Results obtained from human and animal studies suggest that an elevated plasma homocysteine (HCys) concentration may promote the early development of coronary artery disease and ischaemic stroke(Reference Zhou and Austin1). The generation of reactive oxygen species(Reference Lin, Chen and Chen2), subsequently leading to increased LDL oxidation(Reference Kassab, Ajmi and Issaoui3), the generation of vascular injury via the increased expression and liberation of proinflammatory transcription factors and cytokines(Reference Séguin, Abid and Spokes4–Reference Ungvari, Csiszar and Edwards6) as well as reduced vessel relaxation via decreased NO availability(Reference Griffiths, Aldred and Dale7) are discussed as mechanisms underlying the proatherogenic effects of HCys. With the exception of inherited disorders in methionine (Met) and HCys metabolism such as cystathione-β-synthase (CBS) or Met synthase deficiency or polymorphisms in these genes, hyperhomocysteinaemia is assumed as being influenced by various nutritional components(Reference Flicker8). Currently, the influence of folic acid, vitamin B12 and vitamin B6 status as well as the effectiveness of the supplementation of these vitamins in lowering HCys concentration and reducing the risk and recurrence of CVD and ischaemic stroke is the subject of controversial discussion(Reference Bønaa, Njølstad and Ueland9–Reference Lonn, Yusuf and Arnold11). Despite evidence and inconsistent data on the beneficial effects of vitamin B supplementation in CVD and ischaemic stroke, the American Heart and Stroke Association and the German, Austrian and Swiss Homocysteine Society advise treating patients with hyperhomocysteinaemia or with a history of transient ischaemic attacks or stroke daily with a standard multivitamin preparation in order to achieve the recommended daily amounts of 0·4 mg folic acid, 2·4 μg vitamin B12 and 1·7 mg vitamin B6(Reference Malinow, Bostom and Krauss12, Reference Stanger, Herrmann and Pietrzik13). The results of the National Nutrition Survey II revealed a deficit in folic acid intake in the German population also, and subsequently, a fortification of selected foods with folic acid is under consideration(Reference Karlsruhe14). Other important factors determining plasma HCys concentration are age, sex and renal status(Reference Must, Jacques and Rogers15, Reference Heinz, Kropf and Luley16). In recent years, contradictory results have also been reported with regard to the influence of the trace element Se on HCys metabolism. Whereas studies with chicks(Reference Halpin and Baker17) and with mice and rats(Reference Uthus, Yokoi and Davis18–Reference Uthus and Ross21) showed that Se-deficient animals had an impressively reduced plasma HCys concentration than their littermates supplemented with the recommended dietary Se amount or with supranutritional Se, human cross-sectional observational studies with elderly people from Spain(Reference González, Huerta and Alvarez-Uría22), Upper Silesia(Reference Klapcińska, Poprzęcki and Danch23) and the UK(Reference Bates, Thane and Prentice24) uniquely showed an inverse correlation between Se status and plasma HCys concentration. Another study with a middle-aged Inuit population also demonstrated the inverse correlation between Se status and plasma HCys concentration(Reference Bélanger, Dewailly and Berthiaume25). Both in a 20-week intervention trial in New Zealand whereby 189 volunteers aged 18–64 years were supplemented with a placebo or 200 μg Se/d as selenomethionine(Reference Venn, Grant and Thomson26) and in an intervention trial in the UK with 501 elderly people who were supplemented with a placebo or 100, 200 or 300 μg Se/d as high Se yeast for 6 months(Reference Bekaert, Cooper and Green27), Se supplementation had no influence on plasma HCys level. However, the results of a rat study demonstrated a significant reduction of plasma HCys concentration in Se-supplemented rats fed diets supplemented simultaneously with a diet containing a high folic acid concentration of 2 mg/kg(Reference Uthus, Ross and Davis19). Besides the feeding of Se-rich diets, the feeding of Met-rich diets to rats is accepted as a means of increasing their plasma HCys level(Reference Hirche, Schröder and Knoth28, Reference Hirche, Schröder and Knoth29). Results of the impact of dietary Met on plasma HCys in human subjects vary with regard to the critical dietary Met amount. Whereas some reports suggest that about fivefold of the recommended Met amount is necessary to produce a considerable increase in plasma HCys concentration(Reference Ward, McNulty and Pentieva30, Reference Ward, McNulty and McPartlin31), another study with Dutch men found a distinct postprandial increase in plasma HCys concentration lasting about 6 h in healthy men given a high protein diet (21 % energy)(Reference Verhoef, van Vliet and Olthof32). Results from a Finish study with men suggest an association between a slightly increased daily Met intake between 25 and 40 mg/kg body weight and a 31 % higher risk of coronary problems(Reference Virtanen, Voutilainen and Rissanen33). Feeding Met-rich diets to rats and subjecting human subjects to a Met-loading test additionally effected an increase in plasma cholesterol (Chol) concentration, representing another risk factor for the accelerated development of atherosclerosis(Reference Hirche, Schröder and Knoth28, Reference Hirche, Schröder and Knoth29, Reference Hart, Mangoni and Swift34). Moreover, data from two large cross-sectional observational trials from the USA and the UK found that a high Se status was positively associated with increased plasma values for total Chol and LDL Chol(Reference Bleys, Navas-Acien and Stranges35, Reference Stranges, Laclaustra and Ji36). However, combined effects of dietary Se and Met on HCys and Chol metabolism have not been investigated to date. Consequently, we have designed a trial with growing rats to study the effects of different dietary Se concentrations (Se deficiency, half the recommended dietary amount, recommended dietary amount and three times the recommended dietary amount) and of two dietary Met concentrations (recommended dietary amount and five times the recommended dietary amount) on the HCys and Chol metabolism of these animals. Taking into consideration both the recommendations and species differences between human subjects and rats, we have chosen the Se and Met supplementation range in a way that allows transferability of the data.
