Colonic Fermentation and Acetate Production in Youth with and without Obesity

ABSTRACT

Background

In the last few years, there has been a growing interest in the role of gut microbiota in the development of obesity and its complications.

Objectives

In this study, we tested the following hypotheses: 1) lean youth and youth with obesity experience a different capability of their gut microbiota to ferment carbohydrates and produce acetate; and 2) colonic acetate may serve as a substrate for hepatic de novo lipogenesis (DNL).

Methods

Nineteen lean youth [mean ± SE BMI (in kg/m²): 21.8 ± 0.521] and 19 youth with obesity (BMI: 35.7 ± 1.66), ages 15–21 y, frequency-matched by age and sex, underwent a fasting 10-h sodium [d₃]-acetate intravenous infusion to determine the rate of appearance of acetate (Ra_acet) into the peripheral circulation before and after an oral dose of 20 g of lactulose. Pre- and post-lactulose Ra_acet values were determined at a quasi-steady state and changes between groups were compared using a quantile regression model. Acetate-derived hepatic DNL was measured in 11 subjects (6 youth with obesity) and its association with Ra_acet was assessed using Spearman correlation.

Results

Mean ± SE Ra_acet was not different before lactulose ingestion between the 2 groups (7.69 ± 1.02 μmol · kg⁻¹ · min⁻¹ in lean youth and 7.40 ± 1.73 μmol · kg⁻¹ · min⁻¹ in youth with obesity, P = 0.343). The increase in mean ± SE Ra_acet after lactulose ingestion was greater in lean youth than in youth with obesity (14.7 ± 2.33 μmol · kg⁻¹ · min⁻¹ and 9.29 ± 1.44 μmol · kg⁻¹ · min⁻¹, respectively, P = 0.001). DNL correlated with Ra_acet, calculated as changes from the pre- to the post-lactulose steady state (ρ = 0.621; P = 0.046).

Conclusions

These data suggest that youth with obesity ferment lactulose to a lesser degree than youth without obesity and that colonic acetate serves as a substrate for hepatic DNL.

This trial was registered at clinicaltrials.gov as NCT03454828.

Introduction

In the last few years there has been a growing interest in the role of gut microbiota in the development of obesity and its complications (1, 2). In vitro and in vivo studies suggest that gut microbiota are not inert symbionts, but active components of the host metabolism. Some studies propose that the interaction between gut microbiota and host metabolism may occur through the production of SCFAs (3–6), which are products of the bacterial fermentation of undigested carbohydrates. Acetate (or acetic acid) is produced in much higher amounts during the colonic fermentation of undigested carbohydrate than other SCFAs (7). So far, human and animal studies have provided contrasting results on the association between acetate and adiposity. In fact, whereas some studies showed that individuals with obesity tend to have greater concentrations of acetate in stools and plasma (7–10), other studies have shown that colonic acetate might actually favor weight loss (11). An elegant mechanistic study in mice by Perry et al. (7) has suggested that high colonic acetate production might promote weight gain by increasing ghrelin secretion and therefore decreasing the sense of satiety. Conversely, more recent data in lean and obese adults suggest that during the fasting state, lean individuals show greater acetate production than those with obesity (12). Given the inconclusive results and the lack of data concerning the role of colonic fermentation in the development of obesity in pediatrics, in the present study we aimed to assess whether colonic acetate production is different between lean youth and youth with obesity and explored whether colonic fermentation may modulate lipid and glucose metabolism.

