Antibiotics that reduce gut bacteria may improve blood sugar control through increased GLP-1 signaling

The animals used in this study were male C57BL/6N, GF, GLP-1 receptor knockout (GLP1R KO), Cyp2c70 knockout, and Ucp1Dtr2Gfp (Ucp1DTR) mice, aged 12 weeks at the start of experiments. Mice were housed in groups of two to five per individually ventilated cage under specific pathogen-free (SPF) or GF conditions on a reversed 12h dark/light cycle at room temperature (RT, 23°C), thermoneutrality (TN, 30°C), or cold (CE, 8°C) depending on the experimental setting. Mice had ad libitum access to a standard chow diet and water.

For UCP1⁺ cell ablation, Ucp1DTR or wild-type control mice were subcutaneously (sc.) injected with 3 doses of 100 ng Diphtheria Toxin (DT; Sigma-Aldrich, 322326) every 6h on day 0. For sustained UCP1⁺ cell deletion, one dose of 100 ng DT was injected subcutaneously twice per week. Gut microbiome depletion was induced by providing an antibiotic cocktail in sterile drinking water ad libitum to Ucp1DTR, wild-type, and GLP1R KO mice. The antibiotic water in the drinking bottle was refreshed twice per week, and fresh antibiotic water was prepared weekly. The antibiotic cocktail consisted of 100 µg/ml neomycin (Sigma-Aldrich, N6386), 50 µg/ml streptomycin (Sigma-Aldrich, S9137), 100 U/ml penicillin (Sigma-Aldrich, P7794), 50 µg/ml vancomycin (Sigma-Aldrich, V1130), 100 µg/ml metronidazole (Sigma-Aldrich, M1547), 1 mg/ml bacitracin (Sigma-Aldrich, 11702), 125 µg/ml ciprofloxacin (Sigma-Aldrich, 17850), and 100 µg/ml ceftazidime (Sigma-Aldrich, C3809).

The animals were euthanized by CO₂ overdosage with a 50% flow rate of cage volume per minute. All animal experiments conducted in our lab were approved by the Cantonal Veterinary Office of Zurich, Switzerland.

Mice were fasted for 6h before measuring their baseline glucose levels (Roche, Accu-Chek Aviva). 2 g/kg body weight glucose was administered either intraperitoneally (ip.) or orally, and blood glucose levels were recorded at indicated timepoints.

During the period of temperature recording (total 6h), mice were fasted. The tip of the thermometer was dipped in Vaseline. The anal sphincter was stimulated with the tip of the thermometer, which was ultimately inserted into the anus while the mouse was restrained to avoid movement until the temperature measurement was completed.

Energy expenditure and metabolic parameters were recorded using the Phenomaster metabolic cage system (TSE system).

Mice were sacrificed after 6h of fasting. Inguinal white adipose tissue (ingWAT), interscapular brown adipose tissue (iBAT), and the liver were snap-frozen in liquid nitrogen and stored at −80°C until further processing. The small intestinal parts (duodenum—first 8 cm; jejunum—middle; ileum—last 8 cm), caecum, and colon were dissected. The fecal content was pushed out and collected. The intestinal tissue got flushed with PBS, snap-frozen in liquid nitrogen, and stored at −80°C until further processing. The gall bladder was punctured with a 30G insulin syringe, and bile was collected in a PCR tube. For the GLP-1 time course experiment, blood was collected from the tail vein. Blood was immediately mixed with 100 µl/ml 5 mM EDTA and, if needed for GLP-1 measurement, with an additional 10 µl/ml DPP4 inhibitor (Sigma-Aldrich, DPP4-M) and 11.1 µl/ml 550 µM aprotinin (Sigma-Aldrich, A1153). This was followed by subsequent centrifugation at 8000 rcf at 4°C for 10 min, and plasma was transferred into new tubes and stored at −80°C.

Plasma insulin, GLP-1, glucagon, and lipopolysaccharides (LPS) were measured using the Ultra Sensitive Mouse Insulin ELISA Kit (Crystal Chem, 90080), the V-PLEX Plus GLP-1 (active) Kit (mesoscale, K1503OG), the Mouse Glucagon ELISA Kit (Crystal Chem, 81518), and the Lipopolysaccharides (LPS) ELISA Kit (abbexa, abx150357) following the manufacturer's instructions.

