Biochemistry 1987, 26, 4535-4540
4535
Effects of Cholesterol Analogues and Inhibitors on the Heme Moiety of Cytochrome P-45OsCc:A Resonance Raman Study? Motonari Tsubaki,* Atsuo Hiwatashi, and Yoshiyuki Ichikawa Department of Biochemistry, Kagawa Medical School, Kita-gun, Kagawa 761 -07, Japan Received December 2, 1986; Revised Manuscript Received March 9, 1987
ABSTRACT: Interactions of cholesterol analogues and inhibitors with the heme moiety of cytochrome P-45OS, were examined by resonance Raman spectroscopy. The Raman spectra of ferric cytochrome P-450,,, complexed with inhibitors such as cyanide, phenyl isocyanide, aminoglutethimide, and metyrapone were characteristic of low-spin state and were very similar. However, the effect of exchange of the sixth ligand from the oxygen atom (ferric low-spin state) to the nitrogen atom upon aminoglutethimide and metyrapone binding was seen as down-frequency shifts of the v3 band from 1503 to 1501 and 1502 cm-', respectively, while cyanide and phenyl isocyanide binding caused an up-frequency shift of the v3 band to 1505 cm-'. The effects of cholesterol analogues [22(R)-hydroxycholesterol, 22(S)-hydroxycholesterol, 22-ketocholesterol, 20(S)-hydroxycholesterol, and 25-hydroxycholesterol] on a Fe2+-C0 stretching frequency of cytochrome P-45OS, in ferrous CO form were examined. The 22(R)-hydroxycholesterol complex could not give a clear Fe2+-C0 stretching Raman band due to a strong photodissociability. 22(S)-Hydroxycholesterol and 25hydroxycholesterol complexes gave the Raman bands a t 487 and 483 cm-', respectively, whereas 20(S)hydroxycholesterol and 22-ketocholesterol complexes gave Fe2+-C0 stretching frequencies (478 cm-') almost identical with that without substrate (477 cm-'). These findings suggest the existence of the following physiologically important natures of the cytochrome P-450,, active site: (1) there is a strong steric interaction between heme-bound carbon monoxide and the 22(R)-hydroxyl group or the 22(R)-hydrogen of the steroid side chain and (2) the hydroxylation a t the 2 0 s position may cause a conformational change of the side-chain group relative to the heme.
T e C20-C22 bond of cholesterol side chain is cleaved by a cytochrome P-450,, located in the inner mitochondrial membrane of adrenal cortex. Cholesterol side-chain cleavage involves three consecutive hydroxylation steps. Each step requires two electrons, which are supplied by NADPH' via NADPH-adrenodoxin reductase and adrenodoxin, and one molecule of oxygen (Shikita & Hall, 1974; Jefcoate, 1986). The first hydroxylation occurs at the 22R position to yield 22(R)-hydroxycholesterol and the second at the 2 0 s position to give 20(R), 22(R)-dihydroxycholesterol. The final step results in an oxidative cleavage of the diol to form pregnenolone and isocaproic aldehyde (Burstein et al., 1975; Larroque et al., 1981; Hume et al., 1984). The efficient conversion of cholesterol to pregnenolone is ensured by the facts that the hydroxylated intermediates bind to the enzyme up to 300 times more tightly than cholesterol does (Lambeth et al., 1982) and the stability of the ferrous dioxygen complex increases in each successive step (Tuckey & Kamin, 1982, 1983). These observations suggest the existence of a strict stereochemistry of the oxygen molecule bound to ferrous heme and the side-chain group of cholesterol to advance the successive hydroxylation steps properly. Interestingly, the reaction intermediates 22(R)-hydroxycholesteroland 20(R),22(R)-dihydroxycholesterol, stabilize the ferrous-dioxygen complex (2- and 4-fold, respectively) relative to cholesterol. In contrast, isoelectric CO complexes are destabilized by 150- and 300-fold, respectively, relative to cholesterol, indicating a unique interaction between the 22-hydroxyl group and ferrous heme-bound dioxygen, while +This investigation was supported in part by Grants for Scientific Research from the Ministry of Education, Science, and Culture, Japan, and by Grants-in-Aid from The Naito Foundation and from The Mochida Memorial Foundation for Medical and Pharmaceutical Research. * Author to whom correspondence should be addressed.
