DNA damage and oxidative stress response to selenium yeast in the non-smoking individuals: a short-term supplementation trial with respect to GPX1 and SEPP1 polymorphism

By means of advertisement in different public places, 517 residents of Lodz in the age 18–60 were recruited for the study. Questionnaire data and 4 mL of fasting blood were collected from the subjects, followed by DNA isolation and prospective genotyping for GPX1 rs1050450 and SEPP1 rs3877899. In a whole group of 517 recruited subjects, GPX1 allelic variants were in Hardy–Weinberg equilibrium (p = 0.582), and the genotype distribution was as follows: 46, 43, and 11 % for ProPro, ProLeu, and LeuLeu, respectively. SEPP1 genotype frequencies were 60, 32, and 7.5 % for AlaAla, AlaThr, and ThrThr, respectively, with a slight deviation from Hardy–Weinberg equilibrium (p = 0.013). The exclusion criteria for the supplementation trial included current smoking, chronic diseases such as cancer, diabetes, or cardiovascular disease, BMI higher than 35 as well as incomplete data for genetic analyses. Altogether 137 subjects were excluded. Further selection of the remaining 380 volunteers was based on the approximate major and minor allele distribution in order to obtain an as high as possible number of individuals with rare genotypes (GPX1 Leu/Leu and SEPP1 Thr/Thr) and to increase statistical power of comparison tests. Namely, all the subjects with rare genotypes were asked to participate in the supplementation trial, whereas subjects with common alleles were matched for age and BMI. Altogether 95 subjects agreed to participate in the trial. The study group characteristics are presented in Table 1. Genotype distribution is shown in Table 2, whereas Figure S1 presents the numbers of subjects with different genotype combinations. The selected 95 subjects, including 43 men and 52 women at the mean age of 35.6 years, were receiving 200 µg of selenium in a form of selenium yeast for 6 weeks. Se yeast tablets were obtained commercially, and all were from the same batch. During the supplementation trial, as well as during the washout period, the participants were asked not to take any supplements containing vitamins and selenium or other elements. Fasting blood was collected during the study at four time points: before the supplementation assigned as baseline (the first day of the trial, before taking the first tablet), after 2 and 6 weeks of supplementation, and after 4 weeks of the washout period. Blood samples were collected from each participant in two heparin tubes and two EDTA tubes. The material collected in the first heparinized tube (7.5 mL) was fractioned by centrifugation into plasma, buffy coat, and erythrocytes and frozen at −20 °C until biochemical analyses. Whole blood (4.5 mL) from the second heparin tube was used for oxidative burst assessment, which was performed at the day of blood collection (this analysis was performed only at two time points: at baseline and after 2 weeks of supplementation). EDTA whole blood (1.5 mL) was used for leukocyte lysate preparation, which was subsequently stored at −70 °C until mRNA isolation (for gene expression analysis). Whole blood (2.7 mL) from the second EDTA tube was used to prepare agarose slides for the comet assay (this analysis was performed at three time points: at baseline, after 6 weeks of supplementation, and after 4 weeks of the washout period). Along with the blood collection, detailed questionnaire data were collected each time, concerning current health status (well-being with respect to adverse health effects), intake of medications, dietary supplements, herbs or pharmacological treatment with hormones, antibiotics, statins, and other. All the analyses, apart from the oxidative burst assessment, were conducted after completing the study, and all the samples were blinded in terms of genotype and the time point of their collection. All the study participants gave written informed consent, and the subjects who were enrolled for the supplementation trial received detailed, written, and oral information about the trial. The study was approved by the Local Ethics Committee (Ethical Institutional Review Board at the Nofer Institute of Occupational Medicine, Lodz, Poland).

