In vivo mouse models to study bile acid synthesis and signaling

Abstract

The synthesis of bile acids (BAs) is carried out by complex pathways characterized by sequential chemical reactions in the liver through various cytochromes P450 (CYP) and other enzymes. Maintaining the integrity of these pathways is crucial for normal physiological function in mammals, encompassing hepatic and neurological processes. Studying on the deficiencies in BA synthesis genes offers valuable insights into the significance of BAs in modulating farnesoid X receptor (FXR) signaling and metabolic homeostasis. By creating mouse knockout (KO) models, researchers can manipulate deficiencies in genes involved in BA synthesis, which can be used to study human diseases with BA dysregulation. These KO mouse models allow for a more profound understanding of the functions and regulations of genes responsible for BA synthesis. Furthermore, KO mouse models shed light on the distinct characteristics of individual BA and their roles in nuclear receptor signaling. Notably, alterations of BA synthesis genes in mouse models have distinct differences when compared to human diseases caused by the same BA synthesis gene deficiencies. This review summarizes several mouse KO models used to study BA synthesis and related human diseases, including mice deficient in Cyp7a1, Cyp27a1, Cyp7a1/Cyp27a1, Cyp8b1, Cyp7b1, Cyp2c70, Cyp2a12, and Cyp2c70/Cyp2a12, as well as germ-free mice.

Introduction

The chemical structure and functions of bile acids (BAs) have been studied since the 1930s. In recent decades, the emerging role of BAs as more than just endocrine signaling molecules has shifted focus to better understand their importance in the onset of metabolic syndrome, hepatic and intestinal diseases, and cancer. Well-known functions of BAs include: (1) intestinal absorption of dietary lipids and fat-soluble vitamins, (2) major route of cholesterol catabolism and enhancing the biliary excretion, (3) regulating energy metabolism, (4) modulation of bacterial flora in the intestine and biliary system, and (5) endogenous ligands of farnesoid X receptor (FXR) and G protein-coupled bile acid receptor 1 (GPBAR1) [1–7]. While different BAs carry out similar functions, they vary in their ability to activate FXR, with chenodeoxycholic acid (CDCA) as the most potent ligand. The relative potencies of BAs in activating FXR are as follows: CDCA > lithocholic acid (LCA) = deoxycholic acid (DCA) > cholic acid (CA) (Fig. 1) [8, 9]. Alternatively, the primary BAs in mice, muricholic acids (MCAs), are potent FXR antagonists [10]. BAs are also known to activate GPBAR1 at varying potencies as follows: LCA > DCA > CDCA > ursodeoxycholic acid (UDCA) > CA [11]. FXR and GBPAR1 regulate several metabolic signaling pathways that are involved in lipid metabolism, glucose, and energy homeostasis, as well as inflammation. Therefore, any change or perturbation in the composition and size of BA pool may affect the degree of activation of these receptors, especially FXR, which will have a cascade of effects on a multitude of pathways [12, 13].

Fig. 1.

The classical and alternative pathways for bile acid (BA) synthesis. CYP7A1 and CYP8B1 are the major BA synthesis enzymes in the classical pathway. The classical pathway predominantly produces the primary BAs, cholic acid (CA) in the liver. CA and CDCA are potent FXR agonists. CA is further converted to a secondary BA, deoxycholic acid (DCA), in the small intestine. DCA is an FXR agonist, though it is less potent than CA. In mice, there is an additional BA synthesis enzyme CYP2A12 that converts DCA to CA, allowing an interconversion between secondary and primary BAs. CYP27A1 and CYP7B1 are the major BA synthesis enzymes in the alternative pathway. The alternative pathway predominantly generates the primary BA, chenodeoxycholic acid (CDCA) in the liver, and CDCA is a potent FXR agonist. CDCA is further converted to the secondary BA, lithocholic acid (LCA), in the intestines. LCA is ambiguous in its FXR activation. In mice, CDCA produced by the alternative BA pathway can be converted to another primary BA known as muricholic acid (MCA) by CYP2C70. α-MCA is the intermediate MCA found in smaller quantities. α-MCA is converted to β-MCA which acts as an FXR antagonist. In mice, α/β-MCA can also be converted to LCA.

BAs are predominantly synthesized in the liver by two pathways, the classical and alternative pathways. The reactions from cholesterol to BAs in classical pathway need 17 hepatic enzymes [14]. Under physiological conditions, the classical pathway generates approximately 60%–75% of BAs, and is initiated by the rate-limiting enzyme, cholesterol 7α-hydroxylase (CYP7A1), followed by subsequent modification by sterol 12α-hydroxylase (CYP8B1) to form CA. The alternative pathway accounts for the remaining 25%–40% of BAs and is initiated by sterol 27-hydroxylase (CYP27A1) and 7α-hydroxylase (CYP7B1) to form CDCA. CA and CDCA are primary BAs, which are conjugated by taurine or glycine (glycine exclusively in mice) (Fig. 1). Highly soluble conjugated BAs are readily outflowed from the liver by bile salt export pump and stored in the gallbladder. BAs are released postprandially into the small intestine to facilitate lipid absorption. In the ileum, conjugated BAs are taken up into enterocytes through the apical sodium-dependent bile salt transporter and are discharged across the basolateral membrane and released into portal circulation by organic solute transporters α/β (OSTα/β). In the liver, BAs are taken up by hepatic sinusoidal transporters including sodium taurocholate co-transporting polypeptide and organic anion transporting polypeptides. BAs not absorbed will reach the colon where they interface with microorganisms and are subjected to deconjugation and biotransformation (oxidation, epimerization or dehydroxylation) by microbial enzymes yielding secondary BAs, such as DCA and LCA.

