The effect of sex and endurance exercise training on the incretin signaling pathway in 17 rat tissues

Abstract

Incretin-based pharmacotherapies, particularly glucagon-like peptide-1 (GLP-1) receptor agonists, have transformed the treatment of type 2 diabetes, with demonstrated benefits across multiple organ systems. Their success has driven investigation of related gut-derived hormones, most prominently dual GLP-1/glucose-dependent insulinotropic polypeptide (GIP) receptor agonists, but extend to other targets with similar metabolic functions. For this class of drugs, the extent to which organ health improvements are secondary to improved systemic glycemic control versus direct tissue signaling remains unclear, partly because receptor availability across tissues is poorly annotated. We leveraged data from the Molecular Transducers of Physical Activity Consortium to annotate incretin receptor expression across 17 tissues in Fischer 344 rats and the Genotype-Tissue Expression (GTEx) Portal for human-level receptor expression. Further, given the role of exercise in the preservation of muscle mass during weight loss, we analyzed the effects of 1, 2, 4, or 8 weeks of treadmill exercise training on incretin-related signaling at the epigenetic, transcript and protein levels. Endurance training elicited sex- and tissue-specific changes in incretin receptor expression, including downregulation of Gcgr across brown adipose, adrenal glands, and white adipose tissues. Training-induced Gipr regulation occurred in the adrenal glands, brain cortex, and hippocampus. Collectively, these findings contribute to the map of incretin receptor biology and identify exercise-responsive regulatory axes that may underlie synergistic effects of exercise and incretin-based therapies on weight management and metabolic health.

New and noteworthy

This study provides the first multi-omic, multi-tissue description of incretin signaling receptor expression and regulation in response to endurance exercise training. We identify timepoint and sex-specific changes in incretin signaling across tissues, highlighting training effects on Gcgr, Gipr, and Sctr regulation in the adrenals, scWAT, and brain. These findings help establish an exercise-responsive incretin signaling axis that may identify interactions from incretin–based therapies and exercise-based lifestyle interventions.

Graphical Abstract

Introduction

The increasing usage of GLP-1 receptor (GLP-1R) medications has further emphasized the need for a comprehensive understanding of the role of the incretin signaling pathways across tissues. To date, the majority of literature dedicated to incretin physiology, primarily for GLP-1R, has focused on the gut, pancreas and brain¹, with evidence of expression in other peripheral tissues varying between studies^2–32. While the insulinotropic effects of the GLP-1R in pancreatic β-cells and its role in regulating the gut–brain axis are becoming well established, the role of GLP-1 in extra-pancreatic tissues is preliminary. For instance, studies have reported GLP-1R-mediated effects on renal sodium excretion³³, myocardial contractility^34,35, and neuroprotection³⁶. These and other benefits have prompted GLP-1RAs use for managing various kidney, cardiovascular, autoimmune, and other conditions^37–39. The extent to which these benefits are secondary to improved glucose management from primary signaling in the gut-brain axis is unclear; the degree of extra-pancreatic primary signaling is not well annotated, partly due to an incomplete understanding of GLP-1R expression across tissues.

Table 1 summarizes current literature on GLP-1R expression across organs and organ systems in humans, rodents, and non-human primates. Current literature suggests that GLP-1R is not expressed in skeletal muscle^5,16,19,21, the esophagus⁷, adipose tissue^16,19,21,23, bone marrow^15,21, the uterus²¹, or the prostate^5,21. Reports conflict as to whether lymph nodes express GLP-1R. One study did not detect receptors via immunohistochemistry⁵, whereas another identified GLP-1R mRNA in mesenteric lymph nodes¹⁵. Generally, existing research demonstrates that GLP-1R is absent in human and primate thyroid^2,3,5,13,20, except when neoplastic¹³. GLP-1R does, however, appear to be expressed in murine thyroid^2,5,7.

Table 1:

[Multi-omic analysis of GLP-1 receptor expression across tissues]. This table summarizes the findings from the current literature on GLP-1 receptor expression of various extra pancreatic tissues. Species examined included rats, mice, and human tissues with unrestricted tissue types. Proteomic and transcriptomic data were analyzed, and qualitative observations of GLP-1 receptor expression were recorded and tabulated. Column 1 lists the PubMed ID for each study, while row 1 includes the names of the examined tissues.

