https://orcid.org/0000-0002-1652-8083
Abstract
The brown adipose tissue (BAT) is a potential target for the treatment of obesity and metabolic disorders. Its activation by cold exposure or adrenergic drugs can increase systemic insulin sensitivity and improve lipid metabolism; however, little is known about the effects of specific dietary components on BAT activity.
We asked if a short-term (4 weeks) dietary intervention with olive oil could modify BAT activity in lean and overweight/obese volunteers.
This was a 4-week open clinical trial in which all participants underwent a dietary intervention with extra-virgin olive oil supplementation. As the initial intake of olive oil was controlled all the participants were controls of themselves.
The intervention resulted in significant increase in blood monounsaturated fatty acid levels, which was accompanied by increased BAT activity in lean but not in overweight/obese volunteers. In the lean group, an increase in leptin was detected after the intervention, and low leptin values at the beginning of the study were predictive of greater BAT activity after intervention. In addition, increase in leptin concentration was associated with increased BAT activity. Three known endogenous mediators of BAT activity, secretin, fibroblast growth factor 21 (FGF21), and 12,13-dihydroxy-9Z-octadecenoic acid (12,13-diHOME) were increased by intervention in lean, whereas only secretin and FGF21 were increased in subjects with excessive weight.
This study provides clinical evidence for the impact of monounsaturated fatty acids on BAT activity and an advance in the understanding of the beneficial health effects of olive oil.
The brown adipose tissue (BAT) is regarded as a potential target for the treatment of metabolic disorders (1-4). Upon stimulus by cold, BAT dissipates energy in order to produce heat (5-7). During this process, fatty acids and carbohydrates are used as fuel impacting on whole-body energy status and metabolism (8-10). This has been tested in clinical interventions providing evidence for the role of BAT in glucose and lipid metabolism, as well as in body adiposity (10-13).
Unfortunately, cold exposure is not a feasible intervention for large-scale, population-based management of metabolic diseases. Thus, studies have focused on the identification of other strategies that could stimulate BAT activity resulting in health benefits (14). Adrenergic agonists and physical activity are examples of pharmacological and behavioral strategies, respectively, that can activate BAT (3, 15, 16). However, little is known about the impact of specific nutritional interventions on BAT activity (17). There is evidence for the BAT-stimulating effects of the nutrient components capsaicin and catechins (12, 18, 19); nevertheless, data on their actual impact on clinical management of metabolic diseases are still missing (17).
Here, we tested the hypothesis that a dietary intervention with oleic acid–rich olive oil could increase BAT activity in humans. There is undisputed evidence for the beneficial metabolic and cardiovascular effects of mono- and polyunsaturated fatty acids (MUFAs and PUFAs, respectively) in human health. They can improve blood lipid composition (20), prevent primary cardiovascular events, and increase systemic insulin sensitivity (21). For these reasons, the clinical and nutritional management guidelines from major metabolism and cardiovascular societies encourage the increased dietary consumption of MUFAs and PUFAs (22, 23). Logistically, dietary intervention with olive oil is feasible and could be applied for large-scale, population-based purposes. In this proof-of-concept study, we evaluated 41 subjects with low baseline olive oil intake who consumed approximately 0.6 L of the oil during a 4-week intervention.
Methods
Human studies
Study participants.
The study was conducted from April 2017 to May 2019. All participants were recruited by electronic advertisements. Initial screening of +200 subjects resulted in the pre-selection of 109 candidates, with the final inclusion of 49 subjects. Inclusion criteria were: (i) men and women from 25 to 40 years of age; (ii) for the lean group body mass index (BMI) of 18.5 to 24.9 kg/m2; (iii) for overweight/obese group BMI of 25 to 35 kg/m2 or waist circumference +80 cm for women and +88 cm for men, according to the International Diabetes Federation metabolic syndrome criteria for South American people (24). Exclusion criteria were: (i) current individual olive oil intake larger than 250 mL/month; (ii) ongoing pregnancy; (iii) diagnosis of severe neurologic or psychiatric diseases; (iv) use of anti-obesity or lipid-lowering medication; (v) use of adrenergic or benzodiazepine drugs; (vi) current diagnosis of cancer, communicable diseases, rheumatic diseases, liver or kidney failure, thyroid dysfunction or diabetes mellitus; (vii) +5% change in body mass during the previous 6 months; (viii) currently on any dietary approach aimed at body mass modification; (ix) metallic prosthesis. Because there were no previous studies evaluating the impact of dietary fatty acids supplementation in human BAT, we calculated the size of the sample on 50% of detectable BAT from previous studies (7, 25). A total of 44 volunteers concluded the study; however, 3 were excluded: 2 due to 1 positron emission tomography (PET) data missing, and 1 was regarded as an outlier due to a very high BAT glucose uptake at baseline (Suppl. Fig. 1) (26). All volunteers read and signed a written informed consent prior to inclusion in this study. This study was approved by the local ethical committees. This study is registered in ClinicalTrials.gov (NCT03024359) as of January 2017.
Study design.
The study design is depicted in Fig. 1A and 1B. This was a 4-week open clinical trial in which all participants underwent a dietary intervention with extra-virgin olive oil supplementation. Since the amount of olive oil consumption was an inclusion criteria (less than 250 mL/month), the subjects were their own controls. For the dietary intervention, participants received 2 liters of extra-virgin olive oil for 4 weeks. The participants were recommended to replace all the vegetable oil used in their meals for olive oil; they were not encouraged to consume all the olive oil given, but to increase their usual consumption by changing their habitual vegetable oil for olive oil. During the intervention, volunteers were recommended to maintain their regular physical activity and diet except for the olive oil used in all preparations. All subjects underwent 18F-fluorodeoxyglucose (18F-FDG) PET/magnetic resonance imaging (MRI) acquisitions before and at the end of the intervention. In addition, anthropometry, body composition, metabolic and inflammatory parameters, as well as BAT, white adipose tissue (WAT), and muscle activity (as standardized uptake value; SUV) were determined before and at the end of the intervention.
General aspects related to intervention and PET/MRI scan protocols. (A) Schematic representation of study design. (B) Schematic representation of protocols employed in static and dynamic PET/MRI scans. (C) Mean olive oil consumption during the intervention. Mean daily energy intake (D), mean relative macronutrient energy intake (E), mean relative fatty acid intake (F), plasma fatty acid concentration (G), physical activity (H), mean vest temperature during cold exposure (I), and cold sensation (J), at baseline and at the end of the intervention. (K), Mean outdoor temperature during the 1-month period preceding the beginning of the intervention and during the month intervention occurred. *P < 0.05.
Olive oil selection.
In order to select the olive oil that was given to participants, 3 brands were tested by chromatography (Suppl. Table 1) (26). The brand with the highest content of monounsaturated fatty acids (a Portuguese extra-virgin olive oil) was selected, and the 2 batches acquired for the study were also analyzed. The composition of the 3 brands analyzed and the 2 batches of the selected olive oil are shown in Supplementary Tables 1 and 2 (26).
Clinical variables.
All clinical and dietary data were collected by trained staff. Body composition was assessed by dual-energy x-ray absorptiometry (DXA Lunar, General Electrics) before and at the end of the 4-week intervention. Visceral fat mass between the L3 and L4 vertebrae was calculated using the MRI fat fraction image. A threshold of 40% of triglycerides was used to determine the fat area between the 2 vertebrae.
Dietary and physical activity data.
