Research Article

Maternal gut microbiota in pregnancy influences offspring metabolic phenotype in mice

See allHide authors and affiliations

Science  28 Feb 2020:
Vol. 367, Issue 6481, eaaw8429
DOI: 10.1126/science.aaw8429

Mouse mothers transfer metabolic mode

Obesity and metabolic diseases tend to go together, and humans who become obese are also prone to type 2 diabetes and cardiovascular problems. Starting with the observation that offspring of germ-free mice tended to become obese on high-fat diets, Kimura et al. investigated how the presence of the microbiota might be protective in mice (see the Perspective by Ferguson). Short-chain fatty acids (SCFAs) from the microbiota are known to suppress insulin signaling and reduce fat deposition in adipocytes. Further experiments showed that SCFAs in the bloodstream were able to pass from a non–germ-free mother's gut microbiota across the placenta and into the developing embryos. The authors found that in the embryos, the SCFA propionate mediates not only insulin levels through GPR43 signaling but also sympathetic nervous system development through GPR41 signaling. A high-fiber diet promoted propionate production from the maternal microbiota, and maternal antibiotic treatment resulted in obese-prone offspring.

Science, this issue p. eaaw8429; see also p. 978

Structured Abstract

INTRODUCTION

In recent decades, the rapid expansion of antibiotic use and intake of high-calorie, low-fiber diets have contributed to disturbances in the gut microbial community, predisposing humans to various diseases such as metabolic syndrome. Although the influence of microbiota on the postnatal environment has been well documented, much less is known regarding the impact of gut microbiota at the embryonic stage. Although accumulating evidence supports the notion of the developmental origins of health and disease (DOHaD), the underlying mechanisms remain obscure. In this study, we explored the impact of maternal gut microbiota on embryonic development and disease susceptibility late in life.

RATIONALE

Gut microbiota–derived metabolites represented by short-chain fatty acids (SCFAs; e.g., acetate, propionate, and butyrate) not only fuel host cells but also serve as signaling molecules between the gut microbiota and extraintestinal organs. GPR41 and GPR43 belong to the free fatty acids receptor (FFAR) family and are receptors for SCFAs. We have previously corroborated the biological importance of FFARs in energy metabolism through interactions with dietary ingredients, as well as gut microbiota–derived metabolites. Gut microbial SCFAs regulate host energy homeostasis via GPR41 and GPR43 in the sympathetic nervous system, adipose tissues, pancreas, and intestine. A recent study further showed that the gut microbiota of pregnant mice influence immune and brain functions of offspring. These findings raise the possibility that maternal SCFAs play a key role in the regulation of disease susceptibility during postnatal life in the context of the DOHaD theory.

RESULTS

We found that maternal microbiota during pregnancy imparts resistance to obesity to their offspring. Pregnant mice were bred under specific pathogen–free (SPF) and germ-free (GF) conditions, after which newborns were raised by foster mothers under conventional conditions to align growth environments after birth. The offspring from GF mothers were highly susceptible to metabolic syndrome characterized by an exacerbation of obesity and glucose intolerance in association with reduced energy expenditure upon high-fat diet consumption during adulthood. A similar phenotype was also observed in offspring from mice fed a low-fiber (LFi) diet during pregnancy. Treatment of pregnant GF or LFi-fed mice with SCFA rendered adult offspring resistant to obesity. SCFA in the colonic lumen of pregnant mice reached the embryos via the maternal liver and bloodstream. Notably, the sympathetic nerves, intestinal epithelium, and pancreas of embryos highly expressed GPR41 and/or GPR43 to sense SCFAs originating from the maternal gut microbiota. Deficiency in embryonic GPR41 and GPR43 signaling compromised energy metabolism because of sympathetic dysfunction and hyperglycemia during the prenatal period. The SCFA-GPR41 and SCFA-GPR43 axes facilitate the development of neural cells, GLP-1–expressing enteroendocrine cells, and pancreatic β cells, thereby shaping embryonic energy metabolism. This developmental process contributes to maintaining postnatal energy homeostasis.

CONCLUSION

We determined that, during pregnancy, the maternal gut microbiota confers resistance to obesity in offspring via the SCFA-GPR41 and SCFA-GPR43 axes. During pregnancy, SCFAs from the maternal gut microbiota were sensed by GPR41 and GPR43 in the sympathetic nerve, intestinal tract, and pancreas of the embryo, influencing prenatal development of the metabolic and neural systems. These findings indicate that the maternal gut environment during pregnancy is a key contributor to metabolic programming of offspring to prevent metabolic syndrome. Thus, the gut microbiota of pregnant mice provides an environmental cue that fine-tunes energy homeostasis in offspring to prevent the developmental origin of metabolic syndrome.

During pregnancy, maternal gut microbiota influences offspring propensity for obesity via embryonic SCFA receptors.

The maternal gut microbial SCFA-embryonic GPR41 and GPR43 axes facilitate the development of neural cells, GLP-1–expressing enteroendocrine cells, and pancreatic β cells to shape the development of energy metabolism in offspring, even as adults. GPCR, G protein–coupled receptor.

Abstract

Antibiotics and dietary habits can affect the gut microbial community, thus influencing disease susceptibility. Although the effect of microbiota on the postnatal environment has been well documented, much less is known regarding the impact of gut microbiota at the embryonic stage. Here we show that maternal microbiota shapes the metabolic system of offspring in mice. During pregnancy, short-chain fatty acids produced by the maternal microbiota dictate the differentiation of neural, intestinal, and pancreatic cells through embryonic GPR41 and GPR43. This developmental process helps maintain postnatal energy homeostasis, as evidenced by the fact that offspring from germ-free mothers are highly susceptible to metabolic syndrome, even when reared under conventional conditions. Thus, our findings elaborate on a link between the maternal gut environment and the developmental origin of metabolic syndrome.

The gut microbiota substantially contributes to energy extraction from indigestible polysaccharides, such as oligosaccharides and soluble dietary fibers, and influences host energy homeostasis during infancy and adulthood (16). Therefore, maintenance of the symbiotic microbial community is of paramount importance for host health. Changes in microbial composition may cause dysbiosis, leading to disease-prone phenotypes in the host. Dysbiosis has been implicated in a growing number of systemic disorders, including metabolic syndrome (79). Obesity is a major risk factor for metabolic syndrome, predisposing patients to cardiovascular disease and type 2 diabetes. In addition to genetic and epigenetic factors, changes in the gut microbiota have been implicated in the development of obesity, in combination with dietary factors. Gut microbiota–derived metabolites represented by short-chain fatty acids (SCFAs; e.g., acetate, propionate, and butyrate) (914) not only fuel host cells but also serve as signaling molecules between the gut microbiota and extraintestinal organs.

We have previously shown that SCFAs suppress insulin signaling in adipocytes and ultimately inhibit fat deposition via adipose GPR43 (15). Furthermore, GPR41 stimulation by propionate potently activates sympathetic neurons to regulate energy expenditure (16). A recent study further showed that feeding pregnant mice a high-fiber or acetate-supplemented diet decreases offspring susceptibility to allergic airway disease (17). These findings raise the possibility that maternal SCFAs play a key role in the regulation of disease susceptibility during postnatal life in the context of the developmental origins of health and disease theory (18). However, the underlying mechanisms and the biological importance of the maternal-embryonic cross-talk via microbial metabolites remain obscure.

