SEARCH

SEARCH BY CITATION

Keywords:

  • antagonist;
  • apical sodium-dependent bile acid transporter;
  • fibroblast growth factor 15;
  • muricholic acid;
  • steroid 12-alpha hydroxylase

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

Objective

Bile acid (BA) synthesis is regulated by negative feedback end-product inhibition, initiated by farnesoid X receptors (FXRs) in liver and gut. Studies on cholic acid (CA)-free Cyp8b1−/− mice have concluded that CA is a potent suppressor of BA synthesis. Cyp8b1−/− mice have increased BA synthesis and an enlarged BA pool, a phenotype shared with bile-duct-ligated, antibiotics-administered and with germ-free mice. Studies on such mice have concluded BA synthesis is induced due to reduced hormonal signalling by fibroblast growth factor (FGF)15 from intestine to liver. A mutual finding in these models is that potent FXR-agonistic BAs are reduced. We hypothesized that the absence of the potent FXR agonist deoxycholic acid (DCA) may be important for the induction of BA synthesis in these situations.

Design

Two of these models were investigated, antibiotic treatment and Cyp8b1−/− mice and their combination. Secondary BA formation was inhibited by ampicillin (AMP) given to wild-type and Cyp8b1−/− mice. We then administered CA, chenodeoxycholic acid (CDCA) or DCA to AMP-treated Cyp8b1−/− mice.

Results

Our data show that the phenotype of AMP-treated wild-type mice resembles that of Cyp8b1−/− mice with fourfold induced Cyp7a1 expression, increased intestinal apical sodium-dependent BA transporter expression and increased hepatic BA levels. We also show that reductions in the FXR-agonistic BAs CDCA, CA, DCA or lithocholic acid cannot explain this phenotype; instead, it is likely due to increases in levels of α- and β-muricholic BAs and ursodeoxycholic acid, three FXR-antagonistic BAs.

Conclusions

Our findings reveal a potent positive feedback mechanism for regulation of BA synthesis in mice that appears to be sufficient without endocrine effects of FGF15 on Cyp7a1. This mechanism will be fundamental in understanding BA metabolism in both mice and humans.


Abbreviations
BA

bile acid

FXR

farnesoid X receptor

FGF

fibroblast growth factor

CA

cholic acid

DCA

deoxycholic acid

CDCA

chenodeoxycholic acid

Cyp7a1

cholesterol 7α-hydroxylase

ASBT (SLC10A2 or IBAT)

apical sodium-dependent bile acid transporter

SHP

small heterodimer partner

MCA

muricholic acids

LCA

lithocholic acid

UDCA

ursodeoxy cholic acid

AMP

ampicillin

WT

wild-type

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

Bile acid (BA) metabolism is essential in cholesterol homeostasis, regulating absorption and elimination of cholesterol. BAs also influence glucose and lipid metabolism and energy expenditure [1, 2]. The BA pool is maintained by the balance between synthesis, largely controlled by Cyp7a1 and excretion, controlled by the apical sodium-dependent BA transporter (ASBT or IBAT) [3], resulting in >95% return of intestinal BAs to the liver. These processes are under negative control by the farnesoid X receptor (FXR). BA synthesis is partly controlled in the liver through the hepatic FXR–small heterodimer partner (SHP) axis, and presumably from the intestine where FXR drives the formation and secretion of fibroblast growth factor (FGF)15 inhibiting BA synthesis [4-7].

Two primary BAs, cholic acid (CA) and chenodeoxycholic acid (CDCA), are synthesized in human and mouse liver and converted by intestinal microbes to deoxycholic acid (DCA) and lithocholic acid (LCA), respectively. CDCA is abundant in human bile, but not in mice because it is converted to muricholic acids (MCAs) [8] by mechanisms that remain unclear [9]. It has previously been shown in CA-fed mice that addition of β-MCA or ursodeoxycholic acid (UDCA) to the diet causes a significant twofold stimulation of Cyp7a1 enzymatic activity [10]. Further, exposure of human hepatic cells to physiological levels of MCA strongly and dose-dependently increases CYP7A1 protein expression [11]. We interpret these results as that MCAs and UDCA may stimulate BA synthesis through a positive feedback mechanism. Recently, it was demonstrated that α- and β-MCAs possess FXR-antagonistic properties in an in vitro coactivator recruitment assay [12]; this characteristic is shared with UDCA [10, 13, 14]. However, to our knowledge, whether BAs with FXR-antagonistic properties have any relevance for the regulation of BA synthesis in vivo has not been previously investigated.