Methods and materials
Rats and diets
The protocol of the rat nutrition study was approved by the Regional Council of Giessen and by the Animal Welfare Committee of the Justus Liebig University of Giessen (record token: V54-19c20/15cGI 19/3). Sixty-four healthy growing male rats (initial body weight, 64·2 (se 0·25) g from the Interdisciplinary Research Center, Department of Animal Nutrition and Nutritional Physiology's own strain (HK 51) were randomly assigned to eight groups of 8. The Se-deficient basal diet of group C15 had an analysed Se concentration of approximately 15 μg/kg, and was based on Torula yeast and Se-deficient wheat (Table 1). The diets for groups C50 (one-third of the recommended dietary Se level), C150 (recommended dietary Se level) and C450 (three times the recommended dietary Se level) were supplemented with 50, 150 and 450 μg Se/kg diet as sodium selenate. The diets for groups C15, C50, C150 and C450 were supplemented with 3·0 g dl-Met/kg to adjust the Met level to the requirement of the growing rats. The high Met diets of groups M15, M50, M150 and M450 had dietary Se levels that were the same as those of the diets of the respective C groups, but were supplemented with fivefold Met amount (15 g/kg). With the exception of the variations in Se and Met concentrations, the diets were composed according to the American Institute of Nutrition-93G recommendations(Reference Reeves37). The rats were kept individually, and had ad libitum access to the diet and bidistilled water. After 8 weeks, the animals were decapitated under CO2. The livers were excised, frozen in liquid N2 and stored at − 80°C until further analysis. Blood was collected in heparinised tubes, and was centrifuged for 20 min at 3000 g for plasma preparation. Plasma was stored at − 80°C until analysis.
Composition of the basal diet
* In the diets of groups M15, M50, M150 and M450, the amount of methionine in the basal diet was increased to 15 g/kg by replacing maize starch.
† Added according to the American Institute of Nutrition-93G recommendations(Reference Reeves37).
‡ Considering the high natural concentrations of Ca, K, Mg P, Cu, Fe and Zn of Torula yeast, the following mineral amounts were added to 1 kg complete diet with 35 g/kg of the concentrated premix to achieve the recommended amounts(Reference Reeves37): Ca, 3·0 g (CaCO3); Mg, 228 mg (MgSO4.7H2O); Cu, 2·50 mg (CuSO4.5H2O); Mn, 8·62 mg (MnSO4.H2O); Na, 1·02 g (NaCl); K, 741 mg (K3C6H5O7.H2O); Cr, 1·00 mg (KCR(SO4)2.12H2O); B, 0·51 mg (H3BO3); Ni, 0·49 mg (NiSO4.6H2O); F, 1·00 mg (NaF); Si, 5·00 mg (Na2SiO3.9H2O); Li, 0·10 mg (LiCl); Mo, 0·16 mg (H24Mo7N6O24.4H2O); I, 0·23 mg (KI); V, 0·09 mg (NH4VO3).
§ All vitamins were added according to the recommendations(Reference Reeves37) without any corrections.
∥ In the Se-supplemented groups, maize starch was replaced with 8·33 g (C50 and M50), 25 g (C150 and M150), and 75 g (C450 and M450) of the Se premix containing 6 mg Se/kg as sodium selenate in order to adjust the final Se concentrations of 50, 150 and 450 μg/kg.
Selenium assay
Se concentration in the diets and livers was measured by hydride generation atomic absorption spectrometry (Unicam PU 9400 X; PU 3960 X) as described previously(Reference Mueller, Pallauf and Most38). Certified samples from the National Institute of Standard and Technology (soft winter wheat starch, no. 8438 and bovine liver, no. 1577 b) and from Medichem (control serum, Metalle S; Steinenbronn, Germany) served as the reference materials for Se determination in the different matrices.
Liver glutathione peroxidase 1 and plasma glutathione peroxidase 3
Glutathione peroxidase (GPx) 1 and GPx3 activity was measured spectrophotometrically (Carry 50; Varian, Darmstadt, Germany) using the assay method coupled to GSH reductase at 340 nm(Reference Lawrence and Burk39). NADPH oxidation, which is proportional to GPx-dependent peroxide reduction, was recorded for 3 min. GPx1 activity was analysed in the rat liver cytosol diluted with 50 mmol/l phosphate buffer, and GPx3 activity was analysed in the undiluted plasma. For both enzymes, H2O2 served as the substrate. One unit of GPx1 and GPx3 activity was defined as 1 μmol NADPH oxidised per minute and normalised to 1 mg protein.
Protein concentration in the liver cytosol and in plasma
The protein concentration of the liver cytosol and of plasma was determined according to a standard protocol(Reference Bradford40) in microtitre plates using the plate reader Tecan SpectraFluor Plus (Tecan, Grödig, Austria).
Total GSH and total homocysteine in the liver and plasma
Plasma and liver Hcys were measured with modifications of a procedure described earlier(Reference Vester and Rasmussen41) using reversed-phase HPLC on a 100 RP-18e column (Agilent Technologies, Santa Clara, CA, USA). In brief, 10 μl of the internal standard mercaptopropionylglycine (200 μmol) per litre potassium borate buffer 1 (100 mm-trihydrogen borate and 2·0 mm-EDTA, pH 9·5) were added to 30 μl of undiluted plasma. The samples were mixed thoroughly. To reduce protein disulphide bonds and to liberate reduced thiols, 4·0 μl of a mixture of tri-N-butylphosphate–dimethylformamide (1:1, v/v) were added. After 30 min of incubation and centrifugation, protein precipitation was carried out by the addition of 25 μl of 0·6 m-perchloric acid containing 1·0 mm-EDTA. After 15 min of incubation at 20°C, the samples were centrifuged for 10 min at 10 000 g. Twenty microlitres of the supernatant were diluted with potassium borate buffer 2 (2·0 m-trihydrogen borate and 5 mm-EDTA, pH 10·5). Following thorough mixing, 20 μl of a 0·1 % (w/v) 7-fluorobenzofurazone-4-sulphonic acid ammonium salt solution in potassium borate buffer 3 (2·0 m-trihydrogen borate, pH 9·5) were added. After 60 min of incubation in a thermo mixer at 60°C, samples were cooled on ice, mixed and centrifuged for 10 min at 10 000 g. Sixty microlitres of the supernatant were pipetted into dark HPLC vials and put into the autosampler. The injection volume was 10 μl. The elution profile was as follows: 6·5 min eluent A (0·10 m-sodium acetate buffer containing 2 % (v/v) methanol, pH 4·0), 7·5 min eluent B (0·1 m-sodium phosphate buffer containing 5 % (v/v) methanol, pH 6·0), 3·0 min eluent B and 3·0 min eluent A. Fluorescence detection was carried out at an excitation wavelength of 385 nm and an emission wavelength of 515 nm. A calibrator plasma with known HCys and GSH concentrations was carried along to calculate sample concentrations of total GSH and HCys. For GSH and HCys determination in the livers, 1:5 (w/v) homogenates were prepared in ice-cold PBS. The remaining procedure was the same as described above for the undiluted plasma.
