Intestinal adaptation to cold-induced metabolic demand and feeding requires GLP-1R and GLP-2R signalling

Abstract

Chronic cold exposure in mice increases metabolic demand and food intake; the gut correspondingly expands its absorptive surface area. Gut enteroendocrine cells produce peptide hormones including glucagon-like peptide-1 (GLP-1), GLP-2, and glucose-dependent insulinotropic polypeptide (GIP) in response to a meal to facilitate nutrient absorption and post-prandial metabolism. The requirement of GLP-1, GLP-2, and GIP receptor signaling for small intestinal adaptations to chronic cold stress has not been investigated. Here, we show that male and female wild-type, double incretin receptor knockout (Glp1r^-/-Gipr^-/-; DIRKO), and glucagon-like peptide double receptor knockout (Glp1r^-/-Glp2r^-/-; GLPDRKO) mice consume significantly more food over five weeks in cold (6⁰C) compared to thermoneutral (27 ⁰C; TN) conditions. Jejunal circumference, villi length, and crypt depth are significantly greater with cold-stress in WT and DIRKO mice, but not GLPDRKO mice, compared to TN controls. We show that the GLP-2R is required for jejunal villi length expansion upon cold stress despite significantly elevated plasma active GLP-1 levels. In line with this, GLPDRKO mice fail to gain body weight over the five-week experiment compared to WT controls. Therefore, while GLP-2R action is required for cold stress-induced jejunal villi lengthening, this adaptation is dispensable for body weight gain in the presence of GLP-1R signaling.

Chronic cold stress in Glp1r^-/-Gipr^-/-, Glp1r^-/-Glp2r^-/-, and Glp2r^-/- mice revealed distinct and overlapping roles for GLP-1R and GLP-2R in the expansion of the intestinal absorptive surface area expansion in response to chronic cold stress.

Introduction

As temperatures fall below the thermal neutral zone, body temperature is initially maintained by shivering thermogenesis via skeletal muscle and chronically by non-shivering thermogenesis in brown adipose tissue (BAT)^1–3. Specifically, animals chronically housed in the cold balance this heat loss by increasing their energy expenditure. The inner mitochondrial membrane of BAT contains a high density of uncoupling protein 1 (UCP1), which translocates protons to the mitochondrial matrix, uncoupling substrate oxidation from ATP production and generating heat. Chronic cold exposure in rodents increases BAT mass, mitochondria, and UCP1 content, leading to a browning of white adipose tissue⁴. To meet this increased energy demand, rodents housed in the cold increase food intake proportional to the temperature drop^5,6. Interestingly, small intestinal mass also increases proportionally to the temperature drop⁶. Intestinal circumference and villus length have also been shown to be greater with chronic cold stress, where markers of proliferation increase and markers of apoptosis at the villus tip decrease⁵; evidence of a directional role for the gut microbiota has been implicated in this form of gut plasticity^5,7. However, the hormonal signals underlying the induction of these circuits to mediate this plasticity are not fully elucidated.

Enteroendocrine cells in the gut epithelium sense meal ingestion and secrete gut hormones to maximize nutrient absorption^8,9. These hormones include glucagon-like peptide 1 (GLP-1), GLP-2, and glucose-dependent insulinotropic polypeptide (GIP). GLP-1 and GLP-2 are sister peptides, produced through cleavage of the same preproglucagon polypeptide in intestinal L cells by prohormone convertase 1/3 (PC1/3)^10,11. PC1/3 also cleaves the proGIP precursor in intestinal K cells, yielding GIP¹². GLP-1 and GIP both amplify glucose-stimulated insulin secretion from pancreatic islets¹³, whereas both GLP-1 and GLP-2 promote gut surface area expansion^14,15. Drugs modeled to activate GLP-2 receptor (GLP-2R) signaling are used to manage short bowel disease and are effective at removing or decreasing parenteral nutrition dependence in ~65% of patients¹⁶. However, despite their significant quality of life improvements, they can be accompanied by significant gastrointestinal side effects, including nausea, vomiting, abdominal pain, diarrhea, fatigue, and weight loss¹⁶. GLP-2R signaling increases villus length by promoting intestinal cell proliferation and inhibiting apoptosis at the villus tip¹⁷. By contrast, pharmacological GLP-1 receptor (GLP-1R) signaling achieved through activation by exendin-4 treatment has been shown to increase crypt number in the proximal intestine and colon, leading to increased intestinal circumference and length^15,18. Interestingly, in mice, genetic elimination of the glucagon receptor (Gcgr^−/−) leads to a ~3 to 30-fold increase in physiological circulating GLP-1 and GLP-2 levels^19–21, as well as increased small and large intestinal length¹⁵, indicating that increasing endogenous levels of GLP-1 and GLP-2 can also lead to an expansion of the gut surface area. Therefore, using extreme models of physiology to understand the role of these receptor signaling pathways may uncover important mediators in gut plasticity.

GLP-1R, GLP-2R, and GIPR signaling play clear roles in regulating metabolism, intestinal growth, and inflammation at pharmacological levels. Within minutes, however, the serine protease dipeptidyl peptidase 4 (DPP4) cleaves two amino acids from endogenous GLP-1, GLP-2, and GIP, rendering them inactive^14,22. Both GLP-2 and GIP have circulating half-lives of 7 min^17,23,24, whereas GLP-1 has a circulating half-life of 1–2 min²⁵. Many cells within the gastrointestinal tract, and therefore in proximity to the enteroendocrine cells that produce them, express GLP-1R, GLP-2R, and GIPR. Specifically, duodenal Brunner’s glands, stomach gland parietal cells²⁶, the myenteric nerve plexus, and crypt Paneth cells express Glp1r mRNA^27,28. Glp1r mRNA is also detected in both Tαβ and Tγδ subsets of intraepithelial lymphocytes^29–31, which have been shown to control T cell-induced inflammation^31,32. Vagal afferents, enteric neurons, myenteric plexus nerve fibrils, enteroendocrine cells, and subepithelial myofibroblasts express Glp2r mRNA^33,34. Gut CD146+ stromal cells express Gipr mRNA and play a role in attenuating gut inflammation³⁵. Obesity³⁶ and age³⁷ are associated with enhanced expression and activity of the pro-inflammatory DPP4. The impact of chronic cold stress on DPP4, a gene containing a glucocorticoid response element in its promoter³⁸, expression, and activity levels has not been elucidated.

We hypothesized that GLP-1R, GLP-2R, and GIPR signaling mediate the gut surface area expansion in response to chronic cold stress. Moreover, we investigated the roles of these gut hormone receptor signaling pathways in the adaptations of post-prandial metabolism to chronic cold stress using double incretin receptor knockout (Glp1r^−/−Gipr^−/−; DIRKO), and glucagon-like peptide double receptor knockout (Glp1r^−/−Glp2r^−/−; GLPDRKO) compared to wild-type (WT) controls. Jejunal circumference, villi length, and crypt depth were significantly greater with cold-stress in WT and DIRKO mice, but not GLPDRKO mice, compared to TN controls. Studies in Glp2r^−/− mice revealed that GLP-2R was required for jejunal villi length expansion upon cold stress. In line with this, GLPDRKO, but not Glp2r^−/− mice, fail to gain body weight over the 5-week experiment compared to WT controls. Therefore, while GLP-2R action is required for cold stress-induced jejunal villi lengthening, this adaptation is dispensable for body weight gain in the presence of GLP-1R signaling.