Chenodeoxycholic acid and DCA potently suppress CYP7A1 in human primary hepatocytes [15], and reporter gene assays show that CDCA and DCA are the strongest FXR agonists [11, 14, 16]. However, studies employing CA-free mice (Cyp8b1−/−) have demonstrated that CA is a more potent suppressor of BA synthesis in mice than CDCA [17]. In such mice, induction of Cyp7a1 mRNA is increased 4.5-fold and the BA pool is increased too [17]. However, Cyp8b1−/− mice also lack the potent FXR agonist DCA, whilst their expanded BA pool largely consists of α- and β-MCAs and UDCA [17]. It is interesting that this phenotype (increased MCAs and induced BA synthesis) is shared with mice that have undergone bile duct ligation [18], mice administered antibiotics such as ampicillin (AMP) [19] and germ-free mice [12]. The results of studies on such mice have suggested that the induced BA synthesis is due to reduced hormonal signalling of FGF15 from the intestine to the liver. This mechanism was initially proposed by Inagaki et al. [7] and later supported by others [12, 19]. A mutual denominator in these models that has been overlooked is that potent FXR-agonistic BAs are either absent or strongly reduced. We hypothesized that the absence of the potent FXR agonist DCA may be important for the induction of BA synthesis in these models.

In this study, we investigated two of these models alone and in combination: antibiotic treatment and Cyp8b1−/− mice. Our data show that AMP treatment of wild-type (WT) mice creates a phenotype closely resembling that of Cyp8b1−/− mice. This phenotype is difficult to explain from the alterations of the FXR-agonistic BAs DCA, LCA, CA or CDCA. We conclude that this phenotype is rather due to increases in α- and β-MCAs and UDCA, three BAs that can abolish the effects mediated by FXR-agonistic BAs.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

Reagents

Ampicillin, CA, CDCA and DCA were purchased from Sigma-Aldrich (St Louis, MO, USA), and Isolute SI solid phase extraction columns were from Biotage GB Ltd (Hengoed, UK). Primary antibodies against ASBT, β-actin and GAPDH were purchased from Abcam (Cambridge, UK). All primers for the real-time polymerase chain reaction (PCR) were obtained from Cybergene AB (Huddinge, Sweden).

Animals and treatments

For this study, we used 8- to 10-week-old female Cyp8b1−/− (knockout; KO) mice and their littermates Cyp8b1+/+ (WT), inbred as previously described [17], housed in a temperature-controlled and pathogen-free room under a 12-h light–dark cycle. Groups of WT or KO mice (= 5–6) received single doses of saline (SAL; 100 μL) with/without AMP (100 mg kg−1 day−1) by gavage. Another four groups of KO mice (= 3–6) were given oral gavage of 200 μL saline with AMP (100 mg kg−1 day−1) alone or supplemented with CA, CDCA or DCA at a concentration to provide 9 μmol BA kg−1 day−1. All treatments were administered at 09.00 for 3 days, with one additional dose at 17.00 on day 2. Mice were fasted for 4 h with free access to water before being killed. All animal experiments were approved by the institutional animal care and use committee.

Biochemical analyses

Quantitative real-time assays for mRNA measurements were performed as previously described [20], utilizing the housekeeping gene HPRT.

The enzymatic activity of hepatic Cyp7a1 was determined as described previously [21] using D7-7α-hydroxycholesterol as an internal standard. Data were corrected for the concentration of protein, determined using Bio-Rad DC protein assay (Bio-Rad Laboratories, Hercules, CA, USA).

Hepatic and faecal BAs were assayed by gas chromatography mass spectrometry [22]. The Western blotting methods have been previously described [20], and results were recorded using a Molecular Imager system (Bio-Rad Laboratories).

