Effects of weight loss using supplementation with Lactobacillus strains on body fat and medium-chain acylcarnitines in overweight individuals

Abstract

Our previous study showed that supplementation with a combination of Lactobacillus curvatus (L. curvatus) HY7601 and Lactobacillus plantarum (L. plantarum) KY1032 reduced the body weight, body fat percentage, body fat mass and L1 subcutaneous fat area in overweight subjects. We aimed to evaluate whether the changes in adiposity after supplementation with Lactobacillus strains were associated with metabolic intermediates. A randomized, double-blind, placebo-controlled study was conducted on 66 non-diabetic and overweight individuals. Over a 12-week period, the probiotic group consumed 2 g of probiotic powder, whereas the placebo group consumed the same product without the probiotics. To investigate metabolic alterations, we performed plasma metabolomics using ultra-performance liquid chromatography and mass spectrometry (UPLC-LTQ/Orbitrap MS). Probiotic supplementation significantly increased the levels of octenoylcarnitine (C8:1), tetradecenoylcarnitine (C14:1), decanoylcarnitine (C10) and dodecenoylcarnitine (C12:1) compared with the levels from placebo supplementation. In the probiotic group, the changes in the body weight, body fat percentage, body fat mass and L1 subcutaneous fat area were negatively associated with changes in the levels of C8:1, C14:1, C10 and C12:1 acylcarnitines. In overweight individuals, probiotic-induced weight loss and adiposity reduction from the probiotic supplementation were associated with an increase in medium-chain acylcarnitines.

---

1. Introduction

Studies evaluating the influence of probiotics, prebiotics and synbiotics on body weight have been inconclusive because of the scarcity and low quality of the available studies.^1–3 Recently, Madjd et al.⁴ found that consuming a probiotic yogurt during a 12-week hypoenergetic program resulted in no changes in body weight compared with a low-fat yogurt. However, a meta-analysis by Dror et al.⁵ suggested a role for probiotics in promoting weight loss in adults. In our previous study, we observed reductions in body weight, body fat percentage, and body fat mass after a 12-week treatment with two probiotic strains, Lactobacillus curvatus (L. curvatus) HY7601 and Lactobacillus plantarum (L. plantarum) KY1032, suggesting that the probiotic-induced weight loss was associated with decreased oxidative and inflammatory stress in overweight subjects.⁶ The mechanism behind the weight loss observed in adults likely relates to a reduction in inflammation and a strengthened intestinal epithelial barrier that has been observed to accompany probiotic administration.^7–10 Because all these studies on adults involved the probiotic Lactobacillus, the observed weight loss from probiotics may be a result of anti-inflammatory effects, improved intestinal barrier integrity and increased metabolic activity associated with changes in intestinal flora. However, additional studies are required to determine the relationship between the weight loss from probiotic supplementation and metabolic changes.

Metabolites as intermediates and end products of biological processes are most predictable for the phenotype of an organism.¹¹ Therefore, metabolites may serve as biomarkers for changes in the physiological status, particularly for alterations in metabolic pathways in humans.¹¹ However, studies on endogenous metabolic changes caused by probiotic-induced weight loss are limited. To the best of our knowledge, this is the first human intervention study to investigate the effects of probiotic supplementation on both probiotic-induced weight loss and metabolic changes. Therefore, in this study, we aimed to determine whether supplementation with a combination of L. curvatus HY7601 and L. plantarum KY1032 or placebo resulted in changes from the baseline in fasting metabolic intermediates. Furthermore, we aimed to evaluate whether changes in adiposity after probiotic or placebo supplementation were associated with changes in fasting metabolic intermediates. The plasma metabolic profiles of the probiotic group and the control group were analyzed using ultra-performance liquid chromatography and mass spectrometry (UPLC-LTQ/Orbitrap MS).

2. Materials and methods

2.1 Study subjects, study design and intervention

The details of the study subjects, study design, and interventions were published previously.⁶ Briefly, at the beginning of the study, 120 non-diabetic and overweight individuals were enrolled in a 12-week, double-blind, placebo-controlled, randomized study. The randomization was conducted via computer-generated block randomization (60 subjects in the control and probiotic groups). Among them, 95 individuals were included in the final analysis: a probiotic group (n = 49) that consumed 2 g of probiotic powder twice a day (after breakfast and dinner) containing L. curvatus HY7601 (2.5 × 10⁹ colony-forming units (cfu)) and L. plantarum KY1032 (2.5 × 10⁹ cfu) and a placebo group (n = 46) that consumed the same amount of powder daily without the probiotics (NCT02492698; http://www.clinicaltrials.gov). Only the product (powder) provider (Korea Yakult Co., Ltd, Yongin, Gyeonggi, Korea) retained the control and probiotic group list, which was disclosed after the research and all laboratory analyses were completed. Both the placebo and probiotic products had the same product weight, color, and flavor and were put into an opaque package; thus, the investigators were unaware of the individuals’ treatment allocation. Among the 95 subjects, 66 subjects (34 in the placebo group and 32 in the probiotic group) agreed to the analysis of their plasma metabolomic profiles. Written informed consent was obtained, and the study was performed according to the declaration of Helsinki. The procedures were approved by the institutional review boards of Yonsei University and Yonsei University Severance Hospital.

2.2 Measurements of daily energy intake and physical activity

The details regarding daily energy intake and physical activity were published previously.⁶ Briefly, the subjects were instructed to maintain their existing eating habits and physical activity levels during the study period (pre-ingestion: 2 weeks; ingestion period: 12 weeks) to ensure that the changes in body weight did not result from the changes in diet or physical activity. Each participant completed a standardized 3-day diet and physical activity record at weeks 0, 6 and 12.

2.3 Anthropometric parameters and blood pressure

Detailed information is provided in our previous article.⁶ Briefly, body weight, height and waist circumference were measured at screening, baseline and the 12-week follow-up. The body mass index (BMI) was calculated in units of kilograms per square meter (kg m⁻²). During each testing session, systolic and diastolic blood pressure was assessed in a supine position after a resting period.

2.4 Abdominal fat area and body composition measurements

The details of the abdominal fat area and body composition measurements were published previously.⁶ Briefly, body fat distribution and muscle areas were measured using computed tomography (CT) at the level of the 1^st lumbar (L1) and the 4^th lumbar (L4) vertebrae. The body composition of the study participants was measured using dual-energy X-ray absorptiometry (DEXA) to determine the body fat percentage, body fat mass and lean body mass.

2.5 Blood collection and biochemical analyses

The details of the blood collection and biochemical analyses were published previously.⁶ Briefly, blood samples were collected following an overnight fast of at least 12 h, and the levels of fasting triglycerides, total HDL, and LDL cholesterol, glucose, insulin and high-sensitivity C-reactive protein (hs-CRP) were measured. Insulin resistance (IR) was calculated using the homeostasis model assessment (HOMA) by the following equation: HOMA-IR = [fasting insulin (μIU mL⁻¹) × fasting glucose (mg dL⁻¹)]/405.