0006-2960/87/0426-4535$01.50/0
the same substituent interferes with CO complex formation (Tuckey & Kamin, 1983). Resonance Raman spectroscopy of hemoproteins can provide unique information on the structure of the heme moiety (Tsubaki & Yu, 1981; Tsubaki et al., 1981, 1982; Shimizu et al., 1981; Bangcharoenpaurpong et al., 1986) [for a review, see Spiro (1982)]. In recent Raman studies on ferrous cytochrome P-45OS,-C0 complex, we detected the resonance enhancement of a Fe*+-CO stretching frequency at 477 cm-' (Tsubaki & Ichikawa, 1985). This unusually low Fe*+-CO stretching frequency, compared to those of (carbonmon0xy)hemoglobin and myoglobin (507 and 512 cm-I, respectively) (Tsubaki et al., 1982), was interpreted as a direct indication of a weaker Fe2+-C0 bond strength caused by a cysteinyl thiolate ligand trans to CO and a linear and perpendicular stereochemistry of bound CO to the ferrous heme (Tsubaki & Ichikawa, 1985). It was further demonstrated that the addition of cholesterol to ferrous cytochrome P45O,,,-CO complex caused an increase of the Fe2+-C0 stretching frequency to 483 cm-', which might suggest that cholesterol in substrate-binding sites opposes a linear and perpendicular binding of CO to the ferrous heme (Tsubaki et al., 1986a). In this paper, we have extended our resonance Raman studies to reveal the interactions of cholesterol analogues (hydroxycholesterols and 22-ketocholesterol) with the heme moiety of cytochrome P-450,,,. In addition, we have shown the resonance Raman spectra of cytochrome P-450,,, complexed with various inhibitors including aminoglutethimide, metyrapone, cyanide, and phenyl isocyanide for the first time.
'
Abbreviations: EDTA, ethylenediaminetetraacetic acid: NADPH, reduced nicotinamide adenine dinucleotide phosphate.
0 1987 American Chemical Society
4536
TSUBAKI ET
BIOCHEMISTRY
EXPERIMENTAL PROCEDURES Materials. Emulgen 913 was obtained from Kao-Atlas; CNBr-activated Sepharose 4B was purchased from Pharmacia Fine Chemicals. Aminoglutethimide, metyrapone, 22(R)hydroxycholesterol, 22(S)-hydroxycholesterol, 20(S)hydroxycholesterol, and 22-ketocholesterol were obtained from Sigma and 25-hydroxycholesterol was obtained from Steraroids. Other chemicals, such as glycerol, NaC1, EDTA, and potassium cyanide, were obtained from Wako Pure Chemicals, Inc. Phenyl isocyanide was synthesized according to the method of Schmidt and Stern (1929). Preparation of Cytochrome P-450,,,.Cytochrome P-450, from bovine adrenocortical mitochondria was purified as previously described (Tsubaki et al., 1986a). Endogenous cholesterol and intermediate metabolite, if any, were removed during the sample preparation by continuous use of a nonionic detergent, Emulgen 913, in the buffer system (Tsubaki et al., 1986b). Emulgen 913 in the purified sample was removed by adrenodoxin-Sepharose 4B column chromatography as previously described (Tsubaki et al., 1986a). The resulting sample was practically free from Emulgen 913 as judged by adsorption spectra in the ultraviolet region and was in a pure low-spin form; this fact ensured almost complete absence of endogenous cholesterol in the purified sample (Kido et al., 1979; Tsubaki et al., 1986b). The Emulgen 9 13 depleted sample was dialyzed against 20 mM potassium phosphate buffer (pH 7.4) containing 20% (v/v) glycerol, 100 mM NaCl, and 0.1 mM EDTA. 22(R)-Hydroxycholesterol, 22(S)-hydroxycholesterol, 20(S)hydroxycholesterol, 25-hydroxycholesterol, and 22-ketocholesterol in methanol were added to the dialysate and incubated on ice for more than 24 h to ensure the complete formation of ferric cytochrome P-45OS,,-substrate complex. After inspection by visible absorption spectra of these complexes with a Shimadzu UV-240 spectrophotometer, the sample was concentrated by centrifugation at 3000 rpm using CENTRIFLO membrane cones (Type CF25, Amicon Corp.) to about 200 pM, which was used directly for resonance Raman measurements after it was passed through a cellulose acetate filter (0.45 pm; type DISMIC-25cs, Toyo Roshi). The preparation of inhibitor complexes were done essentially in the same manner as above. In the case of ferrous form, samples were transferred into a Raman cell after the inhibitor complexes were prepared and were reduced by adding solid sodium dithionite. The cells were sealed with a septum to avoid reoxidation. Measurements of Resonance Raman Spectra. Excitation wavelengths used for resonance Raman measurements were 441.6 nm from a He-Cd laser (Kimmon Electric, Model CD 4801R) and 457.9 nm from an Ar ion laser (NEC, Model GLG 3300); the spectra were recorded on a Jasco R-800D Raman spectrophotometer at room temperature (-22 "C). Calibration of the Raman spectrophotometer was carried out with fenchone or indene as the standard. A sample solution in a cylindrical Raman cell was spun at 1000 rpm to minimize local heating, photodecomposition, and photodissociation. Some of the samples were examined by visible absorption spectra after the measurements of resonance Raman spectra to ensure that no degradation had occurred.