Cholesterol concentration in plasma and liver
Total Chol concentration in the liver and plasma was determined using the commercial test kits Ecoline®S+ from DiaSys Diagnostic Systems GmbH (Holzheim, Germany). Ten microlitres of undiluted plasma were used for Chol determination with the test kit. Before liver Chol analysis, total liver lipids were extracted with a hexane–isopropanol (3:2, v/v) mixture(Reference Hara and Radin42). Fifty-microlitre aliquots of these lipid extracts were evaporated under N2 gassing, and the remaining lipids were resolved in 50 μl of a Triton X-100–chloroform mixture (1:1, v/v)(Reference De Hoff, Davidson and Kritchevsky43). After evaporation of the chloroform, 10 μl of the Triton X-100 extracts were used for Chol determination with the test kit. Chol concentration in all the liver and plasma samples was measured in duplicate.
mRNA expression of glutathione peroxidase 1, glutathione S-transferase α 2, glycine-N-methyltransferase, cystathione-β-synthase, S-adenosylmethionine decarboxylase, betaine hydroxymethyltransferase, glutamate cysteine ligase, modulatory subunit, glutamate cysteine ligase, catalytic subunit and glutathione synthetase in the liver
Liver RNA was extracted with TRIzol reagent (Invitrogen, Karlsruhe, Germany) according to the manufacturer's protocol. Following reverse transcription of 3·0 μg of total RNA using a commercial complementary DNA synthesis kit (RevertAid™ First Strand synthesis kit; Fermentas, Vilnius, Latvia), the complementary DNA solutions were diluted 2:3 (v/v) with diethyl pyrocarbonate-treated sterile water for use in real-time detection PCR. For the amplification of specific sequences of the genes investigated, the RotorGene 6000™ real-time detection PCR apparatus (Corbett Research, Mortlake, NSW, Australia) was used.
Amplification data were analysed with the RotorGene 6000™ series software using the amplification efficiencies and the C t values(Reference Pfaffl44). The expression of the single genes was normalised to β-actin expression.
The primers used in PCR and their gene bank accession numbers are as follows: GPx1 (NM030826), primer forward (5′ → 3′): TCA TTG AGA ATG TCG CGT CT, primer reverse (5′ → 3′): CCC ACC AGG AAC TTC TCA AA (annealing temperature (T a): 55°C); glycine-N-methyltransferase (NM017084), primer forward (5′ → 3′): CCA CCG CAA CTA CGA CTA CA, primer reverse (5′ → 3′): TCT TCT TGA GCA CGT GGA TG (T a: 57°C); CBS (NM012522), primer forward (5′ → 3′): ATG CTG CAG AAA GGC TTC AT, primer reverse (5′ → 3′): GCG GTA TTG GAT CTG CTC AT (T a: 55°C); betaine hydroxymethyltransferase (BHMT; NM030850), primer forward (5′ → 3′): GCA CCA GCT TGC AGA CAA TA, primer reverse (5′ → 3′): TGT GCA TGT CCA AAC CAC TT (T a: 55°C); S-adenosylmethionine decarboxylase (NM031011), primer forward (5′ → 3′): CCC AGC AGT TAT GGA CCA GT, primer reverse (5′ → 3′) TCC ATC CGA TTT CAT TCC AT (T a: 55°C); glutathione synthetase (GS; NM012962), primer forward (5′ → 3′): AGA TGG CTA CAT GCC CAG TC, primer reverse (5′ → 3′): TGT CTT TCA GCT GCT CCA GA (T a: 57°C); glutamate cysteine ligase, catalytic subunit (GCLC; NM012815), primer forward (5′ → 3′): CCA CAA ACT GGC AGA CAA TG, primer reverse (5′ → 3′): TCC TTC CCA TTG ATG ATG GT (T a: 55°C); glutamate cysteine ligase, modulatory subunit (GCLM; NM017305), primer forward (5′ → 3′): AGG CAC CTC GGA TCT AGA CA, primer reverse (5′ → 3′): AAA TCT GGT GGC ATC ACA CA (T a: 55°C); glutathione S-transferase α 2 (NM031509), primer forward (5′ → 3′): GGG GAG AAA GAG GCA AGT CT, primer reverse (5′ → 3′): CTT CAG CAG AGG GAA GTT GG (T a: 57°C); sterol-regulatory element-binding protein 2 (SREBP2; NM001033694), primer forward (5′ → 3′): ATC CGC CCA CAC TCA CGC TCC TC, primer reverse (5′ → 3′): GGC CGC ATC CCT CGC ACT G (T a: 65°C); hydroxymethyl-glutaryl CoA reductase (HMGCoAR; NM013134), primer forward (5′ → 3′): AAG GGG CGT GCA AAG ACA ATC, primer reverse (5′ → 3′): ACG GCA CGG AAA GAA CCA TAG T (T a: 57°C); ATP-binding-cassette-transporter 8 (ABCG8; NM130414), primer forward (5′ → 3′): AGA CCC TCA CAC AGG ACA CC, primer reverse (5′ → 3′): CAG TCC GTC CTC CAG TTC AT (T a: 59°C); LDL receptor (LDLR; NM175762), primer forward (5′ → 3′): AGA ACT GCG GGG GCC GAA GAC AC, primer reverse (5′ → 3′): AAA CCG CTG GGA CAT AGG CAC TCA (T a: 60°C); β-actin (NM031144), primer forward (5′ → 3′): ATC GTG CGT GAC ATT AAA GAG AAG, primer reverse (5′ → 3′): GGA CAG TGA GGC CAG GAT AGA G (T a: 60°C).
Relative mRNA expression levels are expressed as x-fold changes compared with group C15.
Statistical analysis
The data were analysed by one-way ANOVA using the SPSS Statistical Package 14.0 for Windows (Chicago, IL, USA). Homogeneity of variances was tested using the Levene test. If variances were homogenous, differences of means (P < 0·05) were evaluated with the least significant difference test, if not the Games-Howell test was used.
