2.1. Study type and characterization of subjects
This is a randomized, double-blind, cross-over, controlled clinical trial with number RBR-4d9dmqz in the Brazilian Clinical Trials Registry (ReBEC). The sample size was established a priori, according to an increase in time-to-exhaustion from 59.2 ± 27.8 minutes to 68.4 ± 29.2 minutes in response to purple grape juice by Toscano et al. . This increase represented an effect size of d = 0.92, so that a minimum of 13 subjects were assigned considering an α error of 0.05 and a β error of 0.80 using GPower 3.1 software program (Franz Faul, Universitat Kiel, Germany). However, as the subjects in this study were categorized by genotype, a higher number of participants was needed. Therefore, the study was conducted with 47 male recreational runners. After genotyping, the group with the smallest sample size (CC of the SOD3 gene) was left with 12 subjects.
Inclusion criteria were defined as recreational runners should have been training for at least one year with a frequency of at least four weekly sessions and for at least two months without interruption and taking part in competitions from 5 kilometers to 21 kilometers or marathon regularly. They also should not have any chronic condition or degenerative diseases, did not smoke, and were not using any long-term medication. Furthermore, subjects should not consume purple grapes or their derivatives often; and/or dietary supplements, vitamins or bioactive substances present in the grape (polyphenols). Athletes who suffered from skeletal muscle injuries, altered dietary habits or had inconsistent training patterns, started using medication or did not perform some of the study experimental procedures were also excluded. Figure 1 shows the randomization, allocation and follow-up of subjects, according to the inclusion criteria.
>> INSERT FIGURE 1 <<
2.2. Study design
Figure 2 shows the experimental design of the study. First, the athletes performed the maximum aerobic capacity test. The next 2 weeks they participated in the experimental procedures by performing the test run to exhaustion, drinking grape juice or control drink prior to the test. The order of these procedures was determined randomly (in blocks) by a researcher who did not participate in the other experimental protocols using www.randomizer.org. Subjects (athletes) and the researcher, who gave the test to exhaustion, were blinded as to the supplementation tested. Heart rate and perception of effort were recorded every 10 minutes throughout the test. Blood samples were taken before supplementation (grape juice or control drink), immediately before the running test until exhaustion (two hours after supplementation) and immediately at the end of the test to measure the redox balance, muscle damage and plasmatic nitrite. Furthermore, quality of sleep, social stress and self-referred recovery were evaluated before each test. An oral mucosa sample was collected at the end of the experiment for DNA genotyping.
>> INSERT FIGURE 2 <<
2.3. Aerobic capacity test
Volunteers performed a 3200-m running test in the week before implementing the experimental protocol in order to characterize their aerobic capacity according to the protocols proposed by Weltman et al. . Running tests to estimate maximum aerobic capacity were carried out with each athlete individually. The adopted 3200-m test protocol was performed on a 400-m lane of an official athletic track certified for international competitions. The athletes performed a 5-minute warm-up on the treadmill with spontaneous speed. They were instructed to start the test and complete the proposed distance in the shortest time possible. VO2max was estimated using the following equation: VO2max (ml.kg-1.min-1) = 118.4 – 4.774 x (T), where T is the time in minutes and decimal fraction.
2.4. Pre-experimental procedures
The athletes did not exercise for 48 hours and remained 10 hours overnight fasting prior to testing and were instructed to not consume grapes or derivatives, nor eat any foods rich in antioxidants. Moreover, they did not consume caffeine or alcoholic drinks in the 24 hours prior to the test sessions.
The volunteers arrived at the laboratory at 6:30 am to start the tests on the two days of experimental protocols. First, they were acclimatized in an air-conditioned room at 22 and 25 °C, RH 65% and sat at rest for 10 minutes before starting the experimental protocol.
2.5. Body composition and nutritional assessment
The athletes were characterized for body composition by measuring height with a portable stadiometer (Sanny, Standard, São Paulo, Brazil), and then body composition was assessed using Bioimpedance (InBody 570 Biospace®, San Francisco, California, USA), with multifrequency analysis on an eight-point tetrapolar electrode system. They were also instructed not to use caffeinated or alcoholic beverages during the previous 24 hours and not to drink water on the morning of the evaluation.