RESULTSAND DISCUSSION Resonance Raman Spectra of Inhibitor Complexes. Two types of inhibitors were used in the present study. One group consists of cyanide and phenyl isocyanide; the other consists of metyrapone and aminoglutethimide.
AL.
Table I: Frequencies (cm-I) of Major Raman Bands of Ferric Cytochrome P-450, Inhibitor Comulexes complex" cyanide (50 mM) phenyl isocyanide (500 PM) metyrapone (500 wM) aminoglutethimide (1 mM) without substrate
y4
y3
VI0
ref
1373 1505 1640 this studv 1373 1505 1639 this study 1373 1502 1639 this study 1372 1501 1640 this study 1372 1503 1638 Tsubaki et ai. (1986al
'Numbers in parentheses are concentrations of inhibitors used in the present study.
Cyanide and phenyl isocyanide are known to bind to both ferric and ferrous forms of cytochrome P-450. The cyanide ion binds, however, more readily to the neutral (ferric) than the negative (ferrous) enzyme. Upon formation of ferric cytochrome P-45OS,-cyanide complex, the Soret band peak showed a shift from 417 [ t = 131.9 mM-' cm-' for ferric low-spin form (Tsubaki et al., 1986b)l to 430 nm (e = 106.2 mM-' cm-l). The Soret peak wavelength of the cyanide complex was different from that reported for cytochrome P-450,,, [439 nm, 6 = 78.5 mM-' cm-' (Sono & Dawson, 1982)l. The cause of the difference is not clear at present. Phenyl isocyanide, on the other hand, has a slightly higher affinity for the ferrous form over the ferric form of cytochrome P-450 (Ichikawa & Yamano, 1968). Addition of phenyl isocyanide to ferric cytochrome P-450,, did not cause a significant effect on the absorption spectrum, except for a slight decrease of the Soret band (peak at 417 nm) intensity (by 2.4%) compared to that of substrate-free cytochrome P-450, in ferric low-spin form. The resonance Raman spectra of cytochrome P-450,,, complexed with these two inhibitors in ferric state were measured, and the frequencies of major Raman bands in the higher frequency region are summarized in Table I. The spectra of these complexes showed Raman bands characteristic of ferric low-spin state; v32 = 1505 cm-' and vl0 = 1640 cm-' for the cyanide-bound form; v3 = 1505 cm-l and vl0 = 1639 cm-' for phenyl isocyanide. The resonance Raman spectra of inhibitor complexes with cyanide and phenyl isocyanide in ferrous state are very interesting in comparison with that of the carbon monoxide complex, since the carbon atom of the C=N moiety of these two inhibitors binds to ferrous heme iron through a a-bond as well as a back-donation of electrons from the occupied iron d-orbitals to the empty r*-orbitals of the C=N moieties of the inhibitors, the same mechanism as observed for carbon monoxide binding to ferrous heme iron (Hanson et al., 1976). Indeed, the visible absorption spectrum of the phenyl isocyanide complex was characterized by a split Soret band with peaks located at 431 ( E = 96.8 mM-' cm-') and 454 nm (e = 106.1 mM-' cm-'); the peak at longer wavelength was very close to that of the Soret peak of the ferrous-CO form [448 nm; e = 129.1 mM-' cm-' (Tsubaki et al., 1986b)l. In contrast, upon reduction of the cyanide-bound form of cytochrome P-450,,, the Soret band peak showed a shift to 428 nm ( E = 141.7 mM-' cm-'). After prolonged standing (up to 3 h) at room temperature, Soret band intensity decreased and a shoulder appeared around 465 nm. Therefore, resonance Raman spectra were measured immediately after the addition of sodium dithionite. The resonance Raman spectra of complexes with cyanide (a) and phenyl isocyanide (b) in ferrous state are presented The designations of porphyrin ring modes are based on Kitagawa et al. (1978) and Abe et al. (1978).
r')
R E S O N A N C E R A M A N O F CYTOCHROME P-45OsCc
a
1
1559
I
I
1500 Frequency (cm-1)
1700
1300
FIGURE 1: Resonance Raman spectra (1 300-1 700 cm-I) of ferrous cytochrome P-450, in the presence of cyanide ion (50 mM) (a) and in the presence of phenyl isocyanide (500 wM) (b). Conditions: protein concentration, 200 wM; excitation wavelength, 441.6 nm; laser power at sample point, 20 mW; slit height, 6 mm; slit width, 250 wm.
