Metabo groups in response to micronutrient intervention: Pilot study

Abstract Micronutrients and their metabolites are cofactors in proteins involved in lipid metabolism. The present study was a subproject of the Harmonized Micronutrient Project (ClinTrials.gov # NCT01823744). Twenty participants were randomly selected from 136 children and adolescents that consumed a daily dose of 12 vitamins and 5 minerals supplementation for 6 weeks. The 20 individuals were divided into two pools of 10 individuals, according to their lipid profile at baseline (Pool 1 with lower triglycerides, LDL, and VLDL). The individuals were analyzed at baseline, after 6 weeks of daily supplementation, and after 6 weeks of a washout period in relation to anthropometric, body composition, food intake, lipid profile, micronutrient levels, and iTRAQ proteomic data. Genetic ancestry and its association with vitamin serum levels were also determined. After supplementation, LDL levels decreased while alpha‐tocopherol and pantothenic acid levels increased in pool 2; lipid profiles in pool 1 did not change but had higher plasma levels of pantothenic acid, pyridoxal, and pyridoxic acid. In pool 2, expression of some proteins increased, and expression of other ones decreased after intervention, while in pool 1, the same proteins responded inversely or did not change their levels. Plasma alpha‐tocopherol and Native American genetic ancestry explained a significant fraction of LDL plasma levels at baseline and in response to the intervention. After intervention, changes in expression of alpha‐1 antitrypsin, haptoglobin, Ig alpha‐1 chain C region, plasma protease C1 inhibitor, alpha‐1‐acid glycoprotein 1, fibrinogen alpha, beta, and gamma‐chain in individuals in pool 2 may be associated with levels of LDL and vitamin E. Vitamin E and Native American genetic ancestry may also be implicated in changes of vitamin E and LDL levels. The results of this pilot study must be validated in future studies with larger sample size or in in vitro studies.

These risk factors are sensitive to nutritional intake. Balanced, nutrient-dense diets can help achieve and maintain an adequate lipid, glycemic, and nutritional status profiles (Güngör, 2014).
Almost all recommendations for micronutrient intake are based on the average in groups of individuals. In many cases, these recommendations are based on fasting levels in (presumably) healthy people and data for children and adolescents are sparse in many populations.
We and others proposed to evaluate metabolic responses to acute challenges (e.g., oral glucose or mixed meal) or short-term interventions (e.g., multiple micronutrient challenges) (Kaput & Morine, 2012;Kaput et al., 2014;Mathias et al., 2018;Ommen, Greef, & Ordovas, 2014;Pellis et al., 2012;Stroeve, Wietmarschen, & Kremer, 2015) to provide additional information about nutritional needs of individuals rather than just at baseline status. Metabolic responses are defined as changes in levels of not only the target metabolite or its surrogate (e.g., vitamin or lipoprotein) but also other biochemical variables (e.g., plasma proteins, micro RNAs, and other metabolites). Supplementing intake with a complex mixture of vitamins and minerals to an otherwise calorie-sufficient diet improved metabolic health of Brazilian children and adolescents (Mathias et al., 2018).
The integration of these metabolic readouts provides a more comprehensive description of the physiological system and a more informative description of health. The ability to analyze different metabolites, proteins, RNA, or other blood molecules depends on the sensitivity of technologies, which may constrain the analysis.
Isobaric Tag for relative and absolute quantitation (iTRAQ) allows for the discovery of blood or plasma proteins altered by nutritional intervention.
Genetic ancestry is also an important factor that can influence metabolite requirements in individuals and also affect population-level-derived averages. We (Mathias et al., 2018) and others (Kehdy et al., 2015;Rolim et al., 2016) have used admixture to better understand genetic and metabolite differences of individuals in subgroups of the Brazilian populations. We tested the relationship between an individual's genetic ancestry and micronutrient levels because the Brazilian population is highly admixed (Amerindians, European colonizers, African slaves, and more recent introgression due to immigration from other world regions, e.g., Asia). Linear regressions between ancestral components and baseline vitamin levels showed higher thiamine monophosphate (TMP) levels with higher European ancestry. Plasma vitamin B12 was negatively associated with increasing Native American ancestry. Finally, Native American ancestry was associated with lower baseline folate levels and greater response to the intervention (Mathias et al., (2018)). These results deserve further evaluation since vitamin levels may be implicated in reduction of LDL (Mathias et al., (2018)), an important predictor of cardiovascular disease.
We hypothesized that a metabolic group with a poor lipid profile would benefit most from micronutrient intervention and thus improve metabolic health through changes in expression of some proteins closely related to lipid metabolism. We also hypothesized that some improvements on vitamins and lipid levels could be associated with genetic ancestry.
The study aimed to: (a) evaluate the changes in proteomic profile, nutritional status, and vitamin serum levels after a micronutrient intervention in two lipid profile groups and (b) associate vitamin and lipid levels with genetic ancestry. The results of this pilot study must be validated in future studies with larger sample size or in vitro and experimental designs.

