Physical activity promotes gut adaptation, nutrient responsiveness, and sensitivity to gut peptides in male mice

Summary

Background

Physical activity helps maintain a healthy stable body weight. One mechanism underlying this beneficial effect could be the improved coupling between energy intake and expenditure, since hunger and satiety are better regulated in active individuals. Whether this enhancement of appetite control by exercise is reflected in overall adaptations in the gut and gut-brain communication remains poorly defined.

Methods

We investigated how increased physical activity alters gut morphology, intestinal endocrine function, and central appetite signalling in C57BL/6NRJ male mice fed ad-libitum chow diet. We assessed intestinal growth, L-cell density, and glucose-stimulated endocrine secretion, as well as circulating levels and gene expression of gut derived peptide hormones. Additionally, we quantified neuronal activity in the brainstem dorsal vagal complex following a fasting–refeeding intervention and examined the effects of PYY, CCK, ghrelin, and GLP-1 administration on food intake in sedentary and physically active mice. Depending on the dataset, unpaired or paired Student's t-tests, two-way ANOVA followed by Bonferroni's post hoc test, simple linear regression or a linear mixed-effects model were applied. Statistical significance was set at P < 0.05.

Findings

Physical activity induced a slight growth of the small intestine, increased L-cell density, and enhanced glucose-stimulated GLP-1 secretion. It altered the circulating levels of PYY and ghrelin and increased the dynamic regulation of neuronal activity in the area postrema and nucleus tractus solitarius. Active mice displayed greater sensitivity to gut-derived hormones PYY, CCK, and ghrelin exerting amplified and prolonged effects on food intake, whereas native GLP-1 showed no effect, likely due to its short half-life. Physical activity prevented post-fasting hyperphagia, thereby promoting sustained maintenance of fasting-induced weight loss.

Interpretation

The findings demonstrate that physical activity promotes adaptations in the gut and gut-to-brain communication possibly enhancing the responsiveness to appetite-regulating signals. Such adaptations may strengthen the alignment between energy intake and expenditure to support body weight maintenance.

Funding

This study was supported by Novo Nordisk Foundation (0059436) and Trygfonden Centre for Physical Activity Research (101390, 20045, 125132, and 177225). P.S. is supported by Lundbeck Foundation (R380-2021-1300).

Research in context.

Evidence before this study

We searched PubMed and Embase for articles without language restrictions, using the terms “exercise OR physical activity”, “weight loss maintenance”, “gut hormones”, “GLP-1”, “PYY”, “CCK”, “ghrelin”, and “gut–brain axis” and “appetite regulation”. We also screened the reference lists of relevant reviews and original articles. Studies were included if they reported on the role of physical activity in regulating gut-derived signals, enteroendocrine adaptations, or central appetite regulation. Although consistent evidence shows that physical activity improves coupling between energy intake and expenditure, the mechanisms underlying this effect remain unclear. Few preclinical or clinical studies have directly addressed how physical activity alters gut morphology, enteroendocrine cell function, or gut-to-brain signalling. Evidence was therefore fragmented, and further research was needed.

Added value of this study

Our study demonstrates that physical activity induces a slight growth of the small intestine, increases L-cell density, and enhances glucose-stimulated GLP-1 secretion. Moreover, physical activity increases the dynamic regulation of neuronal activity in the dorsal vagal complex and enhances sensitivity to pharmacological PYY, CCK, and ghrelin. Physical activity attenuated post-fasting weight regain, suggesting improved regulation of energy balance. The current findings reveal that physical activity induces adaptations directly within the gut, particularly in endocrine functions regulating appetite control. These adaptations may underlie the enhanced appetite regulation observed with exercise in humans, thereby bridging a critical gap in the literature.

Implications of all the available evidence

Taken together, available evidence indicates that physical activity not only promotes energy expenditure but also induces adaptations in the gastrointestinal tract and gut-brain communication that may strengthen the regulatory coupling between energy intake and expenditure. These findings highlight the gut–brain axis as a central mediator of the health benefits of physical activity on appetite regulation and underscore the need for future translational and clinical studies to determine how these mechanisms are regulated in obesity in response to exercise.

Introduction

Physical activity influences energy balance,¹ body weight regulation and overall health.^2,3 In this context, exercise exerts benefits beyond simple caloric expenditure alone, as it improves the regulatory coupling between energy intake and energy expenditure.^4,5 Retrospective studies have demonstrated that moderate to high levels of physical activity are associated with improved weight-loss maintenance and thus reduced risk of body weight regain.^4,6 While the acute caloric deficit induced by a single exercise bout may not be sufficient for body weight maintenance,7, 8, 9 increasing daily physical activity levels may influence appetite regulation.¹⁰ Energy intake has been proposed to follow a J-shaped relationship with physical activity levels, such that lower activity levels are associated with dysregulated energy intake relative to expenditure, whereas higher habitual activity levels are associated with improved appetite control and a better alignment between energy intake and expenditure.¹¹ Indeed, exercise does not only increase food intake because of higher energy expenditure but also regulates subjective hunger feelings, enhances post-meal satiety, and reduces the overconsumption of calorie-dense food.^12,13

Significant progress has been made in understanding how nutrient-stimulated peripheral signals are relayed to the brain to control feeding behaviour and thus body weight maintenance.¹⁴ The gastrointestinal tract, from where several satiety- and hunger-promoting signals derive, is central for sensitive appetite regulation. Among these hormones, glucagon-like peptide-1 (GLP-1), peptide YY (PYY) and cholecystokinin (CCK) are secreted in response to a meal and induce satiety.15, 16, 17, 18, 19, 20 GLP-1 is co-secreted with PYY from L-cells of the small intestine and colon, and CCK is secreted from I-cells in the entire small intestine. The signals from the gut can be transmitted either through the release of gut hormones into the bloodstream or via the afferent fibres of the vagus nerve to the brain to coordinate the feeding response.¹⁴

Acute and chronic exercise may differentially modulate the secretion of gut-derived hormones. Whereas acute exercise increases circulating levels of PYY and GLP-1 in fed-state healthy humans,¹² regular physical activity decreases circulating levels of GLP-1 and enhances nutrient-stimulated GLP-1 secretion, independent of insulin sensitivity, in fasted-overweight men.²¹ CCK increases with acute exercise in healthy males but is not modified by repeated bouts of exercise; however, the evidence is based on a limited number of studies.22, 23, 24, 25, 26 Ghrelin, another nutrient-stimulated peripheral hormone secreted from the stomach, is generally increased following long-term chronic exercise and decreased after high-intensity acute exercise.^27,28 Thus, we hypothesise that increased energy intake resulting from exercise-induced higher expenditure drives gastrointestinal adaptations that modulate the secretion of or sensitivity to nutrient-stimulated peripheral hormones, ultimately improving body weight maintenance.

In the present study, we have uncovered that voluntary wheel running induces adaptations directly in the gut, and in gut-to-brain communication in lean young mice. After 6 weeks of running, active mice displayed higher food intake and positive changes on body composition without an overall effect on body weight as well as longer small intestine which persisted after running cessation, greater density of GLP-1 expressing L-cells and enhanced glucose-stimulated GLP-1 secretion. After daytime fasting, we observed improved satiation and body weight maintenance, as well as improved neuronal responsiveness in the brain stem of physically active mice. Based on these findings, we hypothesised that increased physical activity potentiates sensitivity towards gut-derived hormones, and we confirmed this hypothesis by demonstrating enhanced responsiveness to PYY, CCK, and ghrelin in active mice. Taken together, our data suggest that increased physical activity induces gut adaptations that may, in turn, influence appetite regulation and body weight maintenance.

