Exercise Training Independent of Intensity Lowers Plasma Bile Acids in Prediabetes

Abstract

Introduction:

People with obesity have high circulating bile acids (BA). Although aerobic fitness favors low circulating BAs, the effect of training intensity prior to clinically meaningful weight loss on BA is unclear. We tested the hypothesis that 2-wks of interval (INT) versus continuous (CONT) exercise would lower plasma BAs in relation to insulin sensitivity.

Methods:

Twenty-three older adults with prediabetes (ADA criteria) were randomized to 12 work-matched bouts of INT (n=11; 60.3±2.4y; 32.1±1.2kg/m²) at 3min at 50% HRpeak and 3min at 90% HRpeak or CONT (n=12; 60.8±2.4y; 34.0±1.7kg/m²) at 70% HRpeak cycling training for 60min/d over 2 wks. A 180min 75-g OGTT was performed to assess glucose tolerance (tAUC), insulin sensitivity (Siis) and metabolic flexibility (RER_{post-prandial} – RER_fast; indirect calorimetry). BA (n=8 conjugated and 7 unconjugated) were analyzed at 0, 30, and 60min of the OGTT. Anthropometrics and fitness (VO₂peak) were also assessed.

Results:

INT and CONT comparably reduced BMI (P<0.001) and fasting RER (P<0.001), but raised insulin sensitivity (P=0.03). INT increased VO₂peak as compared to CONT (P=0.01). Exercise decreased the unconjugated BAs CDCA iAUC_60min (P<0.001), DCA iAUC_60min (P<0.001), LCA iAUC_60min (P<0.001), and GCDCA iAUC_60min (P<0.001). Comparable reductions were also seen in the conjugated BAs HDCA iAUC_60min (P=0.01) and TLCA iAUC_60min (P=0.007). Increased VO₂peak associated with lowered UDCA_0min (r=-0.56, P=0.02) and CA iAUC_60min (r=-0.60, P=0.005), while reduced BMI related to higher GDCA_0min (r=0.60, P=0.005) and GCDCA_0min (r=0.53, P=0.01). Improved insulin sensitivity correlated with lower GCDCA iAUC_60min (r=-0.45, P=0.03) and GDCA iAUC_60min (r=-0.48, P=0.02), while increased metabolic flexibility related to DCA iAUC_60min (r=0.64, P=0.004) and GCDCA iAUC_60min (r=0.43, P=0.05).

Conclusions:

Short-term training lowers some BA in relation to insulin sensitivity independent of intensity.

INTRODUCTION

Bile acids are traditionally considered digestive steroid-based molecules in the turnover of cholesterol and lipid digestion (1, 2). Hepatocytes synthesize primary bile acids, namely cholic acid (CA) and chenodeoxycholic acid (CDCA), which can be subsequently conjugated with amino acids (e.g. glycine or taurine) in the liver. In response to a meal, these primary conjugated bile acids are then secreted into the gut and undergo bacterial metabolism. This is physiologically relevant for the dehydroxylation and deconjugation of amino acids that result in secondary bile acids (e.g. lithocholic acid (LCA) and deoxycholic acid (DCA)) that are unconjugated. As part of the enterohepatic circulation, some of these secondary bile acids are reabsorbed to return to the liver via portal circulation and reconjugated (with glycine or taurine) to be secreted as conjugated secondary bile acids (1, 3, 4). In recent years though bile acids have gained recognition for acting as nutrient signaling molecules that impact glucose homeostasis, inflammation, gut hormones and insulin secretion/sensitivity (5). Not surprisingly, alterations in bile acid metabolism have been implicated in obesity, type 2 diabetes (T2D), and non-alcoholic fatty liver disease (NAFLD) (6–8). In fact, people with obesity and/or T2D often have higher fasting as well as post-prandial levels of circulating bile acids than healthy controls in relation to insulin resistance (9–11). Thus, identifying treatments that can effectively lower bile acids may contribute to lower T2D risk.