I
R
I
300
I
100
Frequency (cm-1) FIGURE 2: Resonance Raman spectra (100-500 cm-') of ferrous cytochrome P-450, in the presence of cyanide (a) and in the presence of phenyl isocyanide (b). Other conditions are the same as in Figure
1.
in Figure 1. The overall features of the spectra of these two complexes showed a close similarity to that of the ferrous-CO form (Tsubaki et al., 1986a). These observations can be rationalized easily, since these two inhibitors have the same binding mechanism to the ferrous heme iron as mentioned above. In Figure 2, we present the lower frequency region resonance Raman spectra of these two complexes. The spectrum of the cyanide complex (a) showed a close similarity to that of the ferrous-CO form except for the absence of a strong 477-cm-' band, which was identified as a Fe*+-CO stretching frequency previously (Tsubaki & Ichikawa, 1985). In contrast, the spectrum of the phenyl isocyanide complex differed significantly from that of the ferrous-CO form. Two relatively strong bands appeared at 295 and 391 cm-'; the latter might be overlapped with the 383-cm-' band observed for the cyanide complex in ferrous state. Concomitantly, the feature around 422 cm-', which could be seen in the spectra of carbon monoxide and cyanide complexes, disappeared and a new band appeared at 444 cm-'. However, the resonance Raman spectrum of the phenyl isocyanide complex of sperm whale myoglobin in ferrous state did not show any feature corresponding to these three Raman bands (Uchida et al., 1984). These three new bands observed in phenyl isocyanide complex may be related to Fe2+-ligand stretching or ligand internal vibrations. Active investigation on these complexes is in progress in our laboratory.
VOL. 26, NO. 14, 1987
4537
Aminoglutethimide and metyrapone are nitrogen-containing aliphatic and aromatic compounds. The activity of these agents as inhibitors is governed by both their hydrophobic character and the strength of the bond between their nitrogen lone pair and the prosthetic heme iron (Ortiz de Montellano & Reich, 1986). Aminoglutethimide is known as a strong inhibitor of the side-chain cleavage reaction of cholesterol catalyzed by cytochrome P-45OsCc(Ki= 50 pM) (Dexter et al., 1967; Wilson & Harding, 1973). Metyrapone is knwon as a specific inhibitor (Ki= 0.1-0.2 pM) of both 1 I@- and 18-hydroxylation reactions of steroids catalyzed by cytochrome P-45011, (Sato et al., 1978), but this inhibitor also binds to the heme of cytochrome P-45OsCc.Complex formations with these inhibitors were characterized by a red shift of the Soret peak, from 417 to 421 nm (e = 127.3 mM-' cm-') for metyrapone and to 422 nm (e = 113.5 mM-' cm-') for aminoglutethimide, indicating an exchange of the sixth ligand of oxygenous (probably of the water molecule) with nitrogenous ligand. Addition of an excess amount of metyrapone (up to 500 pM) did not cause any formation of the P-420 form as judged by its spectrum after reduction by sodium dithionite followed by ligation of C O (Sato et al., 1978). Frequencies of major Raman bands of ferric cytochrome P-45OsCccomplexed with metyrapone and aminoglutethimide are also included in Table I. The effects of the exchange of the sixth ligand from oxygen (ferric low-spin state without substrate) to nitrogen upon binding of aminoglutethimide and metyrapone were seen in the resonance Raman spectra as a downshift of the u j band from 1503 to 1501 and 1502 cm-I, respectively. It is interesting to note that both cyanide and phenyl isocyanide complexes (the carbon atom is expected to bind to the heme iron) showed a v 3 band at 1505 cm-'. This v3 band may be sensitive to the exchange of the sixth ligand of ferric cytochrome P-45OsCc. Except for this v 3 band, the Raman spectra of all four inhibitor complexes were very similar (spectra not shown). Lower frequency region resonance Raman spectra of these four inhibitor complexes in ferric state were measured also. There was no indication of influences caused by the replacement of the sixth ligand (spectra not shown). Resonance Raman Spectra of Cytochrome P-450,,,Complexed with Hydroxycholesterols and Analogues. Spin states of ferric cytochrome P-450, can be determined from optical spectra by measuring the Soret region or from the chargetransfer band at 645 nm that is characteristic of ferric highspin state. Cholesterol, hydroxycholesterols, and their analogues can change the spin-state equilibrium greatly, both by perturbing the conformation of the active site of enzyme and, if any, by directly providing