Metabolism of 4β-Hydroxycholesterol in Humans*

One of the major oxysterols in the human circulation is 4β-hydroxycholesterol formed from cholesterol by the drug-metabolizing enzyme cytochrome P450 3A4. Deuterium-labeled 4β-hydroxycholesterol was injected into two healthy volunteers, and the apparent half-life was found to be 64 and 60 h, respectively. We have determined earlier the half-lives for 7α-, 27-, and 24-hydroxycholesterol to be ∼0.5, 0.75, and 14 h, respectively. Patients treated with certain antiepileptic drugs have up to 20-fold increased plasma concentrations of 4β-hydroxycholesterol. The apparent half-life of deuterium-labeled 4β-hydroxycholesterol in such a patient was found to be 52 h, suggesting that the high plasma concentration was because of increased synthesis rather than impaired clearance. 4β-Hydroxycholesterol was converted into acidic products at a much slower rate than 7α-hydroxycholesterol in primary human hepatocytes, and 4β-hydroxycholesterol was 7α-hydroxylated at a slower rate than cholesterol by recombinant human CYP7A1. CYP7B1 and CYP39A1 had no activity toward 4β-hydroxycholesterol. These results suggest that the high plasma concentration of 4β-hydroxycholesterol is because of its exceptionally slow elimination, probably in part because of the low rate of 7α-hydroxylation of the steroid. The findings are discussed in relation to a potential role of 4β-hydroxycholesterol as a ligand for the nuclear receptor LXR.

4␤-Hydroxycholesterol is one of the quantitatively most important oxysterols in human circulation (1). We have recently shown that it is formed by the drug-metabolizing enzyme cytochrome P450 3A4 (CYP3A4) 1 (1). Preliminary experiments showed that the formation of this oxysterol by human liver microsomes was relatively slow. The high plasma levels of the oxysterol are therefore surprising, and we hypothesized that this may be a consequence of slow metabolism. Therefore, in this work, we determined the rate of elimination of deuteriumlabeled 4␤-hydroxycholesterol from plasma. Oxysterols are generally degraded to bile acids, and the rate-limiting step in this conversion is the introduction of a hydroxyl group in the 7␣-position of the steroid. Alternative pathways for bile acid biosynthesis start with oxidation of the steroid side chain by CYP27A1 and CYP46. Therefore, we have studied the possibility that these cytochromes are active toward 4␤-hydroxycholesterol. The metabolism of 4␤-hydroxycholesterol was studied in human primary hepatocytes, control, and transfected cells and by incubations with recombinant enzymes. In addition, fecal samples from three untreated subjects and one subject treated with carbamazepine were analyzed for 4␤-hydroxylated bile acids. Based on these experiments, we present evidence that 4␤-hydroxycholesterol has an unusually long halflife in plasma and that this is the result of slow elimination, particularly slow 7␣-hydroxylation that is the rate-limiting step for further conversion into bile acids.
Two oligonucleotide primers for reverse transcription-PCR were synthesized using the Expedite TM nucleic acid synthesis system followed by high pressure liquid chromatography purification (CyberGene AB, Novum Research Park, Sweden): CYP39A1fo/EcoRI, 5Ј-CGGAATTCC-GTGCTTCTGGAAGGTGCTGG-3Ј, and CYP39A1rv/XhoI, 5Ј-CCGCTC-GAGCGGCCTGGTCCTTGTGAGGCC-3Ј (corresponding to nucleotides 18 -36 and 1480 -1464, respectively, in the cDNA). The primers contained 5Ј overhangs with an EcoRI restriction site in the forward primer and an XhoI site for the reverse primer.
The complete cDNA for 24-hydroxycholesterol 7␣-hydroxylase for transfection was produced by reverse transcription-PCR from human total RNA prepared from liver tissue utilizing the QuickPrep Total RNA Extraction Kit (Amersham Biosciences). The cDNA was inserted into a linearized pcDNA4/HisMax vector and propagated into One Shot Top 10FЈ competent cells. From the recovered clones, the cDNA was sequenced to completion to confirm the cDNA sequence and certify that it was correctly inserted so that the reading frame was maintained.