Results
Diets and rat performance variables (diet consumption, live weight and feed efficiency)
The dietary Se concentrations scheduled in the experimental design were confirmed by Se analysis (C15, 14·8 (se 2·94); C50, 49·7 (se 3·73); C150, 156 (se 7·12); C450, 488 (se 9·51); M15, 16·2 (se 4·73); M50, 47 (se 4·02); M150, 149 (se 2·27); M450, 493 (se 14·65)). At the beginning of the experiment, body weight did not differ among the groups (Table 2). For rats receiving the Se-deficient diets either in combination with the recommended Met concentration or with a high Met concentration (C15 and M15), the total feed intake over 8 weeks and feed efficiency were significantly impaired compared with their respective Se-supplemented companions. In nearly all cases, the mean total feed intake of the rats with the high dietary Met concentration was significantly lower than that of the groups with the recommended Met concentration (C15 v. M15, C50 v. M50 and C450 v. M450). Corresponding to the improved feed conversion, all Se-supplemented rats of both dietary Met levels had significantly higher final body weights than their controls. Whereas the final body weight in the high Met Se-deficient rats (M15) was significantly lower than that in their companions with the recommended Met concentration (C15), no differences in the final body weight existed between the matching Se-supplemented groups at the other dietary Se levels (50, 150 and 450 μg/kg).
(Mean values with their standard errors)
a,b,c,d,e,f Mean values within a row with unlike superscript letters were significantly different (P < 0·05).
* For details of diets and procedures, see Methods and materials.
Selenium status, variables of the selenium- and glutathione-dependent redox system in the liver and plasma, and homocysteine concentration in the liver and plasma
The liver Se concentration in the Se-deficient rats of groups C15 and M15 was reduced to 1·7 and 1·5 % of the concentrations in groups C450 and M450 (Table 3). At both dietary Met levels, the gradual increase in Se supplementation from 50 to 450 μg/kg increased the liver Se concentration approximately twofold in each case. With the exception of groups C150 and M150, the liver Se concentration in the other corresponding groups (C15 v. M15, C50 v. M50 and C450 v. M450) was significantly higher in the C groups with the recommended dietary Met concentration. At both Met levels with Se deficiency (C15 and M15), GPx1 mRNA expression was approximately fivefold lower than that in the Se-supplemented rats (C50, C150, C450, M50, M150 and M450). Rats supplemented with 50 μg Se/kg (C50 and M50) already had a GPx1 mRNA expression, which was comparably as high as that in rats supplemented with 150 and 450 μg Se/kg. No influence of the dietary Met level on GPx1 mRNA expression could be analysed. In contrast to GPx1 expression, the expression of the phase II enzyme glutathione S-transferase (glutathione S-transferase α 2), another sensitive indicator of Se deficiency, was 1·5- to 2·5-fold higher in Se-deficient rats than in their Se-supplemented companions. Liver GPx1 activity in C15 and M15 rats was reduced to approximately 1 % of the activity of groups C150, C450, M150 and M450. Liver GPx1 activity of rats of groups C50 and M50 was 40–60 % lower than that of their companions supplemented with the recommended Se amount (C150 and M150) or with slightly supranutritional Se supplementation (C450 and M450). As observed for the liver Se concentration, the liver GPx1 activity of rats with the high dietary Met supplementation was lower than the activity of their companions fed the diets with the recommended dietary Met concentration. However, with the exception of groups C50 and M50, this effect was not statistically significant. Plasma Se concentration in C15 and M15 rats was decreased to 2·5 and 4·3 % of the concentration in groups C450 and M450. Plasma Se concentration gradually and significantly increased when the dietary Se concentration was raised from 50 to 450 μg/kg. In contrast to the liver Se concentration, plasma Se concentration was not influenced by the dietary Met concentration. GPx3 activity in the Se-deficient groups C15 and M15 was reduced to approximately 3 % of the activity in groups C450 and M450. In groups C50 and M50, GPx3 reached an activity of approximately 70 % of that in the respective groups receiving diets containing the recommended Se concentration (C150 and M150) or a slightly supranutritional Se supplementation (C450 and M450). In the Se-deficient groups C15 and M15, plasma GPx3 activity dropped to 3·0–3·5 % of that in groups C150 and M150 supplemented at the recommended level. However, in contrast to liver GPx1 activity, plasma GPx3 activity in the Se-supplemented rats of groups M50 and M150 fed the high dietary Met concentration was significantly higher than that in their companions in groups C50 and C150 on diets with the recommended Met level. For groups M450 and C450, a distinct trend (P = 0·061) existed with regard to this effect. At both dietary Met levels, total liver GSH concentration was significantly higher in C15 and M15 rats than in the respective Se-supplemented groups. Independent of the dietary Se concentration, rats receiving the diets with the high Met concentration had a significantly higher liver GSH concentration than the corresponding groups receiving diets with the recommended dietary Met level. Liver HCys concentration was 1·4- to 1·5-fold higher in Se-deficient rats of groups C15 and M15 than in their Se-supplemented companions. In contrast to the effect observed for GSH, the higher dietary Met concentration had no influence on liver HCys concentration with the exception of groups C50 and M50. Whereas both the liver GSH concentration and the liver HCys concentration decreased in a dose-dependent manner due to increasing Se supplementation at both dietary Met levels, a completely different regulation profile of the plasma concentration of both compounds could be observed. At both dietary Met levels, increasing the Se supplementation of the diets led to a reduction of total plasma GSH concentration in a dose-dependent manner. Thus, the lowest plasma GSH values could be measured in groups C450 and M450. Moreover, rats receiving the diets with the high Met concentration (groups M15, M50, M150 and M450) had in each case a significantly lower plasma GSH concentration than the respective C groups C15, C50, C150 and C450. In contrast, total plasma HCys concentration was the lowest in the Se-deficient group C15 with the recommended dietary Met level. Se addition to the diets approximately doubled plasma HCys levels. However, a dose-dependent effect of the different dietary Se levels on plasma HCys concentration could not be analysed. The addition of the high dietary Met concentration to the diets (rats of groups M15, M50, M150 and M450) again approximately doubled plasma HCys concentration in comparison to the groups C15, C50, C150 and C450. As analysed in the C groups, also in the M groups, a significant increase in plasma HCys concentration existed only between the Se-deficient group M15 and all groups with Se supplementation (M50, M150 and M450), whereas an increase in dietary Se supplementation did not lead to a further rise in plasma HCys level. On the contrary, in M450 rats, even a small but significant decline in plasma HCys concentration could be analysed compared with groups M50 and M150 with lower Se supplementation.
(Mean values with their standard errors)
prot., Protein.
a,b,c,d,e Mean values within a row with unlike superscript letters were significantly different (P < 0·05).
