Additional pathways of sterol metabolism: Evidence from analysis of Cyp27a1−/− mouse brain and plasma

Cytochrome P450 (CYP) 27A1 is a key enzyme in both the acidic and neutral pathways of bile acid biosynthesis accepting cholesterol and ring-hydroxylated sterols as substrates introducing a (25R)26-hydroxy and ultimately a (25R)26-acid group to the sterol side-chain. In human, mutations in the CYP27A1 gene are the cause of the autosomal recessive disease cerebrotendinous xanthomatosis (CTX). Surprisingly, Cyp27a1 knockout mice (Cyp27a1−/−) do not present a CTX phenotype despite generating a similar global pattern of sterols. Using liquid chromatography – mass spectrometry and exploiting a charge-tagging approach for oxysterol analysis we identified over 50 cholesterol metabolites and precursors in the brain and circulation of Cyp27a1−/− mice. Notably, we identified (25R)26,7α- and (25S)26,7α-dihydroxy epimers of oxysterols and cholestenoic acids, indicating the presence of an additional sterol 26-hydroxylase in mouse. Importantly, our analysis also revealed elevated levels of 7α-hydroxycholest-4-en-3-one, which we found increased the number of oculomotor neurons in primary mouse brain cultures. 7α-Hydroxycholest-4-en-3-one is a ligand for the pregnane X receptor (PXR), activation of which is known to up-regulate the expression of CYP3A11, which we confirm has sterol 26-hydroxylase activity. This can explain the formation of (25R)26,7α- and (25S)26,7α-dihydroxy epimers of oxysterols and cholestenoic acids; the acid with the former stereochemistry is a liver X receptor (LXR) ligand that increases the number of oculomotor neurons in primary brain cultures. We hereby suggest that a lack of a motor neuron phenotype in some CTX patients and Cyp27a1−/− mice may involve increased levels of 7α-hydroxycholest-4-en-3-one and activation PXR, as well as increased levels of sterol 26-hydroxylase and the production of neuroprotective sterols capable of activating LXR.


Introduction
Cerebrotendinous xanthomatosis (CTX) is an autosomal recessive disease caused by a defective sterol (25R)26-hydroxylase enzyme, also known as sterol 27-hydroxylase, (cytochrome P450 27A1, CYP27A1) [1,2]. In early infancy it can present with cholestatic liver disease, in early childhood with chronic diarrhoea and cataracts, in later childhood with tendon xanthomata, learning difficulties or psychiatric illness and in adult life with spastic paraparesis, a fall in IQ or frank dementia, ataxia and/or dysarthia [1]. Patients with CTX often present with premature atherosclerosis. CYP27A1 is the first enzyme in the acidic pathway of bile acid biosynthesis, it oxidises the terminal carbon of the cholesterol isooctyl side-chain first to an alcohol and subsequently to an acid introducing R stereochemistry at C-25 [3,4] (Fig. 1). The resulting products are (25R)26-hydroxycholesterol (cholest-5-ene-3β,(25R)26-diol) and 3β-hydroxycholest-5-en-(25R)26oic acid, respectively. Note, here we adopt the systematic nomenclature [5] recommended by the Lipid Maps consortium [6], although in much of the literature the non-systematic names 27-hydroxycholesterol and cholestenoic acid are adopted for these two products of CYP27A1 oxidation of cholesterol. CYP27A1 is also an essential enzyme of the neutral pathway of bile acid biosynthesis oxidising ring hydroxylated sterols at C-26 ultimately to 25R-acids. In light of its importance in bile acid biosynthesis, it is not surprising that sterol (25R)26-hydroxylase deficiency leads to disease in humans. Interestingly, although the formation of chenodeoxycholic acid (CDCA, 3α,7α-dihydroxy-5β-cholan-24-oic acid) is greatly reduced in CTX patients, biosynthesis of the other primary bile acid, cholic acid (3α,7α,12α-trihydroxy-5β-cholan-24-oic acid), is maintained [7]. This is achieved via the cholic acid precursor 3α,7α,12α,25-tetrahydroxy-5β-cholestan-24-one and elimination of the terminal three carbons as acetone with the formation of cholic acid (see Fig. 1 inset i) [8]. CTX can be treated by bile acid replacement therapy, especially with CDCA [9] and in combination with low-density lipoprotein (LDL)-apheresis and/or statins [10][11][12]. Surprisingly, knockout of the Cyp27a1 gene in mouse (i.e. Cyp27a1−/− mouse) does not lead to a CTX-like phenotype, although production of bile acids is markedly reduced [13].
Considering the importance of CYP27A1 in bile acid biosynthesis in human, and the biological activity of intermediates in the acidic pathway of bile acid synthesis [18][19][20][21], it is intriguing that the Cyp27a1−/− mice do not show a disease phenotype. In the current study we have investigated the profile of metabolites (> 50 sterols, oxysterols and sterol-acids) involved in the bile acid biosynthetic pathways in the Cyp27a1−/− mouse, concentrating on the circulation and the brain. Our results show that the Cyp27a1−/− mouse synthesises both oxysterols and cholestenoic acids with a 25S-stereochemistry. As the 25R-and 25S-epimers of the acids are inter-convertible this provides an additional route to the synthesis of 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid and also to bile acids. We have shown earlier that 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid promotes the survival of motor neurons [19]. Patients with CTX, but not Cyp27a1−/− mice, may present with motor neuron dysfunction, this difference could be explained by the presence of an additional route to synthesis of the neuroprotective acid in mouse. We hereby show the loss of Cyp27a1 in mice increases the levels of 7α-hydroxycholest-4-en-3-one, a pregnane X receptor (PXR) ligand known to increase the levels of CYP3A11 [24], resulting in expression of a murine sterol 26-hydroxylase and a pathway to neuroprotective LXR ligands.