Aerobic training is established to reduce insulin resistance in people at risk for T2D (12–14). In fact, we demonstrated in previous work that short term aerobic training independent of intensity reduces whole-body and adipose insulin resistance (15) and improves β-cell function in relation to glucose tolerance (16) in older adults with prediabetes. In addition, we and others have shown aerobic training exerts hepatoprotective effects (17–19). However, the role of exercise training on plasma bile acids in humans is not fully understood. Healthy women with low fitness level (VO_2peak ≤ 35ml/kg/min) had higher circulating bile acids following an OGTT compared with those with high aerobic capacity (VO_2peak ≥ 45ml/kg/min) (20). In line with this, a 14-wk combined exercise (60–75% VO₂peak) and diet intervention reduced bile acid levels in women with insulin resistance in relation to increased fitness and weight loss (21). However, both a single bout of high volume resistance exercise and moderate intensity (70% VO_2peak) endurance among healthy, trained men reduced bile acids (22). This suggests that short-term exercise can reduce circulating bile acids without weight loss. However, to date, no study has examined whether exercise intensity modifies bile acids, nor has any study determined the role of bile acids after exercise training in people with prediabetes. To fill this knowledge gap, we determined the effect of interval (INT) versus continuous (CONT) training prior to clinically meaningful weight loss on plasma bile acids in both conjugated and unconjugated forms among older adults with prediabetes. We hypothesized that INT would lower bile acid levels compared with CONT training, and the change in bile acids would correlate with gains in VO₂peak, insulin sensitivity and fat oxidation.

METHODS

Participants

Twenty-three older adults with obesity (INT: n=11 (9F); 60.3±2.4y; 32.1±1.2kg/m²; CONT: n=12 (10F); 60.8±2.4y; 34.0±1.7kg/m²) were recruited via newspaper flyers and/or advertisements from the Charlottesville, VA community. Participants were screened for prediabetes according to the American Diabetes Association criteria. Impaired fasting glucose (IFG: 100–125mg/dL) and/or impaired glucose tolerance (IGT: 2-h plasma glucose 140–200 mg/dL) was depicted using a 75g oral glucose tolerance test (OGTT). Participants were sedentary (≤60 minutes/week of structured exercise), free of chronic disease (e.g. cancer, renal, cardiovascular, or any metabolic disease), non-smoking, weight-stable (≤2 kg change) and not using medication affecting weight, insulin sensitivity (e.g. metformin, GLP-1 agonists) or vascular function (e.g. Ca⁺⁺ channel blockers, α-blockers, ꞵ-blockers, nitrates, etc.). Participants further went through a physical exam along with clinical biochemistry assays and resting/exercise electrocardiogram to confirm eligibility and ensure participant safety. Study protocols conformed to Declaration of Helsinki standards. Study was approved by the Institutional Review Board (IRB # 17822) and all individuals provided written and verbal consent prior to participation.

Cardiorespiratory Fitness and Body Weight

Peak oxygen consumption (VO₂peak) was determined using a continuous incremental exercise test on a cycle ergometer with indirect calorimetry (CareFusion, V_max CART, Yorba Linda, CA, USA) as previously described by our laboratory (23). Following a 4h minimum fast, body weight was assessed on a digital scale measured to the nearest 0.01 kg with participants wearing minimal clothing and without shoes. Height was measured with a stadiometer to the nearest 0.10 cm and body mass index (BMI) was then calculated. Waist circumference (WC) was measured 2 cm above the umbilicus using a tape measure to the nearest 0.10 cm twice and averaged.

Metabolic Control

Participants were instructed to refrain from alcohol and caffeine consumption, along with engagement in strenuous physical activity 48-h prior to the study visits. Participants were also instructed to avoid taking any medications and/or dietary supplements 24-h ahead of testing. Additionally, participants were asked to record their dietary intake and consume ~250g carbohydrates on the day before testing to minimize the effect of diet on insulin secretion as well as duplicate the diet prior to post-intervention testing day. The final exercise training bout was performed 24-h before post-intervention metabolic testing.

Oral Glucose Tolerance Test (OGTT)

Following an overnight fast of approximately 10-h, participants reported to the Clinical Research Unit (CRU) where an intravenous line was placed in the antecubital vein. Circulating glucose and insulin were determined at 0, 30, 60, 90, 120 and 180 minutes during a 75-g OGTT to assess glucose tolerance and insulin action was calculated using the simiple index of insulin sensitivity (SIis) (24). Conjugated (n=8) and unconjugated (n=7) bile acids were collected at 0, 30 and 60 min. During the OGTT, respiratory energy ratio (RER) was also measured at 0, 60, 120 and 180 min to assess substrate oxidation and metabolic flexibility (average postprandial - fasting RER). To account for fasting bile acids, incremental area under the curve (iAUC) was assessed by the trapezoidal model.