In this study, we explored the impact of the maternal gut microbiota on energy homeostasis in offspring in a mouse model. We observed that the offspring of germ-free (GF) mothers are more prone to obesity and glucose intolerance than those of specific pathogen–free (SPF) mothers. Maternal microbiota–derived SCFAs translocated to the embryos to facilitate development of the sympathetic nervous system and regulation of insulin levels via GPR41 and GPR43 signaling. Thus, during pregnancy, the gut microbiota provides an environmental cue that fine-tunes energy homeostasis in offspring.

Offspring of GF mothers and obesity development

To investigate the impact of the maternal gut microbiota during pregnancy on offspring, pregnant mice were bred under SPF and GF conditions. On day 18.5 of gestation, pregnant GF mice received a fecal microbiota transplant from SPF mice of the corresponding strain to prevent overgrowth of unfavorable microbes. Newborn animals were raised by foster mothers under conventional conditions to align growth environments after birth. After weaning, the male mice were fed a high-fat diet (HFD) to induce obesity (Fig. 1A). Although the postnatal body weight of newborns from GF ICR mothers was less than that of offspring from SPF ICR mothers (fig. S1A), the offspring developed marked obesity upon HFD consumption during growth (Fig. 1A). In addition, the weight of perirenal or subcutaneous white adipose tissue (WAT) and the liver was significantly higher in offspring derived from GF mothers (GF offspring) than in those from SPF mothers (SPF offspring) at 16 weeks (Fig. 1A), which is in accordance with increases in WAT adipocyte size and hepatic triglycerides (TGs) (fig. S1, B and C). Concomitantly, plasma glucose, TGs, non-esterified fatty acids (NEFAs), and total cholesterol levels were significantly elevated in GF offspring (Fig. 1B). Body temperature and heart rate were significantly reduced (Fig. 1C), whereas plasma insulin levels and pancreatic islet sizes were significantly higher in GF offspring than in SPF offspring (fig. S1D). Moreover, the GF offspring exhibited elevated food intake (fig. S1E), with reduced plasma levels of the gut hormone peptide YY (PYY) and glucagon-like peptide-1 (GLP-1) (Fig. 1D) as well as reduced energy expenditure (Fig. 1E). These results indicate that the GF offspring exhibited an obese phenotype upon HFD feeding. In support of this interpretation, HFD-induced glucose intolerance and insulin resistance were significantly accelerated in GF offspring (Fig. 1F), indicating impaired insulin sensitivity. Notably, female GF offspring also exhibited similar phenotypes (fig. S2, A to H).

Fig. 1 Offspring from GF mothers exhibit severe obese phenotype when fed a HFD.

(A) Experimental scheme (left). Body weight changes during the HFD trial (middle) (n = 14 animals per group) and tissue weights (right) (n = 7 to 9 tissues per group). epi, epididymal; peri, perirenal; sub, subcutaneous. (B) Plasma glucose, TGs, NEFAs, and total cholesterol levels (n = 6 to 11 plasma samples per group). mEq, milliequivalents. (C) Body temperature (left) (n = 7 or 8 animals per group) and heart rate (right) (n = 7 animals per group). (D) Gut hormone PYY (left) (n = 6 or 8 plasma samples per group) and GLP-1 (right) (n = 6 plasma samples per group) levels. (E) Energy expenditure (n = 7 or 8 animals per group). (F) Glucose tolerance test (left) and insulin tolerance test (right) (n = 8 animals per group). Male mice were analyzed at 16 weeks of age. Student’s t test; **P < 0.01 and *P < 0.05. Data are presented as means ± SEM. Offspring from three or four litters per group were used. SPF, conventional offspring derived from ICR SPF mothers; GF, conventional offspring derived from ICR GF mothers.

Mammalian neonates are initially exposed to the vaginal microbiota, which substantially contributes to the establishment of the gut microbial community in infants (19) and thus potentially influences the development of the host metabolic system. However, 16S ribosomal DNA (rDNA) amplicon sequencing showed that the relative abundance of bacterial families constituting the gut microbiota was similar in offspring from SPF and GF ICR mothers during adulthood, although GF offspring in infancy showed significant decreases in Streptococcaceae and Enterococcaceae compared with SPF offspring (fig. S3A). Principal coordinate analysis based on weighted UniFrac distances confirmed that there were no differences between the two groups during adulthood, but not during infancy (fig. S3B). To exclude the possible influence of the vaginal microbiota, newborns from SPF and GF mothers were also delivered by caesarean section. Consistent with vaginally delivered offspring, caesarean GF offspring exhibited severe obesity upon HFD feeding during growth (fig. S4A). WAT and liver weight, plasma metabolic parameters, and insulin levels were also significantly higher in the caesarean GF offspring at 16 weeks than in the caesarean SPF offspring (fig. S4, B to D). Furthermore, the former showed reduced energy expenditure (fig. S4E). Additionally, compositions of the gut microbiota were similar between caesarean-delivered offspring from the SPF and GF mothers during adulthood, whereas infant microbiota may be affected slightly by the delivery modes (fig. S5, A and B), consistent with the findings of a recent study (19). Collectively, these data show that the gut microbiota in adulthood is not a primary factor in the obesity-prone phenotype of the GF offspring.

Interaction between the gut microbiota and host genetics modulates the predisposition to obesity, whereas environment factors such as diet are indispensable for the regulation of host-microbe interaction (2022). Although gut microbial compositions tended to be different between offspring from ICR and C57BL/6J mothers (fig. S6, A and B), they showed a similar obesity phenotype (Fig. 1, A to F and figs. S7, A to H). Furthermore, both male and female GF offspring from C57BL/6J mothers developed severe obesity and metabolic disorders upon HFD feeding (figs. S7, A to H, and S8, A to H). Thus, to a greater or lesser extent, metabolic disorders in GF offspring were commonly observed regardless of strain and sex.

Sensing maternal SCFAs in the embryo

We could not detect bacteria in the amniotic fluid of pregnant SPF ICR mice at our animal facility (fig. S9, A and B). We hypothesized that metabolite signals (2325), derived from the maternal gut microbiota, may translocate to the fetus and influence the development of the metabolic system. Thus, we profiled the plasma levels of hydrophilic and lipophilic metabolites in SPF and GF ICR mothers and their embryos during pregnancy. The levels of 5 metabolites in the mothers and 12 metabolites in the embryos were significantly different between the SPF and GF groups (fig. S10); among them, only 5 metabolites showed similar changes in the mothers and embryos in response to breeding conditions (Fig. 2A). In particular, the plasma levels of SCFAs (acetate, propionate, and butyrate) were significantly lower in GF mothers and embryos than in their SPF counterparts (Fig. 2B). Given that plasma SCFA levels were constant in both mothers and embryos of the SPF group during pregnancy (fig. S11), the maternal gut microbiota appears to constitutively supply SCFAs to embryos via the bloodstream.

Fig. 2 Embryonic SCFAs depend on maternal gut microbiota, with receptors already expressed at the embryonic stage.