2.6 Nontargeted metabolic profiling of plasma

2.6.1 Sample preparation. Detailed information on the sample preparation was published previously.¹² Briefly, for protein precipitation, 100 μL of each sample aliquot was mixed with 800 μL of acetonitrile and then centrifuged at 10 000 rpm for 5 min. A total of 820 μL of the supernatant was dried with nitrogen gas. The remaining pellet was dissolved in 10% methanol and centrifuged at 10 000 rpm for 5 min. Finally, 85 μL of the supernatant was transferred to a fresh vial.

2.6.2 UPLC-LTQ Orbitrap XL mass spectrometry (MS) analysis. Chromatographic separations were performed using an Acquity UPLC™ BEH C18 column (2.1 × 50 mm, 1.7 μm; Waters, Milford, MA, USA) with an Ultimate™ 3000 BioRS UPLC system (Dionex-Thermo Fisher Scientific, Bremen, Germany). The injected samples (4 μL) were equilibrated with LC-MS grade water (Thermo Fisher Scientific, Fair Lawn, NJ, USA) containing 0.1% formic acid (solvent A) and were eluted at a flow rate of 0.35 mL min⁻¹, using LC-MS grade acetonitrile (Thermo Fisher Scientific, Fair Lawn, NJ, USA) containing 0.1% formic acid (solvent B). In this study, the proportion of acetonitrile was altered over time (at 0 min, 1% of B; at 1 min, 1% of B; at 16 min, 90% of B; at 17 min, 90% of B; at 17.2 min, 1% of B; and at 20 min, 1% of B). MS was performed using an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) operating in the ESI-positive mode and full scan mode for Fourier-transform mass spectrometry. The measurement conditions were as follows: capillary voltage: 35 V; tube lens voltage: 120 V; capillary temperature: 360 °C; spray voltage: 5 kV; and the flow rate of the nitrogen sheath gas and the auxiliary gas: 50 and 5 (arbitrary units), respectively. The MS data were collected as the mass-to-charge ratio (m/z) from 50 to 1000. For the system conditioning and quality control, a quality control sample was prepared by pooling aliquots of each sample. The MS/MS collision energy range was from 40 to 55 eV.

2.6.3 Data processing and identification of metabolites. The raw data were analyzed using the SIEVE 2.2 data analysis software (Thermo Fisher Scientific, Waltham, MA, USA), and pertinent data were extracted, including the retention time, the m/z ratio and the intensity of the ions. The parameters used for peak alignment were: m/z range: 50–1000; m/z width: 5 ppm; and retention-time width: 2.5 min. The metabolites were identified using the following databases: Human Metabolome (http://www.hmdb.ca); Lipid MAPS (http://www.lipidmaps.org); KEGG (http://www.genome.jp/kegg); and MassBank (http://www.massbank.jp). MS/MS was conducted to confirm putative metabolites by database searches.

2.7 Statistical analysis

The statistical analyses were performed using the SPSS version 21.0 software (IBM/SPSS, Chicago, IL, USA). The results are expressed as the mean ± standard error (SE). Logarithmic transformation was performed on skewed variables. For descriptive purposes, the mean values are presented as median values ± standard deviation (SD) for skewed variables. A two-tailed P < 0.05 was considered statistically significant. Independent t-tests were used to compare the parameters between the two groups. Paired t-tests were performed to determine the differences between the baseline and 12-week follow-up values within each group. Pearson's correlation coefficient was used to examine the relationship between variables. False discovery rate-corrected q-values were computed using the R package ‘fdrtool’, and q-values less than 0.05 were considered to indicate significance. Heat maps were created to visualize and evaluate the relationship among metabolites and biochemical measurements in the study population.

The metabolomics data were exported to SIMCA-P+ 14.0 (Umetrics, Inc., Umeå, Sweden) for multivariate analysis. The data were Pareto scaled, and an orthogonal partial least squares-discriminant analysis (OPLS-DA) was used to predict our models. The validity of the models was measured using a default cross-validation procedure of 7-fold cross-validation and R²Y and Q²Y parameters.

3. Results

3.1 The effects of 12 weeks of probiotic consumption on anthropometric parameters, clinical characteristics, body composition and abdominal fat areas

The effects of 12 weeks of probiotic consumption on anthropometric parameters, clinical characteristics, body composition, and abdominal fat areas were in agreement with our previous findings.⁶Table 1 shows the body weight, BMI, body fat percentage and fat mass, and the abdominal fat areas at the baseline and at 12 weeks in the placebo and probiotic groups. At the baseline, no significant differences between the two groups were observed for age, gender distribution, smoking and drinking rates (data not shown), body weight, BMI, body fat percentage, body fat mass, and the total, visceral, and subcutaneous fat areas at the L1 and L4 levels. After 12 weeks of treatment, the individuals in the probiotic group showed a significant reduction in the body weight, BMI, total fat mass (717 g), fat percentage (0.67%) and a trend toward a decrease in the L4 subcutaneous fat area (P = 0.080) from the baseline, while a significant increase in visceral fat, subcutaneous fat, and the total fat areas at the L1 level was observed in the placebo group. When we compared the changes from the baseline between the control and probiotic groups, significant differences were observed due to changes in the body weight, BMI, fat percentage, fat mass, L1 total fat area, and L1 subcutaneous fat area, with a trend toward a difference in the L4 total fat area (P = 0.063). The probiotic group showed a reduction in these parameters, whereas the control group showed an increase after the 12-week treatment.

Table 1 Effects of 12-week consumption of probiotics on body weight, fat percentage and fat mass by DEXA evaluation and abdominal fat areas by CT

	Total (n = 66)				P ^a	P ^b	P ^c
	Placebo group (n = 34)		Probiotic group (n = 32)
	Baseline	Follow-up	Baseline	Follow-up
Mean ± SE. P^a-Values derived from independent t-test for baseline values. P^b-Values derived from independent t-test for follow-up values. P^c-Values derived from independent t-test for changed values. P^d-Values adjusting for baseline. ^†P < 0.10, *P < 0.05, **P < 0.01, ***P < 0.001 derived from paired t-test.
Weight (kg)	73.2 ± 1.72	73.6 ± 1.82	71.9 ± 1.43	71.3 ± 1.50*	0.581	0.334
Change	0.46 ± 0.24		−0.60 ± 0.30				0.008
Body mass index (kg m^−2)	27.1 ± 0.27	27.2 ± 0.28	26.6 ± 0.23	26.4 ± 0.25*	0.173	0.024
Change	0.15 ± 0.09		−0.23 ± 0.11				0.008
DEXA evaluation
Fat percentage (%)	31.2 ± 1.00	31.3 ± 0.96	30.6 ± 0.98	30.0 ± 1.03*	0.712	0.357
Change	0.12 ± 0.20		−0.67 ± 0.27				0.020
Fat mass (g)	22 993.5 ± 762.7	23 236.3 ± 782.1	22 157.9 ± 585.1	21 440.6 ± 622.9**	0.392	0.079
Change	242.9 ± 200.1		−717.3 ± 225.1				0.002
CT evaluation (L1)
Total fat area (cm^2)	243.8 ± 11.1	256.0 ± 10.0**	247.3 ± 7.6	246.9 ± 8.14	0.797	0.486
Change	12.2 ± 3.70		−0.43 ± 4.04				0.024
Visceral fat area (cm^2)	101.8 ± 7.87	110.1 ± 7.38**	115.2 ± 6.53	116.8 ± 7.09	0.199	0.519
Change	8.33 ± 2.61		1.64 ± 3.19				0.108
Subcutaneous fat area (cm^2)	142.0 ± 6.87	145.84 ± 6.66*	132.2 ± 5.65	130.1 ± 5.77^†	0.275	0.080
Change	3.84 ± 1.88		−2.07 ± 1.42				0.016
Visceral/subcutaneous fat ratio (%)	0.78 ± 0.08	0.83 ± 0.09*	0.94 ± 0.08	0.97 ± 0.08	0.163	0.251
Change	0.05 ± 0.02		0.03 ± 0.03				0.603
CT evaluation (L4)
Total fat area (cm^2)	308.0 ± 8.60	314.5 ± 8.10	301.6 ± 7.10	296.6 ± 7.40	0.569	0.109
Change	6.46 ± 4.33		−4.99 ± 4.22				0.063
Visceral fat area (cm^2)	88.8 ± 5.51	90.2 ± 5.44	86.5 ± 5.22	83.1 ± 4.29	0.764	0.313
Change	1.32 ± 2.18		−3.43 ± 3.49				0.246
Subcutaneous fat area (cm^2)	219.2 ± 8.08	224.3 ± 8.12	215.0 ± 7.77	213.5 ± 7.58	0.714	0.335
Change	5.14 ± 3.97		−1.55 ± 2.68				0.172
Visceral/subcutaneous fat ratio (%)	0.44 ± 0.04	0.43 ± 0.04	0.43 ± 0.04	0.41 ± 0.03	0.935	0.657
Change	0.00 ± 0.02		−0.02 ± 0.02				0.506