Methods

Subjects and screening

Forty-three youth in good general health between 15 and 21 y of age were enrolled in the study (NCT03454828) after meeting eligibility criteria. Youth with obesity and lean youth had a BMI above the 95th percentile and between the 25th and 85th percentiles, respectively, for age, sex, and height. Per the study protocol, exclusion criteria included use of medication on a chronic basis, baseline blood creatinine >1.0 mg/dL, pregnancy, presence of endocrinopathies or significant chronic illness, or use of antibiotics within the past 6 mo. Subjects were not excluded based on dietary habits; 2 subjects (1 male/1 female) in the lean group were vegetarian and all other subjects had otherwise typical, unrestricted diets. Obese subjects were recruited from the Yale Pediatric Obesity Clinic (New Haven, CT) and lean subjects were recruited from the Yale Research Trial Volunteer Database and by study fliers posted throughout New Haven County. Five subjects were excluded from the final analyses, because 4 subjects underwent different protocols used to fine-tune the experiment, and 1 subject was unable to complete the study. Of the 38 subjects included in the final analyses (mean ± SE age: 17.2 ± 0.31 y), 19 (11 females) were adolescents with obesity and 19 (9 females) were adolescents without obesity; the 2 groups were frequency-matched by age and sex. Supplemental Figure 1 shows a CONSORT diagram outlining participant recruitment. Table 1 shows the characteristics of the study group. The study was approved by the Yale University Human Investigations Committee in accordance with the Helsinki Declaration of 1975 as revised in 1983. Written consent from adults and parents and written assent from minors were obtained for all participants after full explanation of the study.

TABLE 1.

Clinical characteristics of the study population obtained at the first study visit¹

Characteristic	Lean youth (n = 19)	Youth with obesity (n = 19)
Age, y	17.5 ± 0.47	16.9 ± 0.38
Sex (girls/boys), n	9/10	11/8
Race (NHW/NHB/H/O), n	17/0/1/1	6/8/5/0
BMI, kg/m2	21.8 ± 0.521	35.7 ± 1.66
Body fat, %	17.8 ± 1.39	40.6 ± 2.32
Waist circumference, cm	72.6 ± 3.06	96.2 ± 5.11
Fasting plasma glucose, mg/dL	86.8 ± 1.53	92.4 ± 2.70
Fasting plasma insulin, μM/mL	11.3 ± 0.23	21.7 ± 0.70
Fasting plasma C-peptide, ng/mL	1.94 ± 0.16	2.77 ± 0.24
Fasting plasma FFAs,2 mM	25.2 ± 2.88	19.4 ± 2.85
HbA1c, %	5.02 ± 0.05	5.46 ± 0.10

¹Values are means ± SE unless otherwise indicated. FFA, free fatty acid; H, Hispanic; HbA1c, glycated hemoglobin; NHB, non-Hispanic black; NHW, non-Hispanic white; O, other ethnicities/races.

²FFAs in lean youth were available for 18 patients.

Experimental design and study procedures

Volunteer subjects were admitted at the Hospital Research Unit of Yale New Haven Hospital at 07:00 after a 12-h overnight fast and were to remain fasting throughout the 10-h infusion study. Our principal goal was to determine the rate of appearance of acetate (Ra_acet) into the peripheral blood after the ingestion of lactulose, a putative indicator of colonic production. In fact, lactulose is an indigestible carbohydrate that is fermented in the intestine by gut microbiota and converted into SCFAs. Among the SCFAs produced by the gut microbiota, acetate is the most abundant (7). This experiment was developed and modified based on the experiments done by Ferchaud-Roucher et al. (13) and by Pouteau et al. (14). In particular, we used lower acetate doses because, as determined from our pilot studies, the doses used by Ferchaud-Roucher et al. and Pouteau et al. resulted in very high enrichments, reaching ∼80% in some cases.

Starting at 08:00, we began a 10-h, primed, continuous, peripheral venous infusion of 99% sodium [d₃]-acetate (Cambridge Isotope Laboratories, Inc.) with a priming dose of 200 μg · kg⁻¹ · min⁻¹ for 4 min and a continuous infusion rate of 59.5 μg · kg⁻¹ · min⁻¹ for 10 h. After 180 min from the start of the infusion, subjects received 20 g of lactulose per os, dissolved in 30 mL of water. Lactulose, a disaccharide of fructose and galactose, is not digested by humans and quantitatively reaches the colon where it is fermented (15). To assess Ra_acet, blood samples were collected every 30 min until 480 min, and then at 540 and 600 min during the final 2 h of the study in order to measure deuterated enrichment of acetate in blood plasma. The primary outcome of the study was the Δ of acetate turnover between the basal steady state and postlactulose steady state. The basal steady state Ra_acet was obtained by averaging the points between 120 and 180 min, with the quasi-steady state determined by inspection. Similarly, the postlactulose steady state Ra_acet was calculated by averaging the points between 420 and 600 min at the quasi-steady state. We will refer to the prelactulose phase as the time between 0 and 180 min and the postlactulose phase as the time between 180 and 600 min.

