On the substrate speciﬁcity of human CYP27A1: implications for bile acid and cholestanol formation

The mitochondrial sterol 27-hydroxylase (CYP27A1) is required for degradation of the C 27 -sterol side chain in bile acid biosynthesis. CYP27A1 seems, however, to have roles beyond this, as illustrated by patients with a deﬁcient sterol 27-hydroxylase due to mutations of the CYP27A1 gene [cere-brotendinous xanthomatosis (CTX)]. These subjects have symptoms ranging from accumulation of bile alcohols and cholestanol to accelerated atherosclerosis and progressive neurologic impairment. The present work describes a de-tailed investigation on the substrate speciﬁcity of recombinant human CYP27A1. In accordance with some previous work with rat liver mitochondria, the activity in general increased with the polarity of the substrate. An obvious example was the ﬁnding that cholesterol was 27-hydroxylated more efﬁciently than cholesterol oleate but less efﬁciently than cholesterol sulfate. The oxysterols 24 S -hydroxycholesterol and 25-hydroxy-cholesterol were 27-hydroxylated less efﬁciently than cholesterol, possibly due to steric hindrance. Surprisingly, sterols with a 3-oxo- (cid:2) 4 structure were found to be hydroxylated at a much higher rate than the corresponding sterols with a 3 (cid:3) -hydroxy- (cid:2) 24-hydroxycholesterol. The results show that recombinant CYP27A1 expressed in E. coli has the same catalytic properties as CYP27A1 in isolated liver mitochondria. analysis of products formed by

Mitochondrial sterol 27-hydroxylase (CYP27A1; P450c27, mitochondrial vitamin D 3 25-hydroxylase) is a multifunc-tional P450 enzyme that catalyzes hydroxylations in both bile acid biosynthesis and bioactivation of vitamin D 3 (1)(2)(3)(4)(5). The conversion of cholesterol into bile acids and the biliary excretion of cholesterol are the most important mechanisms for the ultimate removal of cholesterol from the body. In the classical or neutral pathway of bile acid biosynthesis, starting with the 7 ␣ -hydroxylation of cholesterol, CYP27A1 initiates degradation of the C 27 -steroid side chain. In the acidic pathway, it catalyzes the initial 27hydroxylation of cholesterol (6). Recent data suggest that CYP27A1 plays several important roles in cholesterol homeostasis and affects atherogenesis. A novel mechanism has been described for the elimination of cholesterol from human lung macrophages and cells in arterial endothelium. This mechanism involves 27-oxygenation of cholesterol by CYP27A1 (7). Patients with the rare inherited lipid storage disease, cerebrotendinous xanthomatosis (CTX), have a sterol 27-hydroxylase deficiency due to point mutations of the CYP27A1 gene. Manifestations of this genetically determined CYP27A1 deficiency range from accelerated atherosclerosis to progressive neurologic impairment. The major symptoms in CTX are due to the generalized accumulation of cholesterol and cholestanol in almost every tissue, including the nervous system (8,9). CYP27A1 is expressed in many tissues, and both hepatic and extrahepatic CYP27A1 are active in the 27-hydroxylation of several C 27 -sterols (1,3). Besides cholesterol and various intermediates in bile acid biosynthesis, CYP27A1 may also catalyze 27-hydroxylation of different oxysterols. These oxysterols have been reported to affect the expression of genes in cholesterol homeostasis, at least under in vitro conditions, and a 27-hydroxylation may alter their biological activity (10). Currently, it is not clear if oxysterols are metabolized by CYP27A1. Several 3 ␤ -hydroxy-⌬ 5 sterols are converted by 3 ␤ -hydroxy-⌬ 5 -C 27 steroid dehydrogenase(s) in liver and other tissues into the corresponding 3-oxo-⌬ 4 sterols. 7 ␣ -Hydroxy-4-cholesten-3-one, an intermediate in the classical pathway for bile acid biosynthesis, is known to be 27-hydroxylated by the mitochondrial sterol 27-hydroxylase (11,12). However, there is no information regarding whether, or to what extent, 4-cholesten-3-one or 3-oxo-⌬ 4 metabolites of oxysterols can be metabolized by CYP27A1.
The objectives of the current study were to examine the substrate specificity of human CYP27A1 with respect to C 27 -steroids and to get more information on the relationship between sterol structure and CYP27A1 activity. Considering that cholesterol, having a 3 ␤ -hydroxy ⌬ 5 structure, is 27-hydroxylated much less efficiently by CYP27A1 than the saturated sterols 5 ␤ -cholestane-3 ␣ ,7 ␣ ,12 ␣ -triol and 5 ␤ -cholestane-3 ␣ ,7 ␣ -diol (3,11,12), it was hypothesized that sterols with a 3-oxo-⌬ 4 structure might be hydroxylated at a higher rate. In this paper, we report that CYP27A1 has a low activity toward 24-hydroxycholesterol and 25-hydroxycholesterol, and that 3-oxo-⌬ 4 sterols are 27-hydroxylated much more efficiently than the corresponding 3 ␤ -hydroxy-⌬ 5 sterols. The finding that 4-cholesten-3-one, a suggested precursor to cholestanol, is 27-hydroxylated is discussed in relation to bile acid and cholestanol formation.