the sixth oxygen ligand to the heme iron (Jefcoate, 1977). In the absence of substrate, cytochrome P-45OsCcis completely in low-spin state, and cholesterol combines with cytochrome P-45OsCcto produce a predominantly high-spin state (80%) (Orme-Johnson et al., 1979; Hume et al., 1984). This spin conversion was confirmed by resonance Raman spectroscopy (Tsubaki et al., 1986a). In the present study, the complexation of steroids to cytochrome P-45OS, was ensured spectrophotometrically for 25-hydroxycholesterol and 22(R)-hydroxycholesterol; the former caused an almost complete conversion of spin state to high spin from low spin, whereas the latter caused a slight decrease in the population of the low-spin component and a concomitant increase in that of the high-spin state. Other steroids were low-spin inducers and, therefore, it was impossible to ensure the complete complexation by conventional spectroscopic examination. However, the reported K , values (or K, values)
4538
BIOCHEMISTRY
Table 11: Frequencies (cm-I) of Major Raman Bands of Ferric Cvtochrome P-450.,, Steroid Com~lexes complex v4 y3 y10 ref 22(R)-hydroxychole1368 1503 1640 this study sterol 22(S)-hydroxychole1372 1504 1638 this study sterol 25-hydroxycholesteroI 1372 ND" ND" this study 2O(S)-hydroxychole1370 1503 1638 thisstudy sterol 1368 1500 1637 Shimizu et al. (1981) 22-ketocholesterol 1371 1503 1640 thisstudy cholesterolb 1370 1485 1617 Shimizu et al. (1981) cholesterol + 1370 1485 1620 Tsubaki et al. (1986a) adrenodoxin without substrate 1372 1503 1638 Tsubaki et al. (1986a) " N D means not detected. bRaman spectrum was measured for high-spin ferric cytochrome P-450,, as purified and, therefore, endogenous cholesterol might be bound to the enzyme.
for steroid binding to cytochrome P-450,,, assured that the samples used in this study were complexed completely with steroids (Orme-Johnson et al., 1979; Lambeth et al., 1982; Lambeth, 1983; Hey1 et al., 1986). Frequencies of major Raman bands for these steroid complexes are summarized in Table I1 in comparison with previously reported data. As expected, the resonance Raman spectrum of the 25-hydroxycholesterol complex was quite featureless due to weak Raman intensities since the Soret band peak located at 392 nm is characteristic of high-spin state. The spectrum of the 22(R)-hydroxycholesterol complex was found to have low-spin character but showed a differences in the v4 band, which showed a slightly lower frequency (1368 cm-I) than those of other complexes (1 370-1 372 cm-') (Table 11). The resonance Raman spectra of other steroid complexes were very similar and were very close to that without substrate (Tsubaki et al., 1986a) (Table 11). Effect of Steroids on Ferrous-CO Form. Upon reduction of these steroid complexes with sodium dithionite in the presence of carbon monoxide, all the complexes could be converted to the ferrous-CO form. I n Figure 3, we present resonance Raman spectra of the ferrous-CO form of cytochrome P-450, complexed with various steroids in the higher frequency region (1 300-1700 cm-'). As reported previously, cholesterol binding to the substrate-binding site caused an increase in the photodissociability of heme-bound carbon monoxide (Mitani et al., 1985; Tsubaki et al., 1986a). Other cholesterol derivatives used in this study showed an increase in the photodissociability of bound carbon monoxide to various extents relative to the substrate-free form. The Raman spectrum of the 22(R)-hydroxycholesterol complex was very close to that of the ferrous form (Tsubaki et al., 1986a), indicating a substantial photodissociation of bound carbon monoxide, although a precise quantitation of the photodissociability was not carried out in the present study. On the other hand, the spectra of the complexes with 20(S)-hydroxycholesterol (d) and 22-ketocholesterol (e) were very similar to that of the ferrous-CO form of cytochrome P-450, without substrate (Tsubaki et al., 1986a); the latter indicated a very slight photodissociation of bound carbon monoxide. The effect of steroid binding to the substrate-binding site of cytochrome P-450,, upon the Fe2+-C0 stretching Raman band was examined. The results are shown in Figure 4. Addition of 22-ketocholesterol caused practically no change in the lower frequence region spectrum of the ferrous cytochrome P-45OS,,-C0 complex compared to that without substrate. The Fe2+-C0 stretching frequency was observed at 478 cm-' for this complex, very close to that without substrate
TSUBAKI ET AL.
I
a
I
I
Cyt. P-45O,,(FeZ'.
CO)
3.