Methods

Animal housing

Male and female wild-type C57BL/6J mice were purchased (The Jackson Laboratory, Bar Harbor, USA) and bred in-house. Double incretin receptor knockout (DIRKO) mice³⁹, and glucagon-like peptide double receptor knockout (GLPRDKO) mice were generously provided by Dr. Daniel Drucker (University of Toronto Lunenfeld-Tanenbaum Research Institute, Toronto, Canada). Glp1r^−/− mice⁴⁰ were crossed with Glp2r^−/− mice to generate Glp1r^+/−Glp2r^+/−, then bred to generate Glp1r^−/−Glp2r^−/− and wild-type controls. At 8–12 or 28–35 weeks of age, mice were weighed and assigned to groups such that the mean starting body weights were similar in the two temperature housing groups within a genotype. Mice were individually housed with nesting materials in temperature-controlled chambers set to 6 °C (cold) or 27 °C (thermoneutral; TN), under a 12-h light/dark cycle for 5 weeks at the University of Ottawa Heart Institute. Mice were fed a standard laboratory diet (22% kcal from fat, 55% kcal from carbohydrates, 23% kcal from protein; Envigo 2019). An estimate of average food intake was measured in grams by the change in food mass from the week prior and divided by the number of days. Fat and lean mass were determined by EchoMRI (EchoMRI-100 machine, Echo Medical Systems, Houston, USA) scanning of individual awake mice. All experiments undertaken were approved by the University of Ottawa Animal Care Committee (AUP 2909 and 4379). We have complied with all relevant ethical regulations for animal use.

Metabolic tests

Lipid tolerance test: After 2 weeks of temperature housing, mice were fasted for 5 h and gavaged with 200 µL of olive oil. To minimize potential order effects, mice housed at cold and TN temperatures were gavaged in an alternating sequence, such that each consecutive animal represented a different housing condition. Blood from the tail vein was collected from baseline (prior to gavage), 10 min, 1, 2, and 3 h after gavage in ethylenediaminetetraacetic acid (EDTA)-coated capillary tubes for total GLP-1 and GIP measurements at baseline and 15 min post-gavage. Plasma triglycerides were measured using the colorimetric assay with Infinity Triglyceride Liquid Stable Reagent. Intestinal transit test: After 4 weeks of temperature housing, mice were fasted for 2 h and administered a 300 μL oral gavage of 6% w/v carmine red solution in 0.5% w/v methyl cellulose, and food was available ad libitum for the remainder of the experiment⁴¹. Intestinal transit time was defined as the interval from gavage to the first red fecal pellet.

Blood and tissue collection

Blood was taken at the start of the light cycle in the ad libitum state at weeks 1, 2 and 4 in the Glp2r^−/− mice and WT controls, where 10% TED (5000 kIU/mL Trasylol, 1.2 mg/mL ethylenediaminetetraacetic acid (EDTA), 0.1 nmol/L Diprotin A) (vol/vol) was added for active plasma GLP-1 (K1503OD-2, MSD) measurements. After 5 weeks of temperature housing, mice were euthanized by CO₂ inhalation and cervical dislocation. To minimize potential order effects, tissue was collected from mice housed at cold and TN temperatures in an alternating sequence, such that nearly all consecutive animals represented a different housing condition. Blood was taken at endpoint via cardiac puncture in EDTA-coated syringes. Blood was centrifuged at 4 °C and 12,000 rpm for 10 min. Plasma was stored at −80 °C until analysis. Total plasma GLP-1 levels (K15171C-2, MSD, total plasma GIP levels (81527, Crystal Chem), and leptin levels (90030, Crystal Chem) were measured according to the manufacturer’s instructions. Plasma DPP4 activity was measured as previously described³⁶. Lipids were extracted from ~100 mg of liver using the Folch method with cholesteryl oleate [Cholesteryl-1,2-³H(N)] (Perkin Elmer) for recovery calculations, as described previously^42,43. Liver triglyceride levels were quantified using the Infinity triglyceride liquid stable reagent, and liver total cholesterol levels were quantified using the Infinity cholesterol liquid stable reagent. The small intestine was divided into duodenum, proximal jejunum, distal jejunum, and ileum (1:1:1:1 ratio) and flushed with ice cold phosphate-buffered saline (PBS). Tissue segments used for histology were fixed in 4% paraformaldehyde for 24 hours at room temperature prior to paraffin embedding. Tissues were flash-frozen and stored at −80 °C until analysis. Fecal short-chain fatty acid (SCFA) analyses were performed as previously described and expressed as a percent of total SCFA (acetate, propionate, and butyrate)⁴⁴.

Histology and immunostaining

Five-micron sections of jejunal samples were stained with hematoxylin and eosin. Photomicrographs were obtained with a Leica Aperio Versa slide scanner. Circumference (2–3/mouse), villi length (15–30/mouse), and crypt depth (20–35/mouse) measurements were performed using ImageScope software. One sample did not have full villi lengths and was therefore excluded from the Fig. 1 data set, a criterion established a priori. Four samples did not have full villi lengths and were therefore excluded from the Supplementary Fig. 2 data set. Five-micron sections of jejunal samples were used for immunostaining. Sections were deparaffinized with an autostainer, and antigen retrieval was performed with a Decloaking chamber (BioCare Medical) in 10 mM citric acid buffer (pH 6.0). Slides were cooled at room temperature for 10 min followed by five washes in distilled water. Sections were blocked with 10% donkey serum for 30 min at room temperature, followed by incubation with primary antibody against Ki67 (Abcam ab16667; 1:200) or PBS as a negative control at 4 °C overnight. Sections were then washed twice with PBS for 5 min each on a shaker, followed by incubation with a secondary antibody (donkey anti-rabbit, Invitrogen A32794, 1:500) for 30 min at room temperature. Sections were then washed twice with PBS for 5 min on a shaker and mounted for microscopy. Photomicrographs were obtained with a Leica Aperio Versa slide scanner and analyzed by a custom Cell Profiler pipeline, where the number of Ki67+ nuclei was expressed as a percentage of the total nuclei number of the tissue section.

Fig. 1. Small intestinal adaptations to increased food intake with cold stress require GLP-1R and GLP-2R signalling.

