Microbial Suppression of Gpr41 Expression.

Ligand-induced down regulation is a hallmark of GPCR activation (13). Therefore, we generated Gpr41−/− mice (Fig. S3 A and B), rederived them as GF, and examined whether colonization of 8- to 10-week-old male GF knockout (Gpr41−/−) mice and their WT (+/+) littermates (mixed C57Bl6/J:129/Sv background) for 28 days with Bt and Ms (see Methods) affected ileal expression of Gpr41, or three other known GCPRs that bind fatty-acid ligands: Gpr43, which, as noted above, is also responsive to short-chain (C2-C6) fatty acids, plus Gpr40 and Gpr120, which recognize ligands with longer (>C6) acyl chains (14–16).

Quantitative PCR assays established that levels of colonization of the distal gut (cecum) with each microbial species were not significantly affected by the presence or absence of Gpr41 (mean 8.2 ± 4.3 × 10¹² organisms per gram of luminal contents for B. thetaiotaomicron; 2.4 ± 1.5 × 10⁶ for M. smithii; n, 7–8 mice per genotype). Therefore, any phenotypic differences observed between gnotobiotic WT and knockout animals could not be attributed to differences in their gut microbial ecology.

qRT-PCR assays of ileal RNAs revealed that compared with GF +/+ controls, cocolonization of WT mice produced statistically significant twofold reductions in the steady state levels of Gpr41, Gpr43, and Gpr120 mRNAs (P < 0.05; ANOVA) (Fig. S4). Expression of Gpr40 mRNA in +/+ mice was also reduced by colonization, although the observed change did not quite achieve statistical significance (see Fig. S4). Importantly, the magnitude of the reduction in Gpr40, Gpr43, and Gpr120 gene expression was not affected by the absence of Gpr41 (see Fig. S4).

Together, these findings indicate that Gpr41−/− mice have a specific deficiency affecting only one of these four fatty-acid binding GPCRs and therefore can, in principle, be used to assess the role of Gpr41 in mediating the effects of the microbiota on host energy homeostasis.

Gpr41 Is Needed for Microbiota-Induced Increases in Host Adiposity.

Eight to ten-week-old male GF Gpr41−/− mice, maintained on a standard polysaccharide-rich chow diet, exhibited no significant differences in their epididymal fat pad or total body weights compared with +/+ littermates (P > 0.05; n = 8–14 per group) (Fig. 1A and data not shown). In contrast, Gpr41−/− mice cocolonized with Bt/Ms had significantly lower epididymal fat pad weights than +/+ controls (11.4 ± 0.6 versus 14.4 ± 0.9 mg/g body weight, respectively; P < 0.05) (see Fig. 1A), gained significantly less body weight per day (0.08 ± 0.03 versus 0.19 ± 0.02 g/day, respectively; P < 0.05) (Fig. 1B), and weighed significantly less at the end of the 28-day colonization period (24 ± 0.4g versus 26 ± 0.4g; P < 0.05) (n, 13 to 14 animals per group, representing two independent experiments).

Open in a separate window

Fig. 1.

Microbiota-mediated increase in adiposity is blunted in Bt/Ms cocolonized gnotobiotic Gpr41−/− mice. (A) Weights of both epididymal fat pads in Gpr41−/− and +/+ littermates that are GF, or raised GF and then colonized at 4 to 6 weeks of age with Bt and Ms (n, 8–14 males per group; three independent experiments). (B) Weight gain (n, 4–9 mice per group; followed one to two times per week for up to 4 weeks, between 5 and 9 weeks of age in the case of GF animals, and for 4 weeks after gavage in the case of Bt/Ms cocolonized gnotobiotic animals). Mean values ± SEM are plotted. *, P < 0.05, **, P < 0.01. N.S., not significantly different.

These gut microbiota-dependent differences were not a unique feature of the Bt/Ms gnotobiotic model. CONV-R Gpr41−/− animals, maintained on the same polysaccharide-rich, low-fat chow diet as their cocolonized gnotobiotic counterparts, also exhibited statistically significant decreases in weight gain, total body weight, and fat-pad weight compared with age- and gender-matched CONV-R +/+ littermates (P < 0.05) (Fig. 2 A and B plus data not shown). Dual energy X-ray absorptiometry (DEXA) confirmed their reduced adiposity (13 ± 1% versus 19 ± 1% of body weight in +/+ controls; P < 0.005; n, 9 to 13 per group) (Fig. 2C). The differences in body weight and adiposity observed in CONV-R Gpr41-deficient versus WT mice were not attributable to differences in their locomotor activity or body temperature (n, 4 animals per group; P > 0.05) (Fig. S5 A and B).