Exercise Training

Participants were randomized into either a CONT or INT training that comprised of 12 supervised, work-matched 60 min sessions on a cycle ergometer over 13 days. A rest day was provided on about day 8 for recovery. The first two exercise sessions were performed at 30 and 45 minutes, respectively, to acclimate participants to exercise. CONT training comprised of exercise at 70% HRpeak while INT group alternated between 3 mins of 90% HRpeak and 50% HRpeak for the 60 minute session. Heart rate (Polar Electro, Woodbury, NY) and rating of perceived exertion was recorded throughout each session to confirm exercise intensity.

Biochemical Analysis

Plasma glucose was analyzed immediately after sample collection using the YSI 2300 StatPlus Glucose Analyzer system (Yellow Springs, OH, USA). Blood samples were collected in chilled 3ml EDTA vacutainers with aprotinin centrifuged at 4◦C for 10mins at 3000rpm, aliquoted and stored at −80°C until further analysis. Plasma insulin was analyzed using the enzyme-link immunosorbent assay (ELISA) kits (Millipore, Billerica, MA). All conjugated (TCA, TCDCA, GCA, GDCA, HDCA, TLCA, GUDCA, TUDCA) and unconjugated (CA, CDCA, LCA, DCA, UDCA, TDCA, GCDCA) bile acids were measured from human serum samples using ultra performance Liquid Chromatography – tandem Mass Spectrometry (LC-MS/MS) (Biomarkers Core Lab, IICTR, Columbia University Irvine Medical Center). Bile acids were extracted by spiking human serum samples with deuterated internal standards and mixing with chilled acetonitrile for protein precipitation. After incubation at 4◦C for 15min, the mixture was centriguged and the organic layer was transfereed to a LCMS vial and evaporated under nitrogen stream. The extracted bile acids were resuspended in methanol for LCMS analysis. LC-MS/MS analysis was performed using a Waters TQS Xevo triple quadrupole mass spectrometer equipped with an electrospray ionization source and integrated with a Waters Acquity UHPLC system (Milford, MA). Ten microliters of the sample were injected onto a Phenomenex Kinetex C18 column (50×2.1mm, 1.7u, 100A) maintained at 40◦C and at a flow rate of 0.250ml/min. The initial flow conditions were 40% Solvent A (water containing 5mM ammonium formate) and 60% Solvent B (Methanol containing 5mM ammonium formate). Solvent B was raised to 80% linearly over 8min, increased to 97% in 2 min and returned to initial flow conditions by 11.30 min with a total run time of 14 min. Quantitative measurements were done in selective ion monitoring (SIM) mode and negative electrospray ionization.

Statistical Analysis

Data were analyzed using SPSS Statistical Software (IBM SPSS Statistics Version 28.0). Non-normally distributed data were log-transformed for analysis. An independent two-tailed Student’s t-test was used to compare baseline group differences. A two-way repeated measures analysis of variance (ANOVA) was used to examine group × time interactions, and covaried with baseline differences when appropriate. Weight loss and VO₂peak (ml/kg/min) were also used as co-variates to assess group × time interaction if bile acid main effects were observed. Pearson’s correlation was used to examine associations. Data are reported as mean ± SEM, and significance was accepted as P ≤ 0.05.

RESULTS

Anthropometrics and Fitness

CONT and INT training reduced body weight (P<0.001), although INT promoted more weight loss than CONT (P=0.05; Table 1). Further, INT training increased VO₂peak compared to CONT (P=0.01; Table 1).

Table 1.

Effect of INT or CONT training on anthropometrics, aerobic fitness, and metabolic health.