Male wild-type (WT) (TN, n = 10; C, n = 11), Glp1r^−/−Gipr^−/− (DIRKO) (TN, n = 4; C, n = 4), and Glp1r^−/−Glp2r^−/− (GLPDRKO) (TN, n = 10; C, n = 8) and female WT (TN, n = 9; C, n = 8), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 9; C, n = 10) mice were housed at thermoneutrality (TN; 27 °C) or cold (6 °C) temperatures for 5 weeks. Body weight gain over time in a male and b female mice, where stars indicate statistical differences for that time point; GLPDRKO vs. WT controls (gray) and GLPDRKO vs. DIRKO (blue). Average daily food intake estimates in c male and d female mice. Mice were fasted for 5 h prior to endpoint. Jejunal RNA was isolated from male WT (TN, n = 7; C, n = 7), DIRKO (TN, n = 4; C, n = 4), and GLPDRKO (TN, n = 6; C, n = 5), and female WT (TN, n = 6; C, n = 5), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 5; C, n = 6). Jejunal Glp1r mRNA levels in e male and f female mice. Jejunal Glp2r mRNA levels in g male and h female mice. Jejunal Gipr mRNA levels in i male and j female mice. Small intestinal tissue was collected for morphological measurements in male wild-type (WT) (TN, n = 10; C, n = 10–11), DIRKO (TN, n = 4; C, n = 4), and GLPDRKO (TN, n = 10; C, n = 8), and female WT (TN, n = 9; C, n = 8), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 9; C, n = 10). Small intestinal length in k male and l female mice. Small intestinal weight in m male WT (TN, n = 7; C, n = 10), DIRKO (TN, n = 4; C, n = 4), and GLPDRKO (TN, n = 8; C, n = 6) and n female mice WT (TN, n = 6; C, n = 6), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 7; C, n = 7). Jejunal circumference in o male and p female mice with q representative images. Jejunal villi length in r male and s female mice. Representative images of villi t were cropped and rotated to align the villi in the same orientation. Uncropped images are available in the Data Supplement. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype. For qPCR analyses, “n.d.” denotes not detected. Data are shown as mean ± standard error of the mean (SEM).

Jejunal mitochondrial DNA quantification

Jejunal DNA was isolated as previously described⁴⁵. Briefly, jejunum (10 mg) was homogenized in DNA lysis buffer (pH 8.0) [100 mM Tris–HCl (pH 8.0), 5 mM EDTA, 0.2% SDS, 200 mM NaCl] and incubated with proteinase K solution (Sigma) at 55 °C overnight. RNAse A was added, and samples were incubated at 37 °C for 30 min. DNA was pelleted using phenol/chloroform/isoamyl alcohol (25:24:1) (PCIAA) (Sigma) and isopropanol, and pellets were washed with 70% ethanol and dissolved in UltraPure Distilled Water (Invitrogen). Mitochondrial to nuclear DNA ratios were measured with quantitative real-time PCR (CFX96, Bio-Rad) with mitochondrial cytochrome c oxidase (mt-Co1) and mitochondrial NADH dehydrogenase 1 (mt-Nd1) and nuclear DNA hexokinase 2 (Hk2) and beta-2 microglobulin (B2m) primers (Table 1). Ratios were calculated using the 2^−ΔΔCT method.

Table 1.

List of primers

Gene	Sequence or assay ID
MTCOI (mtDNA)	M_MTCOI_FWD: 5’-TGC TAG CCG CAG GCA TTA C-3’ M_MTCOI_REV: 5’-GGG TGC CCA AAG AAT CAG AAC-3’
ND1(mtDNA)	ND1 FWD: 5′-CTAGCAGAAACAAACCGGGC-3′ ND1 REV: 5′-CCGGCTGCGTATTCTACGTT-3′
HK2 (genomic DNA)	HK2 FWD: 5′-GCCAGCCTCTCCTGATTTTAGTGT-3′ HK2 REV: 5′-GGGAACACAAAAGACCTCTTCTGG-3′
Glp1r	Mm00445292_m1
Glp2r	Mm01329477_m1
Gipr	Mm01316349_g1
Fgf7	Mm00433291_m1
Igf1	Mm00439560_m1
Igf1r	Mm00802831_m1
Actb	Mm00607939_s1
Dpp4	Mm01329189_m1
Rps9	Mm00850060_s1

High-resolution respirometry

Mitochondrial respiration was assessed in fresh jejunum tissue from 3 to 4 mice per group, where 3–4 samples were processed per day from mice housed in cold and TN conditions using an Oxygraph-2k system (OROBOROS Instruments, Innsbruck, Austria). Jejunal tissue was isolated, first cleaned with PBS, then immediately flushed and stored in cold biopsy preservation buffer (BIOPS) (pH 7.1 (5.77 mM Na²⁺ATP, 10 mM Ca-EGTA buffer, 0.1 µM free calcium, 6.56 mM MgCl₂⋅6H₂O, 20 mM taurine, 20 mM imidazole, 60 mM K-lactobionate, 15 mM phosphocreatine, 0.5 mM DTT, 50 mM MES). Tissue was gently teased apart in BIOPS with forceps on ice, followed by permeabilization with saponin (50 µg/mL) for 30 min on a rocker at 4 °C. Samples were transferred to Miro5 buffer (pH 7.1, 110 mM sucrose, 60 mM potassium lactobionate, 20 mM taurine, 10 mM monobasic potassium phosphate, 3 mM magnesium chloride, 20 mM HEPES, 0.5 mM EGTA, 1 mg/mL BSA) and washed for 3 × 10 min on a rocker at 4 °C. Samples were quickly blotted on filter paper, weighed, and placed in 2 mL Miro5 in the chambers. All measurements were collected in duplicate at 37 °C with a constant stirring speed of 750 rpm. Oxygen concentration in the chambers was maintained at 400 μM throughout the experiment. The assay involved the sequential addition of 2 mM malate, 5 mM glutamate, and 10 mM pyruvate (complex I leak respiration), 5 mM ADP (complex I oxidative phosphorylation), 10 mM succinate (complex I + II oxidative phosphorylation), 2.5 µM oligomycin (complex I + II leak respiration), 0.5 µM FCCP (maximal respiration), and 2.5 µM antimycin A (non-mitochondrial respiration). All values were corrected for non-mitochondrial respiration. Data from one sample was removed due to mechanical over-permeabilization, a criterion established a priori. DatLab 7.4 (OROBOROS Instruments, Innsbruck, Austria) was used for analysis.

RNA isolation and gene expression

Jejunal RNA was isolated with TRIzol Reagent (Ambion) as per the manufacturer’s protocol and reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Fisher Scientific 43-688-14). Quantitative Real Time Polymerase chain reaction (qRT-PCR) was performed with Applied Biosystems TaqMan Gene Expression Master Mix (4369016) on the QuantStudio 5. mRNA abundance was calculated using the standard curve qRT-PCR method and normalized to the housekeeping gene (Ribosomal protein S9; Rps9), denoted as absolute mRNA (Table 1). Confirmation of genotype by qPCR was performed in all mice.