* For details of diets and procedures, see Methods and materials.
mRNA expression of key enzymes of homocysteine and glutathione metabolism
To investigate the molecular causes underlying the present results for liver and plasma GSH and HCys concentrations, we analysed the mRNA expression of the key enzymes of GSH and HCys metabolism in the liver of the rats (Fig. 1(a)–(g)). Glycine-N-methyltransferase expression was affected neither by the dietary Se concentration nor by the dietary Met level (Fig. 1(a)). In the C groups with the recommended dietary Met concentration, liver CBS expression in the Se-supplemented rats was 38–43 % lower than that in the Se-deficient group. However, no significant differences in CBS gene expression occurred between rats with different dietary Se supplementations (C50, C150 and C450). The high Met concentration in the diet did not significantly change the CBS expression level in the Se-deficient group M15 compared with group C15. In Se-supplemented rats fed the high Met diet, CBS expression was further reduced to 22–35 % of that in C15 rats (17–30 % of that in M15 rats). Thus, Se-supplemented rats with high Met diets (M50, M150 and M450) had in each case a significantly lower liver CBS expression than groups C50, C150 and C450 (Fig. 1(b)). A very similar expression profile as analysed for CBS could also be observed for liver S-adenosylmethionine decarboxylase expression. S-adenosylmethionine decarboxylase expression in the Se-supplemented rats was reduced to approximately 40 % (C groups) and 5–20 % (M groups) of that measured in their respective Se-deficient companions in groups C15 and M15 (Fig. 1(c)). In the C groups, BHMT responsible for betaine-dependent remethylation of HCys to Met was increased in the Se-supplemented groups C50, C150 and C450 up to 1·6-fold compared with group C15. With the exception of groups C150 and M150, BHMT expression was slightly but not significantly higher in all M groups compared with their respective C groups (Fig. 1(d)). The key enzymes of the GSH biosynthesis pathway were also strongly influenced by the dietary Se and Met concentrations. The expression of the GCLM was reduced by Se supplementation to 30–60 % of that in the Se-deficient rats (C15 and M15), with the lowest expression in group M450 (Fig. 1(e)). In contrast to the expression of GCLM, the expression of the GCLC was in addition strongly regulated by the dietary Met concentration. Whereas the expression of liver GCLC in the C groups was reduced to 40–55 % of that in Se-deficient C15 rats by Se supplementation at all the dietary levels investigated (C50, C150 and C450), the expression of the enzyme in the Se-supplemented groups with high dietary Met (M50, M150 and M450) was reduced to 4–27 % of that in C15 rats and to 5–30 % of that in M15 rats. Moreover, in Se-supplemented high Met rats of groups M50, M150 and M450, a significant influence of the dietary Se concentration on GCLC expression could be analysed (Fig. 1(f)). GS expression in the liver was the highest in the Se-deficient C15 rats. Independent of the Se concentration in groups C50, C150 and C450, a reduction of GS expression to 50–70 % of the level in group C15 was analysed. The addition of the higher Met level to the diets drastically reduced GS expression. Thus, in M15 rats, GS expression was only 30 % of that in C15 rats (Fig. 1(g)).
(a)–(g) mRNA expression of the key enzymes of homocysteine metabolism ((a)–(d)) and GSH metabolism ((e)–(g)) of rats fed diets containing different dietary concentrations in combination with two dietary methionine levels (values are means with their standard errors). a,b,c,d,e Mean values with unlike letters were significantly different (P < 0·05). n 8 animals are considered. For details of diets and procedures, see Methods and materials. GLNMT, glycine-N-methyltransferase; CBS, cystathione-β-synthase; SAMDC, S-adenosylmethionine decarboxylase; BHMT, betaine hydroxymethyltransferase; GCLM, glutamate cysteine ligase, modulatory subunit; GCLC, glutamate cysteine ligase, catalytic subunit; GS, glutathione synthetase. (a)–(g) 
, C groups; 
, M groups.
Cholesterol concentration in the liver and plasma, and mRNA expression of genes involved in the regulation of cholesterol metabolism
Both Se-deficient groups (C15 and M15) had a 18–27 % (C groups) and a 37–44 % (M groups) lower liver Chol concentration than the respective Se-supplemented groups C50, C150, C450, M50, M150 and M450 (Table 4). At both dietary Met concentrations, the rats with the highest dietary Se supplementation of 450 μg/kg (C450 and M450) had a slightly higher liver Chol concentration than the other Se-supplemented groups. However, at the same dietary Met level, no significant differences in the liver Chol level existed between the Se-supplemented groups. With the exception of the Se-deficient groups C15 and M15 and groups C50 and M50, the Se-supplemented groups M150 and M450 had a significantly higher liver Chol concentration than their companions in groups C150 and C450 receiving diets with the recommended dietary Met concentration. In contrast to the liver Chol concentration at both dietary Met levels, plasma Chol concentration in the Se-supplemented groups C50, C150, C450, M50, M150 and M450 was 15–25 % lower than that in the Se-deficient groups. The gene expression of enzymes involved in the regulation of whole body Chol metabolism showed a differential expression pattern (Table 4). No significant differences in gene expression could be analysed for the transcription factor SREBP2 between all the experimental groups. The expression of HMGCoAR was also not affected by the different dietary conditions. The expression of liver LDLR was the lowest in group C15 with Se deficiency and the recommended dietary Met level. All the Se-supplemented rats receiving diets with the recommended dietary Met concentration (C50, 150 and C450) had an approximately 2·5-fold higher expression of liver LDLR than the C15 rats. The rats of the M groups had a 1·5- to 2·0-fold higher expression of liver LDLR than their companions in the respective C groups (M15 v. C15, M50 v. C50, M150 v. C150 and M450 v. C450). The hepatobiliary Chol exporter ABCG8 showed the highest expression level in the Se-deficient M15 rats. The second highest expression was analysed in the corresponding Se-deficient C15 group fed diets containing the recommended dietary Met level. Se supplementation reduced the expression of ABCG8 to approximately 50 % (groups C50, C150 and C450) and 30 % (groups M50, M150 and M450) of the level in the corresponding Se-deficient groups C15 and M15.
(Mean values with their standard errors)
prot., Protein.
a,b,c,d,e Mean values within a row with unlike superscript letters were significant different (P < 0·05).
* For details of diets and procedures, see Methods and materials.
