Introduction

The consumption of tree nuts (e.g. peanuts, almonds, hazelnuts, walnuts, Brazil nuts, pistachios, cashews, and macadamias) has been associated with decreased risk of cardiovascular disease in human trials [1,2,3,4]. The benefits of tree nut consumption are probably associated with their nutritional composition: rich in monounsaturated (MUFA) and polyunsaturated fatty acids (PUFA), magnesium, copper, selenium, vitamin E, folic acid and other bioactive compounds such as phytosterols and phenolic acids [3, 5, 6]. The analysis of two human trials, the Nurse’s Health Study (NHS) (female subjects) and The Health Professionals Follow-up Study (HPFS) (male subjects), both performed in the USA, revealed that the intake of tree nuts was associated with lower risk of cardiovascular disease [7]. Comparable results were found in the analyses of the National Health and Nutrition Examination Survey (NHANES) from 2005 to 2010 [3].

There are some mechanisms that could explain the positive effect of tree nuts on cardiovascular health. The most accepted is the lipid-lowering effect, since tree nuts have a high content of PUFA that could explain the reduction of total cholesterol and LDL-c, observed for almonds, peanuts, walnuts and pecan nuts [4]. The lipid-lowering effect of Brazil nuts has been investigated in several studies. The supplementation of one Brazil nut per day during 18 weeks in obese females reduced cardiovascular risk by increasing HDL-c concentrations [8]. The increase in HDL-c and the decrease in LDL-c were observed in healthy adults after an ingestion of 4 units of Brazil nuts for 4 weeks [9], and a decrease in total cholesterol and LDL-c was observed in obese female adolescents after the ingestion of 3–4 units per day during 16 weeks [10]. Brazil nut supplementation did not affect HDL-c and LDL-c in another study with healthy adults after ingestion of ten units of nuts per day during 2 weeks; however, it did affect the transfer of cholesteryl esters into HDL pool, which is important to reverse cholesterol transport and its elimination into bile. Overall, the increase of cholesteryl esters into HDL pool after consumption of Brazil nuts can be considered as an antiatherogenic effect [11].

Brazil nuts are the richest source of selenium (Se) known in nature [12]. As the amino acid selenocysteine (Sec), Se is inserted into selenoproteins, which have important functions in the antioxidant system, lipid peroxidation, immune function, brain function, diabetes risk, among others [13]. The insertion of Sec into selenoproteins occurs during translation, and it requires the presence of a specific structure in the 3′untranslated region (3′UTR) of the mRNA, a specific RNA for Sec (tRNA[Ser]Sec), and other structures [14].

In contrast to other selenoproteins, Selenoprotein P (SePP) has ten Sec residues, which are expressed primarily in liver but also in other tissues such as brain, gut, heart and kidneys [15]. In Se-deficient conditions, SePP is preferentially directed to the brain and testis, which implies the presence of SePP receptors APOER2 in these two tissues [15]. Plasma Se and SePP concentrations are the most used biomarkers of Se status. SePP concentration and mRNA expression have been associated with insulin resistance. A higher concentration of plasma SePP was found in people with type 2 diabetes, and in rodent models of type 2 diabetes, hepatic mRNA of SEPP1 was elevated [16]. In addition, treatment of mice with two intraperitoneal injections of 1 mg/kg body weight of purified human SePP induced glucose intolerance and insulin resistance [17]. Se and selenoprotein levels have been related to lipid and carbohydrate metabolisms [18] and also to diabetes risk, where patients with type 2 diabetes have higher plasma concentrations of SePP [16]. Moreover, plasma SePP was positively associated with carotid intima media thickness [16]. In animal studies, Se supplementation either with selenite or selenate decreased plasma triglyceride levels; however, only selenate decreased plasma cholesterol and suppressed the gene expression of gluconeogenic enzymes in the liver. It is noteworthy that high levels of selenate were necessary to achieve these effects; therefore, its use in clinical practice is not applicable [18].

SePP is encoded by the gene SEPP1 located in chromosome 5p12. It has two functional single nucleotide polymorphisms (SNPs) [19, 20]. Both are a G to A substitutions—the first one is located in the coding region of the gene changing the amino acid Alanine to Threonine at position 234 of the protein (rs3877899). This polymorphism modulates plasma Se levels in healthy adults in response to supplementation with sodium selenite [20]. The second one (rs7579) is located in the 3′UTR, important for Sec insertion, and it modulates plasma SePP concentrations [20]. Genetic variants in SEPP1 have not been previously associated with lipid profile or glucose levels in response to supplementation with Brazil nuts. Therefore, due to the evidence associating SePP with glucose and lipid metabolisms, the hypothesis of this study was that functional polymorphisms in selenoprotein genes would influence the response of serum lipids and glucose after supplementation with a high-Se and high-lipid nut such as the Brazil nut.