	INT			CONT		ANOVA (P)
	Pre	Post	Pre		Post	T	G × T
Subjects (F)	11 (9)			12 (10)
Age (yr)	60.3 ± 2.4			60.8 ± 2.4
Anthropometrics
Weight (kg)	89.9 ± 3.1	88.9 ± 3.2	93.1 ± 4.0		93.0 ± 4.1	0.006	0.01
BMI (kg.m−2)	32.1 ± 1.2	31.9 ± 1.2	34.0 ± 1.7		33.9 ± 1.7	0.01	0.05
WC (cm)	105.6 ± 3.1	104.9 ± 3.2	104.8 ± 3.3		105.1 ± 3.3	0.47	0.51
Aerobic Fitness
VO2peak (L.min−1)	1.9 ± 0.1	2.0 ± 0.1	1.8 ± 0.2		1.7 ± 0.1	0.11	0.23
VO2peak (ml.kg−1.min −1)	20.6 ± 1.0	22.2 ± 0.8	18.0 ± 0.5		17.6 ± 0.8	0.22	0.01
Glucose (mg.dL −1 )
Fasting	103.1 ± 2.0	102.7 ± 2.6	104.1 ± 2.6		101.0 ± 3.1	0.30	0.40
120 min	140.5 ± 12.8	126.3 ± 12.8	147.4 ± 9.1		132.9 ± 9.3	0.03	0.98
180 min iAUC	4943.9 ± 995.2	5065.0 ± 1208.1	7160.2 ± 1364.6		5399.9 ± 1267.3	0.19	0.17
Insulin (μU.mL −1 )
Fasting	9.1 ± 1.7	11.5 ± 2.4	11.2 ± 1.5		11.3 ± 2.1	0.66	0.41
120 min	53.3 ± 7.7	50.6 ± 10.4	88.6 ± 9.8		82.8 ± 10.6	0.11	0.91
180 min iAUC	9819.5 ± 1101.4	9552.7 ± 1681.9	10482.4 ± 1099.2		8088.5 ± 859.4	0.01	0.68
SIis	0.2 ± 0.002	0.2 ± 0.003	0.19 ± 0.004		0.2 ± 0.003	0.03	0.50

Data are mean ± SEM. CONT, continuous exercise; INT, interval exercise; F, female; BMI, body mass index; WC, waist circumference; VO₂peak, peak oxygen consumption; tAUC, total area under the curve.; SIis, simple index of insulin sensitivity; T, treatment; G × T, G × T, group × treatment.

Glucose Metabolism

INT and CONT training did not influence fasting glucose levels but comparably lowered post-prandial glucose at 120min (P=0.03) and glucose tAUC_180min (P=0.04) (Table 1). Neither intervention affected fasting or 120min insulin levels, however, insulin tAUC_180min (P=0.03) was significantly reduced (Table 1). As such, insulin sensitivity increased following both CONT and INT, independent of intensity (P=0.03; Table 1). Training lowered fasting RER (P<0.001) and increased metabolic flexibility (P=0.008) independent of intensity (Table 2).

Table 2:

Effect of CONT or INT training on fasting and postprandial substrate oxidation.

	INT		CONT		ANOVA (P)
	Pre	Post	Pre	Post	T	G × T
RER
0 min	0.81 ± 0.01	0.77 ± 0.01	0.82 ± 0.02	0.80 ± 0.01	<0.001	0.48
Average PP	0.84 ± 0.02	0.84 ± 0.01	0.87 ± 0.01	0.85 ± 0.01	0.24	0.11
Metabolic flexibility	0.03 ± 0.01	0.06 ± 0.01	0.05 ± 0.01	0.07 ± 0.01	0.03	0.16

Data are mean ± SEM. CONT, continuous exercise; INT, interval exercise; Metabolic flexibility was calculated by subtracting fasting RER from average of post-prandial RER_60–180min.

PP, post-prandial; T, treatment; G × T, group × treatment.

Circulating Bile Acids

Training did not affect fasting bile acids except for GDCA, which was reduced with INT and elevated with CONT training (P=0.02; Table 3). However, among unconjugated bile acids (Figure 1), training reduced CDCA iAUC_60min (P<0.001), DCA iAUC_60min (P<0.001), LCA iAUC_60min (P<0.001), and GCDCA iAUC_60min (P<0.001). There were comparable reductions in the conjugated bile acids HDCA iAUC_60min (P=0.01) and TLCA iAUC_60min (P=0.007) independent of intensity (Figure 2). Collectively, co-varying for weight loss or fitness gains did not alter the results.

Table 3:

Effect of CONT or INT training on fasting unconjugated and conjugated bile acids.