In addition, secondary outcomes included the measurement of breath hydrogen (H₂) and methane (CH₄) and plasma glucose, insulin, C-peptide, and free fatty acids (FFAs). Breath samples were collected hourly in order to measure H₂ and CH₄ production from colonic fermentation. Plasma samples were collected hourly during the infusion in order to measure plasma concentrations of glucose, FFAs, insulin, and C-peptide.

Measurement of the isotopic enrichment of acetate

Deuterium enrichment of acetate was determined by modification of a previously described procedure for measurement of plasma concentration of SCFAs (16, 17). Plasma (100 μL) was acidified with 20 μL 1 M HCl followed by addition of ∼0.5 g NaCl, and derivatized as the amide of 2,4-difluoroaniline by addition of 0.2 mL 1,3-dicyclohexylcarbodiimide (0.2 M in toluene) and 0.2 mL 2,4-difluoroaniline (0.2 M in hexane). After 1 h (room temperature), 2 mL NaHCO₃ (1 M) was added, and the SCFA amides of 2,4-difluoroanaline were extracted (2×) into ethyl acetate, dried under nitrogen gas, and redissolved in 50 μL ethyl acetate for GC-MS analysis. Analyses were performed on an Agilent gas chromatograph (HP6890)–mass spectrometer (HP5973) with a 12-m HP-1 column. In the electron impact mode, we monitored ions m/z 171–175 (m0–m4) to determine plasma [d₃]-acetate enrichment.

Calculation of Ra_acet in peripheral blood

Ra_acet was calculated using the following equation:

(1)

where i is the infusion rate of the [d₃]-acetate tracer (μmol · kg⁻¹ · min⁻¹) and E_t is the plateau enrichment of acetate in peripheral blood. E_t is expressed as mole percentage excess (MPE)/100. With this method, we assessed the Ra_acet in the peripheral circulation. Acetate arising from colonic fermentation will undergo some uptake by the colon and the liver. Thus, there is a level of uncertainty regarding the reliability of Ra_acet to indicate formation (production) of acetate in the colon lumen. Our model has 2 underlying assumptions that determine the validity of using intravenous infusion of acetate tracer to assess relative colonic production of acetate: 1) colonic epithelial absorption of acetate formed in the colon lumen is not different between lean and obese subjects; 2) first-pass uptake of acetate by the liver via the portal vein is also not different between the lean and obese groups. Detailed comparisons of various animal models suggest that colonic mucosal uptake of acetate (as opposed to butyrate) is not likely an important pitfall of our first assumption (18). Using completely different methodology from ours, and also using assumptions about endogenous hepatic production of acetate, Fernandes et al. (19) concluded that the percentage of rectally infused acetate that reached the peripheral circulation and the rate of hepatic uptake of acetate were not different between 2 groups of volunteers who differed by BMI. Thus, we feel that our approach is valid for assessing relative differences in colonic production of acetate between the lean and obese subjects in our study.

Measurement of colonic production of H₂ and CH₄

Breath samples were collected at baseline (0 min) and hourly for the duration of the infusion. Patients were asked to exhale into a glass vial through a straw and then the vial was immediately closed. Collections were obtained in 10-mL Exetainer glass vials. Samples were analyzed on an Agilent 7890B Customized Gas Chromatograph with a PAL RSI 85 autosampler with OpenLab version 2.3 software (Agilent Technologies). The gas chromatograph was fitted with 2 columns, a 0.5-m × 2.5-cm o.d. × 2-mm i.d. HayeSep Q 80/100 mesh column and a ShinCarbon ST column 2 m × 2.5-cm o.d. × 2-mm i.d. Both columns used ultra-high-purity nitrogen carrier gas and were maintained at 140°C throughout the 5-min run. The gas chromatograph used flame ionization detector (FID) for methane analysis and a thermal conductivity detector for hydrogen and carbon dioxide analysis. Three calibration gases were used in each batch run, ranging from 10 to 150 parts per million (ppm) hydrogen, from 10 to 100 ppm methane, and from 1% to 6% CO₂. Two control gases were used with each batch run (30 and 90 ppm hydrogen, 20 and 50 ppm methane, and 2.5 and 5% CO₂). Breath H₂ and CH₄ concentrations were normalized to correct for variable contamination with room air by dividing the gas concentration by the respective carbon dioxide concentration and multiplying by 5 (alveolar air ∼5% carbon dioxide).