The short version of the International Physical Activity Questionnaire was used to assess total physical activity before and at the end of the intervention period. Dietary intake was assessed by 2 separate 24-hour recalls in nonconsecutive days (1 weekday and 1 weekend day) using the multiple-pass method before the 2 exam days (before and after the intervention). Dietary data were analyzed using the Nutrition Data System for Research software (NDSR, Nutrition Coordinating Center, University of Minnesota, Minneapolis, USA).
Biochemical, hormonal, and immunological determinations.
Eight-hour fasting venous blood samples were collected before and at the end of the intervention. Volunteers were instructed to abstain from alcohol and moderate to intense physical activity 48 hours prior to blood and 18F-FDG PET/MRI data collection. Fasting glucose was assessed using an Accu-Chek blood glucose monitoring system (Roche, Bâle, Switzerland). Biochemical parameters were determined by enzymatic methods. Insulin, leptin, monocyte chemoattractant protein-1 (MCP-1), interleukin-6, CXCL14, 12,13-diHOME, secretin, and FGF21 were measured by enzyme-linked immunosorbent assays (R&D Systems, Minneapolis, MN, USA; RayBiotech, Peachtree Corners, GA, USA; MyBiosource, San Diego, CA, USA; Cayman Chemical, Ann Arbor, MI, USA). Insulin resistance (IR) was estimated by homeostasis model assessment of insulin resistance (HOMA-IR).
18F-FDG PET/MRI scanning protocol—static images.
Acquisition protocols are shown in Fig. 1B. Blood samples were collected from fasting volunteers, before cold exposure. Volunteers were submitted to individual air conditioning and a cooling vest. During the cold exposure, the subjects were placed seated in air-conditioned room (19 °C) wearing a cooling vest (Polar Products, Stow, OH, USA) set for 10 to 16 °C for 2 hours. To avoid shivering, a visual scale of cold sensation was used every 30 minutes (27). In addition, researchers checked for visual shivering every 15 minutes. For the static protocol, subjects received an intravenous bolus of 185 MBq of 18F-FDG 1 hour after the beginning of cold exposure; thereafter, subjects remained under cold exposure for 1 additional hour until PET/MRI started. PET/MRI was performed in accordance with protocols published elsewhere (28). Volunteers were placed supine in a headfirst position inside the PET/MRI scan (General Electrics). This protocol was employed for 33 volunteers.
18F-FDG PET/MRI scanning protocol—dynamic images.
Volunteers were submitted to the same cold exposure protocol as described in the previous section. For the dynamic protocol, subjects received an intravenous bolus of 185 MBq 18F-FDG 2 hours after the beginning of cold exposure with the subjects inside the PET/MRI machine; thereafter the PET/MRI started immediately after the FDG injection. The dynamic uptake of 18F-FDG was determined in the neck and upper thoracic region as follows: 4 × 30 seconds; 1 × 60 seconds; 1 × 120 seconds; 3 × 300 seconds; and 2 × 600 seconds. The dynamic scan of the abdominal area was performed just after the upper thoracic region as following: 5 × 180 seconds. After the dynamic scans, a single static image was taken for the static analysis (approximately 1 hour after the FDG injection) to compare with the static images. During the scanning, 4 plastic bags filled with ice were placed against the feet to maintain cooling inside the scanner room. This protocol was employed for 8 volunteers (5 lean and 3 overweight).
18F-FDG PET/MRI data analysis.
All PET/MRI images were analyzed using CARIMAS 2.9 software (Turku PET Centre, Turku, Finland). For BAT, visceral WAT, perirenal WAT, muscle (trapezius), aorta arch, and cerebellum activity, the mean standardized uptake values (SUVs) were estimated by drawing a volume of interest (VOI) with the mask tool of the software on the fused image of the static 18F-FDG PET image and the MR image (as an anatomical reference). The sphere tool was used only for liver and subcutaneous WAT in order to improve data analysis. For BAT activity, the VOIs were drawn in transaxial planes bilaterally in the cervical and supraclavicular areas, and the mean value between the 2 measurements, adjusted to body weight, was used as the mean SUV. The fused PET/MRI images were used for the estimation of BAT volume in the cervical upper thoracic region (including cervical, supraclavicular, and axillary adipose tissue). First, the VOIs of potential BAT sites were drawn manually from the chin to the armpit in both sides, and a lower bound threshold of 40% of triglyceride content was used in the fat fraction MR images. Following the first threshold, the VOI was masked to the background (PET), and the second threshold of voxels with SUVs higher than 1.5 was used. The volume of cervical, supraclavicular, and axillary tissue with more than 40% of triglyceride content and SUV higher than 1.5 g/mL was considered to be brown fat, according to data published elsewhere. For the dynamic PET scans, the metabolic rate of glucose uptake was calculated. The calculation of BAT metabolic rate of glucose uptake was done using Patlak plot (29), while the uptake rate of glucose of white adipose tissue (WAT) was calculated using a fractional uptake rate mathematical model. The blood radioactivity was determined by drawing a VOI in the arch of the aorta. The VOIs of the tissues of interest (BAT, WAT, and muscle) were drawn in the fused images. The plasma time-activity curve (TAC) acted as an input function for the Patlak and fractional uptake rate models for the calculation of tissue-specific influx rate (Ki) of 18F-FDG. Glucose uptake rates of the tissues were calculated by multiplying blood glucose concentration with tissue-specific influx rates (Ki), as described previously. The calculation of BAT metabolic rate of glucose uptake using Patlak plot (29) and the uptake rate of glucose of WAT (using fractional uptake rate) was performed by drawing a VOI in the aorta arch to estimate blood radioactivity. The VOIs of the tissues of interest (BAT, WAT, and muscle) were drawn in the fused images, and the aorta input was used to estimate the glucose uptake of BAT, WAT, and muscle as previously described.
Chromatographic analyses.
The fatty acids present in olive oil and serum were esterified. Chromatographic analyses were performed using a gas chromatograph-mass spectrometer (model GCMS-QP2010 Ultra; Shimadzu).
Statistical analyses.
Statistical analyses were performed using SPSS version 22.0 and Graphpad Prism 8.0. All values are presented as mean and standard error of the mean. The Kolmogorov-Smirnov test was used to check the normal distribution of variables. For nonparametrical variables, the equivalent nonparametric test was used. For comparisons between lean and obese subjects, an unpaired Student t test, or Mann-Whitney for variables without normal distribution, was used. Variables before and after the intervention were compared using paired Student t test or Wilcoxon. Correlations were tested by Pearson or Spearman coefficients. Multiple regression analysis was used to analyze independent associations of variables with changes in BAT activity. A P value of less than 0.05 was considered as statistically significant.
Single-cell.
RNA sequence analytical methods are described in Supplementary data (26).
Results
Dietary intervention with olive oil increased plasma MUFA both in lean and overweight/obese subjects
At baseline, overweight/obese volunteers presented with significantly higher BMI, waist circumference, fat mass, lean mass, blood pressure, blood insulin, HOMA-IR, blood leptin, total and low-density lipoprotein cholesterol, and triglycerides compared with lean volunteers; high-density lipoprotein cholesterol was lower, whereas BAT activity and volume were not significantly different from lean volunteers (Table 1). During intervention (Fig. 1A), total energy (Fig. 1D) and macronutrient (Fig. 1E) intake were similar between the groups; however, olive oil (Fig. 1C) and MUFA (Fig. 1F) intake were significantly increased in both lean and overweight/obese subjects at the end of the intervention. Also, the amount of olive oil consumed per body weight was increased by nearly 300% in both groups (Suppl. Fig. 2) (26), being greater in the lean group before and after the intervention period. Gas chromatography revealed a significant increase in plasma MUFA, a significant decrease in plasma PUFA, and no change in plasma saturated fat in both groups (Fig. 1G). Volunteers did not change physical activity during the intervention (Fig. 1H). Cold exposure protocol is detailed in Fig. 1B; there were no differences in mean vest temperature (Fig. 1I), in cold sensation during cold exposure (Fig. 1J), and mean outdoor temperature in the previous month of the day when the PET scan was performed (Fig. 1K).