(A) Volcano plot showing the significance and magnitude of differences in the relative abundances of plasma metabolites in mothers or embryos from ICR SPF and GF mothers (n = 5 plasma samples of mothers and n = 5 plasma samples of embryos from five litters per group). 1, 1,5-anhydro-d-glucitol; 2, 13-hy-6c,9c-18:2. (B) Levels of plasma SCFAs, as determined by GC-MS [n = 8 plasma samples of mothers (Mo) and n = 8 plasma samples of embryos from eight litters per group]. (C) Expression of Gpr41, Th, Nes, and Gfap during growth in the SCG (n = 8 tissues from four litters per group). (D) Expression of colonic Gpr41, Gpr43, Gcg, Pax4, and Pax6 during growth (n = 7 or 8 tissues from four litters per group). (E) Expression of pancreatic Gpr41, Gpr43, Ins2, and Nkx6.1 during growth (n = 7 or 8 tissues from four litters per group). Values are shown relative to 18S rRNA gene expression [(C) to (E)]. Student’s t test (B); **P < 0.01. Data are presented as means ± SEM [(B) to (E)].

Regulation of host energy metabolism through activation of GPR41 and GPR43 by gut microbial SCFAs has been documented (15, 16, 26, 27). We detected Gpr41 mRNA in the sympathetic ganglia of embryos (Fig. 2C and fig. S12, A and B) with biphasic expression in the embryonic and adult stages (Fig. 2C). Such an expression pattern was not observed for nestin (Nes; an undifferentiated neural marker), tyrosine hydroxylase (Th; a sympathetic neuronal marker), or glial fibrillary acidic protein (Gfap; a glial marker) (Fig. 2C). Meanwhile, Gpr43 mRNA was detected in the intestinal tract from embryonic day 15.5 (E15.5) onward (fig. S12C), although its expression was restricted to enteroendocrine cells in the adult stage (fig. S12D), as previously reported (28). Biphasic increases in Gpr43 expression were observed in the colon in the embryonic and adult stages, with expression peaking later than that of Pax4 and Pax6 (regulators of intestinal enteroendocrine cell differentiation) and at a similar time point to that of Gcg (intestinal enteroendocrine cell maker, GLP-1) (Fig. 2D). These SCFA receptors are also expressed in the pancreas and regulate insulin secretion in adult mice (27, 29). Levels of pancreatic Gpr43 mRNA, but not Gpr41, transiently increased during the perinatal-postnatal period, with expression peaking later than that of Nkx6.1 (early β cell differentiation–related factor) and at a similar time point to that of Ins2 (pancreatic β cell maker) (Fig. 2E). Notably, Gpr43 mRNA expression in both the colon and pancreas of GF embryos was significantly lower than that in SPF embryos, although Gpr41 mRNA expression in the sympathetic ganglia of embryos was similar among the two groups (fig. S12, E to G). These findings imply that the embryonic metabolic tissues, such as the sympathetic nervous system, intestinal tract, and pancreas, may sense maternal gut microbe–derived SCFAs by expressing GPR41 and GPR43.

Sympathetic development via GPR41

We further investigated the functions of GPR41 in the sympathetic nervous system and of GPR43 in the intestine and pancreas during embryonic development. We found that sympathetic nerve projections to the heart were significantly reduced in Gpr41−/− C57BL/6J embryos compared with wild-type (WT) embryos, and such an abnormality was also apparent in GF embryos with the same background (Fig. 3A). Moreover, sympathetic nerve projections to the heart were significantly reduced in Gpr41−/− pups on postnatal day 1 (P1), even though these pups were delivered from mothers maintained under SPF conditions (Fig. 3B). Gut microbial compositions were similar for WT, Gpr41−/−, and Gpr43−/− pregnant C57BL/6J mice (fig. S13, A and B). Plasma SCFA levels were also comparable among the three groups (fig. S13C).

Fig. 3 Propionate promotes sympathetic neuronal differentiation via embryonic GPR41.

(A) TH expression in E18.5 whole heart (n = 7 or 8 tissues of embryos from three or four litters per group) and representative Western blots. (B) Sympathetic nerve projection in P1 whole heart (upper panels) (TH immunostaining) and ventricle (lower panels) (TH, green; DAPI, blue). TH density was quantified (n = 9 or 10 tissues of pups from four litters per group). (C) Effects of propionate (1 mM) on sympathetic neuronal differentiation in Gpr41−/− mice (TH, red; DAPI, blue; n = 10 independent experiments from three biological replicates per condition). (D) Development of nerve projections in the P1 ventricle quantified by Western blotting (n = 8 to 11 tissues of pups from three litters per group) and representative Western blots. Abx., antibiotic treatment. (E) Heart rate (left) (n = 7 to 11 animals from three to five litters per group) and body temperature (right) (n = 8 animals from three or four litters per group) in 4-week-old offspring. (F) Energy expenditure in 4-week-old offspring (n = 8 animals from three litters per group). Student’s t test [(A) and (B)] and Tukey–Kramer’s test [(C) to (F)]; **P < 0.01 and *P < 0.05. NS, not significant. Data are presented as means ± SEM.

We further sought to investigate the contribution of GPR41 signaling to sympathetic neuronal differentiation. In a primary culture of embryonic superior cervical ganglion (SCG)–derived neural cells, each of the tested SCFAs, especially propionate, significantly promoted sympathetic neuronal differentiation (fig. S14A). This finding is consistent with the notion that the most potent agonist of GPR41 is propionate, followed by butyrate and acetate, with median effective concentration (EC50) values of approximately 10, 40, and 2000 μM, respectively (30). Propionate-induced sympathetic neuronal differentiation was abolished in sympathetic neural cells from Gpr41−/− embryos (Fig. 3C). Furthermore, Gi/o-mediated (but not Gi/oα-mediated) GPR41 propionate signaling increased sympathetic neurite length via Gβγ-mediated mitogen-activated protein kinase activation (fig. S14, B and C). These findings strongly suggest that propionate-mediated GPR41 activation promotes sympathetic neuronal differentiation. To test this notion, we treated WT pregnant mice with a cocktail of antibiotics to eliminate intestinal microbiota. Sympathetic nerve projections to the heart were significantly attenuated in pups from antibiotic-treated mice on P1. Notably, this abnormality was ameliorated by administration of propionate during pregnancy (Fig. 3D). Heart rate and body temperature after weaning, an index of sympathetic nerve projection, exhibited a similar trend, as did oxygen consumption (Fig. 3, E and F). Treatment with tyramine markedly decreased oxygen consumption in offspring from untreated or propionate- or antibiotic-treated WT mothers, whereas these effects were weakened in offspring from antibiotic-treated WT mothers or untreated Gpr41−/− mothers (Fig. 3F). On the basis of these observations, we reasoned that propionate from the maternal gut microbiota facilitates sympathetic nerve development via GPR41. Deprivation of propionate resulted in sympathetic dysfunction, including a reduction in body temperature and heart rate fluctuations, as observed in the GF offspring (Fig. 1C).

Embryonic insulin regulation via GPR43

GPR43 is expressed in adult enteroendocrine L cells and promotes the secretion of gut hormones upon activation (28). Because Gpr43 was also detected in the embryonic colon (fig. S12C), we investigated the effect of GPR43 on enteroendocrine cell differentiation at the embryonic stage. Pax4 and Pax6 were significantly up-regulated in the colon of Gpr43−/− C57BL/6J embryos in comparison with WT embryos (Fig. 4A and fig. S15A). Such changes were also observed in GF embryos with the same background. By contrast, Gcg as well as GLP-1 were down-regulated in Gpr43−/− and GF mice (Fig. 4A and fig. S15A). This result suggests a retardation of enteroendocrine cell differentiation in Gpr43−/− and GF mice. To directly assess the role of the SCFA-GPR43 axis in enteroendocrine cell differentiation, we employed intestinal organoids, which recapitulate cell differentiation in vivo by reducing canonical Wnt signaling (fig. S15B). Propionate, which is a more potent (EC50: ~30 μM) GPR43 ligand than acetate or butyrate (EC50: ~50 or 100 μM, respectively) (31), significantly promoted differentiation of GLP-1+ enteroendocrine cells in intestinal organoids from WT embryos (Fig. 4, B and C). In sharp contrast, this effect was abrogated in organoids from Gpr43−/− embryos (Fig. 4, B and C). Notably, the effect of propionate on enteroendocrine cell differentiation was not observed in organoids from adult WT mice (fig. S15C), illustrating that propionate-mediated GPR43 signaling is a prerequisite for the development of enteroendocrine cells during the prenatal period.