DNA was isolated from buffy coat, using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). Allelic discrimination for the two studied polymorphisms was performed using the real-time PCR method and CFX96™ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). GPX1 rs1050450 genotyping was conducted using the high-resolution melt curve (HRM) technique. Oligonucleotide sequences for PCR primers, designed by Beacon Designer™ (PREMIER Biosoft, Palo Alto, CA, USA), were as follows: 5′-GCCGCTTCCAGACCATTG-3′ (forward) and 5′-GGTGTTCCTCCCTCGTAG-3′ (reverse). Data analysis was performed using Bio-Rad CFX Manager and Bio-Rad Precision Melt Analysis Software. Identification of particular genotypes recognized by HRM was based on the comparison with the method introducing specific fluorescent probes, as described previously [27]. For SEPP1 rs3877899 analysis, the TaqMan^® SNP Genotyping Assay (C_8709053_10, Life Technologies, Life Technologies, Carlsbad, CA, USA) was used. The call rate for two SNPs was >99 %, and the concordance of the blinded QC (n = 12) was 100 %.

Plasma sample used for determination of selenium concentration was initially diluted with Triton X-100 (0.2 %, Sigma-Aldrich, St. Louis, MO, USA) in a proportion 1:1. Concentration was assessed using Zeeman flameless atomic absorption spectrometry (instrument Z-5000, Hitachi) with graphite tube and palladium/magnesium matrix modifier (Pd—1500 ppm, Mg—900 ppm). The limit of detection was 1.4 µg/L. The certified reference material Seronorm Trace Elements Serum (SERO AS, Billingstad, Norway) was used for the quality control assessment. The accuracy of the method and coefficient of variance (CV) were 1.3 and 4.7 %, respectively.

Sepp1 concentration was determined in plasma using the immunochemical Sandwich ELISA test (USCN Life Science Inc kit, Hu, China). In brief, plasma samples were diluted 500–1000 times and added to the plate wells with biotin-conjugated polyclonal antibody specific for Sepp1. Next, avidin-conjugated horseradish peroxidase was added, and after incubation with TMB (3,3′,5,5′-tetramethylbenzidine), the substrate solution color change was observed proportionally to Sepp1 concentration. The reaction was terminated with sulfuric acid solution, and color intensity was measured at a wavelength of 450 nm. For each test, a standard curve was determined (range 0.78–50 ng/mL), and Sepp1 concentration was calculated after plotting logarithmic curve. Intra-assay and inter-assay variations were 8.0 and 12.7 %, respectively.

Spectrophotometric methods were used to analyze blood compartments for the activity of GPx1, GPx3, SOD1, and Cp. The activity of glutathione peroxidases was determined in erythrocytes (GPx1) and plasma (GPx3) using the method of Paglia and Valentine [28] with t-butyl hydroperoxide as a substrate and following the rate of NADPH oxidation by the coupled reaction with glutathione reductase. The rate of decrease in the absorbance, being proportional to the GPx activity, was read at a wavelength of 340 nm. The activity of SOD1 was determined in erythrocytes by the use of the method of Beauchamp and Fridovich [29], which relies on the inhibition by SOD1 of the reduction of nitro blue tetrazolium (NBT) by xanthine and xanthine oxidase. The concentration of the reduced form of NBT was measured at a wavelength of 540 nm. The oxidase activity of Cp was determined in plasma according to the method described by Sunderman and Nomoto [30], with a PPD (p-phenylenediamine) as a substrate. The absorbance of the oxidation product was read at a wavelength of 535 nm. The activity of Cp was expressed as the amount of product formed per minute per 1 L of plasma. All the absorbance values were read using the Unicam UV4 UV/Vis spectrophotometer (Cambridge, UK).

The total antioxidant capacity of plasma was determined calorimetrically by the use of the Antioxidant Assay Kit (Cayman Chemical Company, Ann Arbor, MI, USA). Absorbance measurement was taken on a spectrophotometer MultiScan GO (Thermo Scientific, Waltham, MA, USA).

Plasma concentration of thiobarbituric acid-reactive substances (TBARS) was determined using the spectrofluorometric method, optimized by Wasowicz et al. [31]. TBA-reactive compounds were extracted to butanol. The value of fluorescence of butanol layer was read at an excitation wavelength of 525 nm and an emission wavelength of 547 nm, using the PerkinElmer Luminescence Spectrometer LS50B (Norwalk, Ct, USA).