Bioelectrical impedance analysis

A TANITA digital scale was used to determine body fat percentage by bioelectrical impedance analysis. All patients were assessed upon admission to the research unit.

Assessment of hepatic de novo lipogenesis

The incorporation of deuterium into plasma VLDL during administration of deuterium-labeled acetate was used to determine the fractional synthetic rate of fatty acids by modification of mass isotopomer distribution analysis (MIDA) (20). After centrifugation at 75,000 RPM and 4°C for 45 min, plasma was split into 2 parts; 1 part was processed to determine [d₃]-acetate enrichment, and the other part was processed by TLC for plasma triglyceride extraction (21). Lipids from plasma (1 mL) were extracted by the method of Folch et al. (22). Extracted lipids were dissolved in ethanol:chloroform (1:2 vol:vol) and spotted on silica gel G thin-layer plates (Merck). The plates were developed with hexane:diethyl ether:acetic acid (80:20:1, by vol). Lipids were visualized using Fluorescein vapor against a standard and scraped off the silica gel plate. Deuterium enrichment in plasma acetate and palmitate from plasma triglycerides was measured by GC-MS with acetate analysis as aforementioned and palmitate as the methyl ester using methanolic BF3 (25 m HP-1 column, chloride ionization with isobutane, monitoring ions m/z 271–276: m0–m5). Fractional rates of de novo lipogenesis (DNL) were calculated from the [d₃]-acetate enrichment and the mass isotopomer distribution of palmitate. DNL was calculated to account for the potential to incorporate 7 d₂- and 1 [d₃]-labeled acetate moieties into palmitate from plasma [d₃]-acetate (20, 23).

Measurement of plasma glucose, insulin, C-peptide, and FFAs

Plasma concentrations of glucose were measured at the bedside using the YSI 2300 Stat Plus Glucose Analyzer (Yellow Springs Instruments). Plasma concentrations of insulin and C-peptide were measured by ELISA (Millipore Sigma) and plasma concentrations of FFA were measured by an enzymatic colorimetric method assay (Wako Chemicals Inc.). FFA plasma concentrations in the lean group were available in 18 patients.

Statistics

Data are reported as means ± SEs, except for Δ changes that are reported as medians and IQRs. Unadjusted differences between the groups were analyzed using the Mann–Whitney U test and differences within groups were analyzed using the Wilcoxon signed rank test. The AUC for breath H₂ and CH₄ concentration was calculated according to the trapezoidal rule. AUC was calculated in Excel (Microsoft Office 2016).

The prelactulose basal acetate turnover was calculated by averaging the points at the steady state between 120 and 180 min and is referred to as the basal steady state, whereas postlactulose acetate turnover was calculated by averaging the points at the steady state between 420 and 600 min and is referred to as the postlactulose steady state. We will refer to the prelactulose phase as the phase from 0 to 180 min and the postlactulose phase as the phase from 180 to 600 min. The main outcome of the study was Δ acetate turnover, which is the within-person change of Ra_acet between the basal steady state and the postlactulose steady state.

To compare the medians of Δ changes of a variable (including Δ acetate turnover between basal steady state and postlactulose steady state), we modeled the 50th percentile using a quantile regression approach (PROC QUANTREG), in which Δ changes were the dependent variable and group (lean youth compared with youth with obesity), race [non-Hispanic white (NHW), non-Hispanic black, Hispanic, and other, dummy coded into NHW (1/0)], and the basal value of the variable were used as covariates. We used race as a covariate, because, among the clinical characteristics shown in Table 1, the 2 study groups were different only in the distribution of race.

The correlation between DNL and acetate kinetics was assessed using a Spearman correlation (ρ). In order to control the overall Type 1 error rate, we chose 1 primary outcome, and the rest were considered as secondary. Therefore, the primary hypothesis test was conducted using the 2-sided α level of 0.05, and the same false positive rate of 0.05 was assumed for the other outcomes, but care should be taken in interpreting the findings for the nonprimary outcomes, because these should be viewed as corroborating the finding for the primary outcome.