Expression and purification of recombinant human CYP27A1 in Escherichia coli
The methods used for expression of human CYP27A1 in E. coli were essentially the same as those described by Pikuleva et al. (14), with the exception that DH5 ␣ cells were used instead of TOPP3 cells. Recombinant CYP27A1 was partially purified using the initial steps, including sodium cholate solubilization and octyl-Sepharose as described by Pikuleva, Björkhem, and Waterman (14). The adrenodoxin-Sepharose chromatography was omitted, and cytochrome P450 was eluted from the hydroxylapatite column with 500 mM potassium phosphate buffer (pH 7.4) containing 20% glycerol and 0.15% sodium cholate instead of a linear gradient of potassium phosphate (10-400 mM) in 20% glycerol and 0.15% sodium cholate. Fractions with a spectrophotometric index (A 418 /A 280 ) higher than 0.9 were pooled, dialyzed against 50 mM phosphate buffer containing 20% glycerol and 0.1 mM EDTA, and used as source of CYP27A1.
In a separate set of experiments, incubation mixtures with cholesterol and cholesterol sulfate were directly analyzed after acid solvolysis as described previously (16). Ether extracts of the mixtures were washed with water until neutral, converted into trimethylsilyl ether, and analyzed by gas chromatography-mass spectrometry as described below. The recovery of cholesterol in connection with solvolysis of cholesteryl sulfate by this method was shown to be more than 90%. In the calculation of the conversion of cholesterol sulfate into 27-hydroxycholesterol sulfate, it was assumed that the recovery of 27-hydroxycholesterol is the same as that of cholesterol.

Analysis by combined gas chromatography-mass spectrometry
Lipid extracts or isolated chromatographic fractions were converted into trimethylsilyl ether and analyzed by combined gas chromatography-mass spectrometry under the conditions previously described (3,7,16). In the cases of the incubations with cholesterol and cholesterol sulfate, the extent of conversion of the substrate into the corresponding 27-hydroxylated product was calculated from the height of the peaks corresponding to cholesterol and 27-hydroxycholesterol in the total ion current by guest, on July 20, 2018 www.jlr.org Downloaded from chromatogram. The total ion current signal was found to be similar for cholesterol and 27-hydroxycholesterol, and no correction factor was used to correct for different responses.

Other methods
An oxysterol 7 ␣ -hydroxylase (CYP7B1) from pig liver microsomes was purified as described previously (17). Recombinant human CYP7A1 was expressed in E. coli and purified as described (15). Purification of adrenodoxin and adrenodoxin reductase from bovine adrenal mitochondria and preparation of mitochondria from pig liver were carried out as described previously (11,18).