Fig. 4 Propionate promotes enteroendocrine and pancreatic β cell differentiation via embryonic GPR43.

(A) Pax4, Pax6, and Gcg mRNA expression and GLP-1 protein levels in the colon (E18.5) (n = 7 or 8 tissues of embryos from three litters per group). (B and C) Expression of Gcg (B) and distribution of GLP-1–positive cells (C) after 24-hour treatment of embryonic organoids (E15.5) in the presence (+) or absence (-) of Wnt with GPR43 ligands [1 and 10 mM (B); 100 μM propionate (C)]. GLP-1, green; DAPI, blue; n = 5 to 7 independent experiments from two biological replicates per condition. (D) Expression of Nkx6.1 and Ins2 mRNA and insulin protein levels in the mouse pancreas (E18.5) (n = 7 or 8 tissues of embryos from three litters per group). (E) Expression of Ins2 (left) and localization (middle; immunostaining) of insulin, and insulin positive cell count (right) in the presence or absence of GPR43 ligands (10 μM PA-1, 1 mM propionate) after 72-hour treatment (insulin, green; DAPI, blue; n = 4 to 10 independent experiments from two or three biological replicates per condition). For induction of β cell differentiation, cells were cultured with betacellulin and activin A. (F) Plasma insulin (left) and plasma glucose (right) levels in mothers and embryos of E18.5 (n = 8 plasma samples and n = 8 plasma samples of embryos from eight litters per group). GF and SPF mice with C57BL/6J background were analyzed [(A), (D), and (F)]. Student’s t test [(A), (D), and (F)] and Tukey–Kramer’s test [(B) and (E)]; **P < 0.01 and *P < 0.05. NS, not significant. All data are presented as means ± SEM.

Because GPR43 regulates insulin secretion in pancreatic β cells (27, 29) and this receptor was expressed in the pancreas during the perinatal-postnatal period (Fig. 2E), we subsequently investigated the effect of GPR43 on β cell differentiation during embryonic development. Nkx6.1 was significantly up-regulated in the pancreas of embryos from Gpr43−/− and GF mice compared with those from WT and SPF mice, respectively, whereas Ins2 expression and insulin level were down-regulated in Gpr43−/− and GF mice (Fig. 4D and fig. S15D). Moreover, during the induction of differentiation of the rat pancreatic tumor cell line AR42J into β cells, expression levels of Gpr43 mRNA, but not Gpr41 mRNA, were significantly elevated in line with those of Ins2 (fig. S15E). Notably, propionate and the synthetic GPR43 agonist phenylacetamide-1 (PA-1) promoted differentiation into insulin+ β cells; however, this effect was compromised by RNA interference–mediated knockdown of Gpr43 (Fig. 4E and fig. S15F). Thus, propionate-mediated activation of GPR43 facilitates differentiation into pancreatic β cells as well as GLP-1–positive enteroendocrine cells. GF ICR as well as GF C57BL/6J embryos also demonstrated abnormalities in these differentiation and functional markers of sympathetic neuron, enteroendocrine, and pancreatic β cells (fig. S15, A, D, and G).

These findings raise the possibility that SCFA-GPR43 signaling may play a vital role in embryonic insulin secretion. Plasma insulin levels in Gpr43−/− embryos were markedly lower than in WT embryos, although there were no differences between WT and Gpr43−/− mothers (Fig. 4F). Likewise, reduced insulin levels were also observed in GF embryos (fig. S16, A and B). Correspondingly, plasma glucose levels in Gpr43−/− (Fig. 4F) and GF embryos (fig. S16, A and B) were significantly higher than those in their control groups. Given that dysregulation of fetal glucose levels renders offspring susceptible to metabolic syndromes, such as obesity and type 2 diabetes (32), we speculate that the lack of SCFA-GPR43 signaling during the prenatal period causes metabolic syndrome in adulthood, most likely by compromising energy homeostasis in the embryos. Such retardation of the differentiation of the sympathetic nerve, intestinal tract, and pancreas was also evident in Gpr41−/−Gpr43−/− double-mutant C57BL/6J mice (fig. S17, A to E).

Dietary fiber intake during pregnancy

To provide further evidence of the importance of SCFAs in the developmental origin of obesity resistance, we performed a dietary intervention study in which pregnant ICR mice were fed a high-fiber (HFi) or low-fiber (LFi) diet under conventional conditions, after which the susceptibility of their offspring to obesity was examined (Fig. 5A). Although the body weight of postpartum offspring from HFi-fed mothers (HFi offspring) was significantly higher than that of offspring from LFi-fed mothers (LFi offspring) (fig. S18A), HFi intake suppressed the HFD-induced body weight gain from 13 weeks of age onward, in accordance with reduced subcutaneous WAT and liver weights (Fig. 5A). However, the effect of a HFi diet was abrogated when antibiotics were administered to pregnant mice to eradicate the gut microbiota (Fig. 5A and fig. S18A), indicating that microbial fermentation of dietary fiber contributes to obesity suppression. Plasma metabolic parameters were also improved in HFi offspring compared with LFi offspring (Fig. 5B and fig. S18, B and C). Likewise, HFi offspring were resistant to HFD-induced glucose intolerance and insulin resistance (fig. S18D), in association with improved energy expenditure (fig. S18E). Furthermore, sympathetic dysfunction, such as reduction in body temperature and heart rate fluctuations, in LFi offspring was ameliorated in HFi offspring (Fig. 5C). Compositions of the gut microbiota were comparable between LFi offspring and HFi offspring during adulthood but were significantly different between the two groups during infancy (fig. S19, A and B). Metabolome profiling of maternal and embryonic plasma samples revealed that 11 and 4 metabolites were significantly altered between the LFi- and HFi-fed groups (fig. S20), with 4 metabolites commonly increased in the HFi-fed mothers and their embryos (Fig. 5D). Among these 4 metabolites, SCFAs were the only common factor in the SPF versus GF and LFi versus HFi comparisons. SCFA levels were significantly higher in the embryos of HFi-fed mice (HFi embryos) than in those of LFi-fed mice (LFi embryos) (Fig. 5E). We also observed that plasma insulin levels were significantly higher in HFi embryos than in LFi embryos (Fig. 5F), and thereby plasma glucose levels were significantly decreased in HFi embryos (Fig. 5F). Thus, SCFAs generated by the maternal gut microbiota through the fermentation of dietary fiber are provided to embryos via maternal circulation, improving fetal glucose homeostasis and imparting resistance to obesity in the offspring.

Fig. 5 Dietary fiber supplementation of pregnant mothers results in offspring resistance to obesity.