The data analysis for this article was generated using SAS software, version 9.4 of the SAS System for Windows (SAS Institute Inc.), and GraphPad Prism8 software, version 9.0.0 for Windows (GraphPad).

Results

Assessment of acetate kinetics, colonic fermentation, FFAs, insulin, and C-peptide in the whole cohort

In the whole cohort, a quasi-steady state was achieved between 120 and 180 min (basal steady state) after the start of the infusion, with a mean plasma acetate enrichment of 12.3 ± 0.984 MPE (Supplemental Figure 2A). After lactulose ingestion at 180 min, a quasi-steady state was achieved between 420 and 600 min (postlactulose steady state), with a mean plasma acetate enrichment of 7.67 ± 0.623 MPE, implying a rise in Ra_acet after lactulose ingestion (Supplemental Figure 2A). The mean Ra_acet during the basal steady state and postlactulose steady state was 7.55 ± 0.989 μmol · kg⁻¹ · min⁻¹ and 12.1 ± 1.34 μmol · kg⁻¹ · min⁻¹, respectively, which was an ∼48% increase after lactulose ingestion.

In the whole cohort, the mean breath H₂ concentration during the prelactulose phase (0–180 min) was 17.6 ± 3.09 ppm, increasing soon after lactulose ingestion and peaking at 420 min (Supplemental Figure 2B). The mean breath CH₄ concentration during the prelactulose phase was 12.6 ± 3.46 ppm, which increased soon after lactulose ingestion and also peaked at 420 min (Supplemental Figure 2C). Plasma FFA concentrations declined at 120 min and reached a nadir at 300 min. After 300 min, there was a rebound in plasma FFA concentrations, with a peak at 600 min in the whole cohort (Supplemental Figure 3A). Plasma FFA concentrations were significantly greater at 540 and 600 min than at 180 min (P = 0.0002 and P < 0.0001, respectively) (Supplemental Figure 3A). In addition, glucose concentrations started declining at 240 min, and continued to decline until the end of the study (600 min) (Supplemental Figure 3B). Likewise, plasma insulin and C-peptide concentrations declined throughout the study in the whole cohort (Supplemental Figure 3C, D).

Differences in acetate kinetics between lean youth and youth with obesity

During the basal steady state, mean Ra_acet was similar between lean youth and youth with obesity (7.69 ± 1.02 μmol · kg⁻¹ · min⁻¹ and 7.40 ± 1.73 μmol · kg⁻¹ · min⁻¹, respectively, P = 0.343) (Supplemental Figure 4). The mean Ra_acet during the postlactulose steady state was 14.7 ± 2.33 μmol · kg⁻¹ · min⁻¹ in lean youth and 9.29 ± 1.44 μmol · kg⁻¹ · min⁻¹ in youth with obesity (P = 0.001) (Supplemental Figure 4).

Δ Ra_acet between the basal steady state and postlactulose steady state phases was 5.57 μmol · kg⁻¹ · min⁻¹ (2.82–10.3 μmol · kg⁻¹ · min⁻¹) in lean youth and 2.05 μmol · kg⁻¹ · min⁻¹ (1.10–3.04 μmol · kg⁻¹ · min⁻¹) in youth with obesity, which was independent from ethnicity and basal acetate turnover (P = 0.018) (Figure 1A).

FIGURE 1.

Outcomes from colonic bacterial fermentation during a 600-min [d₃]-acetate infusion in lean youth and youth with obesity. (A) Δ Ra_acet, (B) breath H₂ concentration, and (C) breath CH₄ concentration. The ↓ symbol shows the timing of lactulose ingestion. (A) Changes in Ra_acet were compared using a quantile regression model and are shown as medians and IQRs. (B, C) Error bars represent SEMs. n = 38 (lean: n = 19; obese: n = 19). ppm, parts per million; Ra_acet, rate of appearance of acetate.