Statistical and reproducibility

Statistical analysis was performed using GraphPad Prism (version 10.2.2). Data are shown as mean ± standard error of the mean (SEM). Each data point indicates a biological replicate within a group. Two-way ANOVA with post-hoc Tukey’s multiple comparisons correction was performed on analyses with three genotypes and two housing temperatures. Two-way ANOVA with repeated measures and post-hoc Tukey’s multiple comparisons correction was performed on analyses with three genotypes and housing temperature over time. Two-way ANOVA was performed on analyses with two genotypes and two housing temperatures with an uncorrected Fisher’s LSD test. All analyses are based on 95% confidence intervals. Assumptions for normality were assessed using the D’Agostino-Pearson omnibus, Anderson-Darling, Shapiro–Wilk, and Kolmogorov–Smirnov statistical tests or visualization of the quantile-quantile (Q-Q) plot. Four cohorts of male 8–12-week-old wild-type (WT) (TN, n = 10; C, n = 11), Glp1r^−/−Gipr^−/− (DIRKO) (TN, n = 4; C, n = 4), and Glp1r^−/−Glp2r^−/− (GLPDRKO) (TN, n = 10; C, n = 8) and female WT (TN, n = 9; C, n = 8), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 9; C, n = 10) mice were housed at TN (27 °C) or cold (6 °C) temperatures for 5 weeks. Two cohorts of male 25–30-week-old WT (TN, n = 6; C, n = 6), DIRKO (TN, n = 6; C, n = 6), and GLPDRKO (TN, n = 4; C, n = 5) mice were housed at TN (27 °C) or cold (6 °C) temperatures for 5 weeks. Two cohorts of male WT (TN, n = 4; C, n = 5) and Glp2r^−/− (TN, n = 6; C, n = 6) mice were housed at TN (27 °C) or cold (6 °C) for 5 weeks. A total number of 146 mice were used in this study.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

Small intestinal adaptations to increased food intake with cold stress require GLP-1R and GLP-2R signaling

To determine the impact of GLP-1R, GLP-2R, or GIPR on gut surface area expansion with cold stress, we used double receptor knockout mice for Glp1r^−/−Gipr^−/− (DIRKO) and Glp1r^−/−Glp2r^−/− (GLPDRKO) to account for overlapping or compensatory roles in the intestinotrophic action of these gut hormones^15,35,46. Male and female DIRKO, GLPDRKO, and WT controls were housed for 5 weeks at TN or cold temperatures. Cold stress did not significantly impact body weight within any genotype for either sex (Fig. 1a, b). GLPDRKO mice, however, weighed significantly less compared to WT and DIRKO mice (Fig. 1a, b). Continuing previous reports^6,47, cold stress increased average daily food consumption in all genotypes and both sexes (Fig. 1c, d). Jejunal Glp1r mRNA levels were significantly greater in cold-stressed male and female WT mice compared to TN controls (Fig. 1e, f). Jejunal Glp1r mRNA levels were not detected in male or female DIRKO or GLPDRKO mice (Fig. 1e, f). Jejunal Glp2r mRNA levels were significantly greater in cold-stressed male and female WT mice compared to TN controls, and ablated in GLPDRKO mice (Fig. 1g, h). Jejunal Gipr mRNA levels were unchanged in cold-stressed male and female WT mice, and not detected in DIRKO mice (Fig. 1i, j). Fasting plasma total GLP-1 levels were greater with cold stress in male (t-test, P = 0.0633) and female (t-test, P = 0.0320) WT mice and GLPDRKO mice (significant), but not in DIRKO mice (Supplementary Fig. 1a, b), consistent with the engagement of adrenergic stimulation of GLP-1 secretion^48,49. Additionally, the greater mRNA expression of jejunal Glp1r and Glp2r with chronic cold stress in WT mice underlined a biological response from one or both receptors.

Consistent with previous reports^5,6,50, small intestinal length and weight expanded with the increased metabolic demand and food intake in response to cold stress in male and female WT mice (Fig. 1k–n). Small intestinal length failed to increase in male DIRKO and GLPDRKO with cold stress (Fig. 1k). In female mice, small intestinal length was greater with cold stress in all genotypes (Fig. 1l). While small intestinal weight was significantly greater in both DIRKO and GLPDRKO mice with cold stress, this increase was significantly blunted in GLPDRKO mice, independent of sex (Fig. 1m, n). Consistent with previous reports^5,50, jejunal circumference and villi length expanded with the increased metabolic demand and food intake in response to cold stress in male and female WT mice (Fig. 1o–t). Unlike male DIRKO mice, GLPDRKO mice failed to increase jejunal circumference with cold stress, suggesting an impact from the loss of GLP-2R signaling (Fig. 1o, q). In female mice, while jejunal circumference was significantly greater with cold stress in all genotypes, the extent of this increase was significantly ablated in GLPDRKO mice (Fig. 1p). Unlike DIRKO mice, both male and female GLPDRKO mice failed to increase jejunal villi length with cold stress, indicating an impact from the loss of GLP-2R signaling (Fig. 1r–t). Additionally, the impact of the loss of GLP-2R signaling on the intestinal response to increased food intake with cold stress was maintained in the duodenum and jejunum of male mice aged to 25 weeks, demonstrating conservation of this response with age (Supplementary Fig. 2a–d). Overall, these data point to a requirement of GLP-2R signaling to expand jejunal villi length and circumference, impacting small intestinal weight in response to the increased food intake upon chronic cold housing.

Jejunal increases in proliferation with cold stress require GLP-2R

Intestinal stem and progenitor cell proliferation within intestinal crypts drives cell migration along the villus⁵¹. Consistent with the expansion of intestinal surface area observed with the increased food intake that accompanies chronic cold stress, crypt depth was significantly greater with cold stress in WT mice (Fig. 2a, b). Unlike in DIRKO mice, jejunal crypt depth failed to expand in GLPDRKO mice, independent of sex (Fig. 2a, b). Curiously, duodenal crypt depth was significantly greater with cold stress in WT mice only, and jejunal crypt depths were not different with housing temperature in male mice aged to 25 weeks (Supplementary Fig. 2e, f), suggesting that increased crypt depth did not necessitate greater jejunal circumference or villi length at this age. To determine if the increased jejunal surface area was mediated by increased proliferation, the percentage of Ki67-positive nuclei was quantified in jejunal sections. The percentage of Ki67-positive nuclei localized to crypt cells was significantly greater in cold-stressed WT but not GLPDRKO mice, independent of sex (Fig. 2c–e). Overall, these results are in line with enhanced GLP-2R action, previously shown to increase villus length by promoting intestinal cell proliferation and inhibiting apoptosis at the villus tip¹⁷.

Fig. 2. Jejunal increases in proliferation with cold stress require GLP-2R.