Statistical analyses

Data are presented as means ± SEM. Differences in the effects of SAL and AMP treatment between WT and KO mice were analysed by two-way anova or unpaired t-test followed by post hoc Fisher's LSD test; differences between BA treatments were analysed by one-way anova or unpaired t-test followed by post hoc Fisher's LSD test (statistica software; StatSoft, Tulsa, OK, USA; GraphPad Prism, La Jolla, CA, USA). Correlations were tested by calculation of Spearman's rank correlation coefficients. P values <0.05 were considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

AMP-treated WT mice resemble Cyp8b1−/− mice

Administration of AMP to WT mice increased Cyp7a1 mRNA levels 4.4-fold and cholesterol 7α-hydroxylase activity 3.5-fold (< 0.01, Fig. 1a,b). These values were similar to those of SAL-treated Cyp8b1−/− mice. By contrast, there were no significant changes in Cyp7a1 mRNA levels and cholesterol 7α-hydroxylase activity as a result of AMP treatment of Cyp8b1−/− mice, although trends towards increases in both were seen.

image

Figure 1. Effect of ampicillin (AMP) on liver bile acids in wild-type (WT) and Cyp8b1−/− (KO) mice. Mice (n = 5–6 per group) were treated with saline (SAL) or AMP for 3 days. (a) Real-time polymerase chain reaction (PCR) analysis of liver Cyp7a1. (b) Enzymatic activity of liver cholesterol 7α-hydroxylase. Real-time PCR analysis of liver small heterodimer partner (SHP; c), ileal fibroblast growth factor 15 (d) and ileal SHP (e). Data represent mean ± SEM. *< 0.05; **< 0.01.

Download figure to PowerPoint

Because Cyp7a1 is partly regulated by the hepatic FXR–SHP axis, we quantified hepatic SHP mRNA but found no significant differences between groups, although there were consistent trends towards 50% lowered levels in AMP-treated WT animals and in all Cyp8b1−/−mice (Fig. 1c); the latter finding is in agreement with previous reports [17, 23]. Ileal SHP mRNA showed a similar pattern to that in the liver, although changes in the ileum were statistically significant (Fig. 1e). Another factor linked to repression of Cyp7a1 enzymatic activity was ileal FGF15 expression. In WT animals, AMP treatment reduced the ileal FGF15 mRNA level to 35% of that in SAL-treated WT animals (Fig. 1d). In SAL-treated Cyp8b1−/− mice, the levels were 45% of those in WT mice treated with SAL, and treatment of these animals with AMP decreased FGF15 mRNA to only 15% of the levels in SAL-treated WT mice. The gene expression of FGF15 in all animals showed inverse correlations with both mRNA and enzymatic activity of Cyp7a1 as well as with the gene expression of ASBT (Figure S1).

As the BA pool is increased in Cyp8b1−/− mice [17], we measured the level of the BA transporter ASBT in the distal ileum. Compared to SAL-treated WT mice, the ASBT protein in SAL-treated Cyp8b1−/− mice was increased ~2.5-fold, and AMP-treated WT mice had a fourfold increase in ASBT (Fig. 2b). AMP treatment of Cyp8b1−/− mice further increased the ASBT expression to similar levels as in AMP-treated WT mice. The mRNA levels of ASBT showed similar expression patterns compared to the protein, but the differences between groups were somewhat smaller. The highest expression level was seen in the group with lowest FGF15 expression: the AMP-treated Cyp8b1−/− mice (Fig. 2a).

image

Figure 2. Effect of ampicillin on intestinal apical sodium-dependent bile acid transporter (ASBT) and faecal bile acid (BA) excretion in wild-type (WT) and Cyp8b1−/− (KO) mice. (a) Real-time polymerase chain reaction analysis of ileal ASBT mRNA and (b) representative western blot (pooled samples) of ileal ASBT/GAPDH with quantified bands. (c) Faecal total BA excretion (24 h). (d–f) Individual BA composition (grouped as hydrophobic, hydrophilic and CA) (n = 5–6/group). DCA, deoxycholic acid; LCA, lithocholic acid; CDCA, chenodeoxycholic acid; CA, cholic acid; α-MCA, α-muricholic acid; β-MCA, β-muricholic acid; UDCA, ursodeoxycholic acid. Data represent mean ± SEM. Statistical significance in panels d–e represents comparisons of total hydrophobic/hydrophilic BAs; *P < 0.05; **P < 0.01.