To date, mouse models are the most common models to study BA signaling. However, the BA pools of humans and mice have notable species differences. Mouse liver produces CA, CDCA and UDCA, with the latter two BAs further converted to α- and β-muicholic acids (α/β-MCAs) by CYP2C70, which are not expressed in human. The α- and β-MCA are subjected to further biotransformation in the lumen of the intestine where microbial enzymes generate hyodeoxycholic acid (HDCA) and ω-MCA (Fig. 1). Another notable species difference is that mice express CYP2A12 that backconverts DCA to CA, whereas human livers do not have CYP2A12 (Fig. 1) [15].

BAs enterohepatically circulate, approximately 95% of the total BAs in the intestine are reabsorbed into portal circulation and approximately 5% are excreted in feces. The importance of BAs is evident as their synthesis and homeostasis have been shown to be tightly regulated. The regulation is carried out mainly by the FXR-fibroblast growth factor 19/15 (FGF19/15; 19 human ortholog; 15 mouse ortholog) mediated liver-gut axis through a negative feedback loop to prevent excessive BA levels. BAs are the endogenous ligands of FXR. In the distal small intestine, activation of FXR induces FGF19/15. FGF19/15 enters portal circulation and binds to fibroblast growth factor receptor 4/b-klotho complex in hepatocytes which downstream activates extracellular signal-regulated kinase 1/2 and c-Jun N-terminal kinase 1/2 pathways to strongly suppress the expression of genes in classical pathway, CYP7A1/Cyp7a1 and CYP8B1/Cyp8b1 [16–19]. To a lesser extent, the activation of FXR in the liver induces small heterodimer partner (SHP) that can suppress CYP7A1/Cyp7a1 and CYP8B1/Cyp8b1 [16, 20].

To understand BA functions in vivo and overlaps between humans and mice in BA composition and functions, mouse models have been used extensively to study BA homeostasis. To date, a handful of genetically modified mouse models have been developed to better elucidate the functions of BAs, the mechanisms by which they modulate lipid, glucose and energy metabolism. Furthermore, it is well understood that the dysregulation of BA homeostasis is associated with severe systemic and liver diseases, such as cerebrotendinous xanthomatosis (CTX), metabolic dysregulation-associated fatty liver diseases, atherosclerosis, insulin resistance and liver and colon cancers. This review aimed to highlight the current known mouse models available that researchers are utilizing to further advance our understanding of BAs and associated diseases and how they may serve as therapeutic targets to treat these diseases.

Cyp7a1 knockout (KO) model

CYP7A1 is a primary mediator of BA synthesis in the classic pathway with its regulation critically affecting many aspects of BA homeostasis. In humans, deficiency or deletion of the CYP7A1 gene causes atherogenic hyperlipidemia and an increases susceptibility to cholesterol gallstone disease (Fig. 2) [21]. However, mouse models with a genetic ablation of Cyp7a1, are not widely used to study the human disease of CYP7A1 deficiency. Instead, Cyp7a1 KO mice have been predominantly used to study the mechanisms of BA synthesis and cholesterol homeostasis.

Fig. 2.

Comparison of classical and alternative bile acid (BA) synthesis gene deficiencies between humans and mice. The mouse gene knockout (KO) models can vary greatly with human presentations of BA gene mutations as a result of stark species differences in the BA synthesis pathways. CYP7A1 mutations in humans lead to atherogenic hyperlipidemia, which can increase the risk of coronary artery disease. CYP7A1 mutations in humans also cause a higher incidence of cholesterol gallstones. Cyp7a1^−/− mouse models present with similar, if not more pronounced presentation metabolic imbalance. Cyp7a1^−/− mice present with lower birth weight, oily coats, vision defects, hyperkeratosis and an overall high mortality rate within the first few days of life. This presentation is consistent with an inability to absorb fat-soluble vitamins in the absence of BAs. Humans with mutations in CYP27A1 present with a serious autosomal recessive disorder known as cerebrotendinous xanthomatosis (CTX). CTX has many clinical manifestations including accelerated atherosclerosis, neurological abnormalities such as seizures and dementia, osteoporosis and elevated lesions of fat on the skin known as xanthomas. Cyp27a1^−/− mouse models, however, exhibit no human symptoms of CTX, with only moderate adrenal hypertrophy. The adrenal enlargement seen in mice has not been investigated in CTX patients. CYP8B1 mutations in humans have a notably positive effect on metabolic health. Human CYP8B1 deficiencies have been linked to greater whole body insulin sensitivity and these individuals are therefore at a reduced risk of diabetes. Cyp8b1^−/− mouse models similarly experience protection against metabolic disease. Cyp8b1^−/− mice experience lesser liver inflammation on a high fat diet (HFD) when compared to wild-type (WT) mice on a HFD diet. CYP7B1 mutations in humans cause serious liver diseases. Patients with deficiencies in CYP7B1 can experience accelerated end-stage liver disease as early as in infancy leading to liver cirrhosis and liver failure. Cyp7b1^−/− mouse models, however, have no liver injury and are relatively indistinguishable from WT mice.