Methods

Table of materials

Reagents or resources	Source	Identifier
Antibodies
GLP-1 rabbit-anti-mouse	Lab of Jens Juul Holst	2135–8
c-Fos rabbit-anti-mouse	Abcam	Cat# ab190289
Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Thermo Fisher, DK	Cat # A-31573
Donkey Anti-Rabbit IgG (H + L) Alexa Fluor® 555 antibody	Invitrogen	Cat # A-31572
GLP-1 antibody	Lab of Jens Juul Holst	#89390 RRID: AB_289219529,30
CCK antibody	Lab of Jens Rehfeld	#92128 RRID: AB_289300831,32
Chemicals, peptides, and recombinant proteins
Paraformaldehyde 4%	Thermo Fisher, DK	CAS: 30525-89-4
TRIzol	Thermo Fisher, DK	CAS: 593-84-1
SYBR Green Master Mix	Thermo Fisher, DK	Cat# 4309155
Chloroform	Merck	CAS: 67-66-3
Peptide YY (PYY)	Novo Nordisk	NNC0165-1273
Ghrelin	Bio-Techne	CAS: 258338-12-4
Cholecystokinin (CCK)	Bachem	CAS: 25126-32-3
GLP-1 (7–36) amide	Bachem	CAS: H-6795-GMP
Semaglutide	Novo Nordisk	NNC0113-0217
I125 GLP-1	Novo Nordisk
I125 PYY	Perkin Elmer	NEX2400
PYY	Bachem	Cat# H-6045
Hydrogen peroxide	Bio-Techne	Cat# 322381
Target retrieval	Bio-Techne	Cat# 322000
Protease plus	Bio-Techne	Cat# 322331
DAPI mounting medium VECTASHIELD	Vector Laboratories	CAT# H-1200
IBMX (3-isobutyl-1-methylxanthine)	Thermo Fischer, DK	CAT# 228420010
Plasma-coated charcoal	Merck	Cat# 1.02186.0250
Critical commercial assays
U-PLEX Obesity Combo 2	MSD, Rockville USA	Cat# K15301K-1
Acetaminophen L2K Assay	Sekure Chemistry	Cat# 506-10
Multiplex Fluorescent Reagent Kit v2	Advanced Cell Diagnostic	Cat# 323100
Experimental models: organisms/strains
Mouse: C57BL/6NRJ	Janvier	N/A
Oligonucleotides
See Table 1	TAG Copenhagen A/S	https://tagc.com/
Software and algorithms
ImageJ	https://doi.org/10.1038/nmeth.2019	https://imagej.nih.gov/ij/
GraphPad Prism 10	GraphPad Prism	https://www.graphpad.com/features
BDAS software		https://www.harvardapparatus.com/bdas-basic-data-acquisition-software.html
EVOS™ M5000 Imaging System	Thermo Fisher, DK	N/A
Other
Chow diet	Safe Diets	SAFE D30

Experimental model

Ethical considerations

Animal studies were conducted with permission from the Danish National Committee for Animal Research (2020-15-0201-00599), from the local animal use committee (SUND, EMED, P24-183 and P24-252) and in accordance with the guidelines of Danish legislation governing animal experiments and in adherence with the ARRIVE guidelines.³³

Animals

186 male C57BL/6NRJ mice (7 weeks old) were purchased from Janvier (Saint Berthevin Cedex, France). Experiments were carried out in multiple cohorts (see design of studies). The mice were single-housed in a temperature (22 ± 2 °C) and humidity controlled (55%) room and kept in a 12:12 h light/dark cycle with a preceding 30 min half-light period. The mice had ad-libitum access to standard chow (SAFE D30, Safe Diets, 3.389 kcal/g; 22% protein, 5.3% fat, and 50.8% carbohydrates) and drinking water. Mice were housed with small plastic shelter, nesting material, a cardboard tunnel and either a functional or non-functional wheel. C.B-L. and P.S. were aware of the group allocation since in vivo studies were performed without allocation masking.

Method details

Experimental setup

Body weight and food intake were recorded daily. Body composition, including lean and fat mass, was assessed weekly using EchoMRI (EchoMRI™ 4-in-1-500 Body Composition Analyzer, USA). Based on a habituation period of one week approximately, we assigned mice to active (EX; voluntary wheel running) and sedentary (CON; blocked running wheel) groups based on body weight and body composition. To minimize baseline differences, mice were ranked according to body weight, body composition and food intake and then alternately allocated to the exercise (EX) or sedentary control (CON) groups, resulting in comparable baseline mean values between group. Functional 4¾″ × 4.5″ running wheels (Starr, Life Sciences Corp.) were opened after approximately one week of baseline measurements, and mice were euthanised following 6 weeks of running. Non-functional wheels were placed in the cages of sedentary mice to mimic the environmental conditions of the running group. Energy expenditure and energy balance were estimated in a subset of mice (Fig. 1), whose energy intake and body weight were recorded daily and body composition weekly using virtual calorimeter v1.82.³⁴

Fig. 1.

Physical activity impacts metabolic traits and small intestinal length. (A) Colour scheme; sedentary mice (mice with a blocked running wheel) are shown in light blue and active mice (voluntary wheel running) are shown in dark blue. (B) Simple linear regression analysis of total 6 weeks food intake (kcal) in relation to total 6 weeks running distance (km). (C) Mean 24-h food intake (kcal) and (D) mean body weight (g) in sedentary controls and active mice at baseline and after 6 weeks of running compared by mixed model for repeated measurements followed by Bonferroni multiple comparison. (E) Lean mass gain (g) and (F) fat mass gain (g) after 6 weeks intervention compared by unpaired t-test (E = 30). (G) Representative picture of small intestine with cecum and colon attached after 6 weeks intervention from a sedentary and an active mouse. (H) Small intestine length in ad-libitum fed mice after 6 weeks intervention compared by unpaired t-test (E = 32). (I) Simple linear regression analysis of total 6 weeks running distance (km) in relation to small intestine length (cm). (J) Simple linear regression analysis of total 6 weeks food intake (kcal) in relation to small intestine length (cm). (K) Colon length in ad-libitum fed mice after 6 weeks intervention compared by unpaired t-test. (L) Simple linear regression analysis of total 6 weeks running distance (km) in relation to colon length (cm). (M) Simple linear regression analysis of total 6 weeks food intake (kcal) in relation to colon length (cm).

Panels (A) was created using Biorender.

Study 1: Effect of voluntary wheel running on food intake, body weight, body composition and gastrointestinal tract

Mice (7 weeks-old C57BL6/N, n = 21 CON; n = 21 EX) were kept on ad-libitum chow diet with daily measurement of body weight and food in the cage to calculate food intake. At 6 weeks of running mice were sacrificed to collect serum and tissue: small intestine and colon.

Study 2: In situ small intestine perfusion

Mice (7 weeks-old C57BL6/N, n = 16) were kept on an ad-libitum chow diet with weekly measurement of body weight food in the cage to calculate food intake. At 6 weeks of running mice were used for in situ small intestine perfusion to assess intestinal endocrine secretion. All 16 mice went through surgery; however, because of technical issues during the surgery we have results from 13 mice (n = 7 CON; n = 6 EX).

Study 3: Oral glucose tolerance test

At 4 weeks of running mice from study 2 underwent an oral glucose tolerance test (OGTT). Mice were fasted from 06:00 AM (ZT0) to 12:00 PM (ZT6). At T = 0 mice received an oral gavage of a glucose (2 mg/kg BW) with acetaminophen (100 mg/kg) (Merck, CAS: 103-90-2). Blood glucose was measured in the tail vein by a hand-held glucometer at timepoint T = 0, 7.5, 15, 30, 60 and 90 min post oral gavage. At T = 0, 7.5 and 30 minutes post oral gavage 20 μL of blood was drawn from the tail vein into a capillary tube. Concentration of acetaminophen in the blood samples was used as an indicator of gastric emptying. Acetaminophen was measured using a spectrophotometric method kit (Sekure Chemistry, Cat: 506-10).

Study 4: Mixed meal tolerance test

At 4 weeks of running mice from study 7 underwent an oral liquid mixed meal tolerance test (MMTT). Mice were fasted from 06:00 AM (ZT0) to 12:00 PM (ZT6). At T = 0 mice received an oral gavage of liquid mixed meal (0.2 mL; 2.4 kcal/mL) with acetaminophen (100 mg/kg BW). Blood glucose was measured from the tail vein by a hand-held glucometer at timepoint T = 0, 7.5, 15, 30, 60 and 90 min post oral gavage. To measure gastric emptying 20 μL of blood was drawn from the tail vein into a capillary tube at timepoint T = 0, 7.5 and 30 minutes post oral gavage. Concentration of acetaminophen in the blood samples was used as indicator of gastric emptying and was measured using a spectrophotometric method kit (Sekure Chemistry, Cat: 506-10).

Study 5: Effect of fasting and refeeding on post-fasting food intake and body weight gain

At 4 weeks of running mice from study 7 underwent a fasting and refeeding experiment. Body weight and food intake were measured, followed by removing the food from their cages just before the light cycle starts (06:00 AM). Body weight was measured, and food was re-introduced at 12-h post-fasting (06:00 PM). Food intake and body weight were measured at 12-, 24- 48- and 72-h post-refeeding.