Discussion
Rat performance parameters and selenium status
In the present study, we investigated for the first time the combined effect of four dietary Se concentrations and two Met concentrations on the metabolism of HCys and Chol in rats. In accordance with our previous studies with rats on the effects of Se on the regulation of protein tyrosine phosphatase-1B and TAG metabolism(Reference Mueller, Klomann and Wolf45, Reference Mueller, Bosse and Most46), our current data confirm a reduced feed conversion of Se-deficient rats at both dietary Met levels tested. With regard to Met and HCys metabolism, other dietary factors (e.g. folate, vitamin B12 and choline) influencing these pathways remained constant in the present experiment. The parameters of Se status (liver and plasma Se, liver GPx1 and plasma GPx3) confirm that the basal diets C15 and M15 produced a marked Se deficiency, and that the graded amounts of Se added to the other diets led to a stepwise increase in liver and plasma Se concentrations. However, the plateau in GPx1 expression and activity as well as in GPx3 activity was definitely reached with the recommended dietary Se level of 150 μg/kg(Reference Barnes, Evenson and Raines47, Reference Bosse, Pallauf and Hommel48). The phase II enzyme glutathione S-transferase α 2, another sensitive indicator of Se deficiency, also indicated Se-deficient conditions of groups C15 and M15 by its higher expression in Se deficiency(Reference Bosse, Pallauf and Hommel48).
Influence of selenium and methionine on GSH and homocysteine metabolism
In accordance with studies on rats investigating the influence of Se status on HCys metabolism(Reference Uthus, Yokoi and Davis18–Reference Uthus and Ross20), we also found a doubling of total plasma HCys concentration in the Se-supplemented groups at both dietary Met levels compared with the respective Se-deficient groups C15 and M15. An interesting finding of the present study is that the addition of 15 g Met/kg diet (M15, M50, M150 and M450) again doubled plasma HCys concentration compared with the respective groups with the recommended dietary Met amount (C15, C50, C150 and C450). In this context, it is noteworthy that the HCys ratio between Se-deficient rats (C15 and M15) and Se-supplemented rats of the respective dietary Met level remained nearly constant at 1:2. The fact that dietary Met concentration is an important factor that influences plasma HCys concentration was reported for rats as well as for human subjects through an increased dietary Met intake by the consumption of a high protein diet, or subjected to oral Met loading(Reference Hirche, Schröder and Knoth28, Reference Hirche, Schröder and Knoth29, Reference Verhoef, van Vliet and Olthof32, Reference Hart, Mangoni and Swift34). The inclusion of two dietary Met levels in our experimental design, the measurement of liver and plasma HCys and GSH concentrations, and the measurement of the expression of an increased spectrum of the key enzymes of HCys and GSH homoeostasis contribute to an improved understanding of HCys metabolism in rats (Table 2 and Fig. 1). In these studies, the main conclusion explaining the higher plasma HCys concentration was deduced from the fact that Se-deficient rats had a higher expression and activity of the key enzymes of the GSH biosynthesis pathway in the liver (confirmed in the present study by GCLM, GCLC and GS expression) than their Se-supplemented companions, leading to a ‘pull’ on transsulphuration, and consequently, to lower plasma HCys levels in Se-deficient rats. However, in these studies, liver HCys concentration was not analysed(Reference Uthus, Yokoi and Davis18, Reference Uthus and Ross20, Reference Uthus and Ross21). Additional analysis of glycine-N-methyltransferase expression (not influenced by Se status and dietary Met level), CBS expression (strongly reduced in Se-supplemented C and M groups compared with the Se-deficient C15 and M15 rats) and BHMT expression (slightly increased in Se-supplemented C and M rats compared with their Se-deficient companions) and of the liver HCys concentration (reduced by Se supplementation in C and M rats) in the present study gives rise to some new hypotheses regarding the regulation of the HCys metabolism in rats. In comparison to previous studies on rat GSH or HCys metabolism, the present results suggest the following analogies and new aspects:
(1) In accordance with previous work on rat GSH and HCys metabolism(Reference Uthus, Yokoi and Davis18–Reference Uthus and Ross21, Reference Hill, Burk and Lane49–Reference Hill and Burk51), our data confirmed that both the plasma and liver GSH concentrations were increased in Se deficiency, and may result from a pull on the transsulphuration pathway (liver GSH) and a higher export into the plasma (plasma GSH). In these earlier studies, plasma GSH values in Se-deficient rats were nearly twofold increased compared with controls with adequate Se supply(Reference Hill, Burk and Lane49, Reference Hill and Burk50). In the present experiment at both dietary Met concentrations, the reduction of plasma GSH values to half of that in the Se-deficient groups was obtained with a Se supplementation of 450 μg/kg (groups C450 and M450). The increased expression levels of CBS as well as of GCLC, GCLM and GS in Se deficiency support these findings.
(2) Contrary to expectations, the highest expression of CBS in the Se-deficient groups C15 and M15 and almost unchanged expression levels of glycine-N-methyltransferase and BHMT compared with the Se-supplemented groups produced the highest liver HCys concentrations. Therefore, this effect seems to be the consequence of a higher liver HCys disposal under Se-deficient conditions and an increased HCys export into the plasma by Se supplementation and particularly by a higher dietary Met intake rather than deriving from the reduction of the key enzymes of GSH biosynthesis. The dependency of HCys export into the plasma on dietary Met concentration was impressively demonstrated in a rat study(Reference Stead, Brosnan and Brosnan52).
(3) However, in rats, the Se-dependent manipulation of HCys export into the plasma only responds to switching over from Se deficiency to Se supply. A further increase in plasma HCys concentration due to an increase in dietary Se supplementation and plasma Se values cannot be observed. This particular aspect may explain the contradictory results for the interaction of Se status and plasma HCys level in rats(Reference Uthus, Yokoi and Davis18–Reference Uthus and Ross21) and human subjects(Reference González, Huerta and Alvarez-Uría22–Reference Bekaert, Cooper and Green27), since in none of the human studies was a group with severe Se deficiency included. However, in the human studies, in which an inverse relationship between plasma Se and plasma HCys has been identified, the groups with an increased plasma HCys concentration had very low plasma Se values (0·56–0·79 μmol/l)(Reference González, Huerta and Alvarez-Uría22–Reference Bélanger, Dewailly and Berthiaume25) than those in the studies in which Se status or Se supplementation remained without an effect on plasma HCys concentration (>1·0 μmol/l)(Reference Venn, Grant and Thomson26, Reference Bekaert, Cooper and Green27). In this context, it is of interest that the summation of plasma and liver HCys concentrations from the present experiment leads to an inverse relationship between Se status and the summated HCys concentration (Table 2). Moreover, independent of the dietary Met concentration, nearly identical summated HCys concentrations can be observed for the different dietary Se concentrations (C15 v. M15, C50 v. M50, C150 v. M150 and C450 v. M150) (Table 2). Thus, future investigations into the influence of Se, Met or other nutrients on HCys metabolism should consider species differences in hepatic HCys export, and investigate the influence of nutrients on specific carriers involved in HCys export(Reference Blom53).