In 1996, Ishibashi et al. [22] generated one of the earliest Cyp7a1 KO models. Heterozygous KO pups (Cyp7a1^+/−) were phenotypically normal, however, homozygous KO pups (Cyp7a1^−/−) had a high perinatal mortality rate. Cyp7a1^−/− pups presented normally at birth but died within the first 18 days of life. The BA pool composition of Cyp7a1^−/− mice revealed a ~50% decrease in CA, but no change in β-MCA compared to wild-type (WT) mice. HDCA, CDCA and α-MCA were detected only in the KO mice. While the total concentration of BAs in adult Cyp7a1^−/− mice was similar to that in WT mice, the types of BAs differed and fecal BA levels were reduced by 80% [23].

In addition, Cyp7a1^−/− pups had steatorrhea, indicating poor absorption of fats. Cyp7a1^−/− also had higher levels of bilirubin compared to WT mice, but maintained normal liver histology. There were no significant differences in serum triglycerides and cholesterol contents between Cyp7a1^−/− and WT mice, regardless of age. Cyp7a1^−/− pups maintained a low body weight, hyperkeratosis, and vision defects (Fig. 2). Cyp7a1^−/− mice also had lower levels of vitamin D3 and E compared to WT mice and the phenotypic abnormalities were consistent with fat-soluble vitamin deficiencies. Interestingly, nursing mothers as well as Cyp7a1^−/− pups presented a “Brylcream phenotype”, also known as an excessively oily coat. Diet supplementation with 1% CA reversed this phenotype while vitamin diet supplementation did not. The altered phenotype of Cyp7a1^−/− mice only presented in newborn mice. Cyp7a1^−/− pups exceeding 3 weeks of age or more were found to be nearly phenotypically indistinguishable to WT mice [21]. This phenotypic presentation was correlated to the induction of the alternate BA synthesis pathway. The Cyp27a1 gene encoding for the enzyme sterol 27-hydroxylase was not present at birth but was present by day 21 in both Cyp7a1^−/− and WT mice. The presence of sterol 27-hydroxylase at day 21 was correlated with the normal phenotype of Cyp7a1^−/− mice that could survive 3 weeks. While this study was conducted in mice of both sexes, the later study refutes the compensatory upregulation of sterol hydroxylase in Cyp7a1^−/− [23].

In 2001, Schwarz et al. [23] conducted a study with the objective of determining sex differences and the role of the alternative pathway in Cyp7a1 KO mice. This study found the opposite effect regarding alternative pathway responsiveness in the absence of Cyp7a1 in mice when compared to the study conducted by Ishibashi et al. [22]. Cyp7a1^−/− mice fed with cholesterol and cholestyramine, the well-known BA binding resin in the gut, had a significantly reduced BA pool and only retained half of the basal BA synthesis rate when compared to WT mice. While BA synthesis was overall decreased across both sexes of Cyp7a1^−/− mice, female mice maintained a larger BA pool with a higher content of CA. Dietary feeding of cholesterol in Cyp7a1^−/− mice in males caused a slight increase in hepatic cholesterol levels whereas in females, the level of hepatic cholesterol increased 3-fold. The synthesis and maintenance of BA homeostasis in Cyp7a1^−/− mice were not at all attributed to Cyp27a1 or the alternative pathway in this study, as CYP27A1 activity or mRNA levels did not vary as a function of Cyp7a1 genotype, sex, or dietary cholesterol intake [23].

In 2016, Ferrell et al. [24] generated a Cyp7a1^−/− model through backcrossing Cyp7a1^−/− mice in the mixed C57BL/6J/129Sv genetic background to mice with pure C57BL/6J background. The BA pool size was reduced by around 50%–60% in this strain of that in WT mice. Analysis of gallbladder BA composition revealed reduced taurocholic acid (TCA) and taurodeoxycholic acid (TDCA) concentrations with nearly double T-α-MCA concentration. The alteration of BA composition increased the hydrophilicity of the BA pool. Due to the reduced BA pool, feedback inhibition of BA synthesis in Cyp7a1^−/− mice was reduced as well, leading to greater expression of Cyp8b1 and Cyp7b1. This study also found that Cyp27a1 gene expression was upregulated with Cyp7a1 deficiency. The Cyp7a1^−/− mice on the Western diet also had a ~5-fold increase in the Cyp27a1 mRNA expression. Cyp8b1 and Cyp7b1 expressions were also further induced. Liver FXR mRNA expression was unaltered, but SHP mRNA expression was greatly increased. The substantial increase in T-α-MCA and T-β-MCA compared to other diets in Cyp7a1^−/− mice can be attributed to the high cholesterol content in the Western diet that can further stimulate the alternative BA synthesis pathway and increase BA hydrophobicity. Another interesting observation in the study is the relationship between CA levels and insulin resistance in Cyp7a1^−/− mice. An intraperitoneal (ip) injection of glucose into Cyp7a1^−/− mice resulted in significantly reduced blood glucose levels compared to WT mice [24]. As stated previously, the Cyp7a1^−/− mice with unsupplemented diets have markedly reduced concentration of TCA, the conjugated form of CA. This suggests the reduced CA concentration in Cyp7a1^−/− mice may improve glucose and insulin sensitivity.