I 1
1600
I
I
1500 1400 Frequency (cm-l)
'\
131
FIGURE 3: Resonance Raman spectra (1300-1700 cm-') of ferrous cytochrome P-45OS,,-C0 complex in the presence of 22(R)hydroxycholesterol (a), 22(S)-hydroxycholesterol (b), 25-hydroxycholesterol (c), 20(S)-hydroxycholesterol (d), and 22-ketocholesterol (e). Other conditions are the same as in Figure 1.
at 477 cm-I. Addition of 20(S)-hydroxycholesterol caused a slight increase in the photodissociability as evidenced by a slight decrease in intensity of the Fe2+-C0 stretching Raman band, but its band peak was still located at 478 cm-'. 25Hydroxycholesterol, a high-spin inducer that has a hydroxyl group far from the 22-position and is otherwise very similar to cholesterol, caused the photodissociation of bound carbon monoxide and an upward shift of the Fe2+-C0 stretching frequency to 483 cm-I, identical with the case of the cholesterol complex (483 cm-I). The effect of 22(S)-hydroxycholesterol was more significant. The Fe2+-C0 stretching frequency was shifted up to 487 cm-I. The Fe2+-C0 stretching frequency of the ferrous-CO form of cytochrome P-450,,, complexed with 22(R)-hydroxycholesterol could not be identified clearly due to a strong photodissociation (Figure 4). Proposed Model for the Active Site of Cytochrome P-450,,,. Proximity of the 22-position to the heme iron of cytochrome P-450,,, has been suggested by studies using cholesterol analogues containing amino substituents in the side-chain group (Sheets & Vickery, 1982, 1983). Close proximity of the 22position to the heme iron was ensured by electron spin echo envelope modulation (ESEEM) study, which revealed a separation of 4 f 1 A between the deuteron at the 2 2 s position of 22(R)-hydroxycholesterol and the unpaired spin of ferric
R E S O N A N C E R A M A N O F C Y T O C H R O M E P-450,,, I
' a lk
I
I
Cyt. P-45OscC(Fe2*CO)
.-c> v)
C
c
5 E
d
.i,p+*
0
I 500
I I 400 300 Frequency (cm-1)
,, 2
FIGURE 4: Resonance Raman spectra (100-500 cm-') of ferrous cytochrome P-45O,,,-CO complex in the presence of 22(R)hydroxycholesterol (a), 22(S)-hydroxycholesterol (b), 25-hydroxycholesterol (c), 20(S)-hydroxycholesterol (d), and 22-ketocholesterol (e). Other conditions are the same as in Figure 1.
iron (Groh et al., 1983). Interestingly, by the same technique, the distance between the deuteron at the 22R position of 22(S)-hydroxycholesterol was estimated to be more than 6 A from the heme iron (Groh et al., 1983). Furthermore, two diastereomeric aminocholesterols, (22R)-22-aminocholesterol and (22S)-22-aminocholesterol, were both found to be potent inhibitors of the biosynthesis of pregnenolone from cholesterol by cytochrome P-450,,. Both steroids were competitive with cholesterol, but the stereochemically correct analogue (22R)-22-aminocholesterol was bound approximately 1000 times more tightly than (22S)-22-aminocholesterol (Nagahisa et al., 1985). All these observations suggest a strict stereochemical requirement for the interaction between the heme and the side-chain group, which ensures consecutive side-chain hydroxylation and side-chain cleavage. Heyl et al. (1986) demonstrated that the specific binding of 22(R)-hydroxy steroids [22(R)-hydroxycholesterol and 22(R),20(R)-dihydroxycholesterol] to the substrate-binding site of cytochrome P-450, was disrupted by ferrous heme-bound carbon monoxide (but not by dioxygen). They suggested that the 22(R)-hydroxyl group of 22(R)-hydroxy steroids is located on a line perpendicular to the heme plane, between 2 and 3 A from the heme iron. Their proposal is consistent with our present results. The interference of carbon monoxide binding to the ferrous heme iron with 22(R)-hydroxycholesterol was evidenced by the strong photodissociation. The steric hindrance of carbon monoxide binding to the heme iron seemed to be most significant with the 22(R)-hydroxyl group of 22(R)-hydroxycholesterol, but not so much with the smaller 22(R)-hydrogen
V O L . 26, N O . 14, 1987
4539