(4) Another interesting finding of the present study is that it was not only liver HCys disposal that was reduced by Se and Met. The Se- and above all Met-dependent distinct down-regulation of the polyamine-synthesising enzyme S-adenosylmethionine decarboxylase seems to have a protective function for the organism since polyamines are discussed as being involved in the generation of a number of diseases from cancer and psoriasis to parasitic infections(Reference Wallace and Fraser54).
Influence of selenium and methionine on cholesterol metabolism
In the present study, we also aimed to elucidate the molecular links between the amount of dietary Se and Met supplementation and Chol metabolism. In the literature, there exists inconsistent information on the effects of both nutrients on Chol metabolism. Thus, for Se, a large recent US cross-sectional observational trial with 7129 participants reported a distinct positive correlation between serum Se concentration and serum Chol levels. In the present study, participants in the top quartile (serum Se >1·70 μmol/l) had 8 % higher serum Chol levels, 8 % higher LDL Chol levels and 5 % higher HDL Chol levels than participants in the bottom quartile (serum Se < 1·43 μmol/l)(Reference Bleys, Navas-Acien and Stranges35). The data of a very recent cross-sectional observational trial from the UK(Reference Stranges, Laclaustra and Ji36) including 1042 white female and male adults aged 19–64 years confirm the findings of the above-mentioned US study. Despite the overall somewhat lower plasma Se concentrations (bottom quartile: < 0·98 μmol/l, top quartile: 1·20–2·79 μmol/l) compared with the US study, the participants of the top quartile had 11 % higher total plasma Chol values and 12 % higher LDL Chol values than those in the bottom quartile(Reference Bleys, Navas-Acien and Stranges35, Reference Stranges, Laclaustra and Ji36). Another study from Lebanon confirms these results. A higher Se status correlated significantly and positively with a number of markers of the metabolic syndrome including plasma Chol(Reference Obeid, Elfakhani and Hlais55). In contrast, a study from Japan in pre- and postmenopausal women found a decrease of Se status and an increase in plasma Chol level after menopause, and suggested a beneficial effect of Se supplements in lowering Chol values(Reference Karita, Yamanouchi and Takano56). Several studies with rats with a dietary induced hypercholesterolaemia found that Se supplementation with 1·0 mg/kg diet clearly reverses hypercholesterolaemia by lowering liver apo B and HMGCoAR expression and by increasing LDLR expression(Reference Dhingra and Bansal57–Reference Dhingra and Bansal59). A similarly unclear effect as found for Se on Chol metabolism also exists for Met and HCys on Chol metabolism.
In this context, a hypothesis report(Reference Obeid and Herrmann60) and a study with transgenic mice(Reference Zhao, Su and Jacobs61) suggest that a high dietary Met concentration increases the need for phosphatidylethanolamine (PE)-N-methyltransferase for the formation of HCys. Through this increased use of PE-N-methyltransferase for HCys metabolism, the reaction actually catalysed by the enzyme, namely the conversion of PE to phosphatidylcholine, is decelerated. Phosphatidylcholine is again an important phospholipid for the formation of VLDL particles. Therefore, high Met and HCys may increase lipid storage in the liver, and contribute to the development of non-alcoholic fatty liver disease. Moreover, high HCys concentrations may lower liver HDL synthesis, and therefore reduce reverse Chol transport(Reference Obeid and Herrmann60). A study with transgenic PE-N-methyltransferase − / − mice showed that these animals had a reduced liver VLDL secretion than their PE-N-methyltransferase +/+ companions(Reference Zhao, Su and Jacobs61). Accordingly, Met loading in the elderly led to a fourfold increase in plasma HCys concentration and a 22·5 % increase in total Chol:HDL ratio in the plasma(Reference Hart, Mangoni and Swift34). In a study with growing rats, the gradual increase of the dietary Met concentration from 2·6 to 6·0 g/kg diet led to the stepwise increase in plasma HCys and in both plasma and liver Chol. In the present study, the most distinct increase in plasma and liver Chol was analysed between rats consuming the low Met diet (2·6 g/kg) and rats fed the next higher dietary Met level of 3·5 Met/kg diet. A by far lower and NS increase in plasma and liver Chol was observed between rats consuming the other diets containing up to 6·0 g Met/kg diet. An increased expression of the SREBP2 and of HMGCoAR was analysed as the molecular cause for the increasing effect of Met/HCys on liver and plasma Chol. However, the expression of the liver LDLR, another target gene of SREBP2, showed no response to the different Met concentrations(Reference Hirche, Schröder and Knoth29). In another study with tissue cultures and mice, it was postulated that the increased liver Chol synthesis by HCys derives from the activation of lipogenic SREBP transcription factors and VLDL secretion by endoplasmatic reticulum stress(Reference Werstuck, Lentz and Dayal62). In contrast to the above-mentioned studies, describing a Chol-increasing effect of Met supplementation to diets, Chol-lowering effects of Met have also been investigated. Thus, in a rat study, the addition of 3·0 g Met/kg to a 10 % soya protein diet effected a 40 % reduction of plasma Chol, whereas the Met addition to a 20 % soya protein diet effected a small and NS increase in plasma Chol. A significant reduction of liver HMGCoAR activity and a significant increase in Chol-7-α-hydroxylase activity, the key enzyme for bile acid synthesis, were analysed as the molecular causes for the Chol-lowering effect of Met(Reference Taniguchi, Nagao and Inoue63). In another study, rats with an implanted ascites hepatoma cell line developed severe hypercholesterolaemia, based on the reduction of bile acid excretion. Met addition to the diets restored Chol metabolism by increasing Chol excretion via the bile(Reference Kawasaki, Funabiki and Yagasaki64).
