ARTICLES b o+-warm c o Ldi-Lup7-Cre+warm d o Ld-Lrp1-Cre+ ●Ldr ●Ld-cold ●Lal-Lrp1-Cre+co 00 TRL iii elis ile. atus. i,ies lato sh i CDCA ●Ld-wamo f8 ●Ld-ccd● 4000 ● Lair"col O Ldr-Lrp1-Cre+warm 音 Figure 4 Hepatic uptake of cholesterol-rich lipoproteins determines fecal bile acid excretion in cold-exposed mice (a-d)Plasma lipid analysis in a (LdIr-Lrpl-Cre, n=6 mice per group)(c, d). Plasma triglyceride and cholesterol levels (a, c)and lipoprotein profiles from corresponding pooled plasma cold-housed mice(WT, n= 13 mice per group; warm-housed Ldirk-, n= 13 mice; cold-housed Ldlr-, n= 12 mice; Ldir-Lrpl-Cre, n=6 mice per group). Throughout, data are mean ts.e. m. P<0.05, " P<0.01, **P<0.00l; by unpaired two-tailed Students t-test (a, c)or two way ANova (e-g) 2 However, hierarchical cluster analysis showed a separation from the Unexpectedly, we observed a clear separation of the gut microbiota other groups for only the cold-housed mice that were not treated with from WT mice versus those from Cyp7b1--mice, irrespectively of a without manipulating its absorption rate, we performed a study using on the gut microbiome seen in the mDS plot may explain the rela mice that were fed a chow diet supplemented without or with choles- tively weak effect of cold also in the WT mice. Notably, Cyp Zbl over- terol Relative to the warm-housed controls, cold-exposed mice showed expression was associated with changes in the composition of the gut substantial changes in the gut microbiome on both diets(Fig. 5j microbiome. We observed that the effects on gut bacteria were present and Supplementary Fig. 10a, b). Notably, even in the absence of dietary in mice that were housed at 22C but not in mice that were housed at cholesterol, significantly higher fecal bile acid excretion was observed thermoneutrality(Fig. 51, m and Supplementary Fig. 1la-c), indi- in cold-housed versus warm-housed mice(Supplementary Fig 10c, d ), cating that both Cyp 7bI upregulation and accelerated cholesterol indicating that endogenously synthesized cholesterol is sufficient metabolism via activated BAT are important for changes in the for some cold-induced bile acid production. Despite Cyp 7bl induc- microbiome. To investigate the relevance of cold-induced bile acids tion and higher bile acid levels in liver in response to B3-adrenergic for changes in the gut microbiome by an independent approach,we receptor activation(Fig. 2g, h), we observed only a trend toward performed experiments with an additional genetic model, Abcb4-- higher fecal bile acid levels without a clear effect on the gut microbi-(also known as Mdr2--)mice, which lack the canalicular phospholi ome(Supplementary Fig. 10e-g) Taking into account the diurnal pid transporter MDR2, resulting in abrogated cold-induced changes food intake patterns of mice, the resulting intermittent B3-adren- in fecal bile acid levels(Supplementary Fig. lld ). Consistent with ergic receptor stimulation by the CL316, 243 supplied with the diet the observation of similar bile acid levels in the feces of warn likely explains the less pronounced effects. Next we asked whether housed and cold-housed Mdr2-/- mice, we did not observe differ the loss of cold-induced fecal bile acids observed in Cyp7b1-/-mice ences in the gut microbiota( Fig. 5n and Supplementary Fig. lle) (Fig 3c, d) had an effect on the composition of the gut microbiota. In summary, these data show that the cold-induced changes in the DNLINE PUBLICATION© 2017 Nature America, Inc., part of Springer Nature. All rights reserved. a r t i c l e s nature medicine advance online publication  However, hierarchical cluster analysis showed a separation from the other groups for only the cold-housed mice that were not treated with EZ (Supplementary Fig. 9g). To address the role of dietary cholesterol without manipulating its absorption rate, we performed a study using mice that were fed a chow diet supplemented without or with choles￾terol. Relative to the warm-housed controls, cold-exposed mice showed substantial changes in the gut microbiome on both diets (Fig. 5j and Supplementary Fig. 10a,b). Notably, even in the absence of dietary