Study	Liver	CNS	Heart	Lung	Skeletal Muscle	Stomach	Small bowel	Large bowel	Kidney	Adipose tissue	Transcript/Protein	Method of Assessment	Other	PMID
				Each cell indicates a manuscript where the tissue(s) in question is studied.
Richards et al., 2014	Mu	Mu	Mu			Mu	Mu				Protein	GLP1RA fluorescent reporter mouse	Mu: pancreatic α-, β-, and δ-cells, vascular smooth muscle, cardiac atrium, gastric antrum/pylorus, enteric neurons, and vagal and dorsal root.	24296712
Pyke et al., 2014	Mk		Whole heart (Mk), SA Node (Hu)	VSMC (Mk)		Mk	Mk	Mk	VSMC (Mk), Hu		Protein	Immunohistochemistry	Pancreatic beta cells and to lesser extent acinar cells (Mk, db & not db). Ampulla of Vater (Mk). Pancreas (Hu), Thyroid (Mk)	24467746
Körner et al., 2007	Hu	Hu	Hu	Hu, R, Mu	Hu	Hu	Hu	Hu	Hu	Protein	GLP1R Autoradiography, Immunohistochemistry	Pancreas, Breast (Hu). Thyroid (R, Mu), Thymus (Hu), Spleen (Hu), Lymph nodes (Hu), Prostate (Hu), Adrenal gland (Hu)	17475961
Cork et al., 2015		Mu									Protein	Transgenic Mouse Model		26500843
Andersen et al., 2021	Mu		Mu	Mu		Mu	Mu	Mu	Mu		Transcript, Protein	GLP1RA reporter mouse, Immunohistochemistry, in-situ hybridization	Tracheal cartilage chondrocytes, skin fibroblasts (Mu), pancreas, thyroid, esophagus, Thymus, Spleen	33508122
Grunddal et al., 2021						Mu	Mu	Mu			Transcript, Protein	Competition Binding, Autoradiography, In situ Hybridization, Immunohistochemical staining	COS-7 cells (from monkey kidney tissue) transfected with Hu/Mu GLP1R. Mu Pancreas & enteric nervous system.	34662392
Guedes et al., 2015							Hu				Protein	Immunohistochemistry		26048514
Graham et al., 2020		Mu								Transcript, Protein	Transgenic Mouse Model-- Immunohistochemistry and In-situ hybridization		32170734
Hamilton et al., 2009		Mu									Protein	Immunohistochemistry		19617854
Jensen et al., 2018		Mu									Transcript, Protein	Immunohistochemistry, In-situ Hybridization.		29095968
Gier et al., 2012											Protein	Immunohistochemistry	Thyroid (Hu)	22031513
Farr et al., 2016		Hu									Protein	Immunohistochemistry		26831302
Yusta et al., 2015							Mu	Mu			Transcript, Protein	Functional Activity, Flow Cytometry, mRNA Expression, Immunohistochemistry	Bone marrow, thymus, spleen, mesenteric lymph node	25735732
Bullock et al., 1996	R	R	R	R	R	R	R		R	R	Transcript	RNAse Protection Assay, RT-PCR, In-situ Hybridization	R Pancreatic islets, spleen, testes	8770921
Campos et al., 1994	Mu	Mu		Mu			Mu	Mu	Mu		Transcript	RT-PCR, Southern Blot	Mu Pancreas	8156917
Smith et al., 2022		Mu									Transcript	Transgenic Mouse Model, single cell RNA seq		35430298
Wei et al., 1995	Hu	Hu	Hu	Hu	Hu	Hu			Hu	Hu	Transcript	RNAse Protection Assay, RT-PCR	Hu Pancreas	7843404
Caltabiano et al., 2020											Transcript, Protein	Immunohistochemistry, In-situ Hybridization, RT-PCR, Western Blot	Testes (Hu, Mu), Thyroid (Hu), Pancreas (Hu), Tumor leydig cell line (R)	33460247
Yamada et al., 2016	Mu	Mu	Mu	Mu	Mu	Mu	Mu	Mu	Mu	Mu	Transcript	RT-PCR	Pancreas, Skin, Spleen, Testes, Placenta, Ovary, Uterus, Eye, Spinal Cord, Bone Marrow, Prostate, Thymus, Adrenal Gland	27186360
Jun et al., 2014	Mu	Mu	Mu	Mu		Mu	Mu		Mu		Transcript, Protein	Immunohistochemistry, Western Blot, Immunoprecipitation, RT-PCR, Functional Assays	Pancreas (Hu, Mu)	24695667
Campbell et al., 2022										Mu	Transcript	PCR of genomic DNA and RNA, Whole-Mount Confocal Microscopy		35192688
Alvarez et al., 1996		R									Transcript	PCR, Southern blot, Northern blot, In-situ hybridization	N2A Neuroblastoma cells (Mu), RINm5F pancreas cells (R)	8769850
Broide et al., 2013						H					Transcript, Protein	RT-PCR, Immunohistochemistry		23803499
Yamato et al., 1997	Mu	Mu					Mu			Mu	Transcript	RT-PCR	Pancreas (Mu)	9105899
Baggio et al., 2018			Hu								Transcript, Protein	PCR, Southern blot, Western blot, Immunohistochemistry		29444223
Nakagawa et al., 2004	R	R		R							Transcript	RT-PCR, Northern Blot, In-situ Hybridization	Bilateral nodose ganglia (R), Pancreas (R)	14766323
Alvarez et al., 2005		Hu									Transcript	In-situ Hybridization, Binding Assays		15686481
Wismann et al., 2017						Mu	Mu	Mu			Transcript	In-situ hybridization		28702324
Kedees et al., 2013							Mu	Mu			Transcript, Protein	qRT-PCR, Immunohistochemistry, Western Blot		23927844
Koehler et al., 2015							Mu	Mu			Transcript	qPCR, Radiolabeling, Microarray		25738454