---

Furthermore, no significant differences existed at the baseline, and no significant mean changes were observed after the 12-week treatment in systolic and diastolic blood pressure, levels of triglycerides, total HDL and LDL cholesterol, glucose, insulin, hs-CRP and the HOMA-IR index (data not shown) in the placebo and probiotic groups. Additionally, the estimated total calorie intake, the amount of physical activity, the percentage of protein intake, the percentage of fat intake and the percentage of carbohydrate intake did not differ significantly between the two groups at the baseline, week 6 and week 12 (data not shown).

3.2 Plasma metabolic profiling using UPLC-LTQ Orbitrap mass spectrometry

3.2.1 Nontargeted metabolic pattern analysis. The MS data of plasma metabolites were obtained at the baseline and week 12 and were analyzed using an OPLS-DA score plot. OPLS-DA is the most appropriate technique to search for metabolic profiles that define the characteristics of the probiotic-induced weight loss because the predictive component in the OPLS-DA describes the treatment effect excluding the variance between samples in the same group.¹³ OPLS-DAs were conducted for the following three combinations of groups: (1) the comparison between the baseline levels and the 12-week follow-up levels of the placebo group (n = 34) (Fig. 1a); (2) the comparison between the baseline levels and the 12-week follow-up levels of the probiotic group (n = 32) (Fig. 1b); and (3) the comparison between the 12-week follow-up levels of the placebo group (n = 34) and the probiotic group (n = 32) (Fig. 1b).

Fig. 1 Identification of plasma metabolites with significantly altered levels. (a) Score plots from OPLS-DA models in the placebo group (n = 34); comparison between baseline (red) and 12-week follow-up (green) levels in the placebo group. (b) Score plots from OPLS-DA models in the probiotic group (n = 32); comparison between baseline (blue) and 12-week follow-up (yellow) levels in the probiotic group. (c) Score plots from OPLS-DA models at the 12-week follow-up; comparison between 12-week follow-up in the placebo (n = 34, green) and probiotic (n = 32, yellow) groups, respectively. (d–f) S-plots for covariance [p] and reliability correlation [p(corr)] from OPLS-DA models.

The quality of each OPLS-DA model was examined using the R² and Q² values to confirm that the models were not over-fitted and evaluate the predictive ability of each model. R² represents the goodness-of-fit parameter and is defined as the proportion of variance in the data described by the model. Q² represents the predictive ability parameter and is defined as the proportion of variance in the data predicted by the model. R² and Q² values greater than 0.5 indicate a high-quality OPLS-DA model.¹⁴ The two-component OPLS-DA scatter plots of the plasma metabolites for the first combination of groups did not show distinct clustering or a clear separation of participants in the placebo group at the baseline or at the 12-week follow-up assessment (R²Y = 0.729, Q²Y = 0.093). These analyses, however, revealed that the second and third models (Fig. 1) displayed greater than 86% goodness-of-fit (R²Y = 0.948 and R²Y = 0.864 for the second and third combinations of groups, respectively) and a predictive ability greater than 61% (Q²Y = 0.642 and Q²Y = 0.614 for the second and third combinations of groups, respectively), which demonstrated that the second and third OPLS-DA models in the present study were well-fitted and displayed an acceptable predictive ability. The present results clearly showed that plasma metabolomic profiles could distinguish groups based on probiotics, changes in probiotic-induced biochemical characteristics, or both. To extract potential variables contributing to the detected differences, S-plots from the OPLS-DA models were constructed. S-Plots of p(1) and p(corr)(1) were generated using centroid scaling (Fig. 1d–f). The S-plots revealed that metabolites with higher or lower p(corr) values were more relevant for discriminating between the two groups.

3.2.2 Identification of plasma metabolites. Among the 1471 plasma variables, the variables that contributed to the separation between the groups were selected according to the variable importance in the projection (VIP) method; VIP values >1.0 indicate a high relevance for differences between the sample groups. A total of 238 variables were identified with VIP values >1.0, of which 51 were previously identified metabolites and 187 were unknown variables. The results are shown in Table 2. No significant differences were observed in the metabolites at the baseline between the placebo and probiotic groups. At the 12-week follow-up assessment, the placebo group showed changes in the following plasma metabolites: 8 metabolites showed significant increases [adrenoyl ethanolamide, bilirubin, lactosylceramide, leucine/isoleucine, phenylalanine, tyrosine and valine, phosphatidylcholine (PC) (P-38:6)], whereas anandamide and PC (34:1) showed significant decreases. At the 12-week follow-up assessment, a significant increase was observed in the levels of 2-octenoylcarnitine (C8:1), adrenoylethanolamide, chenodeoxycholic acid glycine conjugate, cis-5-tetradecenoylcarnitine (C14:1), decanoylcarnitine (C10), linoleyl carnitine (C18:2), PC (34:1), PC (38:5), PC (P-38:6) and dodecenoylcarnitine (C12:1) in the probiotic group (Table 2). Next, we compared the plasma metabolite changes from the baseline between the placebo and probiotic groups. The probiotic group showed a significant increase in octenoylcarnitine (q = 0.003), tetradecenoylcarnitine (q = 0.001), decanoylcarnitine (q < 0.001) and dodecenoylcarnitine (q = 0.002). At the 12-week follow-up assessment, the probiotic group showed significantly higher peak intensities of octenoylcarnitine (q < 0.001), tetradecenoylcarnitine (q < 0.001), decanoylcarnitine (q < 0.001), dodecenoylcarnitine (q < 0.001) and PC (34:1) and lower peak intensities of lysophosphatidylcholine (lysoPC) (14:0), lysophosphatidylethanolamine (lysoPE) (18:1) and lysoPE (20:5) (Table 2).