(A) Experimental scheme (left). Body weight changes during HFD trial (middle) (n = 8 to 13 animals from four litters per group). Tissue weight (right) (n = 8 to 13 tissues from four litters per group). (B) Plasma glucose, TGs, NEFAs, and total cholesterol levels (n = 8 to 10 plasma samples from four litters per group). (C) Body temperature (left) (n = 7 or 8 animals from three litters per group) and heart rate (right) (n = 8 or 10 animals from three litters per group). (D) Volcano plot showing the significance and magnitude of differences in the relative abundances of plasma metabolites in the mothers or embryos from LFi and HFi mothers (n = 5 plasma samples and n = 5 plasma samples of embryos from five litters). 1, galactose. (E) Plasma SCFAs determined by GC-MS (n = 8 plasma samples of mothers and n = 8 plasma samples of embryos from eight litters per group). (F) Plasma insulin (left) and plasma glucose (right) levels in the mothers and embryos of E18.5 (n = 8 plasma samples of mothers and n = 8 plasma samples of embryos from eight litters per group). Male mice were analyzed at 16 weeks of age [(A) to (C)]. Student’s t test [(B), (C), (E), and (F)] and Tukey–Kramer’s test (A); **P < 0.01 and *P < 0.05 (LFi versus HFi); ##P < 0.01 and #P < 0.05 (HFi versus HFi + Abx). Data are presented as means ± SEM [(A) to (C), (E), and (F)]. LFi, offspring derived from ICR LFi mothers; HFi, offspring derived from ICR HFi mothers [(A) to (C)].

SCFA supplementation during pregnancy

Plasma propionate levels in HFi embryos were likely sufficient to activate the GPR41 and/or GPR43 receptors, considering that the determined values were superior to their EC50 values (Fig. 5E). Positron emission tomography (PET) imaging showed that [11C]-labeled propionate in the colonic lumen reached the embryos via the maternal liver and bloodstream within 40 min after infusion (fig. S21, A to D). Thus, to rigorously examine the role of propionate in obesity resistance of the offspring, we fed pregnant ICR mice a LFi diet supplemented with propionate (Fig. 6A). The intake of this diet raised plasma levels of propionate in both mothers and embryos (fig. S22A). Treatment with propionate suppressed the HFD-induced increases in body weight, perirenal or subcutaneous WAT mass, and liver weight of adult offspring (Fig. 6A). Plasma metabolic parameters were also improved in the offspring of propionate-treated mothers (Pro offspring) compared with control LFi offspring (Fig. 6B and fig. S22, B and C). HFD-induced glucose intolerance and insulin resistance were also significantly ameliorated in Pro offspring (fig. S22D), and energy expenditure was improved in Pro offspring (fig. S22E). Furthermore, sympathetic dysfunction in the LFi offspring was rescued in Pro offspring (Fig. 6C). Compositions of the gut microbiota were similar between LFi offspring and Pro offspring during both infancy and adulthood (fig. S23, A and B). Additionally, maternal intervention with propionate reversed the retardation of sympathetic nerve projections to the heart as well as the retardation of GLP-1+ enteroendocrine cell and pancreatic β cell differentiation in embryos of LFi-fed mothers (Fig. 6, D and E); it also enhanced plasma insulin levels in the embryos, restoring them to levels comparable to those of HFi embryos (Fig. 6F). Consequently, an increase in plasma glucose levels in control embryos was efficiently suppressed in embryos of mothers given propionate (Fig. 6F). Consistent with the offspring from HFi-fed mothers treated with antibiotics, offspring from HFi-fed GF ICR mothers recapitulated the obesity-prone phenotype, which was rescued by administration of propionate during pregnancy (fig. S24, A to H). Together, these observations define the importance of maternal propionate, which renders offspring resistant to obesity.

Fig. 6 Propionate supplementation of pregnant mothers results in offspring resistance to obesity.

(A) Experimental scheme (left). Body weight changes during HFD trial (middle) (n = 11 to 13 animals from four litters per group). Tissue weight (right) (n = 11 to 13 tissues from four litters per group). (B) Plasma glucose, TGs, NEFAs, and total cholesterol levels (n = 8 to 13 plasma samples from four litters per group). (C) Body temperature (left) (n = 6 or 8 animals from four litters per group) and heart rate (right) (n = 6 or 9 animals from four litters per group). (D) TH expression in E18.5 whole heart (n = 8 tissues of embryos from three litters per group) and representative Western blots. (E) Gcg mRNA expression and GLP-1 protein levels in the colon (E18.5) (left) (n = 8 tissues of embryos from three litters per group). Expression of Ins2 mRNA and insulin protein levels in the pancreas (E18.5) (right) (n = 8 tissues of embryos from three litters per group). (F) Plasma insulin (left) (n = 8 plasma samples of mothers and n = 8 plasma samples of embryos from eight litters per group) and plasma glucose (right) (n = 8 plasma samples of mothers and n = 8 plasma samples of embryos from eight litters per group) levels in mothers and embryos (E18.5). Male mice were analyzed at 16 weeks of age [(A) to (C)]. Student’s t test [(A) to (F)]; **P < 0.01 and *P < 0.05. Data are presented as means ± SEM. LFi, offspring derived from ICR LFi mothers; propionate, offspring derived from ICR propionate-supplemented mothers [(A) to (C)].

Discussion

In this study, we determined that maternal gut microbiota during pregnancy confers resistance to obesity in offspring via the SCFA-GPR41 and SCFA-GPR43 axes. During pregnancy, SCFAs from the maternal gut microbiota are sensed by GPR41 and GPR43 in the sympathetic nerve, intestinal tract, and pancreas of the embryo. SCFAs are known to exert pleiotropic effects through several mechanisms (10, 11, 13, 15, 16, 24, 3336) such as histone deacetylase (HDAC) inhibition by butyrate [median inhibitory concentration (IC50): ~90 to 170 μM] (37, 38) and activation of GPR109A by butyrate (EC50: ~700 μM) (39) and Olfr78 by acetate (EC50: ~2300 μM) and propionate (EC50: ~1000 μM) (33), in addition to GPR41 and GPR43. Considering the low concentrations of SCFAs (acetate: ~400 μM, propionate: ~50 μM, and butyrate: ~10 μM) in the embryonic circulating plasma, SCFAs were unlikely to interact with Olfr78 and GPR109A. Meanwhile, the concentrations of SCFAs were sufficient to activate GPR41 and GPR43 (30, 31). Activation of embryonic GPR41 and GPR43 by SCFAs promoted sympathetic neuronal, enteroendocrine, and pancreatic β cell differentiation, which were essential for maintaining energy homeostasis (e.g., thermogenesis and heart rate) via the sympathetic nervous system and fetal glucose homeostasis. Furthermore, given that the expression of Gpr43 tended to be down-regulated in the pancreas and colon of GF offspring, SCFAs such as butyrate and propionate [which also serves as an HDAC inhibitor (40)] may regulate embryonic development by regulating Gpr41 and Gpr43 gene expression through epigenetic modifications.

Several metabolites other than SCFAs showed similar changes in the plasma of mothers and embryos between SPF and GF conditions. However, in HFi-fed mothers and their embryos, only SCFAs were commonly elevated compared with those in the LFi-fed counterparts. Treatment of LFi-fed mothers with propionate repaired the defects in the differentiation of intestinal enteroendocrine cells and sympathetic neurons in the embryos, and the effects of propionate were abolished under Gpr41 and Gpr43 deficiencies. These observations underscore the contribution of the SCFA-GPR41 and SCFA-GPR43 axes in prenatal development of the metabolic and neural systems. Additionally, although offspring from GF-ICR and GF-C57BL/6J mothers developed obesity, the severity of the obese phenotype was slightly different between the two groups. Such variation may be attributed to gut microbial composition, which is substantially affected by host genetic backgrounds (20, 22). Consistent with this notion, we found that plasma SCFA levels differed between ICR and C57BL/6J mothers (Fig. 2B and fig. S13C).