Oxidative burst (generation of reactive oxygen species, ROS) was measured using flow cytometry and fluorescent labeling with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH₂-DA, Sigma-Aldrich, St. Louis, MO, USA). The fluorescence was read in the FACSCanto II flow cytometer (BD Biosciences, San Jose, CA, USA) and expressed as the mean fluorescent intensity (MFI). The final results were expressed as the MFI ratio, calculated as the ratio of PMA stimulated cells MFI to MFI of basal ROS production. In each sample, at least 20,000 cells were examined.

DNA damage, including the strand breaks (SB) and alkali-labile sites (ALS), was assayed in whole blood using alkaline single-cell gel electrophoresis (SCGE; comet assay) method as described by Singh et al. [32] and modified by Mc Kelvey–Martin et al. [33]. One aliquot of whole blood was mixed with nine aliquots of RPMI-1640 medium and 10 aliquots of 2 % (in PBS) molten agarose type VII (low gelling temperature). The mixture was then spread on a slide earlier covered with agarose type I (low EEO) at 1 % concentration in water and dried. The cells embedded in the agarose gel were lysed in cold lysing solution (2.5 M NaCl, 100 mM Na2EDTA, 10 mM Tris base, pH 10, with 1 % Triton X-100 added just before use) at 4 °C for 1 h. Subsequently, DNA was unwound in an alkaline electrophoresis buffer (1 mM Na2EDTA, 300 mM NaOH, pH 13) for 20 min and electrophoresed in the same alkaline conditions (30 min, 25 V, 300 mA, 1.04 V/cm). The gels were then neutralized by rinsing three times with 0.4 M Tris buffer (pH 7.5), and the slides were dried for storage. In parallel analyses, oxidatively generated damage to DNA bases was additionally identified as formamidopyrimidine glycosylase (FPG)-sensitive sites using modified comet assay as described earlier by Collins et al. [34]. FPG enzyme was purchased from New England Biolabs (Hithchin, UK). After lysis, the slides were washed three times with an enzyme buffer (0.1 M KCl, 0.5 mM Na2EDTA, 40 mM HEPES–KOH, 0.2 mg/ml bovine serum albumin, pH 8) and incubated with FPG at 2.7 U/mL in this buffer (kept at −80 °C) for 30 min at 37 °C. The slides were then electrophoresed and neutralized as described above. Finally, the slides were stained with 5 μg/mL DAPI, and 50 cells from each slide were analyzed using an Olympus fluorescence microscope (a BX40 instrument; Olympus, Tokyo, Japan) equipped with an image analysis system (Comet IV, Perceptive Instruments, UK). For each participant, four slides were prepared simultaneously: two for assessment of DNA strand breakage and the other two, which included also the FPG treatment, for the assessment of total DNA damage (i.e., DNA strand breakage and oxidatively generated DNA damage). Respective DNA damage was inferred based on the relative amount of DNA in the comet tail (henceforth referred to as % tail DNA) obtained via computer-aided image analysis. DNA oxidation damage was expressed as the difference between the total DNA damage and DNA strand breaks, i.e., “net FPG-sensitive sites.”

Data normality was assessed using Shapiro–Wilk test. In the case of abnormal distributions, an analysis was performed on the log-transformed values. Baseline differences between males and females were assessed by the use of the t test, while correlations among baseline parameters and individuals’ characteristics were evaluated by the Pearson correlation coefficient. Differences in the distribution of baseline parameters (age, BMI, baseline selenium, and plasma Sepp1 concentration) according to GPX1 rs1050450 and SEPP1 rs3877899 genotypes were evaluated by one-way ANOVA. Repeated measures analysis of covariance (ANCOVA, MANCOVA) was carried out to test the effects of particular genotype or genotypes interaction on the studied biological parameters. With MANCOVA analysis, parameters measured at different time points were considered as dependent variables, while genotypes and covariates were included in the models as independent variables. Covariates included age, sex, BMI, and baseline selenium. The effect of Se supplementation (time effect/within subjects effect) as well as the effect of interaction between genotype and Se supplementation was assessed using MANCOVA test methods (Wilk’s lambda statistics) with the unstructured covariance matrix. Post hoc comparisons between pairwise time points were performed by contrast tests. All the assays were performed at a significance level of α = 0.05. In order to take into account the problem of multiple comparisons, along with original p values, the false discovery rate (FDR), adjusted p values were also calculated and reported in the text, where appropriate. Statistical analyses were conducted using SAS software, version 9.2 (SAS Institute, Cary, NC, USA).