Open in a separate window

Fig. 2.

Microbiota-mediated increase in adiposity is blunted in CONV-R Gpr41−/− mice. (A) Epididymal fat-pad weights (values from both fat pads were combined for each animal; n, 12 males per group). (B) Weight gain (n, 6 per group; followed one to two times per week for up to 4 weeks, between the ages of 5 and 9 weeks). (C) Adiposity in CONV-R WT and knockout mice, defined by DEXA (n, 9–13 males per group; 8–10 weeks old). (D and E) Fasting (4 h) serum leptin levels plotted against percentage total body fat (n, 5 per group). *, P < 0.05, **, P < 0.01 (Student's t test).

Fasting serum levels of leptin were similar in GF Gpr41−/− and +/+ littermates, but significantly lower in Bt/Ms cocolonized and CONV-R Gpr41−/− animals (P < 0.05; n, 5–7 mice per genotype per treatment group) (Table S1). Moreover, serum leptin levels were significantly lower in CONV-R Gpr41−/− animals than would be expected based solely on the observed decrease in their adiposity (P = 0.02) (see Fig. 2 D and E). Together, these findings implicate Gpr41 in microbiota-dependent regulation of host adiposity and leptin production.

Loss of Gpr41 Is Associated with Increased Intestinal Transit Rate and Reduced Efficiency of Energy Harvest from the Diet.

Signals communicated from the gut to the brain via enteroendocrine-cell derived hormones are important regulators of satiety and energy balance. Compared with GF +/+ controls, we found that Bt/Ms-colonization of +/+ mice led to significantly increased circulating levels of peptide YY (PYY). This increase was significantly blunted in their Gpr41−/− littermates (n, 4–8 per group; P < 0.05) (Fig. 3A).

Open in a separate window

Fig. 3.

Increased rate of intestinal transit in Bt/Ms colonized gnotobiotic Gpr41−/− mice. (A) Fasting serum PYY levels in GF and Bt/Ms cocolonized, Gpr41−/− and +/+ mice (n = four to eight per group; two independent experiments; all samples assayed in duplicate). (B and C) Gut transit time measured by oral gavage of a fluorescently labeled nonabsorbable substrate (fluorescine isothiocyanate-dextran; 70,000 MW) in GF and cocolonized WT and Gpr41-deficient animals. (B and C) Distribution of fluorescence signal intensity 60 min after gavage along the length of the gut; data are plotted as a fraction of total fluorescence in B. No signal was observed in the ceca or colons in any of the animals. Panel (C) presents the calculated geometric center of the fluorescence (n = 4–8 animals analyzed per group). *, P < 0.05, **, P < 0.01, ***, P < 0.005.

Despite decreases in levels of several anorexigenic hormones (leptin, PYY), Bt/Ms cocolonized and CONV-R Gpr41−/− animals consumed equivalent amounts of chow as their cocolonized and CONV-R +/+ counterparts (n, 11–16 animals per group; P > 0.3) (Fig. 4A and Fig. S5C). Although energy input was similar, bomb-calorimetric assays of feces demonstrated that the efficiency of caloric extraction from the diet was significantly reduced in cocolonized Gpr41-deficient versus +/+ animals (4.5 ± 0.1 versus 3.4 ± 0.2 kcal/g; n, 6–7 per group; P < 0.0001) (Fig. 4B).

Open in a separate window

Fig. 4.

Gnotobiotic Gpr41−/− mice extract fewer calories from their polysaccharide-rich chow diet and excrete more SCFAs. (A) Dietary energy intake in GF and Bt/Ms cocolonized +/+ and Gpr41−/− littermates (calories of chow consumed per day monitored concurrently with body weight, as in Fig. 1). (B) Bomb calorimetry-based assessment of remaining calories in the feces of colonized (Bt/Ms) Gpr41−/− and +/+ mice (n = 6–7 per group). (C) GC-MS assay of cecal SCFA levels (n = 5–7 per group; each sample assayed in duplicate). (D) GC-MS study of fecal SFCAs from the same mice assayed in (C). *,P < 0.05, **, P < 0.01, ***, P < 0.005. (Student's t test used for comparisons of two groups.)