Synthesis of 4␤,7␣-Dihydroxycholesterol-4␤,7␣-dihydroxycholesterol was synthesized as described previously (4). Cholesta-4,6-dien-3␤-ol (11 mg) dissolved in 1.0 ml of acetone was mixed with 0. 8  Infusion of 4␤-Hydroxycholesterol in Human Healthy Volunteers-Deuterium-labeled 4␤-hydroxycholesterol (500 g) was dissolved in ethanol and mixed with human serum albumin (Biovitrum Plasma Products) and physiological sodium chloride solution (0.9% w/v). The mixture was administered intravenously in a first experiment to a healthy male volunteer 60 years of age (body mass index 25) and in a second experiment to a healthy female volunteer 64 years of age (body mass index 24). Blood samples were taken before and at specific time points after the administration.
Infusion of 4␤-Hydroxycholesterol in a Carbamazepine-treated Volunteer-Deuterium labeled 4␤-hydroxycholesterol (500 g) was administered under the same conditions as described above to a female volunteer 49 years of age (body mass index 25) who was treated for epilepsy with carbamazepine (600 ϩ 750 mg/day) during the last 10 years.
Analysis of Blood Samples from Volunteers Receiving Deuteriumlabeled 4␤-Hydroxycholesterol-The deuterium enrichment in blood samples was monitored by gas chromatography mass spectrometry (GC-MS) in selected ion monitoring mode with the same instrumentation and conditions as previously described but without the addition of internal standards (1).
Fecal Samples-Fecal samples were obtained from three healthy male volunteers, ages 29, 50, and 60 (body mass index of 22, 23, and 25 with plasma 4␤-hydroxycholesterol concentrations of 31, 37, and 17 ng/ml). In addition, one sample was obtained from a female volunteer 49 years of age (body mass index 25) treated with carbamazepine for Ͼ10 years. Her plasma 4␤-hydroxycholesterol concentration was 468 ng/ml.
Metabolism of 4␤-Hydroxycholesterol in Primary Human Hepatocytes-Hepatocytes were isolated from human liver tissue obtained from patients undergoing a liver resection because of liver tumors. The cells were isolated according to a two-step perfusion technique with EGTA and collagenase (Type XI, Sigma) solutions as described previously (5). The hepatocytes were cultured in 6-cm dishes, which had been precoated with 200 l of Engelbreth-Holm-Swarm Matrigel containing 3 ml of culture medium (William E medium supplemented with glutamine (292 g/ml)), insulin (2 international milliunits/ml), penicillin G sodium (100 units/ml), streptomycin sulfate (100 g/ml), and gentamycin (85 g/ml). The hepatocytes were cultured for 4 days without any additions. On the fourth day, the hepatocytes were incubated with 4 g of tritium-labeled 4␤-hydroxycholesterol (3 ϫ 10 3 cpm) and in parallel with 4 g of tritium-labeled 7␣-hydroxycholesterol (2 ϫ 10 6 cpm). The substrates were dissolved in 10 l of ethanol. After the addition of substrates, the incubation was continued for 48 h in an atmosphere containing 5% CO 2 .
Analysis of Bile Acids in Hepatocytes-Hepatocytes and medium were analyzed as described previously (6). After incubation, the medium was hydrolyzed at 120°C for 18 h with 6 g of potassium hydroxide dissolved in 12 ml of 50% ethanol. After hydrolysis, a two-step extraction was performed. The first extraction isolated neutral steroids by adding 80 ml of physiological sodium chloride solution, 80 ml of 50% ethanol, and 100 ml of diethyl ether. The ether phase was transferred into a round bottom flask and evaporated to dryness under reduced pressure and dissolved in 3 ml of methanol. The water phase containing bile acids was acidified with 15 ml of 6 M hydrochloric acid and extracted with 100 ml of diethyl ether. The ether phase was evaporated under reduced pressure, and the residue was dissolved in 3 ml of methanol. An aliquot from each fraction was mixed with scintillation mixture (Lumasafe, Lumac, Groningen, The Netherlands) and counted in a liquid scintillation counter (Model 1414 Winspectral, Wallac Oy, Turku, Finland).