Oxysterol analysis
We adopted a charge-tagging approach utilising "enzyme-assisted derivatisation for sterol analysis" (EADSA) to enhance liquid chromatography (LC) separation and mass spectrometry (MS) detection of oxysterols [17,25,26]. This involves the stereospecific enzymatic oxidation of the 3β-hydroxy-5-ene function in oxysterols (and sterols) to a 3-oxo-4-ene group and subsequent reaction with the cationic Girard P (GP) hydrazine reagent to give charged GP-hydrazones compatible with chromatographic separation using reversed phase solvents and highsensitivity analysis by electrospray ionisation (ESI)-MS and MS with multistage fragmentation (MS n ) (Fig. S1). GP-derivatives give intense [M] + ions in ESI and informative MS 2 and MS 3 spectra. As some oxysterols naturally contain a oxo group they give GP-derivatives even in the absence of oxidising enzyme. However, oxysterols containing a native 3-oxo group are readily differentiated from oxysterols oxidised to contain one by dividing each sample in two and performing derivatisation without oxidising enzyme on one portion of the sample (Fraction B) and performing derivatisation with added enzyme on the second portion (Fraction A), and by exploiting differentially isotope labelled GP reagents to allow discrimination by mass (Fig. S1).

CYP3A4 and CYP3A11 incubations
Recombinant mouse CYP3A11 in bactosomes (Cypex Ltd., Dundee, UK) or recombinant human 3A4 in baculosomes (Life Technologies now Thermo Fisher Scientific) (10 pmol) was incubated with 1-10 μg of 7αhydroxycholesterol (cholest-5-ene-3β,7α-diol), 2 mM NADPH, 5 mM glucose-6-phosphate and 0.4 U glucose-6-dehydrogenase in 0.1 M potassium phosphate buffer, pH 7.4, at a final volume of 500 μL for 16 h at 37°C. The reaction was quenched with ethyl acetate and the organic phase dried down for oxysterol analysis as above. Negative control experiments were performed in the absence of enzyme or NADPH. Incubations with CYP125, a known 26-hydroxylase, were performed to provide a positive control. Bile acid biosynthesis via the neutral, acidic, 25-hydroxylase and (25S)26-hydroxylase pathways. The 25-hydroxylase pathway is shown in inset (i), the mouse (25S)26-hydroxylase pathway in inset (ii). R = OH in acids or SCoA in CoA thioesters. R 1 = H or OH. Where known, mouse enzymes are indicated in bold, CYP3A11 was found in the present work to introduce a (25S)26-hydroxy group to 7α-hydroxycholesterol as indicated by underlining of the enzyme symbol. The enzyme which converts the (25S)26-primary alcohol to a carboxylic acid is indicated as a sterol oxidase (SO). Abbreviations: CYP, cytochrome P450; HSD, hydroxysteroid dehydrogenase; AKR, aldo-keto reductase; BACS, bile acyl-CoA synthetase (SLC27A5); VLCS, very long chain acyl-CoA synthetase (SLC27A2); AMACR, alpha-methylacyl-CoA racemase; ACOX2, branched chain acyl-CoA oxidase 2, also called branched-chain acyl-CoA oxidase; DBP, D-bifunctional protein or multifunctional enzyme type 2 (HSD17B4); SCPx, sterol carrier protein x. [3,4]. Metabolites of increased or decreased abundance in the Cyp27a1−/− mouse are indicated by upward or downward arrows. Red arrows are used to indicate changes in plasma, blue arrows for brain. A solid horizontal line indicates detected but not significantly changed. *, P < 0.05; ** P < 0.01; *** P < 0.001. P < 0.05 is considered significant. The low levels of di-and tri-hydroxycholesterols and of dihydroxycholestenoic acids in brain makes it difficult to distinguish between these compounds and their 3-oxo equivalents using EADSA as their differentiation is based on peak area difference between samples treated with and without cholesterol oxidase (see Fig. S1). Hence, for these metabolites the combined values for the two structures are used.