Food intake was evaluated using a 24-hour food record  employed twice before the time-to-exhaustion tests. This was done to characterize food consumption in the 24 hours prior to the experimental tests (Avanutri®, Rio de Janeiro, Brazil).
2.6. Characterization of grape juice and supplementation protocol
The experimental beverage employed in this study was purple grape juice from Cooperativa Vinícola Garibaldi (Garibaldi, Serra Gaúcha, Brazil) produced from grapes of the Isabel, Bordeaux and Concord (V. labrusca) varieties. According to the manufacturer, the juice was characterized as being a natural, whole (100% grape juice) and non-alcoholic drink with no added sugar, water, flavorings or preservatives. Furthermore, 200 mL of the grape juice contains 130 kcal and 32 g of carbohydrates and does not contain a significant amount of protein, total fat, saturated fat, trans fat, dietary fiber or sodium.
The antioxidant composition was determined by free radical scavenging activities using DPPH (2,2 diphenyl-1-picrylhydrazyl) according to Brand-Williams, Cuvelier and Berset , and ABTS (2,2′-azinobis-3-ethylbenzthiazoline-6 sulfonic acid) [24, 25]. The total phenolic content was determined according to the methodology described by Folin-Ciocateau . The quantification of specific classes of polyphenols was performed in the present study, in which the quantification of the phenolic compounds was performed by high-performance liquid chromatography (HPLC) using a Waters 2695 Alliance system (Milford, MA, USA) equipped with a diode array detector (DAD) and fluorescence detector (FLD) according to a method validated by EMBRAPA. The control drink was (maltodextrin) artificial grape flavor, which presented the same amount of carbohydrates, calories and volume , but without polyphenols.
The athletes had a standardized breakfast which consisted of a sandwich (50 g of whole meal bread + 34 g of processed white light cheese = 152 kcal; 21.8 g of carbohydrates; 7.4 g of proteins; 3.6 g of fat; 4.0 g and purple grape juice or control drink, according to previous randomization. They received 10 mL/kg/day  of supplementation of grape juice or control drink two hours before the running test to exhaustion, so that they could reach greater bioavailability of the polyphenolic compounds present in the drink . Volunteers and researchers involved in the experimental procedures were blinded to supplementation. The beverages also had similar color and flavor.
2.7. Subjective stress, sleep and recovery/rest
To characterize the physiological conditions in the two pre-experimental test moments, the Profile of Mood States questionnaire (POMS)  was given to each athlete to examine the psychometric state of Total Humoral Disturbance (THD). In addition, the Stress and Recovery Questionnaire for athletes (RESTQ-Sport)  was used to evaluate the state of stress and recovery from everyday situations that are potentially stressful and restful. The EPWORTH-Brazilian Sleepiness Scale (ESS-BR)  was employed to assess the occurrence of daytime sleepiness.
2.8. Run to exhaustion test
The athletes performed a free warm-up of between 5 and 7 minutes before starting the test. Next, they performed the running test to exhaustion on a treadmill ergometer (Moviment, São Paulo, Brazil), with the speed set at 80% of VO2max [31, 32], converting this value to belt speed according to a previous study . Heart Rate (Polar FT1, Kempele, Finland) and the Borg Rating of Perceived Exertion Scale  was monitored every 10 minutes during the test. The test was interrupted when the athlete demonstrated inability to keep up with the treadmill speed and through verbal confirmation between 19 and 20 on the effort scale. The total running time was recorded at the end of the test. The electronic panel of the treadmill was covered so that athletes could not see the race data, especially the time elapsed in order to ensure methodological rigor. The test was performed under controlled temperature and relative humidity (RH) (22 and 25 °C, RH 65%) respectively, measured by a thermohygrometer (TFA HT-7429, São Paulo, Brazil).