All the individuals completed the supplementation trial. There was also 100 % responsiveness to the questionnaires introduced at each time point of blood collection, and all the subjects reported to take Se yeast tablets regularly (1 tablet/day) for 6 weeks. Nobody declared smoking tobacco and taking Se containing supplements during the study. Dietary questionnaires concerned frequency of consumption of food products which contain the highest Se content in the Polish diet (according to our previous study [26]). As shown in supplementary figures (Figure S2), the overall consumption frequency of eggs, fish, and nuts was similar during 6-week period of supplementation as compared to the washout. Additionally, we collected data on factors (consumption of drugs, hormones, dietary supplements etc.) that could potentially affect results of the study (Table S2). Subjective side effects observed during supplementation, as reported by two female participants included short nausea directly after taking the supplement (one female) and vaginal thrush (the other female).

Significant differences between males and females at baseline were observed for BMI (24.9 vs. 22.8, respectively, p < 0.0001) and GPx1 activity (16.1 vs. 18.2 U/g Hb, p = 0.01). Significant positive correlations with age were found in females for BMI (ρ = 0.34, p = 0.01), plasma Se or Sepp1 concentrations (ρ = 0.51, p = 0.0001; ρ = 0.39, p = 0.02, respectively), and TBARS (ρ = 0.43, p = 0.002). In males, positive significant correlations with age were observed only for TBARS (ρ = 0.54, p = 0.0002).

Numbers of male and female subjects within each genotype subgroup were similar (Table 2). There were no significant differences at baseline in terms of age, BMI, and Se status between the subjects with different GPX1 genotype. SEPP1 polymorphism was shown to significantly affect the baseline plasma Se concentration (in a linear manner), whereas no baseline differences were observed in terms of age, BMI, and plasma Sepp1 concentration.

Table 3 presents median values and a range of markers for selenium status (plasma concentrations of Se and Sepp1), oxidative stress (the activities of GPx1, GPx3, SOD1 and Cp, TAC, TBARS, and ROS generation), and DNA damage (at the level of DNA strand breaks and DNA oxidation) measured at baseline, during the supplementation trial and after 4 weeks of the washout period. Both plasma Se and Sepp1 median concentrations were significantly increased after 2 weeks of supplementation (98.00 µg/L vs. 62.65 µg/L and 7.36 vs. 3.86 ng/mL, respectively; p < 0.0001 and p < 0.0001, respectively) and after 6 weeks of supplementation (93.84 µg/L and 6.14 ng/mL, respectively; p < 0.0001 and p < 0.0001 vs. baseline, respectively). After 4 weeks of washout, Se started to decrease (74.61 µg/L, p < 0.0001 vs. 6 weeks); however, it was still significantly higher as compared to the baseline (p < 0.0001), whereas Sepp1 was not statistically different as compared to the median value observed after 6 weeks of supplementation (6.00 ng/mL, p = 0.270). GPx1 and GPx3 activities were significantly higher as compared to the baseline after 6 weeks of supplementation (24.43 vs. 17.10 U/gHb and 0.20 vs. 0.18 U/mL, respectively; p < 0.0001 and p < 0.0001, respectively), and both started to decrease slightly after 4 weeks of the washout period, though statistically significant difference as compared to the sixth week’s value was observed only for GPx3 (0.19 U/mL, p < 0.0001). Total plasma antioxidant capacity as well as lipid peroxidation was significantly increased already after 2 weeks of supplementation (p = 0.02 and p = 0.0002 vs. baseline, respectively) and remained significantly higher as compared to the baseline after 6 weeks of supplementation (p < 0.0001 and p < 0.0001, respectively) and after 4 weeks of the washout period (p < 0.0001 and p < 0.0001 vs. baseline, respectively). Plasma Cp activity was significantly increased after 6 weeks (p = 0.003 vs. baseline) and remained significantly higher as compared to the baseline after 4 weeks of washout (p = 0.01). SOD1 activity remained unaffected during the supplementation trial. The ability of whole-blood granulocytes to generate ROS upon PMA stimulation, measured only after 2 weeks of supplementation, was significantly decreased as compared to the baseline (p < 0.0001). A statistically significant increase in DNA damage, assessed at three time points, as compared to the baseline, was observed after 4 weeks of the washout period, both at the level of DNA strand breaks (p < 0.0001 vs. the baseline) and at the level of DNA oxidation (p < 0.0001 vs. baseline).