Table 2 Identification of plasma metabolites in the placebo and probiotic groups at the baseline and the 12-week follow-up assessment

Identity	Molecular formula	Mono-isotopic mass	Peak intensities				VIP		q ^a	q ^b	q ^c
			Placebo group (n = 34)		Probiotic group (n = 32)		Baseline vs. follow-up in test group	Placebo follow-up vs. test follow-up
			Baseline	Follow-up	Baseline	Follow-up
Median ± SD. VIP: variable importance in projection. All q-values were tested following a logarithmic transformation; q^a-value is the adjusted P-value, which is derived from an independent t-test for baseline values and controls the false discovery rate. q^b-Value is the adjusted P-value, which is derived from an independent t-test for follow-up values and controls the false discovery rate. q^c-Value is the adjusted P-value, which is derived from an independent t-test for changed value differences and controls the false discovery rate.
2-Octenoylcarnitine	C15H27NO4	286.200	34 895 ± 16 022	42 300 ± 16 991	35 439 ± 16 109	67 961 ± 27 567***	4.081	2.527	0.632	0.001
Change			7669 ± 19 372		27 443 ± 27 416						0.003
Adrenoyl ethanolamide	C24H41NO2	376.317	18 664 ± 15 997	34 458 ± 33 611**	18 285 ± 8048	30 357 ± 35 093**	2.891	0.581	0.376	0.755
Change			8849 ± 37 046		12 804 ± 36 139						0.975
Anandamide (18:4, n-3)	C20H33NO2	320.255	98 113 ± 115 684	37 985 ± 60 917*	69 127 ± 101 656	23 682 ± 140 925	3.282	1.605	0.436	0.708
Change			−57 216 ± 143 517		−40 050 ± 174 760						0.977
Bilirubin	C33H36N4O6	585.268	8186 ± 5437	12 071 ± 9414**	12 353 ± 5718	12 743 ± 15 375	1.068	0.421	0.253	0.497
Change			4819 ± 9914		519 ± 16 216						0.984
Chenodeoxycholic acid glycine conjugate	C26H43NO5	450.319	15 861 ± 29 076	22 169 ± 31 643	8581 ± 11 756	23 227 ± 28 845**	2.206	0.543	0.220	0.697
Change			2572 ± 35 008		6540 ± 31 414						0.947
cis-5-Tetradecenoylcarnitine	C21H39NO4	370.294	15 536 ± 13 064	18 224 ± 16 025	20 161 ± 11 738	42 025 ± 24 076***	3.426	2.518	0.466	<0.001
Change			1821 ± 13 554		18 803 ± 23 697						0.001
Decanoylcarnitine	C17H33NO4	316.247	41 353 ± 23 692	51 906 ± 45 196	50 266 ± 31 771	127 509 ± 58 183***	6.787	4.897	0.373	<0.001
Change			7920 ± 40 471		72 628 ± 54 934						<0.001
Lactosylceramide (d18:1/12:0)	C42H79NO13	806.563	577 449 ± 182 928	635 365 ± 218 635*	601 958 ± 186 962	600 138 ± 250 879	4.148	3.561	0.497	0.636
Change			50 481 ± 164 836		48 511 ± 237 180						0.983
Linoleyl carnitine	C25H45NO4	424.340	51 128 ± 17 619	53 972 ± 17 097	49 633 ± 16 365	59 268 ± 18 501*	1.424	1.174	0.574	0.386
Change			5896 ± 22 938		5012 ± 20 913						0.964
L-Leucine/L-isoleucine	C6H13NO2	132.101	259 346 ± 47 212	277 810 ± 61 050***	257 218 ± 43 954	254 271 ± 44 994	1.070	2.095	0.604	0.200
Change			24 598 ± 39 509		425 ± 42 335						0.453
L-Phenylalanine	C9H11NO2	166.085	298 196 ± 38 248	308 185 ± 42 274*	286 880 ± 35 234	298 838 ± 37 365	1.098	1.350	0.418	0.317
Change			16 836 ± 36 555		18 846 ± 47 793						0.978
L-Tryptophan	C11H12N2O2	205.096	420 022 ± 61 488	405 622 ± 66 081	387 127 ± 68 946	380 937 ± 70 598	1.453	2.009	0.444	0.285
Change			6171 ± 65 997		−7153 ± 55 062						0.964
L-Tyrosine	C9H11NO3	182.080	128 753 ± 22 084	142 806 ± 22 605**	128 158 ± 24 439	142 174 ± 24 097	1.062	0.823	0.550	0.602
Change			8322 ± 20 579		10 705 ± 26 690						0.980
L-Valine	C5H11NO2	118.086	106 915 ± 13 888	113 757 ± 17 977***	107 229 ± 14 443	109 364 ± 13 897	0.453	1.167	0.546	0.200
Change			9254 ± 14 011		4818 ± 15 304						0.441
LysoPC (14:0)	C22H46NO7P	468.306	191 259 ± 62 602	197 769 ± 69 984	160 679 ± 59 919	138 275 ± 41 111	1.371	3.692	0.220	0.010
Change			−2249 ± 92 314		−4864 ± 71 319						0.961
LysoPC (15:0)	C23H48NO7P	482.322	136 071 ± 34 911	135 438 ± 42 891	125 379 ± 49 937	123 569 ± 35 075	1.299	1.720	0.426	0.288
Change			1279 ± 43 434		−700 ± 58 029						0.980
LysoPC (16:0)	C24H50NO7P	496.337	7 911 302 ± 981 548	7 800 353 ± 1 100 088	7 186 667 ± 1 651 882	7 401 677 ± 987 676	8.505	9.633	0.320	0.282
Change			−309 358 ± 1 383 345		340 441 ± 1 836 755						0.981
LysoPC (16:1)	C24H48NO7P	494.321	358 836 ± 92 212	352 522 ± 115 315	285 779 ± 119 295	320 807 ± 68 597	1.449	2.485	0.284	0.255
Change			−25 519 ± 128 269		35 786 ± 123 142						0.983
LysoPC (17:0)	C25H52NO7P	510.353	169 493 ± 39 705	169 256 ± 55 360	163 555 ± 70 853	168 666 ± 54 476	1.732	1.369	0.516	0.746
Change			8430 ± 58 157		2026 ± 84 403						0.981
LysoPC (18:0)	C26H54NO7P	524.368	2 524 370 ± 444 664	2 523 021 ± 569 624	2 308 683 ± 788 463	2 394 730 ± 596 285	5.657	6.066	0.516	0.496
Change			−72 701 ± 623 538		96 537 ± 893 514						0.979
LysoPC (18:1)	C26H52NO7P	522.353	2 255 002 ± 430 202	2 106 593 ± 523 438	2 008 890 ± 566 552	1 959 515 ± 461 353	4.929	5.711	0.439	0.392
Change			−13 978 ± 519 148		−81 397 ± 690 692						0.982
LysoPC (18:2)	C26H50NO7P	520.337	2 858 939 ± 744 384	2 764 738 ± 933 905	2 568 005 ± 589 188	2 473 871 ± 762 047	5.104	6.841	0.278	0.304
Change			−27 714 ± 724 738		−11 376 ± 766 262						0.982
LysoPC (18:3)	C26H48NO7P	518.319	2 366 847 ± 247 361	2 321 522 ± 278 890	2 153 260 ± 334 090	2 223 654 ± 226 409	4.552	4.522	0.254	0.307