Luciferase reporter assay
The ability of oxysterols and various cholesterol metabolites to activate nuclear receptors i.e. PXR, constitutive androstane receptor (CAR), LXR, farnesoid X receptor (FXR), vitamin D receptor (VDR) and nuclear receptor related protein 1 (NURR1) was tested in luciferase assays. Transient transfections were performed in the mouse substantia nigra-like cell line SN4741. Cells were plated in 24-well plates (5 × 10 5 cells per well) 24 h before transfection, transfected with 1 μg of plasmid DNA per well and complexed with 2 μL of Lipofectamine 2000 (Invitrogen). Cells were transfected with 400 ng of a PXR-, CAR-, LXR-, FXR-, VDR-, or NURR1-responsive luciferase reporter construct and 200 ng PXR, CAR, LXRα, FXR, VDR, or NURR1 [19,24,27]. A reporter gene expressing the Renilla luciferase (pRL-TK, Promega) was cotransfected in all experiments as an internal control for normalization of transfection efficiency. After a 12 h incubation, the lipid/DNA mix was replaced with fresh 2.5% serum medium containing vehicle or appropriate ligand (10 μM), as specified in each experiment. Luciferase activities were assayed 24 h later using the Dual-Luciferase Reporter Assay System (Promega), following the manufacturer's protocol.

Primary midbrain cultures
Brains from E11.5 mouse embryos were obtained, the midbrain region was dissected, mechanically dissociated and plated on poly-Dlysine (150,000 cells/cm 2 ) and grown in serum-free N2 media consisting of 1:1 mixture of F12 and DMEM with 10 ng/mL insulin, 100 μg/ mL apo-transferrin, 100 μM putrescine, 20 nM progesterone, 30 nM selenium, 6 mg/mL glucose and 1 mg/mL BSA. Cells were treated for 3 days in vitro (DIV) with the compounds of interest, fixed with 4% PFA and processed for staining using appropriate antibodies. The fixed cells were washed in PBS and blocked in 5% normal goat serum/PBS for 1 h at room temperature. Primary antibodies were diluted in PBS (pH 7.4), 0.3% Triton X-100, 1% BSA and incubations were carried out overnight at +4°C or at room temperature for 2 h. The antibodies used were anti-: Islet-1 (1:100; Developmental Studies Hybridoma Bank) and Nkx6.1 (1:200; Novus Biologicals) and appropriate secondary antibodies (Jackson ImmunoResearch or Alexa). Cells positive for the corresponding marker were counted directly at the microscope at a magnification of 20×. Cells were counted in every well, in eight consecutive fields (going from one side of the well to the other, passing through the center), in three different wells per experiment and in three different experiments per condition. Random pictures of the wells were taken for every condition to document the result, and representative pictures were subsequently selected to represent the quantitative data. Photos were acquired with a Zeiss Axioplan microscope and a Hamamatsu camera C4742-95 using the Openlab software.