Table 4 presents mean values of the normalized relative gene expression, analyzed at baseline, after 2 weeks of supplementation, after 6 weeks of supplementation, and after 4 weeks of the washout period. Five out of eight analyzed genes (GPX1, GPX4, SEP15, SELS, and SELW) were affected in a negative manner upon Se supplementation. Expression of mRNA for these genes was significantly decreased as compared to the baseline after 6 weeks of supplementation (p = 0.0002 for GPX1, p = 0.001 for GPX4, p < 0.0001 for SEP15, p < 0.001 for SELS, and p < 0.0001 for SELW) and remained significantly decreased as compared to the baseline after 4 weeks of the washout period (p < 0.0001 for GPX1, p = 0.0001 for GPX4, p < 0.0001 for SEP15, p < 0.0001 for SELS, and p < 0.001 for SELW). For two genes (GPX4, SEP15), a significant decrease as compared to the baseline was observed already after 2 weeks of supplementation (p = 0.006 and p = 0.02, respectively). SEPP1 expression was the only case in which the expression was increased upon Se supplementation. However, the increase was not statistically significant as compared to the baseline, and it was significantly decreased after 4 weeks of the washout period (p = 0.001 and p = 0.0002 vs. 2 weeks and 6 weeks, respectively). Expression of the two genes, TRXR1 and SBP2, remained unaffected.

The effects of GPX1 rs1050450 and SEPP1 rs3877899 polymorphisms are presented in Table 5 and Table S3. The effects were analyzed regardless of time and with respect to time interaction to assess the possible modulatory SNP effects upon Se supplementation.

After FDR correction of p values, significant effects were maintained for all the analyzed data regardless of the genotype (Tables 3, 4), whereas most of the effects analyzed after genotype stratification were no more statistically significant (data not shown). The only genotype modulatory effect that remained statistically significant after FDR correction was the effect of GPX1 rs1050450 polymorphism on DNA oxidation, regardless of time (FDR p values: 0.02).