	INT		CONT		ANOVA (P)
	Pre	Post	Pre	Post	T	G × T
Unconjugated Bile Acids (nM)
CA	1.43 ± 0.25	1.37 ± 0.20	1.62 ± 0.20	1.78 ± 0.24	0.76	0.51
CDCA	1.89 ± 0.19	1.48 ± 0.18	1.71 ± 0.20	1.77 ± 0.20	0.74	0.33
DCA	2.46 ± 0.01	2.36 ± 0.08	2.57 ± 0.12	2.61 ± 0.10	0.57	0.20
LCA	0.51 ± 0.09	0.56 ± 0.10	0.68 ± 0.12	0.63 ± 0.10	0.45	0.84
UDCA	1.18 ± 0.23	0.81 ± 0.16	0.99 ± 0.11	0.93 ± 0.14	0.61	0.59
TDCA^	0.93 ± 0.11	0.82 ± 0.10	1.38 ± 0.13	1.46 ± 0.17	0.65	0.12
GCDCA	2.18 ± 0.15	2.18 ± 0.12	2.40 ± 0.15	2.49 ± 0.14	0.53	0.54
Conjugated Bile Acids (nM)
TCA	0.70 ± 0.18	0.53 ± 0.14	0.92 ± 0.17	1.07 ± 0.18	0.58	0.12
TCDCA	1.11 ± 0.17	1.14 ± 0.13	1.41 ± 0.16	1.41 ± 0.16	0.77	0.65
GCA	1.73 ± .12	1.68 ± 0.10	1.91 ± 0.14	1.99 ± 0.16	0.88	0.39
GDCA^	1.77 ± 0.13	1.65 ± 0.15	2.17 ± 0.14	2.33 ± 0.13	0.11	0.02
HDCA	1.05 ± 0.12	0.98 ± 0.11	1.02 ± 0.12	0.99 ± 0.08	0.25	0.54
TLCA	0.25 ± 0.03	0.22 ± 0.03	0.30 ± 0.06	0.44 ± 0.07	0.39	0.09
GUDCA	1.23 ± 0.21	1.33 ± 0.14	1.50 ± 0.13	1.62 ± 0.12	0.14	0.75
TUDCA	0.30 ± 0.08	0.25 ± 0.04	0.48 ± 0.10	0.33 ± 0.06	0.22	0.46

Data are mean ± SEM. CONT, continuous exercise; INT, interval exercise; T, treatment; G × T, group × treatment.. CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; LCA, lithocholic acid; UDCA, ursodeoxycholic acid; TDCA, taurodeoxycholic acid; GCDCA, glycodeoxycholic acid; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; GCA, glycocholic acid; GDCA, glycochenodeoxycholic acid; HDCA, hyodeoxycholic acid; TLCA, taurolithocholic acid; GUDCA, glycoursodeoxycholic acid; TUDCA, tauroursodeoxycholic acid.

^{^}Denotes covarying with baseline.

Figure 1.

Effect of INT and CONT exercise training on unconjugated bile acids. Data are presented as means ± SEM. CONT, continuous exercise; INT, interval exercise; iAUC, incremental area under the curve; CA, cholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; LCA, lithocholic acid; UDCA, ursodeoxycholic acid; TDCA, taurodeoxycholic acid; GCDCA, glycodeoxycholic acid. *Significant treatment effect (P≤0.01).

Figure 2.

Effect of INT and CONT exercise training on conjugated bile acids. Data are presented as means ± SEM. CONT, continuous exercise; INT, interval exercise; iAUC, incremental area under the curve; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; GCA, glycocholic acid; GDCA, glycochenodeoxycholic acid; HDCA, hyodeoxycholic acid; TLCA, taurolithocholic acid; GUDCA, glycoursodeoxycholic acid; TUDCA, tauroursodeoxycholic acid. *Significant treatment effect (P≤0.01).