Hydroxycholestenoic acids
3β-Hydroxycholest-5-en-(25R)26-oic acid is present in appreciable amounts in plasma from the wt animals (3.56 ± 1.22 ng/mL) but is essentially absent from plasma of the Cyp27a1−/− animals (Fig. 2,  Fig. 4A, Table S1). This result is in agreement with the lack of detectable levels of its precursor (25R)26-hydroxycholesterol in plasma from Cyp27a1−/− animals. No evidence for the 25S-epimer of the acid was found in plasma of either genotype.
In wt mouse 7α-hydroxy-3-oxocholest-4-en-(25R)26-oic acid undergoes A-ring reduction and side-chain shortening through a complex series of reactions leading to the formation of CDCA (Fig. 1). Cytosolic aldo-keto reductases (AKR) 1D1 and 1C4 catalyse A-ring reduction, while side-chain shortening of the CoA thioester proceeds in the peroxisome catalysed by AMACR, branched chain acyl-CoA oxidase 2 (ACOX2), D-bifundtional protein (DBP, HSD17B4) and sterol carrier protein x (SCPx) [3]. The substrate for the final step of this series is the CoA thioester of the 24-oxo-(25R)26-carboxylic acid. With our analytical method we tend to observe the carboxylic acids rather than their CoA thioesters and β-keto acids are unstable and decompose by loss of CO 2 to the 26-nor-24-ketones [18]. As might be predicted, the concentration of 7α-hydroxy-26-nor-cholest-4-ene-3,24-dione is reduced from 1.33 ± 0.14 ng/mL in the wt mouse to 0.18 ± 0.05 ng/mL in the Cyp27a1−/− mouse. We do not have an authentic standard for 7αhydroxy-26-nor-cholest-4-ene-3,24-dione, but retention time and MS 3 spectrum are compatible with the proposed structure.

Trihydroxycholestenoic and dihydroxyoxocholestenoic acids
The RICs appropriate to trihydroxycholestenoic and dihydroxyoxocholestenoic acids are shown in Fig. 4C. Each of the four labelled peaks gives essentially an identical MS 3 spectrum. Considering wt plasma first, we assign the peaks at 3.69 and 4.27 min to the syn and anti forms of GP-derivatised 7α,12α-dihydroxy-3-oxocholest-4-en-(25R)26-oic acid (3.39 ± 1.46 ng/mL). In plasma from the Cyp27a1−/− mice the intensities of these two peaks are reduced (1.94 ± 0.18 ng/mL), although not significantly, however, two earlier eluting peaks at 3.53 and 4.06 min are enhanced which we assign to the syn and anti forms of the 7α,12α-dihydroxy-3-oxocholest-4-en-(25S)26oic acid (7.79 ± 0.88 ng/mL). These peaks are only minor in plasma from the wt animals (0.63 ± 0.31 ng/mL). The identification of the 25R-and 25S-epimers was confirmed in studies of the AMACR knockout mouse [32]. In combination, this data reinforces the notion that there is an additional route to introducing a carboxylic acid group to the terminal carbon of the sterol isooctyl side-chain in mouse besides that provided CYP27A1 (Fig. 1, insert ii). Honda et al. have proposed that mouse hepatic CYP3A11 has 26-hydroxylase activity towards 5βcholestane-3α,7α,12α-triol and 5β-cholestane-3α,7α,12α,25-tetrol and that the expression of this enzyme is up-regulated in the Cyp27a1−/− mouse [34]. Our data suggest that this, and/or another enzyme, also oxidises 7α,12α-dihydroxycholest-4-en-3-one at the C-26 position to the (25S)26-sterol acid. Once formed the (25S)26-and (25R)26-sterol acids are inter-convertible [37].

24S,25-Epoxycholesterol
Unlike other oxysterols, 24S,25-epoxycholesterol (3β-hydroxycholest-5-en-24S,25-epoxide) is formed through a shunt of the mevalonate pathway, specifically the Bloch arm of the pathway (Fig. S3). In the shunt pathway squalene epoxidase (SQLE) introduces two oxygen atoms into squalene rather than one and the enzyme 24-dehydrocholesterol reductase (DHCR24) is not involved. We observe low levels of 24S,25-epoxycholesterol (< 1 ng/mL) in plasma of the wt animal, but significantly higher amounts in plasma of the Cyp27a1−/ − animals (4.66 ± 0.72 ng/mL, Figs. 2, 3D, Table S1). Rosen et al. found that hepatic levels of HMG-CoA reductase (Hmgcr) mRNA are 2-3 fold higher in the Cyp27a1−/− animals than wt [13], while Båvner et al. found elevated levels of lathosterol, a marker of cholesterol synthesis, in liver of Cyp27a1−/− mice [38]. Thus, our data supports the concept of an enhanced flow of metabolites through the mevalonate pathway in Cyp27a1−/− mice and an increase in cholesterol biosynthesis. An alternative route to 24S,25-epoxycholesterol has recently been suggested by Goyal et al., who show that the human cholesterol 24S-hydroxylase (CYP46A1) enzyme can oxidise desmosterol (cholesta-5,24-dien-3β-ol) to the 24S,25-epoxide (Fig. S3) [39]. An elevation of 24S,25-epoxycholesterol formed via this route would similarly be compatible with an increase in flow of metabolites through the mevalonate pathway. An alternative explanation could be that knockout of Cyp27a1 removes a route for epoxide metabolism.