Mu = mouse, Hu = human, R = rat, Mk = monkey (or NHP)

VSMC = vascular smooth muscle cells

^*If possible, cell type-specific enrichment is specificed

^*Excluded disease tissues (e.g. tumors)

GLP-1R-positive organs display heterogeneity in receptor distribution. Several organs express the receptor in few cell types. The distribution of these is often so narrow that, with the exception of the lungs, many studies fail to detect GLP-1R at all^{3,5,16,19,21,22,26,28}. Studies show that the liver^4,7, thymus, spleen⁷, and kidneys^3,5,7 express GLP-1R only in blood vessels; the heart in blood vessels and myocytes of the sinoatrial node^3,4,7; the lungs in blood vessels, club cells, and pneumocytes^3,5,7; adrenal glands in chromaffin cells ^7,40; and the testes in Leydig cells²⁰. In the pancreas, GLP-1R is present in acinar cells, β-cells, δ-cells, blood vessels, and approximately 10% of α-cells^3,4,7,8, while epithelial ductal cells are GLP1-R negative²⁰.

In the gastrointestinal tract, reports vary as to which regions and cell types express GLP-1R. Some studies suggest that there is no GLP-1R in the mucosa of the stomach or intestines^4,5. Others posit that stomach and intestinal mucosa do indeed express GLP-1R, albeit with inconsistencies in the cell types proposed to express the receptor^{3,4,7,8,25,31}. Current literature does, however, appear to agree that only immune cells within the lamina propria of the GI tract express GLP-1R^8,25. One study finds that gastric muscularis expresses GLP-1R³⁰, while another suggests that only a subset of gastric smooth muscle cells possess GLP-1R³. A study examining the ileum and large intestine, however, found no GLP-1R in the muscularis⁵.