Metabolism of 4␤-Hydroxycholesterol in 293 Cells Transfected with CYP46 -At a confluence of 60%, 293 cells were transfected with a mixture of 60 l of Tfx TM 20 (Promega), 10 g of cDNA of CYP46, and 2.93 ml of medium prewarmed to 37°C (6). Thereafter, an additional 12 ml of complete medium was added. After 48 h of culturing, the cells were selected with 0.4 mg/ml Geneticin (Sigma) and incubated for 48 h with 20 g of 4␤-hydroxycholesterol or 20 g of 27-hydroxycholesterol dissolved in 20 l of ethanol.
Metabolism of 4␤-Hydroxycholesterol in 293 Cells Transfected with CYP39A1-Human embryonic kidney 293 cells were cultured at 37°C in an atmosphere of 5% CO 2. in Dulbecco's modified Eagle's medium. The medium was supplemented with L-glutamine, D-glucose (1000 mg/ liter), sodium pyruvate (catalog number 31885-023, Invitrogen), heatinactivated fetal calf serum (5% v/v, Invitrogen), and penicillin/streptomycin (100 IU/ml, 100 g/ml, respectively, Invitrogen). The cells were plated out on polylysine-coated dishes (5 g/cm 2 ), and when the cells had reached a confluence of 80 -90%, they were transfected with 10 g of purified pcDNA4/HisMax vector containing the 24-hydroxycholesterol 7␣-hydroxylase cDNA by using the Tfx20 Reagent (Promega, Madison, WI). Control cells were transfected with the same amount of vector without insert. The cells were then cultured for another 48 h in 18 ml of Dulbecco's modified Eagle's medium with L-glutamine supplemented with 2% delipidated fetal calf serum (Sigma) and 20 g of either 24hydroxycholesterol or 4␤-hydroxycholesterol (dissolved in 20 l of ethanol).
Analysis of Steroids in 293 Cells-Cells and medium were sonicated three times for 20 s. Thereafter, the cells and medium were extracted by adding 2 ml of 20% NaCl and 85 ml of chloroform/methanol (2:1). The organic phase was evaporated and dissolved in 1 ml of toluene and then purified on a silica column (International Sorbent Technology, Mid Glamorgan, United Kingdom). The column was activated by the addition of 2 ml of hexane prior to the addition of sample (in 1 ml of toluene). Thereafter, the column was washed with 8 ml of 0.5% isopropyl alcohol in hexane, and the oxysterols were eluted with 5 ml of 50% isopropyl alcohol. The eluate was evaporated and derivatized to trimethylsilyl ether derivatives by treatment with pyridine/hexamethyldisilazane/ trimethylchlorosilane (3:2:1) (v/v/v) at 60°C for 30 min and analyzed by GC-MS in full scan mode.
Microsomes containing human CYP3A4, CYP3A5, and CYP3A7 (50 l, 50 pmol) were pre-incubated for 5 min with 100 nmol of cholesterol dissolved in 10 l of 2-hydroxypropyl-␤-cyclodextrin and 390 l of buffer (0.1 M phosphate buffer, pH 7.4). Finally, 0.5 mg of NADPH dissolved in 50 l of buffer was added. After incubation at 37°C for 120 min, the reaction was stopped by adding 50 l of methanol (1).