Study 6: Effect of fasting and 1-h refeeding on gene expression and neural activity

Mice (7 weeks-old C57BL6/N, n = 23 CON; n = 27 EX) were kept on ad-libitum chow diet with daily measurement of food in the cage to calculate food intake. At 6 weeks, we performed terminal fasting and 1 h refeeding experiments. Body weight and food were measured at ZT0 whereafter cages were changed and food removed from cages and the animals kept fasting until ZT12. Mice were divided into two subgroups (euthanised in the fasted or refed state). Fasted mice were euthanised at ZT12 (n = 12 CON-fasted, n = 13 EX-fasted). Refed mice (n = 11 CON-refed, n = 14 EX-refed) were allowed to refeed for 1 h before they were euthanized. Blood glucose was measured post killing and tissue was dissected and stored for future analysis: small intestine, colon, brain. Results from this study are combined from two cohorts of mice (n = 50, 7 weeks-old C57BL6/N); they were exposed to the same handling and were euthanised in the same manner at the same age.

Study 7: Effect of fasting refeeding; tissue collection

Mice (7 weeks-old C57BL6/N n = 41) were added to supplement tissue collection for analysis in study 6. After 5 weeks of running, mice were euthanised in the following groups: CON-fasting n = 6, CON-1-h refed n = 6, CON-24-h refed n = 6, EX-fasting n = 8, EX-1-h refed n = 7, EX-24-h refed n = 8. Tissue collected: plasma, small intestine, colon, brain, and nodose ganglia.

Study 8: Peptide hormone injections

Peptide hormones related to feeding status were investigated in ad-libitum fed mice. The mice received intraperitoneal (i.p.) injection of peptide hormone PYY (4 nmol/kg BW) (Novo Nordisk, NNC0165-1273), CCK (3.8 nmol/kg BW) (Bachem, CAS: 25126-32-3), Ghrelin (0.3 μmol/kg BW) (Bio-Techne, CAS: 258338-12-4) to measure food induction/suppression and differences between groups. On the day of peptide administration food in all cages was measured 15 min before administration and removed from the cage. Body weight was measured to allow correct dosing, and the food was introduced to the cage again shortly after the injection. Injections were performed in a randomised order. Food was measured 1-, 2-, 4-, 12- and 24-h post peptide administration. Body weight was measured at 12- and 24-h post administration. The study was designed as a crossover, such that mice received either vehicle or peptide injection on day 1, followed by a 3-day washout to avoid carryover effects, after which they received the alternate treatment.

Tissue and serum collection and processing

Mice were sacrificed by decapitation in the evening near lights off time. Feeding status at euthanasia is indicated in the figure legends. To ensure consistent results, the mice were euthanised in a randomised order, and further to minimize an effect of the circadian rhythm killing was sometimes performed over 3 days to ensure that the timing of euthanasia did not differ significantly. Blood was collected from trunk in and Eppendorf tube and quickly placed on ice. The small intestine, colon, brain, and nodose ganglia were surgically removed. Gastrointestinal tissue was weighed and measured before being divided into four sections: duodenum (proximal; first 2 cm after stomach), jejunum (mid; 2 cm approximately 15 cm from stomach), ileum (distal, last 2 cm before the cecum), and colon (last 2 cm before the rectum) (Fig. 2). Depending on future analysis, tissue was either placed in 4% paraformaldehyde (PFA) or snap-frozen in liquid nitrogen. The analysis was conducted blinded.

Fig. 2.

Effect of physical activity on regional intestinal morphology and gene expression. (A) Colour scheme. (B) Villus height (μm), (C) crypt depth (μm), and (D) luminal area (μm²) in the proximal, middle, and distal small intestine were analysed by mixed model repeated measures with exercise (EX) and intestinal segment (Seg) as main factors (E = 11). (E) Illustration of the intestinal segments analysed (prox, mid and dist). Intestinal gene expression in the proximal, mid and distal small intestine (F) Glp-1r, (G) Cck, (H), Cckar, (I) Gcg, (J) Pyy (E = 9). (K) GLP-1 positive cells per villus/crypt unit in the proximal, mid and distal small intestine (E = 10). (L) Representative images of immunofluorescent staining of GLP-1 in the distal small intestine. (M) Intestinal tissue concentration of GLP-1 (pmol/g) in prox, mid and dist small intestine (E = 6). Panels (F)–(M) are compared by two-way ANOVA followed by Bonferroni's multiple comparison with exercise and intestinal segment as main factors. Circulating levels of (N) GLP-1 (pmol/mL), (O) Ghrelin (pg/mL) and (P) PYY (pg/mL) in sedentary and active mice during ad-libitum-feeding compared by unpaired t-test (E = 31). Bar plots show mean ± SEM. P-values are indicated on the individual graphs.

Panels (A) and (E) were created using Biorender.

Serum analysis

As mentioned, whole-blood was collected post decapitation and plasma was saved at −80 °C. Plasma samples were collected from ad-libitum fed mice (study 1), and from fasted versus 1-h refed mice (study 3) and analysed for specific peptide hormones. U-PLEX Obesity Combo 2 (mouse) Multiplex Assay (Meso Scale Discovery, Rockville, USA) was used following the manufacturer's instruction to measure: total Ghrelin, total GLP-1, and total PYY. The kit also measures C-peptide, leptin, and insulin; however, these results will be published elsewhere. The analysis was conducted blinded.

Gene expression analysis

Gene expression analysis was performed in small intestine (proximal, middle and distal sections), colon, and isolated brainstem samples. The tissue was surgically removed and snap-frozen with liquid nitrogen. Total RNA was isolated from tissues with phenol/chloroform extraction method. Briefly, tissue was homogenised with 1 ml of Trizol using Qiagen Tissuelyser Retsch MM300, 200 μL chloroform was added to each sample and spun at 12,000×g for 10 min, 4 °C. After centrifugation, RNA was precipitated using isopropyl alcohol (1:1 reaction), washed with 75% ethanol and dissolved in Ultrapure RNase/DNase free water. RNA purity was measured using Nanodrop spectrophotometer (nanodrop one, Thermo Fischer Scientific, USA). RNA was converted into cDNA using cDNA Reverse Transcription Kit (Cat# 4368813, Life Technologies) following the instructions from the manufacturer and using the same RNA amount for the compared samples. Quantitative PCR (qPCR) were performed with PowerUp™ SYBRTM Green Master Mix (Cat# A25780, ThermoFisher Scientific) and 300 nM of forward and reverse primers and using ViiA 7 Real Time PCR system (Applied Biosystems) for amplification. Relative quantification of the gene was performed using the 2^−ΔCt method, with GAPDH used as the housekeeping gene for all tissues. The analysis was conducted blinded.

Table 1.

Primer sequences for qPCR.

Gene	Concentration	Forward 5′-3′	Reverse 5′-3′
Gapdh	10 pmol/μL	AACTTTGGCATTGTGGAAGG	GGATGCAGGGATGATGTTCT
Gcg	10 pmol/μL	TTACTTTGTGGCTGGATTGCTT	AGTGGCGTTTGTCTTCATTCA
Cck	10 pmol/μL	TAGCGCGATACATCCAGCAGGT	GGTATTCGTAGTCCTCGGCACT
Pyy	10 pmol/μL	ACGGTCGC ATGCTGCTAAT	GACATCTCTTTTTCCATACCGCT
Cckar	10 pmol/μL	CTT TTC TGC CTG GAT CAA CCT	ACC GTG ATA ACC AGC GTG TTC
Ghrelin	10 pmol/μL	CAGAAAGCCCAGCAGAGAAA	GAAGGGAGCATTGAACCTGA
Gpr65	10 pmol/μL	ATGGCGATGAACAGCATGTG	ACGCATAAAGATCCGATGTTGG
Glp-1r	10 pmol/μL	ACG GTG TCC CTC TCA GAG AC	ATC AAA GGT CCG GTT GCA GAA

Haematoxylin and eosin staining with Alcian blue

Intestinal tissue was surgically dissected, washed in PBS and placed in 4% PFA for 24-h at RT. Samples were stored in 4% PFA at 4 °C and then embedded in paraffin and cut by In-Lab (Denmark, https://in-lab.dk/). HE-staining was used to quantify the height and depth of small intestinal villi and crypts. Histological images were acquired using EVOS5000 light microscopy. Villus height and crypt depth was measured manually with Fiji ImageJ. Two sections per intestinal segment from each mouse were analysed. All villus and crypts were measured.

Immunostaining and image analysis of small intestine and brain

The small intestine and brain were surgically removed, and the small intestine was divided in segments (proximal, middle and distal) before being placed in 4% (PFA) for 24 h at RT. After 24-h samples were moved to a 30% sucrose solution for 48-h before being snap frozen in ice-cold isopentane and stored at −80 °C. Intestinal tissue was cut in 16 μm thick sections and placed on glass slides. The brains were surgically dissected and removed from cranial cavity and post-fixed in 4% PFA for 24-h at RT followed by 48-h in 30% sucrose at 4 °C. Brain tissue was sectioned using a cryostat (chamber −24 °C) in 20 μm thick slices and carefully placed in a liquid anti-freeze medium (30% phosphate buffer, 40% ethylene glycol and 30% glycerol) and stored at −20 °C until later use.