Metabolism of cholesterol and 4-cholesten-3-one by recombinant human CYP27A1
Mixtures from incubations of cholesterol with recombinant human CYP27A1 in the presence of NADPH and the mitochondrial electron transporters, adrenodoxin, and adrenodoxin reductase, were analyzed by HPLC (after conversion of product into 3-oxo-⌬ 4 structure prior to analysis). Cholesterol was converted into one major product with the same retention time as authentic 27-hydroxy-4-cholesten-3-one ( Fig. 1A ). CYP27A1 was incubated with 4-cholesten-3-one to investigate the ability of the enzyme to 27-hydroxylate this sterol. The formation of 27-hydroxy-4-cholesten-3-one was found to be about 10-fold higher than with cholesterol (Fig. 1B). No products were formed when incubations were carried out in the absence of the mitochondrial electron transporting proteins (Fig. 1C).

Metabolism of cholesterol sulfate and cholesterol oleate by recombinant human CYP27A1
Incubations with cholesterol sulfate were analyzed by gas chromatography-mass spectrometry. After incubation of cholesterol sulfate with recombinant human CYP27A1 and acid solvolysis, a product peak occurred in the chromatogram corresponding to 27-hydroxycholesterol. When equivalent amounts of cholesterol and cholesterol sulfate were incubated with the same amount of CYP27A1 under identical conditions, the rate of conversion of cholesterol into 27-hydroxycholesterol was 7%, whereas the corresponding conversion of cholesterol sulfate was 19%. In another set of experiments, the rate of conversion of cholesterol and cholesterol sulfate was 6% and 27%, respectively. Cholesterol sulfate is thus a better substrate for CYP27A1 than cholesterol itself. When cholesterol oleate was incubated under identical conditions using the same prepara-tion of enzyme and the same molar concentration of substrate, there was no significant conversion into the 27hydroxylated product ( Ͻ 0.2% conversion).

Metabolism of 24-hydroxycholesterol and 24-hydroxy-4-cholesten-3-one by recombinant human CYP27A1 and liver mitochondria
Incubation of racemic 24-hydroxycholesterol with CYP27A1 and subsequent conversion to 3-oxo-⌬ 4 -derivatives gave three product peaks on HPLC analysis ( Fig. 2A ). Incubation of racemic 24-hydroxy-4-cholesten-3-one with CYP27A1 under the same conditions resulted in the same three product peaks, but the conversion into the products was about 10-fold higher (Fig. 2B). All three products increased with time in parallel ( Fig. 3 ). No products were formed when the substrate was incubated with CYP27A1 in the absence of the mitochrondrial electron transporters (Fig. 2C).
In separate experiments, 24-hydroxycholesterol or 24hydroxy-4-cholesten-3-one were incubated with pig liver mitochondria. The same pattern of products was obtained as with recombinant human CYP27A1. Also, with liver mitochondria the rate of product formation was five to ten times higher with 24-hydroxy-4-cholesten-3-one than with

Gas chromatography-mass spectrometry analysis of products formed by CYP27A1 with cholestenone and 24-hydroxy-4-cholesten-3-one
The product formed from cholestenone by CYP27A1 was found to have the same retention time on gas chromatography (as trimethylsilyl ether) and the same mass spectrum as authentic 27-hydroxy-4-cholesten-3-one formed by oxidation of 27-hydroxycholesterol with cholesterol oxidase. The three products from incubations with CYP27A1 and racemic 24-hydroxycholesterol and subsequent oxidation or directly from 24-hydroxy-4-cholesten-3-one ( Fig. 2A,  B) were also analyzed by combined gas chromatographymass spectrometry as trimethylsilyl ethers. The materials corresponding to Peaks 2 and 3 had identical mass spectra, with prominent peaks at m/z 470 (M-90) and 380 (M-2 ϫ 90), as expected for the 24R-and 24S-isomers of 24,27-dihydroxy-4-cholesten-3-one. The identity of Peak 1 in Fig. 2A and B could not be established.