GLP-1R expression is highly compartmentalized in the CNS, with various regions positive and others negative for GLP-1R^6,10,11,41. GLP-1R is expressed in several major brain compartments, including the cerebral cortex^{6,10,21,22,24,29}, olfactory bulb^{6,10,12,21,29}, hippocampus^{6,10,11,21,29,41}, amygdala^6,10,41, thalamus^10,29,41, and hypothalamus^{4,6,10,14,18,22,24,29,41}. GLP-1R expression in the CNS also varies by cell type. For example, ependymocytes of the CNS are GLP-1R-negative^10,18, while pyramidal neurons often possess GLP-1R^11,14. One report specifically notes that GLP-1R is typically present in GABAergic and glutamatergic neurons, but not in dopaminergic and cholinergic neurons¹⁰. Much like the gastrointestinal tract, studies examining GLP-1R in the CNS often disagree. For example, immunohistochemical data consistently identifies GLP-1R in the cerebral cortex^6,22 and cerebellum^11,41 of adult organisms, conflicting with similar research examining GLP-1R mRNA transcripts in the same brain regions^{6,21,22,24,29}. One possible explanation is that there are development-related effects on the expression of GLP-1R in various brain regions¹⁷. Altogether, GLP-1R expression depends on localization, cell type, and life cycle stage in the CNS.

The success of GLP-1RAs has further spurred evaluation of related satiety and insulin-regulating molecules as standalone or combination targets. Most famously, GLP-1R/GIPR dual agonists have demonstrated improved glucose management outcomes compared to GLP-1RAs alone. Other advancements include GLP-1R/GIPR/glucagon receptor (GCGR) tri-agonists that have advanced to clinical trials^42–44. Vasoactive intestinal peptide (VIP) stimulates glucose-dependent insulin secretion through VIP2 (also known as VPAC2) and has prompted development of selective analogs⁴⁵. Other targets, including cholecystokinin (CCK) or neuropeptide Y (NPY), also represent gut-derived targets with overlapping metabolic functions under active preclinical and clinical investigation. As with GLP-1R, the tissue distribution of these receptors is incompletely annotated, limiting mechanistic interpretation of their pleiotropic effects.

Exercise is routinely co-prescribed with GLP-1 based therapies, yet whether exercise modulates the same receptor axes is poorly described. Exercise training reduces obesity and type 2 diabetes risk^46–48, driven by increasing insulin-independent glucose uptake, improving peripheral insulin sensitivity and preserving β-cell function⁴⁹. Exercise mitigates lean mass loss during GLP-1RA treatment⁵⁰. Combined, incretin agonists and exercise prove superior to medication alone in maintaining bone density and lean body mass, decreasing abdominal obesity, and improving blood glucose regulation^41,51–54. Also similar to exercise training, the efficacy of GLP-1RAs displays sex heterogeneities; women taking GLP-1RAs often lose more weight compared to men⁵⁵, with initial evidence suggesting interactions between GLP-1, androgens⁵⁶ and estrogens⁵⁷.

Here, we leveraged our comprehensive Molecular Transducers of Physical Activity Consortium (MoTrPAC) rat study, that investigated the multi-omic effects of 1, 2, 4 or 8 weeks of progressive endurance exercise training across 17 different rat tissues⁵⁸ to describe receptor expression across tissues and determine how incretin and related gut hormone receptors are regulated by regular exercise.

Methods

Rats, housing and diet.

The experimental design, including animal handling, acclimation, treadmill training, and tissue collection has been detailed elsewhere^58,59. Briefly, female and male Fischer 344 inbred rats were placed into a reverse dark–light cycle housing for at least 10 days prior to treadmill familiarization to allow entrainment to the new light cycle. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Iowa.

Experimental groups, treadmill familiarization and exercise training.

Five experimental groups were studied: sedentary control (SED) or 1, 2, 4, or 8 wk of treadmill training (1W, 2W, 4W, 8W, respectively). Rats went through a 12-d treadmill (Panlab 5-lane rat treadmill; Harvard Instruments, Model LE8710RTS) familiarization protocol. Endurance exercise treadmill training began at 6 months of age. A progressive training protocol that targeted an intensity of ~70%–75% of VO₂max was used. Rats trained 5 days per week, followed by 2 days of rest, while SED rats were only placed on the stationary treadmill.

Tissue collection.

Tissue dissection occurred 48h after the last training session in the dark cycle between the hours of 11:30am and 2:30pm. Food was removed 3h before starting dissections in anesthetized rats. Blood and tissues (heart, vena cava, spleen, gastrocnemius (SKM-GN), vastus lateralis (SKM-VL), subcutaneous white adipose tissue, brown adipose tissue, liver, lung, kidney, adrenal, cortex, hypothalamus, hippocampus, small intestine, and colon) were collected as previously detailed⁵⁹.