Samples of the oral epithelial cells in the subjects were collected by mouth washing with sucrose, and then the DNA was extracted. The genotypes were determined using the RFLP technique (Restriction Fragment Length Polymorphism) in polymerase chain using specific primers. The sense primer (5'-CGCCAGGCGCGGGAACACTCAG-3') and antisense primer (5'-GGCGGACTTGCACTCGCTCTCG-3') were used for the SOD3 Arg213Gly polymorphism. Amplification was performed according to a protocol described by Laddha . The PCR product was digested using the restriction enzyme MwoI (New England 307 Biolabs, Ipswich, MA, USA) and taken to a dry incubation at 60 °C for 3 hours.
The sense (5'-ACAATCACTCCTTAAATATGGTGG-3') and antisense primers (5' -AAGTAGGGACAGACAGGACCAGTA-3') were respectively used for the PPARα 7G/C polymorphism. Amplification occurred according to the method proposed by Pishva et al. . The PCR product was digested using the Taqα I enzyme (New England 316 Biolabs, Ipswich, MA, USA) and taken to incubation at 65 °C for 3 hours. Next, they were evaluated by electrophoresis on 15% polyacrylamide gel and staining with 0.5% silver nitrate.
2.10. Redox balance, plasma nitrite and damage muscle
The oxidative activity was evaluated by lipid peroxidation that could be quantified by the metabolic product of malondialdehyde (MDA) through a thiobarbituric acid reaction in plasma . The antioxidant activity was determined through total antioxidant capacity (TAC). It was quantified in plasma via the free radical scavenging activity of 2,2-diphenyl-1-picrylhydrazyl . Plasma nitrite concentrations were determined by the Griess reaction that quantifies the nitrite in the sample through the diazotization reaction, forming a pinkish chromophore . Muscle damage was measured by Creatine kinase (CK) using the catalytic activity method. Lactate dehydrogenase (LDH) concentrations were measured using the pyruvate-lactate method via specific commercial kits (Labtest, Minas Gerais, Brazil). The absorbances were obtained in Labmax 240 premium automatic analyzer (Labtest, Minas Gerais, Brazil) at a wavelength of 340nm.
2.11. Statistical analysis
Data were presented as mean ± standard deviation (SD). Normality and homogeneity of data were assessed using the Shapiro-Wilk and Levene tests, respectively. Data that did not show a normal distribution were ranked according to Templeton  for normalization. Analyzes were performed by being subdivided according to the genotype considering the presence or absence of the allele with the polymorphism characteristic. Athletes were divided based on the presence of the G allele (GG+CG genotypes) or CC genotype for the SOD3 Ag213Gly polymorphism; while subjects were categorized according to the presence of the C allele (CC+GC genotypes) or GG genotype for the PPARα 7G/C polymorphism. An independent t-test was used in order to compare the results of pre-experimental conditions (nutritional intake, sleep, mood, general stress, physical recovery and plasma biomarkers) and differences between absolute deltas in time-to-exhaustion. The results for biomarker levels were analyzed using two-way ANOVA for multiple comparisons with Bonferroni's post-hoc test. The effect size was calculated using the Cohen’s d test for independent samples using the GPower Statistics 3.1 program and then classified according to Cook, Cook and Therrin  into small (d = 0.00 - 0.40), medium (d = 0.50 - 0.70), large (d = 0.80 - 1.00) or very large (d ≥1.30) effect sizes. This provided information on the power of the practical application of the intervention. Moreover, an individual analysis (simple subject analysis) was performed on the exhaustion time variable to demonstrate the variability of the athletes’ responses in running performance. P-values < 0.05 were considered statistically significant. SPSS was used for data treatment (v. 23, IBM SPSS, Chicago, IL, USA).
2.12. Ethical statement
The study was conducted entirely in accordance with the Declaration of Helsinki. Its protocol was approved by the Research Ethics Committee of the Center for Health Sciences, Federal University of Paraiba (protocol no. 2.196.523). The subjects signed an informed consent form according to Resolution 466/12 of the National Health Council (Brazil), and informed consent was also obtained from all of the subjects before the inclusion in the study.