H₂ and CH₄ production in lean youth and youth with obesity

During the prelactulose phase, lean youth showed a higher production of H₂ (22.6 ± 5.22 ppm) than youth with obesity (10.5 ± 2.40 ppm) (P = 0.048) (Figure 1B). The AUC for breath H₂ concentration during the postlactulose phase was not different between lean youth and youth with obesity (P = 0.198) (Figure 1B). Supplemental Figure 5A shows normalized breath H₂ concentrations.

The mean breath CH₄ concentration during the prelactulose phase was 9.89 ± 2.85 ppm in lean youth and 10.8 ± 4.83 ppm in youth with obesity (P = 0.279) (Figure 1C). During the postlactulose phase, the mean breath CH₄ concentration was 13.8 ± 4.50 ppm in lean youth and 13.1 ± 6.25 ppm in youth with obesity (P = 0.050) (Figure 1C). Not all subjects were CH₄ producers, defined by CH₄ production <3 ppm at all time points during the study, and the prevalence of non-CH₄ producers was similar between the 2 groups (26% of lean youth and 21% of youth with obesity) (P = 0.701). Supplemental Figure 5B shows normalized breath CH₄ concentrations.

Changes in plasma FFA concentrations in lean youth and youth with obesity

The changes in plasma FFA concentrations from baseline (0 min) to the nadir (300 min) were −1.97 μM (−0.377 to −0.095 μM) in lean youth and −0.157 μM (−0.214 to −0.007 μM) in youth with obesity (P = 0.130) (Figure 2A). The rebound, calculated as Δ from nadir (300 min) to peak (600 min), was greater in lean youth than in youth with obesity, independently of ethnicity and plasma FFA concentration at 300 min (P = 0.049) (Figure 2B).

FIGURE 2.

Plasma FFA concentrations during a 600-min [d₃]-acetate infusion in lean youth and youth with obesity. (A) Individual changes in FFAs from baseline to nadir, (B) individual changes in FFAs from nadir to peak. The ↓ symbol shows the timing of lactulose ingestion. (A, B) Changes were compared using a quantile regression model and data are shown as medians and IQRs. n = 37 (lean: n = 18; obese: n = 19). FFA, free fatty acid.

Association between DNL and acetate kinetics

The mean DNL at 180 min was 2.53% ± 0.364% in lean youth and 1.96% ± 0.159% in youth with obesity (P = 0.255). Between 540 and 600 min, mean DNL was 3.38% ± 0.620% in lean youth and 1.97% ± 0.192% in youth with obesity (P = 0.139). Δ DNL was 1.10 (0.250–1.20) in lean youth and 0.00 (−0.500 to 0.600) in youth with obesity (P = 0.085). There was a positive correlation between Δ DNL and Δ Ra_acet (Spearman: ρ = 0.621; P = 0.046) (Figure 3).

FIGURE 3.

Correlation between Δ DNL and Δ Ra_acet in 11 subjects (5 lean youth and 6 youth with obesity). R and P values were obtained from a Spearman correlation, n = 11. DNL, de novo lipogenesis; Ra_acet, rate of appearance of acetate.

Plasma glucose, insulin, and C-peptide concentrations in lean youth and youth with obesity

At the start of the study (0 min), youth with obesity showed higher plasma glucose concentrations than lean youth (P = 0.001) (Table 1). During the prelactulose phase, plasma glucose concentrations declined slightly in youth with obesity, whereas lean youth showed a steadier glucose concentration. The Δ plasma glucose (from 0 to 180 min) during the prelactulose phase was 1 mg/dL (−4 to 3 mg/dL) in lean youth and −3 mg/dL (−8 to 1 mg/dL) in youth with obesity (P = 0.047) (Supplemental Figure 6A). In both groups, plasma glucose concentrations were lower at 600 min than at 0 min (P = 0.00046 in lean youth and P = 0.00024 in youth with obesity) (Supplemental Figure 6B), but the Δs between 0 and 600 min were similar between the groups after adjustment for race and basal plasma glucose concentrations (P = 0.765). The degree of reduction in plasma glucose concentrations during the postlactulose phase was similar between the 2 groups, with a Δ of −8 mg/dL (−12 to −4 mg/dL) in lean youth and −7 mg/dL (−14 to −4 mg/dL) in youth with obesity (P = 0.999) (Supplemental Figure 6A). However, there was a steeper decline in plasma glucose concentrations during the postlactulose phase than during the prelactulose phase in both lean youth and youth with obesity (P = 0.0002 and P = 0.032, respectively) (Supplemental Figure 6A).

