GPR41 deficiency alters the gut microbiota-bile acid axis, reduces ileal expression of Npc1l1, and attenuates hypercholesterolemia in male mice

ABSTRACT

Hypercholesterolemia contributes to the development of atherosclerosis and is a major risk factor for cardiovascular diseases (CVD). Dietary fiber can attenuate CVD, at least in part, by serving as a fermentable substrate for gut bacteria, leading to the production of short-chain fatty acids (SCFAs), such as butyrate and propionate, which have been linked to atheroprotective effects. SCFAs are sensed by G-protein coupled receptors including GPR41, GPR43, and GPR109A. To explore the role of these receptors in hypercholesterolemia and CVD, we examined atherosclerosis progression and lipid metabolism in Gpr41^-/-, Gpr43^-/-, and Gpr109a^-/- mice using a proprotein convertase subtilisin/kexin type 9 adeno-associated virus (PCSK9-AAV) model of hypercholesterolemia. Deficiency of any single SCFA receptor did not significantly affect atherosclerotic plaque burden compared with wild-type (WT) littermates. However, male Gpr41^-/- mice exhibited decreased gonadal fat, plasma triacylglycerol, and low-density lipoprotein cholesterol levels compared to their WT littermates. GPR41 deficiency in males was also associated with increased cecal propionate levels, reduced ileal expression of nutrient transporters such as Npc1l1 and a trend toward increased gut motility. In addition, male Gpr41^-/- mice displayed altered gut microbiota composition and lower levels of microbially generated bile acids relative to their WT counterparts. Together, these findings highlight GPR41 as a key intestinal chemosensor regulating nutrient uptake, lipid storage, and microbiota composition.

Introduction

Cardiovascular disease (CVD) is the leading cause of death worldwide and is expected to rise as individuals with elevated risk factors increase in populous countries.¹ Atherosclerosis is an inflammatory disease resulting from dysregulation of lipid metabolism and chronic inflammation.² Despite major advances in atherosclerosis treatment and prevention, including statin drugs, atherosclerosis continues to be a leading contributor to global morbidity and mortality.³ Moreover, while statins are highly effective in lowering plasma cholesterol, their use can be limited by adverse effects, and accumulating evidence suggests that long-term statin exposure may increase the risk of type 2 diabetes in some patients.⁴ Therefore, additional strategies to address atherosclerosis and hypercholesterolemia are warranted.

One promising target for new CVD therapies is the gut microbiome. Intestinal microbes influence the host via production of a vast array of metabolites. Recent evidence suggests that gut-microbiota-derived metabolites, including trimethylamine N-oxide, imidazole propionate, and short-chain fatty acids (SCFAs) influence cardiometabolic health.^5-8 SCFAs are mainly produced from bacterial fermentation of dietary fibers, resistant starches and other carbohydrates that reach the distal intestine. Acetate, propionate, and butyrate are the most abundant SCFAs in the intestine of humans and mice. Butyrate elicits beneficial metabolic effects including promotion of gut barrier integrity as well as atheroprotection through reduced inflammation.^9-12 Similarly, propionate has been shown to ameliorate atherosclerosis in mice by reducing plasma cholesterol levels through regulation of cholesterol absorption.¹³ Lower abundances of SCFA-producing bacteria are associated with higher risk for atherosclerosis.¹⁴ However, the complete mechanisms behind the atheroprotective effects of butyrate and propionate are not fully understood.

SCFAs are sensed by G-protein coupled receptors (GPRs) including GPR41, GPR43, and GPR109A. GPR41 senses propionate and butyrate (and acetate to a lesser extent) and is expressed in various tissues that are related to cardiometabolic physiology including the pancreas, adipose tissue, dendritic cells, and regulatory T-cells.^15,16 GPR41 deficiency has been shown to alter energy balance and gut motility in mice.^16-18 Moreover, previous studies have demonstrated that GPR41 has roles in insulin sensitivity, obesity, and hypertension, all of which are known to impact atherosclerosis.^18-20 These studies show that GPR41 is involved in physiological processes that modulate cardiometabolic health. Nevertheless, whether GPR41 contributes to the atheroprotective effects of SCFA has not been tested.

We investigated the contributions of SCFA receptors to atherosclerosis using a hyperlipidemic mouse model fed a high-fiber diet to promote short-chain fatty acid (SCFA) production. Mice deficient in GPR41, GPR43, or GPR109A (Gpr41^-/-, Gpr43^-/-, and Gpr109a^-/-) showed no significant differences in atherosclerotic plaque area compared to their respective wild-type littermates, indicating that individual loss of these receptors does not exacerbate atherosclerosis progression under these conditions. Intriguingly, GPR41 deficiency in males resulted in several cardiometabolic alterations including reduced plasma triacylglycerol (TAG) and total cholesterol (TC) levels, increased gonadal adiposity, and increased cecal propionate, most of which were not observed in the Gpr43 or Gpr109a KO strains. Further assessment of Gpr41^-/- males revealed alterations in cecal microbiota composition and microbial bile acid metabolism, and reduced ileal expression of nutrient transporters including the cholesterol transporter Npc1l1. Our findings suggest that GPR41 plays a key role in nutrient uptake, cholesterol metabolism, and gut microbiome function in male mice.

Materials and methods

Gpr-KO animals and diets

Gpr41^-/-, Gpr43^-/-, and Gpr109a^-/- mice were generated as previously described^18,21,22 and generously donated by Karen Ho, Northwestern University (Gpr41^-/- and Gpr43^-/-) and Pamela Martin, Augusta University (Gpr109a^-/-). All knockout (KO) strains had previously been backcrossed with C57BL/6J mice for multiple generations. Each strain of KO mice was paired with C75BL/6J (Jackson Laboratories, Bar Harbor, ME) breeding mates to produce a heterozygous (het) F₁ generation for each strain. The hets were then paired for breeding to produce an F₂ generation which was used for experimentation. All mice were housed in a ventilated cage system (Alternative Design, Siloam Springs, AR) on corn husk bedding with ad libitum access to chlorinated water and chow (Teklad 8604, Inotiv, Madison, WI). All animal experiments were conducted in accordance with the University of Wisconsin-Madison's animal welfare standards and were approved by the university's Animal Care and Use Committee.

Gpr-KO experimental design

The F₂ generation for each group (Gpr41, Gpr43, and Gpr109a) yielded male and female WT, het, and KO littermates. Progeny were segregated by sex and genotyped by PCR. Hets were removed from the study and the WT and KO littermates were separated, with no more than 5 mice per cage, for the duration of the experiment. Each sex/genotype group consisted of at least three cages. At 6 weeks of age, male and female WT and KO littermates were injected with 1.5 × 10¹¹ particles of PCSK9-AAV8 (AAV8-D377Y-mPCSK9, Vector Biolabs, Malvern, PA) via retro orbital injection under isoflurane-induced anesthesia. Mice were also switched to a high-plant polysaccharide, high-cholesterol (1.5% wt/wt) diet supplemented with cocoa butter (6.5% wt/wt) (TD.210423, Inotiv) immediately after injection to induce hypercholesterolemia. This generates plasma total cholesterol levels of >500 mg/dL. Any mice that did not have terminal plasma total cholesterol levels >150 mg/dL were considered to have had abnormal PCSK9-AAV infection and were removed from the study. Mice were maintained on this diet for 12 weeks at which point they were sacrificed for tissue collection. Gonadal adipose tissue was collected from males by dissecting all tissue on the gonadal adipose pad distal from the testes, and from the females by dissecting the entire gonadal adipose pad after first separating the uterine horns. The heart was dissected, submerged in OCT compound and frozen at −80 °C for storage. Sample sizes for each sex/genotype group ranged from 7 to 12 mice, 6 to 9 mice, and 8 to 11 mice within the GPR41, GPR43, and GR109A groups, respectively.