Baseline Phenotypes of Volunteers According to Body Mass Index
| Lean | Overweight/obese | P | |
|---|---|---|---|
| Subjects, n | 26 | 15 | -- |
| Sex, female/male | 15/11 | 8/7 | -- |
| Age, years | 31.8 ± 0.9 | 34.4 ± 1.3 | 0.10 |
| BMI, kg/m2 | 21.7 ± 0.4 | 30.2 ± 1 | <0.01 |
| Waist circumference, cm | 76.6 ± 1.3 | 97.6 ± 1.7 | <0.01 |
| Total fat mass, % | 27.6 ± 1 | 36.9 ± 2.2 | <0.01 |
| Total lean mass, kg | 42.3 ± 1.7 | 51.1 ± 2.8 | <0.01 |
| Visceral adipose tissue, gd | 207.9 ± 33.5 | 1080.4 ± 133.4 | <0.01 |
| L3-L4 visceral fat, gc,d | 32.4 ± 4.8 | 106.9 ± 11.3 | <0.01 |
| Mean blood pressure, mmHgc | 81.1 ± 2.1 | 89.5 ± 4.1 | 0.05 |
| Fasting glucose, mg/dL | 75.4 ± 1.8 | 80.2 ± 2.1 | 0.11 |
| Insulin, mU/L | 3.7 ± 0.2 | 6.0 ± 0.7 | <0.01 |
| HOMA-IR | 0.7 ± 0.1 | 1.2 ± 0.1 | <0.01 |
| Leptin, pg/mL | 61.6 ± 7.1 | 258.8 ± 56.2 | <0.01 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 206.9 ± 8 | 0.02 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 130.6 ± 6.8 | <0.01 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 44.9 ± 2.7 | <0.01 |
| Triglycerides, mg/dLd | 76.5 ± 5.9 | 133 ± 19 | <0.01 |
| BAT activity, mean SUVd | 0.8 ± 0.1 | 0.6 ± 0.2 | 0.33 |
| BAT volume, mLd | 5.0 ± 1.7 | 13.0 ± 8.5 | 0.79 |
| Lean | Overweight/obese | P | |
|---|---|---|---|
| Subjects, n | 26 | 15 | -- |
| Sex, female/male | 15/11 | 8/7 | -- |
| Age, years | 31.8 ± 0.9 | 34.4 ± 1.3 | 0.10 |
| BMI, kg/m2 | 21.7 ± 0.4 | 30.2 ± 1 | <0.01 |
| Waist circumference, cm | 76.6 ± 1.3 | 97.6 ± 1.7 | <0.01 |
| Total fat mass, % | 27.6 ± 1 | 36.9 ± 2.2 | <0.01 |
| Total lean mass, kg | 42.3 ± 1.7 | 51.1 ± 2.8 | <0.01 |
| Visceral adipose tissue, gd | 207.9 ± 33.5 | 1080.4 ± 133.4 | <0.01 |
| L3-L4 visceral fat, gc,d | 32.4 ± 4.8 | 106.9 ± 11.3 | <0.01 |
| Mean blood pressure, mmHgc | 81.1 ± 2.1 | 89.5 ± 4.1 | 0.05 |
| Fasting glucose, mg/dL | 75.4 ± 1.8 | 80.2 ± 2.1 | 0.11 |
| Insulin, mU/L | 3.7 ± 0.2 | 6.0 ± 0.7 | <0.01 |
| HOMA-IR | 0.7 ± 0.1 | 1.2 ± 0.1 | <0.01 |
| Leptin, pg/mL | 61.6 ± 7.1 | 258.8 ± 56.2 | <0.01 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 206.9 ± 8 | 0.02 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 130.6 ± 6.8 | <0.01 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 44.9 ± 2.7 | <0.01 |
| Triglycerides, mg/dLd | 76.5 ± 5.9 | 133 ± 19 | <0.01 |
| BAT activity, mean SUVd | 0.8 ± 0.1 | 0.6 ± 0.2 | 0.33 |
| BAT volume, mLd | 5.0 ± 1.7 | 13.0 ± 8.5 | 0.79 |
Abbreviations: BAT, brown adipose tissue; BMI, body mass index; L3-L4, lumbar vertebrae; SUV, standardized uptake value
a Independent sample t test
b measured by MRI
c n = 33
d nonparametric test
Baseline Phenotypes of Volunteers According to Body Mass Index
| Lean | Overweight/obese | P | |
|---|---|---|---|
| Subjects, n | 26 | 15 | -- |
| Sex, female/male | 15/11 | 8/7 | -- |
| Age, years | 31.8 ± 0.9 | 34.4 ± 1.3 | 0.10 |
| BMI, kg/m2 | 21.7 ± 0.4 | 30.2 ± 1 | <0.01 |
| Waist circumference, cm | 76.6 ± 1.3 | 97.6 ± 1.7 | <0.01 |
| Total fat mass, % | 27.6 ± 1 | 36.9 ± 2.2 | <0.01 |
| Total lean mass, kg | 42.3 ± 1.7 | 51.1 ± 2.8 | <0.01 |
| Visceral adipose tissue, gd | 207.9 ± 33.5 | 1080.4 ± 133.4 | <0.01 |
| L3-L4 visceral fat, gc,d | 32.4 ± 4.8 | 106.9 ± 11.3 | <0.01 |
| Mean blood pressure, mmHgc | 81.1 ± 2.1 | 89.5 ± 4.1 | 0.05 |
| Fasting glucose, mg/dL | 75.4 ± 1.8 | 80.2 ± 2.1 | 0.11 |
| Insulin, mU/L | 3.7 ± 0.2 | 6.0 ± 0.7 | <0.01 |
| HOMA-IR | 0.7 ± 0.1 | 1.2 ± 0.1 | <0.01 |
| Leptin, pg/mL | 61.6 ± 7.1 | 258.8 ± 56.2 | <0.01 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 206.9 ± 8 | 0.02 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 130.6 ± 6.8 | <0.01 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 44.9 ± 2.7 | <0.01 |
| Triglycerides, mg/dLd | 76.5 ± 5.9 | 133 ± 19 | <0.01 |
| BAT activity, mean SUVd | 0.8 ± 0.1 | 0.6 ± 0.2 | 0.33 |
| BAT volume, mLd | 5.0 ± 1.7 | 13.0 ± 8.5 | 0.79 |
| Lean | Overweight/obese | P | |
|---|---|---|---|
| Subjects, n | 26 | 15 | -- |
| Sex, female/male | 15/11 | 8/7 | -- |
| Age, years | 31.8 ± 0.9 | 34.4 ± 1.3 | 0.10 |
| BMI, kg/m2 | 21.7 ± 0.4 | 30.2 ± 1 | <0.01 |
| Waist circumference, cm | 76.6 ± 1.3 | 97.6 ± 1.7 | <0.01 |
| Total fat mass, % | 27.6 ± 1 | 36.9 ± 2.2 | <0.01 |
| Total lean mass, kg | 42.3 ± 1.7 | 51.1 ± 2.8 | <0.01 |
| Visceral adipose tissue, gd | 207.9 ± 33.5 | 1080.4 ± 133.4 | <0.01 |
| L3-L4 visceral fat, gc,d | 32.4 ± 4.8 | 106.9 ± 11.3 | <0.01 |
| Mean blood pressure, mmHgc | 81.1 ± 2.1 | 89.5 ± 4.1 | 0.05 |
| Fasting glucose, mg/dL | 75.4 ± 1.8 | 80.2 ± 2.1 | 0.11 |
| Insulin, mU/L | 3.7 ± 0.2 | 6.0 ± 0.7 | <0.01 |
| HOMA-IR | 0.7 ± 0.1 | 1.2 ± 0.1 | <0.01 |
| Leptin, pg/mL | 61.6 ± 7.1 | 258.8 ± 56.2 | <0.01 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 206.9 ± 8 | 0.02 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 130.6 ± 6.8 | <0.01 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 44.9 ± 2.7 | <0.01 |
| Triglycerides, mg/dLd | 76.5 ± 5.9 | 133 ± 19 | <0.01 |
| BAT activity, mean SUVd | 0.8 ± 0.1 | 0.6 ± 0.2 | 0.33 |
| BAT volume, mLd | 5.0 ± 1.7 | 13.0 ± 8.5 | 0.79 |
Abbreviations: BAT, brown adipose tissue; BMI, body mass index; L3-L4, lumbar vertebrae; SUV, standardized uptake value
a Independent sample t test
b measured by MRI
c n = 33
d nonparametric test
Anthropometric, hormonal, and inflammatory protein measurements. Mean values of (A) body weight, (B) waist circumference and (C) energy efficiency, and mean blood concentrations of insulin (D), leptin (E), interleukin-6 (F) and MCP-1 (G) at baseline and at the end of the intervention in lean and overweight/obese volunteers. *P < 0.05. Abbreviation: MCP-1, monocyte chemoattractant protein-1.