RNA sequencing (RNA-seq)

Detailed description of the RNA-seq procedures are available elsewhere⁵⁸. Briefly, after tissue lysis, we extracted total RNA using a BiomekFx automation workstation. RNA-seq libraries were prepared from ~500 ng of poly(A)-selected RNA using the Universal Plus mRNA-Seq kit (NuGEN/Tecan, #9133) on a Biomek i7 automation system (Beckman Coulter). Libraries were sequenced with 100bp paired-end reads on an Illumina NovaSeq 6000 platform. For the bioinformatics pipeline, reads were demultiplexed using bcl2fastq2 (v2.20.0), adapters trimmed with cutadapt (v1.18)⁶⁰, and quality assessed with FastQC (v0.11.8)⁶¹. Alignment was performed with STAR (v2.7.0d)⁶² and quantification with RSEM (v1.3.1)⁶³. We generated normalized expression values by TMM-normalizing filtered raw counts and converting to log counts per million using edgeR (v4.2.2) ⁶⁴ using edgeR::calcNormFactors and edgeR::cpm.

Assay for Transposase-Accessible Chromatin using Sequencing (ATAC-seq)

We performed ATAC-seq (assay for transposase-accessible chromatin using sequencing) following a modified Omni-ATAC protocol, as detailed elsewhere⁵⁸. From a subset of tissues 50,000 nuclei were collected (or up to 500μL), pelleted in 1 mL ATAC-RSB buffer, and resuspended in 50μL transposition mixture. Transposed DNA was purified, amplified and cleaned before pooled libraries were sequenced (Illumina NovaSeq 6000) with 50bp paired-end reads, targeting 35 million read pairs per sample.

Reads were demultiplexed (bcl2fastq2,v2.20.0) and analyzed using the ENCODE ATAC-seq pipeline (v1.7.0, https://github.com/ENCODE-DCC/atac-seq-pipeline)⁶⁵. Peaks from all samples were concatenated, trimmed to 200bp around summits, and merged using bedtools (v2.29.0)⁶⁶ to create a master peak list. This was intersected with filtered alignments using bedtools coverage (-nonamecheck and -counts options) to generate a peak-by-sample count matrix. Subsequent steps were performed separately for each tissue. Peaks from non-autosomal chromosomes or fewer than 10 reads in at least four samples were removed. Filtered counts were quantile-normalized using limma-voom for downstream analyses. Batch correction was applied by including sample processing batch as a covariate in differential analysis.

Multiplexed immunoassays

Detailed methods have been described previously⁵⁸. In brief, tissues were lysed in cold lysis buffer (550μL: 25mM Tris-HCl pH 7.5, 100mM NaCl, 0.01% Triton-X 100, 0.0025% NP-40, 1x Halt protease inhibitor cocktail) and sonicated. After centrifugation (>20,000xg, 10 minutes, 4°C), protein concentration was measured, samples were normalized to the lowest protein concentration and plated in duplicate into 96-well plates; all tissue samples were analyzed on the same plate. The Stanford Human Immune Monitoring Center performed immunoassays for the Luminex panel Metabolic Hormone (RMHMAG-84K) Magnetic Bead Panels (Millipore). Plates were read on a Luminex FlexMap3D instrument (minimum 50 beads per analyte) and data analyzed using Masterplex QT software. Raw mean fluorescence intensities (MFI) were log2-transformed, and measurements with fewer than 20 beads were removed. Within each panel and tissue, we also removed samples with >50% missing values and features with at least two missing values in any experimental group. Remaining missing values were imputed using k-nearest neighbors (k=5). Within each panel, tissue, and analyte, we removed outliers more than 4 standard deviations from the mean.

Human GTEx Expression Comparison

All human gene expression values were obtained from the “gtexr” R package (v0.2.1)⁶⁷, corresponding to the GTEx v8 data release⁶⁸. For each transcript, the detection cutoff in the MoTrPAC dataset was defined as having >0.5 counts per million in at least 2 samples, which matches the original manuscript. Detection cutoff for GTEx is considered median >0.1 TPM, matching GTEx eQTL minimum expression criteria⁶⁸. BAT and Vena Cava were not available in the human GTEx dataset. Skeletal muscle in the human GTEx dataset was used to compare to both rat SKM-GN and SKM-VL.