| Pilot study design and Population
This study was a subproject of the Harmonized Micronutrient Project (ClinTrials.gov # NCT01823744) that analyzed omics, biochemical, and nutritional status (Mathias et al., 2018) at baseline (time point or visit 1); after 6 weeks of daily supplementation of vitamins and minerals (time point or visit 2); and after 6 weeks of a washout period (time point or visit 3). Participants were healthy children and adolescents (ages 9-13) recruited from the west side of Ribeirão Preto (Brazil) in two county schools and one private school. For this specific pilot study, 20 participants were randomly selected from a sample size of 136 children and adolescents that were previously included following specific exclusion criteria: (a) one or more episodes of axillary temperature higher than 37°C within the 15 days preceding the data collection, (b) three or more episodes of liquid stools within the 24 hr before assessment, (c) current intake of vitamin or mineral supplement; dietary restrictions of any time, including weight-loss interventions, and (d) history of chronic diseases; participation in another clinical trial within the 4 weeks preceding the study (Mathias et al., 2018). The
(n = 10) had higher triglycerides, LDL, and VLDL levels at baseline. The participants were evaluated by a pediatrician to determine their clinical conditions and pubertal stage, according to Tanner's criteria (Tanner, 1962) in visit 1, 2, and 3.
All participants received a daily supplement of 12 vitamins and 5 minerals in a commercial milk bar (Nestrovit ® ) ( Table 1) for 5 days per week for 6 weeks. This product was chosen because it (a) was palatable (which would facilitate acceptance by participants), (b) had low amounts of calories (3 milk bars contains 75 calories), (c) has been commercially available in Switzerland since 1936 but never sold in Brazil, and (d) had a known and standard nutritional composition, all of which met the objectives of this study. Six of the authors individually monitored supplement intake at the beginning of each school period, and therefore, the compliance rate for the individuals in this substudy was 100%.

| Blood collection and Laboratory analyses
Blood samples were taken after 12 hr fasting in EDTA tubes for metabolomics and proteomics, in PAXgene tubes for DNA analysis, and separately in ACD tubes for clinical biochemistry. After centrifugation, plasma was removed and 100 μl was frozen for iTRAQ proteomic analysis. Clinical biochemistry, micronutrient, dietary intake, genotype analyses, and plasma vitamin response were described previously (Mathias et al., 2018).
For the iTRAQ proteomic analysis, 6 sample pools were made: pool 1 at Visit 1 (P1V1), Visit 2 (P1V2), and Visit 3 (P1V3), and pool 2 at Visit 1 (P2V1), Visit 2 (P2V2), and Visit 3 (P2V3). These 6 pooled plasma samples were dilapidated and depleted of the most abundant plasma proteins using the Proteopep Immunoaffinity Albumin and IgG depletion kit (Sigma ® ), according to manufacturer's protocol. Total proteins of each pool were quantified by the method of Bradford (1976). After preparation, the samples were submitted to enzymatic hydrolysis with trypsin. Tryptic peptides from each pool were labeled with isobaric tag for relative and absolute quantitation using the iTRAQ 8-plex kit (AB Sciex ® ) according to the manufacturer's instructions. Each peptide solution was labeled at room temperature for 2 hr with one iTRAQ Vitamin measurements were previously analyzed (Mathias et al., 2018). In this study, vitamin medians (minimum-maximum) were used for individuals in pool 1 and pool 2. Note: Comercial name: "Nestrovit", brand: "Nestlé". 2 milk bars (10 g): 51.3 kcal, 4.4 g of carbohydrate, 0.5 g of protein and 3.4 g of lipid. 3 milk bars (15 g): 77 kcal, 6.7 g of carbohydrate, 0.8 g of protein and 5.2 g of lipid.