Dietary intervention with olive oil reduced visceral fat and systemic markers of inflammation both in lean and overweight/obese subjects
The intervention resulted in no changes in body weight (Fig. 2A), waist circumference (Fig. 2B), and energy efficiency (Fig. 2C). In volunteers with excessive weight, the intervention resulted in increased total fat mass (Table 2) and a trend in body weight (P = 0.06); however, both in overweight/obese and lean groups, visceral fat mass was reduced, which was accompanied by an increase in subcutaneous adipose tissue (between L3 and L4 vertebrae) in the lean group and a trend to increase in the overweight/obese subjects (Table 2); however, no changes were detected in the android/gynoid ratio (Table 2). In lean, but not in overweight/obese subjects, the intervention reduced total cholesterol and reached borderline significance for a decrease of low-density lipoprotein cholesterol levels (Table 2). In addition, only in the lean group, the dietary intervention increased blood insulin (Fig. 2D) and leptin (Fig. 2E), and this was accompanied by increased HOMA-IR (Table 2). Both in lean and overweight/obese volunteers, systemic interleukin-6 was reduced by the intervention (Fig. 2F); in the overweight/obese group, but not in the lean, blood levels of MCP-1 were reduced by the intervention (Fig. 2G).
Baseline and Intervention-Induced Changes in Clinical and Biochemical Parameters
| Lean | Overweight/Obese | |||||
|---|---|---|---|---|---|---|
| Baseline | Intervention | P | Baseline | Intervention | P | |
| BMI, kg/m2 | 21.7 ± 0.4 | 21.8 ± 0.4 | 0.09 | 30.2 ± 1.0 | 30.6 ± 1.0 | 0.05 |
| Total fat mass, kg | 27.6 ± 1.0 | 27.5 ± 1.0 | 0.76 | 36.9 ± 2.2 | 37.4 ± 2.1 | <0.05 |
| L3-L4 visceral AT, gb | 34.4 ± 3.3 | 31.1 ± 3.4 | <0.05 | 121.2 ± 12.5 | 108.8 ± 10.9 | <0.05 |
| Subc. AT between L3-L4, g | 126.1 ± 9.5 | 139.5 ± 8.0 | 0.02 | 230.7 ± 16.7 | 254.2 ± 18.7 | 0.059 |
| Android/gynoid ratio | 0.90 ± 0.04 | 0.88 ± 0.04 | 0.73 | 1.20 ± 0.05 | 1.22 ± 0.05 | 0.18 |
| Total lean mass, kg | 42.3 ± 1.7 | 42.6 ± 1.7 | 0.69 | 51.1 ± 2.8 | 51.6 ± 2.8 | 0.36 |
| Mean BP, mmHgbc | 81.1 ± 2.1 | 79.4 ± 1.4 | 0.17 | 89.5 ± 4.1 | 89 ± 4.4 | 0.63 |
| Fast. glucose, mg/dL | 75.4 ± 1.8 | 75.7 ± 1.9 | 0.89 | 80.2 ± 2.1 | 80.4 ± 2.9 | 0.93 |
| HOMA-IRb | 0.7 ± 0.1 | 1.1 ± 0.1 | <0.01 | 1.2 ± 0.2 | 1.3 ± 0.2 | 0.91 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 178.2 ± 6.8 | <0.05 | 206.9 ± 8.0 | 202.3 ± 8.4 | 0.51 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 103.6 ± 5.4 | 0.06 | 130.6 ± 6.8 | 128.2 ± 6.7 | 0.71 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 55.8 ± 2.0 | 0.54 | 44.9 ± 2.7 | 43.8 ± 2.3 | 0.24 |
| Triglycerides, mg/dLb | 76.5 ± 5.9 | 77.5 ± 5.2 | 0.97 | 133 ± 19.0 | 122.2 ± 10.7 | 0.73 |
| Lean | Overweight/Obese | |||||
|---|---|---|---|---|---|---|
| Baseline | Intervention | P | Baseline | Intervention | P | |
| BMI, kg/m2 | 21.7 ± 0.4 | 21.8 ± 0.4 | 0.09 | 30.2 ± 1.0 | 30.6 ± 1.0 | 0.05 |
| Total fat mass, kg | 27.6 ± 1.0 | 27.5 ± 1.0 | 0.76 | 36.9 ± 2.2 | 37.4 ± 2.1 | <0.05 |
| L3-L4 visceral AT, gb | 34.4 ± 3.3 | 31.1 ± 3.4 | <0.05 | 121.2 ± 12.5 | 108.8 ± 10.9 | <0.05 |
| Subc. AT between L3-L4, g | 126.1 ± 9.5 | 139.5 ± 8.0 | 0.02 | 230.7 ± 16.7 | 254.2 ± 18.7 | 0.059 |
| Android/gynoid ratio | 0.90 ± 0.04 | 0.88 ± 0.04 | 0.73 | 1.20 ± 0.05 | 1.22 ± 0.05 | 0.18 |
| Total lean mass, kg | 42.3 ± 1.7 | 42.6 ± 1.7 | 0.69 | 51.1 ± 2.8 | 51.6 ± 2.8 | 0.36 |
| Mean BP, mmHgbc | 81.1 ± 2.1 | 79.4 ± 1.4 | 0.17 | 89.5 ± 4.1 | 89 ± 4.4 | 0.63 |
| Fast. glucose, mg/dL | 75.4 ± 1.8 | 75.7 ± 1.9 | 0.89 | 80.2 ± 2.1 | 80.4 ± 2.9 | 0.93 |
| HOMA-IRb | 0.7 ± 0.1 | 1.1 ± 0.1 | <0.01 | 1.2 ± 0.2 | 1.3 ± 0.2 | 0.91 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 178.2 ± 6.8 | <0.05 | 206.9 ± 8.0 | 202.3 ± 8.4 | 0.51 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 103.6 ± 5.4 | 0.06 | 130.6 ± 6.8 | 128.2 ± 6.7 | 0.71 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 55.8 ± 2.0 | 0.54 | 44.9 ± 2.7 | 43.8 ± 2.3 | 0.24 |
| Triglycerides, mg/dLb | 76.5 ± 5.9 | 77.5 ± 5.2 | 0.97 | 133 ± 19.0 | 122.2 ± 10.7 | 0.73 |
Abbreviations: AT, adipose tissue; BAT, brown adipose tissue; BMI, body mass index; L3-L4, lumbar vertebrae; SUV, standardized uptake value
aPaired t test
bnonparametric test
c n = 30
Baseline and Intervention-Induced Changes in Clinical and Biochemical Parameters
| Lean | Overweight/Obese | |||||
|---|---|---|---|---|---|---|
| Baseline | Intervention | P | Baseline | Intervention | P | |
| BMI, kg/m2 | 21.7 ± 0.4 | 21.8 ± 0.4 | 0.09 | 30.2 ± 1.0 | 30.6 ± 1.0 | 0.05 |
| Total fat mass, kg | 27.6 ± 1.0 | 27.5 ± 1.0 | 0.76 | 36.9 ± 2.2 | 37.4 ± 2.1 | <0.05 |
| L3-L4 visceral AT, gb | 34.4 ± 3.3 | 31.1 ± 3.4 | <0.05 | 121.2 ± 12.5 | 108.8 ± 10.9 | <0.05 |
| Subc. AT between L3-L4, g | 126.1 ± 9.5 | 139.5 ± 8.0 | 0.02 | 230.7 ± 16.7 | 254.2 ± 18.7 | 0.059 |
| Android/gynoid ratio | 0.90 ± 0.04 | 0.88 ± 0.04 | 0.73 | 1.20 ± 0.05 | 1.22 ± 0.05 | 0.18 |
| Total lean mass, kg | 42.3 ± 1.7 | 42.6 ± 1.7 | 0.69 | 51.1 ± 2.8 | 51.6 ± 2.8 | 0.36 |