Methods

Study population and supplementation protocol

One hundred and thirty unrelated healthy adults (males and females) aged 20–60 years old were selected at University of São Paulo, Brazil. Volunteers taking multivitamins and mineral supplements, anti-inflammatory drugs, with excessive alcohol consumption, athletes, obese (BMI >30) and with chronic diseases such as cancer, diabetes and cardiovascular disease were not included in the study. The participants were invited by electronic correspondence and personal communication when the protocol was explained.

The Supplementation with Brazil Nuts study (SU.BRA.NUT study) was an 8-week dietary intervention with one Brazil nut a day in healthy subjects, followed by eight more weeks without intervention. At the beginning of the study, 20 mL of blood sample were drawn and subsequently the volunteers took a daily supplement of one Brazil nut for 8 weeks. At the end of four and 8 weeks of supplementation, another 20 mL blood sample was taken, and then two more blood samples were taken at 4 week intervals during 8 weeks without intervention (Fig. 1). Volunteers were asked to complete a control calendar and mark with an “x” when they consumed each nut throughout the intervention period. For all participants, weight and height were measured, and the body mass index (BMI) was calculated. Written informed consent was signed by all volunteers before blood sampling. The protocol was approved by Faculty of Pharmaceutical Sciences Ethical Committee (CAE: 00961112.3.0000.0067) and was conducted according to the Declaration of Helsinski.

Fig. 1
figure 1

Supplementation with Brazil nuts protocol. Down arrows indicate blood sampling collection

Full size image

Centesimal Composition and Se content of Brazil nuts

The centesimal composition of a random sample of Brazil nuts representative of the four batches used in the study was analysed according to the methods proposed by the Association of Official Analytical Chemists (AOAC [21]). The Se content of Brazil nuts was determined using hybrid generation flame atomic absorption spectrometry (HGFAAS), as described previously [8].

Blood sampling

Fasting blood samples (20 mL) were drawn by venipuncture into four 5-mL EDTA tubes. An aliquot of 1.5 mL of EDTA whole blood was used for DNA extraction and subsequent genotyping. Another 5 mL of blood were collected in a tube without anticoagulant to obtain serum. Plasma was separated by centrifugation at 3000 rpm for 15 min at 4 °C. The erythrocyte pellet was washed three times with 5 mL sterile 9 g/L NaCl solution, slowly mixed by inversion, and centrifuged at 10,000 rpm for 10 min (Eppendorf, C5408) at 4 °C, and the supernatant fluid was discarded. Aliquots of whole blood, serum, plasma and erythrocytes were frozen at −80 °C in sterile, demineralized tubes until the analyses were performed.

Biochemical parameters

Concentrations of glucose, C-reactive protein and lipid profile (total cholesterol (TC), HDL-c, LDL-c and triglycerides) in fasting serum were measured spectrometrically using commercial kits (Labtest, Minas Gerais, Brazil) adapted to a biochemical analyser (LabMax 240, Labtest, Minas Gerais, Brazil). Malondialdehyde (MDA) was measured in plasma by high-performance liquid chromatography (HPLC) on a Shimadzu (Kyoto, Japan) instrument equipped with a Phenomenex (Torrance, CA, USA) reverse phase C18 column. The instrument was calibrated using a MDA standard stock solution in the following concentrations: 0.0, 0.25, 0.5, 1.0, 2.0, 4.0, 6.0 and 12.0 μM.

Genotyping

Total DNA was extracted from 200 μL of whole blood using the Purelink Genomic DNA Minikit (Invitrogen, Life Technologies, California, USA). The final concentration was measured by using a NanoDrop ND 1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA) and adjusted for further analysis. SNPs in selenoprotein genes were determined by real-time PCR using Taqman SNP Genotyping Assays (Thermo Fisher Scientific, Wilmington, DE, USA). Samples were assayed along with no-template and internal controls for each genotype and run in the StepOne Plus Real Time PCR system under the following conditions: enzyme activation at 95 °C for 10 min, followed by 40 cycles at 92 °C for 15 s and 60 °C for 1 min for annealing and extension. The allelic discrimination was obtained by performing an endpoint read. The SNPs selected were in GPX1 gene (rs1050450), GPX4 gene (rs713041), SEPP1 gene (rs3877899 and rs7579), SELS gene (rs34713741) and SEP15 gene (rs5845).