At the start of the study (0 min), youth with obesity showed higher plasma insulin concentrations than lean youth (Table 1). The mean plasma insulin concentration during the prelactulose phase was 11.3 ± 0.877 μU/mL in lean youth and 21.7 ± 0.318 μU/mL in youth with obesity (P = 0.028) (Supplemental Figure 6C). The Δs during the prelactulose phase were −2.16 μU/mL (−3.24 to 1.33 μU/mL) in lean youth and −3.18 μU/mL (−8.37 to −0.700 μU/mL) in youth with obesity (P = 0.530) (Supplemental Figure 6C). There was an overall reduction in plasma insulin concentrations from 0 to 600 min in lean youth and in youth with obesity (P = 0.003 and P = 0.0007, respectively), but the Δs between 0 and 600 min were similar between the 2 groups (P = 0.176) (Supplemental Figure 6D). The Δs of plasma insulin concentrations during the postlactulose phase were −0.800 μU/mL (−5.61 to 0.335 μU/mL) in lean youth and −3.37 μU/mL (−6.65 to −1.49 μU/mL) in youth with obesity (P = 0.640).

The mean plasma C-peptide concentration during the prelactulose phase was 1.75 ± 0.150 ng/mL in lean youth and 2.60 ± 0.248 ng/mL in youth with obesity (P = 0.011) (Supplemental Figure 6E). The Δs during the prelactulose phase were −0.370 ng/mL (−0.790 to 0.025 ng/mL) in lean youth and −0.495 ng/mL (−0.810 to 0.005 ng/mL) in youth with obesity (P = 0.190) (Supplemental Figure 6E). Plasma C-peptide concentrations declined after the start of the study, with an overall reduction from 0 to 600 min in lean youth and in youth with obesity (P = 0.0003 and P = 0.0004, respectively) (Supplemental Figure 6E), but the Δs between 0 and 600 min were similar between the 2 groups (P = 0.341) (Supplemental Figure 6F). The Δs of plasma C-peptide concentrations during the postlactulose phase were −0.420 ng/mL (−0.565 to −0.060 ng/mL) in lean youth and −0.490 ng/mL (−0.665 to −0.345 ng/mL) in youth with obesity (P = 0.748) (Supplemental Figure 6E).

Discussion

In this study we show, to our knowledge for the first time, that 1) the rate of appearance of plasma acetate consequent to the colonic fermentation of lactulose is higher in lean youth than in youth with obesity and 2) changes in the colonic fermentation of lactulose are directly correlated to changes in DNL, suggesting that colonic fermentation represents an important metabolic pathway leading to adiposity accumulation through the modulation of DNL. The association between colonic fermentation and DNL was suggested initially by studies in animals. In fact, Turnbaugh et al. (24) hypothesized that ob/ob mice salvage more energy from the colon than their lean counterparts. This conclusion was based, in part, on the fact that cecal concentrations of acetate were higher in the obese mice than in the lean, and that energy concentration in cecal fluid was lower in the obese mice (25). So far, though, no evidence had been provided in humans that colonic acetate synthesis might be linked to DNL. This observation is important because it suggests that colonic fermentation can contribute to the development of obesity by modulating DNL to some extent. The differences in Ra_acet observed between lean youth and youth with obesity may suggest that such a mechanism may be more relevant in lean youth. The lower Ra_acet after colonic fermentation in youth with obesity, probably due to changes in gut microbiota composition (26), may be a protective mechanism aimed at limiting excessive energy retrieval from the colon. Conversely, the boost derived from colonic fermentation metabolites to DNL might not be sufficient to lead to massive adiposity accumulation, because increased lipogenesis might be counterbalanced by an increased lipid oxidation.

Moreover, it cannot be excluded that the difference in acetate production may have consequences for hunger and satiety. In fact, the microbiota may also play a role in regulating brain function, perhaps via signals from the colon, including SCFAs (25). Therefore, the mechanism for these effects may not simply relate to colonic salvage of energy derived from undigested carbohydrate.