The comparison of our data with the data from the above-mentioned studies on the role of Se and Met in Chol metabolism indicates both analogies and inconsistencies. In accordance with the studies in which hypercholesterolaemia in rats was induced by feeding high Chol diets, our data confirm the up-regulation of the liver LDLR and the decrease of plasma Chol by Se. In contrast to these studies, we could not find a reduction of HMGCoAR expression by Se. This difference may result from a higher Chol loading of the livers in the cited studies by feeding of high Chol diets in combination with an up-regulated LDLR. However, liver Chol data were not provided in these studies(Reference Dhingra and Bansal57–Reference Dhingra and Bansal59). When the free liver Chol concentration exceeds a critical value, the processing of the transcription factor SREBP2 by the SREBP cleavage-activating protein is blocked via an up-regulation of the insulin-induced genes 1 and 2 (Insig 1 and 2) which bind to SREBP cleavage-activating protein(Reference Yabe, Brown and Goldstein65). Despite significantly higher liver Chol concentrations in the Se-supplemented C and M groups than in the respective Se-deficient groups, the feeding of Chol-free diets in the present study obviously did not result in a liver Chol concentration switching on the Insig–SREBP cleavage-activating protein–SREBP2 system. In this context, it should be mentioned that we could not detect any differences in the expression of Insig 1 and 2 (data not presented). A further new result in the present study was the finding that the hepatobiliary Chol exporter ABCG8, which also plays an important role in the reverse transport of phytosterols into the lumen of the small intestine(Reference Kidambi and Patel66, Reference Duan, Wang and Wang67), was distinctly down-regulated in all Se-supplemented rats, and therefore provides a further explanation for the higher Chol concentration in the livers of Se-supplemented rats and for their lower plasma Chol levels. In contrast to the previous rat studies(Reference Dhingra and Bansal57–Reference Dhingra and Bansal59) and to the present results, recent human studies found a positive correlation between a high Se status and an increased plasma Chol concentration(Reference Bleys, Navas-Acien and Stranges35, Reference Stranges, Laclaustra and Ji36). One possible explanation for this difference is the fact that in the rat studies, young and still growing animals were used, whereas the human data were collected from adults. This fact raises the question whether the long-term uptake of Se above the recommendations via permanent Chol accumulation in the liver would ultimately also increase plasma Chol values. Thus, long-term studies with rats or better with pigs, which have a lipoprotein metabolism which is more closely related to that of human subjects, are needed to confirm or confute the risk of Se supplementation for increased plasma Chol values. In accordance with previous rat studies(Reference Hirche, Schröder and Knoth29, Reference Woo, Siow and Pierce68), in the present study, a positive and significant correlation also existed between the dietary Met level and the liver Chol concentration. However, in contrast to these studies(Reference Hirche, Schröder and Knoth29, Reference Woo, Siow and Pierce68) and to the human study(Reference Hart, Mangoni and Swift34) examining the effect of Met loading on Chol metabolism, we could not find a significant increase in plasma Chol concentration by increasing the dietary Met concentration to fivefold of the recommended amount. Similarly as discussed for Se, it is conceivable that feeding adult rats a high Met diet in combination with Chol addition to the diet also increases plasma Chol concentration. However, the present results showing the up-regulation of the LDLR by Se and Met and the additional down-regulation of ABCG8 by Se suggest that the over-supplementation with both nutrients may rather increase the risk of the early development of non-alcoholic fatty liver disease. Another very interesting issue regarding ABCG8 regulation which should be examined in the future involves the function of this transporter in the small intestine where it is selectively involved in the reverse transport of phytosterols but not of Chol back into the lumen. In the case of a phytosterol-rich diet, this particular aspect would suggest a positive influence of a high Se status on plasma Chol level, since a reduced export of phytosterols back to the intestinal lumen would lead to a replacement of Chol by phytosterols.
Conclusions
The present rat study investigating the combined effects of Se and Met on HCys and Chol metabolism reveals the following new aspects:
(1) Se increases plasma HCys concentration only when Se-deficient conditions are compared with Se supplementation, independent of the dietary Met concentration. Already at a suboptimal Se supply of 50 μg/kg diet, the plasma HCys level reaches a plateau concentration. When compared with Se-deficient conditions, the elevation of plasma HCys level by dietary Se does not only result from the reduction of the liver key enzymes of GSH biosynthesis and the subsequent deceleration of the transsulphuration pathway. Se also reduces liver HCys disposal by an increased export into the plasma. Due to the fact that plasma HCys concentration can only be reduced by severe Se deficiency, which is not desirable in human subjects, in our opinion, it can be assumed that Se supplementation is not clearly a risk factor for homocysteinaemia in human subjects. Moreover, our rat model suggests that the summation of plasma and liver HCys concentration results in an inverse relationship between Se status and the HCys concentration, as was similarly reported from some human trials. A high dietary Met concentration, however, is a substantial risk factor for homocysteinaemia. In summary, the rat, in particular when fed diets with a high Met concentration and sufficient Se, may represent a useful model for further critical study of the controversial role of HCys as a risk factor for atherosclerosis, CVD and ischaemic stroke.
(2) Se increases liver Chol via two mechanisms, the concentration-dependent up-regulation of the liver LDLR and the down-regulation of the hepatobiliary Chol exporter ABCG8. A high dietary Met concentration amplifies the Se effects on liver Chol concentration. The present results do not support the findings from recent human trials in which a high Se status corresponded to elevated levels of total Chol and LDL Chol in plasma, but point rather to the development of non-alcoholic fatty liver disease by the long-term over-supplementation with Se and Met.
However, future investigations with animal models should examine the effects of Se supplementation on Chol metabolism when diets with a high energy and Chol concentration are consumed, as is the case in industrial nations. In this context, the role of phytosterols should also be addressed. Furthermore, the pig may represent a better animal model for future studies on Se and Chol metabolism, since its lipoprotein metabolism is more closely related to that of human subjects than to that of the rat.
Acknowledgements
All the authors declare that they have no conflicts of interest, and that they have significantly contributed to the paper. The present research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. N. M. W. conducted the rat study, prepared the samples, and participated in the analyses of HCys, GSH and enzymes. She was further involved in statistical data analysis and in writing of the manuscript. K. M. carried out RNA extraction and reverse transcription, and supervised real-time PCR experiments. Moreover, she evaluated the real-time PCR data. F. H. supervised HCys and GSH analyses, and evaluated the relevant data. E. M. supervised Se analysis and evaluated the relevant data. J. P. provided laboratory support at the Justus Liebig University of Giessen for the rat nutrition trial and for some of the analyses. He also helped in compiling the manuscript. A. S. M. designed the study, was involved in the gene expression analyses, checked the statistical evaluation, prepared the tables and figures, and has written large passages of the manuscript. We thank our Masters students Doris Lippmann and Anne Cathrin Seltmann for their help with analyses within the scope of their Masters theses.