Intestinal tissue samples stored on glass slides were used for IF staining targeting GLP-1 positive cells. IF were performed directly on the glass slides. Slides were blocked with a blocking solution (BS) (5% BSA, 5% normal donkey serum (NDS) in PBS) and dried with 100% ethanol before overnight incubation with primary GLP-1 AB (2135-8, 1:6000) at 4 °C. Slides were washed with 1× PBS before second incubation with secondary AB (Alexa Fluor 647, 1:200) for 1 h at RT in a dark room. After final incubation slides were washed and dried and added DAPI, a mounting media that stains cell nuclei, and hereafter stored at 4 °C until analysis.

Brain sections stored in anti-freeze media were washed with 1× PBS. Washing and incubation with AB was performed in wells, keeping brain sections wet at all times. Sections were placed in blocking solution (BS) followed by incubation in primary cFOS AB (ab190289, 1:500) overnight at 4 °C on a slow shaking table. Samples were washed and incubated with secondary AB (Alexa flour 647, 1:500) for 2-h at RT in the dark. After the final wash, sections were mounted on glass slides and left to air dry in the dark. When slides were dry DAPI was added to each slide and covered with a coverslip. Cells positive for the target antigen were counted manually with Fiji ImageJ, normalized to the size of the region of interest and reported per image. Representative images were analysed from sections spanning bregma −7.22 to −7.90. Approximately 6 brain and 3 intestine slides from each mouse were used for quantification. The analysis was conducted blinded.

Intestinal peptide extraction

Tissue samples were collected along the small intestine of fasted mice (CON: n = 4, EX n = 4). Tissues samples were collected from the most proximal, the most distal and the mid small intestine. Samples were snap frozen in liquid nitrogen and stored at −20 °C. The samples were later homogenised in 1% trifluoroacetic acid (Thermo Fisher Scientific, catalogue no. TS-28904) prepared in Milli-Q water (Milli-Q Integral water purification system). Homogenization was performed using 5-mm steel beads in a bead mill (TissueLyzer, QIAGEN Instruments AG) at 30 Hz for 4 × 2 minutes. The homogenates were then left to stand for 1-h at room temperature and centrifuged at 9000×g for 20 minutes.

The resulting supernatants were purified using pH tC18 cartridges (Waters, catalogue no. WAT036810) and eluted with 0.1% trifluoroacetic acid in 70% ethanol. Eluates were dried overnight under a stream of air and subsequently reconstituted in buffer containing 100 mmol/L Tris (Sigma–Aldrich; T-3253, T1503), pH 8.5, 0.1% (wt/vol) human serum albumin (Calbiochem; catalogue no. 12666), 20 mmol/L EDTA, and 0.6 mmol/L thimerosal (Sigma–Aldrich; catalogue no. T-5125). Total GLP-1 concentration was measured with radioimmunoassay which is later described.

In situ small intestine perfusion (study 2)

Ad-libitum fed mice were anaesthetised with an i.p. injection of a Ketamine/Xylazine mix (0.1 mL/20 g) (Ketamine 90 mg/kg (Ketaminol Vet.; MSD Animal Health Madison, USA) and Xylazine (10 mg/kg (Rompun Vet.: Bayer Animal Health, Germany). The abdominal cavity was opened, and the small intestine was isolated by excising the surrounding organs: colon, stomach, and spleen, and the vasculature to the kidneys was ligated. A tube was inserted into the proximal small intestine just below the pyloric sphincter; the intestinal lumen was continuously perfused with heated (37 °C) isotonic saline at flow rate 0.035 mL/min. A catheter was placed in the abdominal aorta perfusing the small intestine at 2.5 mL/min flow rate with a modified Krebs Ringer Bicarbonate buffer heated to 37 °C and oxygenated with 95% O₂ and 5% CO₂. The venous effluent was collected by a catheter placed in vena portae. Hereafter, the mouse was euthanised by perforating the diaphragm. The intestine was perfused for 25 minutes before the initiation of the experimental protocol. Venous effluent samples were collected for 1 minute periods. Samples were immediately placed on ice and stored at −20 °C until further analysis. Each protocol started with a 10 minutes baseline period of luminal saline infusion (0.035 mL/min) followed by intra-luminal glucose stimulation. The luminal stimuli were administered for 15 minutes at an initial bolus rate of 0.135 mL/min for 3 minutes (to replace previous solutions in the lumen) followed by 12 minutes at infusion rate 0.035 mL/min. Post glucose stimulus the lumen was infused with saline to allow wash out bolus rates mentioned above. Intra-vascular KCl infusion at 50 mmol/L (final concentration) was used as a positive control in the end of each experiment. The analysis was conducted blinded.

Perfusion system

We used a single-pass perfusion system (Uniper UP-100, Hugo Electronics-Harvard apparatus, March-Hugstetten, Germany) which heats the perfusion buffer to 37 °C. The perfusion buffer was a modified Krebs–Ringer bicarbonate buffer containing 0.1% (w/v) bovine serum albumin (BSA) (Merck KGaA, Darmstadt, Germany), 5% (w/v) dextran T-70 to balance osmolarity (Pharmacosmos, Denmark), 3.5 mmol/L glucose and 5 mmol/L of fumarate, glutamate and pyruvate, and 10 μmol/L IBMX. The buffer was constantly gassed with 95% O₂/5% CO₂ to maximally increase pO₂ and maintain pH around 7.4. Arterial as well as venous perfusate samples were analysed regularly for pH and partial pressures of oxygen and carbon dioxide (Radiometer Acid-Base laboratory). The intestinal perfusion method is described in detail elsewhere.³⁵

Biochemical measurements of perfusion effluent and tissue protein extraction

Total amidated GLP-1 and total PYY concentrations in venous effluents from intestinal perfusions were measured by in-house-developed radioimmunoassay (RIA). Total GLP-1 (7–36amide) was measured with an in-house-developed RIA, based on a C-terminally directed antiserum specific for the amidated GLP-1 form (code no. 89390, RRID: AB_2892195).^29,30 The standard was synthetic GLP-1 7–36NH₂ (cat. no. H-6795-GMP, 4081700, Bachem, Frechen, Germany), and the tracer was monoiodinated ¹²⁵I-labelled GLP-1 (7–36NH₂) (a gift from Novo Nordisk A/S, Bagsværd, Denmark). Total PYY was measured with a porcine antiserum (cat. no T-4093, Bachem, Germany); it should be noted that porcine and murine PYY share amino acid sequence.³⁶ The antibody used recognizes PYY1/3–36 and PYY1/3–34 equally. The standard was synthetic rat/mouse/porcine PYY (cat. no. H-6045, Bachem, Germany), and the tracer was ¹²⁵I-labeled porcine PYY (cat. no. NEX2400). Free and bound peptides were separated with plasma-coated charcoal (cat. no. 1.02186.0250, Merck, Germany). CCK was measured using antiserum (number 92128, RRID: AB_2893008), which binds all bioactive (i.e., α-amidated and tyrosyl O sulphated) forms of CCK with equal potency without cross-reactivity with any gastrin.^31,32 Bolton–Hunter ¹²⁵I-labelled sulphated CCK-8 was used as tracer and sulphated CCK-8 for standards.³² Antigen-bound peptide radioactivity was counted by a gamma counter (Wizard2, Automatic Gamma Counter, PerkinElmer, Denmark), and radioactivity was translated into hormone levels by interpolation on a standard curve with known concentrations increasing from 0 pmol/L to 320 pmol/L an adequate range to detect all samples in this study.

Data presentation and statistical analysis

No statistical methods were applied to predetermine the sample size for experiments as a priori estimates of effect size and variability were not feasible. However, we have employed the Resource Equation approach,³⁷ to assess whether sample size used was adequate or not. The resource equation value is presented for each study in each figure legend (resource equation: E = total number of animals – total number of groups). All mice in the running group were included in the analysis regardless of their running distance. There was great variation in running distance (155 km), but importantly this was balanced across studies. Statistical analyses were conducted using GraphPad Prism 9 (GraphPad, La Jolla, USA) and IBM SPSS Statistics Software 29.0.1.0. The specific statistical tests employed for each dataset are detailed in the corresponding figure legends. These included: simple linear regression analysis used to assess the association between two continuous variables; unpaired student’s t-test for comparison between two independent groups, paired t-tests for matched measured; and two-way ANOVA followed by Bonferroni’s multiple comparison for experiments with two factors and multiple group comparisons. In case of missing values, the data were analysed by fitting a mixed model, rather than by repeated measures ANOVA. All tests were two-tailed, with a significance threshold of P < 0.05. Assumptions of normality and homogeneity of variance were assessed by QQ-plots and residual plots before parametric testing. A linear mixed-effects model was used to assess the effect of training, drug, and time on food intake, with training, injection, and time included as fixed effects and mouse identification as a random effect. Box plots show full data range, and bar graphs depict mean ± SEM. Confidence intervals are presented where they enhance the reader's interpretation of the estimates. The graphs were made using GraphPad prism 9 (GraphPad, La Jolla, USA). P-values are depicted on the graphs.