Download figure to PowerPoint

AMP administration reduces faecal BA excretion in WT and Cyp8b1−/− mice

Faecal total BA excretion was considerably reduced by AMP administration in WT and KO mice (25-fold and 15-fold decrease per 24 h compared to the respective control groups, both < 0.01, Fig. 2c), in line with previous results [19, 24]. This finding indicates that AMP treatment markedly enhances intestinal BA absorption. SAL-treated Cyp8b1−/− mice showed an 85% higher faecal BA excretion compared to SAL-treated WT control animals (< 0.01), in accordance with earlier reports [17]. Analysis of faecal BA composition (Fig. 2d–f) showed that the major BAs in WT mice were DCA, β-MCA and CA (41%, 29% and 13% respectively), whereas the major faecal BAs in Cyp8b1−/− mice were β-MCA, LCA and α-MCA (39%, 24%, and 23%, respectively) (see Table S1 for details of individual faecal BAs).

Effects of genetically or AMP-induced depletion of DCA

Analysis of hepatic BAs showed that total BAs were increased by 77% in Cyp8b1−/− mice (Fig. 3a), in agreement with previous findings of an increased BA pool in these animals [17]. This was due to increases in β-MCA, α-MCA, UDCA and CDCA of 69%, 290%, 330% and 610%, respectively, whilst CA and DCA were undetectable (Fig. 3b–d). Notably, AMP treatment of WT mice increased total hepatic BAs by 62%, due to increases in CA, β-MCA, α-MCA, UDCA and CDCA of 49%, 48%, 190%, 13% and 240%, respectively, whereas DCA levels were reduced. To illustrate the changes in hydrophilic BAs, the sum of the three hydrophilic BAs (α- and β-MCAs and UDCA) was compared between groups. This total mass was increased by 109% in Cyp8b1−/− compared to WT mice (< 0.05) and by 64% following AMP administration in WT mice (< 0.05); in AMP-treated Cyp8b1−/− mice, there was only a 12% increase of these FXR antagonists compared to SAL-treated Cyp8b1−/− mice (see Table S2 for details of individual liver BAs).

image

Figure 3. Effect of ampicillin on hepatic bile acid (BA) content in wild-type (WT) and Cyp8b1−/− (KO) mice. (a) Liver total BAs, (b–d) individual BA content (grouped as hydrophilic, hydrophobic and CA). DCA, deoxycholic acid; LCA, lithocholic acid; CDCA, chenodeoxycholic acid; CA, cholic acid; α-MCA, α-muricholic acid; β-MCA, β-muricholic acid; UDCA, ursodeoxycholic acid. Data represent mean ± SEM, n = 5–6/group. *P < 0.05.

Download figure to PowerPoint

Effects of BA administration on Cyp7a1, hepatic SHP and intestinal FGF15, SHP and ASBT

To investigate the effects of different BAs, AMP-treated Cyp8b1−/− animals received CA, DCA or CDCA by gavage for 3 days. In this model, endogenous CA and DCA are absent, and AMP treatment suppresses synthesis of secondary BAs [19]. In the CA- and DCA-treated groups, Cyp7a1 mRNA levels were suppressed by 81% and 82%, respectively, compared to control animals (Fig. 4a). CDCA treatment was less effective and reduced Cyp7a1 mRNA by only 59%. The changes in enzymatic activity of Cyp7a1 were similar to those for the mRNA (Fig. 4b). In addition, the changes in hepatic SHP mRNA levels paralleled those of Cyp7a1 mRNA; levels were clearly highest in the groups treated with CA or DCA (Fig. 4c) whilst there was no significant difference between the control and the CDCA-treated groups. The suppressed, basal FGF15 expression in AMP-treated KO animals was significantly increased in all BA-treated groups: in the CA and DCA groups by 4.1-fold and in the CDCA group by 3.5-fold (Fig. 4d). Intestinal SHP mRNA expression was increased in all three BA-treated groups (Fig. 4e) showing a similar pattern to that of ileal FGF15 mRNA expression.