| Anthropometric and body composition data
Height and weight were measured immediately following blood collection (Jelliffe, 1968). Body mass index (BMI) was used as criteria for weight status (World Health Organization, 2007). Waist circumference was measured at the level of the imaginary horizontal line in the middle region between the last rib and the iliac crest (Heyward & Stolarczyk, 1996). Body composition analysis was performed by bioelectrical impedance analysis, according to Lukaski, Bolonchuk, and Hall (1986) immediately following the blood draw and before breakfast.

| Food intake data
The usual dietary intake was assessed by a food frequency questionnaire (FFQ) of the preceding month using a previously validated questionnaire for Brazil children at each of the three study time points (Fumagalli, Monteiro, & Sartorelli, 2008

| Genetic ancestry
Genetic analysis and ancestry determination were previously analyzed and reported in Mathias et al. (2018). Average ancestry data for individuals in each pool were calculated by summing percent of individual ancestry and dividing by 10.

| Statistical analysis
SPSS 20.0 program ® was used to analyze metabolic and nutritional data. Mann-Whitney and Student's t tests were used to compare two pools. For longitudinal analysis, ANOVA for repeated measurements was used adjusting by Bonferroni test. Chi-square test was used to compare proportions. The intensities found in iTRAQ analyses were expressed by fold change of the pool and not of the individual. The fold change of the pool is the ratio between the quantitative values of a given protein between baseline and postintervention. Proteins with fold change ≥1.20 or ≤0.80 were considered as differentially expressed proteins, as described in other studies (Duthie, Osborne, & Foster, 2007;Moulder et al., 2010;Seshi, 2006;Unwin et al., 2006). Simple and multiple variate linear regression approaches were used to test associations between the ancestral components and lipid and vitamin levels in the 20 participants. Statistical significance was considered when p < .05.
At baseline, pool 1 had lower gamma-tocopherol (a form of vitamin E), retinol (vitamin A), and higher TMP (thiamine monophosphate, a form of vitamin B1) and vitamin B12 when compared to pool 2. At baseline, average nutrient intake, anthropometric measurements, and body composition did not differ between the pools (Tables 2 and 3) and also did not vary throughout the study in individuals of either pool (p > .05 for all parameters, data not shown). The lipid profile improved only in individuals in pool 2 with a decrease in LDL from V1 to V2 (Table 4). Although many vitamins increased from V1 to V2 after intervention and decreased from V2 to V3 (after wash out), only pantothenic acid (vitamin B5), pyridoxal (a form of vitamin B6), and pyridoxic acid (a catabolic product of vitamin B6) plasma levels in pool 1 and alpha-tocopherol (a form of vitamin E) and pantothenic acid in pool 2 reached statistical significance (Table 5).
Twenty plasma proteins were identified by proteomic analysis that changed expression after micronutrient supplementation in at least one of the pools, and 18 presented fold change ratio ≥1.20