Analysis of brain cholesterol and metabolite levels in the Cyp27a1−/ − mouse
As humans with CTX can show spastic paraparesis, a fall in IQ or frank dementia and ataxia we next examined the oxysterol and sterol profile of brain from Cyp27a1−/− mice and wt controls.

Dihydroxycholesterols and dihydroxycholestenones
The low levels of these metabolites made it difficult to distinguish between dihydroxycholesterols and dihydroxycholestenones, as using EADSA their differentiation is based on peak area difference between samples treated with and without cholesterol oxidase. Hence, for these metabolites the combined values for the two structures are given, and for simplicity we just refer to them as dihydroxycholestenones. The pattern of dihydroxycholestenones in brain of both the Cyp27a1−/− and wt mice is complex (Fig. 5C). In the Cyp27a1−/− and wt mice we observe both 7α,25-and 7α,24-dihydroxycholest-4-en-3-ones. In the Cyp27a1−/− animals we see an elevation in the levels of these metabolites compared to wt. As the first peaks of the syn/anti pairs of both oxysterols almost co-elute we measured the combined amount for the two oxysterols (Figs. 5C, 6, Table S1). This gave a value of 0.12 ± 0.01 ng/mg in the Cyp27a1−/− animals compared with 0.03 ± 0.00 ng/mg in the wt. As in plasma, we find 7α,(25S)26-dihydroxycholest-4-en-3-one in brain of the Cyp 27a1−/− mouse (0.04 ± 0.00 ng/mg) and also traces of the 25R epimer (0.01 ± 0.00 ng/mg). In the wt animal the 25R epimer is present at trace levels (0.01 ± 0.00 ng/mg) while the 25S isomer is essentially absent (< 0.01 ng/mg). The exact origin of these 7α,26-dihydroxy metabolites in brain is not clear; a possible origin in the Cyp27a1−/− mice is that they formed in brain by side-chain hydroxylation of imported 7α-hydroxycholesterol or 7α-hydroxycholest-4-en-3-one by up-regulated CYP3A11 (Fig. 1, inset ii), which is reported to be expressed in brain [52]. Alternatively, CYP46A1, which is abundantly expressed in brain has 25-hydroxylase, (25R)26-hydroxylase as well as 24S-hydroxylase activity to cholesterol and can also use 7α-hydroxycholesterol as a substrate and thus may account for some of the dihydroxy metabolites found in the two genotypes (Fig. S7) [53]. There is likely to be also some direct 7α-hydroxylation of side-chain  hydroxylated substrates by CYP7B1 accounting for 7α,(25R)26-dihydroxy metabolites in the wt brain. CYP7B1 is an oxysterol 7α-hydroxylase which is expressed in brain and 7α-hydroxylates both 25-and (25R)26-hydroxycholesterols but has only minor activity towards 24Shydroxycholesterol [3,54], however, this latter substrate is greatly dominating in brain, and its hydroxylation by CYP7B1 may account for the observation of some of the 7α-hydroxy metabolite. CYP39A1, the oxysterol 7α-hydroxylase which acts on 24S-hydroxycholesterol is expressed in many tissues. The Cyp39a1 gene is reported to be expressed in the somatosensory cortex of adult mouse brain [55]. The major dihydroxycholestenone in Cyp27a1−/− brain is 7α,12α-dihydroxycholest-4-en-3-one, this oxysterol is not detected in wt brain (0.50 ± 0.17 ng/mg cf. < 0.01 ng/mg). Its high level in plasma suggests it may be imported from the circulation into brain.

Trihydroxycholesterols and trihydroxycholestenones
As above, combined values for these two generic structures were recorded, and for simplicity we just refer to them as trihydroxycholestenones. As in plasma from the Cyp27a1−/− mice we see a trihydroxycholestenone in brain, which we annotate with the structure 7α,12α,25-trihydroxycholest-4-en-3-one (0.24 ± 0.02 ng/mg, Fig. 5D). There is no evidence for this molecule in wt brain (Fig. 6, Table S1).

