Biochimica et Biophysica Acta, 1040 (1990) 211-216
211
Elsevier BBAPRO 33717
Resonance Raman study on mutant cytochrome P-450 obtained by site-directed mutagenesis T. Egawa 1, y . Imai
2, T. Ogura
1 and T. Kitagawa 1
1 Institute for Molecular Science, Okazaki'National Research Institutes, Myodaiji, Okazaki and 2 Department of Veterinary Science, Unioersity of Osaka Prefecture, Mozuumemachi, Sakai (Japan)
(Received 28 December1989)
Key words: ResonanceRaman; CytochromeP-450; Laurate(to- 1)-hydroxylase;Carbonmonoxideadduct
Resonance Raman spectra were observed for the threonine-301 to serine or valine mutant as well as the wild type of rabbit liver microsomal cytochrome/'-450 [laurate(0:- 1)-hydroxylase] (P-450(o:- 1)), which were prepared through site-directed mutagenesis. The high-spin marker resonance Raman (RR) bands became similarly stronger for all the P-450s examined in the oxidized form upon addition of laurate, and the RR spectra in the higher frequency region of the oxidized, reduced and CO-adduct forms did not distinctly differ among the P-450s examined. Nevertheless, the Fe-CO stretching mode (Pw-co) of the CO adduct exhibited an upshift for the valine mutant, suggesting positional proximity of Thr-301 to bound CO like Thr-252 of P-450~m, in agreement with the expectation from the sequence analysis. The 1,w.co band was shifted to higher frequency upon binding of normal alkyl fatty acids with Ct0 or longer alkyl chain but little affected by binding of shorter fatty acids.
Introduction Cytochrome P-450 (P-450) is a family of heme proteins which have a monoxygenase activity and give the Soret band around 450 nm in their reduced CO adduct [1]. All P-450s are believed to have a cysteinate as the fifth ligand of the heine iron, but other structural characteristics of P-450s are not known. Since the synthetic ferryloxo iron-porphyrins exhibit the monoxygenase activity like P-450 [2], despite the fact that the ferryloxo heme is found for intermediates of peroxidase catalysis, we are curious to know what kinds of residue in the heine proximity play an essential role in the P-450 catalysis. X-ray crystallographic analysis was completed only for P-450c~ isolated from P s e u d o m o n a s p u t i d a [3]. The catalytic site is surrounded by hydrophobic residues. The X-ray study located Thr-252 at the distal heme
Abbreviations: P-450, cytoehrome P-450; P-450(to-1), P-450 [laurate(to-1)-hydroxylase]; P-450~, P-450 from Pseudomonas putida; P-450scc, adrenal cortex mitochondrial P-450 which catalyzes side-chain cleavage of cholesterol; P-4481, a high spin type /-450 purified from PB- or MC-induced rabbit liver mierosomes; PB, phenobarbital; MC, 3-methylcholantlirene; RR, resonance Raman; Hb, hemoglobin. Correspondence: T. Kitagawa, Institute for Molecular Science, Okazaki National Research Institutes, Myodaiji, Okazaki, 444 Japan.
surface, which might serve as a key residue in the P-450 catalysis. The corresponding threonine is highly conserved in various P-450s [4,5]. The sequence alignment of rabbit liver microsomal P-450 [laurate(to- 1)hydroxylase] [P-450(to - 1)] suggested that Thr-301 corresponds to the distal threonine. In fact, replacement of Thr-301 to an other residue affected the catalytic activity of this enzyme [6,7]. Since P - 4 5 0 ( t o - 1) has not been crystallized, it is interesting to examine spectroscopically whether Thr-301 of P - 4 5 0 ( t o - 1) is really located in the vicinity of the sixth ligand of the heme iron. Resonance Raman (RR) spectroscopy has provided detailed structural information about heme vicinity of various heme proteins and their reaction intermediates (see Refs. 8, 9 for a review). Accordingly, in the present study, we investigated R R spectra for a few Thr-301 mutants of P-450(to - 1) obtained by the sitedirected mutagenesis and found that the F e - C O stretching mode (vFe.CO) is affected by replacement of Thr-301 with Val but not by the replacement with Ser. We also found that binding of normal alkyl fatty acids with longer carbon chains than C10 causes an upshift of the ~'Fe-co frequency.