Analysis of Incubations-CYP7A1 and CYP27A1 metabolites were extracted by adding 5 ml of chloroform/methanol (2:1) and 1 ml of physiological sodium chloride solution to the test tube. The test tube was whirl-mixed and centrifuged. The organic phase was transferred to a new test tube evaporated under N 2 , and the residue was trimethylsilyl-derivatized. CYP27A1 metabolites were converted into methyl esters with trimethylsilyldiazomethane prior to sialylation. 4␤-Hydroxycholesterol was determined in incubations with CYP3A4, CYP3A5, and CYP3A7 as described previously (1). All of the incubations were analyzed by full scan GC-MS with the exception of incubations with recombinant CYP3A4, CYP3A5, and CYP3A7, which were analyzed by GC-MS in selected ion monitoring mode.
Analysis of Fecal Bile Acids-In a round bottom flask, 0.1 g of feces suspended in 1 ml of H 2 O was evaporated to dryness under vacuum in a rotary evaporator. Thereafter, the samples were subjected to solvolysis and subsequent hydrolysis. Solvolysis was performed at 40°C for 1 h by adding 4 ml of 2 M HCl in ethanol/acetone (9:1). After solvolysis, the pH was adjusted to 7 with 1 M NaOH, and the mixture was dried in a rotary evaporator. The residue was dissolved in 8 ml of 4 M NaOH/ methanol (1:1), and the hydrolysis was performed for 16 h at 80°C (9).
The following extraction was performed as described earlier under "Experimental Procedures," with the exception that the residue was methylated and trimethyl-silylated. Bile acids in feces were analyzed by gas chromatography mass spectrometry. The gas chromatograph was a Hewlett-Packard 5890 Series II Plus equipped with an HP-5MS capillary column (30 M ϫ 0.25 mm, 0.25-m phase thickness) connected to an HP 5972 mass selective detector with a HP 7673A automatic sample injector. The oven temperature was as follows: 180°C for 1 min, 20°C/ min to 220°C, and then 3.5°C/min to 290°C where the temperature was kept for 27 min. Helium was used as a carrier gas with a flow rate of 0.8 ml/min. Splitless injection was used (1 l), and the detector temperature was 280°C. The detector transfer line temperature was set to 270°C. The mass spectrometer was used in the full scan mode (m/z 100 Ϫ 700), and the electron ionization energy was 70 eV.
Ethical Aspects-These studies were approved by the Ethics Committee of Karolinska Institutet at Huddinge University Hospital (Huddinge, Sweden).

Elimination of 4␤-Hydroxycholesterol from Human Circulation-
To study the elimination of 4␤-hydroxycholesterol in human circulation, hexadeuterated 4␤-hydroxycholesterol was administered intravenously to three volunteers. One of them had been treated with the antiepileptic drug carbamazepine for more than 10 years. This drug has been shown to increase the concentration of 4␤-hydroxycholesterol in the circulation (1). Before infusion, blood samples were taken to determine basal levels of 4␤-hydroxycholesterol (Table I). Volunteers without carbamazepine treatment had levels in the normal range (1).
As shown in Fig. 1, the apparent half-life for the labeled 4␤-hydroxycholesterol in plasma was determined to be 64 and 60 h in the two healthy volunteers, respectively. In this calculation, it is assumed that there is no change in the pool of unlabeled 4␤-hydroxycholesterol as a consequence of the administration of the labeled compound. The half-lives for 4␤hydroxycholesterol were calculated from the linear (elimination) part of the curves, assuming first order kinetics. In an earlier experiment in the male volunteer, the half-life of 24hydroxycholesterol in plasma was determined to be ϳ10 h (10), which in turn is considerably longer than the half-life for 7␣and 27-hydroxycholesterol (Table II) (11). The apparent halflife of 4␤-hydroxycholesterol in the carbamazepine-treated volunteer was determined to be 52 h.
Metabolism of 4␤-Hydroxycholesterol in Human Primary Hepatocytes-4␤-Hydroxycholesterol and 7␣-hydroxycholesterol were incubated with primary human hepatocytes to study the metabolism into bile acids. The metabolism of 4␤-hydroxycholesterol in primary human hepatocytes was found to be extremely slow. Only 4% of the added 4␤-hydroxycholesterol was converted to polar products under the experimental conditions used. These polar products were recovered in the bile acid containing phase after extraction. Parallel incubations with 7␣-hydroxycholesterol resulted in 59% conversion to bile acids, 70 -90% of which corresponded to cholic and chenodeoxycholic acid (6).