This supplementation trial, conducted among 95 non-smoking individuals, prospectively genotyped for two redox active selenoproteins, confirmed the previously observed effect of GPX1 rs1050450 on GPx1 activity response to Se supplementation [24, 35]. Glutathione peroxidase is an important antioxidant enzyme, which is responsible for reducing hydrogen peroxide in the presence of reduced glutathione [36]. Although GPx1 is Se-dependent enzyme, its activity relies also on other different factors, including age, sex, smoking status, and health condition [37‐39]. To exclude the modifying effect of some of these variables, the study included only the non-smoking individuals who reported themselves to be free from chronic diseases, whereas sex and age were included in the analysis as covariates. Functional effects linked to GPX1 rs1050450 polymorphism, associated with proline (Pro)-to-leucine (Leu) substitution at codon 198 (or 200), were originally observed in MCF7 cells [35] and supported by some [27, 37, 40], though not all [41], human observational studies. Altogether, GPX1 variant (Leu) allele has been suggested to cause lower responsiveness of GPx1 to Se. However, this hypothesis has not been entirely confirmed in human supplementation trials. Miller et al. [24] combined the results of two separate randomized, placebo-controlled double-blind trials (RCT), in which subjects (n = 255) with coronary artery disease were supplemented with 100 µg of Se (as SeMet) per day for 12 weeks. Average increase in GPx1 activity in the patients with at least one Leu allele was significantly lower as compared to ProPro homozygotes, but the difference between the genotypes was most evident in the subjects with the lowest baseline plasma Se concentration (83–99 µg/L). The second, smaller study was conducted among 37 obese women considered as Se-deficient (with the mean baseline plasma Se concentrations in different genotype groups ranging from 54 to 62 µg/L), who were supplemented with Brazil nuts for 8 weeks (providing 290 µg of Se/day) [25]. The authors have observed different magnitude of GPx1 activity increase depending on GPX1 genotype; however, the difference was not statistically significant, possibly due to the low sample size (only five LeuLeu homozygotes). Our study, which was conducted among individuals with a relatively low baseline Se status (mean plasma Se 62.6 µg/L), provides final evidence, indicating that GPX1 Leu allele at rs1050450 is associated with a lower GPx1 response to Se supplementation, specifically in the individuals with low Se dietary intake. This observation excludes the utility of GPx1 activity as a useful biomarker of Se status in supplementation trials, not only in the high Se status but also among the low Se status population, as previously suggested by us [27]. Furthermore, we also observed that GPX1 rs1050450 polymorphism modulated the response of other peroxidase, GPx3, showing no enzyme responsiveness in the case of Leu homozygotes, after 6 weeks of supplementation. Because this SNP effect was not target related, the exact relationship remains unclear, though possibly resulting from strong redox-related interactions between both (erythrocyte and plasma) glutathione peroxidases.

On the basis of their study, Miller et al. [24] have concluded that individuals possessing Leu allele(s) have a higher demand for Se compared to ProPro homozygotes, suggesting that they may especially benefit from Se supplementation in terms of a decreased risk of cancer or other diseases. This hypothesis, though being very promising in terms of personalized nutrition, has never been verified by human long-term supplementation trials, and furthermore, human gene association studies indicate for a rather unclear relationship between GPX1 polymorphism and cancer (detailed review in [42]). Following the hypothesis considering antioxidant activity of selenoproteins as one of the anticancer mechanisms exerted by Se, it is surprising that only few