Change			−27 577 ± 348 379		77 923 ± 397 424						0.975
LysoPC (20:2)	C28H54NO7P	548.368	48 650 ± 14 453	45 271 ± 18 219	43 780 ± 14 320	43 290 ± 12 385	0.738	1.047	0.280	0.333
Change			−5237 ± 13 679		−1080 ± 17 676						0.983
LysoPC (20:3)	C28H52NO7P	546.353	261 010 ± 91 967	215 766 ± 109 297	226 020 ± 79 881	216 478 ± 49 161	1.995	3.072	0.265	0.198
Change			−8869 ± 77 560		−18 003 ± 87 498						0.974
LysoPC (20:4)	C28H50NO7P	544.337	1 309 520 ± 226 881	1 252 848 ± 318 502	1 188 819 ± 287 103	1 273 091 ± 209 477	3.374	3.635	0.413	0.610
Change			−19 723 ± 285 236		93 833 ± 329 350						0.979
LysoPC (20:5)	C28H48NO7P	542.322	900 448 ± 149 034	885 285 ± 195 560	830 623 ± 150 060	807 360 ± 162 594	2.503	4.029	0.425	0.202
Change			41 099 ± 172 802		−1727 ± 172 024						0.937
LysoPC (22:5)	C30H52NO7P	570.353	80 338 ± 23 881	78 258 ± 32 512	71 864 ± 27 196	72 223 ± 19 298	0.951	1.454	0.472	0.237
Change			2394 ± 29 480		−3078 ± 27 556						0.956
LysoPC (22:6)	C30H50NO7P	568.337	347 641 ± 104 527	364 518 ± 158 401	361 487 ± 119 798	328 312 ± 83 996	2.024	2.858	0.631	0.286
Change			30 429 ± 150 946		−7672 ± 117 998						0.935
LysoPC (P-16:0)	C24H50NO6P	480.343	112 472 ± 26 724	106 976 ± 32 924	103 047 ± 37 472	109 072 ± 34 122	1.505	1.141	0.614	0.687
Change			−3702 ± 36 989		8568 ± 46 280						0.978
LysoPE (16:0)	C21H44NO7P	454.291	62 587 ± 18 587	59 544 ± 22 020	49 518 ± 29 359	50 262 ± 15 759	1.034	1.372	0.380	0.221
Change			−4162 ± 20 985		5346 ± 30 519						0.981
LysoPE (18:1)	C23H46NO7P	480.306	59 151 ± 22 856	53 238 ± 28 464	40 245 ± 17 341	37 270 ± 13 029	0.941	1.930	0.219	0.023
Change			689 ± 24 555		−5829 ± 21 298						0.957
LysoPE (18:2)	C23H44NO7P	478.291	149 773 ± 80 948	132 497 ± 85 664	96 380 ± 40 395	97 739 ± 49 668	1.325	2.816	0.220	0.135
Change			−389 ± 68 207		−12 733 ± 57 233						0.974
LysoPE (20:3)	C25H46NO7P	504.304	38 831 ± 11 582	39 069 ± 13 861	34 437 ± 12 637	31 771 ± 8892	0.626	1.076	0.375	0.218
Change			−570 ± 13 767		−783 ± 15 822						0.974
LysoPE (20:4)	C25H44NO7P	502.290	101 015 ± 32 492	99 479 ± 42 240	86 886 ± 27 057	85 829 ± 26 674	0.914	1.546	0.223	0.229
Change			−9526 ± 39 282		−1367 ± 32 385						0.985
LysoPE (20:5)	C25H42NO7P	500.275	42 974 ± 18 239	45 421 ± 17 863	32 976 ± 11 085	32 072 ± 11 452	0.740	1.774	0.223	0.010
Change			1247 ± 18 455		−1354 ± 14 689						0.937
LysoPE (22:6)	C27H44NO7P	526.291	117 205 ± 46 081	141 015 ± 53 875	122 738 ± 43 119	119 554 ± 36 333	1.049	1.862	0.482	0.213
Change			16 636 ± 52 949		−1734 ± 41 114						0.939
Oleamide	C18H35NO	282.278	1 258 806 ± 677 934	1 196 507 ± 875 468	1 108 012 ± 452 940	949 756 ± 437 199	5.943	9.144	0.264	0.222
Change			−73 276 ± 892 773		430 ± 543 359						0.976
Palmitamide	C16H33NO	256.262	283 497 ± 105 409	255 684 ± 179 302	226 556 ± 77 557	195 703 ± 119 138	2.309	4.296	0.234	0.226
Change			8205 ± 169 982		−27 776 ± 119 511						0.972
PC (34:1)	C42H82NO8P	760.581	209 456 ± 134 065	145 127 ± 65 678*	167 808 ± 65 642	197 479 ± 74 849*	2.689	3.688	0.459	0.019
Change			−23 560 ± 118 192		33 155 ± 77 617						0.158
PC (34:3)	C42H78NO8P	756.550	101 569 ± 47 108	104 238 ± 52 437	114 304 ± 39 540	129 799 ± 58 007	1.320	2.079	0.392	0.224
Change			1167 ± 39 344		9921 ± 50 901						0.952
PC (36:2)	C44H84NO8P	786.597	349 783 ± 152 011	306 883 ± 137 471	302 397 ± 136 950	301 049 ± 153 631	2.534	1.839	0.274	0.659
Change			−25 685 ± 130 773		16 086 ± 164 738						0.817
PC (36:3)	C44H82NO8P	784.581	244 534 ± 89 903	268 692 ± 79 760	256 467 ± 103 724	254 206 ± 144 209	2.360	2.070	0.592	0.677
Change			20 843 ± 91 335		5295 ± 107 697						0.969
PC (36:5)	C44H78NO8P	780.549	1 528 074 ± 546 480	1 705 305 ± 511 360	1 603 113 ± 619 869	1 524 850 ± 702 578	6.076	5.518	0.596	0.703
Change			117 492 ± 507 258		83 735 ± 594 039						0.974
PC (38:5)	C46H82NO8P	808.581	161 745 ± 47 093	165 948 ± 53 864	152 075 ± 37 523	163 267 ± 53 008*	2.067	1.111	0.372	0.654
Change			−7615 ± 47 883		9464 ± 43 071						0.886
PC (40:6)	C48H84NO8P	834.597	77 173 ± 23 647	81 893 ± 23 530	73 703 ± 22 060	68 659 ± 37 805	1.007	0.826	0.492	0.727
Change			5324 ± 27 740		7111 ± 32 919						0.982
PC (P-38:6)	C46H80NO7P	790.570	48 623 ± 23 721	57 387 ± 28 949*	47 040 ± 25 569	54 003 ± 32 395*	1.530	1.143	0.486	0.587
Change			5753 ± 31 034		8692 ± 23 566						0.981
SM (d18:0/16:1)	C39H79N2O6P	703.572	57 898 ± 26 461	59 176 ± 29 207	57 906 ± 26 876	66 527 ± 49 231	1.822	1.194	0.561	0.591
Change			44 ± 26 137		2789 ± 47 890						0.960
Taurocholic acid	C26H45NO7S	516.303	109 376 ± 31 186	106 027 ± 39 733	88 557 ± 38 499	99 801 ± 23 475	0.796	1.381	0.318	0.276
Change			−7591 ± 43 832		11 509 ± 40 732						0.985
trans-2-Dodecenoylcarnitine	C19H35NO4	342.262	17 290 ± 11 877	22 674 ± 13 824	21 695 ± 10 693	46 513 ± 34 370***	3.716	2.852	0.427	<0.001
Change			2764 ± 13 198		20 585 ± 34 310						0.002
Uric acid	C5H4N4O3	169.035	146 921 ± 29 685	152 417 ± 33 376	127 808 ± 38 889	138 640 ± 33 804	0.809	1.181	0.437	0.220
Change			10 476 ± 25 637		−989 ± 26 350						0.944