PYY is also a regulator of gut motility. It produces a dose-related inhibition of transit rate along the length of the gut (17). To test whether Gpr41-deficiency and the associated decrease in PYY is accompanied by an increased transit rate, we gavaged GF and Bt/Ms-colonized Gpr41−/− and +/+ mice with a nonabsorbable fluorogenic marker (FITC-dextran; MW 70,000). Although no statistically significant differences in gastric emptying rates were observed (data not shown), intestinal transit rate was significantly faster in Bt/Ms-colonized Gpr41−/− versus +/+ littermates (P < 0.005; see Fig. 3 B and C). The effect of Gpr41-deficiency on intestinal transit rate was microbiota-dependent: No significant differences were noted among GF animals of either genotype (n, 4–8 per group) (see Fig. 3 B and C). The differences between Bt/Ms-colonized Gpr41−/− and +/+ littermates were not attributable to differences in the length of their small intestines, which were similar (1.8 ± 0.06 versus 1.9 ± 0.07 cm/g body weight, respectively; n, four to eight animals per group; P > 0.05).

Based on the observed increase in intestinal transit rate, we hypothesized that more undigested polysaccharides may reach the distal gut in Gpr41−/− versus +/+ mice. This was confirmed by microanalytic biochemical assays of glucans (glucose-containing polysaccharides) in cecal contents. Bt/Ms-colonized, Gpr41-deficient animals had significantly higher cecal glucan levels than their colonized +/+ littermates (19% increase; 2.2 ± 0.1 versus 1.8 ± 0.1 μmoles/g dry weight of cecal contents; P < 0.05; n, five to six per group; P < 0.05), and higher levels of monomeric glucose (28% increase; 2.7 ± 0.2 versus 2.1 ± 0.2 μmol/g dry weight of cecal contents; P < 0.05). In addition, GC-MS-based analyses of cecal SCFA levels revealed that the concentrations of propionate and acetate were significantly increased in Bt/Ms-colonized Gpr41−/− mice (n, 5–7 animals per group; P < 0.05) (Fig. 4C).

Follow-up whole-genome transcriptional profiling of Bt using (i) microbial RNA isolated from the cecal contents of Bt/Ms-colonized Gpr41-deficient and +/+ littermates, and (ii) custom Bt GeneChips containing probe sets specific for >98% of the organism's protein-coding genes, failed to reveal statistically significant differences in the expression of bacterial genes involved in fermentation of polysaccharides to SCFA between the two groups of animals (n, 5 per group; P < 0.05; FDR <1%). Follow-up qRT-PCR assays of these RNAs confirmed that several key genes in the pathway involved in SCFA production, including pyruvate formate lyase (BT4738; EC 2.3.1.54), and acetate kinase (BT3963; EC 2.7.2.1) exhibited no significant differences in their expression between Bt/Ms colonized Grp41−/− and +/+ mice (data not shown).

GC-MS analysis of feces indicated that Bt/Ms-colonized Gpr41−/− animals had a statistically significant (37 ± 9%) increase in total SCFAs compared with their +/+ littermates (n, 5–7 per group; P < 0.05) (Fig. 4D). In contrast, fecal levels of free C6-C18 fatty acids (FFA) and triglycerides were not significantly different between the two groups of mice [0.6 ± 0.1 versus 0.7 ± 0.1 mg/g weight of feces (FFA), and 1.9 ± 0.2 versus 1.9 ± 0.1 mg/g weight of feces (triglycerides); n = seven to eight animals assayed per group; P > 0.05]. qRT-PCR assays of distal small intestinal (ileal) RNA (SI Methods and Table S2) indicated that there were no significant differences in expression of key host genes involved in the active uptake or transepithelial transport of lipids [e.g., Mct1 (mono-carboxylate transporter), CD36 (lipid-binding glycoprotein) or ApoB (chylomicron-mediated transport); data not shown]. These findings are not surprising, given that the majority of long-chain fatty acids are absorbed in the proximal intestine.

We reasoned that the observed increase in fecal SCFAs reflects, at minimum, reduced host-passive absorption. SCFAs stimulate, and are substrates for, de novo lipogenesis in the liver. Bt/Ms-colonization of GF +/+ animals resulted in statistically significant increases in liver triglyceride levels (n, seven to eight per group; P < 0.05) (Fig. S6A). This effect of colonization was not seen in Gpr41−/− animals, nor were any differences observed between GF Gpr41−/− and +/+ mice (n, 7–8 per group; P > 0.05) (see Fig. S6A). qRT-PCR assays confirmed reduced expression of fatty acid synthase (Fas) in the livers of Bt/Ms-colonized Gpr41−/− mice compared to their +/+ littermates (77 ± 12% lower; n, 5–7 per group; P < 0.01) (Fig. S6B). In addition, fasting (4 h) serum triglyceride levels were decreased in Bt/Ms-colonized Gpr41-deficient versus +/+ animals (n, 7–8 per group) (see Table S1). These differences were not attributable to alterations in hepatic expression of genes involved in long-chain fatty-acid transport, trafficking, or fatty-acid reesterification (data not shown). Together, our results indicate that Gpr41-deficient mice have reduced hepatic lipogenesis, consistent with reduced intestinal absorption and delivery of SCFAs.