Atherosclerosis analysis

Atherosclerotic plaque area was measured in the aortic sinus as previously described.²³ Briefly, 10 μm sections of the aortic sinus were systematically collected such that each slide set contained sections from 0, 100, 200, 300, 400, 600, 700, and 800 μm ascending from the base of the aortic sinus where the aortic valve leaflets first appear. Slides containing these sections were fixed with formalin and stained with oil red O (ORO) and imaged on an inverted light microscope. Digitally captured images were analyzed using ImageJ (National Institutes of Health, Bethesda, MD) to measure lipid content (ORO-positive area) and total plaque area. The mean of the values across all sections was used to calculate average lipid content and average plaque area for each mouse. Sections that were damaged were removed from the mean calculation. The aortas from three mice were damaged during collection and could therefore not be analyzed.

Analysis of plasma markers

Blood was obtained via heart puncture of anesthetized mice just prior to euthanasia with 0.5 M EDTA and plasma was collected after centrifugation. Colorimetric assay kits were used to acquire plasma levels of total cholesterol (999-02601, Fujifilm, Tokyo, Japan), HDL-cholesterol (997-01301, Fujifilm), TAG (994-02891, Fujifilm) according to the manufacturer's instructions. Plasma levels of PCSK9 were determined using an ELISA assay kit (ab215538, Abcam, Cambridge, UK).

Plasma fast protein liquid chromatography (FPLC) analysis

Plasma from seven male WT mice and seven male Gpr41^-/- mice was evenly pooled and mixed. A volume of 165 μL of each pool was diluted in 835 µL of PBS and the entire mixture (1 mL total) was loaded onto a AKTA fast protein liquid chromatography (Amersham Pharmacia Biotech, Amersham, UK) fitted with a Superose 6 column (Cytiva, Marlborough, MA) in tandem with a Superdex 200 column (Cytiva) using a mobile phase consisting of 10 mM PBS (pH 7.4) and 0.02% sodium azide. After the first 11 minutes, 0.5 mL fractions were collected every minute into 48 separate elution tubes. Each fraction was analyzed for TAG and cholesterol content using the colorimetric assay kits described above.

Liver lipid content

Lipids were extracted from the liver by bead-beating 20–50 mg of frozen tissue with three 2.8 mm ceramic beads in the presence of lipid extraction buffer from the Abcam Liquid Extraction Kit (ab211044, Abcam) for 2 × 30 seconds followed by agitation for 20 minutes. The extraction slurry was spun down at 10,000 × g for 5 minutes at 4 °C, and the supernatant was transferred to a new tube and allowed to dry overnight. The resulting residue was dissolved in 50 μL of the kit resuspension buffer plus 350 μL of 10% Triton X-100 (X-100, Sigma-Aldrich, Burlington, MA) and heated (37 °C) sonication for 30–45 minutes. The lipid extract was used in the TAG and total cholesterol colorimetric kits described above to determine the lipid content per gram of tissue.

Bile acid measurements by uHPLC-MS/MS

Cecal sample prep. Bile acids were extracted from ~30 mg of frozen cecal content using the same lipid extraction method used for liver tissue described above, except that after drying overnight, the extracts were dissolved in 80% methanol and stored at room temperature (RT). Plasma sample prep. 20 µL of plasma was diluted in 80 µL of 100% methanol, vortexed for 30 seconds, spun down at 14,000 × g for 5 minutes at 4 °C, and the supernatant was collected. Cecal and plasma extracts were further diluted 1:10 using ultrahigh-pressure liquid chromatography (uHPLC)-grade H2O, and 100 μl was transferred to an HPLC vial for analysis. Bile acids were measured as previously described.²⁴ Briefly, samples were analyzed using a Thermo Scientific Vanquish uHPLC system coupled to a Q Exactive Orbitrap high-resolution mass spectrometer with a heated electrospray ionization source (negative polarity). Chromatographic separation was performed on a Waters Acquity UPLC BEH C18 column (1.7 μm, 2.1 × 100 mm) using a binary solvent system: 10 mM ammonium acetate in water (pH 6.0, solvent A) and 100% methanol (solvent B). The gradient ran from 30% to 100% B over 24 minutes, held at 100% for 5 minutes, then returned to 30% for a 2.5-minute re-equilibration, at a flow rate of 200 μl/min. The autosampler and column were maintained at 4 °C and 50 °C, respectively; injection volume was 10 μl. Mass spectrometry was performed using full MS1 scans (290–1000 m/z) from 3 to 20 minutes at a resolution of 17,500. Raw data were converted to mzXML format, and bile acids were identified using El-MAVEN.²⁵ Bile acids with peak areas below 10,000 in more than half of the samples were excluded from further analysis. Each bile acid was quantified and converted to µM (plasma) or mmol/g of cecal content by using standard curves and normalizing by sample input mass. Standards for each bile acid species were loaded on the same run as the samples except ω-muricholic acid, tauroursodeoxycholic acid, taurohyodeoxycholic acid and tauro-β-muricholic acid which were loaded on a separate run. Slopes for these standards were adjusted to account for run-to-run variation. We were unable to obtain a standard for tauro-α-muricholic acid, so the tauro-β-muricholic slope was used for conversion instead.

SCFA analysis of cecal contents

Frozen cecal content (~50 mg) was added to a vial containing 1 mL of 60 mM 2-butanol as an internal control, 2 g of H₂SO₄, and n μL of water (where n = [300 − cecal content mass (mg)]) was added. The vial was immediately sealed and allowed to dissociate at RT for 2–3 days. Prepared vials of each sample, along with standards for each analyte, were loaded onto headspace sampler (HS20, Shimadzu, Columbia, OH) and gas chromatograph (CG-2010 Plus GC, Shimadzu) under previously described conditions.²³ The resulting chromatograms were analyzed using Shimadzu Lab Solution software (version 5.92) and areas under the curve (AUC) were calculated for acetate, butyrate, and propionate. Concentrations of each SCFA, expressed as μmol per g of cecal content, were determined by converting the AUCs to μM and then adjusting for sample input mass.

Microbial community sequencing

DNA from flash-frozen cecal or fecal content was extracted via bead-beating with phenol-chloroform and cleaned as described previously.²³ Purified DNA was subjected to 16S rRNA V4 library prep and sequencing via Illumina MiSeq as described previsously²³ and sequenced at the UW-Madison Next Generation Sequencing Core. The resulting 16S rRNA gene amplicon sequences were demultiplexed and cleaned using Qiime2's²⁶ plugin for the DADA2²⁷ package. One sample was removed from analysis due to low read count (<1000). QC and removal of chimeric reads resulted in an average of 57,815 reads per sample of the Gpr KO experiment and 28,309 reads per sample from the cecal microbiota transplant (CMT) experiment.

Microbiota analysis

Prior to analysis, ASV tables were cleaned to remove spurious ASVs that had less than a total of 20 reads across all samples. Weighted and unweighted UniFrac distances were calculated using ASV relative abundance tables with phylogenetic trees and used to generate principal coordinate analysis (PCoA) ordination plots after removal of low-abundance ASVs (below 0.1% in all samples) using the phyloseq package (version 1.40.0) in R. ASV taxonomy was assigned using a pre-trained SILVA²⁸ classifier (silva-138-99-515-806-nb-classifier) in Qiime2. Richness and Shannon diversity metrics were calculated from ASV tables using the estimate_richness() function in phyloseq. Differential abundance of cecal genus-level features of male Gpr41^-/- mice and their WT littermates was determined using MaAslin2 ²⁹ package (version 1.10.0) in R after removal of genera that were below an average relative abundance cutoff of 0.05%.