Correlations

Increased VO₂peak associated with lowered UDCA_0min (r=-0.56, P=0.02) and CA iAUC_60min (r=-0.60, P=0.005), while weight loss correlated with lower CA_0min (r=0.47, P=0.04), GDCA_0min (r=0.60, P=0.005), GCDCA_0min (r=0.53, P=0.01) and TCA_0min (r=0.58, P=0.04). Reduced fasting glucose correlated with lower CDCA iAUC_60min (r=-0.48, P=0.03) and trended with higher GDCA iAUC_60min (r=0.38, P=0.07). While reduced insulin at 120min were linked to reduced GDCA iAUC_60min (r=0.52, P=0.01), decreased insulin iAUC_180min was associated with higher GCDCA iAUC_60min (r=0.43, P=0.04) and TLCA iAUC_60min (r=0.44, P=0.04). As such, improved insulin sensitivity associated with reduced GCDCA iAUC_60min (r=-0.45, P=0.03) and GDCA iAUC_60min (r=-0.48, P=0.02; Figure 3). Training enhanced metabolic flexibility correlated with DCA iAUC_60min (r=0.64, P=0.004) and GCDCA iAUC_60min (r=0.43, P=0.05; Figure 4).

Figure 3.

Correlations of change (Δ) in insulin sensitivity (SIis) to Δ log GCDCA iAUC_60min and Δ log GDCA iAUC_60min. Grey circles – INT, black circles – CONT. GCDCA, glycodeoxycholic acid; GDCA, glycochenodeoxycholic acid

Figure 4.

Correlations of change (Δ) in metabolic flexibility to Δ log GCDCA iAUC_60min and Δ log DCA iAUC_60min. Grey circles – INT, black circles – CONT. GCDCA, glycodeoxycholic acid; DCA, deoxycholic acid.

DISCUSSION

Short-term exercise training studies in people with obesity have been reported to improve liver unsaturated fat composition (25), reduce circulating markers of hepatic apototsis (26), lower the hepatokine fetuin-A (27) as well as increase hepatic insulin extraction (16) in relation to insulin sensitivity prior to clinically meaningful weight loss. Herein, the central finding in this study adds to this work on liver health following exercise and shows that short-term aerobic training lowered plasma levels of some circulating unconjugated (CDCA, DCA, LCA and GCDCA) and conjugated (HDCA and TLCA) bile acids in older adults with prediabetes independent of exercise intensity. Interestingly, this observation occurred only the in post-prandial state, as there was overall no effect on fasting bile acids (except for GDCA). These observations may be of clinical relevance since lowering of post-prandial GCDCA, GDCA, and DCA correlated with improved insulin sensitivity and metabolic flexibility. In fact, previous work demonstrates that people with obesity and insulin resistance exhibit elevated circulating bile acid concentrations (10, 11, 28). This is consistent with others discussing the metabolic role of bile acids in glucose homeostasis and energy metabolism particularly in the postprandial state (4, 20, 21, 29). Further, it is not necessarily surprising that we noted significant reductions in CDCA, DCA, LCA and GCDCA as they contribute to the majority of circulating bile acids (3). Our work is mostly consistent with other acute and chronic exercise studies in rodents and humans. Indeed, rodents undergoing voluntary wheel running for 2 to 12 weeks increased fecal bile acids excretion and cholesterol turnover, thereby relating to lower circulating bile acids (30, 31). In sedentary women with insulin resistance and obesity undergoing a 14 week exercise and diet intervention reduced total plasma bile acid levels in the fasting state as mainly a result from conjugated bile acids (21). However, post-prandial unconjugated bile acids were elevated. This is somewhat consistent as well with work from the same group reporting that low-fit women have elevated absolute levels of circulating conjugated bile acids during the post-prandial state of an OGTT compared with high fit women (20). Regardless, recreationally active people undergoing a single middle-distance running protocol decreased serum values of CA, DCA, CDCA, UDCA, GUDCA and HDCA (32). In contrast, among young males, acute resistance or aerobic exercise raised LCA, while only resistance exercise lowered total serum bile acids (22). These later data suggest bile acids are modifiable following short-term exercise prior to clinically meaningful weight loss with potential for different exercise modalities to modify bile acid metabolism. Our data add to the existing literature that prior work did not examine exercise intensity nor examine older adults with prediabetes. Subsequently, we show exercise reduces CDCA, DCA, LCA, and GCDCA (unconjugated forms) as well as HDCA and TLCA (conjugated forms) independent of intensity.

