ARTICLES of AbcblI expression, as well as that of Nrob2, and downregulation transporter(ASBT), which is required for efficient CBA re-uptake as of the SHP target genes Cyp 7al and Cyp8b1 in warm-housed, as well part of the enterohepatic circulation- Cold exposure led to higher as cold-housed, mice(Fig. 2e), whereas other genes important for ASBT expression at the mRNA (SIc10a2; Fig 3f), as well as protein synthesis( Cyp27a1)and conjugation(Baat)were unaffected by PX (Fig. 3g), level, which was lost in Cyp7b1-/-mice(Supplementary treatment but were slightly higher after cold exposure. Notably, the Fig. 7c, d). Furthermore, cold-housing resulted in a trend toward upregulation of Cyp bl that we observed after cold treatment was higher bile acid levels in portal(Fig. 3h) and systemic(Fig. 3i)blood. not affected by a FXR-mediated negative feedback loop( Fig. 2e). Taken together, these data argue against diminished CBa re-uptake Consistent with sustained Cyp7b1 expression, PX treatment did not capacity as a cause for higher fecal bile acid levels influence the cold-induced increase in hepatic bile acid levels(Fig Another mechanism for increased fecal concentrations of cBa and only species that arose by the action of the PX-sensitive enzyme could be due to reduced bile acid deconjugation by bile salt hydrolases CYP8B1, such as CA, were slightly lower(Supplementary Fig 4a-c).(BSHs), which are expressed by a number of gut bacteria 26. To address In a human cohort, we found lower hepatic expression of CYP7BI this hypothesis we depleted the intestinal microbiota by using an ant in obese subjects with type 2 diabetes than in non-obese controls, biotic cocktail. Relative to that in the untreated controls, this interven whereas expression of CYP7AI and CYP8BI was not different tion resulted in higher fecal CBA content both in warm-housed and (Supplementary Fig. 5a-c), indicating that there was metabolic cold-exposed mice(Fig. 3j, k), which could be explained by the com- 9 regulation of the alternative pathway in humans. plete eradication of gut BSH activity by antibiotic treatment(Fig. 31) To determine the role of BAT in cold-induced bile acid synthesis, we Although antibiotic treatment did not affect the expression of genes g used the B3-adrenergic receptor agonist CL316, 243 to pharmacologi- related to bile acid metabolism(Supplementary Fig. 8a), hepatic e cally activate BAT in mice that were housed at room temperature. As UBA but not CBA levels were lower(Supplementary Fig 8b, c) w compared to the untreated controls, expression of Cyp7bl, but of no Of note, higher BSH activity was observed in cold-exposed mice other genes involved in hepatic bile acid synthesis, was higher in the in the absence of antibiotics than in warm-housed mice(Fig. 31), CL316, 243-treated mice(Fig. 2g)and was associated with increased resulting in higher fecal taurine levels(Fig. 3m). Cumulatively, these levels of hepatic bile acids( Fig. 2h and Supplementary Fig. 6a, b). results suggest that saturation of ASBT rather than diminished decon Genetic intervention by adeno-associated virus(AAV)-mediated jugation by BSH activity is responsible for higher fecal CBAexcretion E overexpression of Cyp 7bl in WT mice resulted in moderately higher after cold exposure z hepatic bile acid levels at thermoneutral conditions, as compared to those in mice that were treated with GFP-expressing control AAv Bile acid excretion depends on a BAT-liver cholesterol axis E(AAV-GFP)(Fig. 2i and Supplementary Fig. 6c). Conversely, the Cold-activated BAT efficiently processes dietary lipids carried by p a cold-dependent induction in hepatic bile acid levels was blunted prandial lipoproteins and promotes receptor-mediated uptake of the in Cyp7b1-- mice, as compared to that in WT mice(Fig. 2j and respective cholesterol-rich remnants by the liver 5, 