Cyp27a1 KO mice

The Cyp27a1 KO models were generated to better understand the human disease CTX. CTX is an autosomal recessive disorder of BA metabolism and cholesterol homeostasis that causes systemic and neurologic abnormalities. Systemic abnormalities include osteoporosis and accelerated atherosclerosis while progressive neurological abnormalities include seizures and dementia (Fig. 2). CTX patients have a mutation in CYP27A1 gene that encodes for sterol 27-hydroxylase. The defective CYP27A1 gene of CTX patients causes a decreased production of CDCA and an increased production of 25-hydroxylated bile alcohols and cholestanol through the classical pathway of BA synthesis. Cholestanol accumulation results in xanthomas, lesions caused by a buildup of fats and cholesterol under the skin, and accelerated atherosclerosis (Fig. 2). Increased cholestanol level in serum is a clinical biomarker for CTX. Accelerated atherosclerosis in CTX is attributed to a lack of 27-hydroxylase in macrophages which is required to eliminate cholesterol in the reverse cholesterol transport pathway to prevent the formation of foam cells. Despite the xanthomas and premature atherosclerosis, CTX patients have normal levels of circulating serum cholesterol.

To better study the pathology and mechanisms of CTX in humans, a Cyp27a1 KO mouse model was created. In 1998, Rosen et al. [25] reported the characteristics of a Cyp27a1^−/− mouse model. They found that these mice had an appreciably less sensitivity to a defective Cyp27a1 gene. Most noticeably, Cyp27a1^−/− mice did not have the clinical manifestations of CTX in humans, with no visible xanthomas or atheroma produced (Fig. 2). In humans with CTX, BA production was markedly reduced, whereas Cyp27a1^−/− mice had near normal production of BAs. BA formation in the Cyp27a1^−/− mice was only 15% less than the normal, and this reduction of BAs was not associated with symptoms of lipid malabsorption [25]. The decrease in BA production eased negative feedback inhibition of the Cyp7a1 gene, which caused a 9-fold increase in 7α-hydroxylase, acting as an effective compensatory mechanism. Enzymes such as sterol 25-hydroxylase, 26-hydroxylase, and 24-hydroxylase are also thought to contribute to the BA pool in Cyp27a1^−/− mice in a similar compensatory mechanism. Cyp27a1^−/− mice retain normal plasma levels of retinol, tocopherol and vitamin D. Furthermore, Cyp27a1^−/− mice lack an accumulation of cholestanol which is another characteristic of CTX patients. Cholestanol production is increased in CTX patients due to a buildup of the compensatory 7α-hydroxylated intermediates, 7α-hydroxy-4-cholesten-3-one being the intermediate of greatest interest. 7α-hydroxy-4-cholesten-3-one is converted to cholestanol through a microsomal dehydratase enzyme [26]. This enzyme is found in much greater quantities in the human liver than in rodent livers, which may explain how Cyp27a1^−/− mice have less cholestanol accumulation than humans with a deficient CYP27A1 gene and may also explain the lack of visible xanthomas. Cyp27a1^−/− mice also have a less overall induction of the classical BA synthesis pathway via CYP7A1 when compared to CTX patients. This leads to less 25-hydroxylated bile alcohol accumulation in mice and may help explain the lack of CTX clinical manifestations.

A few years later in 2000, Repa et al. [27] studied the Cyp27a1^−/− model through the lens of feedback regulation and nuclear receptors. As seen from mRNA analysis, the lack of sterol 27-hydroxylase in Cyp27a1^−/− mice disrupted pathways regulated by the FXR. The mRNA concentrations of SHP, a downstream target gene of FXR, was reduced by 65% in Cyp27a1^−/− mice. There was a marked induction of Cyp7a1 leading to a 10-fold increase of Cyp7a1 mRNA. A substantial increase of Cyp8b1 mRNA was also seen in Cyp27a1^−/− mice, which helps explain the increase in CA in the intestinal BA pool of Cyp27a1^−/− mice. Repa et al. also highlighted how a depletion of Cyp27a1 maintains adequate BA synthesis and cholesterol homeostasis in both humans and mice. This is especially interesting because a depletion of Cyp27a1^−/− affects both the classical and alternative pathways of BA synthesis. This phenomenon was explained through studying the Cyp27a1^−/− mice and the observation that Cyp27a1^−/− pups did not require BA supplementation in their diets, unlike Cyp7a1^−/− knockout mice. Even though Cyp27a1^−/− mice have adequate BA levels at birth, their BA pool does not reach that of WT mice in adulthood. Regardless, Cyp27a1^−/− mice maintain whole body cholesterol homeostasis and this is because Cyp27a1^−/− mice absorb very small amounts of cholesterol, have compensatory increases of cholesterol synthesis in the liver and intestines, and have a much greater excretion of cholesterol through feces.