Cecal-content DNA samples from 10 Gpr41^-/- and 14 WT males were submitted to the University of Wisconsin-Madison Biotechnology Center. DNA concentration was verified using the Qubit® dsDNA HS Assay Kit (Life Technologies, Grand Island, NY). Samples were prepared according to the QIAGEN FX DNA Library Preparation Kit (QIAGEN). Quality and quantity of the finished libraries were assessed using an Agilent Tapestation (Agilent, Santa Clara, CA) and Qubit® dsDNA HS Assay Kit, respectively. Paired end, 150 bp sequencing was performed using the Illumina NovaSeq X Plus (Illumina, San Diego, CA). The resulting reads were trimmed with trimmomatic (version 0.39) and host DNA was removed using RSEM (version 1.3.1) and bowtie2 (version 2.3.4). To quantify bsh and bai genes, we collected the hidden Markov model (HMM) profiles of each gene from KOfamKOALA³⁰ (version 2023-10-02) and conducted a hmmsearch using HMMER (version 3.3.2) against the Mouse Gastrointestinal Bacterial Catalog³¹ (MGBC). After removing all redundant sequences, the final reference database consisted of 1435 genes, the vast majority of which were bsh. We quantified each gene by mapping the cleaned metagenomic reads to the reference database using RSEM (version 1.3.1) with the – bowtie2 and – very-sensitive parameters. The resulting FPKM count table was filtered based on the hmmsearch score. We initially chose the score threshold for each individual gene recommended by KOfamKOALA, or when not available, used the hmm score generated after running hmmsearch of the hmm profile against the C. scindens genome ASM2089211v1 (Supp. Table 1). However, this resulted in zero detection from of baiA, baiE, and baiF in any mice. We reasoned that the bai genes present in mice may be significantly distinct from those in the literature which are primarily found in human strains. Therefore, we reduced the score threshold by half for these genes (Supp. Table 1). Taxonomy contributions were assigned by tracing each hit back to its genome of origin in MGBC.

RNA isolation and construction of RNA-Seq libraries

Total RNA from the liver and ileum of five Gpr41^-/- and five WT males was extracted using the following procedure: Frozen tissue (20–100 mg) was added to 1 mL Trizol (Cat #15596026, Thermo Fisher Scientific, Waltham, MA) and immediately beat-beated for 2 minutes in a screwcap tube with 0.5 g of 1 mm diameter zirconium beads (Cat. #11079110ZX, Biospec Products, Bartlesville, OK). Homogenates were incubated at room temperature (RT) for five minutes and then 0.2 mL of chloroform was added and shaken by hand for 15 seconds and allowed to rest at RT for 3 minutes. Samples were then centrifuged at 12,000 × g for 15 minutes at 4 °C. A total of 500 µL of the aqueous phase was collected and mixed with 318 µL of chilled 100% ethanol. The entire volume was then added to the Qiagen RNeasy Mini kit (Cat. # 74004, Qiagen, Germantown MD) and processed according to the manufacturer's instructions. RNA extracts were submitted to the University of Wisconsin-Madison Biotechnology Center for service and RNA-Seq libraries were constructed as previously described.³² Briefly, oligo (dT) selection was used to select for mRNA followed by synthesis of cDNA with SuperScript II RT (Invitrogen, Carlsbad, California) and RNAse treatment. The cDNA was cleaned and purified using a SPRI bead kit (Beckman Coulter Genomics, Brea, CA), ligated to Illumina unique molecular index (UMI) adapters. The indexed product was then purified subjected to 10 cycles of PCR amplification using a genomic DNA primer set from Illumina. The resulting libraries were assessed for quality and quantity before sequencing.

Liver and ileum RNA-seq analysis

Sequencing was conducted on a single Illumina NovaSeq6000 run generating paired-end 150 bp sequences. The resulting PE reads were demultiplexed and cleaned with trimmomatic which yielded an average of 34.8 million PE reads per sample. Cleaned reads were then annotated against the Genome Reference Consortium (GRC) build 38 Mus musculusgenome using RSEM. This resulted in a table of transcripts per million reads (TPM) counts for genes assigned with an Ensembl gene ID. Transcripts that summed to less than 10 TPM were removed from further analysis. One Gpr41^-/- liver sample, which exhibited signs of contamination as it had a highly distinct expression profile from all other liver samples, was removed from further analysis. Differential expression analysis was conducted using the DESeq2 (version 1.36.0) R package. DEGs were considered significant if they had P-values < 0.01. All significantly up- or downregulated DEGs were used as input to the Enrichr package (version 3.2) to assess KEGG (“KEGG_2019_Mouse” database) pathway enrichment in WT (upregulated DEGs) and Gpr41^-/- (downregulated DEGs) in R, which generated a P-value for each KEGG pathway.

Liver histology

After sacrifice, liver samples were fixed using 10% formalin at room temperature for 48 hours, followed by storage in 80% alcohol at 4 °C. The fixed samples were embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E), Masson's trichrome, and picrosirius red. For histopathological analysis, sections were examined under light microscopy by a board-certified liver pathologist (YJL), who was blinded to the treatment groups. Hepatic steatosis was graded based on the percentage of the liver section occupied by fat vacuoles (grade 0 ≤ 5%; grade 1 = 5 to 33%; grade 2 = 34 to 66%; grade 3 ≥ 66%). Lobular inflammation (clusters of recruited immune cells in the lobules) was graded based on the foci of chronic inflammatory cells (grade 0 = none; grade 1 = 1-2 foci per 20× field; grade 2 = 2-4 foci per 20× field; grade 3 ≥ 4 foci per 20× field). Hepatocyte ballooning degeneration was graded based on the number of ballooned hepatocytes present in the sections (grade 0 = none; grade 1 = few; grade 2 = many). The stage of hepatic fibrosis was assessed using Masson's trichrome and picrosirius red staining (stage 0 = no fibrosis; stage 1a = zone 3 mild perisinusoidal fibrosis; stage 1b = zone 3 moderate perisinusoidal fibrosis; stage 1c = periportal/portal fibrosis only; stage 2 = zone 3 plus periportal/portal fibrosis; stage 3 = bridging fibrosis; stage 4 = cirrhosis). The grade and stage scoring system adopted here mimics the nonalcoholic fatty liver disease scoring system (NAS), which was designed for the evaluation of steatohepatitis and fibrosis in humans.³³

Experiment testing cholesterol content in diet

A standard natural ingredient base-formula diet (2018 global 18% protein diet, Inotiv) was supplemented with 0%, 1.5%, or 2% cholesterol (wt/wt). This diet was fed to PCSK9-AAV-infected C57BL/6J male mice for three weeks (n = 3–4 mice/group). After three weeks plasma was collected to assess total cholesterol and fresh fecal samples were collected to determine SCFA profiles as described above.