---

3.3 The relationship between the changes in the body weight, body fat percentage, body fat mass and the total L1 subcutaneous fat areas and the major metabolites

Fig. 2 shows the correlations among the changes from the baseline in body weight, total fat percentage and total body fat mass, the total L1 subcutaneous fat areas and the major metabolites that showed significant differences between the baseline and the 12-week follow-up in the placebo or probiotic groups. The changes in the body weight, total body fat percentage, total body fat mass and L1 subcutaneous fat area were strongly and negatively correlated with the changes in octenoylcarnitine, tetradecenoylcarnitine, decanoylcarnitine and dodecenoylcarnitine in the probiotic group. No significant associations between the changes in body weight and body adiposity and the changes in the levels of plasma medium-chain acylcarnitines (ACs) were observed in the placebo group. However, changes in body weight were positively correlated with changes in leucine/isoleucine and valine levels in the placebo group but not in the probiotic group. Positive interrelations were observed among the changes in the medium-chain ACs with positive interrelations among the changes in the levels of leucine/isoleucine, valine, phenylalanine and tyrosine in both the placebo and probiotic groups (Fig. 2).

Fig. 2 Correlation matrix among the clinical parameters and major metabolites in the placebo and test groups. Correlations were obtained by deriving a Pearson correlation coefficient. Red is a positive correlation and Blue is a negative correlation.

4. Discussion

We showed that supplementation with the two probiotic strains resulted in significant differences in octenoylcarnitine (C8:1), tetradecenoylcarnitine (C14:1), decanoylcarnitine (C10) and dodecenoylcarnitine (C12:1) compared with the placebo group (n = 34). Additionally, we found that in the 32 overweight men and women in the probiotic group, the changes in body weight, whole body adiposity and abdominal fat were strongly and negatively associated with the changes in C8:1, C14:1, C10 and C12:1 ACs; however, no significant association between these measures was observed in the placebo group. Therefore, the probiotic-induced weight loss and adiposity reduction in healthy overweight subjects consuming a weight maintenance diet is associated with an increase in medium-chain ACs.

ACs are derived from mitochondrial and peroxisomal acyl-CoA metabolites by the substitution of the carnitine moiety for CoA. They are the product of incomplete fatty acid oxidation in the mitochondria and are elevated in obesity and type 2 diabetes; they are also positively associated with insulin resistance^15–18 even after adjusting the BMI.¹⁹ However, in this study, no significant differences were observed from the baseline in fasting glucose, insulin, C-peptide levels or the HOMA-IR index in the probiotic group at the 12-week follow-up. We observed that the probiotic-induced weight and adiposity loss were coupled with an increase in lipid-derived medium-chain ACs. Therefore, the current study provides new findings on the effects of probiotic-induced weight loss on metabolic intermediates and particularly on increases in plasma medium-chain ACs. Brady et al.²⁰ previously found an increase in ACs after starvation in both lean and obese Zucker rats. Additionally, Redman et al.²¹ observed that the weight loss with calorie restriction in overweight subjects resulted in an increase in serum ACs. Huffman et al.²² also reported that fasting ACs were increased after 3 months of calorie restriction. In a recent study, the assessment of plasma acylcarnitines before and after the weight loss in obese subjects showed increased levels of acylcarnitines and a strong correlation between non-esterified fatty acids (NEFA) and several acylcarnitines during the weight loss. Since plasma NEFA are indicative of lipolysis,²³ increased lipolysis might drive the generation of acylcarnitines.²⁴ Similarly, in the present study, increased ACs in the probiotic group may have resulted from increased lipolysis during body fat reduction. Additionally, tetradecenoylcarnitine (C14:1) increases in altered fatty acid oxidation states, such as very long chain acyl-CoA dehydrogenase deficiency.²⁵ Indeed, Yoo et al.²⁶ found that the combination of the probiotics L. curvatus and L. plantarum was effective in inhibiting the gene expression of various fatty acid synthesis enzymes in the liver in parallel with the decrease in fatty acid oxidation-related enzymatic activity and gene expression in mice. Therefore, increased lipolysis together with decreased fatty acid oxidation may result in elevations in medium-chain acylcarnitines during the weight loss after probiotic consumption.

Several mechanisms regarding the effects of Lactobacillus strains on adiposity reduction have been proposed. In a study reporting the effect of L. plantarum KY1032 on fat mass reduction, L. plantarum KY1032 cell extracts were suggested as directly decreasing fat mass, by modulating the adipogenesis of 3T3-L1 cells.²⁷ Yoo et al.²⁶ demonstrated that two Lactobacillus strains, L. plantarum KY1032 and L. curvatus HY7601, have an adverse effect on fat accumulation in adipose tissue in diet-induced obese mice by changing the gut microbiome composition or modulating cholesterol metabolism. Additionally, another study reported on the roles of L. plantarum KY1032 and L. curvatus HY7601 in the gut; the two probiotics may reduce fat accumulation by inducing competition for nutrients within the gut or by reducing gut microbiota diversity.²⁶ In the present study, we observed weight loss in the probiotic group while they consumed a weight maintenance diet. Therefore, we hypothesize that the administration of the Lactobacillus strains may have partly affected adipogenesis in adipose tissue or the gut microbiome in our study subjects, resulting in decreased fat accumulation.