Recombinant CYP27A1 was found to convert 4␤-hydroxycholesterol into 4␤,27-dihydroxycholesterol and metabolize it further into the corresponding acid. The products 4␤,27-dihydroxycholesterol and the corresponding acid were converted into trimethylsilyl ether and trimethylsilyl ether-methyl ester, . This mass spectrum is consistent with the structure of the trimethylsilyl ether of methyl-3␤,4␤-dihydroxy-5-cholestenoic acid. The conversion of 4␤-hydroxycholesterol into 27-oxygenated products was 50%. When cholesterol was incubated under the same conditions, the degree of 27-oxygenation was 27%. The corresponding turnover numbers were 0.41 and 0.22 nmol/min ϫ nmol cytochrome P450, respectively.
Analysis of Fecal Bile Acids-Feces from the female volunteer treated with carbamazepine was analyzed for bile acids by GC-MS.  (2 ϫ 90)-129), 253 ,147, and 129 (Fig. 3). The mass spectrum is similar to a previously published spectrum of 3,4␤,12␣-trihydroxy-5␤-chol- anoic acid (as methyl ester, trimethylsilyl ether) (13). With respect to the ion at m/z 369, the loss of 89 mass units is characteristic for two vicinal trimethylsiloxy groups (13). The ion at m/z 417 (probably formed by a loss of one trimethylsiloxy group and one 131-fragment from the molecular ion) is highly characteristic for 4␤-hydroxylated bile acids (13). The feces from three healthy volunteers were analyzed for bile acids, and in one sample, the same 4␤-hydroxylated bile acid was found (constituting 0.37% total bile acids), whereas the concentration in the remaining two samples from the healthy volunteers was below the detection limit.

DISCUSSION
In the present investigation, deuterium-labeled 4␤-hydroxycholesterol was injected into volunteers to determine the elimination rate from the circulation. An exceptionally long half-life of ϳ2-3 days was found for 4␤-hydroxycholesterol compared with the oxysterols 7␣-, 27-, and 24-hydroxycholesterol for which the half-lives have been determined to be 0.5-12 h (10,11). Previous in vitro experiments indicated a slow rate of formation for 4␤-hydroxycholesterol (1). If the rate of formation is slow also in vivo, the long half-life could be expected to be the result of slow metabolism of the oxysterol. This was further investigated in primary human hepatocytes and by the use of normal and transfected cells as well as recombinant human cytochromes CYP7A1 and CYP27A1, enzymes known to metabolize oxysterols. Primary human hepatocytes efficiently converted 7␣-hydroxycholesterol to bile acids, whereas only a small fraction of 4␤-hydroxycholesterol was metabolized into polar products. This finding was in accordance with in vitro experiments with recombinant CYP7A1 where 4␤-hydroxycholesterol was 7␣-hydroxylated at a rate only half of that of cholesterol. Cultured 293 cells with a high endogenous CYP7B1 activity efficiently 7␣-hydroxylated 27-hydroxycholesterol, whereas 4␤-hydroxycholesterol was not converted at all. These experiments indicate that 4␤-hydroxycholesterol undergoes the rate-limiting 7␣-hydroxylation step much slower than cholesterol. It is possible that the 4␤-hydroxyl group in some way impairs the 7␣-hydroxylation reaction. The enzymatic hydroxylation of the steroid side chain was not impaired, because 4␤-hydroxycholesterol was 27-oxygenated by recombinant CYP27A1 at almost twice the rate compared with cholesterol, and incubations with 293 cells transfected with CYP46 resulted in a rapid conversion into a 24-hydroxylated metabolite. Because CYP46 is expressed exclusively in brain in humans (14), 24-hydroxylation cannot be a quantitatively important pathway for elimination of 4␤-hydroxycholesterol but may be of local importance in the brain. The different metabolic pathways investigated are outlined in Fig. 4.