Another interesting discovery investigated by Repa et al. was the enlargement of liver and more notably, the enlargement of adrenal glands in Cyp27a1^−/− mice (Fig. 2) [27]. While it was previously discussed that whole body cholesterol homeostasis is maintained in Cyp27a1^−/− mice, a notable exception is the adrenal gland which had a 48% higher concentration of cholesterol as well as an increased concentration of non-cholesterol sterols. While supplemental CA feeding was able to stop the further increase in adrenal size, CA feeding was unable to reverse the large weight of the adrenal glands. It is theorized that the increase of adrenal gland size is due to a compensatory mechanism of absent sterol 27-hydroxylase in Cyp27a1^−/− mice. It is unknown if adrenal hypertrophy is seen in CTX patients.

Continuing the characterization of this KO mouse model, in 2001, Honda et al. [28] published a study yielding similar findings regarding the translatability of the KO model. This study further corroborated the differences in clinical phenotypes between humans with CYP27A1 deficiency and mice with Cyp27a1 deletion. Cyp27a1^−/− female mice had a much lower concentration of intermediates in BA synthesis, 25-hydroxylated BAs and cholestanol, when compared to CTX patients. The study proposes two possible reasons for the lack of sterol accumulation in Cyp27a1^−/− mice. One reason suggests that Cyp27a1^−/− mice have less upregulation and less activity of 7α-hydroxylase compared to CTX patients, leading to less sterol intermediates; the other reason suggested for the lack of sterol accumulation is that Cyp27a1^−/− mice have greater efficiency in metabolizing 25-hydroxylated bile alcohols. Cyp27a1^−/− mice have an appreciable upregulation of triol 25hydroxylase which can break down harmful 25-hydroxylated bile alcohols. This upregulation is not seen in CTX patients. CDCA, MCA and CA were all reduced in Cyp27a1^−/− mice compared to WT mice.

It is also important to note the differential activity of FXR when comparing humans with CYP27A1 deficiency and Cyp27a1 KO mouse models. In humans, CDCA is a major primary BA and a known strong, endogenous FXR agonist. A decrease in the concentration of CDCA in CTX leads to diminished activation of FXR, and a decrease in the expression of the bile salt export pump, which ultimately reduces the transportation of bile salts through the canalicular pathway [29]. Whereas in mice, the alternative pathway makes MCAs FXR antagonists [30]; in this analysis, deletion of Cyp27a1 in mice would not have reduced FXR activity, if not further activated.

Cyp8b1 KO mouse model

The Cyp8b1 gene is responsible for the synthesis of CA through coding the enzyme sterol 12α-hydroxylase. In humans with CYP8B1 deficiencies and mutations, there is an increase in plasma concentration of CDCA and a decrease in CA, which leads to an increase in whole-body insulin sensitivity (Fig. 2) [31]. This ties a compelling link between CYP8B1 and metabolic diseases such as diabetes that can be further explored in mouse models.

In 2018, Patankar et al. [32] published a study determining various pathologies including type 2 diabetes and metabolic dysregulation-associated steatohepatitis (MASH) through the framework of a Cyp8b1 KO model. Before the high fat diet was introduced, Cyp8b1^−/− mice had a lower body weight than the WT mice. The rate at which the mice gained weight on a high fat diet (HFD) was comparable between both Cyp8b1^−/− and WT mice. Notably, the Cyp8b1^−/− mice had less hepatic steatosis, lower liver weight and lower plasma aspartate aminotransferase (AST) activities when compared to WT mice on the same HFD (Fig. 2). The Cyp8b1^−/− mice may also have a reduced rate of de novo lipogenesis as seen in the decreased gene expression of transcriptional regulator, sterol regulatory element-binding protein (SREBP)-1c gene. Additionally, the model displays reduced absorption of lipids in the intestine as seen through increased expression of SREBP2. SREBP2 is a marker for cholesterol biosynthesis in the liver and therefore, an increase of cholesterol synthesis would indicate a compensatory mechanism for the reduction of cholesterol absorption. The degree of lipid droplet accumulation, inflammation, and fibrosis were also reduced in Cyp8b1^−/− mice on HFD. While Cyp8b1^−/− mice had greater protection from liver injury and hepatic steatosis than WT mice on a HFD diet, the same protection did not persist on a methionine-choline deficient (MCD) diet that pathologically mimics MASH in the liver. Though this diet has been replaced with other feeding compositions to induce MASH, the MCD diet suppresses fatty acid oxidation and prevents the release of very-low-density lipoproteins (VLDL). In a 3-week feeding study, both Cyp8b1^−/− and WT mice had similar levels of hepatic steatosis. Serum activities of alanine aminotransferase (ALT), levels of triglycerides and rate of weight change did not differ between the KO and WT groups, but Cyp8b1^−/− mice had a substantially higher release of VLDL than WT mice on the MCD diet [32].

