ARTICLES a negative endocrine feedback signal to th ia fibroblast growth warm-housed mice(Fig.). On the genus level, the abundance of factor(fgf)9inh Parabacteroides spp. was higher, whereas the abundance of undlassi cold- ances at the versus differ- cold in an ghe the 16S rR ent with studie exposurel0.. We examine that th Porphyrom uction© 2017 Nature America, Inc., part of Springer Nature. All rights reserved. Ar t i c l e s  advance online publication nature medicine a negative endocrine feedback signal to the liver via fibroblast growth factor (FGF) 19 in humans or via the mouse ortholog FGF15 (ref. 21). The mechanism involves the activation of the transcriptional co-repressor small heterodimeric partner (SHP; encoded by Nr0b2) in hepatocytes, which confers suppression of CYP7A1 expression and hence lowers bile acid synthesis. Both in liver and adipose tissue, bile￾acid-activated FXR has additional effects, including such as those on cholesterol, fatty acid and glucose metabolism20,21. Bile acids also stimulate signaling at the plasma membrane via the G-protein-cou￾pled bile acid receptor 1 (GPBAR1; also known as TGR5), thereby regulating diverse metabolic pathways22. In the context of energy metabolism, TGR5 stimulation triggers the secretion of the insulin secretagogue and anorexic hormone glucagon (encoded by Gcg) in L-cells of the intestine23. Moreover, in both mice and humans, bile acids increase energy expenditure via the stimulation of TGR5 on brown adipocytes24,25, which suggests a role of liver-derived bile acids as regulators of adaptive thermogenesis. Bile acids show mutual interactions with gut microbiota. On the one hand, the gut bacteria modify bile acids through both dehydroxy￾lation, to form secondary bile acids such as deoxycholic acid (DCA), and through deconjungation via the enzymatic activity of bile salt hydrolases26. On the other hand, bile acids exert species-dependent bacteriostatic effects, thereby affecting the composition of the intes￾tinal microbiota27. As previously mentioned, cold exposure has been found to induce alterations in the gut microbiota of mice10,11, and transplantation of feces from cold-exposed mice was found to improve the metabolic profile of recipient mice that were housed at ambient temperature10. Thus, these studies provided evidence that cold exposure has distinct effects on the bacterial composition of the intestine and can promote a healthy metabolic phenotype. However, how cold conditions change the composition of the gut microbiota remains elusive. Here we demonstrate that during BAT activation by cold expo￾sure, bile-acid-driven alterations in systemic cholesterol metabolism integrate the organ-specific responses in thermogenic adipose tis￾sues, liver and intestine. Cholesterol homeostasis was achieved by the stimulation of lipoprotein flux, increased bile acid synthesis via the alternative pathway and massive fecal excretion of conjugated bile acids, which ultimately shaped the gut microbiome and promoted adaptive thermogenesis in a cold environment. RESULTS Cold alters the gut microbiome Dietary composition is a determinant of the gut microbiome28,29, and cold exposure requires higher food intake to meet the increased energy demands for thermogenesis12. Here we combined dietary and cold intervention by challenging mice with a cholesterol-enriched high-fat diet (HFD) and keeping the mice under thermoneutral (30 °C; hereafter referred to as warm) versus cold (6 °C) conditions, respectively. First, we profiled the fecal microbiota from warm-housed and cold-housed mice by analyzing the DNA sequence encoding the 16S rRNA. Multidimensional scaling (MDS) based on weighted UniFrac distances revealed distinct clustering between microbiota from warm-housed versus cold-housed mice (Fig. 1a), which is consist￾ent with studies using mice that were subjected to prolonged cold exposure10,11. We examined the changes at the family level and found that the abundance of Lachnospiraceae and Deferribacteraceae fam￾ily members was higher, whereas the abundance of Clostridiales and Porphyromonadaceae family members was lower, in cold- versus warm-housed mice (Fig. 1b). On the genus level, the abundance of Parabacteroides spp. was higher, whereas the abundance of unclassi￾fied Porphyromonadaceae family members was clearly lower in cold￾housed versus warm-housed mice (Fig. 1b). These differences could be traced to operational taxonomic unit (OTU) abundances at the species level, which showed distinct patterns in warm-housed versus cold-housed mice (Fig. 1c). Although we could not observe differ￾ences in α-diversity at the genus level (Fig. 1d), we did note that cold exposure resulted in lower richness and a decreased Shannon index at the species level (Fig. 1e). We detected similar cold-dependent differences in chow-fed wild-type (WT) mice (Supplementary Fig. 1a–e) and in HFD-fed diabetic db/db mice (Supplementary Fig. 2a–e). Thus, these data clearly indicated that cold-housing resulted in an altered gut microbiome in mice that were fed either a chow diet or a cholesterol-enriched HFD. Cold induces the alternative bile acid synthesis pathway In accordance with increased energy demands, we observed higher food intake and fecal mass in cold-exposed mice than in mice that were housed at thermoneutrality (Fig. 1f and Supplementary Table 1a). Despite higher dietary lipid uptake, the cold-exposed mice had lower plasma levels of triglycerides and total cholesterol (Fig. 1g), owing to reduced levels of triglyceride-rich lipoproteins (Fig. 1h) and choles￾terol-transporting low-density lipoprotein (LDL) and high-density lipoprotein (HDL) (Fig. 1i). Next we administered radiolabeled cho￾lesterol by oral gavage and observed higher cholesterol uptake in all BAT depots in cold- versus warm-housed mice, despite the slightly lower levels of radioactivity in plasma (Fig. 1j). Notably, cholesterol uptake was also higher in the liver, which took up more than 10% of the administered dose in cold-exposed mice (Fig. 1k). Consistent with the known inefficient absorption of dietary cholesterol, the main proportion of radioactive tracer was still present in the small intes￾tine, irrespective of cold exposure (Fig. 1k). Taken together, these data demonstrate both accelerated lipoprotein processing in BAT and increased cholesterol transport from thermogenic adipose tissues toward the liver, which is consistent with previous studies17. To determine the fate of cholesterol, we measured bile acid levels in the liver by quantitative liquid chromatography coupled to mass spec￾trometry (LC–MS). We observed that the levels of most of the unconju￾gated bile acid (UBA; Fig. 2a) and conjugated bile acid (CBA; Fig. 2b) species were significantly higher in the cold-housed mice than in the warm-housed control mice and noted that the most prominent induc￾tions were for cholic acid (CA), ursodeoxycholic acid (UDCA) and muricholic acids (MCAs). However, the composition of the bile acid pool was not significantly affected (Supplementary Fig. 3a). In gall bladder, higher levels of CBA were observed in cold- versus warm￾housed mice (Fig. 2c), suggesting that there was efficient biliary CBA excretion via the bile salt export pump (BSEP; encoded by Abcb11). Consistent with normal liver function, cold exposure did not provoke liver inflammation or damage (Supplementary Fig. 3b,c). To determine the basis for cold-induced hepatic bile acid synthesis, we performed mRNA analysis of liver tissue to evaluate the expression of components of both the classical and alternative bile acid syn￾thesis pathways (Fig. 2d,e). Whereas Cyp7a1 expression was unal￾tered, Cyp7b1 expression was fourfold higher in cold-housed mice than in warm-housed mice (Fig. 2e), indicating specific upregulation of the alternative bile acid synthesis pathway. To unravel the poten￾tial role of FXR in cold-induced bile acid synthesis, we treated mice with the FXR agonist Phenex20606 (hereafter referred to as PX)30. As expected, treatment with PX resulted in significant induction