of cholesterol or 25-hydroxycholesterol. In this context, it is very interesting to note that 22-ketocholesterol affected neither the photodissociability of carbon monoxide nor the Fe2+-C0 stretching frequency. This steroid exerts an inhibitory effect by acting as a nonmetabolizable analogue, and its Ki value is severalfold lower than the cholesterol K,,, (Lambeth, 1983). Since the location of the oxygen atom of the keto group at position 22 corresponds to being between the 22(S)-hydrogen and the 22(R)-hydrogen of the cholesterol side chain, this oxygen atom has no steric hindrance toward the heme-bound carbon monoxide and has no influence on the photodissociability or on the Fe2+-C0 stretching frequency. Another intersting steroid is 20(S)-hydroxycholesterol. This 20(S)-hydroxycholesterol is not a physiologically correct intermediate in side-chain cleavage reaction, although it has a stereochemically correct hydroxyl group at position 20 (Burstein & Gut, 1976; Larroque et al., 1981; Hume et al., 1984). This fact is somewhat puzzling, but the hydroxylation at the 22R position prior to the hydroxylation at the 2 0 s position may be absolutely required in biological systems to initiate the side-chain cleavage reaction properly. This requirement infers the occurrence of a stereochemical change of the side-chain group relative to the heme upon the hydroxylation at the 2 0 s position. Once the hydroxylation occurs at the 2 0 s position, newly formed interactions (steric or hydrogen-bonding) between 20(S)-hydroxyl and surrounding amino acid residues or the heme plane may change the conformation of the side-chain group relative to the heme; this new adoptive conformation is probably unfavorable for 22(R)-hydroxylation. The presnt Raman data are consistent with this hypothesis; the FeZ+-CO stretching frequency of 20(S)-hydroxycholesterol is located at 478 cm-', almost identical with that without substrate (477 cm-') and 5 cm-' lower than that of the complex with cholesterol or 25-hydroxycholesterol. There are discrepant observations about the interaction between 22(S)-hydroxycholesterol and cytochrome P-450,,: Orme-Johnson et al. (1979) observed a dissociation constant of 22(S)-hydroxycholesterol comparable with those of 20(S)-hydroxycholesterol,20(R),22(R)-dihydroxycholestero1, and 22(R)-hydroxycholesterol for ferric cytochrome P-450,,,; on the other hand, Heyl et al. (1986) claimed that 22(S)hydroxycholesterol has a much reduced affinity for ferric cytochrome P-450,,, compared to that of 20(R)-hydroxycholesterol, 22(R)-hydroxycholesterol, and 20(R),22(R)-dihydroxycholesterol and even lower than that of cholesterol and 25-hydroxycholesterol. On the basis of this very weak interaction, Heyl et al. (1986) proposed that the protein structure around the 22-position of the side chain cannot accommodate a large 22(S)-hydroxyl group, but the smaller 22(S)-hydrogen of cholesterol, leading to a steric exclusion of 22(S)hydroxycholesterol from the active site. Our Raman data, however, indicated the existence of a strong steric interaction between heme-bound carbon monoxide and the side-chain group of 22(S)-hydroxycholesterol, presumably with 22(R)-hydrogen. In addition, results obtained by EPR measurements of both ferric cytochrome P-450,,, and ferrous cytochrome P-45OS,,-NO complex in the presence of 22(S)hydroxycholesterol indicated the existence of strong interactions between the heme iron and the side-chain group (Tsubaki et al., preceding paper in this issue). In conclusion, present resonance Raman data suggest the existence of a strong steric interaction between heme-bound carbon monoxide and the 22(R)-hydroxyl or 22(R)-hydrogen of the cholesterol side chain. It is further inferred that hydroxylation at the 2 0 s position in the first step of the side-chain
4540
BIOCHEMISTRY
cleavage reaction results in a shift or rotation of the side-chain group relative to the heme caused by steric or hydrogenbonding interactions between surrounding amino acid residues or the heme plane and the newly formed hydroxyl group. Under this new conformation, hydroxylation at the 22R position may be not so favorable. This conclusion is consistent with the results obtained by EPR measurements of ferrous cytochrome P-45OS,,-NO complex in the presence of various cholesterol analogues (Tsubaki et al., preceding paper in this issue). ACKNOWLEDGMENTS We thank Dr. Hiroshi Hori of Osaka University, Toyonaka, Japan, for his helpful suggestions and discussions. Registry No. Cytochrome P-450, 9035-51-2; steroid 20-22-desmolase, 37292-81-2; heme, 14875-96-8.