17. Accordingly, in induced the a CyP27aI in BAT(Fig. 2k)and led to increased amounts of plasma cold(Fig.1g-i), we observed higher amounts of circulating cholesterol 27-hydroxycholesterol(Fig. 21), the substrate for hepatic CYP7B1. rich lipoproteins in cold-housed versus warm-housed mice that were Of note, 27-hydroxycholesterol did not accumulate in the liver and deficient in the LDL receptor (LDLr), which is the main receptor for g plasma of cold-exposed Cyp7b1-/-mice(Supplementary Fig. 6f, g), hepatic remnants(Fig 4a, b). When we combined LdIr-genotype with which could be because much of it ed in the feces( Fig. 21 knockout of the the alternative In conclusion, these data reveal that under conditions of BAT activa- lipoprotein receptor, LDLR-related protein 1 (LrpI), we observed on, the alternative pathway is selectively upregulated and increases even higher lipoprotein levels(Fig. 4c, d), indicating blunted remnant bile acid sy clearance in the absence of both receptors. Notably, the cold-induced increase in fecal bile acid excretion in LdIr-mice was only 50% of that old accelerates fecal bile acid excretion via cyP7b1 in WT mice, and it was nearly abolished in mice that lacked LDLR and To further explore the fate of bile acids, we quantified their levels hepatic LRPI (Fig. 4e-g). Taken together, these findings indicate that N in stool samples from warm-housed and cold-housed mice. The fecal bile acid excretion is dependent on hepatic cholesterol that is deliv- amount of excreted bile acids was much higher in cold-housed ered by postprandial lipoproteins generated by cold-activated BAT than in warm-housed mice(Fig 3a, b). Notably, some CBA sp To assess the contribution of dietary cholesterol to cold-induced cies, especially tauro-a/B-MCA (T-a/B-MCA) and tauro-ca bile acid synthesis, we first blocked dietary cholesterol resorption by TCA), were up to 40-fold higher UBA levels were less affected and inhibiting the intestinal cholesterol transporter Niemann-Pick-1-like were increased up to twofold(Fig. 3a, b ) In Cyp7b1-/- mice, the 1(NPCIL1)with ezetimibe(EZ). Notably, combining cold-exposure cold-induced rise in fecal CBA was abrogated, and concentrations and EZ treatment resulted in lower levels of plasma lipids and choles- of some bile acid species were even lower(Fig. 3c, d). Conversely, terol-rich lipoproteins, even more so than with cold exposure alone AAV-mediated Cyp7b1 overexpression resulted in higher fecal bile (Fig 5a, b). Consistent with diminished dietary cholesterol uptak acid species, as compared to that in the AAv-GFP controls(Fig. 3e and hepatic lipid levels after EZ treatment(Fig. 5c), we observed and Supplementary Fig. 7a, b). Notably, the effect was not present compensatory hepatic upregulation of the gene encoding the rate- at thermoneutral conditions but at 22C, and it was even more pro- limiting enzyme of the cholesterol synthesis pathway(Hmgcr),as bounced at 16C(Fig. 3e and Supplementary Fig. 7a, b), indicat- well as of Ldlr(Fig. 5d), whereas expression of Cyp7b1 was unaf ing that increased dietary intake and BAT-dependent processing of fected( Supplementary Fig 9a). Liver bile acid levels of cold-housed cholesterol at lower temperatures is critical for efficient CYP7B1- EZ-treated mice were comparable to those of cold-housed control mediated bile acid production mice(Fig. 5e and Supplementary Fig 9b, c). In contrast, the fecal mice, wea sate the basis of higher fecal CBA levels in cold-housed bile acid concentrations of EZ-treated mice were not higher after rst quantified the ileal expression of apical sodium bile cold exposure, and bile acid levels were almost identical to those in© 2017 Nature America, Inc., part of Springer Nature. All rights reserved. a r t i c l e s nature medicine advance online publication  of Abcb11 expression, as well as that of Nr0b2, and