Previous studies have reported elevated C3 and C5 ACs, which are the by-products of leucine, isoleucine and valine (branched chain amino acid, BCAA) catabolism, and higher levels of BCAAs and other amino acids including phenylalanine, tryptophan and tyrosine in obese individuals than in lean individuals, which is possibly because of higher insulin resistance.^17,28 Prez-Cornago et al.²⁹ showed a significant reduction in serum isoleucine in overweight adults following a weight loss intervention. In the present study, the subjects were overweight and had slightly higher HOMA-IR indexes (mean HOMA-IR index at baseline = 2.7) compared to the reference range for Koreans.³⁰ Since an association exists between the HOMA-IR index and BCAAs,^17,28 we expected that possible associations might exist among the probiotic-induced weight loss, the HOMA-IR index, and the changes of metabolic intermediates, such as BCAAs. However, in the probiotic group of this study, changes in amino acids were not associated with changes in body weight and body adiposity. Only the placebo group showed significant elevations from the baseline in BCAAs, phenylalanine and tyrosine at the 12-week follow-up and positive correlations between the changes in BCAAs and changes in body weight. No significant change was observed in the HOMA-IR index or the fasting glucose, insulin and C-peptide levels after the 12-week follow-up with no correlation observed between the HOMA-IR index and BCAAs in the placebo or probiotic groups. These findings appear to show a lack of association and interaction between the HOMA-IR index and BCAAs and may be inconsistent with previous studies.^17,28 However, the significant increase of BCAAs observed in the placebo group at the 12-week follow-up may imply that the maintenance of an overweight state may accelerate the increase of BCAAs, whereas the consumption of the probiotics may inhibit their increase. Although we did not observe a correlation between the HOMA-IR index and BCAAs, increased BCAAs may influence insulin resistance in overweight individuals as previously described in several studies.^17,28 Therefore, taken together, we propose that the intake of probiotics may help inhibit rising BCAAs in overweight individuals.

This study has some limitations. We specifically focused on overweight individuals without diabetes. Therefore, our data cannot be generalized to obese individuals or diabetic patients. Additionally, numerous metabolic intermediates were observed using UPLC-LTQ/Orbitrap MS in this study; however, most of these metabolites remain unidentified because the databases of endogenous biomolecules have not yet been constructed for this technique.³¹ Therefore, we identified only a limited number of metabolites among the 238 variables that contributed to the separation between the placebo and probiotic groups. Moreover, since the metabolomics analyses were conducted in some of the participants from the previous study, the statistical power to detect changes in metabolites may be reduced. Regarding gut microbiome alterations in our study subjects, we cannot characterize such changes because we did not measure the amount of the probiotics present in the feces of our study participants. Finally, we cannot determine whether the changes in the metabolic intermediates observed in the present study are derived from the direct effects of the probiotics themselves or the probiotic-induced adiposity reduction.

Despite these limitations, we found significant differences in octenoylcarnitine (C8:1), tetradecenoylcarnitine (C14:1), decanoylcarnitine (C10) and dodecenoylcarnitine (C12:1) in the group of healthy overweight individuals who consumed a weight maintenance diet supplemented with the two probiotic strains compared with the placebo-supplemented group. Furthermore, in the probiotic group, a significant decrease in body weight, whole body adiposity (percentage body fat or body fat mass) and the L1 subcutaneous fat area was strongly associated with an increase in C8:1, C14:1, C10 and C12:1 ACs. However, the precise mechanisms underlying the metabolic changes in the probiotic-induced weight loss should be examined in further studies.

Abbreviations

AC	Acylcarnitine
BCAA	Branched chain amino acid
BMI	Body mass index
CT	Computed tomography
DEXA	Dual-energy X-ray absorptiometry
L1	1st lumbar
L4	4th lumbar
lysoPC	Lysophosphatidylcholine
lysoPE	Lysophosphatidylethanolamine
HOMA-IR	Homeostasis model assessment of insulin resistance
hs-CRP	High-sensitivity C-reactive protein
OPLS-DA	Orthogonal partial least squares-discriminant analysis
PC	Phosphatidylcholine
SE	Standard error
UPLC-LTQ/Orbitrap MS	Ultra-performance liquid chromatography-linear-trap quadrupole-Orbitrap mass spectrometry
VIP	Variable importance in projection

Conflict of interest

The authors declare no conflicts of interest.

Acknowledgements

This study was funded by grants from the Korean Health Technology R&D Project, the Ministry of Health & Welfare, Republic of Korea (HI14C2686010115 and HI14C2686); and the Bio-Synergy Research Project (NRF-2012M3A9C4048762) and the Mid-career Researcher Program (NRF-2016R1A2B4011662) of the Ministry of Science, ICT and the Future Planning through the National Research Foundation, Republic of Korea.