The size of the pool of 4␤-hydroxycholesterol in equilibrium with the administered deuterium-labeled material can be estimated from the dilution of the administered labeled 4␤-hydroxycholesterol. The extrapolation of the elimination curves for the two healthy volunteers to time zero (Fig. 1)   hydroxycholesterol. As 500 g of the deuterium-labeled material was infused in the volunteers, the pools of unlabeled 4␤hydroxycholesterol were apparently 1.1 and 1.7 mg, respectively. The half-lives for 4␤-hydroxycholesterol in the circulation in the two volunteers were determined to be 64 and 60 h, respectively. This corresponds to an elimination of ϳ0.2 and 0.3 mg/day, respectively. Assuming that 4␤-hydroxycholesterol is quantitatively converted into bile acids, a maximum of ϳ0.3 mg/day would be formed. Because the normal production of bile acids is around 400 mg/day (15), Ͻ0.1% of this pool can be expected to be derived from such a hypothetical pathway. Such a small amount would not be detectable with the methodology normally used. In a patient with maximally up-regulated CYP3A4, the formation of 4␤-hydroxylated bile acids could be expected to be up to 20-fold higher. In view of this, it is interesting that the patient treated with antiepileptics stud- ied here had a content of one specific 4␤-hydroxylated bile acid in feces corresponding to ϳ0.6% total bile acids. When this bile acid was measured in feces from three healthy volunteers with normal plasma 4␤-hydrocholesterol concentrations, it was identified in one of the volunteers, whereas the concentration in feces from the other two volunteers was below the detection limit. The 4␤-hydroxylated bile acid constituted ϳ0.4% total bile acids, excluding 4␤-hydroxylation of cholesterol as a major pathway for formation of the 4␤-hydroxylated bile acid identified. Thus, the major pathway for the formation of 4␤-hydroxy-lated bile acids remains to be determined. At the present state, the possibility must be considered that CYP3A4 may have some 4␤-hydroxylase activity toward another intermediate in bile acid synthesis.
Because the high plasma concentrations of 4␤-hydroxycholesterol in patients treated with some antiepileptic drugs are the result of increased synthesis and not impaired metabolism, 4␤-hydroxycholesterol may reflect the CYP3A4 activity in vivo. Thus, this cholesterol metabolite is a potential clinical marker for CYP3A4 activity.  4. Metabolism of 4␤-hydroxycholesterol. CYP7A1 (cholesterol 7␣-hydroxylase) is catalyzing the rate-limiting step in the major pathway for bile acid synthesis. CYP7B1 (oxysterol 7␣-hydroxylase) is of importance in an alternative pathway to bile acids. CYP27A1 (sterol 27hydroxylase) present both in the liver and in extrahepatic tissues catalyzes the first steps in an alternative pathway to bile acids. CYP46 (cholesterol 24-hydroxylase) present only in brain in humans catalyzes 24-hydroxylation of cholesterol that facilitates elimination of cholesterol from the brain. CYP39A1 is present in the liver and catalyzes 7␣-hydroxylation of 24-hydroxycholesterol.
In conclusion, the elimination rate of 4␤-hydroxycholesterol from the human circulation was found to be exceptionally slow compared with other oxysterols. This slow rate of elimination may at least in part be because of slow 7␣-hydroxylation, which is the rate-limiting step in bile acid biosynthesis. A 4␤-hydroxylated bile acid was identified in feces from healthy volunteers and from a patient with epilepsy with a high plasma concentration of 4␤-hydroxycholesterol. The levels of the fecal 4␤hydroxylated bile acid were similar in the patient and in the healthy volunteer, making it unlikely that 4␤-hydroxylation of cholesterol by CYP3A4 is a major pathway for the formation of fecal 4␤-hydroxylated bile acids.