Comparison of the rates of hydroxylation of 3␤-hydroxy-⌬ 5 sterols and the corresponding 3-oxo-⌬ 4 sterols by CYP27A1
As shown in the table, the rate of hydroxylation by CYP27A1 was five to ten times higher with the 3-oxo-⌬ 4 sterols than with the corresponding 3␤-hydroxy-⌬ 5 compounds.

DISCUSSION
Previous studies with purified and recombinant CYP27A1 have shown that in general, polar substrates are hydroxylated at a higher rate than less polar substrates. Thus, 5␤cholestane-3␣,7␣,12␣-triol is 27-hydroxylated at the highest rate of all known substrates for CYP27A, and 5␤-cholestane-3␣,7␣-diol is hydroxylated much more efficiently than cholesterol (3,11,14). From results with purified hepatic CYP27A, the relative rates of 27-hydroxylation of 5␤-cholestane-3␣,7␣,12␣-triol, 5␤-cholestane-3␣,7␣-diol, and 7␣- Incubations were performed for 40 min with substrate (15 M), 0.2 nmol of CYP enzyme, two units of microsomal NADPH-cytochrome P450 reductase, and 1 mol of NADPH in 1 ml of 50 mM Tris-acetate (pH 7.4) containing 20% glycerol and 0.1 mM EDTA. No detectable products were formed in negative controls without microsomal NADPH-cytochrome P450 reductase. The data are given as the means from two experiments.  Equivalent concentrations of the different substrates were incubated with CYP27A1, adrenodoxin, and adrenodoxin reductase as described in Materials and Methods. Data are given as the means Ϯ SD from three to five experiments. The differences in hydroxylation rates between sterols with a 3-oxo-⌬ 4 structure and the corresponding sterols with a 3␤-hydroxy-⌬ 5 structure were statistically significant (P Ͻ 0.01). The differences in the reaction rate order were statistically significant (P Ͻ 0.01) in all cases except for the difference between 24-hydroxycholesterol and 25-hydroxycholesterol (P Ͼ 0.1).
a Sum of three products. The present study provides further information on structural features affecting the rate of CYP27A1-catalyzed hydroxylation. It is apparent that sterols with a 3-oxo-⌬ 4 structure are more efficiently 27-hydroxylated by CYP27A1 than the corresponding 3␤-hydroxy-⌬ 5 sterols. Accordingly, cholestenone was 27-hydroxylated at a rate about 10 times higher than cholesterol and 24-and 25-hydroxycholestenone were hydroxylated at a much faster rate than 24-and 25-hydroxycholesterol. The finding that CYP27A1 is able to 27-hydroxylate 24and 25-hydroxycholesterol may have implications for oxysterol-mediated regulation of gene expression. Oxysterols such as 24-hydroxycholesterol and 25-hydroxycholesterol have been suggested to be potential regulators of genes in cholesterol homeostasis (10,21). Metabolism may alter their biological activity and is required for elimination of the oxysterols from the body. The 7␣-hydroxylation of 24-and 25-hydroxycholesterol by CYP7A1, CYP7B1, and CYP39A1 has previously been described (15,19,22). Here we demonstrate that 24S-and 24R-hydroxycholesterol and 25hydroxycholesterol are 27-hydroxylated by human CYP27A1, although at a low rate. Since CYP27A1 is not confined to the liver but is expressed in almost every tissue in the body, this metabolism of oxysterols may be important in extrahepatic cells. It is possible that the 27-hydroxylation, as well as the 7␣-hydroxylation, of oxysterols alters their effects on gene expression. That CYP27A1-mediated hydroxylation is involved in the elimination of oxysterols from the body is supported by a recent report (16). Labeled 24-hydroxycholesterol was found to be converted into cholic and chenodeoxycholic acids and free and conjugated 5-cholestene-3␤,24,27-triol by human primary hepatocytes. The conjugated 27-hydroxylated product was also found in human intestinal content. It should be emphasized, however, that it is possible that conjugated rather than free 24S-hydroxycholesterol may be the normal substrate for CYP27A1. In this connection, it is interesting that cholesterol sulfate was found to be 27-hydroxylated by CYP27A1 at a higher rate than cholesterol itself.