authors have investigated the relationship between GPx1 activity and any disease risk in a prospective manner. To our knowledge, only two such studies have been conducted, indicating significant inverse correlations in breast cancer [43] or coronary artery disease [44]. Thus, it should be noted that even though the possible relationship between GPX1 polymorphism and the risk of cancer may exist, there is no epidemiological evidence that GPx1 upregulation caused specifically by Se supplementation may decrease such a risk, and that such supplementation should be especially beneficial in GPX1 LeuLeu individuals with a low Se status. Some insights into a possible beneficial effect of Se supplementation specifically in GPX1 LeuLeu individuals may be obtained by the analysis of the relationship between GPX1 rs1050450 polymorphism and cancer risk-related biomarkers, such as DNA damage, in the Se supplemented individuals. Such an analysis, so far, has been conducted only in two human supplementation trials. Caple et al. [22] analyzed the effect of GPX1 polymorphism on DNA damage (as assessed by comet assay) in blood lymphocytes of 48 subjects receiving 100 µg of Se (in the form of antioxidant supplement containing also vitamins A, E, and C) for 6 weeks. No modifying effects of this SNP have been observed, either at the level of endogenous DNA damage, or at the level of peroxide-induced DNA damage (as expressed by percent of DNA in comet tail). In contrast, a significant effect of GPX1 polymorphism at the level of DNA damage has been observed in Brazil obese women in the already mentioned study, indicating that after 8 weeks of Brazil nuts consumption, ProPro homozygotes (n = 18) had lower DNA damage (measured by comet length), as compared to the baseline. Surprisingly, LeuLeu homozygotes (n = 5) apparently did not benefit from the supplementation, having significantly higher DNA damage after 8 weeks as compared to the wild-type homozygotes [25]. In our study, we observed that Se supplementation had no effect on DNA damage (regardless of the baseline values, as assessed in the MANCOVA model) but, intriguingly, the levels of both DNA oxidation and DNA strand breaks were significantly increased as compared to baseline, after 4 weeks following discontinuation of supplementation. Notably, this specific time point was during (not after) washout, as both plasma Se and Sepp1 markers (being significantly increased by intervention) still remained high. LeuLeu homozygotes had a significantly higher level of DNA oxidation as compared to ProPro and ProLeu subjects, which could be linked to the lower GPx1 activity as compared to ProPro or ProLeu; however, these genotype differences in DNA oxidation were independent of time, clearly indicating that GPx1 increase caused by Se supplementation did not improve DNA repair specifically in LeuLeu homozygotes (nor did it in two other genotypes). Furthermore, considering DNA oxidation, the LeuLeu individuals seemed also to be the most negatively affected after supplementation discontinuation. In contrast, specifically ProPro homozygotes seemed to be the only genotype which was negatively affected by Se supplementation discontinuation at the level of DNA strand breaks, notwithstanding with the observation from the Brazil study (showing ProPro homozygotes to have decreased DNA damage after consumption of high Se content nuts) or with the observational study in New Zealanders in whom the lower DNA damage as a function of increasing serum Se level has been shown in ProPro homozygotes, but not in the ProLeu or LeuLeu subjects [23, 25]. Nevertheless, we failed to indicate that GPX1 LeuLeu homozygotes may especially benefit from Se supplementation in terms of decreased DNA damage, which is consistent with the observed DNA damage effect in the LeuLeu subjects from the Brazil study [25]. Instead, we observed that GPX1 polymorphism modulated a specific negative “withdrawal effect” linked to Se supplementation discontinuation.