24S,25-Epoxycholesterol
We have reported earlier that the level of 24S,25-epoxycholesterol is reduced in brain of Cyp27a1−/− mice (0.92 ± 0.05 ng/mg, cf. 1.32 ± 0.07 ng/mg, Figs. 5E, 6, Table S1) [56]. This is in contrast to the situation in plasma, but is compatible with reduced cholesterol biosynthesis through the Bloch arm of the cholesterol biosynthesis pathway as observed by Ali et al. who found reduced desmosterol levels in cortex of brain of Cyp27a1−/− animals [43]. We also find reduced levels of desmosterol in whole brain of the Cyp27a1−/− mouse (see section 3.2.7). The current data also supports the recent report of Goyal et al. that shows that 24S,25-epoxycholesterol can be formed from desmosterol in a reaction catalysed by CYP46A1 (Fig. S3) [39].

Enzyme activity of CYP3A11 and CYP3A4
In a previous study Honda et al. showed that hepatic microsomal 23-, 24-, 25-and 26-hydroxylations of 5β-cholestane-3β,7α,12α-triol and 23R-, 24R-, 24S-and 27-hydroxylations of 5β-cholestane-3β,7α,12α,25-tetrol were catalysed by CYP3A enzymes, CYP3A4 in man and predominantly by CYP3A11 in mouse [34]. Here we performed a preliminary study to investigate whether these two enzymes also had activity towards 7α-hydroxycholesterol, the primary product of CYP7A1 catalysed hydroxylation of cholesterol. We found that recombinant human CYP3A4 hydroxylates 7α-hydroxycholesterol predominantly to 7α,25-dihydroxycholesterol and to a minor extent to 7α, (25S) 26  To maintain a single y-axis magnification factors have been applied as indicated. The low levels of di-and tri-hydroxycholesterols and of dihydroxycholestenoic acids in brain made it difficult to distinguish between these compounds and their 3-oxo equivalents using EADSA as their differentiation is based on peak area difference between samples treated with and without cholesterol oxidase. Hence, for these metabolites the combined values for the two structures are given, and for simplicity we just give values for di-and trihydroxycholest-4-en-3-ones and 7α-hydroxy-3-oxocholest-4-enoic acids. Using the EADSA method 24S,25epoxycholesterol isomerises to 24-oxocholesterol, becomes hydrolysed to 24,25-dihydroxycholesterol and undergoes methanolysis to 3β,24-dihydroxycholest-5-ene-25-methoxide. The total 24S,25-epoxycholesterol corresponds to the sum of the individual forms.
(10 μg/500 μL) to 5 μM reduced the product formation by both enzymes. At 5 μM the concentration of 7α,(25S)26-dihydroxycholesterol formed was below the limit of detection for incubations with either enzyme. In 16 h incubations with 50 μM 7α-hydroxycholesterol, turnover of substrate by CYP3A4 (20 nM) and CYP125 (20 nM), a known 26-hydroxylase, were 3% and 7%, respectively. Negative control experiments in the absence enzyme or NADPH confirmed the requirement of enzyme and co-factor for the formation of both dihydroxycholesterols. In future studies we will investigate the activities of CYP3A enzymes towards other substrates of enhanced abundance in Cyp27a1−/− animals. However, this preliminary study confirms mouse CYP3A11 as a (25S)26-hydroxylase to substrates with cholest-5en-3β,7α-diol structure.