Materials and Methods Wild-type and Thr-301 mutated P - 4 5 0 ( t o - 1) were prepared from the transformed yeast [7,10] as described
0167-4838/90/$03.50 © 1990 ElsevierSciencePublishers B.V. (BiomedicalDivision)
212 previously [11] with a small modification for large scale preparation. The yeast precultivated in the synthetic medium (2% glucose and 0.67% yeast nitrogen bases without amino acids supplemented with amino acids and bases [12]) was inoculated to YEPD medium (1% glucose, 2% Bactopeptone and 2% yeast extract). Cultivation was carried out aerobically at 30 ° C for 17 h and the cell was harvested. The P-450s were purified from the microsomal fraction with successive chromatography on ~-aminooctyl Sepharose, hydroxyapatite and CM-Sephadex C-50 columns. P-450 was reduced by a small amount of sodium dithionite and its R R spectrum was measured at around 0 ° C under N2 atmosphere. Since the reduced form appeared to be very unstable, its R R spectrum was measured before completion of reduction with a minimum laser power (5-7 mW at the sample point) and a spinning cell (1800 rpm) within a few minutes (46-115 s). Accordingly, the sample of the reduced form was actually a mixture of unreduced species, reduced P-450 and reduced P-420. After the Raman measurements, CO gas was introduced into the cell and its absorption spectrum was measured to quantify an amount of P-420. Under the present experimental conditions, 70% of the enzyme remained as the active form after the Raman measurements. The RR spectra of the CO adduct of the substrate bound form were measured with the spinning cell after adding CO gas and dithionite to the same samples as used for the measurements of visible absorption spectra for the oxidized form. P-420 was obtained by leaving the dithionite containing enzyme in the N 2filled Raman cell at 4 ° C overnight and it was confirmed with absorption spectrum for the CO adduct. Raman scattering was excited by an H e / C d laser (Kinmon Electrics CDR80SG) or a Kr + ion laser (Spectra Physics, Model 2026) and recorded on a JEOL 400D Raman spectrometer with a cooled photomultiplier (RCA 31034a) for stable samples and a Spex 1404 double monochromator with a diode array detector (PAR 1420) operated by an OMA-2 system (PAR 1215) for unstable samples. The frequencies of Raman bands were calibrated with indene for each measurement. Results and Discussion Fig. 1 shows the R R spectra of oxidized P-450(oa - 1) excited at 406.7 nm. The strongest band at 1372 cm -1 is assigned to the v4 mode [13], which reflects electron delocalization from the iron ion to the porphyrin ~r *(es) orbitals and therefore is sensitive to an axial ligand [14]. There is no shift of this frequency between the wild type and mutants. Some other R R bands can be used as diagnostic markers for the coordination number and spin states of the heme iron [15,16]; the /'Pl0 and v3 bands generally appear around 1640-1636 cm -1 and 1503-1499 cm -1
1625
t6~. Wild
1565 15~
,t4Tl\ A
1"-83
1372
v.9~,
Raman s h i t t / c m -1 Fig. 1. Resonance Raman spectra of oxidized P-450(o - 1)s. (A) wild type, (B) T301V, (C) T301S. S denotes a Raman band of glycerol.The solid and broken vertical lines indicate the high-spin and low-spin marker bands, respectively. Laser, 406.7 nm, 5 mW at sample point; instruments, OMA systems in which accumulation time 660 s.
for the hexa-coordinate ferric low-spin hemes, respectively, around 1631-1627 cm -1 and 1500-1495 cm -1 for the penta-coordinate ferric high-spin hemes, and 1620-1608 and 1485-1480 cm -1 for hexa-coordinate ferric high-spin hemes, although their frequencies are occasionally found outside of the typical regions depending on the axial ligands. R R spectra of various P-450s were previously classified on the basis of such an empirical rule [17]. The R R spectra shown in Fig. 1 gave the low-spin marker bands at 1636 (vl0) and 1499 cm -1 (/"3) and the high-spin ones at 1625 (Vl0) and 1483 cm -1 (v3). The vl0 and u3 frequencies of the high spin species are close to those of P-450sc c (1617 and 1485 cm - t ) [17] and P-450¢~ (1623 and 1488 cm -a) [18], which are considered to adopt the six-coordinate structure in the solution [17], but are distinct from the Vl0 frequencies of P-4481(PB ) (1629 cm -1) [19] and P-4481(MC ) (1630 c m - 1 ) [19] which are very close to those of typical five coordinate high-spin hemes (v 3 is not reported). Therefore, the high-spin species of P-450(oa- 1) is categorized into the six-coordinate heme from R R spectroscopy. The coexistence of the low-spin and high-spin hemes remain unaltered after mutation of Thr-301, although the relative population of the low-spin heme seems to be slightly decreased in mutants. When a saturating amount of laurate was added to the oxidized P-450 solutions, relative intensities of the marker bands were altered. Fig. 2 shows the difference spectra of the substrate-bound form minus its free form. Upon addition of laurate the strongest v4 band exhibits appreciable narrowing and the high-spin marker bands are markedly intensified at the expense of the low-spin bands. Note that the frequencies of the high-spin bands became more reliable and are unaltered from those in Fig. 1. This means that the substrate binding simply shifted the spin equilibrium toward the high-spin form without changing the geometry of the high-spin heme,
213 1625 Wild
~,
P-450/P-420 and
1565
lse~= !