Statistical analyses

Data used in the preparation of this article were obtained from the Molecular Transducers of Physical Activity Consortium (MoTrPAC) database, which is available for public access from the R package MotrpacRatTraining6moData. Specific datasets used were [v2.0.0] of the MotrpacRatTraining6moData package⁵⁸. This includes Z-scores and summary statistics from molecular differential analysis for each gene, tissue, sex, and timepoint combination, and expression levels from molecular assays. For the ATACseq analysis, if multiple sites were annotated to a specific gene, values across sites were averaged. Corresponding epigenetic analysis done treating sites within a gene annotation independently is separately available in the github repository. Pairwise Pearson correlations between the assays were calculated for all selected incretin genes, and the strongest correlations were highlighted in the figures.

P-values were adjusted per tissue/assay/timepoint using the Benjamini-Hochberg method for false discovery rate across features of interest. Statistical significance was considered FDR<0.1. All code can be found here: https://github.com/christopherjin/exercise_glp1.

Results

Incretin receptor expression across tissues

Figure 1A describes the broader set of gut hormone receptors and examples of their downstream cellular effects. Across all tissues and incretin receptor combinations, 92/153 (60.1%) of receptors were detected in rats (Figure 1B) and 68/135 (50.4%) of available receptors were detected in humans (GTEx). The overall agreement between rats and humans for shared investigated incretin receptors detected was 101/135 (74.8%), with the highest agreement found in the hypothalamus, WAT, and skeletal muscle tissues 8/9 (88.9%). The lowest agreement was in the heart and colon tissues (5/9, 55.6%). In the rats, incretin receptors were detected most frequently in the hypothalamus (9/9, 100%), small intestine (8/9, 88.9%), spleen (7/9, 77.8%), WAT (7/9, 77.8%), and hippocampus (7/9, 77.8%). In humans, incretin receptors were detected most frequently in the hypothalamus, (9/9, 88.9%) and cortex (7/9, 77.8%). Vipr2 was detected in every tissue measured in rats and in each measured tissue in humans except blood (Figure 1C). Glp1r was measured in the cortex, hypothalamus, hippocampus, blood, vena cava, lung, spleen, kidney and small intestine tissues in rats, but only in cortex, hypothalamus, and heart tissues in humans.

Figure 1: GLP-1R expression across 17 tissues in sedentary rats.

A. Diagram showing annotated roles of gut signalling molecules.

B. Table showing Rat and Human expression detection as annotated in MoTrPAC and GTEx data, with overall summary of agreements.

C. Triangle heatmap describing which tissue-incretin combinations are detected in human and rat tissues. MoTrPAC rat data is described in the upper right triangle and Human GTEx data is in the lower left. Grey indicates data unavailable from GTEx, and the outlined rectangles indicate disagreement in detection status. Detection limits used are defined by the respective consortia (see: methods).

Effect of endurance training on expression of incretin-related signaling genes across tissues

Across the 17 investigated tissues, endurance training induced relatively few and modest transcriptional changes in incretin and other appetite- and metabolism-regulating peptide hormone receptors 48 hours after the last exercise bout. Nevertheless, several notable patterns emerged. Glp1r expression, highest in the lung at baseline, was consistently downregulated in both sexes following 8 weeks of training (Figure 2A–B). The Gipr showed sex-divergent responses in the adrenal gland, with sustained decreases in females across 2W and 8W but a transient increase in males at 1W (Figure 2A–B), similar to many other metabolism-related transcripts in the adrenal gland⁶⁹. The most consistent training-induced upregulation was observed for Sctr in the male spleen, which increased progressively across multiple timepoints. Gcgr displayed the most sustained tissue-specific response, with progressive downregulation across all training timepoints in the female adrenal gland. Vipr2 was upregulated in male blood and female gastrocnemius at 8W, pointing to potential roles in peripheral metabolic and neuromuscular adaptation (Figure 2A–B). In the colon, Cckar expression diverged by sex, trending toward decreases in females and increases in males (Figure 2A). Finally, Cckbr was downregulated in the male hypothalamus, and its expression in the hippocampus was correlated with chromatin accessibility across timepoints (Figure 2C), suggesting epigenetic regulation as a mechanism underlying exercise-induced transcriptional responses at this locus.