In Cyp8b1^−/− mice, the lack of CA and the increase in MCA may explain the prevention of hepatic lipid accumulation, and therefore, the lesser degree of liver inflammation and injury [32]. The complete depletion of CA caused by the deletion of Cyp8b1 results in a drastic increase in MCA production. MCAs are weak BAs for intestinal cholesterol absorption due to their highly hydrophilic characteristics. This leads Cyp8b1^−/− mice to less intestinal cholesterol absorption while also having increased production of MCA from hepatic cholesterol, which may ultimately reduce cholesterol and lipid accumulation in the liver. MCAs have been identified as antagonists of the nuclear receptor FXR. Since FXR regulates the synthesis of fatty acids through the esterification of cholesterol, administration of FXR agonists like obeticholic acid has been observed to increase low density lipoprotein cholesterol [32]. Therefore, the increase in MCA, an FXR antagonist, may explain the repression of de novo lipogenesis in Cyp8b1^−/− mice. Through these studies, there is evidence that the Cyp8b1 deletion may present a beneficial effect and level of protection against hepatic injury that varies based on diet, at least in mice.

Cyp7b1 KO mouse model

The Cyp7b1 gene is responsible for the synthesis of CDCA through coding the enzyme oxysterol 7α-hydroxylase in the alternative BA synthesis pathway. In humans, CYP7B1 mutations cause severe illness that can rapidly progress to end-stage liver disease even in infants (Fig. 2). These severe liver diseases include progressive cholestatic hepatitis and hepatic fibrosis. Oxysterol 7α-hydroxylase deficiency caused by mutations in CYP7B1 is the rarest genetic metabolic disorder in the inborn errors of BA synthesis classification [33]. Mouse models, however, do not replicate the human pathogenesis of oxysterol 7α-hydroxylase deficiency. In 2000, Li-Hawkins et al. [34] created a Cyp7b1^−/− mouse model through the introduction of a null mutation at the Cyp7b1 locus. These mice had a complete absence of oxysterol 7α-hydroxylase, but did not present with prolific liver injury and were nearly indistinguishable to WT mice (Fig. 2). The oxysterol substrates for 7α-, 25- and 27-hydroxylase, were elevated in Cyp7b1^−/− mice. Plasma cholesterol, major tissue cholesterol and triglyceride levels as well as overall BA metabolism remained normal in Cyp7b1^−/− mice [34]. The only organs with decreased sterol biosynthesis in Cyp7b1^−/− mice are the kidneys of, specifically, male mice. The sex specific phenotype of male mice may be attributed to an inability to metabolize oxysterol intermediates in male kidneys.

Cyp7b1^−/− mice do not exhibit any characteristics of human CYP7B1 mutations. It is hypothesized that the absence of 7α-hydroxylase is compensated by upregulation of other BA synthesis pathways that can metabolize 25- and 27-hydroxycholesterol. Another possible reason for the species differences between mice and humans is the difference of BA composition and the subsequent alteration of FXR activation. Humans with CYP7B1 mutations will have decreased CDCA concentrations and will therefore exhibit less FXR agonism. Cyp7b1^−/− mice, however, will have decreased α/β-MCA concentrations and will therefore exhibit less FXR antagonist activity. Perhaps, differential FXR expression can be one of the major contributors of species differences between humans and mice in regard to CYP7B1 deficiencies, similar to the species differences seen in CYP27A1 deficiencies.

Cyp7a1/Cyp27a1 double KO (DKO) model

The primary difficulty of studying individual BA is the existence of various endogenous BAs in vivo using animal models, whereas in vitro models lack organ crosstalk, a feature which is important in regulating BA homeostasis. It becomes increasingly difficult to isolate and characterize individual BA amongst the presence of a mixed BA pool as seen in animal and human studies. Even with the individual KO of Cyp7a1^−/− and the Cyp27a1^−/− in mice, there still exists a substantial amount of endogenous BAs. In Cyp7a1^−/− KO mice, there still exists 25% of total BA amount when compared to WT mice. In Cyp27a1^−/− KO mice, there still exists 22% of total BA amount compared to WT mice [35]. Though these quantities of BAs in single KO models are significantly reduced, the presence of endogenous BAs even at the quantities they are present in, makes it difficult to further understand individual BA and their roles in receptor signaling. To be able to study individual BAs and the extent of individual BA signaling in vivo, Rizzolo et al. [35] characterized the novel low BA mouse model. This model is a DKO of the Cyp7a1 and of the Cyp27a1 genes, effectively disrupting both the classical and alternative pathways. The BA analysis showed a 45.9% reduction of BAs in the plasma, 60.2% reduction in the liver, 76.3% reduction in the gallbladder, 88.7% reduction in the small intestine and 93.6% reduction in the colon of DKO mice compared to WT mice [35]. Further inspection of the BA pool of DKO mice found surprisingly that the DKO mice maintained a similar composition of the 23 BA types profiled to that of WT mice. This elucidates that while the overall concentration of BAs decreases in DKO mice, the decrease stays proportional to individual BA types resulting in a similar BA pool composition.

