Flohe, R. (1999) Tissue-specific functions of individual expression of SelW mRNA was decreased in the Se-depleted glutathione peroxidases. Free Radic. Biol. Med. 27, 951–965. colon to such an extent that SelW mRNA was barely detect- [6] Gasdaska, J.R., Harney, J.W., Gasdaska, P.Y., Powis, G. and able in samples from animals fed the severely Se-deficient Berry, M.J. (1999) Regulation of human thioredoxin reductase 0 diet (Fig. 4). Density analysis of the autoradiographs expression and activity by 3 -untranslated region selenocysteine insertion sequence and mRNA instability elements. J. Biol. Chem. showed that the ratio of SelW/GAPDH expression ranged 274, 25379–25385. from 1.0 to 1.9 in the rats fed the Se-adequate diet but [7] Esworthy, R.S., Aranda, R., Martin, M.G., Doroshaw, J.H., was zero in the rats fed the severely Se-deficient diet, since Binder, S.W. and Chu, F.-F. (2001) Mice with combined the intensity of the SelW band was below the level of detec- disruption of Gpx1 and Gpx2 genes have colitis. Am. J. Physiol. tion. These data indicate that SelW expression is highly sen- 281, G848–G855. [8] Chu, F.-F., Esworthy, R.S. and Doroshaw, J.H. (2004) Role of sitive to Se-depletion. SelW is expressed in muscle [22] but Se-dependent glutathione peroxidases in gastrointestinal inflam- has not previously been shown to be expressed in either mation and cancer. Free Radic. Biol.Med. 36, 1481–1495. the colon or a gastrointestinal cell line. [9] Bermano, G., Nicol, F., Dyer, J.A., Sunde, R.A., Beckett, G.J., In summary, the present data obtained from both Caco-2 Arthur, J.R. and Hesketh, J.E. (1995) Tissue-specific regulation of selenoenzyme gene expression during selenium deficiency in rats. cells in culture and from rat colon tissue show that Se-deple- Biochem. J. 311, 425–430. tion has differential effects on selenoprotein expression, in par- [10] Wingler, K., Bo¨cher, M., Flohe, L., Kollmus, H. and Brigelius- ticular a much greater effect on the abundance of GPX1 and Flohe, R. (1999) mRNA stability and selenocysteine insertion SelW mRNAs than on that of GPX4 or GPX2 mRNA. SelW sequence efficiency rank gastrointestinal glutathione peroxidase is an anti-oxidant protein [14,22,23] and therefore our hypoth- high in the hierarchy of selenoproteins. Eur. J. Biochem. 259, 149– 157. esis is that the large effect of Se-depletion on the expression of [11] Brigelius-Flohe, R., Muller, C., Menard, J., Florian, S., Schmehl, both SelW and GPX1, together with the smaller effect on K. and Wingler, K. (2001) Functions of GI-GPx: Lessons from GPX4, and despite the maintenance of GPX2 expression, low- selenium-dependent expression and intracellular localization. ers antioxidant protection in the colon; and in turn this would Biofactors 14, 101–106. [12] Sukenaga, Y., Ishida, K., Takeda, T. and Takagi, K. (1987) increase susceptibility to development of inflammatory re- cDNA sequence coding for human glutathione peroxidase. sponses in the gastrointestinal system [8]. Nucleic Acid Res. 15, 7178–7178. [13] Yagi, K., Komura, S., Kojima, S., Sun, Q., Nagata, N., Ohishi, N.and and Nishikimi, M. (1996) Expression of human phospho- hydroperoxide glutathione peroxidase gene for protection of host cells from lipid hydroperoxide-mediated injury. Biochem. Biophys. Res. Commun. 219, 486–491. [14] Jeong, D., Kim, T.S., Chung, Y.W., Lee, B.J. and Kim, I.Y. (2002) Selenoprotein W is a glutathione-dependent antioxidant in vivo. FEBS Lett. 517, 225–228. [15] Erickson, J.M., Rushford, C.L., Dorney, D.J., Wilson, G.N. and Schmickel, R.D. (1981) Structure and variation of human ribosomal DNA. Gene 16, 1–9. [16] Chomczynski, P. and Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidium thiocyanate–phenol–chloro- form extraction. Anal. Biochem. 162, 156–159. [17] Maniatis, T., Fritsch, E.F. and Sambrook, J., Eds., (1982). Fig. 4. Response of SelW mRNA levels in rat colon to Se-depletion. Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Total RNA isolated from rat mucosal scrapings was subjected to Laboratory Press, Cold Spring Harbor, NY. Northern hybridisation (10 lg in each lane). Specific hybridisation to a [18] Bermano, G., Arthur, J.R. and Hesketh, J.E. (1996) Selective SelW probe was detected by autoradiography. Lanes 1–3 correspond control of cytosolic glutathione peroxidase and phospholipid to RNA from the severely Se-deficient group (group A) and lanes 4–8 hydroperoxide peroxidase mRNA stability by selenium supply. to RNA from the Se-adequate group (group D). GAPDH mRNA was FEBS Lett. 387, 157–160. used as internal control and did not show any change due to different [19] Arthur, J.R., Nicol, F., Hutchinson, A.R. and Beckett, G.J. dietary Se. (1990) The effects of selenium depletion and repletion on the 796 V. Pagmantidis et al. / FEBS Letters 579 (2005) 792–796

metabolism of thyroid hormones in the rat. J. Inorg. Biochem. 39, regulate selenoenzyme gene expression in thyroid cells (FRTL-5) 101–108. in culture. FEBS Lett. 438, 81–84. [20] Mitchell, J.H., Nicol, F., Beckett, G.J. and Arthur, J.R. (1997) [22] Gu, Q.P., Ream, W. and Whanger, P.D. (2002) Selenoprotein Selenium and iodine deficiencies: effects on brain and brown W gene regulation by selenium in L8 cells. Biometals 15, 411– adipose tissue selenoenzyme activity and expression. J. Endocri- 420. nol. 155, 255–263. [23] Ream, L.W., Vorachek, W.R. and Whanger, P.D. (2001) in: [21] Villettte, S., Bermano, G., Arthur, J.R. and Hesketh, J.E. (1998) Selenium: Its Molecular Biology and Role in Human Health Thyroid stimulating hormone and selenium supply interact to (Hatfield, D.L., Ed.), Kluwer Press, Boston.