Figure 2: Incretin receptor responses to endurance training across tissues and sexes.

A. Heatmap of differential transcriptomic responses across tissues following endurance training at multiple timepoints. logFCs represent fold changes for transcript-level changes (red=upregulation, blue=downregulation) relative to sex-match controls (male=blue, pink=female) at select timepoints (1w, 2w, 4w, 8w) in selected incretin-related genes. Timepoint of comparison indicated by green top annotation. Column annotations indicate sex, tissue, and timepoint. Grey boxes indicate that expression in a given tissue did not meet necessary quantification thresholds (see methods). Asterisk indicates the tissue/sex/timepoint combination is significant at FDR<0.1.

B. Triangle plots showing tissue-specific transcriptional changes in selected incretin-related genes following 8 weeks of endurance training. The direction of regulation is indicated by triangle orientation (upward or downward), while color and size encode sex-specificity: female-specific (pink), male-specific (blue), and sex-consistent (white). Features passing an FDR of 0.1 were visualized.

C. Chromatin accessibility (ATAC_MEAN) and RNA expression (TRNSCRPT) of Cckbr were measured in hippocampal tissue from male (triangles, blue) and female (circles, pink) rats at 1, 2, 4, and 8 weeks of endurance training. The blue line represents the line of best fit, shown with standard error bounds.

Effect of endurance training on GIP and GLP-1 peptide levels across tissues

Training altered the levels in the adrenal gland; in males after 1W of training, there was an increase in GLP-1 (logFC=0.72, FDR=0.04, Figure 3) and GIP (logFC=0.37, FDR=0.07). Training also increased levels of GIP in the small intestines; In females, GIP increased at 1W (logFC=0.10, FDR=0.03) and 2W (logFC=0.10, FDR=0.03). In both sexes, small intestine GLP-1 showed logFC between 1.05 and 3.67 at every timepoint; this exploratory signal did not meet FDR<0.1 at any individual timepoint and requires replication in an adequately powered follow-up.

Figure 3: Incretin signalling molecule abundance responses to endurance training.

Tissue-specific changes in GIP (left) and GLP-1 (right) peptide levels as measured by immunoassay following endurance training. Each cell represents the log2 Fold change (logFC) relative to sex-matched control in males and females for multiple timepoints (1w, 2w, 4w, 8w). Color intensity reflects the direction and magnitude of change, with blue for downregulation and red indicating upregulation; Asterisk indicates the tissue/sex/timepoint combination is significant at FDR<0.1.

Discussion

This study describes incretin and related gut hormone receptor expression across 17 rat tissues and characterizes their regulation by 1, 2, 4, and 8 weeks of endurance training at the transcript, chromatin, and peptide levels. Receptor expression is broadly distributed but heterogeneous; bulk detection in rats and humans showed substantial concordance in metabolically active tissues including hypothalamus, white adipose, and skeletal muscle, while peripheral tissues such as heart and colon diverged. This cautions against framing GLP-1 and related gut-derived hormones as exclusively pancreatic signals and underscores that tissue-level receptor availability is a prerequisite for interpreting pleiotropic drug effects. Training-induced transcriptional responses were modest in magnitude but tissue- and sex-specific, with the most consistent effects in adrenal Gcgr and Gipr, scWAT Gipr, hippocampal Cckbr, and pulmonary Glp1r. The localization of these effects to tissues with established roles in energy balance and stress signaling suggests that exercise engages the broader incretin axis beyond the canonical gut–pancreas–brain circuit.

Endurance training produced opposing patterns of Gipr regulation across tissues, consistent with the dual peripheral and central roles of GIP signaling in energy balance. In male scWAT, Gipr expression rose at 1W, 2W, and 4W of training, with the 2W timepoint reaching statistical significance. GIP signaling in adipose tissue supports physiological lipid storage, fatty acid uptake, and tissue blood flow following feeding⁷¹, and impaired adipose GIP signaling in humans tracks with insulin resistance and defective lipid handling⁷². The exercise-induced increase in Gipr expression observed in male scWAT may thus reflect enhanced signaling sensitivity and contribute to improved glycemic control during early training adaptations. In contrast, training decreased Gipr expression in the cortex, and in the adrenal gland of females. Consistent with these findings, mouse models with germline or CNS–specific Gipr deletion exhibit reduced body weight and improved glucose metabolism⁷³, suggesting that exercise could affect energy balance partially via central GIP regulation.