cholesterol, significantly higher fecal bile acid excretion was observed in cold-housed versus warm-housed mice (Supplementary Fig. 10c,d), indicating that endogenously synthesized cholesterol is sufficient for some cold-induced bile acid production. Despite Cyp7b1 induc￾tion and higher bile acid levels in liver in response to β3-adrenergic receptor activation (Fig. 2g,h), we observed only a trend toward higher fecal bile acid levels without a clear effect on the gut microbi￾ome (Supplementary Fig. 10e–g). Taking into account the diurnal food intake patterns of mice, the resulting intermittent β3-adren￾ergic receptor stimulation by the CL316,243 supplied with the diet likely explains the less pronounced effects. Next we asked whether the loss of cold-induced fecal bile acids observed in Cyp7b1−/− mice (Fig. 3c,d) had an effect on the composition of the gut microbiota. Unexpectedly, we observed a clear separation of the gut microbiota from WT mice versus those from Cyp7b1−/− mice, irrespectively of housing temperature (Fig. 5k). The dominant effect of the genotype on the gut microbiome seen in the MDS plot may explain the rela￾tively weak effect of cold also in the WT mice. Notably, Cyp7b1 over￾expression was associated with changes in the composition of the gut microbiome. We observed that the effects on gut bacteria were present in mice that were housed at 22 °C but not in mice that were housed at thermoneutrality (Fig. 5l,m and Supplementary Fig. 11a–c), indi￾cating that both Cyp7b1 upregulation and accelerated cholesterol metabolism via activated BAT are important for changes in the gut microbiome. To investigate the relevance of cold-induced bile acids for changes in the gut microbiome by an independent approach, we performed experiments with an additional genetic model, Abcb4−/− (also known as Mdr2−/−) mice, which lack the canalicular phospholi￾pid transporter MDR2, resulting in abrogated cold-induced changes in fecal bile acid levels (Supplementary Fig. 11d). Consistent with the observation of similar bile acid levels in the feces of warm￾housed and cold-housed Mdr2−/− mice, we did not observe differ￾ences in the gut microbiota (Fig. 5n and Supplementary Fig. 11e). In summary, these data show that the cold-induced changes in the 3,000 Ldlr –/– warm Ldlr –/– cold Ldlr –/–Lrp1-Cre + warm Ldlr –/–Lrp1-Cre + cold Ldlr –/–Lrp1-Cre + warm Ldlr –/–Lrp1-Cre + cold 400 300 200 100 0 400 300 200 100 0 0 10 Fraction 20 30 0 10 Fraction 20 30 LDL TRL LDL TRL Cholesterol (mg/dl) Cholesterol (mg/dl) 2,000 1,000 0 3,000 4,000 2,000 1,000 0 *** TG Chol TG * Chol *** Lipids (mg/dl) Lipids (mg/dl) Ldlr –/– warm Ldlr –/– cold a b c d Ldlr –/– warm Ldlr –/– cold Ldlr –/–Lrp1-Cre + warm Ldlr –/–Lrp1-Cre + cold WT warm WT cold 10 *** *** *** *** *** *** *** *** *** *** *** ** *** *** * *** ** 8 Bile acids (ng/mg, fold) 6 4 2 0 CA β-MCA DCA UDCA α-MCA ω-MCA CDCA e Ldlr –/– warm Ldlr –/– cold Ldlr –/–Lrp1-Cre + warm Ldlr –/–Lrp1-Cre + cold WT warm WT cold *** *** *** *** *** *** *** *** *** *** *** *** *** *** * 6,000 4,000 2,000 0 *** * ** * * Bile acids (ng/mg, fold) Bile acids (ng/mg) 200 150 100 50 0 T-α/β-MCA TUDCA TDCA TCDCA GCA Ldlr –/–Lrp1-Cre + warm Ldlr –/–Lrp1-Cre + cold WT warm WT cold Ldlr –/– warm Ldlr –/– cold f g Figure 4 Hepatic uptake of cholesterol-rich lipoproteins determines fecal bile acid excretion in cold-exposed mice. (a–d) Plasma lipid analysis in warm-housed and cold-housed Ldlr−/− mice (n = 7 mice per group) (a,b), as well as in mice that lacked both LDLR and hepatic LRP1 expression (Ldlr−/−Lrp1-Cre, n = 6 mice per group) (c,d). Plasma triglyceride and cholesterol levels (a,c) and lipoprotein profiles from corresponding pooled plasma samples (b,d) are shown. (e–g) Fecal levels of UBA (e) and CBA (f) species, as well as the total amount of fecal bile acids (g) from warm-housed and cold-housed mice (WT, n = 13 mice per group; warm-housed Ldlr−/−, n = 13 mice; cold-housed Ldlr−/−, n = 12 mice; Ldlr−/−Lrp1–Cre, n = 6 mice per group). Throughout, data are mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001; by unpaired two-tailed Student’s t-test (a,c) or two-way ANOVA (e–g)