Cecal transplant experiment

Microbiota transplantation was conducted as previously described.²³ Briefly, Flash-frozen cecal contents from male Gpr41^-/- mice and their WT littermates were pooled from 4 to 5 mice in an anaerobic chamber (5% H₂, 20% CO₂, and 75% N₂) and mixed with anaerobic Mega Medium³⁴ (1 mL per 100 mg), vortexed to make a slurry, and sealed in a hungate tube to maintain an anaerobic atmosphere. The slurry preparations were then retrieved using an anaerobic-purged needle and syringe to limit oxygen exposure before immediately being used to gavage 6-week-old male germ-free C57/BL6 mice (0.1 mL per mouse). Germ-free mice were born and maintained in sterile isolators and fed autoclaved chow diet (LabDiet 5021; LabDiet, St. Louis, MO) and sterile water ad libitum until colonization. Upon colonization, mice were removed from the isolators in sterile cages, gavaged, and fed an irradiated high-plant polysaccharide, high-cholesterol (1.5% wt/wt) diet described above (TD.210423, Inotiv) ad libitum. One week later, mice were injected with 1.5 × 10¹¹ particles of PCSK9-AAV8 by RI as described above. A second slurry preparation and gavage was repeated one week after AAV injection. The mice were sacrificed 8 weeks post-injection. At sacrifice, the mice were fasted for 4 hours before euthanasia. Plasma was collected for cholesterol measurement, and cecal contents were flash-frozen for microbiota analysis. Fresh feces were collected 4 weeks post injection for microbiota analysis.

Gut transit assay

Transit time was determined as described previously.³⁵ Briefly, Gpr41^-/- and wild-type littermates (aged about 8–9 weeks) were maintained on a high-fiber and high-cholesterol diet (TD.210423) for 4 weeks prior to the assay. Mice received a single oral gavage of 0.3 mL of 6% nonabsorbable carmine red dye (Sigma) prepared in 0.5% methylcellulose (Sigma). They were singly housed with minimal bedding during the experiment. Chow diet was provided during and following the experiment. Gut transit time was measured as the duration (in minutes) between the gavage and the appearance of the first red fecal pellet. The experiment was conducted in three independent batches each consisting of 2–4 littermates per genotype.

Cholesterol excretion assay

One week after the transit time assay, Gpr41^-/- and wild-type littermates (aged ~13–14 weeks) were administered a single dose of cholesterol as a 20 mg bolus (0.4 mL from 50 mg/mL stock) prepared in coconut oil (Thermo Scientific). Following gavage, mice were singly housed with minimal bedding and provided chow diet during the experiment. All fecal pellets were collected and pooled from each mouse every hour from 0 through 8 hours, post-gavage. Fecal pools were homogenized by bead-beating and lipids were extracted using the Abcam kit (ab211044). Total cholesterol was quantified colorimetrically (Fujifilm, catalog number 999-02601) to examine the fecal cholesterol concentration at each timepoint and the cholesterol excretion rate. The cholesterol excretion rate was calculated by multiplying cholesterol concentration of each fecal pool by the pool's total mass to get total cholesterol excreted per hour.

Statistical analysis

The Student's T-test was used to compare means between genotypes within each strain and sex unless otherwise stated. A Wilcoxon rank-sum test was used for comparison of histology scores. Similarities of microbiota profiles were assessed by PERMANOVA using the adonis function in the vegan R package (version 2.6-4). Spearman's rank coefficients and P-values for associations between microbial features and host phenotypes were generated using the cor.test() function in the stats package in R (version 4.3.3). P-values for the differentially expressed gene transcripts (liver and ileum) and differentially abundant taxa were generated by R packages DESeq2 (1.36.0) and MaAslin2 (version 1.10.0), respectively. Enrichment analysis of KEGG pathways was conducted using a Fisher's exact test with the Enrichr package in R.

Results

GPR41 deficiency alters cardiometabolic phenotypes and SCFA production, but not atherosclerosis

We tested the role of SCFA receptors in atherosclerosis by comparing disease progression in mice lacking GPR41, GPR43, or GPR109A (Gpr41^-/-, Gpr43^-/-, and Gpr109a^-/-) with their respective wild-type (WT) littermates. Atherosclerosis was induced in male and female mice by increasing plasma cholesterol levels via administration of an adeno-associated virus (AAV) carrying a proprotein convertase subtilisin/kexin (PCSK9) payload (PCSK9-AAV) at 6 weeks of age (Figure 1a). Mice were maintained on a high plant polysaccharide (HPP) diet supplemented with 1.5% cholesterol (wt/wt) for 12 weeks. This diet was selected because it increases blood cholesterol in PCSK9-AAV infected C57BL/6J mice while simultaneously generating high levels of cecal SCFAs (Supp. Figure 1a–d). Deficiency in any of the three SCFA receptors (Gpr41^-/-, Gpr43^-/-, and Gpr109a^-/-) did not exacerbate atherosclerotic plaque size or lipid content relative to their WT counterparts in either sex (Figure 1b and c, Supplemental Results and Supp. Figure 2a–d). Interestingly, Gpr109a^-/- mice had reduced plaque lipid area, but not macrophage infiltration compared to their WT littermates (Supplemental Results and Supp. Figure 2e).

Figure 1.

Atherosclerotic and metabolic phenotypes of Gpr41^-/- mice and their WT littermates. (a) Schematic of the experimental design. Atherosclerotic plaque area (b) and lipid content (c) measurements by sex and genotype. Plasma total cholesterol (d), liver mass (e), gonadal white adipose tissue mass (f) as a proportion of body weight, liver mass (g) as a proportion of body weight, liver TAG content (h), and liver total cholesterol content (i). Sample size = 6–12/group. The level of significance is indicated with (*) P < 0.05 and (**) P < 0.01. ORO, oil red O; TC, total cholesterol; TAG, triacylglycerol; gWAT, gonadal white adipose tissue.

Unexpectedly, GPR41 deficiency resulted in multiple alterations associated with enhanced cardiometabolic health. These included reduced plasma total cholesterol (TC) in males and females relative to their respective WT littermates as well as reduced plasma triacylglycerol (TAG), gonadal white adipose tissue (gWAT) mass, and liver mass in Gpr41^-/- males compared to their WT littermates (Figure 1d–f). There were no differences in plasma high-density lipoprotein (HDL) between genotype in either sex (Supp. Figure 4g−i). The reduced liver mass in Gpr41^-/- males did not seem to be a result of differences in lipid accumulation as there was no difference in liver TAG or TC content, although female Gpr41^-/- mice had reduced TC content in the liver compared to their WT counterparts. Histological analysis of both male and female livers revealed no differences in steatosis, inflammation, or lipid ballooning between Gpr41^-/- and WT littermates (Supp. Figure 3a−d).

Considering the alterations of plasma TC and TAG observed in male Gpr41^-/- mice, we decided to conduct lipoprotein analysis on this group via fast protein liquid chromatography (FPLC) of pooled plasma samples. We found that Gpr41^-/- males had a reduced IDL/LDL-cholesterol fraction (Supp. Figure 3e). LDL and IDL particles have high cholesterol and TAG content, respectively, which mirrors the lower TC and TAG concentration found in the plasma. Taken together, these results suggest that GPR41 does not play an important role in SCFA-induced atheroprotection but does serve key roles, particularly in males, in lipid metabolism and adiposity in hypercholesterolemic mice.

Cecal propionate and secondary bile acid profiles are altered in male Gpr41^-/- mice

SCFAs affect energy balance through their role as signaling molecules as well as metabolic substrates, so we compared cecal acetate, propionate, and butyrate levels of Gpr41^-/- and WT littermates and found that propionate was significantly elevated with Gpr41 deficiency in male, but not female mice (Supp. Figure 5a–c). This is notable since propionate is among the strongest agonists of GPR41. We did not find differences in acetate or butyrate in either sex, nor did we detect significant changes in cecal SCFA between Gpr43^-/- and Gpr109a^-/- mice and their respective WT among either sex. We did not detect significant differences between genotypes for cecal isovalerate and isobutyrate (Supp. Figure 5d and e).