An ingWAT or iBAT depot was homogenized in RIPA buffer (50 mM Tris/HCl pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X100, 0.5% sodium deoxycholate, 10% glycerol) containing 100X protease inhibitor (Roche, 5056489001) and 100X phosphatase inhibitor (Thermo Fisher Scientific, Zug, Switzerland). The lysate was centrifuged at 4°C for 10 min at maximum speed. Total protein was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225) following its instructions. 30 µg of protein was loaded onto an SDS-polyacrylamide gel (4% stacking, 10% separation density). Following separation, protein was transferred onto a 0.2 µm nitrocellulose membrane (Bio-Rad, 1620112). Blocking was done in 5% milk for 1h and subsequently incubated with the UCP1 Polyclonal Antibody (1:1000, Invitrogen, PA1-24894) in 5% milk at 4°C overnight. The HRP-linked Anti-rabbit IgG Antibody (1:4000, Cell Signaling, 7074) was used as a secondary antibody. Imaging was done with the ChemiDoc MP Imaging System (Bio-Rad).

mRNA isolation of liver and ingWAT was done using GENEzol (Geneaid, GZR200) following the manufacturer's instructions. Reverse transcription of 100 ng RNA into cDNA was performed with the High-Capacity Reverse Transcription kit (Applied Biosystems, 4368813). cDNA was diluted to 5 ng/µl, and gene expression was measured with quantitative RT-PCR using the KAPA Sybr Fast kit (Roche, SFUKB) on a 384-well format Viia 7 machine (Applied Biosystems). Primer sequences are listed in Table 1.

SCFAs were extracted from cecal contents after homogenization in PBS using a Tissue Lyzer at maximum speed (30 Hz/s, 2.5 min). A standard solution was prepared containing the six compounds of interest, with final concentrations of 1 mM for acetate, lactate, propionate, butyrate, and succinate, and 10 mM for formate, using 50% isopropanol. Eight serial dilutions of the standard solution were made in 50% isopropanol. An internal standard solution was prepared with isotope-labeled versions of the compounds in LC-water. A master mix was prepared containing methanol, LC-water, 100 mM aniline, 50 mM EDC, 2 M HCl, and internal standard solution. For the reaction, 75 µl of the master mix was added to 25 µl of each standard/sample in a 96-well polypropylene plate (Thermo Fisher). The mixture was incubated for 30 min at 4°C, followed by quenching the reaction with 500 mM β-mercaptoethanol. The samples were incubated for an additional 30 min at room temperature before being analyzed with the LC-MS.

BAs were extracted from caecal samples using methanol containing BHT and deuterated internal standards added prior to extraction. Sample weights were recorded for data normalization after measurement. An external calibration curve, ranging from 0.0006 µM to 27 µM, was used for quantification. The analytes were separated using reversed-phase chromatography on a C18 column, with a mobile phase consisting of methanol/acetonitrile, 5 mM ammonium acetate, and 0.1% formic acid. The eluents were analyzed by MRM mass spectrometry, with measurements taken in either positive or negative electroionization mode.

STC-1 cells were kindly provided by Boehringer-Ingelheim. Cells were kept in full medium containing high-glucose (4.5 g/l) DMEM (Gibco, 41966-029) supplemented with 10% fetal bovine serum (Gibco, 10500-064) and 1% penicillin/streptomycin (Gibco, 15140-122) in a humidified cell culture incubator (37°C, 5% CO₂). Full medium was changed every 2–3 days, and cells were split 1:5 upon reaching 70% confluency.

For GLP-1 secretion assays, cells were seeded in 24-well plates and used upon reaching 80% confluency. Cells were washed two times with prewarmed secretion buffer (138 mM NaCl, 4.5 mM KCl, 4.2 mM NaHCO₃, 1.2 mM NaH₂PO₄ (2H₂O), 2.5 mM CaCl₂ (2H₂O), 1.2 mM MgCl₂ (6H₂O), 10 mM HEPES; supplemented with fresh 0.1% fatty acid–free bovine albumin serum) and subsequently incubated in secretion buffer for 1h in the cell culture incubator. Meanwhile, a stimulation buffer containing secretion buffer supplemented with 20 µl/ml DPP4 inhibitor (Sigma Aldrich, DPP4-M) was prepared and used as a control. For positive control, 10 mM glucose was added to the stimulation buffer. For stimulation with BAs, stimulation buffer was supplemented with 30 µM of the respective BA. The following BAs were used: cholic acid (CA, Sigma Aldrich, C1129), ursocholic acid (UCA, Avanti, 700229P), taurodeoxycholic acid (TDCA, Avanti, 700250P), taurocholic acid (TCA, Sigma Aldrich, T4009), hyocholic acid (HCA, Avanti, 700159P), ursodesoxycholic acid (UDCA, Sigma Aldrich, U5127), alpha-muricholic acid (α-MCA, Avanti, 700232P), beta-muricholic acid (β-MCA, Sigma Aldrich, SML2372), omega-muricholic acid (ω-MCA, Avanti, 700231P), tauro-alpha-muricholic acid (tauro-α-MCA, Avanti, 700243P), and deoxycholic acid (DCA, Sigma Aldrich, D5670). The secretion buffer was replaced with the respective stimulation buffer, and cells were incubated for 3h in the cell culture incubator. Supernatants were collected, spun down at 900 rcf for 5 min at 4°C, and stored at −80°C. Active GLP-1 was measured using the V-PLEX Plus GLP-1 (active) Kit (mesoscale, K1503OG) according to the manufacturer's protocol. The cell plate was washed with PBS and stored at −80°C for the following protein isolation and total protein determination used for data normalization (see Section 2.5).