During the study, we also observed a decline in plasma FFA and glucose concentrations that was independent from the effect of insulin, because plasma insulin and C-peptide concentrations were also declining during the study. Although the difference in Δ FFA from baseline to the nadir between lean youth and youth with obesity was not statistically significant, there was a trend toward a greater suppression in lean than in obese youth. The effect of colonic fermentation on adipose tissue lipolysis has been known for some time. In particular, it has been shown that lactulose ingestion and colonic administration of SCFAs in adults affects adipose tissue lipolysis (13, 27). Ferchaud-Roucher et al. (13) studied acetate turnover in 8 overweight adults by infusing ¹³C-acetate and inducing colonic fermentation with an oral dose of 30 g of lactulose. The investigators showed that the colonic fermentation of lactulose causes a rapid decline of adipose tissue lipolysis (measured using glycerol turnover), with a nadir ∼90 min after lactulose ingestion (13). Moreover, similarly to what we observed, there was a rebound of FFAs after the nadir with a progressive increase in adipose tissue lipolysis (13). In vitro studies have shown that the effect of colonic fermentation on adipose tissue lipolysis is mediated mostly by acetate through the free fatty acid receptors (FFARs) (27). In the present study, the decline of plasma FFA tended to be more pronounced in lean youth than in youth with obesity, and the increase of FFA concentrations after the nadir was higher in lean youth than in youth with obesity. The marked increase of plasma FFA concentrations after the nadir could be due to the prolonged fasting, because the subjects were required to be fasting for 12 h before the start of the study, and throughout the 10-h study as well.

A decline in plasma glucose, insulin, and C-peptide concentrations was observed in the whole group of subjects. This could be due to prolonged fasting; however, the steeper decline in plasma glucose concentrations after the ingestion of lactulose suggests that there may be an effect of the colonic production of acetate on carbohydrate metabolism. However, it has to be noted that Steudle et al. (28) compared the effect of different lactulose doses on glucose concentrations in 12 adults and reported no changes within the 3 h after lactulose ingestion.

This study has several strengths, such as the detailed assessment of colonic acetate synthesis and DNL measured by using stable isotopes, the length of the study, and the novelty of the findings in this age group. Despite these strengths, some limitations need to be acknowledged. It would have been beneficial to assess fecal microbiota composition and metabolomics and to relate these results to the data shown. In addition, other limitations include the lack of a longitudinal assessment of adiposity, the lack of more sophisticated measures of lipolysis through the use of deuterated tracers, and the small sample size. However, it must be noted that despite the small sample size, our data on the differences of rate of appearance of plasma acetate between lean youth and youth with obesity are comparable with what was observed by Fernandes et al. (19). They showed that with the administration of acetate through the rectum, subjects with hyperinsulinemia (who also manifested a higher BMI than lean subjects) exhibited lower colonic acetate production than their leaner counterparts.

In conclusion, these data suggest that the colonic production of acetate is different between lean youth and youth with obesity as a consequence of the colonic fermentation of lactulose, and that this difference may have an influence on adipose tissue lipolysis and hepatic DNL.

Notes

Supported by National Institute of Diabetes and Digestive and Kidney Diseases (NIH) grants R01-DK114504 (to NS) and P30DK045735. This publication was also made possible by Clinical and Translational Science Award grant UL1 TR001863 from the National Center for Advancing Translational Science (NCATS), a component of the NIH. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the NIH.

Author disclosures: The authors report no conflicts of interest.

Supplemental Figures 1–6 and Supplemental Table 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn.

Abbreviations used: AUC, area under the curve; bf%, body fat percent; CH₄, methane; DNL, de novo lipogenesis; FFA, free fatty acid; FID, flame ionization detector; GC, gas chromatograph; H, Hispanic; H₂, hydrogen; MIDA, mass isotopomer distribution analysis; MPE, moles per cent excess; NHB, non-Hispanic Black; NHW, non-Hispanic White; O, other ethnicity; ppm, parts per million; Ra_acet, rate of acetate appearance; SCFA, short chain fatty acid.

Data Availability

The unidentified raw data used for this article are available as supplemental material (Supplemental Table 1).