Sterols as nuclear receptor ligands
Sterols, including oxysterols and cholestenoic acids, are known ligands to nuclear receptors, including the LXRs, FXR, PXR, (also known as the steroid xenobiotic receptor, SXR) and VDR [18,19,24,[58][59][60][61]. Another nuclear receptor, CAR, or constitutive androstane receptor, as the name implies, exhibits an intrinsically high transcriptional activity and provokes activation of target gene expression in the absence of ligand binding, but can also be activated by cholesterol precursors [62]. Of these nuclear receptors, LXRα and β and PXR are known to be expressed in midbrain [63,64], and in previous studies we have identified 24S,25-epoxycholesterol to be a potent LXR ligand that enhances midbrain dopamine neurogenesis in the developing midbrain and  Fig. 8. Analysis of the PXR activational capacity of sterols and oxysterols of enhanced or changed abundance in Cyp27a1−/− mouse. Luciferase activity in SN4741 neural cells transfected with (A) a PXR-responsive luciferase reporter construct (PXRE) and PXR, and (B) an LXR-responsive luciferase reporter construct (LXRE) and LXRα, and stimulated for 24 h with the compounds indicated (10 μM). Cholest-4-en-3-one and 7α,24-dihydroxycholest-4-en-3-one were increased in brain but not in plasma, 7α,12α-dihydroxycholestan-3one was elevated in plasma but not in brain, 7α, (25S)26-dihydroxycholest-4-en-3-one and 7α,12α,25-tihydroxycholest-4-en-3-one are not commercially available, while 7α-hydroxy-3oxocholest-4-en-(25S)26-oic is only available as an unresolved mixture with the 25R epimer. (25R)26-Hydroxycholesterol and desmosterol were reduced in both brain and plasma of the Cyp27a1−/− mouse, 7α-hydroxy-3-oxocholest-4-en-(25R)26-oic acid was reduced in plasma only. 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid, an intermediate in the pathway from cholesterol to bile acids, to be a ligand to the LXRs which enhances motor neuron survival in the CNS via activation of these receptors [19,63]. Some patients with CTX, characterised by deficiency in CYP27A1 and an inability to biosynthesise cholestenoic acids, present with motor neuron disease [65], however, the Cyp27a1−/− mouse does not show a motor neuron disease phenotype. In an attempt to explain this anomaly, we have investigated if cholesterol metabolites of enhanced abundance in the Cyp27a1−/− mouse are nuclear receptor ligands, and whether they are protective towards motor neurons through activation of nuclear receptors.
Of the compounds of increased abundance in Cyp27a1−/− plasma or brain none activated LXR or PXR in luciferase assays performed in mouse neuronal cells, with the exception of cholest-4-en-3-one and 7αhydroxycholest-4-en-3-one which both activated PXR (Fig. 8A, Table  S2). Goodwin et al. also found that cholest-4-en-3-one and 7α-hydroxycholest-4-en-3-one both activate mouse PXR [24]. CYP3A11 in mouse and CYP3A4 in human are both PXR target genes [24,60] [24,60]. This data coupled with Honda et al.'s finding that CYP3A11 is the predominant enzyme responsible for side-chain hydroxylations of 5β-cholestane-3α,7α,12α-triol and 5β-cholestane-3α,7α,12α,25-tetrol in mouse liver microsomes [34] and our data that CYP3A11 hydroxylates 7α-hydroxycholesterol at C-26, provides additional routes for cholesterol metabolism in the Cyp27a1−/− mouse, one of which may go through (25S) 26-carboxylic acids (Fig. 1, inset ii). (25S)26-CoA thioesters, generated from the corresponding acids, are substrates for the epimerase AMACR, thus the (25S)26-acids can be converted to their 25R-epimers. This is the likely explanation for the presence of both epimers in mouse plasma. Although we did not differentiate the neuroprotective molecule 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid or its 25S-epimer from their 3-oxo metabolites in mouse brain from either the Cyp27a1−/− or Where known enzymes are indicated in bold. CYP3A11 was found in the present work to introduce a (25S)26-hydroxy group to 7α-hydroxycholesterol as indicated by underlining of the enzyme symbol. The enzyme which converts the (25S)26-primary alcohol to a carboxylic acid is indicated as a sterol oxidase (SO). Metabolites of increased or decreased abundance in the Cyp27a1−/− mouse are indicated by upward or downward arrow. Red arrows are used to indicate changes in plasma, blue arrows for brain. A horizontal solid line indicates detected but not changed significantly. *, P < 0.05; ** P < 0.01; *** P < 0.001. P < 0.05 is considered significant. The low levels of di-and trihydroxycholesterols and of dihydroxycholestenoic acids in brain makes it difficult to distinguish between these compounds and their 3-oxo equivalents using EADSA as their differentiation is based on peak area difference between samples treated with and without cholesterol oxidase. Hence, for these metabolites the combined values for the two structures are considered. The metabolites identified in brain are enclosed within the blue box. Abbreviations are as in Fig. 1.
the wt genotype, the current study establishes a route to the formation of the neuroprotective compound, even in the absence of CYP27A1, through up-regulated CYP3A11 as summarised in Fig. 9. It is not known whether CYP3A11, like CYP27A1, can oxidise primary alcohols to carboxylic acids, so an additional sterol oxidase may be required for the second oxidation. The additional pathway to 3β,7α-dihydroxycholest-5en-(25R)26-oic acid illustrated in Fig. 9 could explain the absence of a CTX motor neuron phenotype in the Cyp27a1−/− mouse.