TA B L E 2
Comparison of anthropometric measurements, body composition profile, and gender at baseline between the pools or ≤0.80. In addition, most of the identified proteins had different levels between the pools after the intervention (i.e., the same protein had increased expression in a pool and had decreased or unchanged expression in the other pool after the supplementation) ( Table 6).
Average genetic admixture differed between pools with a higher percentage of Native American ancestry in pool 2 compared to a higher percentage of ancestry from Europe in pool 1 (Table 7). African genetic ancestry was greater in pool 2 although the difference did not reach statistical significance. However, simple linear regression analysis applied to all subjects (n = 20) shows that genetic ancestry alone could not explain statistically different vitamin levels at baseline (Table 8).
Multiple regression analysis was used to find associations among metabolite levels in response to the intervention using alpha-tocopherol and pantothenic acid (whose plasma levels as well as predict 29% of the fold change variation for alpha-tocopherol from visit 1 to visit 2 (r = .62; R 2 = .38; adjusted R 2 = .29; ANOVA p = .007). Regression analysis did not find any association

TA B L E 3 Comparison of energy and macronutrients intake between pools
Variable pool 1 (n = 10) pool 2 (n = 10) p value

TA B L E 4 Comparison of lipid profile between the pools and throughout the study
between the percentage of Native America ancestry and sex with plasma pantothenic acid in V1, V2, or for the fold change (V2-V1).
Statistically significant associations were also not found between the percentage of European ancestry and sex with plasma alpha-tocopherol or pantothenic acid in V1, V2, and for fold change (V2-V1).
Plasma pantothenic acid fold change and differences in sex can predict 34% of the variation in fold change for LDL (r = .64; R 2 = .41; adjusted R 2 = .34; ANOVA p = .012).
We tested whether pantothenic acid, alpha-tocopherol, sex, and genetic ancestry could together predict baseline LDL and after intervention. The fold change in plasma alpha-tocopherol, fold change in plasma pantothenic acid, differences in sex, and the percentage of American and Europeans ancestry explained 39% of the variation in LDL levels (r = .75; R 2 = .56; adjusted R 2 = .39; ANOVA p = .04). We could not link genetic ancestry with proteomics because the samples were pooled. α-tocopherol (Vit E) V2 (µg/ml) 5.6 (3.1-7.5) 8.0 (3.1-10.1) a .063
body composition, or nutritional status (which did not change throughout the study).
In pool 2, expression of alpha-1 antitrypsin, haptoglobin, Ig alpha-1 chain C region, and plasma protease C1 inhibitor increased. These changes may be associated with the improvements in plasma LDL. The above proteins have been shown to be associated with positive physiological effects in lipid/glucose metabolism, micronutrients transport/ metabolism, and in the immune system (Carter & Worwood, 2007;Davis et al., 2008;Toonen et al., 2016;UniProt, 2016UniProt, , 2017a. In pool 2, expression of alpha-1-acid glycoprotein 1 and fibrinogen alpha-, beta-, and gamma-chains decreased in response to the intervention. Alpha-1-acid glycoprotein 1 is a positive acute phase plasma protein (Luo et al., 2015;Tesseromatis et al., 2011), and high fibrinogen (Gruys et al., 2005) levels were positively associated with atherothrombotic disease (Aleman, Walton, & Byrnes, 2014;Perl et al., 2016;Poredoš & Ježovnik, 2015). A decrease in levels of these markers after intervention may TA B L E 6 Expression of the proteins in pool samples* identified by iTRAQ proteomic analysis benefit individuals in pool 2. Individuals in pool 1 did not show any improvements in lipid profile, and the analyzed proteins responded inversely or did not change their levels.
Multiple linear regression analysis applied to all subjects (n = 20) showed that sex and plasma alpha-tocopherol predicted 32% of LDL plasma variation at baseline and 11% of LDL plasma levels variation in V2. Sex and fold change in alpha-tocopherol and fold change in pantothenic acid plasma levels explained 31% and 34% of the fold change variation in plasma LDL levels, respectively. In addition, the percentage of America genetic ancestry and differences in sex could together predict 33% of alpha-tocopherol plasma levels variation in V2, as well as 29% in fold change variation of alpha-tocopherol that was found between visit 1 to visit 2. This is the first study showing a possible association between American genetic ancestry and vitamin E in children and adolescents. The role of vitamin E as an antioxidant is well known, but it also contributes to anti-inflammatory responses through interleukin-4, interleukin 8, TNF-α, and inhibition of lipopolysaccharide secretion (Wu, Liu, & Ng, 2008). Moreover, vitamin E may protect against cardiovascular disease, improve lipid profiles, and reduce LDL oxidation (Burdeos, Nakagawa, & Kimura, 2012;Daud et al., 2013;Heng et al., 2013;Qureshi, Salser, & Parmar, 2001;Wu et al., 2008). Micronutrients, including pantothenic acid (vitamin B5), play an important role in lipid metabolism (Al-Attas et al., 2014;Evans et al., 2014;Hadjistavri et al., 2010;Heng et al., 2013;Kelishadi et al., 2014Kelishadi et al., , 2010, which supports the changes in LDL metabolism in the pool 2. Differences in sex and genetic ancestry from Americans and from Europeans predicted 43% and 30% of fold change LDL plasma levels variation. Others have found association between American and Europeans genetic ancestry with LDL, even after adjusting for interactions with vitamin E (Dumitrescu et al., 2012(Dumitrescu et al., , 2010).
The present pilot study found that the fold change in plasma alpha-tocopherol, fold change in plasma pantothenic acid, differences in sex, and the percentage of American and Europeans ancestry explained 39% of the variation in fold change for LDL, corroborating some studies (Burdeos et al., 2012;Daud et al., 2013;Dumitrescu et al., 2012Dumitrescu et al., , 2010Evans et al., 2014;Heng et al., 2013;Qureshi et al., 2001;Wu et al., 2008). To our knowledge, this is the first study that found these variables explained variation in LDL plasma levels after micronutrient supplementation.
This study has some limitations. Samples for pooling were selected based on differences in lipid profiles and were few in number.
In addition, pooling eliminated the possibility of analyzing samples individually or testing the association of vitamin levels, proteomic data, and ancestry. These experimental choices were due to the high cost and time for this procedure. However, the use of pooling samples has been successfully used in several studies of proteomic analysis (Karp & Lilley, 2009;Kaur, Rizk, & Ibrahim, 2012;Weinkauf, Hiddemann, & Dreyling, 2006).