~,~l, s !
~3~ , i
,
I i
Raman
i
i i
,
shift/cm
-1
Fig. 2. Resonance Raman spectra of laurate-bound forms of oxidized P-450(~0 - 1) which were obtained as difference spectra. (A) wild type, (B) T301V, (C) T301S. Meanings of S, and solid and broken vertical lines and experimental conditions are the same as those in the caption of Fig. 1.
in agreement with the observation in the absorption spectra [10]. This trend remained practically unchanged in the mutated P-450(to - 1), although the T301S species seems to resist slightly against the spin conversion. Reduced P-450(to - 1)s were not so stable that their RR spectra were observed with a low laser power (5-7 mW at the sample point) within a few minutes (46-115 s) after addition of dithionite. Nevertheless, P-420 were contaminated in addition to the unreduced P-450. Since high-quality RR spectra could be obtained for each of reduced P-420 and oxidized P-450, the R R spectra of P-450 were determined as the difference between the observed spectra and those of reduced P-420 and oxidized P-450. Fig. 3 shows an example of the reduced
oxidized P-450 mixture (B), pure reduced P-420 (C), oxidized P-450 (D) and their difference (A) of the T301S mutant. After the measurement of spectrum (B) the absorption spectrum of its CO adduct was measured. This indicated that 70% of the enzyme remained as the active P-450. The RR spectra of the reduced forms of the wild type and the T301V mutant were similar to Fig. 3A, which is close to those of other reduced P-450s [18-22]; the v4 band appears at an unusually low frequency (1342-1340 cm-1), and when it is converted to P-420, the v4 band returns to the normal frequency of reduced hemeproteins. These characteristics of P-450 are not altered by replacement of Thr-301 with Val or Ser. When CO gas was introduced into the cell containing the reduced P-450, all of the mutant P-450s similarly gave the Soret band at 450 nm and their v4 frequencies were coincident within ___1 cm -1 at 1365 cm -1. The RR spectra in the low frequency region are shown in Fig. 4, where the 441.6 nm excited RR spectra in the 600-300 cm -1 region of the wild type (a), T301V (b) and T301S (c) species are displayed. The band at 473-476 cm -1 exhibited an isotopic frequency shift by - 8 cm-1 with ~2ClsO and by - 3 . 5 cm -~ with 13C160. Therefore, the band was assigned to the F e - C O stretching vibration (VF¢_CO). This frequency is distinctly lower tfian the corresponding modes of other heme proteins, which appear generally around 490-520 c m - ] [23,24], but are close to those of other P-450s [25,26].
1628 158/,
-~)
Rarnan shift/cm-1
Fig. 3. Resonance Raman spectra of the T301S species. (A) Spectrum of P-450 (Ire 2+ ) obtained as difference [ = ( B ) - ( C ) - (D)]. (B) Observed spectrum for the mixture of oxidized P-450, reduced P-450 and reduced P-420. (C) Spectrum of reduced P-420. (D) Spectrum of oxidized P-450. Laser, 406.7 nm, 7 mW at sample point; instrument, OMA system in which accumulation time is 66 s.
Raman shift/c rn"t Fig. 4. Resonance Raman spectra of the CO adduct of reduced P-450(00-1)s. (A) Difference spectrum [ = spectrum ( c ) - spectrum (b)]; (B) difference spectrum [ = spectrum ( a ) - spectrum (c)]; (a) wild type, (b) T301V and (c) T301S. Spectra (a), (b) and (c) were observed with a scanning spectrometer and an excitation line at 441.6 nm (12 roW). Scanning speed, 10 cm-1/min.