Role of funders

The funders had no role in study design, data collection, data analysis, interpretation of data, or manuscript writing.

Results

Physical activity promotes a slight growth of the small intestine, which persists after running cessation

First, we assessed the physiological response to regular physical activity by exposing male mice to 6 weeks of voluntary wheel running. Ad-libitum-fed mice were single-housed, and based on equal phenotypic distribution, they were assigned to active (functional running wheel) or sedentary control (blocked running wheel) groups (Fig. 1A). Active mice ran 257 km on average over a 6 weeks period corresponding to 6.1 km/day. The standard deviation was 155.6 km, indicating a large variation in total running distance; however, this was balanced across all studies. The running distance correlated positively with food intake, increasing by 0.43 g per km (Fig. 1B). As we observed previously,³⁸ the estimated daily energy expenditure was higher in the active group, which resulted in a null energy balance (Figure S1A–B) due to an increase in food intake since active mice consumed 2.64 ± 0.36 kcal/day, in relative terms 20%, more food compared to sedentary controls (Fig. 1C). Body weight increased by 1.18 ± 0.25 g in sedentary mice and by 1.55 ± 0.25 g in active mice during the 6 weeks (Fig. 1D). After 6 weeks of wheel running, active mice displayed increased lean mass (1.33 ± 0.2 g) and decreased fat mass (−0.52 ± 0.2 g), whereas sedentary mice showed increased lean mass (0.34 ± 0.1 g) and increased fat mass (0.30 ± 0.2 g) (Fig. 1E and F). Together, these results highlight that voluntary wheel running promotes food intake and improves body composition without overall changes in body weight.

Given the observed daily increase in food intake in the active mice, we assessed whether increased physical activity induces adaptations in the intestinal tract. In ad-libitum fed mice we measured the small intestinal length in both active and sedentary mice. Although not significant, active mice had 1.3 ± 0.71 cm longer small intestine than sedentary (CI: [−0.1; 2.7], P = 0.076 for the difference) (Fig. 1G and H). Interestingly, the small intestine length was positively correlated with total 6 weeks running distance and total food intake (Fig. 1I and J). There was no correlation between small intestinal length and total 6 weeks food intake in sedentary mice (Fig. 1J), revealing that increased physical activity may promote intestinal growth possibly driven by increased food intake. The colon was 0.57 ± 0.26 cm longer in active mice (CI: [0.33; 1.0]) compared to sedentary (Fig. 1K); however, the colon length was not significantly correlated with total running distance (Y = 0.002× + 6.6, CI for slope: [−0.0003; 0.005], P = 0.077) (Fig. 1L). Colon length was not correlated with food intake in neither active nor sedentary mice (Fig. 1M). Of significance, active mice displayed a longer small intestine even at 17 days post running cessation (Figure S1D–E); in this inactivity period the active mice was still displaying increased food intake compared to sedentary controls.³⁸ Notably, the colon was not significantly longer in active mice after running cessation (Figure S1F). Overall, voluntary wheel running slightly increases the length of the small intestine, and this adaptation is maintained even after cessation of running.

Increased physical activity amplifies the intestinal and circulating nutrient-stimulated signals

Food digestion is one of the main functions of the small intestine; however, nutrient digestion and absorption are distinct to the intestinal region. Therefore, seeing that physical activity promotes small intestinal growth we investigated region-specific changes promoted by physical activity. We did not observe any differences in villus height, luminal area nor crypt depth between ad-libitum-fed sedentary and active mice (Fig. 2A–D). However, physical activity did regulate the expression of gut-derived appetite regulating peptides and their receptors in the small intestine depending on intestinal region. In particular, physical activity upregulated the expression of Glp-1r (mean difference in the proximal small intestine: 0.013 ± 0.003 CI: [0.006; 0.02], P = 0.0004) (Fig. 2F), the gene encoding pro-glucagon (Gcg) (mean difference in the proximal small intestine: 0.003 ± 0.001 CI: [0.0007; 0.006], mean difference in the mid small intestine: 0.005 ± 0.001 CI: [0.001; 0.008]) (Fig. 2I), and Cck in the proximal small intestine (mean difference 0.002 ± 0.0007 CI: [0.0001; 0.004]) (Fig. 2G). No changes were observed on Cckar or Pyy expression in any sections of the small intestine (Fig. 2H, J). Of note, Glp-1r and Gcg expression was not modified during or beyond the exercise period in the colon (Figure S1D–E). We observed molecular changes induced by activity predominantly in the proximal small intestine, a region essential for digestion and nutrient sensing.

The intestinal L-cells play a pivotal role in nutrient sensing and endocrine secretion. Given the observed effect of physical activity on the small intestine possibly related to nutrient sensing and absorption, we assessed the number of L-cells by immunostaining for GLP-1 (Fig. 2L). Density of GLP-1 positive L-cells increased along the small intestine, reaching a peak in the distal ileum (Fig. 2K), as previously described.³⁹ Overall, there was an effect of physical activity on the number of L-cells across all segments (mean difference: 0.3 ± 0.1 CI: [0.02; 0.59] P = 0.038], the difference being most pronounce in the distal small intestine (mean difference: 0.5 ± 0.2 CI: [0.07; 0.92], P = 0.04). Despite that active mice had a greater number of GLP-1 positive cells compared to sedentary mice in the small intestine (Fig. 2K–L), they show similar levels of circulating GLP-1 (Fig. 2N), which may reflect technical constraints affecting the detection of circulating GLP-1, which is known to be difficult to measure in mouse plasma.⁴⁰ Peptide analysis of intestinal tissue showed no effect of physical activity on GLP-1 concentration in fasted mice (mean difference: 6.5 ± 2.7 CI: [−0.06; 13.06], P = 0.052) (Fig. 2M); however, this may be a result of insufficient power in the analysis. In serum we measured circulating levels of PYY which showed no effect of physical activity (mean difference: −8.2 ± 4.1 CI: [−16.6; 0.22], P = 0.056) (Fig. 2O), although it tended to be lower in the active mice. Lastly, we observed lower circulating ghrelin (mean difference: −87.49 ± 41.82 CI: [−172.4;−2.6]) in active mice compared to sedentary (Fig. 2P). Some appetite regulating signals are relayed through the nodose ganglion to the brain stem. GLP-1R, CCKaR, and GGR65 are expressed on vagal afferent nerves innervating the gastrointestinal tract and are described to be important for nutrient sensing and appetite regulation.^18,41,42 Increased physical activity had no effect on expression of these nutrient-stimulated receptors in the nodose ganglia or brain stem at the end of the light phase during ad-libitum feeding (Figure S2A–B). By contrast, during postprandial satiety (ZT0), physical activity was associated with downregulation of Gcg (mean difference: 0.58 ± 0.1) and Cckar (mean difference: 0.28 ± 0.1) expression in the brain stem. Cck and Glp-1r expression in the brain stem was not significantly modulated by physical activity (Cck mean difference: −0.2 ± 0.06 CI: [−0.46; 0.06], P = 0.078) and (Glp-1r mean difference: −0.13 ± 0.07 CI: [−0.25; 0.13], P = 0.13) (Figure S2C) indicating an adaptive response to gut adaptations potentially dependent on feeding status.

Increased physical activity strengthens the postprandial intestinal adaptive response to nutrients

Showing changes in intestinal expression and circulating levels of nutrient-stimulated gut-derived peptides in active mice (Fig. 2), we speculated that the intestinal endocrine response may be enhanced by increased physical activity. Therefore, we investigated glucose-stimulated endocrine secretion from the small intestine using in situ isolated small intestine perfusion (Fig. 3A). GLP-1 secretion increased in response to a 15 minutes luminal infusion of glucose in both sedentary and active mice (Fig. 3B). Glucose-stimulated GLP-1 secretion was 179.1 ± 53.4 pmol higher in active compared to sedentary mice in the same 15 minutes period (Fig. 3C). The greater GLP-1 secretion was independent of small intestine length (Fig. 3D), and in support of this finding we did observe a correlation between total GLP-1 secretion and total running distance (slope estimate: 1.57 CI: [0.82; 2.3], P = 0.004) (Figure S2D). Glucose did not promote PYY secretion neither in active nor sedentary mice (Fig. 3E and F). CCK secretion was induced by glucose in both sedentary and active mice (Fig. 3G) but there was no significant effect of physical activity (mean difference: 132.0 ± 82.7 pmol CI: [−49.9; 313.9], P = 0.14) (Fig. 3H).