image

Figure 4. Effects of bile acid (BA) administration to ampicillin (AMP)-treated Cyp8b1−/− (KO) mice on BA metabolism in liver and intestine. KO mice were treated with AMP. Groups of mice received, by gavage, saline (KO+AMP, n = 3), cholic acid (KO+AMP+CA, n = 6), chenodeoxycholic acid (KO+AMP+CDCA, n = 5) or deoxycholic acid (KO+AMP+DCA, n = 6) for 3 days. (a) Real-time polymerase chain reaction (PCR) analysis of liver Cyp7a1 mRNA. (b) Liver Cyp7a1 activity. Real-time PCR analysis of liver SHP (c), ileal FGF15 (d) and ileal SHP (e). Data represent mean ± SEM. *< 0.05; **< 0.01.

Download figure to PowerPoint

Finally, ASBT gene expression was reduced by 45% in the CA and CDCA groups, whilst a somewhat stronger 65% reduction was observed in the DCA group (Fig. 5a). ASBT protein expression was suppressed by 40% in the CA and the DCA groups and by 60% in the CDCA group (Fig. 5b).

image

Figure 5. Effect of bile acid administration to ampicillin-treated Cyp8b1−/− (KO) mice on ileal apical sodium-dependent bile acid transporter (ASBT). (a) Real-time analysis of ileal ASBT mRNA. (b) Representative Western blot (pooled samples) of ileal ASBT/β-actin protein expression with densitometry quantification of the representative bands. Data represent mean ± SEM. *P < 0.05; **P < 0.01.

Download figure to PowerPoint

Effects of BA administration on hepatic BA levels

Total hepatic BA levels were reduced by 30–34% in CA- and DCA-treated mice, whereas the CDCA group showed a 30% increase compared with control animals (Fig. 6a). Analysis of individual BAs showed that total hydrophilic BAs (α- and β-MCAs and UDCA) (Fig. 6b) were 39% less abundant in the CA group; the same comparison for the DCA group showed 34% lower mass, whereas the CDCA group had slightly increased levels of hydrophilic BAs (+22%) compared to controls. Notably, α-MCA, the most potent FXR antagonist [12], was reduced by 68% in both CA- and DCA-treated mice. By contrast, CDCA increased by 66% following administration of this BA, compared to control animals, but was reduced by 89% in the CA and the DCA groups (Fig. 6c), indicating that this potent FXR activator is unlikely to explain why Cyp7a1 suppression was highest in the CA and DCA groups. In DCA-treated mice, DCA was detectable (Fig. 6c) at similar low levels as initially observed in WT animals (Fig. 3c). This finding suggests that this level of hepatic DCA does not suppress Cyp7a1 and therefore does not support our initial hypothesis that lack of DCA could be the cause of the induction of BA synthesis in Cyp8b1−/− or in AMP-treated WT mice. (see Table S3 for details of individual liver BAs).

image

Figure 6. Effect of bile acid (BA) administration to ampicillin-treated Cyp8b1−/− (KO) mice on liver BAs. KO mice were treated as described in the legend to Fig. 4. (a) Total BAs. (b–d) Individual BAs (grouped as hydrophobic and hydrophilic BAs and CA). DCA, deoxycholic acid; CDCA, chenodeoxycholic acid; CA, cholic acid; α-MCA, α-muricholic acid; β-MCA, β-muricholic acid; UDCA, ursodeoxycholic acid; ND, not detectable. Data represent mean ± SEM. Statistical significance in panels b and c represents comparisons of total hydrophobic/hydrophilic BAs; *P < 0.05.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

The present data expand current knowledge not only regarding the phenotype of Cyp8b1−/− mice, but also in particular of the regulation of BA synthesis in mice in general. Cyp8b1−/− mice have an enlarged total BA pool [17], reflected by the hepatic BA contents (Fig. 3a). Further, their total BA synthesis is increased fourfold, in agreement with previous data [17]. An important novel finding was that Cyp8b1−/− mice have a more than twofold increase in intestinal expression of ASBT, a protein reported to be negatively regulated by FGF15 [25], the latter in line with the current findings. The increased expression of this highly efficient BA pump should severely reduce the faecal excretion of BAs, as demonstrated (Fig. 2c).