214 We emphasize that the gee-co frequency of the T301V species is higher by 3 cm-1 than those of the wild type and T301S. Since the band is broad, one may doubt the observation of the frequency shift. In order to demonstrate it more explicitly, difference spectra were calculated. As shown by the upper traces in Fig. 4, spectrum (c) minus spectrum (b) yields a derivative-like curve around 473 cm-1 as shown by trace (A) while spectrum (a) minus spectrum (c) gives a flat curve in that region as shown by trace (B). Although the derivative-like curve is obscured by the inclined base line, the presence of the band shift is evident from trace (A). The frequency shift was observed for all isotopes of ~3C160 and 12C180 and repeatedly confirmed with different spectrometers. Note that the difference between spectra (b) and (a) gave a similar differential type curve around 473 cmbut with a less clear shape than that of (c)-(b), suggesting that the wild type contains a small amount of the higher frequency component besides the main band of the lower-frequency counterpart. Since the RR spectra of the oxidized and reduced forms scarcely depended on the mutation of Thr-301, the structure of a whole protein would be little different between T301V and others. Accordingly, the frequency shift of VFo_COis probably due to some local effect near bound CO. One of the plausible interpretations is to assume that the frequency shift occurs owing to direct interaction between bound CO and the distal residue. Yu and co-workers [27,28] systematically studied the correlation between the VFe.CO frequencies and steric hindrance to bound CO and concluded that the increase of the steric hindrance raises the VFe.CO frequency; the vFe_COband is observed at 509 cm -1 for the 15-strapped heroes with less steric hindrance and at 514 cm -1 for 13-strapped heroes with larger steric hindrance. For P-450sc c, which gives the VF,.CO band at 477 cm -1 in the substrate-free state, the vFe.co frequency was up-shifted to 487 and 483 cm -1 upon binding of 22(S)-hydroxycholesterol and 25-hydroxycholesterol, respectively, but not shifted upon binding of 20(S)-hydroxycholesterol and 22-ketocholesterol [26], and the up-shift of VFe_COwas ascribed to steric hindrance. For P-450cam, the presence of camphor raised the VF~_COfrequency from 464 to 481 cm -1 [25]. On the basis of these observations, it is inferred that the T301V protein imposes appreciable steric interaction to the sixth coordination site and accordingly bound CO is more tilted in T301V than in T301S and the wild type. This postulates direct interaction of the methyl side chain of Val-301 with CO. An alternative interpretation assumes that the /)Fe,-CO frequency depends upon the hydrophobicity around CO, This idea arised from the observation of the vF~.co bands for mutant COHbs [29]; Val-Ellfl of Hb occupies a part of the CO binding site of the fl-chain of normal Hb, and when Val-Ellfl was replaced by a larger or smaller nonpolar residue, its band position
remained unshifted but when it was replaced by a polar group, it exhibited a frequency shift. Based on this idea, removal of the side-chain OH group from the distal residue of P-450 is considered to cause a change of hydrophobicity at the CO binding site and thus the frequency shift. However, it is puT~,ling that deletion of the OH group caused an upshift for P-450(0~ - 1) contrary to the fact that mutation of distal histidine (HisE7fl) of Hb to a nonpolar Gly or Val lowered the vFe.CO frequency [29]. In the Thr to Val mutation of P-450(to- 1), the increase of steric hindrance by the c n 3 group should raise the VFe_COfrequency but the hydrophilic to hydrophobic change of environments should lower its frequency. It is most probable that both effects are competing and the magnitude of the former is slightly larger in this case. Then it is understandable that the Thr to Ser mutation does not induce a frequency shift, since the hydrophobicity remains unchanged. The presence of the higher frequency component in the wild type may suggest the presence of two rotamers for the CH3-CH(OH )- side-chain of Thr; the CH 3 group does not impose the steric hindrance to CO in one rotamer, but it does in the other and the latter rotamer is a minor component. The minor component of the rotamer ceases to exist upon removal of the CH 3 group, in agreement with the observed symmetric band shape of the Ser mutant. The VFe.CO frequency of P - 4 5 0 ( ~ - 1) (473 cm -1) is higher than those of P-450cam (464 cm -1, [25]) and rabbit liver microsomal