We also evaluated bile acids in the cecum and plasma since enterohepatic circulation of bile acids plays an important role in whole-body sterol flux. To assess how bile acid metabolism might be altered between genotypes, we assessed bile acid profiles of male WT and Gpr41^-/- mice. We measured plasma and cecal levels of bile salts (bile acids conjugated to taurine or glycine) and unconjugated bile acids by uHPLC/MS/MS. Deoxycholic acid (DCA), tauro-conjugated DCA (TDCA) as well as tauro-conjugated hyodeoxycholic acid (THDCA) were significantly reduced (P < 0.05) in the plasma of Gpr41^-/- mice compared to their WT counterparts (Figure 2a). There were no significant differences in either conjugated or unconjugated primary bile acid levels in the plasma. In the cecum, two primary conjugated bile acids were detected; tauro-β-muricholic acid (T-β-MCA) and taurocholic acid [TCA], neither of which differed between genotype. Four unconjugated bile acids, which result from microbial hydrolysis of primary conjugated bile acids, were detected in the cecum; chenodeoxycholic acid (CDCA), cholic acid (CA), β-muricholic acid (β-MCA), and ursodeoxycholic acid (UDCA) (Figure 2b). Among these, UDCA levels were significantly reduced and CDCA trended lower (P = 0.06) in Gpr41^-/- cecal contents compared to WT. In the cecum, eight microbially derived secondary bile acid species were detected, of which 5 were significantly reduced in Gpr41^-/- mice; lithocholic acid (LCA), 12-oxo-LCA, 3-oxo-DCA, hyodeoxycholic acid (HDCA), and DCA; and one, ω-muricholic acid (ω-MCA), trended down (Figure 2b). Notably, all of the bile acid species that were significantly different between genotype were reduced in the Gpr41^-/- mice relative to their WT counterparts. Moreover, all of these species were microbially-derived secondary bile acids. These results indicate that Gpr41 deficiency leads to a global reduction of circulating and intestinal unconjugated and secondary bile acids suggesting that the alterations in the bile acid pool are driven by microbial activity in the intestine. These results, along with the differences in propionate, suggest that GPR41 plays a role in influencing the structure and function of the intestinal microbiota.

Figure 2.

Plasma and cecal bile acid measurements in male Gpr41^-/- and their WT littermates. Concentrations of bile acids in the plasma (a) and cecal contents (b) from male WT and Gpr41^-/- mice. Comparisons of means was conducted via T-test between genotypes. Sample size = 6–11/group. The level of significance is indicated with (‡) P < 0.1, (*) P < 0.05, (**) P < 0.01, (***) P < 0.001. 12-oxo-LCA, 12-oxo-lithocholic acid; 3-oxo-DCA, 3-oxo-deoxycholic acid; 7-oxo-DCA, 7-oxo-deoxycholic acid; β-MCA, β-Muricholic acid; CDCA, Chenodeoxycholic acid; CA, Cholic acid; DCA, Deoxycholic acid; HDCA, Hyodeoxycholic acid; LCA, Lithocholic acid; T-α-MCA, Tauro_a_Muricholic acid; T-β-MCA, Tauro-β-Muricholic acid; TCDCA, Taurochenodeoxycholic acid; TCA, Taurocholic acid; TDCA, Taurodeoxycholic acid; THDCA, Taurohyodeoxycholic acid; TUDCA, Tauroursodeoxycholic Acid; UDCA, Ursodeoxycholic acid; ω-MCA, ω-Muricholic acid.

GPR41 regulates cecal microbiota composition in males

We next assessed the cecal microbiota composition and function in male Gpr41^-/- mice and their WT littermates. First, we characterized the cecal bacterial community composition by 16S rRNA gene amplicon sequencing. Principal coordinates analysis (PCoA) of weighted UniFrac distances showed that cecal community structures of Gpr41^-/- mice were highly distinct from their WT counterparts (PERMANOVA, P = 0.0001) (Figure 3a). This is in contrast to male Gpr43^-/- or Gpr109a^-/- mice which did not exhibit strong differences in microbial composition compared to their respective WT littermates (Supp. Figure 6a) highlighting GPR41's influence over the cecal microbiota. Compared to their WT littermates, Gpr41^-/- mice had a significant reduction in ASV richness and Shannon diversity index (Figure 3b and c). Differential abundance analysis showed that Gpr41-deficiency led to lower relative abundances of 17 genera belonging to the families Lachnospiraceae, Oscillospiraceae, Ruminococcaceae, Erysipelotrichaceae, Butyricicoccaceae, Rikenellaceae, Deferribacteraceae, and Clostridiales (Figure 3d). In contrast, only four genera, Bacteroides, Lactobacillus, Lachnospiraceae FCS020 group, and Streptococcus, were significantly higher in Gpr41^-/- mice (Figure 3d).

Figure 3.

Characterization of cecal microbiota composition of male WT and Gpr41^-/- littermates. (a) PCoA plot using weighted UniFrac distances of 16S rRNA ASV profiles from male WT and Gpr41^-/- cecal content. ASV richness (b), and Shannon diversity index values (c). (d) MaAsLin2 effect sizes (top axis, dots) and relative abundances (bottom axis, bars) of genera (abbreviated family name is bracketed) that were differentially abundant (P < 0.05). High-abundance taxa were those that averaged above 1% relative abundance across all mice, whereas Low-abundance taxa were those below 1% (0.01% minimum cutoff). The negative (left) direction indicates abundances in Gpr41^-/- mice (light blue bar) while the positive (right) direction indicates the abundance in WT mice (dark blue bar). MaAsLin2 effect sizes to the left of the origin denotes significantly (open dots = P < 0.05 but adjusted P > 0.1; solid dots = adjusted P < 0.1) greater abundance in Gpr41^-/- mice, whereas effect sizes to the right of the origin indicates significantly greater abundance in WT mice. (e) Weighted UniFrac PCoA of 16S rRNA profiles from all male WT littermates of the Gpr41, and Gpr43, and Gpr41 KO mice, respectively. (f) Spearman coefficients and significance between taxa and plasma TAG and total cholesterol using all male WT mice. Comparisons of means for alpha diversity metrics were conducted via T-test between genotypes, sample size = 8–12/group. The level of significance is indicated with P < 0.1, (*) P < 0.05, (**) P < 0.01, (***) P < 0.001. ASV, amplicon sequence variant; TAG, triacylglycerol.

We next searched for genera that were correlated with plasma TAG and total cholesterol levels. To do this, we included all male WT littermates from each SCFA-receptor KO group in our analysis (n = 32). The rationale behind this analysis was to capture as many mice with as much microbial diversity as possible to increase the power of the associations. Despite all being C57BL/6J, the WT mice from the different groups exhibited distinct microbial signatures (Figure 3c), likely a remnant from their vivarium of origin (see Methods). Spearman correlation analysis revealed eight significantly (P < 0.05) positive associations with plasma cholesterol and/or TAG; Bifidobacterium, Xylanophilum group, “uncultured” Lachnospiraceae genus, and Clostridia VadinBB60 were positively correlated with plasma cholesterol, Roseburia was positively associated with plasma TAG, whereas Lachnospiraceae A2, UCG-006, and Lachnoclostridium were positively correlated with cholesterol and TAG (Figure 3f). One genus, Odoribacter, was found to be negatively correlated with plasma cholesterol (Figure 3f). Notably, of the taxa that were positively associated with plasma lipids, five (Lachnospiraceae A2 genus, “uncultured” Lachnospiraceae genus, Roseburia, Lachnospiraceae UCG-006, and Clostridia VadinBB60) had lower abundances in Gpr41^-/- mice compared to their WT littermates (Figure 3d). This provides evidence that GPR41 selects for taxa that are correlated with increased circulating lipid concentrations.