downregulation of the SHP target genes Cyp7a1 and Cyp8b1 in warm-housed, as well as cold-housed, mice (Fig. 2e), whereas other genes important for synthesis (Cyp27a1) and conjugation (Baat) were unaffected by PX treatment but were slightly higher after cold exposure. Notably, the upregulation of Cyp7b1 that we observed after cold treatment was not affected by a FXR-mediated negative feedback loop (Fig. 2e). Consistent with sustained Cyp7b1 expression, PX treatment did not influence the cold-induced increase in hepatic bile acid levels (Fig. 2f), and only species that arose by the action of the PX-sensitive enzyme CYP8B1, such as CA, were slightly lower (Supplementary Fig. 4a–c). In a human cohort, we found lower hepatic expression of CYP7B1 in obese subjects with type 2 diabetes than in non-obese controls, whereas expression of CYP7A1 and CYP8B1 was not different (Supplementary Fig. 5a–c), indicating that there was metabolic regulation of the alternative pathway in humans. To determine the role of BAT in cold-induced bile acid synthesis, we used the β3-adrenergic receptor agonist CL316,243 to pharmacologi￾cally activate BAT in mice that were housed at room temperature. As compared to the untreated controls, expression of Cyp7b1, but of no other genes involved in hepatic bile acid synthesis, was higher in the CL316,243-treated mice (Fig. 2g) and was associated with increased levels of hepatic bile acids (Fig. 2h and Supplementary Fig. 6a,b). Genetic intervention by adeno-associated virus (AAV)-mediated overexpression of Cyp7b1 in WT mice resulted in moderately higher hepatic bile acid levels at thermoneutral conditions, as compared to those in mice that were treated with GFP-expressing control AAV (AAV-GFP) (Fig. 2i and Supplementary Fig. 6c). Conversely, the cold-dependent induction in hepatic bile acid levels was blunted in Cyp7b1−/− mice, as compared to that in WT mice (Fig. 2j and Supplementary Fig. 6d,e). Cold exposure induced the expression of Cyp27a1 in BAT (Fig. 2k) and led to increased amounts of plasma 27-hydroxycholesterol (Fig. 2l), the substrate for hepatic CYP7B1. Of note, 27-hydroxycholesterol did not accumulate in the liver and plasma of cold-exposed Cyp7b1−/− mice (Supplementary Fig. 6f,g), which could be because much of it was excreted in the feces (Fig. 2m). In conclusion, these data reveal that under conditions of BAT activa￾tion, the alternative pathway is selectively upregulated and increases bile acid synthesis. Cold accelerates fecal bile acid excretion via CYP7B1 To further explore the fate of bile acids, we quantified their levels in stool samples from warm-housed and cold-housed mice. The amount of excreted bile acids was much higher in cold-housed than in warm-housed mice (Fig. 3a,b). Notably, some CBA spe￾cies, especially tauro-α/β-MCA (T-α/β-MCA) and tauro-CA (TCA), were up to 40-fold higher. UBA levels were less affected and were increased up to twofold (Fig. 3a,b). In Cyp7b1−/− mice, the cold-induced rise in fecal CBA was abrogated, and concentrations of some bile acid species were even lower (Fig. 3c,d). Conversely, AAV-mediated Cyp7b1 overexpression resulted in higher fecal bile acid species, as compared to that in the AAV-GFP controls (Fig. 3e and Supplementary Fig. 7a,b). Notably, the effect was not present at thermoneutral conditions but at 22 °C, and it was even more pro￾nounced at 16 °C (Fig. 3e and Supplementary Fig. 7a,b), indicat￾ing that increased dietary intake and BAT-dependent processing of cholesterol at lower temperatures is critical for efficient CYP7B1- mediated bile acid production. To investigate the basis of higher fecal CBA levels in cold-housed mice, we first quantified the ileal expression of apical sodium bile transporter (ASBT), which is required for efficient CBA re-uptake as part of the enterohepatic circulation20. Cold