Whereas compositions of the gut microbiota during adulthood were comparable in the SPF and GF offspring, compositions in infancy were different between the two groups. A similar trend was also observed in infant HFi offspring compared with LFi offspring, raising the possibility that altered infant microbiota partially contributes to development of the obesity phenotype of GF and LFi offspring. Nevertheless, propionate administration to LFi mothers during pregnancy ameliorated the obesity-prone phenotype of the offspring without affecting the infant microbiota (fig. S23). Furthermore, propionate treatment also improved hyperglycemia in GF embryos and lower energy expenditure in LFi and GF offspring. Although GPR41 deficiency also caused lower energy expenditure, propionate administration during pregnancy failed to prevent this phenotype. On the basis of these observations, we reason that maternal microbiota–derived SCFAs, particularly propionate, play a vital role in preventing the development of metabolic disorder in offspring. Meanwhile, it should be noted that GF offspring exhibited the abnormality in energy expenditure in both light and dark cycles, whereas LFi offspring displayed the abnormality only in the dark cycle. Therefore, we cannot formally exclude the possibility that the presence of intestinal microbiota and/or their products, other than SCFAs, may contribute to the enhancement of energy expenditure in the light cycle.

Embryonic insulin regulation was impaired in embryos from GF mothers, and insulin levels were significantly elevated in the adult stage. Excessive insulin levels in adult GF offspring are most likely attributed to metabolic adaptations in response to low birth weight and the retardation of pancreatic β cell differentiation, eventually increasing susceptibility to obesity upon HFD feeding. This abnormality is reminiscent of catch-up growth, whereby children born small for gestational age (SGA) face a risk of excessive body weight gain and metabolic syndrome later in life (18, 41), as evidenced by multiple birth cohort studies (4244). Although the etiology of SGA births remains to be clarified, maternal factors, including malnutrition, smoking, and alcohol consumption, have been implicated (45). We further propose that deprivation of SCFAs as a result of disturbances in intestinal microbiota may be another causative factor for SGA births, leading to catch-up growth and susceptibility to obesity. Our study provides evidence for the crucial contribution of the maternal gut environment during pregnancy to the metabolic programming of offspring to prevent metabolic syndrome. This finding opens new research avenues into preemptive therapies for metabolic disorders by targeting the maternal gut microbiota.

Materials and methods

Animal study

All animal diets unless otherwise indicated were provided by Research Diets. GF ICR (Sankyo Labo Service) and C57BL/6J (Clea Japan) mice were housed in vinyl isolators under a 12-hour light-dark cycle and given regular chow (CMF, Oriental Yeast). For fecal microbiota transplantation, pregnant GF mice (day 18.5) received oral gavage of fecal suspension in phosphate-buffered saline (PBS) from strain-matched pregnant SPF mice. In caesarean section experiments, progesterone (2 mg per mouse) was subcutaneously injected into pregnant GF or SPF IQI mice (Clea, Japan) on days 17.5 and 18.5 to avoid preterm delivery. On day 19.5, newborns were delivered by caesarean section and reared by foster IQI mothers under conventional conditions for 4 weeks. Then, 4-week-old male mice were fed a HFD (D12492) for 12 weeks and housed individually (table S1). In separate experiments, conventional ICR mice (SLC, Inc.) were fed AIN-93G formula–based LFi or 10% inulin–containing HFi diets during pregnancy (table S2). To eliminate commensal bacteria, HFi-fed ICR mice were treated with 1 mg/ml neomycin (Nacalai Tesque) in drinking water during pregnancy. Offspring were reared by CMF-fed conventional foster mothers for 4 weeks; weaned mice were fed a HFD for 12 weeks and housed individually. In SCFA-supplementation experiments, conventional ICR mice were given an AIN-93G–based diet supplemented with 5% (w/w) propionate for the pregnancy period (table S3). Offspring were reared by CMF-fed conventional foster mothers for 4 weeks. Over the course of the experiments, newborns [litter size aligned to 10 ± 2 (ICR) or 6 ± 2 (C57BL/6J) mice] from the GF and SPF mothers were reared by strain-matched foster mothers under conventional conditions for 4 weeks. Then, more than three groups of littermates from each mother were analyzed in individual experiments.

Heart rates were determined in conscious mice using a tail-cuff system (Softron and Muromachi Kikai), and body temperature was determined in conscious mice using Lifechip and a pocket reader (Destron Fearing), as described elsewhere (16).

C57BL/6J-background WT, Gpr41−/−, and Gpr43−/− mice and Gpr41−/−Gpr43−/− double-mutant pregnant mice were given CMF. Gpr41−/− and Gpr43−/− mice were generated as described previously (15, 16); Gpr41−/−Gpr43−/− double-mutant mice were generated by the CRISPR-Cas9 system (fig. S17, A and B). Two guide RNAs were identified using the publicly available optimized CRISPR design tool (http://crispr.mit.edu). Transgene insertion was established by homology-directed DNA repair recombination, and mutations were analyzed using genotyping primers. Pregnant WT and Gpr41−/− mice were treated with propionate (150 mM) and antibiotics (1 mg/ml neomycin) in drinking water. After birth, pups were analyzed on P1.

Plasma samples were obtained from embryos (6 to 12 per litter) delivered by WT and mutant mice at E18.5 and subjected to metabolome analysis. Plasma glucose concentrations were measured in individual embryos at E18.5. The SCG, heart, colon, and pancreas were collected from WT or mutant mice at E16.5, E18.5, P1, P7, P14, P28, or P49.

All experimental procedures involving mice were performed according to protocols approved by the Institutional Animal Care and Use Committee of Keio University School of Medicine [permission no. 09036-(12)]; the Committee on the Ethics of Animal Experiments of the Tokyo University of Agriculture and Technology (permit no. 28–87); and the Animal Care and Use Committee, Okayama University (permission no. OKU-2018346). Mice were treated with lethal anesthesia with somnopentyl, and all efforts were made to minimize animal suffering.

Histology

Livers, pancreas, and whole embryos were embedded in OCT compound (Sakura Finetek), and adipose tissues were embedded in paraffin. These samples were sectioned into 7-μm-thick slices that were stained with oil red O (Sigma-Aldrich) or hematoxylin and eosin for microscopic examination. Sections and cells were fixed in 4% paraformaldehyde and immunostained using primary antibodies raised against tyrosine hydroxylase (TH) (Millipore) to detect sympathetic neurons, nestin (BD Biosciences) to detect undifferentiated neural cells, GLP-1 (Abcam) to identify enteroendocrine cells, and insulin (Sigma-Aldrich) to identify differentiated pancreatic exocrine cells. This was followed by signal development using secondary antibodies conjugated with a fluorescent marker. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Roche).

In situ hybridization

Mouse embryos (transverse and sagittal sections) and adult tissues were frozen in OCT compound, after which 18-μm sections were cut using a cryostat and stored at –80°C until hybridization. Mouse antisense Gpr41 and Gpr43 RNA probes were transcribed using T7 RNA polymerase and uridine 5′-α-[35S]-thio-triphosphate (Perkin Elmer). Tissue sections were examined as described previously (15, 16) and counterstained with hematoxylin and eosin and TH antibodies (Millipore) to visualize SCGs.