Fig. 3.

Effect of physical activity on intestinal endocrine function. (A) Colour scheme and illustration of the in situ intestine perfusion setup. (B) GLP-1 secretion (pmol/L/min) from in situ isolated perfused small intestine stimulated with 20% luminal glucose and positive control KCl. (C) Total GLP-1 secretion (pmol/15 min) during 15 minutes luminal glucose stimulation compared by unpaired t-test (E = 11). (D) Total GLP-1 secretion (pmol/15 min) during 15 minutes luminal glucose stimulation standardised to intestinal length compared by a mixed model with repeated measures with exercise (EX) and glucose as main factors. (E) PYY secretion (pmol/L/min) from in situ isolated perfused small intestine stimulated with 20% luminal glucose and positive control KCl. (F) Total PYY secretion (pmol/15 min) during 15 minutes luminal glucose stimulation compared by t-test (E = 11). (G) CCK secretion (pmol/L/min) from the in situ isolated perfused small intestine. (H) Total CCK secretion (pmol/15 min) compared by t-test (E = 11). (I) Blood glucose in sedentary and active mice following OGTT (J) AUC of blood glucose following OGTT compared by unpaired t-test (E = 14). (K) Acetaminophen concentration in serum following OGTT, and (L) AUC of acetaminophen absorption following OGTT compared by unpaired t-test (E = 14). (M) Blood glucose in sedentary and active mice following a liquid MMTT. (N) AUC of blood glucose following a MMTT compared by unpaired t-test (E = 25). (O) Acetaminophen serum concentrations following MMTT, and (P) AUC of acetaminophen absorption following MMTT compared by t-test (E = 20). Data are shown as mean ± SEM. P-values are depicted on the graphs.

Panel (A) was created using Biorender.

GLP-1 suppresses food intake and slows gastric emptying.⁴³ Given the enhanced glucose-stimulated GLP-1 secretion in active mice, we next examined whether running influences glucose tolerance and gastric emptying. We assessed this using an oral glucose tolerance test (OGTT) and a liquid mixed meal tolerance test (MMTT) combined with acetaminophen as an indicator of gastric emptying rates. Active mice had significantly better glucose tolerance following an OGTT (Fig. 3I and J); however, there was no apparent effect of running on gastric emptying rates in the OGTT (Fig. 3K–L). Glucose tolerance appeared similar between groups during the MMTT (Fig. 3M–N). Nonetheless, there was a slight delay (mean difference: −1259 ± 661.4 CI: [−2639; 120], P = 0.071) in acetaminophen absorption rates in active mice compared to sedentary (Fig. 3O–P). In support of the suggested effect of physical activity on gastric emptying following MMTT we observed a correlation between total acetaminophen absorption and total running distance (slope estimate: 3.94 CI: [1.3; 18.8], P = 0.029) (Figure S2E). The results suggest that running affect glucose tolerance, and possibly gastric emptying, depending on the gastric load employed. In summary, these results indicate that increased physical activity improves the ability to regulate nutrient-stimulated peripheral signals.

Increased physical activity improves satiation and satiety responses after calorie restriction

It is well established that exercise supports long-term weight-loss maintenance after dietary intervention.44, 45, 46 We therefore hypothesised that running may improve satiation and satiety after caloric restriction. Using the fasting-refeeding paradigm,⁴⁷ we examined food intake during the transition from negative energy balance to acute energy intake. Mice were fasted for 12-h during the light phase to avoid the strong dark-cycle feeding drive that could otherwise mask between-group differences. Following fasting, sedentary mice compensated for the energy deficit by increasing food intake by 46% over the subsequent 24-h (mean increase: 4.5 ± 0.8 CI: [2.4; 6.5]), whereas active mice showed no significant changes in food intake (mean difference: 0.9 ± 1.1 CI: [−1.9; 3.8], P = 0.7) (Fig. 4A and B). Sedentary mice continued to overconsume at 48 h, with food intake remaining elevated by 25% (mean increase: 2.4 ± 0.6 CI: [0.8; 4.0]). By 72-h post-fasting, increase in food intake in sedentary mice was no longer significant (mean increase: 1.2 ± 0.5 CI: [−0.2; 2.7], P = 0.11). In contrast, active mice maintained food intake throughout the refeeding period (Fig. 4B). Both sedentary and active mice lost body weight during the daytime fast. Sedentary mice fully recovered this weight loss within 24 h of refeeding and continued to gain weight, exceeding baseline body weight at 48-h (mean increase: 0.39 CI: [−0.02; 0.81], P = 0.07) and at 72-h (mean increase: 0.44 CI: [0.03; 0.85], P = 0.03). In contrast, active mice exhibited improved body-weight maintenance with body weight remaining significantly lower than baseline at 72-h of refeeding (mean difference: 0.67 ± 0.16 CI: [0.3; 1.1], P = 0.0004) (Fig. 4C). Collectively, these results indicate that active mice exhibit more robust regulation of satiation and satiety following fasting-induced body weight loss, thereby protecting against rebound body weight gain.

Fig. 4.

Effect of physical activity on gut-to-brain communication. (A) Colour scheme. (B) Food intake in sedentary and active mice 24-h before (baseline) and 24-, 48, and 72-h post a 12-h daytime compared by Bonferroni multiple comparison compared to baseline food intake. (C) Body weight at baseline, after 12-h daytime fasting and after 24-, 48-, and 72-h of refeeding compared by two-way ANOVA for repeated measurement with Bonferroni multiple comparison comparing to baseline. (D) In a new cohort we investigated fasting and acute 1-h refeeding. (E) 1-h food intake, and (F) body weight change post 1-h refeeding following a 12-h daytime fast compared by unpaired t-test (E = 32). (G) Blood glucose during fasting and after 1-h refeeding compared by two-way ANOVA with Bonferroni multiple comparison (E = 87). Circulating concentrations of (H) GLP-1, (I) ghrelin and (J) PYY in fasted and 1-h refed mice compared by two-way ANOVA with Bonferroni multiple comparison (E = 32). (K) Representative pictures of immunofluorescent cFOS staining in the dorsal vagal complex of a sedentary and an active mouse. cFOS positive cells/mm² in the dorsal vagal complex including (L) AP, (M) NTS, and (N) DMV of fasted and 1-h refed mice (CON-fasted n = 4, CON-refed n = 4, EX-fasted n = 3, EX-refed n = 4, E = 11). Main effect of refeeding (F) and exercise (EX) is compared by two-way ANOVA with Bonferroni multiple comparison. Data are shown as mean ± SEM. P-values are depicted on the graphs.

Panels (A) and (D) were created using Biorender.

To investigate the physiological adaptations underlying the observed enhanced satiation, we performed an acute 1-h refeeding experiment following 12-h daytime fasting (Fig. 4D). As expected, we observed no differences in food intake (Fig. 4E) or body weight gain (Fig. 4F) between sedentary and active mice, despite the usual higher food intake of active mice during this period (Figure S3A–B), confirming the anticipated overcompensation in sedentary mice. Blood glucose increased after 1 h of refeeding in both groups, but active mice exhibited significantly lower levels (Fig. 4G). Gastrointestinal adaptations to fasting and refeeding were then examined: refeeding increased stomach and small intestine wet weights, independent of running, while cecum and colon weights remained unchanged (Figure S3C–F). Notably, small intestinal length was greater in active mice but unaffected by feeding (mean difference during fasting: 2.1 ± 1.1 CI: [−0.1; 4.2], P = 0.063, and during refeeding: 1.9 ± 1.1 CI: [−0.3; 4.2], P = 0.089) (Figure S3G), indicating that running-induced intestinal growth are independent of acute feeding changes. Although gene expression analysis of Glp-1r, Cckar, and Gcg showed no effect of feeding status or of physical activity, Glp-1r was non-significantly upregulated in the proximal small intestine of refed active mice (mean difference Glp-1r: 0.01 ± 0.005 CI: [−0.00005; 0.02], P = 0.08) (Figure S3H–J), in line with our observations during ad-libitum-feeding. Glp-1r was also observed to be downregulated in the middle section in fasted active mice, whereas there was no effect of physical activity on Glp-1r expression in the distal small intestine (mean difference: −7.3e-005 ± 4.8e-005 CI: [−1.9e-004; 4.3e-005]) (Figure S3H). Circulating GLP-1 and PYY increased, and ghrelin decreased in the acute refed state, irrespective of exercise (Fig. 4H–J).