Male and female WT and GLPDRKO mice were housed at TN (27 °C) or cold (6 °C) temperatures for 5 weeks. Small intestinal tissue was collected for morphological measurements in male wild-type (WT) (TN, n = 10; C, n = 11), DIRKO (TN, n = 4; C, n = 4), and GLPDRKO (TN, n = 10; C, n = 8), and female WT (TN, n = 9; C, n = 8), DIRKO (TN, n = 5; C, n = 4), and GLPDRKO (TN, n = 9; C, n = 10). Jejunal crypt depth in a male and b female mice. Jejunal sections from male WT (TN, n = 7; C, n = 7) and GLPDRKO (TN, n = 6; C, n = 5), and female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 5; C, n = 6) were immunostained for c Ki67 and quantified as a positive percentage area, normalized to total nuclear area in d male and e female mice. Jejunal RNA and mitochondrial DNA were extracted from male WT (TN, n = 7; C, n = 7) and GLPDRKO (TN, n = 6; C, n = 5), and female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 5; C, n = 6). Jejunal gene expression levels Actb in f male and g female mice. Jejunal mitochondrial DNA levels were measured in h male and i female mice. Jejunal high-resolution respirometry flux per mg tissue was measured in j male WT (TN, n = 3; C, n = 3) and GLPDRKO (TN, n = 4; C, n = 4), and k female WT (TN, n = 3; C, n = 3) and GLPDRKO (TN, n = 3; C, n = 4) mice. CI Complex I, CII Complex II, OXPHOS oxidative phosphorylation. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype. Data are shown as mean ± standard error of the mean (SEM).

We next investigated molecular signatures associated with the intestinal action of gut hormones that may accompany the changes in intestinal surface area observed with the increase in food intake that accompanied the chronic cold stress. Since the intestinotrophic effects of pharmacological GLP-1R and GLP-2R activation both require keratinocyte growth factor (KGF/Fgf7)^15,34 and pharmacological GLP-2 also requires insulin-like growth factor-1 (IGF) action to increase villus length and small intestinal weight⁵², we measured the mRNA expression of these target genes in response to cold stress. Jejunal Fgf7 mRNA levels were greater, but not significantly (P = 0.0637), in cold-stressed male WT mice compared to TN controls; however, this change was not conserved in female WT mice (Supplementary Fig. 3a, b). Cold stress did not impact jejunal Fgf7 mRNA levels in GLPDRKO mice, independent of sex (Supplementary Fig. 3a, b). Cold stress did not significantly increase jejunal Igf1 or IGF1 receptor (Igf1r) mRNA levels in WT mice in both sexes (Supplementary Fig. 3c-f). Intestinal beta-actin (Actb) is required for the maintenance of the intestinal epithelial barrier⁵³. Consistent with small intestinal surface area expansion, jejunal Actb mRNA levels were significantly greater with cold stress in male WT, but not in GLPDRKO mice or in female mice of either genotype (Fig. 2f, g). Gut DPP4 action includes the hydrolysis of dietary peptides⁵⁴. Male WT mice displayed significantly greater jejunal Dpp4 mRNA levels with cold stress compared to TN controls (Supplementary Fig. 3g). However, this change was not conserved in female WT or GLPDRKO mice (Supplementary Fig. 3g, h). Together, these data underscore the complexity of the molecular mechanisms driving intestinal surface area expansion in response to physiological increases in gut hormone action and food intake, where changes in mRNA signatures cannot be detected at the whole tissue level.

In intestinal organoids, mitochondrial copy number increases during differentiation, and mitochondrial electron transport chain activity is required for the initiation of crypt formation⁵⁵. In various species, cold stress can change mitochondrial gene expression and increase BAT mitochondrial DNA content, shape, and membrane permeability^4,56. Jejunal mitochondrial content, shown by the ratio of mitochondrial (mt-Co1) DNA levels to nuclear (Hk2) DNA levels, was not impacted by cold stress in WT mice (Fig. 2h, i; Supplementary Fig. 3i, j). In GLPDRKO mice, however, jejunal mitochondrial content was significantly lower with cold stress, independent of sex (Fig. 2h, i; Supplementary Fig. 3i, j). These data suggest that GLP-1R/2 R signaling is required to maintain jejunal mitochondrial levels. To investigate the impact of this lower mitochondrial content, jejunal oxygen consumption rates per milligram of tissue were determined by high-resolution respirometry. Jejunal oxygen consumption was not different in male cold-stressed mice, independent of genotype (Fig. 2j). Therefore, the lower jejunal mitochondrial content in male GLPDRKO mice did not significantly impact jejunal oxidative capacity. In female mice, oxygen consumption was significantly greater in the tissues of cold-stressed GLPDRKO mice compared to TN controls in response to malate, pyruvate, and glutamate, indicating an increased complex I leak respiration (Fig. 2k). Similarly, complex I-driven oxidative phosphorylation was significantly greater in the tissues of cold-stressed GLPDRKO mice compared to TN controls (Fig. 2k). The tissue response to succinate and oligomycin, targeting complex I and II oxidative phosphorylation and leak, respectively, was not different in the tissues of cold-stressed mice compared to controls, independent of genotype (Fig. 2k). These data suggest that a compensatory increase in jejunal oxidative capacity in female GLPDRKO mice is upstream to the proliferative response with cold stress.

Cold stress in male GLPDRKO mice does not significantly impact plasma leptin levels, body weight gain, or lipid tolerance but does significantly shorten intestinal transit time

Leptin-deficient mice (ob/ob), a genetic model of overeating, have greater small intestinal length, jejunal perimeter, and jejunal villi length⁵⁰. Consistent with the overeating phenotype and intestinal adaptations observed in cold-stressed mice, fasting plasma leptin levels were significantly lower with cold stress in WT mice compared to TN controls (Fig. 3a). The impact of cold stress to lower plasma leptin levels in GLPDRKO mice, however, was minimal and not significant (Fig. 3a). Chronic cold stress did not significantly impact body weight or lean mass gain from baseline to endpoint in WT mice, however, led to a significant decrease in fat mass consistent with lower plasma leptin levels (Fig. 3b–d). Overall, male GLPDRKO mice were significantly lighter than WT controls, independent of housing temperature (Fig. 3b). GLPDRKO mice experienced an increase in lean mass and maintained fat mass, leading to an overall neutral impact of cold stress on body weight gain (Fig. 3b–d). Liver mass was not different with housing temperature, independent of genotype, but fasting liver triglyceride levels were significantly lower in male cold-stressed mice, independent of genotype, compared to TN controls (Fig. 3e, f). Liver total cholesterol levels were not significantly impacted by housing temperature or genotype (Supplementary Fig. 4a).

Fig. 3. Cold stress in male GLPDRKO mice does not significantly impact plasma leptin levels, body weight gain, or lipid tolerance.

Male WT (TN, n = 10; C, n = 11) and GLPDRKO mice (TN, n = 10; C, n = 8) were housed at TN (27 °C) or cold (6 °C) temperatures for 5 weeks. Fasting plasma a leptin levels. Body b weight, c lean mass, and d fat mass were measured at baseline and endpoint. Liver e mass and f triglyceride levels in WT (TN, n = 7; C, n = 7) and GLPDRKO mice (TN, n = 6; C, n = 5). After 2 weeks of temperature housing, WT (TN, n = 7; C, n = 7) and GLPDRKO mice (TN, n = 6; C, n = 5) were fasted for 5 h and administered a dietary fat challenge of 200 μL of olive oil. Fasting plasma g triglyceride levels. Lipid tolerance test plasma h triglyceride levels with AUC (inset). The P value indicates statistical differences among genotypes at TN. Plasma i GLP-1 and j GIP levels at baseline (0) and 10 min after the olive oil gavage (10). Fasting plasma k DPP4 activity. After 4 weeks of temperature housing, WT (TN, n = 7; C, n = 7) and GLPDRKO mice (TN, n = 6; C, n = 5) were fasted for 2 h and administered an oral gavage of carmine red. Intestinal l transit time was measured from gavage to the first red fecal pellet. Short-chain fatty acids were extracted from the feces of male WT (TN, n = 10; C, n = 11) and GLPDRKO mice (TN, n = 10; C, n = 7) for m butyrate measurements expressed as a percentage of total short-chain fatty acids. Two-way ANOVA with repeated measures and a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype for both body weight and composition, as well as plasma GIP levels. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype. Data are shown as mean ± standard error of the mean (SEM).