There are several possible explanations for how exercise lowered bile acids. Obesity has been suggested to raise circulating bile acids (9–11, 33). Lowering body weight therefore would be a reasonable mechanism contributing to reduced bile acids. Interestingly, INT reduced weight by ~ 1 kg more than CONT exercise. The relevance of this amount of weight loss with INT is unclear though as it correlated with lower fasting bile acid levels and did not promote greater circulatory changes in bile acids relative to CONT exercise. Nonetheless, while the subtle weight loss after exercise is consistent with bariatric surgery work reporting decreased total fasting bile acid levels in people with prediabetes (34), we noted that post-prandial, not fasting, bile acids were reduced with exercise training. This suggests another explanation is likely playing a role. Young adult women with healthy BMI but differing levels of fitness varied in bile acid levels in response to an OGTT (20). Those with VO₂max >45 ml/kg/min depicted a gradual increase in bile acids, particularly LCA, along with lower fecal concentrations, while women with VO₂max <35 ml/kg/min consistently showed high levels of both circulating and fecal bile acids. Although we report that elevations in VO₂peak were associated with reduced CA iAUC_60min and UDCA_0min, it is worth noting that 14 weeks of exercise and diet decreased fasting bile acids in women with obesity, despite covarying for changes in fitness (21). Herein, we noted that INT improved fitness more than CONT exercise, yet this did not translate into differential post-prandial bile acid results and covarying for fitness did not alter the results. Together, this highlights that fitness per se may in part, but not solely, explain reductions in bile acids. We did not design the study to determine the mechanism by which exercise changes biles acids, but prior work had shown that 7α-hydroxy-4-cholesten-3-one (C4), a surrogate of bile acids synthesis, was not statistically different between low and high fit women (20). However, the enteroendocrine hormone cholecystokinin (CCK) that is secreted upon meal intake to stimulate gallbladder release of bile acids and slow gastric emptying was lower in the high fit women (20). Slower gastric emptying could influence gut transit time, thereby impacting reabsorption/recycling of bile acids or gut microbiome breakdown to unconjugated forms. Conjugated bile acids facilitate their reuptake in the small intestine and provide feedback regulation via enterohepatic circulation. Although the reduction in HDCA and TLCA in our study during the OGTT after training may reflect to some extent lower bile acid reabsorption, we identified no change in fasting bile acids. This is noteworthy given that during the fasting state bile acids are not needed to be reabsorbed/synthesized for digestion. Thus, exercise may modulate bile acids by intestinal cells and/or gut microbiome during the post-prandial state to support glucose absorption for glycogen replenishment after exercise training (35, 36).

The G protein coupled receptor (TGR5) is located in intestinal cells and is considered a key mechanism by which bile acids favor glucose homeostasis. Bile acids stimulate glucagon-like peptide-1 (GLP-1) release from intestinal L-cells through in part TGR5 activation for modulation of circulating insulin levels (37, 38). Insulin is considered an important modulator of the bile acid pool, and Haeusler et al (11) showed in humans that insulin infusion during a euglycemic-hyperinsulinemic clamp lowered circulating bile acids, although this effect was attenuated in people with obesity. In our study, we observed that lowered insulin levels at 120 min after training were linked to reduced GDCA iAUC_60min. Additionally, decreased insulin iAUC_180min after training was associated with lower GCDCA iAUC_60min and TLCA iAUC_60min. As such, improvements in insulin sensitivity were associated with reduced GCDCA iAUC_60min and GDCA iAUC_60min. Together, this is consistent with work conducted in mice showing that insulin signaling activates FoxO1 and maintains bile acids via up-regulation of CYP8b1 (enzyme key for bile acid synthesis) and normal farnesoid X receptor activity. In contrast, obese and insulin-resistant mice had impaired insulin and Fox-O1 signaling, thereby leading to elevated bile acids (39). This observation with insulin singaling defects would be consistent too with elevated hydrophobic bile acids (mainly glycine and taurine conjugated based) increasing risk for not only cancer, gallstones and gastrointestinal disease, but also induce oxidative stress and cardiac arrhythmia as well as atheroscloersis (40). It is important to acknowledge that glucose itself may also impact bile acid concentrations though promoting CYP7a1 expression (41) and we did observe a lowering 120min plasma glucose. However, we did not observe direct correlations with post-prandial glucose and bile acids. This indicates glucose is likely not a primary factor in our study contributing to bile acid changes. Taken together nonetheless these findings highlight that exercise mediated improvements in glucose regulation are at least partially modified by bile acids in older adults with prediabetes.