Biochemical analyses

Plasma samples were obtained from mice fasted for 5 hours. Plasma glucose levels were determined using OneTouch Ultra (LifeScan, Inc.). Determinations of plasma TGs, free fatty acids, and total cholesterol levels were performed using commercial kits (TG, LabAssay Triglyceride; free fatty acids, LabAssay NEFA; and total cholesterol, LabAssay Cholesterol; Wako Chemicals). Levels of plasma PYY [Mouse/Rat PYY enzyme-linked immunosorbent assay (ELISA) kit; Wako Chemicals], GLP-1 [GLP-1 (Active) ELISA kit; Shibayagi], and insulin [Insulin ELISA kit (RTU); Shibayagi] were determined using ELISA kits, following the respective manufacturer’s instructions. Levels of total GLP-1 in colonic lysates were determined using a commercial kit (Multi Species GLP-1 Total ELISA; Millipore). For GLP-1 determinations, plasma and tissue samples were treated with dipeptidyl peptidase IV inhibitor (Merck Millipore) to prevent the degradation of active GLP-1.

Glucose- and insulin-tolerance tests

For the glucose-tolerance test, mice fasted for 24 hours were intraperitoneally (i.p.) administered 1.5 mg of glucose (Wako Chemicals) per gram of body weight. For the insulin-tolerance test, mice fasted for 3 hours were administered human insulin (3 mU/g, i.p.; Sigma-Aldrich). Plasma glucose concentration was monitored before injection and 15, 30, 60, 90, and 120 min after injection using OneTouch Ultra.

PET and CT imaging

Ten-week-old female Sprague-Dawley rats (Charles River) were housed under a 12-hour light-dark cycle and provided an AIN-93G diet during pregnancy. For PET imaging, rats on day 16.5 and 17.5 of pregnancy were used. Synthesis of [11C]-propionate was performed by modifying [11C]-acetate preparation as previously described (46). Briefly, [11C]-CO2 was bubbled into a mixture of 0.1 M ethyl magnesium bromide and tetrahydrofuran (Wako Chemicals) at −10°C, and then the mixture was purified using solid-phase extraction columns (OnGuard II Ag and OnGuard II H, DIONEX). The resultant solution was converted to sodium salt with NaHCO3aq (Maylon). [11C]-propionate was confirmed by high-performance liquid chromatography [Partisil-10-SAX (4.6 mm by 250 mm) and 20 mM phosphate buffer (pH 4) containing KH2PO3 and phosphoric acid (for adjusting pH) as an eluent at 0.7 ml/min and at 40°C] at 96.5% radiochemical purity and 5.77% radiochemical yield. [11C]-Propionate was diluted with physiological saline to 20 megabecquerels (250 to 300 μl per rat) and administered to the colonic lumen via the anus using a feeding needle for rats under isoflurane anesthesia. PET and computed tomography (CT) scans were performed using ClairvivoPET (Shimadzu Co., Ltd.) and Aquilion TSX-01A (Tokyo Medical Systems) instruments, respectively. After PET and CT imaging, rats were laparotomized to collect blood from the inferior vena cava under isoflurane anesthesia. Subsequently, portions of the liver, kidney, spleen, left hind paw muscle, and fetus with placenta were dissected. Each tissue was washed twice with physiological saline. After sufficiently wiping off physiological saline, the radioactivity of each organ was measured using a γ-counter. The distribution of [11C]-propionate in each organ and in embryos was measured using an AccuFLEX γ7001 instrument (ARC-7001; Hitachi Aloka Medical). Thereafter, the weight of each tissue was measured. The obtained radioactivity value was calculated using the following equationValue of attenuation correction = radioactivity value/(2{(T1 − T2)/20.4})where T1, T2, and 20.4 (in minutes) indicate the time when each tissue was measured, [11C] propionic acid administration time, and 11C half-life, respectively. After attenuation correction, each radioactivity value was divided by the corresponding weight to calculate the radioactivity value per fixed weight. The pharmacokinetics of the tracer was evaluated as counts per minute per gram of each tissue relative to that in the liver taken from the same pregnant rat.

SCFA determinations

Plasma SCFAs were determined as previously described (47). The SCFA-containing ether layers were collected and pooled for gas chromatography–mass spectrometry (GC-MS) analysis using a GCMS-QP2010 Ultra instrument (Shimadzu). The concentration of each SCFA was determined using external standard calibration over an appropriate concentration range.

Commensal bacteria composition

Fecal, cecal, placental, colonic, and embryonic DNA were extracted from frozen samples using the FastDNA SPIN Kit for Feces (MP Biomedicals) or Gentra Puregene Mouse Tail Kit (Qiagen) according to the manufacturer’s instructions. Bacteria (Bacteroides vulgatus JCM5826T) were provided by the Japan Collection of Microorganisms of RIKEN BRC and used as standards specifically for the DNA-based determination of bacterial counts. Bacterial DNA was isolated using the MonoFas Bacterial Genomic Kit IV (GLC Science) according to manufacturer’s instructions. Quantitative polymerase chain reaction (PCR) analysis was performed using SYBR Premix Ex Taq II (TaKaRa Bio) and a StepOne Real-Time PCR System (Applied Biosystems). Standard curves for quantification consisted of 10-fold serial dilutions in the range of 108 to 100 copies of the target 16S ribosomal RNA (rRNA) gene. Universal bacterial primer sequences were as follows: 5′-CRAACAGGATTAGAACCCT-3′ (forward) and 5′-GGTAAGGTTCCTCGCGTAT-3′ (reverse) (47).

For bacterial culture, amniotic fluids and cecal contents from E16.5 pregnant ICR mice were cultured in MRS (BD Biosciences) and BL medium (Eiken Chemical) at 37°C for 24 hours under anaerobic conditions. The samples were then suspended at 0.1 mg/ml in sterile PBS and diluted in 10-fold series.

For 16S rDNA amplicon sequencing, sequencing of the 16S rRNA variable regions 3 and 4 was performed as described previously (47). The libraries were sequenced using the MiSeq system (Illumina) with 2 × 300 base pair protocols. Microbial diversity and composition analyses were performed using QIIME with the Greengenes reference database clustered at 97% identity.

Indirect calorimetry

Energy expenditure was calculated as the product of the calorific value of oxygen (3.815 + 1.232 × the respiratory exchange ratio) and oxygen consumption (VO2) as previously described (15). Tyramine (100 mg/kg), a catecholamine releasing agent, was administrated at a volume per i.p. injection as described (16).

RNA isolation and quantitative reverse transcription PCR

Total RNA was extracted using the RNeasy Mini Kit (Qiagen) and ISOGEN (Nippon Gene). cDNA was reverse transcribed using isolated RNA samples as templates and Moloney murine leukemia virus reverse transcriptase (Invitrogen). Expression analyses were performed using SYBR Premix Ex Taq II and the StepOnePlus Real-Time PCR System. Real-time PCR cycling conditions were as follows: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, 58°C for 30 s, and 72°C for 1 min. In addition, dissociation was examined at 95°C for 15 s, followed by 1 cycle of 60°C for 1 min and 95°C for 15 s. The 18S rRNA gene was used as an internal control. Each sample was tested in duplicate for the average Ct value. Relative mRNA expression was calculated after normalization to the 18S rRNA reference gene using the 2-ΔΔCt method. Primer sequences are listed in tables S4 and S5.