Together, these data strongly suggest that the total mass of BAs in the enterohepatic circulation is increased several fold in AMP-treated WT mice and in Cyp8b1−/− mice with or without AMP. Nevertheless, the intestinal FXR-driven FGF15 expression was suppressed by 65–90% in all these groups of mice. This finding is important as it strongly suggests that the FXR-signalling ability of this large mass of BAs is severely impaired. The cause for this could either be the decrease in the level of potent FXR-agonistic BAs and/or the increase in FXR-antagonistic BAs. Interestingly, analysis of hepatic BAs in Cyp8b1−/− mice showed, in line with previous data [17], that DCA was undetectable and that the FXR-antagonistic BAs MCA and UDCA together accounted for an impressive increase of 109% compared to WT mice.

It may be argued that the absence of CA in Cyp8b1−/− mice may cause this reduced FXR signalling. However, the results from the AMP-treated WT mice are not compatible with this hypothesis. Of note, AMP-treated WT mice displayed a phenotype similar to that of Cyp8b1−/− mice: (i) the same level of induction of Cyp7a1; (ii) the same reductions in FGF15 and trends towards reduced hepatic SHP; (iii) similar increases in the gene and protein expression of ASBT; and (iv) a 62% increase in total hepatic BAs of which 82% was due to increased amounts of MCA+UDCA. It is important to note that all these changes occurred concomitantly with sustained levels of hepatic CA. Therefore, in Cyp8b1−/− mice, the absence of CA, defined as a weak FXR activator [11, 14, 16], is highly unlikely to explain why BA synthesis is derepressed and, consequently, does not support the notion that CA is an important suppressor of BA synthesis in mice. It may also be argued that an AMP-induced reduction in the highly potent FXR agonist LCA in the liver derepresses BA synthesis within LCA levels in the liver that are undetectable. However, CDCA, the precursor of LCA, was strongly increased ninefold in Cyp8b1−/− mice. Furthermore, the fact that BA synthesis was not significantly induced in AMP-treated Cyp8b1−/− mice, although faecal LCA was eliminated in these mice, does not support this possibility. Therefore, we finally questioned whether the deficiency of the potent FXR agonists CDCA and/or DCA could explain why BA synthesis was increased in Cyp8b1−/− mice.

Insight into this was obtained by administration of CA, CDCA or DCA to AMP-treated Cyp8b1−/− mice, in which CA and DCA are absent and the intestinal formation of secondary BAs is blocked by eradication of intestinal microbiota. Only administration of CA or DCA, but not CDCA, strongly and equally reduced Cyp7a1, in agreement with previous findings [17]. This was in concert with the hepatic expression of SHP as well as with the FGF15 expression data. In addition, the intestinal expression of FGF15, which suppresses ASBT [25], was increased in all three BA-treated groups in line with the reduction in ASBT expression. Hepatic BA analysis showed that DCA was only detected in the DCA group. A striking feature of the CA and DCA groups was that total hepatic BAs were reduced by 30–34%. This was due to a 33% and 39% reduction in hydrophilic BAs in the CA and DCA groups, respectively. This in turn was almost solely due to a 70% reduction in α-MCA, which is a more potent FXR antagonist than β-MCA [12]. In CDCA-treated mice, hydrophilic BAs were instead increased, particularly α-MCA which increased threefold (< 0.05), compared to the CA- and DCA-treated groups. This expansion of hepatic FXR-antagonistic BAs in the CDCA-treated group would be expected to strongly counteract the suppression of Cyp7a1, mediated by CDCA, which was 15-fold higher in CDCA-treated mice compared to the CA or DCA groups. Therefore, changes in CDCA levels cannot explain why Cyp7a1 was suppressed. The fact that Cyp7a1 expression was equally suppressed in CA- and DCA-treated mice, in which hepatic contents of FXR-antagonistic BAs were reduced by the same amount and CDCA levels were similar, indicates that the observed presence of hepatic DCA is unlikely to have any significant suppressive effect on Cyp7a1. Notably, a similar level of DCA was present in SAL-treated WT mice (cf. Figs 3c and 6c). Therefore, the elimination of DCA in AMP-treated WT mice or in Cyp8b1−/− mice (Fig. 3c) is unlikely to explain why BA synthesis is induced fourfold in these animals. It was recently suggested that the increased BA synthesis in germ-free mice may involve intestinal FXR-antagonistic MCA effects. However, the important finding that FXR-agonistic BAs were eliminated or strongly reduced in the entire germ-free animal was never addressed [12]. Therefore, the contribution of the elevated MCAs in germ-free mice, if any, to the induction of BA synthesis in these animals could not be assessed in that study.