Statistical analysis

Continuous variables were tested for normality using the Kolmogorov–Smirnov test. The data were presented as geometric means (CI 95%). Concentrations of blood lipids, glucose and C-reactive protein were compared in the different time points using ANOVA repeated measures or Friedman’s test. For MDA, paired t Student test was used. A genetic dominant model was used to assess differences in the presence of the rare allele. In this model, individuals with the rare allele were combined in one category, leaving the common genotype in another category. Multivariate linear regression models were created using total cholesterol and glucose at each intervention as dependent variables. Age, body fat composition, gender, tertiles of plasma Se, and six SNPs were included as independent variables. Repeated measures analysis of covariance (ANCOVA) was performed to investigate the effect of the genotypes for SNPs appointed in the multivariate linear regression models. Covariates included for outcome total cholesterol was age and for outcome glucose were gender, tertiles of plasma Se and body fat composition. The Chi square test with continuity correction was used to determine whether genotype frequencies followed the Hardy–Weinberg Equilibrium. The haplotype distribution and linkage disequilibrium were done in the software Haploview 4.2. SNPs were considered in linkage disequilibrium when D′ was >0.5. Differences were considered significant at p < 0.05. The analyses were performed using the Statistical Package for the Social Sciences software version 17.0 for Windows (SPSS, Chicago, IL, USA) and GraphPad Prism (GraphPad Prism version 5.00 for Windows, GraphPad Software, San Diego, CA, USA).

Results

Demographic and anthropometric characteristics of the participants

A total of 130 healthy volunteers completed the entire study protocol. Females constituted 75% of the group and 72% of the group self-reported as being Caucasian. Mean age for females was 28.4 years (95% CI 26.9–30.0) and for males was 29.2 years (95% CI 26.4–32.3). Family history of chronic diseases, such as cancer, diabetes mellitus and cardiovascular disease, was reported by 87% of the volunteers. At baseline, there was a difference between females and males for BMI (22.5 (95% CI 21.8–23.1) vs. 24.7 (95% CI 23.5–26.0), respectively (p < 0.001) and body fat percentage (26.8 (95% CI 25.6–28.0) vs. 21.4 (95% CI 19.0–24.1), respectively (p < 0.001).

Centesimal composition and Se content of Brazil nuts

The Se content and centesimal composition of Brazil nuts are shown in Table 1. Four different batches were used during the supplementation. The mean ± standard deviation for Se content of these four batches was 100.4 ± 5.3 μg/g. The average weight of the nuts was from 3 to 4 g, therefore each nut provided approximately 300 μg of Se, which is approximately five times higher than the RDA for adults (55 μg/day).

Table 1 Centesimal composition and selenium contents in Brazil nuts used during the protocol
Full size table

Lipid profile alterations after supplementation with Brazil nuts

The lipid profile and MDA levels are reported in Table 2. During the intervention, fasting glucose concentrations decreased after 4 and 8 weeks of daily consumption of Brazil nuts (p < 0.001). Total cholesterol concentrations also decreased after 8 weeks of supplementation. No significant differences were observed for triglycerides, HDL-c, LDL-c and MDA levels. After interruption of Brazil nut intake, fasting glucose concentrations at four and 8 weeks were still lower than baseline (p < 0.001). C-reactive protein levels at 8 weeks after interruption were higher than baseline (p = 0.023). No significant differences were observed for total cholesterol, triglycerides, HDL-c, LDL-c and MDA levels.

Table 2 Biochemical parameters during and after diary intervention with one unit of Brazil nut for 8 weeks in healthy subjects
Full size table

Influence of age on total cholesterol concentrations

Multivariate linear regression models for total cholesterol were created in order to explain the variations observed during Brazil nut supplementation (Table 3). Age was the only variable associated with total cholesterol at 4 weeks of intervention, in which the increase of 1 year of age was related to the increase of 1.123 mg/dL of total cholesterol.

Table 3 Multivariate linear regression models for dependent variable total cholesterol during intervention with one unit of Brazil nut in healthy subjects
Full size table

Body fat percentage, gender, plasma Se and rs3877899 on SEPP1 gene influence on glucose concentrations