Bile acids are consider key regulators of energy metabolism through also TGR5 activation in skeletal muscle and adipose tissue (2). In fact, bile acid binding of TGR5 in skeletal muscle as well as brown adipose tissue leads to induction of cAMP activation of thyroid hormone to raise thermogenesis (42). LCA, a secondary bile acids derived from CDCA, is the most potent TGR5 agonist and a key regulator of energy metabolism (43). In our study, we show that exercise training lowers LCA and CDCA as well as GCDCA independent of intensity. This is relevant since lowering of CDCA and GCDCA correlated with reduced metabolic flexibility and suggests that a lower concentration of bile acids following short-term training related to less carbohydrate oxidation during the OGTT. This likely reflects a role of bile acids in shifting oxidative to non-oxidative glucose metabolism for support of glycogen synthesis after exercise training (44, 45). Indeed, bile acids within the liver have been shown to stimulate of glycogen storage and inhibit of hepatic glycolytic and lipogenic gene expression as evident by carbohydrate responsive element–binding protein (ChREBP) and sterol responsive element–binding protein 1 (44, 46, 47). Since fasting glucose is primarily regulated by the liver, it is of interest to note that reduced fasting glucose after training correlated with lower CDCA iAUC_60min. This is consistent with work in mice with obesity and diabetes who were treated with TGR5 agonists displaying decreased hepatic glucose production (48). Concurrently, conjugated and nonconjugated bile acids also bind to the farnesoid X receptor. This is physiologically relevant given CDCA is the most potent agonist of this receptor. Studies in rodents indicate that bile acid interaction with farnesoid X receptor does not directly affect liver insulin sensitivity though, but rather effects peripheral insulin sensitivity in tissues such as adipose tissue and skeletal muscle (10, 47). Subsequently, our findings of lower LCA and CDCA highlights that exercise may mediate multi-organ benefits for glucose tolerance improvements.

There are several limitations in this study to acknowledge. We have a modest sample size that comprise of primarily of women. Thus we may not be able to generalize our findings to other populations, nor are we able to identify sex specific effects on plasma bile acid levels in response to exercise training despite some (49) but not all (39) noting such differences. Bile acids were collected at points 0, 30, and 60min during the 180min OGTT, and it is unclear whether this would differ through the 180min duration. Interestingly, previous works (20, 21) show that bile acid levels in response to a 75g OGTT decreased after 30min and likely plateaued after 60min time point, which is consistent with our findings. We did not strictly control dietary fat or fiber intake in this study, which could influence the production and metabolism of both primary and secondary bile acids. While work is needed using mixed-meals or high fat tolerance tests to elucidate how plasma glucose and lipid profiles (e.g. cholesterol/triglycerides) interact with bile acids during the post-prandial state, data from 3-day food logs from our prior work (50) using a similar design showcased no difference in macronutrients following training. Therefore, it is unlikely these nutrients differentially impacted our ability to test exercise specific effects. Moreover, we did not determine gut microbiota as a potential mechanism contributing to bile acid changes with exercise (51). Lastly, we did not use the euglycemic clamp to depict insulin sensitivity, although the SIis has been validated as a surrogate measure against the clamp in people with obesity we we previously described (52). Nonetheless, a strength of the study is that people did not consume dietary supplements (pre-/pro-biotics) and people consumed the same diet 24 hour prior to testing, thereby increasing confidence that bile acids were impacted by exercise.

CONCLUSIONS

In conclusion, we report that 2 weeks of aerobic training decreased several bile acids in older adults with prediabetes, independent of intensity. Among these, the post-prandial reduction was observed mostly in secondary bile acids (unconjugated: DCA, LCA, GCDCA, conjugated: HDCA, TLCA) except for the primary bile acid CDCA. These alterations in bile acids following exercise seem clinically relevant as they related to gains in fitness, weight loss and insulin sensitivity. In fact, pharmaceutical use of bile acid sequestrants are reported to reduce plasma cholesterol and HbA1c levels in people with type 2 diabetes (25). Additional work is thus warranted to understand the role of bile acids following exercise and their impact on T2D risk reduction.