Results are presented as mean ± SEM. Statistical analyses other than ANCOVA were performed with GraphPad Prism 9 and 10. An unpaired two-tailed t-test was done for comparison of two groups. An ordinary one-way analysis of variance (ANOVA) was applied for comparisons of three or more groups. Two-way ANOVA was applied for comparisons of two groups across multiple parameters. The ANCOVA analysis was done using the Energy Expenditure Analysis page1 provided by NIDDK Mouse Metabolic Phenotyping Centers2. Statistical significance is indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

To study the contribution of brown and beige adipocytes to enhanced glucose tolerance upon gut microbiota deletion, we used the Ucp1Dtr2Gfp mouse line (Ucp1DTR), which allows for transient ablation of UCP1⁺ cells upon diphtheria toxin (DT) injections as described previously (21) (Figure 1A). Wild-type (WT) mice served as control for the inducible Ucp1DTR mouse model, and all animals received subcutaneous (sc.) DT injections twice per week. This ensured the ablation of UCP1⁺ cells in Ucp1DTR mice, while WT mice retained brown and brite adipocytes. Within both genotypes, one group received an antibiotic cocktail (Abx) administered in the drinking water for 3 weeks before starting experiments to deplete the gut microbiota (Ctrl + Abx, DTR + DT + Abx), while the other groups (Ctrl, DTR + DT) retained their gut microbiome. Gut microbiota depletion resulted in reduced adipose tissue weight (Figure 1B), which is in line with the reported reduced adiposity (12, 13). Furthermore, UCP1⁺ cell depletion resulted in decreased depot mass, as expected (Figure 1B). For confirmation, we measured UCP1 protein expression in both ingWAT and iBAT. Abx treatment indeed increased UCP1 protein expression in ingWAT, however, injection of DT significantly reduced UCP1 protein levels in both ingWAT and iBAT in Ucp1DTR mice at room temperature (Figures 1C–E). Mice treated with Abx experienced a significant body weight loss within the first seven days of treatment but recovered and even gained weight compared to their controls at room temperature (RT; Figure 1F), which aligns with previous studies (22). Interestingly, mice at either thermoneutrality (TN; Supplementary Figure S1A) or cold exposure (CE; Supplementary Figure S1B) showed increased body weight after 4 weeks of Abx treatment. This finding is likely to be explained by the increased food intake that was observed in mice treated with Abx for 4 weeks at CE (Supplementary Figure S1C). Following the 3 weeks of Abx treatment, we conducted an intraperitoneal glucose tolerance test (ipGTT). Surprisingly, the significant improvement in glucose tolerance was independent of UCP1⁺ cell presence in all condition tested – at RT (Figures 1G, H), at TN (Supplementary Figures S1D, S1E) and at CE (Supplementary Figures S1F, S1G). Blood insulin levels were significantly decreased in mice upon antibiotic-induced gut microbiota depletion at RT (Figure 1I), but not at TN (Supplementary Figure S1H) and CE (Supplementary Figure S1I). These data demonstrate that gut microbiota depletion-dependent improved glucose metabolism is independent of UCP1⁺ cells.

To assess the potential contribution of gut microbiota to thermogenesis independent of the thermogenic role of UCP1⁺ cells, we compared rectal body temperature in all experimental groups (Ctrl, Ctrl + Abx, DTR + DT, DTR + DT + Abx) during fasting at room temperature (0h) and cold exposure (4h and 6h). Mice with antibiotic-induced gut microbiota depletion experienced a stronger decrease in rectal body temperature during cold exposure than their controls (Figure 1J). This body temperature drop was even more pronounced in mice lacking both gut microbiota and UCP1⁺ cells (Figure 1J), suggesting that gut microbiota contributes to thermogenesis independently of UCP1⁺ cells. Surprisingly, energy expenditure was improved in Abx-treated mice compared to WT controls upon 5 days of CE (Figure 1K). Due to severity constraints, energy expenditure was not measured in mice with both microbiota depletion and UCP1⁺ cell ablation, resulting in the exclusion of these experimental groups from analysis. Altogether, these findings suggest that gut microbiota plays a significant role in thermogenesis, influencing both cold sensitivity and energy expenditure.