| CON CLUS IONS
Ten individuals with similar high lipid profile at baseline responded positively (i.e., decreased LDL) to the intervention and also had increased alpha-tocopherol and pantothenic acid levels. Changes after the intervention in the level of alpha-, beta-, and gammafibrinogen chains, haptoglobin, Ig alpha-1 chain C region, plasma protease C1 inhibitor, alpha-2-HS-glycoprotein, alpha-1 antitrypsin, and alpha-1-acid glycoprotein1 may be associated with changes of plasma LDL. Many of the proteins differed inversely between individuals in each of the pools, that is, while a protein had increased expression in one pool, the same protein had decreased or unchanged expression in the other pool. These results were consistent with the emerging awareness that individuals differ in response to the same nutritional intervention. The use of pools allowed for the identification of proteins correlated with changes in LDL levels using iTRAQ methodology. In addition, differences in sex, plasma alpha-tocopherol, plasma pantothenic acid, and genetic ancestry directly or indirectly predicted LDL plasma levels in the total sample. The results of this pilot study must be validated in future studies in vitro, with animal models, and in human studies with larger sample sizes.

ACK N OWLED G EM ENTS
The authors thank the children, teens, and parents who participated in this study as well as the principals, teachers, and school district officials who made the schools available for activities related to the project. We also thank team from Protein Chemistry Center (Medical School of Ribeirão Preto, University of São Paulo, Brazil), which collaborated with the proteomic analyzes.

I N FO R M E D CO N S E NT
Written informed consent was obtained from all study participants. Each participant signed an assent form, and their parents (or legal guardians) signed a consent form prior to participation in this study.

TR A N S PA R EN C Y D ECL A R ATI O N
The lead author affirms that this manuscript is an honest, accurate, and transparent account of the study being reported. The reporting of this work is compliant with STROBE guidelines. The lead author affirms that no important aspects of the study have been omitted and that any discrepancies from the study as planned (ClinTrials.gov # NCT01823744) have been explained.