| Mean BP, mmHgbc | 81.1 ± 2.1 | 79.4 ± 1.4 | 0.17 | 89.5 ± 4.1 | 89 ± 4.4 | 0.63 |
| Fast. glucose, mg/dL | 75.4 ± 1.8 | 75.7 ± 1.9 | 0.89 | 80.2 ± 2.1 | 80.4 ± 2.9 | 0.93 |
| HOMA-IRb | 0.7 ± 0.1 | 1.1 ± 0.1 | <0.01 | 1.2 ± 0.2 | 1.3 ± 0.2 | 0.91 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 178.2 ± 6.8 | <0.05 | 206.9 ± 8.0 | 202.3 ± 8.4 | 0.51 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 103.6 ± 5.4 | 0.06 | 130.6 ± 6.8 | 128.2 ± 6.7 | 0.71 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 55.8 ± 2.0 | 0.54 | 44.9 ± 2.7 | 43.8 ± 2.3 | 0.24 |
| Triglycerides, mg/dLb | 76.5 ± 5.9 | 77.5 ± 5.2 | 0.97 | 133 ± 19.0 | 122.2 ± 10.7 | 0.73 |
| Lean | Overweight/Obese | |||||
|---|---|---|---|---|---|---|
| Baseline | Intervention | P | Baseline | Intervention | P | |
| BMI, kg/m2 | 21.7 ± 0.4 | 21.8 ± 0.4 | 0.09 | 30.2 ± 1.0 | 30.6 ± 1.0 | 0.05 |
| Total fat mass, kg | 27.6 ± 1.0 | 27.5 ± 1.0 | 0.76 | 36.9 ± 2.2 | 37.4 ± 2.1 | <0.05 |
| L3-L4 visceral AT, gb | 34.4 ± 3.3 | 31.1 ± 3.4 | <0.05 | 121.2 ± 12.5 | 108.8 ± 10.9 | <0.05 |
| Subc. AT between L3-L4, g | 126.1 ± 9.5 | 139.5 ± 8.0 | 0.02 | 230.7 ± 16.7 | 254.2 ± 18.7 | 0.059 |
| Android/gynoid ratio | 0.90 ± 0.04 | 0.88 ± 0.04 | 0.73 | 1.20 ± 0.05 | 1.22 ± 0.05 | 0.18 |
| Total lean mass, kg | 42.3 ± 1.7 | 42.6 ± 1.7 | 0.69 | 51.1 ± 2.8 | 51.6 ± 2.8 | 0.36 |
| Mean BP, mmHgbc | 81.1 ± 2.1 | 79.4 ± 1.4 | 0.17 | 89.5 ± 4.1 | 89 ± 4.4 | 0.63 |
| Fast. glucose, mg/dL | 75.4 ± 1.8 | 75.7 ± 1.9 | 0.89 | 80.2 ± 2.1 | 80.4 ± 2.9 | 0.93 |
| HOMA-IRb | 0.7 ± 0.1 | 1.1 ± 0.1 | <0.01 | 1.2 ± 0.2 | 1.3 ± 0.2 | 0.91 |
| Total cholesterol, mg/dL | 184 ± 5.7 | 178.2 ± 6.8 | <0.05 | 206.9 ± 8.0 | 202.3 ± 8.4 | 0.51 |
| LDL cholesterol, mg/dL | 108.2 ± 4.6 | 103.6 ± 5.4 | 0.06 | 130.6 ± 6.8 | 128.2 ± 6.7 | 0.71 |
| HDL cholesterol, mg/dL | 56.5 ± 2.0 | 55.8 ± 2.0 | 0.54 | 44.9 ± 2.7 | 43.8 ± 2.3 | 0.24 |
| Triglycerides, mg/dLb | 76.5 ± 5.9 | 77.5 ± 5.2 | 0.97 | 133 ± 19.0 | 122.2 ± 10.7 | 0.73 |
Abbreviations: AT, adipose tissue; BAT, brown adipose tissue; BMI, body mass index; L3-L4, lumbar vertebrae; SUV, standardized uptake value
aPaired t test
bnonparametric test
c n = 30
Dietary intervention with olive oil increased brown adipose tissue activity in lean but not overweight/obese subjects
In static PET imaging acquisitions, the intervention resulted in increased BAT activity in lean but not volunteers with excessive weight (Fig. 3A, 3J, 3K). There were no changes in BAT volume (Fig. 3B), BAT maximal 18F-FDG uptake (Fig. 3D), BAT triglyceride content (Fig. 3E), WAT triglyceride content (Fig. 3G), trapezius SUV (Fig. 3H), and trapezius triglyceride content (Fig. 3I). A nonsignificant increase in BAT metabolic activity (Fig. 3C) was detected in lean, but not overweight/obese volunteers. In addition, there was a reduction in WAT activity in lean, but not obese subjects (Fig. 3F). Cerebellum and liver glucose uptake (mean SUV) decreased after the intervention in lean volunteers only (Suppl. Table 3) (26). In a subset of volunteers, we performed dynamic PET acquisition for the evaluation of the rate of glucose uptake in BAT, as depicted in Fig. 3L and 3M. There was a trend for increased BAT glucose uptake rate in lean but not in overweight/obese subjects after the intervention period (P = 0.08). There were no differences in glucose uptake rate in subcutaneous WAT and trapezius (Suppl. Fig. 3) (26). In order to identify factors potentially involved in the effect of olive oil dietary intervention to induce BAT activity, we performed multiple linear regression analyses with selected factors previously identified as involved in BAT activation (30-33). Leptin presented the most relevant coefficients; however, this was restricted to lean volunteers (Suppl. Table 4) (26). To further determine the putative correlation between leptin changes and BAT activity, we first categorized leptin levels according to baseline concentrations; categorization was made separately for men and women because of the known sexual dimorphism regarding blood leptin levels (34) (Suppl. Fig. 4) (26). Subjects with low leptin levels presented a significant increase in BAT activity after the intervention (Fig. 3N). In addition, a trend of increased BAT metabolic activity (SUV mean*volume) was detected (P = 0.11; Fig. 3O). Next, we performed a multiple regression analysis stratified by leptin category at the beginning of the study; we observed a positive and significant correlation between changes in leptin concentrations and changes in BAT SUV that was maintained after adjustments for age, sex, and BMI (Table 3). Similar findings were detected when employing changes in BAT volume (Suppl. Table 5) (26). In addition, we evaluated blood concentrations of other markers related to BAT activity. The intervention resulted in increased 12,13-diHOME in lean, but not obese (Fig. 3P; n = 20 volunteers), reduced CXCL14 in overweight/obese, but not lean (Fig. 3S), and increased FGF21 (Fig. 3Q) and secretin (Fig. 3R) in both lean and overweight/obese. The summary of the main findings is depicted in Fig. 4.