Sympathetic neural cell culture

SCGs were dissected from P1 mice, trypsinized in 0.05% trypsin in Hanks’ balanced salt solution for 20 min at 37°C, and dissociated by trituration. Dissociated cultures were plated onto dishes coated with poly-l-lysine (20 μg/ml; Sigma-Aldrich) in DF medium (Gibco; Thermo Fisher Scientific, Waltham, MA) containing 1% penicillin-streptomycin solution (Gibco). Cells were cultured in conditioned medium containing nerve growth factor (10 ng/ml; Upstate Biotech) and 10% fetal bovine serum (FBS) as described elsewhere (16). Neuronal differentiation was quantified by counting TH-positive cells; cells with outgrowths longer than the cell body diameter were scored as positive for neurites. ImageJ (NIH, Bethesda, MD) was used to determine neurite outgrowth (48).

Western blotting and cAMP determination

Hearts were homogenized in 0.1 M sodium phosphate (pH 7.4) and centrifuged at 10,000g for 20 min at 4°C. Supernatants were analyzed by Western blotting as previously described (15). Proteins were detected by Western blotting using anti-TH (Millipore) and anti-α-actin antibodies (Sigma-Aldrich) for normalization. For cyclic adenosine monophosphate (cAMP) determinations, cells were lysed in 0.1 N HCl. After acetylation, cAMP levels were determined in duplicate using enzyme immunoassay kits (Cayman) as previously described (16).

Pancreatic cell line culture

AR42J cells (rat pancreatic exocrine cells) were purchased from the American Type Culture Collection (ATCC) and cultured in F-12K (Thermo Fisher Scientific) containing 20% FBS and maintained at 37°C under 5% CO2. For cell differentiation, the cells were plated in 24-well plates (1 × 105 cells per well) and cultured with betacellulin (2 nM; Wako Chemicals) and activin A (1 nM; Wako Chemicals), in the presence or absence of propionate (1 mM; Sigma-Aldrich) or PA-1 (10 μM) for 72 hours. Then, total RNA was isolated using ISOGEN. For small interfering RNA (siRNA) knockdown experiments, AR42J cells were transfected with 30 nM siRNA (Gpr43; target sequence: GGCCAUUGCACCAUCGUCA; GE Healthcare) using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturers’ instructions.

Organoid culture

Colonic crypts were isolated from 7-week-old mice or embryos on day 15.5 and cultured on Matrigel, as previously described (49). Culture medium contained advanced DMEM/F12 (Life Technologies) supplemented with gentamicin/amphotericin B solution (Life Technologies), 10 mM HEPES (Nacalai Tesque), 1% (v/v) N2 (Life Technologies), 1% (v/v) B27 (Life Technologies), 1 μM N-acetylcysteine (Sigma-Aldrich), 2 mM l-alanyl-l-glutamine (Nacalai Tesque), 500 nM A-83-01 (Wako Chemicals), 10 nM [Leu15]-gastrin 1 (Sigma-Aldrich), 1 mM nicotinamide (Wako Chemicals), 10 μM SB202190 (Wako Chemicals), 2.5 μM CHIR99021 (StemRD), and 2.5 μM thiazovivin (StemRD) supplemented with the growth factors EGF (epidermal growth factor; 50 ng/ml; Peprotech), noggin (100 ng/ml; Sigma-Aldrich), and R-spondin (1 μg/ml; Sigma-Aldrich). The medium was incubated in the presence or absence of Wnt3a for 24 hours. Organoid cultures were then treated with propionate (1 or 10 mM) for 24 hours, and total RNA was isolated using the RNeasy Mini Kit.

Metabolite analysis

Liquid chromatography–tandem MS (LC-MS/MS)–based lipidomics analysis was performed as described elsewhere (50). Briefly, samples were subjected to solid-phase extraction on a Sep-Pak C18 cartridge (Waters) with deuterium-labeled internal standards (arachidonic acid-d8, leukotriene B4-d4, 15-hydroxyeicosatetraenoic acid-d8, and prostaglandin E2-d4). Lipidomics analyses were performed using a Waters UPLC system with a linear ion-trap quadrupole mass spectrometer (QTRAP 5500; AB SCIEX) equipped with an Acquity UPLC BEH C18 column (1.0 mm by 150 mm by 1.7 μm; Waters). MS/MS analyses were conducted in negative-ion mode, and lipophilic metabolites were identified and quantified by multiple-reaction monitoring. Hydrophilic metabolites from biological samples were extracted and derivatized as described previously (51). GC-MS analysis was performed using a GCMS-QP2010 Ultra instrument (Shimadzu) with a fused silica capillary column (CP-SIL 8 CB low bleed/MS; 0.25 mm by 30 m by 0.25 μm; Agilent Technologies) as described previously (51). Acquired data were exported in the CSV file format and analyzed using in-house analytical software (AI output).

Statistical analysis

All values are presented as means ± SEM. Statistical significance of differences between groups was determined using two-tailed unpaired Student’s t test (two groups) or two-tailed one-way analysis of variance followed by Tukey-Kramer’s post hoc test and Dunnett’s post hoc test (three or more groups). P values < 0.05 were considered statistically significant. False discovery rates (q value) of the metabolomics and 16S rDNA amplicon sequencing data were estimated with the Benjamini-Hochberg procedure. The Smirnov-Grubbs’ test was used for evaluating outliers. The similarity of microbiomes was tested using PERMANOVA (permutational multivariate analysis of variance).

Supplementary Materials

References and Notes

Acknowledgments: We thank K. Igarashi for metabolome analysis; A. Nagata for SCFA analysis; M. Arita and S. Kasuga for in vitro assay; and T. Sasaki, M. Akehi, and H. Hirano for PET imaging. Funding: This work was supported by research grants from the JSPS KAKENHI (JP17H05344 to I.K., JP15H05897 and JP15H05898 to M.A., and JP18H04680 and JP17KT0055 to K.H.), AMED (JP18gm1010007 to I.K. and JP18gm1010004h0103 to K.H.), the Lotte Foundation (to I.K.), the Institute for Fermentation Osaka (to I.K.), the Takeda Science Foundation (to H.K.), the Asahi Grass Foundation (to K.H.), and the Yakult Science Foundation (to K.H.). Author contributions: I.K., J.M., R.O.-K., K.W., T.Y., M.O., R.A., Y.I., D.K., D.I., A.I., Y.T., S.T., S.K., M.W., M.I., F.N., H.K., M.S., K.I., and K.H. performed the experiments. I.K., J.M., R.O.-K., and K.H. wrote the manuscript. I.K., J.M., R.O.-K., J.I., G.T., H.O., M.A., H.I., and K.H. interpreted the data. I.K. and K.H. supervised the project. All authors read and approved the final manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: The source data underlying Figs. 1 to 6, figs. S1 to S24, and metabolome analysis have been deposited into the Dryad repository (52). The raw data for 16S rDNA amplicon sequencing have been deposited at the DNA Data Bank of Japan (DDBJ) under the accession nos. DRA007699 (fig. S3, A and B), DRA007700 (fig. S5, A and B), DRA009267 (fig. S6, A and B), DRA007701 (fig. S13, A and B), DRA009265 (fig. S19, A and B), and DRA009266 (fig. S23, A and B). All other data generated or analyzed during this study are included in this article and its supplementary materials.

Stay Connected to Science

Navigate This Article