In this study, we were able to systematically exclude changes in LCA, DCA, CDCA or CA as the cause of induction of BA synthesis in Cyp8b1−/− or in AMP-treated WT mice. To what extent CDCA and CA suppress BA synthesis in vivo in mice cannot be determined from the present results as their effects are clearly overridden by the presence of the antagonists. Reporter assay results show that CDCA is a much stronger FXR agonist than CA [11, 14, 16]. Therefore, CDCA should account for the major FXR agonism in the mouse under basal conditions. It is noteworthy that reduced hepatic levels of MCAs in AMP-treated Cyp8b1−/− mice following administration of CA or DCA, but not CDCA, are also seen in normal C57Bl/6 mice after administration of these BAs [8]; thus, hepatic MCAs are reduced >90% by CA or DCA treatment. Furthermore, CDCA administration increased total hepatic levels of MCAs and UDCA by 27%, indicating that this positive feedback mechanism is likely to be an important reason why CA, although a weak FXR ligand, has similar in vivo suppressive effects on Cyp7a1 compared to DCA. These results suggest a general significance of this positive BA feedback mechanism in the regulation of Cyp7a1 and indicate that regulation of BA synthesis from altered MCA levels in the liver may be at least as important as that of FXR-agonistic BAs.

The phenotype of Cyp8b1−/− mice may appear as outlined in Fig. 7. Absence of CA creates a reduction in endogenous FXR agonists, such as DCA. This reduction does not alter total BA synthesis, since the basal DCA-FXR-activity is low in WT mice as found in our study. Concomitantly, CDCA synthesis is induced that in turn is largely converted to MCAs and UDCA, BAs with FXR-antagonistic properties. This reduces intestinal FGF15 levels causing a derepression of ASBT [25], which increases the BA pool. In the liver, Cyp7a1 is derepressed by local antagonistic action of MCAs against FXR-bound CDCA, thereby inducing the synthesis of BAs that will further increase the pool. In this way, the positive feedback circle is accomplished causing the BA pool to expand until the maximal ASBT-dependent BA transportation capacity is exceeded thereby increasing BA excretion.

image

Figure 7. Schematic diagram of BA metabolism in WT and Cyp8b1−/− mice. Green arrows, direction of flow of BAs; magenta crossover, compounds absent in Cyp8b1−/−mice; magenta vertical arrows, increases/decreases of structures/compounds/BA flow). Cyp8b1−/− mice lack CA and DCA. Therefore, CDCA production is increased and converted to MCAs in the liver. As increased amounts of MCAs and CDCA enter the small intestine, their intestinal absorption increase together, presumably via both apical sodium-dependent bile acid transporter (ASBT) and ASBT-independent mechanisms. In the enterocyte, MCAs exert FXR-antagonistic effects overriding the agonistic effects of CDCA, thereby strongly reducing ileal FGF15 expression. This derepresses ASBT expression, which increases BA absorption enlarging the BA pool; this pool will be further enriched with MCAs from hepatic conversion of CDCA to MCAs. Likewise, in the hepatocyte, MCAs will exert FXR-antagonistic effects, derepressing Cyp7a1 expression and thereby increasing hepatic BA synthesis. In parallel, reduced FGF15 signalling from the gut to the liver may contribute to derepression of hepatic Cyp7a1 expression, although evidence for this is circumstantial. CA, cholic acid; CDCA, chenodeoxycholic acid; MCAs, muricholic acids; BA, bile acid; FXR, farnesoid X receptor; FGF15, fibroblast growth factor 15.