Male Gpr41^-/- mice have a reduced genetic capacity for bile acid metabolism.

We conducted shotgun metagenomic sequencing analysis on cecal contents from male Gpr41^-/- mice and their WT counterparts to assess differences in microbial function potential, specifically abundance of bile acid-modifying genes. The first step of bile salt modification by intestinal microbes is hydrolysis of the amino acid to form unconjugated primary bile acids. This reaction is catalyzed by bile salt hydrolase (BSH) which is present in many gut bacterial species.³⁶ The subsequent modifications that produce DCA and LCA rely on the bile acid 7α-dehydroxylation encoded in the bai operon.³⁷ While this pathway has been extensively researched in human microbiota such as Clostridium scindens and C. hylemonae, assessing the abundances of homologous genes in mouse metagenomes is understudied.^24,38-40 To assess bile acid modifying genes in mice metagenomes, we used hidden Markov models (HMM) profiles of bsh and the genes in the bai operon (baiA2, baiB, baiCD, baiE, baiF, baiG, baiH, baiI, and baiN) to search for homologs within the Mouse Gastrointestinal Bacteria Catalog (MGBC)³¹ genome database. The resulting homologs were used as reference sequences to map the cecal metagenomic reads to quantify the cecal abundances of each gene. The abundance of bsh trended higher in Gpr41^-/- mice compared to WT mice (P = 0.093, Figure 4a). In contrast, Gpr41^-/- mice had significantly lower abundances of baiCD, baiE, baiF, baiI, and baiH than their WT counterparts (Figure 4b). Nearly all of the bai gene reads detected in these metagenomes were mapped to Dorea spp. in the Lachnospiraceae family along with other Lachnospiraceae genera (Supp. Figure 6b). Interestingly, the detected bai gene hits were highly distinct from those of the human strains C. scindens and C. hylemonae suggesting that Dorea and other Lachnospiraceae, instead of Clostridium (from the Clostridiaceae family), may dominate the 7α-dehydroxylation niche in mouse intestinal microbial communities. The bsh gene, on the other hand, was detected from a diverse collection of taxa across multiple phyla (Supp. Figure 6b). These findings indicate that microbiota in Gpr41^-/- mice have a reduced genetic capacity to make secondary bile acids, explaining their reduced concentrations in the cecum and plasma observed in Gpr41^-/- mice.

Figure 4.

Cecal abundances of bile acid modifying genes in male WT and Gpr41^-/- mice. Fragment per kilobase million (FPKM) counts of bsh (a), the gene coding for protein that removes the amino acid groups from conjugated bile acids, in the cecal metagenomes of male Gpr41^-/- mice and their WT littermates. FPKM counts of the bai operon (b) which encodes the 7α-dehydroxylation pathway that converts CA to DCA, and CDCA or UDCA to LCA. Comparisons of means were conducted via T-test between genotypes (n = 7–11/group). The level of significance is indicated with (‡) P < 0.1, (*) P < 0.05, (**) P < 0.01.

Male Gpr41^-/- mice show reduced expression of ileal Npc1l1 and other nutrient transporters

To assess how GPR41 deficiency may impact lipid metabolism and handling in the intestine and liver, we compared the hepatic and ileal transcriptome profiles between WT and Gpr41^-/- males. We observed 645 differentially expressed genes (DEGs; unadjusted P < 0.01) in the liver, with 445 upregulated in WT and 200 upregulated in Gpr41^-/- livers. Pathway enrichment analysis of liver DEGs showed that 44 KEGG pathways were significantly enriched (P < 0.01) in the WT livers, while only two (Retinol metabolism and Steroid hormone biosynthesis) were upregulated in Gpr41^-/- livers (Supp. Figure 7a). The WT-enriched pathways included various immune and inflammatory signaling processes, the most significant of which included Chemokine signaling pathway, Cytokine–cytokine receptor interaction, and TNF signaling pathway (Supp. Figure 7a). A subset of DEGs were present in multiple pathways enriched in WT mice, of which the most frequently represented DEGs were Nfkbia (NFKB inhibitor α), Fos (Fos proto-oncogene), Ccl2 (monocyte chemoattractant protein-1), Icam1 (intercellular adhesion molecule 1), and Tlr2 (Toll-like receptor 2) all of which are involved in immune signaling (Supp. Figure 7b). We also examined expression of the rate-limiting genes involved in cholesterol synthesis (Hmgcr, Sqle) and its regulation (Srebf1, Srebf2) but did not detect any significant (P < 0.05) differences although there was a trend toward reduced Hmgcr mRNA in the livers of Gpr41^-/- mice (P = 0.053, Supp. Figure 7c). These results suggest that the elevated LDL-cholesterol observed in WT mice is likely not due to differences in hepatic cholesterol production. These results also indicate that the vast majority of DEGs in the liver that are downregulated in Gpr41^-/- mice involve genes that are overrepresented in immune processes.

In the ileum, we detected 486 DEGs (P < 0.01) with 354 upregulated in WT mice and 132 upregulated in Gpr41^-/- mice. KEGG pathway enrichment analysis revealed that WT-enriched DEGs were overrepresented in 12 pathways (Figure 5a). Interestingly, two pathways, Cholesterol metabolism and Fat digestion and absorption pathways were enriched in both the WT and Gpr41^-/- small intestine (Figure 5a). Upon further examination of the DEGs that contributed to Cholesterol metabolism pathway enrichment in both genotypes, we observed overexpression of genes involved in cholesterol uptake and lipoprotein formation in WT mice (Cyp27a1, Abca1, Abcg8, Abcg5, Npc1l1, and Apoe), and genes involved in lipase activity and lipoprotein uptake in Gpr41^-/- mice (Lipg, Pcsk9, and Ldlr) (Figure 5b). Importantly, in the context of our PCSK9-AAV model, there was no difference in expression in the liver or circulating PCSK9 protein (Figure 2f). The DEGs that contributed to Fat digestion and absorption pathway enrichment between genotypes similarly centered around overexpression of lipid transport genes in WT mice (Abca1, Abcg8, Abcg5, and Npc1l1) and lipase genes in the Gpr41^-/- mice (Pnliprp2, Clps, and Acat2) (Figure 5d). NPC1L1 is the primary transporter of cholesterol in the intestine and can influence plasma cholesterol levels. Given that all mice were fed a high cholesterol diet, our results suggest that downregulation of Npc1l1 in the small intestine of Gpr41^-/- mice may contribute to the observed reduction in circulating cholesterol levels. Ileal expression of bile acid transporters Ost-α and Asbt were also reduced in Gpr41^-/- mice suggesting that Gpr41 deficiency results in a general reduction of sterol uptake (Figure 5e). Interestingly, various pathways that were enriched in the ileum of WT mice are linked by their involvement in temporal (Circadian rhythm) as well as nutrient uptake processes (Protein digestion and absorption, Retinol metabolism, Cholesterol metabolism, Vitamin digestion and absorption, Mineral absorption, Fat digestion and absorption) indicating that GPR41 may play a role in sensing nutrient availability and coordinating nutrient harvest.