References

1. M. N. Mugambi, A. Musekiwa, M. Lombard, T. Young and R. Blaauw, Synbiotics, probiotics or prebiotics in infant formula for full term infants: a systematic review, Nutr. J., 2012, 11, 81.
2. S. Park and J. H. Bae, Probiotics for weight loss: a systematic review and meta-analysis, Nutr. Res., 2015, 35, 566–575.
3. S. Rao, R. Srinivasjois and S. Patole, Prebiotic supplementation in full-term neonates: a systematic review of randomized controlled trials, Arch. Pediatr. Adolesc. Med., 2009, 163, 755–764.
4. A. Madjd, M. A. Taylor, N. Mousavi, A. Delavari, R. Malekzadeh, I. A. Macdonald and H. R. Farshchi, Comparison of the effect of daily consumption of probiotic compared with low-fat conventional yogurt on weight loss in healthy obese women following an energy-restricted diet: a randomized controlled trial, Am. J. Clin. Nutr., 2016, 103, 323–329.
5. T. Dror, Y. Dickstein, G. Dubourg and M. Paul, Microbiota manipulation for weight change, Microb. Pathog. DOI:10.1016/j.micpath.2016.01.002 , in press.
6. H. Y. Ahn, M. Kim, J. S. Chae, Y. T. Ahn, J. H. Sim, I. D. Choi, S. H. Lee and J. H. Lee, Supplementation with two probiotic strains, Lactobacillus curvatus HY7601 and Lactobacillus plantarum KY1032, reduced body adiposity and Lp-PLA₂ activity in overweight subjects, J. Funct. Foods, 2015, 19, 744–752.
7. M. Le Barz, F. F. Anhê, T. V. Varin, Y. Desjardins, E. Levy, D. Roy, M. C. Urdaci and A. Marette, Probiotics as complementary treatment for metabolic disorders, J. Diabetes Metab., 2015, 39, 291–303.
8. A. Moya-Pérez, M. Romo-Vaquero, F. Tomás-Barberán, Y. Sanz and M. T. García-Conesa, Hepatic molecular responses to Bifidobacterium pseudocatenulatum CECT 7765 in a mouse model of diet-induced obesity, Nutr., Metab. Cardiovasc. Dis., 2014, 24, 57–64.
9. P. D. Cani, J. Amar, M. A. Iglesias, M. Poggi, C. Knauf, D. Bastelica, A. M. Neyrinck, F. Fava, K. M. Tuohy, C. Chabo, A. Waget, E. Delmée, B. Cousin, T. Sulpice, B. Chamontin, J. Ferrières, J. F. Tanti, G. R. Gibson, L. Casteilla, N. M. Delzenne, M. C. Alessi and R. Burcelin, Metabolic endotoxemia initiates obesity and insulin resistance, Diabetes, 2007, 56, 1761–1772.
10. J. Amar, C. Chabo, A. Waget, P. Klopp, C. Vachoux, L. G. Bermúdez-Humarán, N. Smirnova, M. Bergé, T. Sulpice, S. Lahtinen, A. Ouwehand, P. Langella, N. Rautonen, P. J. Sansonetti and R. Burcelin, Intestinal mucosal adherence and translocation of commensal bacteria at the early onset of type 2 diabetes: molecular mechanisms and probiotic treatment, EMBO Mol. Med., 2011, 3, 559–572.
11. H. H. Chen, Y. J. Tseng, S. Y. Wang, Y. S. Tsai, C. S. Chang, T. C. Kuo, W. J. Yao, C. C. Shieh, C. H. Wu and P. H. Kuo, The metabolome profiling and pathway analysis in metabolic healthy and abnormal obesity, Int. J. Obes., 2015, 39, 1241–1248.
12. S. Y. Lee, M. Kim, S. Jung, S. H. Lee and J. H. Lee, Altered plasma lysophosphatidylcholines and amides in non-obese and non-diabetic subjects with borderline-to-moderate hypertriglyceridemia: A case-control study, PLoS One, 2015, 10, e0123306.
13. S. H. Lee, S. Park, H. Kim and B. H. Jung, Metabolomic approaches to the normal aging process, Metabolomics, 2014, 10, 1268–1292.
14. M. Bylesjö, D. Eriksson, A. Sjödin, S. Jansson, T. Moritz and J. Trygg, Orthogonal projections to latent structures as a strategy for microarray data normalization, BMC Bioinf., 2007, 8, 207.
15. M. J. Patel, B. C. Batch, L. P. Svetkey, J. R. Bain, C. B. Turer, C. Haynes, M. J. Muehlbauer, R. D. Stevens, C. B. Newgard and S. H. Shah, Race and sex differences in small-molecule metabolites and metabolic hormones in overweight and obese adults, OMICS, 2013, 17, 627–635.
16. K. M. Huffman, S. H. Shah, R. D. Stevens, J. R. Bain, M. Muehlbauer, C. A. Slentz, C. J. Tanner, M. Kuchibhatla, J. A. Houmard, C. B. Newgard and W. E. Kraus, Relationships between circulating metabolic intermediates and insulin action in overweight to obese, inactive men and women, Diabetes Care, 2009, 32, 1678–1683.
17. C. B. Newgard, J. An, J. R. Bain, M. J. Muehlbauer, R. D. Stevens, L. F. Lien, A. M. Haqq, S. H. Shah, M. Arlotto, C. A. Slentz, J. Rochon, D. Gallup, O. Ilkayeva, B. R. Wenner, W. S. Yancy Jr., H. Eisenson, G. Musante, R. S. Surwit, D. S. Millington, M. D. Butler and L. P. Svetkey, A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance, Cell Metab., 2009, 9, 311–326.
18. S. H. Adams, C. L. Hoppel, K. H. Lok, L. Zhao, S. W. Wong, P. E. Minkler, D. H. Hwang, J. W. Newman and W. T. Garvey, Plasma acylcarnitine profiles suggest incomplete long-chain fatty acid beta-oxidation and altered tricarboxylic acid cycle activity in type 2 diabetic African-American women, J. Nutr., 2009, 139, 1073–1081.
19. E. S. Tai, M. L. Tan, R. D. Stevens, Y. L. Low, M. J. Muehlbauer, D. L. Goh, O. R. Ilkayeva, B. R. Wenner, J. R. Bain, J. J. Lee, S. C. Lim, C. M. Khoo, S. H. Shah and C. B. Newgard, Insulin resistance is associated with a metabolic profile of altered protein metabolism in Chinese and Asian-Indian men, Diabetologia, 2010, 53, 757–767.
20. L. J. Brady, P. S. Brady, L. Albers, A. T. Davis and C. L. Hoppel, Carnitine metabolism in lean and obese Zucker rats during starvation, J. Nutr., 1986, 116, 668–674.
21. L. M. Redman, K. M. Huffman, L. R. Landerman, C. F. Pieper, J. R. Bain, M. J. Muehlbauer, R. D. Stevens, B. R. Wenner, V. B. Kraus, C. B. Newgard, W. E. Kraus and E. Ravussin, Effect of caloric restriction with and without exercise on metabolic intermediates in nonobese men and women, J. Clin. Endocrinol. Metab., 2011, 96, E312–E321.
22. K. M. Huffman, L. M. Redman, L. R. Landerman, C. F. Pieper, R. D. Stevens, M. J. Muehlbauer, B. R. Wenner, J. R. Bain, V. B. Kraus, C. B. Newgard, E. Ravussin and W. E. Kraus, Caloric restriction alters the metabolic response to a mixed-meal: results from a randomized, controlled trial, PLoS One, 2012, 7, e28190.
23. S. Klein, Y. Sakurai, J. A. Romijn and R. M. Carroll, Progressive alterations in lipid and glucose metabolism during short-term fasting in young adult men, Am. J. Physiol., 1993, 265, E801–E806.
24. M. G. Schooneman, A. Napolitano, S. M. Houten, G. K. Ambler, P. R. Murgatroyd, S. R. Miller, C. E. Hollak, C. Y. Tan, S. Virtue, A. Vidal-Puig, D. J. Nunez and M. R. Soeters, Assessment of plasma acylcarnitines before and after weight loss in obese subjects, Arch. Biochem. Biophys., 2016, 606, 73–80.
25. C. G. Costa, E. A. Struys, A. Bootsma, H. J. ten Brink, L. Dorland, I. Tavares de Almeida, M. Duran and C. Jakobs, Quantitative analysis of plasma acylcarnitines using gas chromatography chemical ionization mass fragmentography, J. Lipid Res., 1997, 38, 173–182.
26. S. R. Yoo, Y. J. Kim, D. Y. Park, U. J. Jung, S. M. Jeon, Y. T. Ahn, C. S. Huh, R. McGregor and M. S. Choi, Probiotics L. plantarum and L. curvatus in combination alter hepatic lipid metabolism and suppress diet-induced obesity, Obesity, 2013, 21, 2571–2578.
27. D. Y. Park, Y. T. Ahn, C. S. Huh, S. M. Jeon and M. S. Choi, The inhibitory effect of Lactobacillus plantarum KY1032 cell extract on the adipogenesis of 3T3-L1 Cells, J. Med. Food, 2011, 14, 670–675.
28. J. Y. Kim, J. Y. Park, O. Y. Kim, B. M. Ham, H. J. Kim, D. Y. Kwon, Y. Jang and J. H. Lee, Metabolic profiling of plasma in overweight/obese and lean men using ultra performance liquid chromatography and Q-TOF mass spectrometry (UPLC-Q-TOF MS), J. Proteome Res., 2010, 9, 4368–4375.
29. A. Perez-Cornago, L. Brennan, I. Ibero-Baraibar, H. H. Hermsdorff, A. O'Gorman, M. A. Zulet and J. A. Martínez, Metabolomics identifies changes in fatty acid and amino acid profiles in serum of overweight older adults following a weight loss intervention, J. Physiol. Biochem., 2014, 70, 593–602.
30. K. J. Yun, K. Han, M. K. Kim, Y. M. Park, K. H. Baek, K. H. Song and H. S. Kwon, Insulin Resistance Distribution and Cut-Off Value in Koreans from the 2008-2010 Korean National Health and Nutrition Examination Survey, PLoS One, 2016, 11, e0154593.
31. C. Ezzili, K. Otrubova and D. L. Boger, Fatty acid amide signaling molecules, Bioorg. Med. Chem. Lett., 2010, 20, 5959–5968.

---