The overall observation of increased DNA damage during washout (shown regardless genotype) was somewhat unexpected and points to unrecognized properties of Se that may be assigned to specific chemical forms of Se present in Se yeast. The composition of Se yeast largely depends on the supplier [45], and considering that different chemical forms of Se have different cytotoxic and biochemical effects [46], it may not be excluded that some forms of Se exert deleterious effects on genomic stability. Blessing et al. indicated for example that reducible Se compounds such as phenylseleninic acid, phenylselenyl chloride, selenocystine, ebselen, and 2-nitrophenylselenocyanate are able to affect zinc finger proteins involved in DNA repair [47].

Considering daily dietary intake in Polish population (20–59 µg), the average daily intake of the element in this study during supplementation should have not exceeded the upper tolerable limits of 300 or 400 µg/day, set by the European Scientific Committee on Food or US Food and Nutrition Board [4, 48]; thus, no toxic effects were expected. Furthermore, post-supplementation plasma Se concentrations in the study individuals (55–149 µg/L) were close to or lower than those reported in plasma or serum in other short-term Se supplementation trials, such as SELGEN (109 µg/L, plasma), Danish study (126–154 µg/L, serum), New Zealand study (142.1 µg/L, plasma), or Brazil study (127–148 µg/L, plasma) [21, 24, 25, 49], and notably, it was even lower than presupplementation values observed in the long-term supplementation trials such as NPC (114 µg/L, plasma) or SELECT (135 µg/L, serum) [13, 50]. However, we observed that Se supplementation was associated with a significant increase in lipid peroxidation, which occurred already after 2 weeks of intervention and which did not diminish after 4 weeks of washout. This prooxidant effect was accompanied by the overall increased ability to challenge oxidative stress (as indicated by higher activities of GPx1, GPx3, Cp, and higher plasma antioxidant capacity), suggesting some overlapped antioxidant and prooxidant processes, possibly resulting from the overlapped activity of different Se compounds present in Se yeast [45]. At the same time, mRNA expression in white blood cells (WBC) for five selenoprotein-encoding genes (GPX1, GPX4, SEP15, SELS, and SELW) was significantly decreased, suggesting that such a negative transcriptional feedback could have been induced by excess of Se. In other words, additional intake of 200 µg of Se may be considered too high in the study group. Selenoprotein mRNA expression in WBC has been proposed as a potential marker of Se status [51, 52]. However, studies conducted in humans point to no correlation between selenoprotein gene expression in blood and plasma/serum Se in the subjects supposed to be already Se repleted or possessing relatively high baseline Se status [49, 52, 53]. In contrast, Pagmantidis et al. [54] have observed a subtle increase in mRNA expression for SPS1, SEP15, and SELK in the individuals with baseline plasma Se = 93.9 µg/L (n = 39,) who were supplemented with selenate for 6 weeks. The only study which is partially consistent with our results is a 12-week Se supplementation trial joined with trivalent influenza vaccination at week 10th. In the trial, 119 non-smoking men (age 50–64 years) with the mean plasma Se concentration of 95.5 µg/L were supplemented with Se in a form of Se yeast (50,100, or 200 µg/day) or Se-enriched onions (50 µg), and a significant decrease in mRNA expression has been observed after 10 weeks of supplementation (before vaccination) for SELW in the highest Se dose group (200 µg/day Se yeast) as compared to the placebo. For the two other analyzed selenoproteins (SELS and SELR), no changes in mRNA expression have been reported during Se supplementation (before vaccination) [55]. Notably, both SELW and SELS were decreased upon Se supplementation in our study. Altogether, the effect of Se intervention on mRNA selenoprotein expression in humans seems to be selective and possibly depends on the chemical form of Se (as supported by in vitro study [56]) as well as on the baseline Se status. Our study, which was conducted among the subjects with the lowest baseline Se status, as compared to other trials discussed indicated that a high dose intervention in the subjects with a relatively low dietary intake induced selective selenoprotein mRNA decrease. One cannot exclude that this negative transcriptional feedback resulted from a specific redox dysregulation linked to a too high difference in the redox state between presupplementation and (post)supplementation period, which consequently prevented the adaptation process to occur. This hypothesis seems in line with the new concept of reductive stress, which may be induced by upregulated GPx1 activity [42].

The second SNP of interest in this study, SEPP1 rs3877899, associated with the alanine (Ala) into threonine (Thr) change in the polypeptide chain at codon 234, was shown to affect expression of different Sepp1 isoforms, showing that variant (Thr) allele is associated with lower proportions of 60-kDa isoform, and that plasma Sepp1 concentration may depend on the interaction between this SNP and sex [21, 57]. In our study, we failed to find any effect of SEPP1 polymorphism on Sepp1, either at the level of protein or mRNA. However, we observed that the SNP affected plasma Se concentration at baseline, with the lower levels in the subjects possessing Thr allele(s), and this effect was also observed during a supplementation trial at the level approaching statistical significance (p = 0.08). Meplan et al. have observed that SEPP1 polymorphism modified the levels of plasma Se in a response to supplementation in BMI-dependent manner, suggesting that SEPP1 may affect response to Se specifically in the individuals with higher BMI (higher than 30). In this study, as assessed by MANCOVA model, we failed to find such an interaction, possibly due to the small number of individuals with BMI higher than 30 (no BMI effects on plasma Se response were observed, either alone or in combination with SEPP1 genotype, data not shown). Meplan et al. have observed several other, not target related, effects of SEPP1 rs3877899 polymorphism, linked to the affected expression or activity of specific selenoproteins in blood compartments (TRxR1, GPx1, or GPx4). The authors have concluded that different expression of Sepp1 associated with SEPP1 rs3877899 polymorphism may affect expression and activities of other selenoproteins by influencing Se supply to different tissues. In our study, such effects were minor and they concerned combined and time-dependent effect of GPX1 and SEPP1 polymorphisms on SEP15 expression, which may reflect the complexity of interactions within human selenoproteome, pointing to its dependence on Se supply and Se-dependent redox regulation. The other SEPP1 effect, observed in the interaction with time and linked to the activity of Se-independent antioxidant enzyme, superoxide dismutase 1, was rather not consistent and difficult to explain, though it may confirm some metabolic link between Sepp1 expression and redox pathway related to glutathione peroxidases as a product of SOD1 (hydrogen peroxide) is a substrate for GPx1 and GPx3.