Figure 5.

Transcriptomic analysis of ileum samples from male WT and Gpr41^-/- littermates. (a) Significantly enriched KEGG pathways (P < 0.05) in the ileum, expressed as the log of the P-value (leftward bars denoting WT and rightward bars denoting Gpr41^-/- P-values), resulting from pathway enrichment analysis of differentially expressed genes (P < 0.05) of male Gpr41^-/- mice and their WT littermates. Normalized expression (TPM) of individual genes involved in cholesterol metabolism (b), amino acid transport (c), fat digestion and absorption (d), and bile acid transport (e) with unadjusted P-values. Sample size = 5 per group.

Additionally, we noticed that a disproportionate number of DEGs in the ileum belonged to the solute carrier (Slc) family of genes which code for membrane transport proteins. Given the observation that WT mice had enriched expression of genes in the Protein digestion and absorption pathway, we compared the expression of Slc genes belonging to gene classes that are involved in amino acid and peptide transport (Slc1a, Slc6a, Slc7a, Slc16a, Slc36a, Slc38a, Slc43a, and Slc15). A total of 44 AA transport genes had expression in the ileum above a detection threshold of 100 TPM. Of these, seven (16%) were significantly differentially expressed (P < 0.05) between WT and Gpr41^-/- mice and all seven were upregulated in the WT mice (Figure 5c). This further supports the hypothesis that Gpr41^-/- mice have an altered capacity for nutrient uptake which may lead to differences in nitrogen availability in the luminal content between WT and Gpr41^-/- mice.

GPR41 deficiency alters expression of FXR gene targets in male mice

Consistent with the reductions in bile acids observed in Gpr41^-/- mice relative to WT, we also observed reduced expression of genes regulated by the transcription factor farnesoid X receptor (FXR) which is activated by bile acids.⁴¹ We observed significant (P < 0.05) downregulation of the FXR gene targets⁴² Shp, Fabp6, Baat, Ostα, Apoc3, Apoa1, Apoe, and Abcc2 in Gpr41^-/- mice in the ileum (Supp. Figure 8a). We also observed differences in liver expression of the FXR gene targets Vldlr, Abcb11, Insig2, Shp, Slc10a2, Apoc2, Pltp, and Sdc1 (Supp. Figure 8b). This may signify reduced FXR activation by bile acids in Gpr41^-/- mice; however, the expression patterns of these genes do not match canonical FXR regulatory patterns; for example, according to a previous study,⁴² reduced FXR activity would lead to a reduction in Shp, Fabp6, Baat, Ostα, Apoe, and Abcc2, with a concomitant increase in Apoa1 and Apoc3 expression, not a decrease as we observed. These discrepancies may be the result of a complex network of signals induced by secondary bile acids impacted by ablation of Gpr41.

Transplantation of male WT and Gpr41^-/- cecal microbiota did not result in distinct microbial communities or transfer of donor cholesterol phenotypes

To test whether alterations in the gut microbiota induced by GPR41 deficiency caused the differences in plasma cholesterol levels between genotypes, we conducted a cecal microbiota transplant (CMT) from male Gpr41^-/- mice and their WT littermates into germ-free C57BL/6J male mice infected with PCSK9-AAV (Supp. Figure 9a). Eight weeks after transplantation we tested plasma levels of total cholesterol but did not observe a difference between recipient groups (Supp. Figure 9b). Notably, 16S rRNA community compositions were indistinguishable between WT-CMT and Gpr41^-/--CMT recipient mice at both 4 weeks (fecal communities) and 8 weeks (cecal communities) post-transplant (Supp. Figure 9c,d). This suggests that Gpr41 deficiency is required to maintain the distinct community structures observed between Gpr41^-/- mice and their WT littermates.

A trend toward increased gut transit rate and cholesterol excretion was observed in GPR41 deficient mice

We next tested whether the reduced plasma lipids could be the result of altered gut motility and dietary lipid excretion. To test gut motility, we performed a gut transit assay by gavaging a non-absorbable carmine red dye in Gpr41^-/- male mice along with their male WT littermates and monitored the time taken for the appearance of the first red fecal pellet.⁴³ Our data shows a trend toward faster gut transit in Gpr41^-/- mice (P = 0.13, Supp. Figure 10a) which aligns with a previous report from Samuel et al. ³¹. To test whether this difference in transit rate was associated with alterations in cholesterol excretion we administered a 20 mg bolus of cholesterol to these mice and assessed cholesterol content in the feces. Gpr41^-/- mice exhibited a trend toward higher cholesterol excretion compared to wild-type littermates after receiving a 20 mg oral bolus of cholesterol (P = 0.15; Supp. Figure 10b and c). Together, these findings suggest that gut transit alone likely does not fully explain the differences in plasma cholesterol between WT and Gpr41^-/- mice but may be a contributing factor.

Discussion

SCFAs, particularly butyrate and propionate, have been shown to protect against CVD and atherosclerosis.^9,13,44,45 The aim of the current study was to assess whether the SCFA receptor GPR41, which senses both butyrate and propionate, mediates the protective effects of SCFAs against hypercholesterolemia and atherosclerosis. Using a PCSK9-AAV-induced hypercholesterolemia model in mice fed a high-fiber, cholesterol-rich diet, we found that global genetic ablation of Gpr41 did not potentiate atherosclerosis. We also evaluated GPR43 and GPR109A, but despite their role in ameliorating hypertension and promoting regulatory T cell differentiation^20,46,47 neither appeared to contribute to atheroprotection in this model. Unexpectedly, we found that GPR109A deficiency led to reduced plaque lipid content, indicating that GPR109A may promote atherosclerosis in our model. These findings suggest that SCFA receptor signaling is not required for atheroprotection under hypercholesterolemic conditions.

Another unexpected finding was that Gpr41^-/- male mice exhibited alterations in cardiometabolic phenotypes compared to their WT littermates, including reduced plasma TAG and LDL cholesterol, reduced fat mass, and reduced liver weight. Despite this, Gpr41^-/- mice did not have reduced atherogenesis suggesting that whole-body ablation of Gpr41 may induce proatherogenic effects that counteract the protective impact of reduced plasma LDL. The reduced adiposity observed in GPR41-deficient mice in the current study is consistent with a previous study by Samuel et al. ¹⁸, but differed from the results observed in another study by Bellahcene and colleagues.⁴⁸ This discrepancy may be explained by differences in diet, disease model, or breeding strategies⁴⁹ between studies.

We also detected sex-dependent effects of SCFA receptor deficiency on some metabolic phenotypes, most notably in plasma TAG (GPR41, GPR43) gonadal adiposity (GPR41, GPR109A) and plasma total cholesterol (GPR43). This may be a consequence of the interaction between cholesterol metabolism and steroidal hormones such as estrogen which are derived from cholesterol. Estrogen regulates adipose tissue function and circulating lipoproteins^50,51 so the loss of SCFA receptors, which also impacts energy homeostasis and lipid metabolism, may alter the effect of estrogen on lipid storage. These findings may reflect important sex-associated differences regarding the role of SCFAs in metabolic diseases which warrants further study.