The multivariate linear regression models created for glucose can be seen in Table 4. The body fat percentage was associated with glucose concentrations at three time points. An increase of one unit of body fat percentage increased 0.429, 0.495 and 0.583 mg/dL of glucose at baseline, 4 and 8 weeks of intervention, respectively. The presence of the SNP rs3877899 in SEPP1 gene was associated with glucose concentrations at baseline, in which the presence of the variant allele A was related to a reduction of 4.520 mg/dL of glucose (p = 0.025). The increase of plasma Se concentrations was related to an increase in glucose concentrations at 4 weeks of intervention (p = 0.046). Gender was associated with glucose concentrations at 8 weeks of intervention: being male was related to an increase of 8.145 mg/dL in glucose concentrations (p < 0.001). The frequency of the genotypes and alleles for rs3877899 in SEPP1 gene were GG 54% (n = 70), GA 36% (n = 47) and AA 10% (n = 13) and for rs7579 were GG 38% (n = 50), GA 42% (n = 55) and AA 19% (n = 25). Genotype distribution did not significantly deviate from Hardy–Weinberg equilibrium for any of the SNPs. Haplotype analysis showed evidence of linkage disequilibrium for the two SNPs in the SEPP1 gene (rs7579, rs3877899) (D′ = 1.0 and r 2 = 0.15) with three haplotypes observed: haplotype a (common) GG (44%), haplotype b AG (28%) and haplotype GA (28%).

Table 4 Multivariate linear regression models for dependent variable glucose during intervention with one unit of Brazil nut in healthy subjects
Full size table

Influence of SNPs in SEPP1 on serum cholesterol and glucose concentrations

Total cholesterol and glucose were stratified by SNPs in SEPP1 (Fig. 2). The SNP rs7579 modulated cholesterol response to supplementation. During the intervention, total cholesterol decreased in both groups, but carriers of the rare allele A had higher cholesterol concentrations during the supplementation, almost reaching statistical significance after 4 weeks of intervention (p = 0.054). Glucose concentrations were modulated by the coding SNP rs3877899. During intervention, glucose concentrations decreased significantly in both groups; however, carriers of the rare allele A had lower values at baseline (p = 0.004) and after 8 weeks of intervention (p = 0.013).

Fig. 2
figure 2

Total cholesterol and glucose stratified by polymorphisms in SEPP1 gene. Values are means ± st error. Two way ANCOVA repeated measures adjusted for multiple comparisons by Bonferroni test for total cholesterol (a) and glucose (b) levels. Covariates used: age for cholesterol and body fat composition, gender and plasma Se tertiles for glucose levels

Full size image

Discussion

Previous studies have demonstrated that the intake of Brazil nuts improves blood lipid profile in adults [8,9,10, 22]. The present results support and extend these observations by showing that supplementation with one Brazil nut per day for 8 weeks decreases total serum cholesterol and glucose concentrations in healthy adults. In addition, genetic variations in SEPP1 (rs3877899 and rs7579) modulated the effect of Brazil nut supplementation on plasma glucose and total cholesterol concentrations. This study is the first to observe an association between plasma cholesterol and glucose levels and the polymorphisms in SEPP1 after Brazil nut supplementation.

The supplementation with one Brazil nut provided an average of 300 μg of Se for each nut. This is a high amount of Se and after a long-term supplementation (e.g. >1 year) could potentially increase the risk of metabolic diseases. Some epidemiological studies have demonstrated that high plasma Se levels (above 150 μg/L) increase the risk of all-cause mortality and cardiovascular disease. In the US adult population with baseline plasma Se of 120 μg/L, there is an increased risk for cancer (mostly lung cancer), coronary and cardiovascular diseases after supplementation [23]. The SELECT trial, designed to investigate the role of Se and vitamin E supplementation in prostate cancer prevention, also observed an increased risk for diabetes after supplementation with 200 μg of selenomethionine for 5 years. The population was 50 years old or older American males with high baseline Se status (135 μg/L) [24]. Nevertheless, in our study, the population had lower Se status at baseline (90 μg/L), and the duration of the intervention was only 2 months, which is much shorter than the duration of the SELECT trial. Moreover, our intervention was with one Brazil nut and not an isolated compound as the trial used, and we considered the genetic constitution of the individuals, which the SELECT trial had not considered on its first stages. All these differences could justify the absence of potential adverse effects of the Brazil nut supplementation in our study.

Recent reports have demonstrated that the intake of Brazil nuts could increase HDL-c concentrations in obese females [8] and healthy adults [9], decrease total cholesterol and LDL-c in obese female adolescents [10] and decrease total cholesterol in dyslipidaemic adults [22]. In the present work, we found that the daily intake of one Brazil nut for 8 weeks decreased total cholesterol levels. Although the decrease in total cholesterol concentration can be considered of limited clinical application because these volunteers were not dyslipidaemic, this reduction after 8 weeks of ingestion and the following increase during the washout period suggest that Brazil nuts may interfere with cholesterol metabolism and this might be relevant for populations at risk for metabolic syndrome or dyslipidaemia.