In regard to FXR signaling, DKO mice had a lower basal function of FXR as seen with lower mRNA expression of FXR target genes such as hepatic SHP and ileal Fgf15 [35]. This can be attributed to the lower concentrations of BAs in DKO mice as lowering the endogenous ligand for FXR would decrease its activity. Though there is a decrease in FXR activity, FXR expression and response to activation are maintained as FXR is fully activated in the DKO mice with the use of a FXR synthetic agonist, GW4064. This is a promising development for the study of individual BA signaling as individual BA feeding may help understand the extent of individual BA functions, including the activation of FXR. Cyp7a1/Cyp27a1 DKO mice have a significantly reduced quantity of BAs while still supporting a similar BA pool makeup and similar FXR activity as compared to WT mice.

Cyp2c70 KO mouse model

Cyp2c70 is a murine specific BA synthesis gene not found in humans, and this gene encodes an enzyme that converts CDCA to MCAs, the mouse BA species from the alternative pathway. Therefore, a Cyp2c70^−/− mouse model will result in a more humanized model. In 2020, de Boer et al. [36] generated a hepatic Cyp2c70^−/− (Cyp2c70^ako) mouse model using CRISPR/Cas9 to better understand the role of Cyp2c70 in BA signaling and metabolism. In mice, Cyp2c70 converts CDCA to MCAs through the epimerization of the C7 hydroxyl group of CDCA. In Cyp2c70^ako mice, there was a significant reduction of β-MCA with a compensatory increase in CDCA. This shifts to a more hydrophobic and a less diverse BA pool, producing a more humanized BA composition. An increase in CDCA also leads to more humanized cholesterol absorption patterns. The stimulation of FXR in Cyp2c70 KO mice decreased fecal cholesterol deposition compared to that in WT mice. The alteration of fecal cholesterol deposition can be attributed to different BA properties and a difference in BA composition. MCA mixed micelles lack the ability to solubilize and absorb cholesterol, which in turn leads to more cholesterol deposition in the fecal matter. With a more humanized BA pool, human BA signaling and the pharmacological effects of FXR can be better modeled in Cyp2c70^−/− mice.

A follow-up study by Honda et al. [15] found a similar BA composition in Cyp2c70 KO mice. The Cyp2c70^−/− mice lacked MCAs and had a distinctly higher concentration of CDCA. Interestingly, the study by Honda et al. found that Cyp2c70^−/− mice had chronic liver inflammation, the extent of which depended on the unconjugated hepatic CDCA. Overall, there was also a total reduction in the BA pool of Cyp2c70^−/− mice and a subsequent decrease in FXR activation. The decrease in FXR activation was reported by a down-regulation of sodium taurocholate co-transporting polypeptide and multidrug resistance-associated protein 4, which was somewhat of an unexpected outcome. It is known that CDCA is the most potent endogenous agonist of FXR and because CDCA is at a higher concentration in Cyp2c70^−/− mice, it was hypothesized that there would be an increase of FXR activation in this model. However, there was no increase in FXR activation observed, even with a slight decrease of activation. This is an important finding which may suggest that in pathological conditions, involving higher liver weight and liver injury, prefer inflammation activated signaling pathways over the FXR pathway for BA biosynthesis and regulation [15]. A decrease in FXR signaling while also observing an increase in BA regulation and negative feedback must indicate that another, more inflammatory pathway is at work to regulate the BA pool, which in the Cyp2c70^−/− mice, leads to a higher liver weight and more liver injury.

Cyp2a12 KO mouse model

Similar to Cyp2c70, Cyp2a12 is a murine specific gene not found in humans. In humans, the gut microbiota converts the primary BAs, CA and CDCA, to secondary BAs, DCA and LCA. In mice, however, CYP2A12 converts the secondary BA, DCA, back to a primary BA, CA, a mechanism known as 7α-rehydroxylation. This species difference creates a notable difference in BA pool composition between humans and mice. Since secondary BAs are hydrophobic and are far more toxic than primary BAs, the Cyp2a12 gene in mice plays a potentially protective role in preventing the accumulation of secondary BAs. A Cyp2a12^−/− mouse model would remove this potentially protective mechanism and generate a more humanized model to better represent human disease states.

As a series of models to better study hydrophobic BA composition, Honda et al. [15] generated a Cyp2a12^−/− mouse model using CRISPR-Cas9 system. While the total BA pool size was similar to that in WT mice, the Cyp2a12^−/− mice had an elevated DCA accumulation. DCA is a known potent endogenous agonist of GP-BAR1. Robust elevations of DCA have the potential to alter GP-BAR1 signaling, especially in the context of metabolic syndrome where activation of the receptor is known to improve steatohepatitis, glucose homeostasis, inflammation in adiposity, atherosclerosis as well as energy expenditure [11, 37, 38]. The model also confirmed that Cyp2a12 is the primary gene responsible for 7α-rehydroxylation. This was demonstrated by the ~80% depletion of TDCA 7α-hydroxylase in Cyp2a12^−/− mice [15].