Effect of PXR ligands on oculomotor neurons
In our previous study of the effect of LXR ligands on oculomotor neurons we found that the LXR ligand 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid increased the number of Islet-1 expressing neurons in mouse primary midbrain cultures [19]. Islet-1 is a transcription factor expressed in all postmitotic motor neurons. The increase in numbers of Islet-1+ cells was found to be a consequence of increased neuronal survival. According to the scheme presented in Fig. 9 we would predict that treatment of midbrain primary cultures with a PXR ligand would increase the synthesis of 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid through up-regulated CYP3A11 and thus increase the number of Islet-1+ cells in culture. This is exactly what is observed for the more efficacious activator of PXR, 7α-hydroxycholest-4-en-3-one (Fig. 10A & B). This data substantiates the concept that neuroprotective compounds are formed in the Cyp27a1−/− mouse through up-regulated CYP3A11. In human, not all CTX patients suffer motor neuron signs, and perhaps the existence of an additional route to the neuroprotective acid through CYP3A4, the human equivalent of CYP3A11, may provide protection in these patients. Future experiments will explore the importance of these pathways in samples from CTX patients.

Conclusions
From the analysis of > 50 sterols, oxysterols and sterol-acids we can conclude that a sterol hydroxylase other than CYP27A1, probably CYP3A11, is responsible for the formation of the 25S-epimers of 7α,26dihydroxycholest-4-en-3-one, and perhaps 7α-hydroxy-3-oxcholest-4en-26-oic and 7α,12α-dihydroxycholest-4-en-26-oic acids and their 3βhydroxy-5-ene precursors in the Cyp27a1−/− mouse (Fig. 9). In this mouse the 25S-sterol-acids dominate over their 25R-epimers, while the reverse is true in the wt mice where CYP27A1 is active. In both genotypes the 25S-and 25R-sterol-acids are inter-convertible in a reaction catalysed AMACR after activation of the acids with Co-enzyme A. Thus, in the Cyp27a1−/− mouse an additional (25S)26-hydroxylase pathway may account for some of the primary bile acids formed in this mouse. It has yet to be confirmed whether CYP3A11 can convert primary alcohols to carboxylic acids in a manner similar to CYP27A1. If not, an alternative sterol oxidase must be responsible for this conversion in the Cyp27a1−/− mouse. One candidate is CYP24A1 which has been shown to catalyse similar reactions during vitamin D 3 metabolism [66].
Interestingly, we also find low levels of 7α,24-dihydroxycholest-4en-3-one in both Cyp27a1−/− and wt mouse brain (Figs. 5C, 6). This oxysterol has not previously been detected in brain. Its absence from plasma indicates it is formed in brain rather than being imported from the circulation.
The concurrent increase in plasma and brain of levels of many oxysterols with a 7α-hydroxycholest-4-en-3-one structure is compatible with their passage across the BBB down a concentration gradient. The current data also supports the hypothesis of others [38] that oxysterols imported to brain with a 7α-hydroxy or 7α-hydroxy-4-en-3-one structure are the precursors of cholesta-4,6-dien-3-ones in brain of Cyp27a1−/− mice and also humans with a similar deficiency.
We were not able to distinguish between low levels of 3β,7α-dihydroxycholest-5-en-26-oic acids and their 3-oxo-4-ene equivalents in mouse brain, but we were able to confirm their combined presence in both Cyp27a1−/− and wt animals. The expression in brain of the necessary enzymes to interconvert the 25R-and 25S-epimers of 3β,7αdihydroxycholest-5-en-26-oic acids provides a biosynthetic route to the neuroprotective compound 3β,7α-dihydroxycholest-5-en-(25R)26-oic acid [18,67], providing an explanation for the lack of a motor neuron disease phenotype in the Cyp27a1−/− mouse.

Conflict of interest
The EADSA technology utilised in this study is licenced to Cayman Chemical Company and Avanti Polar Lipids Inc. by Swansea Innovations, a wholly owned subsidiary of Swansea University.
The technology "Kit and method for quantitative detection of steroids", US9851368B2, is patented by Swansea University.

Transparency document
The Transparency document associated with this article can be found, in online version.