Potential mechanisms underlying this improvement in blood lipid profile include the nutritional composition of Brazil nuts [25], the connection between Se metabolism with the mevalonate pathway [26] and with cholesterol biosynthesis [27]. Although Brazil nuts have higher saturated fatty acids (SFA) concentrations compared with other nuts, the MUFA and PUFA contents are sufficient to exert their cholesterol lowering effect [4].

Another explanation for the decrease observed in total cholesterol concentration is the relation between cholesterol and selenoprotein biosynthesis. The two metabolisms are connected through the mevalonate pathway [26]. In order to be functional, selenoproteins need to have the amino acid Sec inserted in their structure during translation. This requires the synthesis and activation of the tRNA[Ser]Sec which is dependant on four bases’ modifications. One of these modifications is the isopentenylation of adenosine 37, which requires isopentenyl pyrophosphate (IPP), a direct metabolite of mevalonate during cholesterol biosynthesis [26]. Our hypothesis is that during the supplementation with a high Se content nut, the synthesis of selenoproteins might be stimulated, which could be directing the IPP for the isopentenylation of adenosine 37 of the tRNA[Ser]Sec, to the detriment of cholesterol synthesis. Selenoproteins that are ranked high in the hierarchy of selenoprotein expression, such as GPx4, could be preferentially synthesized under such conditions [28]. The metabolisms of Se and cholesterol are also connected via sterol response element binding protein 2, SREBP2. Previously, it was observed that Se supplementation increased expression of 15-deoxy-12,14-prostaglandin J2 [29], a ligand of the peroxisome proliferator-activated receptor-γ, PPAR-γ, which reduces SREBP2, decreasing cholesterol biosynthesis [30]. Moreover, studies in mice deleted for the Trsp gene that encodes the specific transfer RNA for Sec, the tRNA[Ser]Sec, prevented selenoproteins expression and increased plasma cholesterol [31]. All these results indicate that one or more selenoproteins may be involved in the regulation of cholesterol metabolism.

Experimental evidence has demonstrated that SNP rs3877899 in SEPP1 influences plasma Se concentration after supplementation with 200 μg/day of sodium selenite for 4 weeks [20]. Our results extend these earlier observations, indicating that individuals with the rare allele A (GA + AA) had lower concentrations of glucose at baseline, and after 8 weeks of supplementation with Brazil nuts in comparison with subjects of homozygous GG. As observed for total cholesterol, values were within the range expected for healthy adults, limiting the clinical application of this result. However, the reduction after just 8 weeks and the maintenance of lower concentrations during the washout period could be relevant as a higher glucose lowering effect after Brazil nut intake should be expected for pre-diabetic or insulin-resistant individuals. Recently, a meta-analysis of randomized controlled trials investigating the influence of nut consumption on glycaemic control in diabetes demonstrated that diets with tree nuts per day significantly reduced fasting glucose concentration in individuals with type 2 diabetes [32]. Furthermore, SePP has been associated with the insulin metabolism in vitro and in vivo studies. SePP is expressed in different tissues in mice [33], but it is mainly synthesised and secreted to the bloodstream by the liver [34]. The pancreas of mice also expresses SePP in the α-cells which produces glucagon and in the β-cells which produces insulin [33]. SePP expression was decreased in isolated islets after high glucose treatment [35]. Our results indicate that after supplementation with Brazil nuts the plasma SePP concentrations increased (data not shown), and the glucose levels decreased, which is in accordance with the previous findings observed in mice. The mechanism underlying this connection of SePP and glucose metabolism may be explained by the presence of a binding site for FoxO1 transcription factor in the promoter gene of SEPP1 [35]. This transcription factor regulates the expression of gluconeogenic enzymes, such as glucose-6-phosphatase and phosphoenolpyruvate carboxykinase. These observations suggest that SePP is an important factor regulating glucose metabolism. Therefore, the presence of genetic polymorphisms in SEPP1 gene could modify how this selenoprotein regulates the insulin metabolism.

In conclusion, this study indicates that supplementation with one Brazil nut per day for 8 weeks can reduce total plasma cholesterol concentration in healthy adults, and this response is apparently modulated by the coding SNP rs3877899 in SEPP1. In addition, Brazil nut supplementation can also reduce fasting plasma glucose, and this effect is possibly modulated by rs7579 in SEPP1. Although the observed reductions were still within the normal range for clinical practice, these results may be important in future nutritional interventions with the goal of investigating the effect of supplementation with Brazil nut on cholesterol and glucose metabolisms. Any future Brazil nut supplementation study conducted for long-term periods should be carried out so as to produce a more modest increase in plasma Se levels in order to achieve concentrations not higher than 200 μg/L, with the aim of avoiding adverse effects observed in previous studies. These future studies should consider the baseline nutritional status, the gender and the genetic background of the participants.