Cyp2a12/Cyp2c70 DKO mouse model

The primary BAs in WT mice are CA and MCA. Cyp2a12 and Cyp2c70 are two of the most important genes that contribute to species differences between mice and humans. As a part of a series of mouse models, Honda et al. generated a Cyp2a12/Cyp2c70 DKO model to mimic a more humanized model than individual murine specific gene KO models [15]. The DKO model had more extensive differences in BA composition when compared to the individual KO models. The DKO mice showed a marked elevation of CDCA, DCA and LCA. Similar to Cyp2c70^−/− mice, the DKO mice also presented chronic liver inflammation, a degree of which was proportional to unconjugated CDCA concentrations similar to Cyp2c70^−/− mice, and the DKO mice had a significant reduction in BA pool and no FXR activation. The lack of FXR activation in DKO mice suggests that alternate signaling pathways like the pregnane X receptor and the cytokine/c-Jun N-terminal kinase signaling pathway are the primary mechanisms for modulating BA signaling in hydrophobic BA environments.

Germ free (GF) model

Microorganisms that comprise the gut microbiome have a prolific impact on BA metabolism. The gut microbiota, primarily located in the distal small intestine and the colon, facilitate the deconjugation, dehydrogenation, oxidation-reduction and dehydroxylation of primary BAs. The metabolic modifications of BAs in the gut microbiota result in more chemically diverse secondary BAs. To better understand the characteristics and signaling utility of secondary BAs, a mouse model lacking the microorganisms of the gut microbiota can be used. In 2013, Sayin et al. [10] generated a GF mouse model and studied the differences in BA composition as compared to WT or conventionally raised (CONV-R) mice. CONV-R mice had a dramatic 71% reduction in BA pool compared to GF mice. This provides evidence that the gut microbiota has an essential function in preventing an overpopulation of BAs. Furthermore, CONV-R mice predominantly contained TCA in their proximal small intestine and CA in their distal small intestine [10]. This different BA composition highlights the efficacy of microbial deconjugation in the small intestine which then allows to further metabolite into secondary BAs in the colon. In GF mice, the BA pool almost exclusively included taurine conjugated BAs like Tβ-MCA and TCA as well as some TUDCA. This confirms that microbes are needed for BA deconjugation and establishes UDCA as a primary BA in mice as it is present without the microbes that cause secondary BA production.

The study by Sayin et al. also assessed FXR signaling in GF mice. Taurine conjugation plays an important role in FXR signaling with conjugated BAs being investigated as antagonists of FXR. Since CONV-R mice can deconjugate BAs rapidly, GF mice provide an ideal model to study the role of BA conjugation in FXR regulation. Both CONV-R and GF mice had TCA and Tα-MCA as the most common BAs in the liver. However, GF mice had lower levels of TCA and higher levels of Tβ-MCA. Due to the predominance of Tβ-MCA in GF mice, FXR was relatively inactive. The study was able to conclude that Tα-MCA & Tβ-MCA are competitive and reversible antagonists for FXR. Through additional comparisons, it was also found that the gut microbiota regulates FXR in the ileum through Fgf15 but does not influence hepatic FXR activation [10]. The gut microbiota, through metabolic modifications, increases the hydrophobicity of the BA pool. A more hydrophobic BA pool lends itself to greater FXR activation, while the absence of the gut microbiota and the accumulation of taurine conjugated BAs antagonize FXR activation.

Conclusions

BAs are universally acknowledged for their role in lipid digestion and absorption. However, the utility of BAs as signaling molecules, as well as the pathology of BAs in various metabolic diseases, is a frontier of utmost importance. Through the generation of various in vivo mouse models, BA synthesis and composition can be manipulated to better understand the characteristics of individual BAs and their role in enterohepatic signaling. There are vast differences in BA composition between humans and mice; through generating KO models with genes like Cyp2a12 and Cyp2c70, in vivo mouse models are getting more representative of human BA physiology. By using a GF mouse model, the importance of the gut microbiome in the metabolism of BA could be better understood. Through knocking out essential BA synthesis genes like Cyp7a1 and Cyp27a1, BA synthesis can be manipulated allowing for lower concentrations of endogenous BAs and the study of specific lipid imbalance disease states, respectively. The Cyp7a1/Cyp27a1 DKO mouse model is novel in this field for being able to extensively eliminate endogenous BAs and allow for exogenous BA feeding to isolate and characterize BA subtypes in vivo. A better understanding of BAs as endogenous signaling molecules, in lipid homeostasis, and in pathology is essential for the advancement of research into metabolic diseases. In vivo models allow for continued investigations into BAs for therapeutic benefits/targets and for potential clinical markers of disease.