Some intestinal satiety signals are relayed through the nodose ganglion; thus, we next assessed the expression of nutrient-sensing markers in the ganglia. Feeding did not affect Glp-1r, Cckar or Gpr65 expression; however, Glp-1r was downregulated in active mice independent of feeding (mean difference: 0.001 ± 0.0006 CI: [8.8e-005; 0.003], P = 0.038) (Figure S3K). Further, did we observe downregulated expression of Cckar in the active mice (mean difference: −0.006 ± 0.002 CI: [−0.01; 0.0002]). Taken together with intestinal Glp-1r and Cckar expression, these findings suggest that physical activity influences nutrient sensing mechanisms. Signals from the nodose ganglion induce satiation by acting on the brain stem dorsal vagal complex, including the area postrema (AP), the nucleus tractus solitarius (NTS), and the dorsal motor nucleus of the vagus (DMV).⁴⁸ Therefore, we assessed acute neuronal activity in these regions following the fasting-refeeding paradigm using immunofluorescent cFOS staining (Fig. 4K). Refeeding increased neuronal activity in the AP and NTS, with no overall significant differences between sedentary and active mice. However, exercise tended to influence neuronal activity in both the AP and NTS (mean difference in AP: 6.0 ± 3.6 CI: [−1.1; 13.2]; in NTS: 6.0 ± 3.8 CI: [−1.5; 13.6]). Comparisons between fasted and refeeding conditions revealed significant increases in the number of active cells in the AP (mean difference: 20.9 ± 4.6 CI: [11.6; 30.2]) and NTS (mean difference: 12.8 ± 4.9 CI: [1.4; 24.3]) only in the exercise group, suggesting that physical activity modulates the dynamic regulation of feeding-related central signals (Fig. 4L–M). No significant effects of feeding status or physical activity were observed in neuronal activity within the DMV (Fig. 4N). In summary, physical activity enhances the regulation of satiation and satiety in mice, limiting compensatory overeating after fasting and maintaining stable food intake. These effects are associated with running-induced adaptations in the nutrient-sensing pathways in the nodose ganglion and a dynamic modulation of neuronal activity in the brain stem dorsal vagal complex.

Increased physical activity enhances sensitivity to gut-derived pharmacological peptides

Given that circulating GLP-1, PYY, and ghrelin levels were similar between sedentary and active mice after fasting, while dynamic regulation of neuronal activity in the dorsal vagal complex was higher and food intake remained stable in active mice, we suggest that active mice may be more sensitive to gut-derived appetite signals, requiring less hormonal input to achieve the same satiation effect. We therefore evaluated the effects of pharmacological appetite-regulating peptides on food intake.

We conducted a crossover study to assess food intake following PYY, CCK, and ghrelin injections at the onset of hunger (ZT12) and fullness (ZT0) in ad libitum-fed active and sedentary mice. At ZT12, PYY suppressed food intake by 40%, CCK by 55%, and ghrelin increased intake by 70% in active mice during the first hour post-injection (Fig. 5A–C). At ZT0, PYY suppressed intake by 75% and ghrelin increased intake by 110% in active mice (Fig. 5D–F), at this dose there was no statistical significant effect of CCK in either groups (mean difference in active mice: −0.329 ± 0.16 CI: [−0.68; 0.03], P = 0.063). Importantly, the effects of peptide injections persisted for several hours in active mice (Figure S4A–G). Interestingly, active mice generally consumed more during the dark-phase and less during the light-phase, consistent with enhanced pre-meal hunger signalling as well as post-meal satiety signalling (Figure S4A–G), as has been reported in humans.^12,13,49,50 In line with this, PYY, CCK, and ghrelin had greater effects on food intake in active mice, particularly during the light-phase 1-h post-injection. Native GLP-1 had no detectable effect (Figure S4H), likely due to its short half-life.51, 52, 53 These results demonstrate that increased physical activity enhances sensitivity to gut-derived appetite-regulating hormones, with active mice showing stronger responses to PYY, CCK, and ghrelin. This suggests that exercise not only modulates feeding behaviour but may also fine-tune the hormone-appetite pathways controlling hunger and satiety.

Fig. 5.

Peptide hormone sensitivity. Schematic overview of experimental setup. (A) 1-h food intake post saline or PYY injection at ZT12 (CON n = 7, EX n = 13, E = 18), (B) 1-h food intake post saline or CCK injection at ZT12 (CON n = 8, EX n = 8, E = 14), (D) 1-h food intake post saline or ghrelin injection at ZT12 (CON n = 7, EX n = 7, E = 12). (E) 1-h food intake post saline or PYY injection at ZT0 (CON n = 14, EX n = 14, E = 26), (F) 1-h food intake post saline or CCK injection at ZT0 (CON n = 14, EX n = 14, E = 26), (G) 1-h food intake post saline or Ghrelin injection at ZT0 (CON n = 8, EX n = 8, E = 14). 1-h food intake following peptide injection was compared to food intake following vehicle injection by paired t-test.

Panel (A) was created using Biorender.

Discussion

Here, we show that exercise-induced increase in food intake is associated with adaptations of the gastrointestinal tract which may represent a key component of the mechanisms underlying the beneficial effects of exercise on body weight maintenance.⁵⁴

During high energy demands, appetite regulation is predominantly homeostatic.⁵⁵ Previous work from our group demonstrated that running-induced energy compensation depends on the initial running distance and functions as a compensatory response to an initial running-induced fat loss, whereas the subsequent increase in lean mass occurs following the rise in energy intake.³⁸ In the present study, we found that once energy balance is restored in chronically active mice, which was characterized by reduced fat mass and increased lean mass, habitual running induces an increase in small intestinal length, likely driven by elevated food intake. Notably, this intestinal elongation, along with elevated food intake, persisted beyond the running period. The increase in small intestine length was positively correlated with total running distance, and with total food intake in active but not sedentary mice, pointing to a key role of physical activity in driving the slight growth of the small intestine by promoting food intake. The intestine is the largest endocrine organ of the body, and the gut-derived hormones exert peripheral and central effects that regulate energy homeostasis. Therefore, exercise-induced intestinal growth could influence physiological functions,⁵⁶ although further studies are needed to confirm this.⁵⁶ Intestinal growth has previously been described in energetic diet or cold challenges associated with energy store depletion.⁵⁷ Since running-induced hyperphagia serves as a defence against energy loss,³⁸ the accompanying intestinal adaptation may support this response by enhancing nutrient absorption capacity, as has previously been hypothesised.⁵⁷ Although running-induced colonic growth has not previously been described, we speculate that exercise-related changes in inflammatory signalling may modulate this adaptation. As gut inflammation is associated with colonic shortening, we propose that inactive mice exhibit greater intestinal inflammation than active mice. Consequently, the increased colonic length observed here may reflect the absence of inflammation-induced shortening rather than active growth.

Not only did exercise induce morphological adaptations of the intestine, but we also observed that exercise increased L-cell density and enhanced glucose-stimulated GLP-1 secretion. Notably, running enhanced GLP-1 secretion independently of small intestine length, consistent with the greater number of GLP-1 positive cells per villus/crypt unit observed in the distal intestine. This provides evidence for a direct, exercise-induced regulation of GLP-1 production in the small intestine. Despite these intestinal adaptations, running had no effect on circulating GLP-1 levels during ad-libitum feeding. However, active mice exhibited lower circulating ghrelin and reduced circulating PYY levels compared to sedentary controls. Importantly, both the nutrient-stimulated intestinal responses and peripheral hormonal signals influenced by running are dependent on feeding-status, highlighting the importance of carefully controlling dietary conditions as potential confounders in exercise studies. In humans, acute exercise slows down gastric emptying,⁵⁸ an effect possibly mediated by GLP-1^43,59 since circulating GLP-1 is shown to increase in response to an acute exercise bout.^60,61 Habitual physical activity is associated with lower fasting and higher meal-induced GLP-1 secretion.⁴⁴ In the present study, increased physical activity tended to slow gastric emptying following a mixed meal test, but not following a glucose tolerance test. Taken together, these data suggest that physical activity alters both basal and meal-induced secretion of appetite-related hormones, which may contribute to improved appetite regulation over time.^17,21,62 Importantly, this nutritional signalling may be impaired in obesity since it has been demonstrated that people living with obesity have impaired pre- and post-meal GLP-1 secretion,63, 64, 65 and that both obese humans and mice may have lower Glp-1 expression.66, 67, 68 Based on our findings in lean mice, it would be essential to determine whether running can restore this nutritional signalling in the context of obesity. Although caution is warranted when extrapolating these findings to humans, our results are consistent with recent evidence showing that a 1 year moderate-to-vigorous intensity exercise programme increased late-phase postprandial GLP-1 secretion in obese humans following weight loss induced by a low-calorie diet.⁶⁹