The most intriguing finding in the present work is the efficient 27-hydroxylation of cholestenone by CYP27A1 and its relation to bile acid and cholestanol biosynthesis. Cholestenone is a precursor in the two pathways described for cholestanol formation. It is generally believed that the major pathway in biosynthesis of cholestanol in healthy subjects involves an oxidation of cholesterol into cholestenone by an NAD-dependent 3␤-hydroxy-⌬ 5 -steroid dehydrogenase [reviewed in ref. (8)]. Cholestenone is then converted to cholestanol by the action of a 5␣-reductase and a 3␤-hydroxysteroid dehydrogenase. In patients with CTX, an alternative pathway has been described involving intermediates in bile acid biosynthesis as precursors. This pathway starts with dehydration of 7␣-hydroxy-4-cholesten-3-one by a specific liver enzyme to yield cholesta-4,6-dien-3-one, which is converted into 4-cholesten-3-one. The final steps are the same as in the major pathway. It has been calculated that the alternative pathway is responsible for at most 20% of the cholestanol formed in healthy humans (8).
Cholestanol is known to be degraded to 5␣-bile acids, allocholic, and allochenodeoxycholic acid by liver enzymes similar to those involved in biosynthesis of 5␤-bile acids (8). Interestingly, cholestanol is also a substrate for CYP27A1 and is 27-hydroxylated by this enzyme at a rate similar to that of cholesterol (23). The results of the present study suggest the possible existence of a minor pathway for bile acid biosynthesis involving 27-hydroxylation of cholestenone by CYP27A1. The 27-hydroxycholestenone formed may be 7␣-hydroxylated by CYP7A1, and the resulting 7␣,27-dihydroxy-4-cholesten-3-one may be further converted into bile acids (Fig. 5). It may be mentioned that we have shown that cholestenone is a substrate for human CYP7A1, although 7␣-hydroxylation of this substrate occurs at a rate considerably lower than that of cholesterol and 27-hydroxycholesterol (M. Norlin et al., unpublished observations). Most probably, a pathway including this reaction is of little or no importance from a quantitative point of view.
Formation of cholestanol is dramatically increased in patients with CTX having a CYP27A1 deficiency. Currently, there are three hypotheses for the mechanisms behind the increased synthesis of cholestanol in patients with CTX: 1) the activity of the 3␤-hydroxy-⌬ 5 -steroid dehydrogenase catalyzing the first step in the major pathway may be increased in CTX patients; 2) the increased biosynthesis of cholestanol may be secondary to the increased cholesterol biosynthesis; and 3) the increased biosynthesis of cholestanol may be due to increased utilization of bile acid intermediates as precursors for cholestanol (8). In addition to the increased formation of cholestanol in CTX, the accumulation of cholestanol in this disease may in part be due to a reduced removal of this steroid from macrophages as a consequence of the lack of CYP27A1 in the macrophages (23).
The results of the present study provide an additional explanation for the increased levels of cholestanol in CTX patients. The ability of CYP27A1 to 27-hydroxylate cholestenone indicates that cholestenone may be continuously drained from the cholestanol-forming pathways under normal conditions, probably leading to formation of a bile acid (Fig. 5). According to these results, the increased cholestanol levels in CTX patients may in part be the consequence of the abolished CYP27A1-catalyzed 27-hydroxylation of cholestenone directing cholestenone exclusively into cholestanol formation.
The skillful technical assistance of Kerstin Rönnqvist is gratefully acknowledged. The authors also express gratitude to Manfred Held for performing the GC-MS analysis. This work was supported by grants from the Swedish Research Council Medicine and the Strategic Foundation.