Dietary fat and fiber significantly impact intestinal surface area and lipid handling^50,57,58. To assess the impact of the failed gut surface area expansion in male GLPDRKO mice on dietary fat handling, a lipid tolerance test was performed following a 5 h fast. Additionally, both GLP-1R and GLP-2R signaling modulate the rate at which dietary triglycerides appear in plasma, however, in opposing manners^59–61. Cold stress did not significantly impact fasting plasma triglyceride levels or lipid tolerance upon an acute dietary fat challenge, independent of genotype (Fig. 3g, h). However, TN-housed GLPDRKO mice displayed significantly worse lipid tolerance compared to WT controls, phenocopying the Glp1r^−/− mouse⁵⁹ (Fig. 3h). Plasma total GLP-1 levels 10 min post-oil gavage increased significantly from baseline in cold-stressed WT mice (Fig. 3i). Plasma total GIP levels significantly increased from baseline with the oil gavage in WT mice, independent of housing temperature, and cold-stressed male GLPDRKO mice (Fig. 3j). Fasting plasma DPP4 activity levels were similar in cold-stressed male mice compared to TN controls (Fig. 3k). Together these data indicate that while the loss of GLP-2R signaling was required for gut surface area expansion with cold stress, this did not significantly impact lipid tolerance.

To investigate the impact of the failed surface area expansion in GLPDRKO mice, an intestinal transit time test was performed. Cold stress significantly shortened intestinal transit time to less than 200 min, independent of genotype (Fig. 3l). The magnitude of transit time shortening from TN controls, however, was significantly greater in cold-stressed GLPDRKO mice (Fig. 3l). Consistent with greater food consumption, the percent abundance of fecal butyrate levels, primarily produced by gut bacteria⁶², was significantly greater in cold-stressed male mice, independent of genotype (Fig. 3m). Correspondingly, the percent abundance of fecal acetate levels was significantly lower with cold stress, independent of genotype (Supplementary Fig. 4b). The percent abundance of fecal propionate levels was unchanged with housing temperature and genotype (Supplementary Fig. 4c).

Cold stress in female GLPDRKO mice does not significantly impact body weight gain or lipid tolerance but does significantly shorten intestinal transit time

In female mice, plasma leptin levels were significantly lower with cold stress independent of genotype (Fig. 4a). Chronic cold stress did not significantly impact body weight or lean mass gain from baseline in female WT mice; however, it led to a significant decrease in fat mass, consistent with lower plasma leptin levels (Fig. 4b–d). Overall, female GLPDRKO mice were significantly lighter than WT controls, independent of housing temperature (Fig. 4b). Female GLPDRKO mice experienced a significant increase in lean mass and maintained fat mass, leading to an overall neutral impact of cold stress on body weight gain, independent of housing temperature (Fig. 4b–d). Liver mass was not different with housing temperature, independent of genotype, and liver triglyceride levels were only lower (trend; P = 0.0683) in cold-stressed WT, but not in GLPDRKO mice, compared to TN controls (Fig. 4e, f). Liver total cholesterol levels were not different in female mice, independent of genotype or housing temperature (Supplementary Fig. 4d).

Fig. 4. Cold stress in female GLPDRKO mice does not significantly impact plasma leptin levels, body weight gain, or lipid tolerance and intestinal transit time.

Female WT (TN, n = 9; C, n = 8) and GLPDRKO (TN, n = 9; C, n = 10) mice were housed at TN (27 °C) or cold (6 °C) temperatures for 5 weeks. Fasting plasma a leptin levels. Body b weight, c lean mass, and d fat mass were measured at baseline and endpoint. Liver e mass and f triglyceride levels from female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 5; C, n = 6). After 2 weeks of temperature housing, female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 5; C, n = 6) mice were fasted for 5 h and administered a dietary fat challenge of 200 μL of olive oil. Fasting plasma g triglyceride levels. Lipid tolerance test plasma h triglyceride levels with AUC (inset). Plasma i GLP-1 and j GIP levels at baseline (0) and 10 min after the olive oil gavage (10). Fasting plasma k DPP4 activity. After 4 weeks of temperature housing, female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 5; C, n = 6) mice were fasted for 2 h and administered an oral gavage of carmine red. Intestinal l transit time was measured from gavage to the first red fecal pellet. Short-chain fatty acids were extracted from the feces of female WT (TN, n = 6; C, n = 5) and GLPDRKO (TN, n = 6; C, n = 6) for (m) butyrate measurements expressed as a percentage of total short-chain fatty acids. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature, genotype, and time for both body weight and composition, as well as plasma GIP levels. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype. Data are shown as mean ± standard error of the mean (SEM).

Fasting plasma triglyceride levels were significantly lower in female WT, but not GLPDRKO cold-stressed mice compared to TN controls (Fig. 4g). Female WT mice displayed significantly impaired lipid tolerance with cold stress compared to TN controls (Fig. 4h). In female GLPDRKO mice, however, this difference with cold stress was lost (Fig. 4h). Plasma total GLP-1 levels in females were significantly greater in cold-stressed WT mice in response to the oil gavage compared to baseline (Fig. 4i). Plasma total GIP levels increased significantly in response to the oil gavage in the female cold-stressed WT and GLPDRKO mice and to levels significantly greater than TN controls (Fig. 4j). Plasma DPP4 activity levels were significantly greater in cold-stressed female mice compared to TN controls (Fig. 4k), independent of genotype, suggesting that the stress response element³⁸ for Dpp4 in female mice was more sensitive than in males to cold stress.

Cold stress significantly shortened intestinal transit time to less than 200 minutes in both WT and GLPDRKO female mice (Fig. 4l). The magnitude of this shortened transit time from TN controls was significantly greater in cold-stressed GLPDRKO mice (Fig. 4l). The percent abundance of fecal butyrate levels was significantly greater in both cold-stressed female WT and GLPDRKO mice compared to TN controls (Fig. 4m). Correspondingly, the percent abundance of fecal acetate was significantly lower with cold stress, independent of genotype (Supplementary Fig. 4e). The percent abundance of fecal propionate levels was unchanged by housing temperature and genotype (Supplementary Fig. 4f). Together, these data highlight the impact of increased food intake on transit time and SCFA production.