Multiple Linear Regression Analysis for the Association Between Delta Mean Standardized Uptake Value and Delta of Selected Variables According to Leptin Category at Baseline.
| β | Model 1 CI | P | β | Model 2 CI | P | |
|---|---|---|---|---|---|---|
| Low-leptin | ||||||
| Visceral adipose tissue | −0.063 | −0.006; 0.005 | 0.79 | −0.067 | −0.007; 0.005 | 0.78 |
| Insulin | −0.035 | −0.223; 0.193 | 0.88 | −0.150 | −0.285; 0.156 | 0.54 |
| Leptin | 0.493 | 0.002; 0.022 | 0.02 | 0.504 | 0.001; 0.023 | 0.03 |
| Secretin | −0.067 | −0.007; 0.006 | 0.77 | −0.049 | −0.007; 0.006 | 0.84 |
| FGF21 | −0.216 | −0.020; 0.008 | 0.36 | −0.101 | −0.019; 0.013 | 0.70 |
| High-leptin | ||||||
| Visceral adipose tissue | −0.328 | −0.005; 0.001 | 0.16 | −0.072 | −0.001; 0.002 | 0.75 |
| Insulin | −0.202 | −0.096; 0.039 | 0.39 | −0.036 | −0.074; 0.064 | 0.88 |
| Leptin | −0.328 | −0.005; 0.001 | 0.16 | −0.269 | −0.004; 0.001 | 0.22 |
| Secretin | 0.061 | −0.009; 0.011 | 0.97 | −0.134 | −0.013; 0.007 | 0.61 |
| FGF21 | −0.255 | 0.006; 0.002 | 0.28 | −0.249 | −0.006; 0.008 | 0.25 |
| β | Model 1 CI | P | β | Model 2 CI | P | |
|---|---|---|---|---|---|---|
| Low-leptin | ||||||
| Visceral adipose tissue | −0.063 | −0.006; 0.005 | 0.79 | −0.067 | −0.007; 0.005 | 0.78 |
| Insulin | −0.035 | −0.223; 0.193 | 0.88 | −0.150 | −0.285; 0.156 | 0.54 |
| Leptin | 0.493 | 0.002; 0.022 | 0.02 | 0.504 | 0.001; 0.023 | 0.03 |
| Secretin | −0.067 | −0.007; 0.006 | 0.77 | −0.049 | −0.007; 0.006 | 0.84 |
| FGF21 | −0.216 | −0.020; 0.008 | 0.36 | −0.101 | −0.019; 0.013 | 0.70 |
| High-leptin | ||||||
| Visceral adipose tissue | −0.328 | −0.005; 0.001 | 0.16 | −0.072 | −0.001; 0.002 | 0.75 |
| Insulin | −0.202 | −0.096; 0.039 | 0.39 | −0.036 | −0.074; 0.064 | 0.88 |
| Leptin | −0.328 | −0.005; 0.001 | 0.16 | −0.269 | −0.004; 0.001 | 0.22 |
| Secretin | 0.061 | −0.009; 0.011 | 0.97 | −0.134 | −0.013; 0.007 | 0.61 |
| FGF21 | −0.255 | 0.006; 0.002 | 0.28 | −0.249 | −0.006; 0.008 | 0.25 |
Multivariate analyses were performed using age, sex, and BMI as confounding variables based on literature.
Abbreviations: CI, confidence interval; FGF21, fibroblast growth factor 21.
aModel 1: crude
b Model 2: adjusted for age, sex, and BMI
Multiple Linear Regression Analysis for the Association Between Delta Mean Standardized Uptake Value and Delta of Selected Variables According to Leptin Category at Baseline.
| β | Model 1 CI | P | β | Model 2 CI | P | |
|---|---|---|---|---|---|---|
| Low-leptin | ||||||
| Visceral adipose tissue | −0.063 | −0.006; 0.005 | 0.79 | −0.067 | −0.007; 0.005 | 0.78 |
| Insulin | −0.035 | −0.223; 0.193 | 0.88 | −0.150 | −0.285; 0.156 | 0.54 |
| Leptin | 0.493 | 0.002; 0.022 | 0.02 | 0.504 | 0.001; 0.023 | 0.03 |
| Secretin | −0.067 | −0.007; 0.006 | 0.77 | −0.049 | −0.007; 0.006 | 0.84 |
| FGF21 | −0.216 | −0.020; 0.008 | 0.36 | −0.101 | −0.019; 0.013 | 0.70 |
| High-leptin | ||||||
| Visceral adipose tissue | −0.328 | −0.005; 0.001 | 0.16 | −0.072 | −0.001; 0.002 | 0.75 |
| Insulin | −0.202 | −0.096; 0.039 | 0.39 | −0.036 | −0.074; 0.064 | 0.88 |
| Leptin | −0.328 | −0.005; 0.001 | 0.16 | −0.269 | −0.004; 0.001 | 0.22 |
| Secretin | 0.061 | −0.009; 0.011 | 0.97 | −0.134 | −0.013; 0.007 | 0.61 |
| FGF21 | −0.255 | 0.006; 0.002 | 0.28 | −0.249 | −0.006; 0.008 | 0.25 |
| β | Model 1 CI | P | β | Model 2 CI | P | |
|---|---|---|---|---|---|---|
| Low-leptin | ||||||
| Visceral adipose tissue | −0.063 | −0.006; 0.005 | 0.79 | −0.067 | −0.007; 0.005 | 0.78 |
| Insulin | −0.035 | −0.223; 0.193 | 0.88 | −0.150 | −0.285; 0.156 | 0.54 |
| Leptin | 0.493 | 0.002; 0.022 | 0.02 | 0.504 | 0.001; 0.023 | 0.03 |
| Secretin | −0.067 | −0.007; 0.006 | 0.77 | −0.049 | −0.007; 0.006 | 0.84 |
| FGF21 | −0.216 | −0.020; 0.008 | 0.36 | −0.101 | −0.019; 0.013 | 0.70 |
| High-leptin | ||||||
| Visceral adipose tissue | −0.328 | −0.005; 0.001 | 0.16 | −0.072 | −0.001; 0.002 | 0.75 |
| Insulin | −0.202 | −0.096; 0.039 | 0.39 | −0.036 | −0.074; 0.064 | 0.88 |
| Leptin | −0.328 | −0.005; 0.001 | 0.16 | −0.269 | −0.004; 0.001 | 0.22 |
| Secretin | 0.061 | −0.009; 0.011 | 0.97 | −0.134 | −0.013; 0.007 | 0.61 |
| FGF21 | −0.255 | 0.006; 0.002 | 0.28 | −0.249 | −0.006; 0.008 | 0.25 |
Multivariate analyses were performed using age, sex, and BMI as confounding variables based on literature.
