Experimental procedures
On the day prior to the experiment, subjects were instructed to refrain from strenuous physical exercise and alcohol. They were also instructed to avoid any spicy food or beverages except water during and after dinner. On the day of the experiment, all experiments were performed in the morning from 7:50 to 11:30 AM to minimize the circadian influence. Subjects were given a simple breakfast (energy content 300 kcal or approximately 1,254 kJ), which is a traditional Japanese food about 3 h before the exercise. This meal contained 10, 20, and 70 % of energy as fat, protein, and carbohydrate, respectively. The subject was placed in an upright seated condition for at least 30-min before the experiment in a room of the temperature of 24-25^oC. The ECG data of each subject were recorded for 5-min as baseline data. At rest, all subjects breathed in synchrony with an electric metronome at 15 beats/min (0.25 Hz) to ensure that respiratory-linked variations in heart rate did not overlap with low-frequency heart-rate fluctuations (<0.15 Hz) from other sources. Following the resting protocol subjects ingested placebo (CON) or capsaicin (CAP) supplementation (150 mg, Cerebos Co., Ltd., Taiwan) chosen at random, separated by a week for washing-out. The amount of capsaicin used in this study was the dose of 1.5 times per time ingestion. We also performed the preliminary experiment on CAP supplementation at rest before this study and confirmed to be optimal time for capsaicin effect at 60 and 90-min after ingestion of capsaicin in young healthy humans. Based on these results, our obese subjects ingested CON or CAP supplementation 1 h before the exercise. After 1 h, the subjects performed an aerobic exercise using a stationary cycle ergometer (CB-X 1000, NAPS, Japan) for 30-min at 60 rpm with intensity of 50%VTmax, ranging 60 to 78 Watt for each subject. During exercise, subject could breathe freely as the respiratory rate would easily exceed the 0.25 Hz, so that low frequency components of heart rate variability (HRV) would not be affected by respiration. The ECG data were collected for 30-min during exercise for both trials. Our ECG R-R interval power spectral analysis has been fully described elsewhere (Nagai et al., 2003; Moritani et al., 1993; Matsumoto et al., 1999). The procedures of measurement of cardiac depolarization-repolarization interval have also been fully described elsewhere (Benhorin et al., 1990; Haws and Lux, 1990; Shimizu et al., 1994; Ue et al., 2000).

Genetic analysis
The distribution of the genotypes defined by the -866 G/A polymorphism of UCP2 gene and -55 C/T polymorphism of UCP3 gene in the present study is presented in Table 2. A noninvasive genotyping sampling method has been implemented for collecting buccal mucosa cells using buccal swab brushes. After the phenol- extraction procedure, 0.2 to 2 ?g of DNA per subject was obtained. Genotypes were determined with a fluorescence-based allele-specific DNA primer assay system (Toyobo Gene Analysis, Tsuruga, Japan) (Yamada et al., 2002). The analytical measurement of -866 G/A polymorphism of UCP2 gene and -55 C/T polymorphism of UCP3 gene has been fully described elsewhere (Hamada et al., 2008a, 2008b).

Statistical analysis
All statistical analyses were performed using a commercial software package (SPSS version 11.5 for Windows, SPSS Inc., Chicago, IL). Statistical differences were assessed using two-way ANOVA with repeated measurements for time, treatment, and time× treatment. When a significant difference was found, Student's paired t-test was used for comparisons between the trials at certain specific times. Student's unpaired t-test was used to compare differences in measurements between wild type + heterozygous type and homozygous type of UCP2 or UCP3 polymorphism, respectively. P values <0.05 were considered to be statistically significant. Data are expressed as mean and standard errors (SE).

Effects of capsaicin on the power spectrum of the R-R intervals
The baseline ANS data of capsaicin (CAP) or placebo (CON) trials were not significantly different (TOTAL power: CON, 1089.2 ± 312.6 ms², CAP, 1259.7 ± 345.8 ms²; LF power: CON, 268.0 ± 53.2 ms², CAP, 393.2 ± 115.4 ms²; HF power: CON, 830.2 ± 281.6 ms², CAP, 866.6 ± 257.3 ms²; SNS index: CON, 0. 42 ± 0.10, CAP, 0.52 ± 0.16). ECG R-R interval power spectral results also showed that the TOTAL power representing overall ANS activity in both trials did not reach statistical significance during exercise (Figure 1A). In LF power containing the sympatho-vagal component, there were no significant differences in these trials at any points (Figure 1B). The HF power associated with parasympathetic activity, also, was not significantly different between trials (Figure 1C). Moreover, the SNS index presenting the global sympathetic nervous activity was not significantly different during exercise in CON or CAP supplementation (Figure 1D).

Distribution of genotypes
Table 2 shows the distributions of genotypes of our subjects on - 866 G/A variant of UCP2 gene and -55 C/T variant of UCP3 gene. In the -866 G/A variant of UCP2 gene, GG allele had an overall frequency of 55.5 %, whereas subjects with the G/A allele and AA allele had frequency of 11.1 % and 33.3 %, respectively. In the -55 C/T variant of UCP3 gene, subjects with CC allele had frequency of 33.3 %. The CT allele and TT allele had frequency of 44.4 % and 22.2 %, respectively. The subject with GG allele of UCP2 and TT allele of UCP3 has also frequency of 11.1 %. The AA allele with CT allele and AA allele with TT allele in our subject has frequency of 22.2 % and 11.1 %, respectively. Although no significant difference between genotypes of UCP2 polymorphism (no Figure), G/G + G/A allele increased in SNS index compared to AA allele at EXE 15-20 min during the exercise (p > 0.068) However, genotypes of UCP3 on SNS index were not different (no Figure).

Cardiac depolarization-repolarization interval
The baseline RTc and QTc data of both trials were not significantly different. Figures 2A-B show the changes of cardiac depolarization-repolarization interval after ingestion of CAP or CON during exercise in our obese subjects. There were no significant differences in RTc and QTc during exercise in both trials.

• Obese individuals possess reduced cardiac autonomic nervous activities, especially sympathetic nervous activity associated with thermogenesis induced by capsaicin.
• Lower sympathetic nervous activity may associate with -866 G/A variants of UCP2 polymorphism.
• Capsaicin ingestion, however, may consider as a safe nutrient-aid with no adverse effects of cardiac electrical stability.