Husbandry of Gnotobiotic Mice.

Gpr41−/− mice and their +/+ littermates (mixed C57BL/6J:129/Sv background) were rederived as GF and housed in flexible film plastic gnotobiotic isolators (19), where they were maintained on a strict 12-h light cycle (lights on at 0600 h) and fed a standard autoclaved polysaccharide-rich chow diet (B and K Universal) ad libitum. Four- to six-week-old male mice were inoculated with a single gavage of 10⁸ CFU of the sequenced human gut-derived B. thetaiotaomicron strain VPI-5482 [harvested from overnight culture in TYG medium (6)], and M. smithii strain PS [cultured at 37°C for 5 d in 125-ml serum bottles containing 15 ml of MBC medium (8) supplemented with 3 g/L formate, 3 g/L acetate, and 0.3 ml of a freshly prepared oxygen-free solution of filter-sterilized 2.5% Na₂S (note that the remaining volume or headspace in the bottle contained a 4:1 mixture of H₂ and CO₂, and the headspace was rejuvenated every 1 to 2 d) (8)]. All colonized mice were killed 28 d after gavage. The density of colonization was determined in cecal contents using qPCR assays that used species-specific primers (8). Age-matched, male CONV-R WT and Gpr41-deficient male mice were also fed the same autoclaved polysaccharide-rich chow diet ad libitum as the gnotobiotic animals. All experiments performed with gnotobiotic mice used protocols approved by the Washington University Animal Studies Committee.

Analysis of Host Adiposity and Energy Harvest.

All mice were fasted (4 h) before being killed. Epididymal fat pads, livers, and segments of the distal intestine (ileum) and colon were removed and flash-frozen in liquid nitrogen. Epididymal fat pad and liver weights were recorded before freezing.

Before killing, total body fat content was measured by DEXA (Lunar PIXImus Mouse, GE Medical Systems), 5 min after mice had been anesthesized with an intraperitoneal injection of ketamine (10 mg/kg body weight) and xylazine (10 mg/kg). Weight gain and chow consumption were monitored weekly in mice who were individually caged for the duration of the experiment. The energy content of fecal samples (freeze-dried immediately after collection) was defined using a bomb-calorimeter (Parr Instruments) and previously established methods (8).

Measurement of Physiological Parameters.

Locomotor activity and body temperature were assessed for 5 d using a telemetry device (minimitter PDT-4000; Mini Mitter) beginning 7 d after implantation (20). Locomotor activity data were processed using VitalView software (Mini Mitter).

Gastric emptying and gastrointestinal transit time was measured in GF and Bt/Ms-colonized Gpr41−/− and +/+ littermates using established methods, after an 18-h overnight fast (21, 22). FITC-labeled dextran (70,000 MW; Molecular Probes) was administered by gavage (100 μl of a 5-mg/ml solution prepared in PBS). Sixty minutes later, the entire GI tract from stomach to rectum was removed and placed in ice-cold PBS for 30 s to inhibit motility. The stomach, small intestine (divided into 10 equal length segments), cecum, and colon (subdivided into two equal-length segments) were each placed in a separate tube containing 1 ml of PBS (5 ml for stomach and cecum). The segments were coarsely chopped with scissors, and luminal contents suspended using a combination of vigorous washing and vortexing. A dilution series was completed for each sample (1:10 to 1:1000 in PBS) and the fluorescent signal quantified in a multiwell fluorescence plate reader (Stratagene Mx3000; excitation at 485 nm; emission at 530 nm). A histogram of the fluorescence signal distributed along the gastrointestinal tract was then plotted and the geometric center determined (SUM [% of total fluorescence per segment × segment number])/100) (23). Gastric emptying was calculated based on the amount of FITC-dextran left in the stomach compared with the total amount of fluorescence in the intestine.

The methods used for measurement of serum proteins, cecal, and fecal SCFAs, plus triglycerides are described in the SI Materials and Methods.

The authors thank Sabrina Wagoner, David O'Donnell, Maria Karlsson, Trey Coleman, Shelley Dixon, Randal Floyd, Amber Skach, and Marcus Thornton for technical assistance; Peter Turnbaugh, Peter Crawford, and Robert Heuckeroth for helpful advice; and Laura Kyro for graphics support. This work was supported in part by Graduate Research Fellowship DGE-0202737 from the National Science Foundation (to B.S.S.), by Grants DK70977 and DK30292 from the National Institutes of Health, and by grants from the W.M. Keck Foundation, Exploratory Research for Advanced Technology program of the Japan Science and Technology Agency, and The Howard Hughes Medical Institute.