Abbreviations: CI, confidence interval; FGF21, fibroblast growth factor 21.
aModel 1: crude
b Model 2: adjusted for age, sex, and BMI
BAT activity and determination of BAT-related blood markers. Mean values of brown adipose tissue (BAT) activity (mean SUV) (A), BAT volume (B), BAT metabolic activity (C), BAT maximal SUV (D) and BAT content of triglycerides (E) at baseline and at the end of the intervention in lean and overweight/obese volunteers. White adipose tissue activity (mean SUV) (F) and content of triglycerides (G), trapezius activity (mean SUV) (H), and content of triglycerides (I) in both groups at baseline and at the end of the intervention. PET/MRI images of 2 lean volunteers (J) and 2 overweight/obese volunteers (K) at baseline and at the end of the intervention. Glucose uptake rate in lean (L) and overweight/obese (M) volunteers at baseline and at the end of the intervention. Figure (N) shows BAT activity (SUV mean) and (O) shows BAT metabolic activity at baseline and at the end of intervention according to the leptin category at the beginning of the study. Mean blood concentrations of BAT markers 12,13-diHOME (P), FGF21 (Q), secretin (R), and CXCL14 (S) at baseline and at the end of the intervention in lean and overweight/obese volunteers. *P < 0.05. Abbreviation: SUV, standardized uptake value.
Graphical abstract of the main findings of the study. Schematic representation of the effects of 4 weeks of olive oil consumption in brown adipose tissue activation and BAT-related blood markers in lean and overweight/obese subjects.
Thermogenic markers regulated by dietary intervention with olive oil are enriched in cells belonging to the thermogenic spectrum irrespectively of brown or white adipose tissue origin
In order to determine the cell specificity of thermogenic markers undergoing regulation by the dietary intervention with olive oil, we employed a dimensionality reduction method to analyze single-cell RNA sequencing data from 2 public datasets (35, 36). First, analyzing WAT data (Suppl. Fig. 5) (26), we obtained similar clustering as described in the original study (35) (Suppl. Fig. 5A). The thermogenic clusters were identified by the expressions of Ucp1 (Suppl. Fig. 5B) and adiponectin (Suppl. Fig. 5C) transcripts. We also evaluated the expression of Slc27a1 that encodes a protein involved in the transport of long-chain fatty acids into cells and is enriched in adipocytes with thermogenic activity (Suppl. Fig. 5D). Next, we evaluated transcripts related to the markers we measured in the subjects submitted to olive oil intervention: Lep that encodes for leptin (Suppl. Fig. 5E); Ephx2 that encodes a protein involved in the metabolism of 12,13-diHOME (Suppl. Fig. 5F); Cxcl14 (Suppl. Fig. 5G) and the transcripts encoding for its receptors, Cxcr4 (Suppl. Fig. 5H) and Ackr2 (Suppl. Fig. 5I); secretin receptor, Sctr (Suppl. Fig. 5J); and, receptors for FGF21, Fgfr1 (Suppl. Fig. 5K) and Egfr (Suppl. Fig. 5L). Lep (Suppl. Fig. 5E), Ephx2 (Suppl. Fig. 5F), and Sctr (Suppl. Fig. 5J) were the transcripts with the most relevant expression in the thermogenic cluster. Next, we combined data from WAT and BAT and performed another round of analysis (Suppl. Fig. 6) (26). Supplementary Figure 6A depicts clustering, and 6B depicts the origin of the cells. Thermogenic clusters were defined by the expressions of Ucp1 (Suppl. Fig. 6C), adiponectin (Suppl. Fig. 6D), and Slc27a1 (Suppl. Fig. 6E). In addition to the transcripts related to leptin, 12,13-diHOME, CXCL14, secretin, and FGF21 (Suppl. Fig. 6G-6O) as described in Fig. 4, here, we also analyzed Cyp2e1 (Suppl. Fig. 6F), a transcript encoding another protein involved in the metabolism of 12,13-diHOME. As for cells obtained from WAT, in the BAT/WAT combined analysis, the transcripts related to leptin (Suppl. Fig. 6L), 12,13-diHOME, Cyp2e1 (Suppl. Fig. 6F), and Ephx2 (Suppl. Fig. 6G); and, the transcript related to secretin, Sctr (Suppl. Fig. 6K) were the ones with preferential expression in the thermogenic cluster.
Discussion
In this study, we provide the first evidence for the effect of dietary olive oil increasing BAT activity in humans. We showed that a short-term dietary intervention with olive oil leads to increased BAT activity in lean volunteers, that this is accompanied by increased blood levels of leptin, secretin, and FGF21, and that low leptin concentration at baseline is predictive of greater BAT activity after intervention. This result was obtained by the evaluation of a considerably large number of volunteers who were analyzed by PET/MRI scans at baseline and at the conclusion of the intervention. In addition, a subsample of volunteers was also submitted to dynamic PET scans, providing further strength to our findings. Dynamic PET scans are known to provide a more accurate quantification of BAT activity, as determined by the rate of glucose uptake (14, 37).
Nutrients have been previously implied as potential modulators of BAT activity. However, only a few studies have looked into this question in depth (38). The acute ingestion of catechin or caffeine promoted increased metabolic rate in lean individuals with detectable BAT (19); in addition, the consumption of capsaicinoids for 6 weeks led to an increase in cold-induced thermogenesis and a reduction of body mass in lean volunteers (12, 18). There are no previous records of studies that evaluated the effects of unsaturated fatty acids on the activity of the human BAT. Nevertheless, some studies have assessed the association between unsaturated fats and BAT activity in rodents (39-42). As a whole, they showed that consumption of olive oil and its phenolic compounds resulted in increased interscapular UCP-1 expression (39, 40), increased BAT catecholamine concentration (40), greater BAT oxygen consumption, and greater BAT heat production (41), thus providing experimental evidence to support the hypothesis of our work.
An encouraging aspect of clinical interventions with olive oil is that, different from other dietary interventions that rely on the restriction of macronutrients and/or calories, it is easily incorporated as a lifestyle change by most people (43). In our sample, compliance was high, as demonstrated by a substantial increase in MUFA intake, which resulted in an approximately 9% increase in plasma MUFA levels. Moreover, the effectiveness of the intervention was validated by the reproduction of previously described actions of MUFA, such as reduction of total cholesterol and reduction of systemic inflammatory markers (44, 45). As previously reported (46), we detected reductions in visceral fat in both groups which was accompanied by an increase in the subcutaneous fat. Considering the favorable metabolic profile of subcutaneous as compared with visceral fat (47), it is possible that changes in the distribution of adipose tissue could explain, at least in part the metabolic benefits of the intervention. The reduction in the FDG uptake in the liver could also be an effect of MUFA. There are studies showing that hepatic steatosis is associated with increased liver FDG uptake (48, 49) and that olive oil consumption is associated with reduced liver steatosis (50, 51).