Next, we aimed to identify the main driver of gut microbiota depletion-dependent improved glucose metabolism. It has been shown previously (7, 15, 23, 24) that active GLP-1 is significantly improved upon antibiotic-induced gut microbiota depletion. Indeed, mice treated with Abx for 4 weeks showed significant elevation of baseline active GLP-1 in plasma compared to control mice in a fasted state (Figure 2A). Studies have shown that antibiotic-induced depletion does not fully erase the gut microbiota but significantly reduces its abundance and alters its composition (7, 22, 25, 26). To further confirm that the increase in active GLP-1 is independent of the remaining gut microbiota, we measured active GLP-1 in GF mice. Notably, active GLP-1 was also significantly elevated in fasted GF mice compared to controls (Figure 2B). These results suggest that the elevation of active GLP-1 is driven by the absence of gut microbiota, independent of any remaining microbial presence. To identify the optimal Abx treatment duration, we performed a treatment time course. Abx was given to the treatment group, and plasma was collected on days 1, 3, 7, and 10. Surprisingly, even one day of Abx treatment was sufficient to yield a significant elevation of active GLP-1 in the plasma compared to control mice and increased further until day 7 before stabilizing at a level already seen on day 3 (Figure 2C). Thus, we performed the following experiments after at least 3 days of antibiotic treatment. GLP-1 is primarily secreted in response to nutrient ingestion and plays a key role in the incretin effect. To assess the role of GLP-1 in improving glucose tolerance following antibiotic-mediated gut microbiota depletion, we first performed an oral glucose tolerance test (oGTT) in control and whole-body GLP-1 receptor knockout (GLP1R KO) mice with or without antibiotic-induced gut microbiota depletion. These mice also exhibited significantly elevated active GLP-1 levels in a fasted state after Abx-mediated gut microbiota depletion (Figure 2D). While gut microbiota depletion in control mice resulted in improved glucose tolerance, this effect was fully reversed upon a whole-body knockdown of the GLP-1 receptor (Figures 2E, F), suggesting that GLP-1 signaling is essential for the observed glucose tolerance improvement. Given the elevated basal levels of GLP-1 following gut microbiota depletion independent of nutrient ingestion, we hypothesized that the observed improvement in glucose tolerance would also be abolished in GLP1R KO mice during an ipGTT. Surprisingly, however, the improvement in glucose tolerance was still observed in GLP1R KO mice following antibiotic treatment (Figures 2G, H). These results indicate that there is a GLP1R-independent mechanism contributing to enhanced peripheral glucose homeostasis upon gut microbiota depletion.

A possible mediator is glucagon, which is known to be regulated by GLP-1 (27, 28). Thus, we measured plasma glucagon levels at different timepoints during antibiotic administration and measured Gcgr gene expression in liver and ingWAT. Gcgr gene expression was not changed in either ingWAT or liver (Supplementary Figure S2A–S2F). Surprisingly, fasting plasma glucagon levels were increased 7 days upon antibiotic treatment compared to control mice (Figure 2I). As this effect is not observed at the earlier timepoints, we conclude that glucagon is not mediating the secondary effects on glucose metabolism improvement.