Intestinal expression of genes involved in nutrient uptake were significantly downregulated in Gpr41^-/- males. These included Npc1l1, which encodes for the main cholesterol transporter, potentially explaining the mechanism underpinning the reduced plasma TC in Gpr41^-/- mice. In addition, we observed reduced expression of multiple transporters involved in amino acid absorption indicating that GPR41 deficiency leads to a state of reduced absorption. We also found that gut transit rate and fecal cholesterol excretion both trended higher in Gpr41^-/- mice further reflecting a hypo-absorptive state. In agreement with these results, a previous study demonstrated GPR41's role in promoting energy harvest and slowing intestinal transit in mice¹⁸ and another showed that propionate, the primary agonist of GPR41, reduces intestinal transit in guinea pigs.⁵² Together, these findings suggest that GPR41 coordinates intestinal motility in response to luminal nutrient availability via SCFA sensing to promote an absorptive state.

Microbiota profiling further revealed that GPR41 deficiency is associated with reduced diversity and altered community composition, including decreased abundance of SCFA-producing and bile acid–modifying taxa. This was accompanied by a significant reduction of secondary bile acids levels in both plasma and cecum. The genetic capacity for 7α-dehydroxylation (bai operon) was diminished in the Gpr41^-/- cecal microbial population, suggesting that GPR41 selects for bile acid-metabolizing microbiota. Notably, the bai operon genes detected in these mice were predominantly represented by Lachnospiraceae, especially Dorea spp., which is thought to be only a minor contributor of bai capacity in humans.⁵³ In line with this, we found evidence of reduced signaling by the bile acid-sensitive FXR transcription factor in Gpr41^-/- mice. FXR is activated by bile acids during a fed state,⁵⁴ so the altered FXR signaling associated with GPR41 deficiency may reveal an underlying mechanism by which GPR41 promotes absorption. GPR41 deficiency also led to an overall reduction of the secondary bile acid pool in the cecum and plasma but primary bile acid profiles were mostly unaltered. Although microbially derived deoxycholic acid (DCA), which we found to be reduced in Gpr41^-/- males, promotes adiposity and liver weight in mice,⁵⁵ we are unable to discern here whether this is a causal factor for the improved cardiometabolic state in GPR41-deficient mice. Further studies are needed to elucidate this.

Transplantation of male Gpr41^-/- gut microbiota into male WT germ-free mice failed to recapitulate the cholesterol-lowering phenotype. Moreover, the Gpr41^-/- microbial communities reverted to the WT conformation suggesting that GPR41 deficiency is required to maintain the differences in microbiota observed in the donor mice. This suggests that the metabolic effects in male Gpr41^-/- mice appear to arise from host-microbe interactions that require intact GPR41 function rather than being solely microbiota-driven.

Exogenous propionate has been shown to reduce intestinal NPC1l1 expression and atherosclerosis in mice,¹³whereas administration of the propionate prodrug tripropionin reduces plasma cholesterol and adiposity in mice fed a high-fat diet.⁵⁶ Our finding that GPR41, a propionate receptor, promotes Npc1l1 expression and cholesterol absorption appears to contradict these studies at first glance, however, the apparent discrepancy may be the consequence of GPR41-mediated mechanism that modulates propionate levels in the intestine. We observed higher cecal concentrations of propionate in Gpr41^-/- mice. This, in conjunction with our observations that Gpr41^-/- mice had higher levels of propionate-producing microbiota and faster gut transit, suggests that propionate-GPR41 signaling results in reduced gut motility and more host absorption while ultimately reducing microbial propionate levels, thereby creating a potential negative feedback loop on nutrient uptake (Supp. Figure 11). We propose that disrupting this equilibrium, either by loss of Gpr41 or by elevated exogenous propionate, propionate-NPC1L1 signaling predominates resulting in reduced cholesterol absorption. Further studies are needed to test this hypothesis.

Our study is limited by the use of whole-body KO mice which prevents the examination of the tissue-level effects of these receptors. Given their unique expression profiles in different organs,¹⁶ tissue-specific KOs will be important to dissect competing effects on lipid metabolism and disease. Additionally, since GPR41, GPR43, and GPR109A have some redundancy in sensing SCFAs, assessment of double or triple KOs for these receptors could better control for compensatory effects and detect synergies. Finally, we relied on PCSK9-AAV to induce hypercholesterolemia that may introduce PCSK9-related confounders. Although we did not find any evidence that plasma PCSK9 levels were explanatory of our lipid-metabolism phenotype observations, we cannot rule out that exogenous Pcsk9 could not have any off-target effects, since PCSK9 is known to interact with other surface receptors besides the LDL-receptor.⁵⁷

In summary, we demonstrate that GPR41 plays a previously unrecognized role in regulating intestinal absorption, microbial ecology, and systemic lipid levels. These effects are not sufficient to alter atherogenesis in this model, but they implicate GPR41 as a key integrator of host–microbiota signaling in metabolic regulation. Our data also raise the possibility that intestinal GPR41 coordinates intestinal programming toward an absorptive state, in line with the notion of SCFAs and their receptors are key regulators of intestinal energy harvest and that modulation of this axis could represent a therapeutic target for metabolic disease.

Supplementary Material

Supplementary material

Hutchison_Supp_Table1_Accepted

Supplementary material

Supplemental_Results_Accepted

Supplementary material

Supplementary figures

Funding Statement

This work was supported by the National Institutes of Health (NIH) grants HL144651 (FER), HL148577 (FER). ERH and LNL were supported in part by the Metabolism and Nutrition Training Program NIH T32 (DK007665). ERH was supported by the University of Wisconsin–Madison Food Research Institute (Robert H. and Carol L. Deibel Distinguished Graduate Fellowship in Probiotic Research). This work was also supported by a grant from a Transatlantic Networks of Excellence Award from the Leducq Foundation (17CVD01). Fondation Leducq.

Ethics declarations

We used ChatGPT to check for grammatical errors and improve conciseness in limited sections of the manuscript. None of the data, figures, findings, interpretation, or conclusions were generated or altered by AI.

Disclosure of potential conflicts of interest

The authors report there are no competing interests to declare.

Acknowledgments

We would like to thank Dr. Karen Ho at Northwestern University and Dr. Pamela Martin at Augusta University for generously donating the mice used for this study and Drs. Barbara Mickelson and Jake Lusis for their expertise in the diet formulation. We also extend our thanks to the University of Wisconsin–Madison Biotechnology Center's DNA Sequencing Facility (Research Resource Identifier – RRID:SCR_017759) and the Gene Expression Center for providing expertise and resources. Finally, we thank the University of Wisconsin–Madison Experimental Animal Pathology Lab and the Medical Sciences Center vivarium staff for their assistance with this study. Much of the current work is based on Chapter 4 of ERH's PhD dissertation which is publicly available on ProQuest.⁵⁸

Author contributions

ERH and FER conceived of and designed the study. ERH, MJ, JHB, and MMT bred the mice and conducted the mouse experiments. Tissues were collected by ERH, MJ, MMT, JHB, and KK. LNL, MJ, and DAN analyzed bile acids. YL conducted histological analysis of the livers. ERH and BP conducted FPLC analysis and VAL provided resources for atherosclerosis assessment. Gut transit and cholesterol absorption experiments were carried out by MJ and MMT. ERH conducted all statistical analyzes with assistance from MJ and QZ. The manuscript was written by ERH, MJ, and FER.

Data availability statement

The data used in this study is available upon request to the corresponding author. The sequencing data generated in this study is available for download from the NCBI Sequence Read Archive (SRA). The 16S rRNA gene sequencing data are available under accession number PRJNA1284957; Shotgun metagenomic sequencing data are available from the NCBI SRA under accession number PRJNA1285464; Liver and ileum RNA-seq sequencing data are available from the NCBI SRA under accession number PRJNA1285089.

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/19490976.2025.2598957.