Despite similar compliance among lean and overweight/obese subjects, it was only in the lean that the olive oil promoted increased BAT activity. This could be explained by the fact that, as a rule, obese people have very low, or even absent BAT activity as measured by FDG uptake (7, 25, 52); and also, because recent data indicate that they possess less metabolically active BAT in the supraclavicular region (53), although it is worthwhile mentioning that this is a controversial concept (54). In addition, the intake of olive oil per body weight was lower in the obese group both before and after the intervention, which may have mitigated the benefits of the oil even considering the fact that the increase in the consumption was similar in both groups. In fact, it has been shown that the benefits of olive oil increase with the amount of consumption (55); moreover, it has been shown that in obese people BAT is hyporesponsive to interventions that are efficient to induce its activity in otherwise lean subjects, such as cold exposure and insulin (37). The hyporesponsiveness of BAT is related to body adiposity, as shown by the reversibility of its activity following body mass reduction (56, 57). A recent experimental study proposed the existence of a hypothalamus-adrenal-medulla-BAT axis in which leptin acts promoting BAT activation only if its baseline levels are low, as in leanness (33). Moreover, experimental studies have provided evidence for the central effects of leptin to induce BAT activity, which can be blunted by hypothalamic leptin resistance (58-61). This is in concert with our data as a baseline and intervention-induced shift in blood leptin levels were associated with increased BAT activity in response to olive oil. In addition to BAT activity, we also detected nonsignificant increases in BAT volume and metabolic activity in lean volunteers, which reached borderline significance when stratified by baseline leptin levels. Increased blood insulin was yet another outcome of increased consumption of olive oil. The activation of GPR40, which is responsive to oleate, is a well-known mechanism that increases pancreatic β-cell capacity to secrete insulin (62). This mechanism is impaired in obesity (63), which could explain why only the lean volunteers presented a shift in insulin levels during the intervention. Together with leptin, FGF21, and secretin, the increased insulin levels in the lean volunteers could provide an additional stimulus to BAT activity, as previously demonstrated (64).
We also detected increases in markers previously associated with browning: secretin (30, 65), FGF21 (66), and 12,13-diHOME (31). Secretin and FGF21 levels increased by 10% and 30%, respectively, in almost 85% of the sample, indicating an important role for olive oil in stimulating both of these browning markers. However, the association was lost when looking specifically into the overweight/obese population, suggesting the association is lost by excessive adiposity and high leptin concentrations at baseline. In the case of FGF21, this could be explained by the fact that obese and insulin-resistant subjects can present resistance to the actions of FGF21, which is regulated by leptin (67, 68). Regarding secretin, its implication in the regulation of BAT was recently described (30, 65) and the impact of adiposity on its capacity to stimulate BAT is still unknown; thus, we cannot rule out the possibility that both secretin and FGF21, together with leptin and insulin, have a synergistic role in increasing BAT activity in lean but not in overweight/obese individuals (30, 65, 66). In concert with these findings, recent studies have suggested a role for the gut peptides secretin and cholecystokinin, stimulating the secretion of glucagon-like peptide-1, stimulating insulin secretion, and activating the postprandial BAT thermogenesis in an AMPK-SIRT1-PGC-1α–dependent pathway (38). In addition, it has been shown that MUFAs can allosterically activate SIRT1 and enhance PGC-1α/PPARα signaling, leading to increased oxidative metabolism (69). Another recently described browning marker, 12,13-diHOME, was also increased in response to olive oil; however, its increase occurred only in the lean volunteers. Recent data have indicated that 12,13-diHOME can act as a stimulator of BAT activity by increasing fatty acid uptake into brown adipocytes (31), and in our study, this could have been enhanced by MUFA. The only BAT marker not modified by olive oil intervention was the chemokine CXCL14. It has been reported that CXCL14 acts as a regulatory factor that is secreted by BAT in response to thermogenic activation (32). In our study, the overall low BAT activity at baseline could have dampened CXCL14 response or, otherwise, it could simply not be responsive to MUFA.
In the final part of the study, we used a single-cell RNA sequencing analytical system to ask whether the factors identified as responsive to olive oil intervention could potentially interact with cells presenting the brown and beige adipocyte transcriptional signatures. We found that, in addition to leptin, Ephx2, the transcript encoding a protein involved in the metabolism of 12,13-diHOME, and the receptor for secretin, were particularly enriched in brown and beige adipocytes. Some of these cells also expressed the receptor for FGF21. Thus, we provided a cellular basis for the putative role of these substances in the activation of brown and beige adipocytes. These specific questions were never asked before using single-cell RNA sequencing data.
Very few studies have investigated the effects of dietary components in human BAT and even fewer studies have compared lean and overweight/obese subjects. The present study was performed in a relatively large sample (41 volunteers); however, the smaller number of overweight/obese as compared with lean subjects and the lack of a placebo group represent weaknesses that could have impacted in the interpretation of results. Additionally, we acknowledge that indirect calorimetry analysis could have been performed adding interesting data to evaluate changes in energy expenditure.
Conclusion
The increased consumption of olive oil induces the activation of BAT in lean but not overweight/obese subjects. This is related to baseline leptin levels and to changes in leptin during the intervention. This is also related to changes in blood levels of secretin and FGF21, 2 endogenous substances previously known as inducers of BAT activity. Thus, activation of BAT is yet another health-promoting effect of MUFA-rich olive oil and should be further explored in combined approaches to treat obesity and metabolic associated conditions.
Abbreviations
- 12,13-diHOME
12,13-dihydroxy-9Z-octadecenoic acid
- 18F-FDG
18F-fluorodeoxyglucose
- BAT
brown adipose tissue
- BMI
body mass index
- CXCL14
C-X-C motif chemokine 14
- FGF21
fibroblast growth factor 21
- HOMA-IR
homeostasis model assessment of insulin resistance
- MCP-1
monocyte chemoattractant protein 1
- MRI
magnetic resonance imaging
- MUFA
monounsaturated fatty acids
- PET
positron emission tomography
- PUFA
polyunsaturated fatty acids
- SUV
standardized uptake value
- VOI
volume of interest
- WAT
white adipose tissue
Acknowledgments
The authors thank all the volunteers of the study. We thank the Nuclear Medicine Division, Laboratory of Emergency Medicine, the Dietary Assessment Research Group and Laboratory of Population Studies of the University of São Paulo. We also thank Joseane Morari, Giulianna Silva, Gerson Ferraz, Erika Anne Roman, Kalle Koskensalo, Prince Dadson, and Bruna Bombassaro for the technical support and contribution to the manuscript. The Laboratory of Cell Signaling belongs to the National Institute of Science and Technology – Neuroimmunomodulation.
Financial Support: This study was financially supported by grants from São Paulo Research Foundation. Grant numbers: #2013/07607-8; #2016/10616–7; #2018/05479-6; #2019/02055-3.
Author Contributions: M.M-P., G.A.N., J.A-F., D.S.O., J.C.L.J., M.S., D.E.C., H.P.S., S.R.G.F., and M.T.S. performed data collection. M.M-P, M.U.D, G.A.N, J.A-F, D.S.O., M.T.S., K.A.V., and L.A.V. performed data analysis and interpretation of results. M.M-P, M.U.D., M.T.S., K.A.V., and L.A.V. designed and developed the study. M.M.P. and L.A.V. drafted the manuscript and L.A.V. supervised the study. All authors edited and approved the final manuscript.
Clinical Trial Information: ClinicalTrials.gov registration no. NCT03024359.
Additional Information
Disclosure Summary: The authors have nothing to disclose.
Data Availability
All datasets generated and analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
References