Gcgr expression in the adrenals decreased at all training timepoints in female rats. Glucagon is well-known for its general role in energy homeostasis, e.g. by stimulating hepatic glucose production in response to the usage of glucose during acute exercise⁷⁴. The reduction in adrenal Gcgr may reflect ligand-induced receptor downregulation secondary to repeated exercise-driven elevations in circulating glucagon. Another possibility for the decrease in adrenal Gcgr expression is an adaptive dampening of the glucagon–catecholamine–cortisol stress axis, reflecting a reduced stress response following rat familiarization with the exercise training regimen⁷⁵.

In the hippocampus, a progressive strengthening of the correlation between chromatin accessibility (ATAC) and transcript expression for Cckbr was observed with increasing training duration. Cckbr expression in the hippocampus is predominantly localized to the dentate gyrus and on GABAergic interneurons, where CCK induces GABA release⁷⁶; exercise has been shown to increase hippocampal GABA release⁷⁷. Thus its decreased accessibility and expression may be driven by negative feedback regulation due to reduced need for CCK-mediated GABA release. The correlation between chromatic accessibility and transcript expression increased at each timepoint up to 8W of training. While these data do not establish causality, the pattern supports an exercise dose-dependent regulatory relationship. Further experimental validation is needed to determine whether the observed ATAC–RNA relationships represent a direct causal pathway or a correlated adaptation within broader hippocampal remodeling processes.

Several observed changes, including Glp1r downregulation in the lung or Sctr upregulation in the spleen, are difficult to interpret because the functional roles of these receptors in their respective tissues are not well established. In the lung, GLP-1R has been implicated in airway tone, vascular function, and inflammation in mice, but metabolic signaling through pulmonary GLP-1R remains undefined, and substantial interspecies differences further complicate interpretation⁷⁸. In the spleen, whether splenic Sctr participates in metabolic regulation or immune modulation is unresolved. Within the gut-brain-axis, endurance training did not significantly alter Glp1r expression, though the limited sample size constrains the ability to detect modest effects and precludes definitive conclusions.

Study limitations include lack of specific targeting of incretin receptors. As such, the absence of significant incretin receptor-tissue combinations does not necessarily imply a lack of exercise-incretin interactions. Further, receptor expression does not infer downstream signaling or physiological effects. In several cases, tissue peptide levels as detected by Luminex were elevated in the absence of local receptor induction, and vice versa. Incretin receptors were not identified in the mass spectronomy based proteomics platforms performed in the MoTrPAC dataset. Hormone concentrations do not distinguish between active and inactive forms, nor do they indicate cell-specific secretion or uptake. The analyses were performed in bulk tissue and thus it is not possible to distinguish in which individual cell types incretin or its receptors were located. Tissues were collected 48 hours after the last exercise bout to isolate training adaptations from acute responses, which would miss rapid post-exercise kinetics. Absence of effect at this timepoint does not preclude meaningful acute regulation. Future studies sampling at multiple post-exercise intervals will be needed to resolve whether the receptor-level adaptations identified here represent the residue of repeated acute responses or a distinct chronic remodeling program.

In conclusion, we provide a comprehensive assessment of incretin receptor expression and exercise-induced regulation across 17 tissues in male and female rats. Given the widespread clinical use of GLP-1 and dual GLP-1/GIP receptor agonists, and the routine recommendation to incorporate exercise alongside pharmacotherapy, understanding how training influences incretin receptor expression across tissues carries direct translational relevance. These data raise the possibility that exercise alters tissue sensitivity to incretin-based therapies through receptor-level regulation, potentially contributing to the metabolic benefits of combined lifestyle and pharmacological interventions. Future studies examining cell type-specific receptor dynamics and downstream signaling will be critical to elucidate the mechanistic interplay between exercise and incretin biology.