GLP-2R signaling is indispensable for jejunal villi length expansion in response to the increased food intake that accompanies chronic cold temperature housing

The requirement of GLP-2R signaling to expand jejunal villi length and circumference in response to increased food intake with chronic cold stress was observed when comparing morphological results in DIRKO and GLPDRKO mice compared to WT controls. As such, the impact of the loss of GLP-2R action was only assessed in the absence of GLP-1R signaling. To confirm the requirement of GLP-2R action and unveil the potential compensatory role of GLP-1R action in the expansion of intestinal surface area, male Glp2r^−/− mice and WT controls were housed at TN (27 °C) or cold (6 °C) for 5 weeks. Daily body weight measurements revealed no significant impact of housing temperature over time, independent of genotype (Fig. 5a). As expected, average daily food intake was significantly greater with cold stress, independent of genotype (Fig. 5b). Jejunal Glp2r mRNA levels were undetectable in Glp2r^−/− mice (Supplementary Fig. 4g). Jejunal Glp1r mRNA levels were unchanged in Glp2r^−/− mice compared to WT controls (Supplementary Fig. 4h). Plasma active GLP-1 levels obtained in the ad libitum fed state were higher, but not significantly, in male WT mice (Fig. 5c). Loss of GLP-2R signaling, however, led to significantly greater plasma active GLP-1 levels at week 1, 2, and 4 of cold temperature housing (Fig. 5d). Loss of GLP-2R signaling did not impact the increase in small intestinal length or weight observed with cold stress (Fig. 5e, f). Loss of GLP-2R action did not impact jejunal circumference or crypt depth (Supplementary Data 1), however, it significantly blunted the increase in jejunal villi length observed with cold stress compared to WT controls (Fig. 5g, h). Loss of GLP-2R action did not significantly impact the effect of cold stress on body weight, lean mass, or fat mass (Fig. 5i–k). Taken together, these data revealed that GLP-2R signaling alone was indispensable for jejunal villi length expansion in response to the increased metabolic demand and food intake that accompany chronic cold temperature housing. Moreover, compensation for the loss of GLP-2R action, including by GLP-1R signaling, maintained expansion of intestinal length, weight, and preserved body weight gain in response to chronic cold stress.

Fig. 5. GLP-2R signaling is indispensable for jejunal villi length expansion in response to the increased food intake that accompanies chronic cold temperature housing.

Male WT (TN, n = 4; C, n = 5) and Glp2r^−/− (TN, n = 6; C, n = 6) mice were housed at TN (27 °C) or cold (6 °C) for 5 weeks. a Daily body weight gain over time. b Average daily food intake estimates. Blood was taken at the start of the light cycle for ad libitum plasma active GLP-1 levels in c WT and d Glp2r^−/− mice. Mice were fasted for 5 h prior to endpoint. Small intestinal e length f weight. Jejunal g villi length and h representative images of villi were cropped and rotated to align the villi in the same orientation. Uncropped images are available in the Data Supplement. Body i weight, j lean mass, and k fat mass were measured at baseline and endpoint. Two-way ANOVA with a Tukey test correction for multiple comparisons was used to determine statistical significance between housing temperature and genotype. Data are shown as mean ± standard error of the mean (SEM).

Discussion

Previous studies have shown that housing mice at cold temperatures significantly increases food intake, intestinal length and mass, as well as absorptive surface area^5–7. In our study, housing mice at 6 °C also led to a doubling of food intake, which was accompanied by increases in small intestinal length, weight, jejunal circumference, jejunal villi length, and jejunal crypt depth. We tested the requirement of gut hormone receptors GLP-1R, GIPR, and GLP-2R using double receptor knockout mice to account for overlapping and compensatory roles of gut hormone action. While the intestinal changes observed with cold stress were maintained in DIRKO mice, GLPDRKO mice failed to increase jejunal villi length and crypt depth, indicating that, in the context of GLP-1R deficiency, GLP-2R action was required for this adaptation. Moreover, GLPDRKO mice weighed less than WT controls and did not gain body weight from baseline with cold stress. Studies in Glp2r^−/− mice confirmed the requirement for jejunal villi length expansion; however, body weight gain was not impacted by the loss of GLP-2R action, highlighting the compensatory role for GLP-1R action and the extreme adaptability of the small intestine.

Treatment of male mice with GLP-1R agonist liraglutide, and female mice with exendin-4, increases small intestinal length and weight¹⁵. In our study, the requirement of endogenous GLP-1R signaling to increase small intestinal length was only observed cold-stressed male, but not female mice, suggesting a sex difference in the identity or action of compensatory intestinotrophic signaling molecules. Cold stress still led to a significant increase in small intestinal weight in both male and female GLPDRKO mice, but this increase was significantly attenuated compared to that in cold-stressed WT and DIRKO mice. The intestinotrophic effects of GLP-2 in the colon require KGF produced by gut GLP-2R+ myofibroblasts, as shown in Fgf7^−/− mice³⁴. Pharmacological GLP-2 also increases expression and secretion of insulin-like growth factor-1 (IGF-1) from GLP-2R+ myofibroblasts⁶³. Similarly, physiological increases in GLP-2 with refeeding increase villus length and crypt-cell proliferation in an epithelial-IGF-dependent manner⁵². In our study, despite the increased abundance of Ki67+ cells and intestinal surface area in cold-stressed WT mice, we only observed a small trend for increased jejunal Fgf7 mRNA levels in male WT mice, and no differences in jejunal Igf1 or Igf1r mRNA levels. Similarly, Fgf7, Igf1, and Igf1r mRNA levels were unchanged with housing temperature in male GLPDRKO mice. Therefore, while GLP-2R action requires the expression of these key proteins to elicit its intestinotrophic effects, physiological increases in GLP-2R action with cold stress did not significantly increase the gene expression of these markers in the jejunum. Studies exposing mice with cell-specific deletions of GLP-2R, such as in subepithelial myofibroblasts, to chronic cold stress are required to define the signaling axis involved.

Increases in intestinal mass with cold stress have been reported to occur by 2 weeks⁶. Interestingly, body mass over 2 weeks was shown to remain constant across 11 different ambient temperatures, between 5 and 35 °C, in both male and female mice⁶. Specifically, the decreases in white adipose tissue were accompanied by increases in liver and small intestinal weight by the same magnitude⁶. Our 5-week study demonstrated significant increases in body weight from baseline in male and female WT mice with cold stress, despite significant decreases in adiposity, consistent with lower plasma leptin levels. However, body weight gain from baseline was blunted in GLPDRKO male and female mice. This attenuated body weight gain and significantly lower body weight may be driven by decreases in nutrient absorption with the loss of GLP-2R signaling⁶⁴. However, as shown in the Glp2r^−/− mice, which display a similar body weight gain to WT controls, this loss can be compensated for. Indeed, Glp2r^−/− mice displayed significantly greater plasma active GLP-1 levels, as previously shown⁶⁵, highlighting a feedback loop with its expression on enteroendocrine cells, including those that are GLP-1+ and GLP-2+ in the small intestine^66,67. Together, these results highlight a critical redundancy between GLP-1R and GLP-2R for metabolic adaptation to increased energy demands. Another example of the role of food volume to promote intestinal length is in the hyperphagic ob/ob mice, who display significantly longer small intestines⁵⁰. By contrast, shorter small intestines are observed in diet-induced obese mice compared to controls^58,68. Reduced sensitivity to endogenous GLP-1R signaling is reported in metabolic disease^69,70, where impairment of the intestinotrophic actions of GLP-1R and GLP-2R could extend to obesity. Interestingly, the GLP-1R/GLP-2R dual agonist dapiglutide is being investigated in the DREAM Trial (NCT05788601) to lower body weight and low-grade inflammation in obesity. Altogether, these findings highlight essential roles for GLP-1R and GLP-2R action as well as gut adaptation in the face of increased metabolic demand.