exposure led to higher ASBT expression at the mRNA (Slc10a2; Fig. 3f), as well as protein (Fig. 3g), level, which was lost in Cyp7b1−/− mice (Supplementary Fig. 7c,d). Furthermore, cold-housing resulted in a trend toward higher bile acid levels in portal (Fig. 3h) and systemic (Fig. 3i) blood. Taken together, these data argue against diminished CBA re-uptake capacity as a cause for higher fecal bile acid levels. Another mechanism for increased fecal concentrations of CBA could be due to reduced bile acid deconjungation by bile salt hydrolases (BSHs), which are expressed by a number of gut bacteria26. To address this hypothesis we depleted the intestinal microbiota by using an anti￾biotic cocktail. Relative to that in the untreated controls, this interven￾tion resulted in higher fecal CBA content both in warm-housed and cold-exposed mice (Fig. 3j,k), which could be explained by the com￾plete eradication of gut BSH activity by antibiotic treatment (Fig. 3l). Although antibiotic treatment did not affect the expression of genes related to bile acid metabolism (Supplementary Fig. 8a), hepatic UBA but not CBA levels were lower (Supplementary Fig. 8b,c). Of note, higher BSH activity was observed in cold-exposed mice in the absence of antibiotics than in warm-housed mice (Fig. 3l), resulting in higher fecal taurine levels (Fig. 3m). Cumulatively, these results suggest that saturation of ASBT rather than diminished decon￾jugation by BSH activity is responsible for higher fecal CBA excretion after cold exposure. Bile acid excretion depends on a BAT–liver cholesterol axis Cold-activated BAT efficiently processes dietary lipids carried by post￾prandial lipoproteins and promotes receptor-mediated uptake of the respective cholesterol-rich remnants by the liver15,17. Accordingly, in contrast to the decline in circulating lipids that occurs in WT mice with cold (Fig. 1g–i), we observed higher amounts of circulating cholesterol￾rich lipoproteins in cold-housed versus warm-housed mice that were deficient in the LDL receptor (LDLR), which is the main receptor for hepatic remnants (Fig. 4a,b). When we combined Ldlr−/− genotype with a liver-specific knockout of the gene encoding the alternative hepatic lipoprotein receptor, LDLR-related protein 1 (Lrp1), we observed even higher lipoprotein levels (Fig. 4c,d), indicating blunted remnant clearance in the absence of both receptors. Notably, the cold-induced increase in fecal bile acid excretion in Ldlr−/− mice was only 50% of that in WT mice, and it was nearly abolished in mice that lacked LDLR and hepatic LRP1 (Fig. 4e–g). Taken together, these findings indicate that fecal bile acid excretion is dependent on hepatic cholesterol that is deliv￾ered by postprandial lipoproteins generated by cold-activated BAT. To assess the contribution of dietary cholesterol to cold-induced bile acid synthesis, we first blocked dietary cholesterol resorption by inhibiting the intestinal cholesterol transporter Niemann-Pick-1-like 1 (NPC1L1) with ezetimibe (EZ). Notably, combining cold-exposure and EZ treatment resulted in lower levels of plasma lipids and choles￾terol-rich lipoproteins, even more so than with cold exposure alone (Fig. 5a,b). Consistent with diminished dietary cholesterol uptake and hepatic lipid levels after EZ treatment (Fig. 5c), we observed compensatory hepatic upregulation of the gene encoding the rate￾limiting enzyme of the cholesterol synthesis pathway (Hmgcr), as well as of Ldlr (Fig. 5d), whereas expression of Cyp7b1 was unaf￾fected (Supplementary Fig. 9a). Liver bile acid levels of cold-housed EZ-treated mice were comparable to those of cold-housed control mice (Fig. 5e and Supplementary Fig. 9b,c). In contrast, the fecal bile acid concentrations of EZ-treated mice were not higher after cold exposure, and bile acid levels were almost identical to those in