REFERENCES Abe, M., Kitagawa, T., & Kyogoku, Y. (1978) J. Chem. Phys. 69, 4526-4534. Bangcharoenpaurpong, O., Rizos, A. K., Champion, P. M., Jollie, D., & Sligar, S. G. (1986) J . Biol. Chem. 261, 8089-8092. Burstein, S., & Gut, M. (1976) Steroids 28, 115-131. Burstein, S., Middleditch, B. S., & Gut, M. (1975) J. Biol. Chem. 250, 9028-9037. Dexter, R. N., Fishman, L. M., Ney, R. L., & Liddle, G. W. (1967) J . Clin. Endocrinol. Metab. 27, 473-480. Groh, S. E., Nagahisa, A., Tan, S. L., & Orme-Johnson, W. H . (1983) J . Am. Chem. SOC.105, 7445-7446. Hanson, L. K., Eaton, W. A., Sligar, S. G., Gunsalus, I. C., Gouterman, M., & Connel, C. R. (1976) J. Am. Chem. SOC. 98, 2672-2674. Heyl, B. L., Tyrrell, D. J., & Lambeth, J. D. (1986) J. Biol. Chem. 261, 2743-2749. Hume, R., Kelly, R. W., Taylor, P. L., & Boyd, G. S. (1984) Eur. J . Biochem. 140, 583-591. Jefcoate, C. R. (1977) J. Biol. Chem. 252, 8788-8796. Jefcoate, C. R. (1986) in Cytochrome P-450 Structure, Mechanism and Biochemistry (Ortiz de Montellano, P. R., Ed.) pp 387-428, Plenum, New York. Kido, T., Arakawa, M., & Kimura, T. (1979) J . Biol. Chem. 254, 8377-8385. Kitagawa, T., Abe, M., & Ogoshi, H. (1 978) J. Chem. Phys. 69, 4516-4525. Lambeth, J. D. (1983) Mol. Pharmacol. 23, 743-747. Lambeth, J. D., Kitchen, S. E., Farooqui, A. A., Tuckey, R., & Kamin, H. (1982) J . Biol. Chem. 257, 1876-1884. Larroque, C., Rousseau, J., & van Lier, J. E. (1981) Biochemistry 20, 925-929. Mitani, F., Iizuka, T., Shimada, H., Ueno, R., & Ishimura, Y . (1985) J . Biol. Chem. 260, 12042-12048.
TSUBAKI ET AL.
Nagahisa, A., Foo, T., Gut, M., & Orme-Johnson, W. H. (1985) J . Biol. Chem. 260, 846-851. Orme-Johnson, N . R., Light, D. R., White-Stevens, R. W., & Orme-Johnson, W. H. (1979) J . Biol. Chem. 254, 2103-21 11. Ortiz de Montellano, P. R., & Reich, N. 0. (1986) in Cytochrome P-450 Structure, Mechanism and Biochemistry (Ortiz de Montellano, P. R., Ed.) pp 273-314, Plenum, New York. Sato, H., Ashida, N., Suhara, K., Itagaki, E., Takemori, S . , & Katagiri, M. (1978) Arch. Biochem. Biophys. 190, 307-314. Schmidt, P., & Stern, D. (1929) in Beilstein's Handbuch der Organischen Chemie, Band XII, 4th ed., p 191, Springer Verlag, Berlin. Sheets, J. J., & Vickery, L. E. (1982) Proc. Natl. Acad. Sci. U.S.A. 79, 5773-5777. Sheets, J. J., & Vickery, L. E. (1983) J . Biol. Chem. 258, 11446-1 1452. Shikita, M., & Hall, P. F. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 1441-1445. Shimizu, T., Kitagawa, T., Mitani, F., Iizuka, T., & Ishimura, Y. (1981) Biochim. Biophys. Acta 670, 236-242. Sono, M., & Dawson, J. H . (1982) J . Biol. Chem. 257, 5496-5502. Spiro, T. G. (1982) in Iron Porphyrins (Lever, A. B. P., & Gray, H. B., Eds.) Part 11, pp 89-159, Addison-Wesley, Reading, MA. Tsubaki, M., & Yu, N.-T. (1981) Proc. Natl. Acad. Sci. U.S.A. 78, 3581-3585. Tsubaki, M., & Yu, N.-T. (1982) Biochemistry 21, 1140-1 144. Tsubaki, M., & Ichikawa, Y. (1985) Biochim. Biophys. Acta 827, 268-274. Tsubaki, M., Srivastava, R. B., & Yu, N.-T. (1981) Biochemistry 20, 946-952. Tsubaki, M., Srivastava, R. B., & Yu, N.-T. (1982) Biochemistry 21, 1 132-1 140. Tsubaki, M., Hiwatashi, A., & Ichikawa, Y. (1986a) Biochemistry 25, 3563-3569. Tsubaki, M., Tomita, S., Tsuneoka, Y., & Ichikawa, Y. (1986b) Biochim. Biophys. Acta 870, 564-574. Tsubaki, M., Hiwatashi, A., Ichikawa, Y., & Hori, H. (1987) Biochemistry (preceding paper in this issue). Tuckey, R. C., & Kamin, H . (1982) J . Biol. Chem. 257, 9309-93 14. Tuckey, R. C., & Kamin, H. (1983) J . Biol. Chem. 258, 423 2-4237. Uchida, K., Sato, E., Kimura, T., Usami, M., & Harada, I. (1984) Raman Spectrosc. Proc. Int. Conf., 9th, 1984, 754-755. Wilson, L. D., & Harding, B. W. (1973) J. Biol. Chem. 248, 9-14.