We also assessed molecular changes in tissues involved in appetite signalling. Glp1-r and Cckar are highly expressed on gut-innervating enteric nerves and are essential for sensitive appetite regulation by sensing both nutrients and stretch in the gastrointestinal tract.^42,70, 71, 72, 73, 74 In the present study, Glp-1r expression was upregulated in the proximal small intestine of active mice compared to sedentary mice, independent of feeding-status whereas it tended to be downregulated in the mid small intestine in fasted active mice. Although GLP-1 is secreted in similar amounts from the proximal and distal small intestine, the density of GLP-1 positive L-cells increases along the small intestine. Given that the intestinal feeding response is primarily driven by proximally secreted GLP-1,^39,75 we propose that the physical activity-induced increase in food intake promotes upregulation of Glp-1r expression in the proximal small intestine, potentially contributing to an enhanced feeding response in active mice. Gut-derived signals stimulating the enteric nerves are relayed to the brainstem via the nodose ganglion. While no effects were observed in the nodose ganglia or brainstem at the end of the light phase, expression of Gcg and Cckar in the brainstem was downregulated by running at the beginning of the light phase, corresponding to a postprandial satiation phase. These findings underscore that the effects of physical activity depend on feeding status. To evaluate how the feeding state influences gut-derived hormones, we assessed both circulating hormone levels and RNA expression in the intestine, nodose ganglion, and brainstem during fasting and 1-h after refeeding in sedentary and active mice. Although no differences were observed in the intestinal gene expression and circulating levels of appetite hormones between fasting and acute refeeding, Glp-1r expression in the nodose ganglion was downregulated by exercise, independent of feeding status. Additionally, active mice exhibited greater neuronal response to refeeding in the AP and NTS compared to sedentary mice. Thus, even though circulating levels of PYY, ghrelin and GLP-1 were similar between groups, the neural responsiveness of specific brainstem neurons appears to be enhanced by running. These findings led us to propose that physical activity enhances sensitivity to satiety signals, such that less hormonal input is required to achieve comparable satiation in active mice.

To study the sensitivity to gut-derived peptides, we assessed food intake following administration of pharmacological peptide hormones. Low doses of PYY and CCK suppressed food intake, whereas ghrelin stimulated food intake only in active mice; this pattern was observed during both the light and dark phases. Exercise also exercise influenced feeding patterns by increasing food intake during the dark phase, resulting in higher total 24-h food intake; notably, however, active mice consumed less food during the light cycle than sedentary mice. Together, the data underpin that exercise enhances not only pre-meal hunger but also post-meal satiation and satiety. This is further supported by the observation that active mice exhibited greater sensitivity to pharmacological administration of both anorectic and anorexigenic drugs. Considering that running initially elevates energy intake³⁸ and that appetite signalling is enhanced after four weeks of running, these findings may reflect a long-term adaptation that improves the alignment of energy intake with expenditure.³⁵ These findings are consistent with self-reported hunger and satiety in human studies, which have shown that individuals with high levels of physical activity exhibit both increased pre-meal hunger as well as post-meal satiety.^10,55,62 In particular, physically active individuals were better able to incorporate the calories of a calorie-dense preload meal into their total energy intake, preventing consumption in excess of energy requiremments.^62,76 This aligns with our results, in which sedentary mice exhibited a compensatory eating pattern following daytime fasting, resulting in additional body weight gain compared to active mice. Overall, our study demonstrates that, in addition to enhancing hunger signalling, increased physical activity improves post-meal satiation and satiety signalling. These adaptations may facilitate a better alignment between energy intake and expenditure, thereby contributing to long-term body weight maintenance.

Energy homeostasis is a biological system that is conserved both in mice and humans.⁷⁷ However, the extent to which these physiological and molecular findings fully translate to humans is uncertain, as our experiments were not conducted at thermoneutrality. In mice, energy balance is primarily regulated in mice by total energy expenditure, particularly under standard housing conditions at ∼22 °C,^78,79 as they increase heat production to maintain core temperature. In contrast, humans live within their thermoneutral comfort zone, where thermoregulation minimally influences total energy expenditure, and therefor energy balance is likely to rely proportionally more on energy intake.^78,80 Moreover, energy demand and body composition change throughout life,³⁸ which may modulate the effects of exercise on appetite regulation and tissue-level adaptations. In the present study, we did not examine tissue-level adaptations during aging. Another limitation is the relatively small sample size in some comparisons and the exclusive use of male mice. Metabolic responses to exercise differ between sexes, both at baseline and in response to exercise,^81,82 and a previous study reported that voluntary exercise increases food intake more in females than males. Furthermore, in females, exercise does not induce changes in body weight or body composition, whereas in males it reduces both body weight and fat mass.⁸³ Therefore, gut adaptation, responses to nutrients, and sensitivity to gut peptides in response to exercise may differ in females. In this study, we used voluntary wheel running, which limits precise control over activity volume, intensity and timing. We observed large variation in total running distance, likely reflecting a combination of biological differences. When excluding runners averaging less than 2 km/day, who likely did not engage meaningfully in running activities, the standard deviation decreased to 97.8 km, approximately 30% of the mean, which is within the expected biological range for voluntary wheel running. We did not investigate the biological factors underlying this variation; therefore all running mice were included in the statistical analysis. Although voluntary wheel running is characterised by intermittent bouts, resembling interval training and active lifestyle in humans, the total daily running time and distance are relatively high, which may not fully resemble structured human exercise programmes.⁸⁴ Factors such as wheel design, sex, strain, diet, and age can influence running pattern, and therefore study outcomes.⁸⁴ Nevertheless, voluntary wheel running avoids the stress responses typically associated with treadmill or swimming exercises in mice,⁸⁵ and aligns with their natural rhythmic activity patterns.^85,86 This low-stress model was particularly important for our assessment of neuronal activity in specific brain regions, as stress-induced activation would overwise confound these measurements.^82,83

In conclusion, in mice, physical activity was accompanied by increased energy intake, which likely drives gastrointestinal adaptations that modulate the intestinal responses to nutrients and enhance sensitivity to gut-derived peptides. Moreover, the number of GLP-1 positive L-cells was higher in active mice, suggesting an increased capacity for nutrient sensing and hormone secretion, which may contribute to the beneficial effects of exercise on long-term appetite regulation and body weight maintenance.⁸⁶ We propose that the morphological and molecular intestinal adaptations observed in active mice reflect improved intestinal health and function. Overall, these results indicate more sensitive nutrient sensing and a potentially improved alignment between energy intake and expenditure, which could underlie enhanced appetite regulation and body weight control in physically active individuals.

Contributors

Conceptualization, C.B.-L., J.J.H., B.K.P., and P.S.; Methodology, C.B-L., J.J.H., J.F.R. and P.S.; Formal analysis, C.B.-L. and P.S.; Investigation, C.B.-L., J.V.U., K.D.G., J.L., H.K., J.F.R. and P.S.; Visualization, C.B.-L.; Verification, C.B.-L. and P.S.; Writing – original draft, C.B.-L.; Writing – review & editing, C.B.-L., J.J.H., B.K.P., J.F.R. and P.S.; Supervision, C.B.-L., J.J.H., B.K.P. and P.S. All authors read and approved the final version of the manuscript.

Data sharing statement

Lead contact

Further information requests regarding resources and reagents should be directed to and will be answered by the Lead Contact, Paula Sanchis (Paula.Sanchis.Tortosa@regionh.dk)

Materials availability

This study did not generate new unique materials.

Data and code availability

Raw data is available upon request to lead contact Paula Sanchis. This study does not generate any codes.

Declaration of interests

Peptide hormone PYY (NNC0165-1273) and semaglutide (NNC0113-0217) were provided by Novo Nordisk Compound Sharing. All authors revised the manuscript critically for important intellectual content and gave their approval for the current version to be published. Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, is supported by Novo Nordisk Foundation (NNF23SA0084103). J.J.H. reported consulting fees from Third Bridge, Metaphore Biotechnologies, Vial Health Technology, Crinetics Pharmaceuticals, Immunic Therapeutics, Guidepoint, Fractyl Health, Alcimed, Alphasights, Soffinova Partners, Jefferies International; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from AstraZeneca, Eli Lilly and Decheng Capital Global Lifesciences Fund; support for attending meetings and/or travel from Endocrine Society and Eli Lilly; participation on a Data Safety Monitoring Board or Advisory Board for Novo Nordisk A/S; leadership or fiduciary role in other board, society, committee, or advocacy group, paid or unpaid on Antag Therapeutics and Bainan Biotech; and stock or stock options in Antag Therapeutics and Bainan biotech.