Download figure to PowerPoint

Of note, both Cyp8b1−/− and AMP-treated WT mice share a phenotype that can be explained by BA-mediated intestinal and hepatic responses without any FGF15 signalling from gut to liver. Importantly, FGF15−/− mice, used when the endocrine FGF15 hypothesis was outlined [7], also share this phenotype with an extremely large increase of 180% in total hepatic BA levels (70% or higher increase of total BAs) [26] and a 2.5-fold increase in Cyp7a1 mRNA and enzymatic activity [7]; a similar phenotype is also seen in beta-klotho deficiency [27]. Therefore, our proposed positive feedback model of regulation of BA synthesis, demonstrated both in Cyp8b1−/− and in AMP-treated mice, may also operate in FGF15−/− mice. Although intestinal ASBT expression has not been shown for FGF15−/− mice, it is likely to be difficult to maintain such a large BA pool if ASBT is not induced. Of particular interest in this regard are the findings of a recent study by Vergnes et al. [28] in mice with a natural mutation in the Diet1 gene. The Diet1 protein supports the gene and protein expression of intestinal FGF15. The authors noted that these mice resemble FGF15−/− animals having reduced intestinal FGF15 expression, twofold increased hepatic BA levels, a fourfold increase in Cyp7a1 mRNA and a 3.5-fold increase in intestinal ASBT mRNA levels.

Thus, it appears that induction of both ASBT and Cyp7a1 is important for the positive feedback mechanism. This is apparently the situation in both AMP-treated and Cyp8B1−/− mice, and presumably in FGF15−/− and beta-klotho−/− mice too and has indeed been reported for germ-free mice [12] and Diet1 gene mutant mice [28]. Although attempts to identify circulating FGF15 have failed [29], an endocrine action of an intestinal protein such as FGF15 can definitively not be excluded. However, it should be noted that the concept of a gut-derived inhibitor of BA synthesis is based on experiments in bile fistula rats infused either intravenously or into the duodenum respectively, with taurocholic acid [30, 31]. An important difference between these administration routes is that the duodenal route will enable deconjugation of BAs in the intestine whereas intravenous administration does not. Thus, the effect of BA composition is compared in addition to the route of administration. In our opinion, conclusions regarding inhibitory effects on BA synthesis by non-BA intestinal factors cannot be drawn from those results. Interestingly, data on liver regeneration in FGF15-deficient mice [26] and observations in FGFR4-deficient mice [32] also support the notion that bioactive FGF15 may not be present in mouse blood [33] provided that FGFR4 is the only FGF15 receptor. Thus, for the concept of FGF15 signalling from gut to liver to remain, the presence of bioactive FGF15 in portal blood must now be shown. Alternatively, FGF15 may also bind to other receptors.

In conclusion, our data indicate that the complex regulation of BA synthesis in mice, until now considered to be controlled by end-product inhibition by BAs, is also largely controlled by a highly potent BA-mediated positive feedback mechanism. The recognition of this mechanism should be a fundamental key for understanding BA metabolism in mice and humans as FGF15/19 is currently thought to operate similarly in both species. It will now be important to understand how and to what extent BA synthesis in humans is regulated by BA-mediated positive feedback as humans are highly deficient in MCAs [34]. In addition to UDCA, other BAs such as the dominant BAs, CDCA, CA and DCA are also likely to be involved in this regulation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank Ingela Arvidsson and Lisbet Bentin for expert technical assistance and Ingemar Björkhem for helpful criticism and advice. This work was supported by grants from the Swedish Research Council, the Swedish Heart-Lung Foundation, Stockholm County Council (ALF) and the Cardiovascular Program, Karolinska Institutet/Stockholm County Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
joim12140-sup-0001-TableS1-3.docxWord document19K

Table S1. Effect of AMP on fecal BA excretion in WT and Cyp8b1−/− mice.

Table S2. Effect of AMP on hepatic BA levels in WT and Cyp8b1−/− mice.

Table S3. Effect of BA administration on hepatic BAs in AMP-treated Cyp8b1−/− mice.

joim12140-sup-0002-FigureS1.pdfapplication/PDF177KFigure S1. Correlations between ileal FGF15 mRNA and Cyp7a1 mRNA, Cyp7a1 enzymatic activity or ileal ASBT mRNA.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.