The impact of cold stress on intestinal lipid metabolism in this study was sex dependent. In male mice, despite the greater energy demand by peripheral tissues, plasma triglyceride levels were similar to those observed in TN-housed mice in response to an acute dietary fat challenge. Rather, the changes observed in this study were driven by genotype. The phenotype of significantly worse lipid tolerance observed in our male GLPDRKO mice resembles that of a Glp1r^−/− mouse⁵⁹, as Glp2r^−/− mice display similar lipid tolerance compared to controls⁶⁵. Together, these results suggest that the role of endogenous GLP-1R signaling in the control of intestinal lipoprotein secretion in male mice dominates, consistent with results from hamsters with prolonged co-infusion of GLP-1 and GLP-2⁷¹. In female mice, by contrast, cold-stressed WT mice had greater post-oil triglyceride levels compared to TN controls. Cold stress, however, did not worsen lipid tolerance in female GLPDRKO mice, suggesting that cold-stress-induced increases in intestinal-triglyceride secretion require either GLP-1R or GLP-2R. Moreover, female cold-stressed GLPDRKO mice displayed significantly greater jejunal oxygen consumption, which may limit the availability of fatty acids for chylomicron production. Finally, while male GLPDRKO mice more closely phenocopy the Glp1r^−/− mouse⁵⁹ in terms of lipid tolerance, this genotype impact was not observed in female mice, highlighting a potential sex difference in the balance of GLP-1 and GLP-2 control of intestinal lipid metabolism. The specific intestinal GLP-1R/GLP-2R+ cell types and mechanisms underlying this maintenance in body composition and lipid tolerance remain unsolved.

Cold stress in mice has also been shown to change the composition of the gut microbiota, and fecal transplants from cold-stressed donors to room temperature-housed recipients elicit the same increases in intestinal morphology, independent of food intake^5,7. These studies also demonstrate an association of the fecal and cecal SCFA levels with the expansion of the gut surface area⁵. While our data reflect a shift in the proportion of SCFA with cold stress, these shifts persisted in our GLPDRKO mice, who fail to expand the absorptive surface area despite increased food intake. The increase in fecal butyrate levels observed likely reflects the increase in gross energy of feces observed as ambient temperature drops⁶. Certainly, microbial products beyond fecal SCFA levels, including bile acids, play important roles in host metabolism⁷². Additionally, L cells are equipped with GPCRs to detect microbes and their metabolites, and several studies find impacts of GLP-1R and GLP-2R signaling on the gut microbiota composition^31,73,74. Specifically, GLP-1R agonist semaglutide has been shown to induce pathways in the gut microbiota associated with glucose and amino acid metabolism³¹. The addition of oligofructose as a prebiotic to the diet in ob/ob mice significantly increases plasma GLP-1, colonic L cell abundance, and colon proglucagon mRNA levels^73,74, and GLP-2R antagonism blunts the anti-inflammatory effects of this prebiotic treatment⁷³. Overall, the findings from the current study support the requirement of GLP-1R and GLP-2R action downstream of changes in the gut microbiota to expand the absorptive surface area in response to the increased fuel availability.

In our study, we found that cold stress significantly reduced intestinal transit time compared to TN controls, and to a greater extent in GLPDRKO mice. A reduction in intestinal transit time has also been observed in both male and female C57BL/6NJ mice housed at 22 °C compared to 30 °C for 2 weeks, although food intake was comparable among housing temperatures⁷⁵. Rather, this change in intestinal transit at 22 °C was corticotropin-releasing hormone-dependent⁷⁵. In our study, daily food intake was similar among cold-stressed mice, independent of genotype, suggesting that this magnitude of change from the TN to cold-stressed GLPDRKO mice was driven by factors beyond the volume of food introduced to the system. GLP-1 slows gastric emptying and gut motility, while both GLP-1 and GLP-2 have been shown to act as ileal “brakes”, slowing nutrient transit in the small intestine to facilitate nutrient absorption⁷⁶. Similarly, GLP-1R and GLP-2R dual agonist dapiglutide has been shown to dose-dependently slow intestinal barium sulfate transit in healthy rats⁷⁷. This suggests that compensatory mechanisms manifest to slow nutrient transit for nutrient absorption in TN-housed GLPDRKO mice. Nevertheless, these data suggest that cold stress, where gut hormone receptor signaling is high, uncouples the feedback mechanisms that GLP-1R and GLP-2R govern for intestinal transit in the post-prandial state.

In conclusion, in the state of increased metabolic demand that accompanies cold stress, increased food intake drives intestinal surface area expansion, in part by crypt-cell proliferation and villus lengthening in a GLP-2R-dependent manner. These findings also highlight sex differences in the metabolic response to cold stress. GLP-2 analogs are already approved therapies for short bowel syndrome and are effective at reducing the need for parenteral nutrition in ~65% of individuals¹⁶. Understanding the exact pathways engaged during the intestinotrophic effects of GLP-2R to expand the absorptive surface area may explain the heterogeneity of the GLP-2 analog response and may provide more therapeutic targets for individuals living with short bowel syndrome, as well as other diseases that impact the small bowel, including atrophy induced by parenteral nutrition, inflammatory bowel diseases, and obesity. Exploring intestinal adaptations to complementary physiological challenges is required to uncover deeper mechanistic insight into metabolic disease.

Author contributions

Study concept and design: N.M.M., A.A.H., and E.E.M.; Acquisition of data: N.M.M., A.H., C.F.M., D.L., H.O., L.H., S.M.P., N.A.T., C.A.A.L., M.N.B., E.M.Z., S.M.C., I.A., X.Z., I.L.S., and E.E.M. Analysis of data: N.M.M., A.A.H., C.F.M., D.L., H.O., L.H., and E.E.M., Drafting of the manuscript: N.M.M., A.A.H., C.F.M., A.M., and E.E.M. Critical revision of the manuscript for important intellectual content: N.M.M., C.F.M., D.L., S.M.P., C.A.A.L., I.A., I.L.S., A.M., K.A.P., M.E.H., E.E.M. Obtained funding: E.E.M, M.E.H, and K.A.P.

Peer review

Peer review information

Communications Biology thanks Changting Xiao, Kaare Villum Grunddal and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Joao Valente.

Data availability

All data supporting the findings of this study are available within the Supplementary Information files.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09281-4.