First, we aimed to determine the main source of increased GLP-1 secretion. WT mice were treated with an antibiotic cocktail for 3 days. As expected, gut microbiota-depleted mice showed decreased body weight compared to the control mice (Figure 3A). We measured active GLP-1 levels in the different intestinal parts, namely, duodenum, jejunum, ileum, caecum, and colon. Intestinal tissue weight was not different; however, a tendency of increased caecum weight was observed compared to control mice (Supplementary Figure S3A). This is in line with previous reports that showed that antibiotic-induced gut microbiota depletion leads to a highly enlarged caecum (29). Significantly increased active GLP-1 was measured in the ileum upon Abx-dependent gut microbiota depletion compared to control mice (Figure 3B). Depletion of the gut microbiome with Abx modulates the presence of luminal secondary metabolites such as SCFAs and BAs, which consequently affect gut signaling and potentially whole-body glucose metabolism. Thus, we next examined the SCFA composition, key gut-derived metabolites modulated by the composition of the gut microbiome and potential stimulators of GLP-1 release (30, 31). We collected the feces of mice treated with antibiotics for 3 days and analyzed the SCFAs. In line with previous studies, we found that acetate and butyrate were significantly reduced upon Abx-induced gut microbiota depletion compared to control mice (Figure 3C). As both are known to act as GLP-1 stimulators (30, 31), we conclude that while gut microbiota depletion leads to a significantly altered SCFA composition, this change does not drive the increase in active GLP-1. Given that antibiotics not only deplete the gut microbiota but also damage the gut barrier and overall lead to gut dysbiosis (32, 33), we investigated whether lipopolysaccharide (LPS) levels changed in the intestinal regions that exhibited altered GLP-1 secretion following 3 days of antibiotic-induced gut microbiota depletion, as well as in plasma (Supplementary Figures S3B, S3C). However, we observed no significant changes in LPS levels either in plasma (Supplementary Figure S3B) or in the ileum, caecum or colon (Supplementary Figure S3C). This is in line with a previous study, which found that gut-derived LPS is not essential for the improvement in glucose tolerance observed after gut microbiota depletion (13). BAs are modified by the gut microbiota and known to act as GLP-1 secretion regulators, thereby modulating glucose metabolism and insulin sensitivity (23, 34 –39). Therefore, we aimed to identify the BAs that are regulated upon antibiotic-induced gut microbiota depletion. BA profiling (Table 2) was conducted in the bile, feces, and plasma of mice treated with Abx for 3 days compared to control mice (Figures 3D–K and Supplementary Figures S3D–S3F). Various BAs were changed in concentration upon antibiotic-induced gut microbiota depletion as shown in (Figures 3D–K). We first focused on the muricholic acids (MCAs) ω-MCA and α-MCA, which were most significantly regulated amongst all BAs in feces (Figure 3J), as potential GLP-1 modulators. We used the Cyp2c70 knockout mouse model, which lacks the Cyp2c70 enzyme responsible for the synthesis of MCAs, leading to an altered, more human-like BA pool (40 –43). Plasma was collected in fasted mice and 15 min following oral mixed meal ingestion. Interestingly, basal active GLP-1 was not different, but active GLP-1 secretion upon meal ingestion was blunted in mice lacking MCAs (Figure 3L). This data indicates that MCAs play a role in postprandial active GLP-1 release in mice but demonstrates that MCAs do not regulate basal active GLP-1.

Next, we sought to examine the effects of all significantly regulated BAs on GLP-1 secretion. Thus, we performed a GLP-1 secretion assay in STC-1 cells, stimulating starved cells with the different BAs at a concentration of 30 µM. We were able to identify three inhibitors of GLP-1 secretion, namely taurocholic acid (TCA), hyocholic acid (HCA), and ursodesoxycholic acid (UDCA; Figure 3M). As shown in our BA profiling data (Figures 3H, I), HCA and UDCA were significantly reduced in feces, while TCA was increased in gall bladder (Figure 3G). In the BA pool of GF mice, which have elevated basal active GLP-1 (as shown in Figure 2B), it has been shown that taurine-conjugated BAs are present while secondary BAs such as HCA and UDCA are likely reduced due to the absence of gut microbiota (23, 44). Hence, we propose that HCA and UDCA may be potential inhibitors of basal GLP-1 secretion, and their absence following antibiotic-induced gut microbiota depletion could drive the observed improvement in glucose metabolism by relieving the inhibition of basal GLP-1 secretion.

While our study provides new insights into how gut microbiota depletion modulates glucose metabolism through GLP-1 signaling, some limitations should be acknowledged.

First, representative Western blots of UCP1 protein expression were obtained from two biological replicates per group. It should be noted however that the reliability efficiency of the Ucp1DTR mouse model has been previously validated (21).

Second, we did not directly quantify microbial load by 16S rRNA sequencing or bacterial qPCR. However, we followed a previously published protocol, which demonstrated that this antibiotic regimen efficiently depletes gut bacteria (12). Moreover, consistent physiological indicators of microbiota depletion, including enlarged caecum and reduced fecal SCFA levels, together with prior validation of the same antibiotic cocktail, strongly support the effectiveness of the treatment.

Third, while we demonstrate a GLP-1–dependent improvement in oral glucose tolerance, our data indicates the involvement of additional pathways beyond canonical GLP1R and GCGR signaling, which remains yet to be determined.

Finally, while bile acid-induced GLP-1 secretion has often been linked to TGR5- or FXR-dependent signaling, basal GLP-1 secretion regulation may be governed by distinct mechanisms, such as modulation of ion channels, other GPCRs expressed in L cells, or changes in bile acid transporters. Further studies will be required to define these mechanisms.