P-450a (467 cm-1, Egawa et al., unpublished data) but slightly lower than that of bovine mitochondrial P-450sc c (477 cm-1, [22]). It was noted previously for P-450scc that when the vvo_co band appears at a higher frequency as in the substrate bound forms, the bound CO is more photodissociable [26]. The CO adduct of P-450(t~- 1) was considerably photostable and may correspond to the least photodissociable adduct of P-450sc c. This may suggest that the relatively low frequency of vF~_CO implies weaker sterical hindrance in addition to a more hydrophobic environment. The VF~.CO band exhibited a frequency shift upon binding of normal alkyl fatty acids. The normal fatty acids with the carbon number of C9 to Ca3 serve as a substrate for this P-450(t0- 1) (Imai, unpublished resuits). It is known that binding of a substrate causes a shift of Sorer band from 417 to 393 nm for the oxidized P-450(o~- 1) [10] and accordingly the relative amount of the substrate-bound P-450(~0 - 1) was monitored by the ratio of absorbances at 393 and 417 nm. The lower part (B) of Fig. 5 plots of A393//A417against the carbon number of the fatty acids observed when an excess amount of them were added to the oxidized P-450(to 1) solution. The VF¢.COfrequencies which were observed after CO gas and dithionite were incorporated into the
215 ,80
5
\
[ . . . . . . -]- --- ~
subs.free
t ,0tI
"o
0.5t . . . . . . . . . . . . T--tl
130N subs. freei
i
8
9
2
not exactly parallel with the sterical interaction of CO with the substrate. In conclusion, it is deduced that the up-shift of the VFe.CO frequency upon binding of fatty acids longer than Ct0 and C12 arises from the steric hindrance by those substrates. Similarly, the small up-shift of the 1,Fe.CO frequency of the T301V mutant would also arise from the steric hindrance by the methyl group of Val-301 to bound CO. It became certain that Thr-301 of P4 5 0 ( t o - 1 ) is located in the vicinity of the oxygen binding site similar to Thr-252 of P-450cam. Acknowledgments
t
10 1'1 12
Carbon number Fig. 5. The Fe-CO stretching frequencies and the absorbance ratios at 393 and 417 run for the CO adduets of two kinds of P-450(t0 -1)s in the presence of the saturated amount of normal alkyl fatty acids. (A) The Fe-CO stretching frequencies vs. the carbon number of fatty acids. Error bars denote general reproducibility of band positions in this measurement. (B) The ratio of absorbances at 393 and 417 nm vs. the carbon number of fatty acids. Circles and triangles denote the data for wild type and the T301V species, respectively. Errors in absorbance are as small as 0.01 and therefore error bars are smaller than individual markers. Solid lines and broken lines indicate the values for the substrate free form of each P-450.
P-450(to-1) (wild-type and T301V) solutions, are plotted in the upper part (A) of Fig. 5. It is apparent that the C8 acid (caprylic acid) binds little to P-450(to - 1) and its ~'F,-co frequency is almost unshifted. The increase of ,/1393/,4417 implies that an appreciable amount of C 9 acid (pelargonic acid) binds to the enzyme, but its ~'F~-COfrequency is practically the same as that of the Ca acid. The C10 (capric acid) and C12 acids (lauric acid) definitely bind to the enzyme (approx. 60% of the enzyme are in the bound form from the Aa93/A417 value). We note that the ~'F~-COfrequencies for the C~o and C12 acid-bound enzymes are alike and distinctly higher than those of the shorter acidbound ones. In the case of the C12 acid, only the to - 1 position was hydroxylated but for the Clo acid both to and t o - 1 positions were hydroxylated, although the latter has higher activity [10]. These results suggest that the normal fatty acids are held by the enzyme at the COOH terminus in the geometry of directing its toterminus toward proximity of the heine iron and the to-terminus of the fatty acids shorter than C 9 does not reach the proximity sufficient for the bound CO to interact with it. This is compatible with the fact that the C 9 to C13 fatty acids can he hydroxylated by this P-450(to - 1). Since the binding geometry of 02 is fairly different from that of CO, it is not unreasonable that the oxygenation activity for a variety of fatty acids does
This study was supported by a Grant-in-Aid for Scientific Research on Priority Areas (63635005) from the Ministry of Education, Science and Culture. T.E. is indebted to Professor T. Masuda of Tokyo Metropolitan University for encouragement during the course of this study. References 1 0 r t i z de Montellano, P.R. (ed.) (1986) Cytochrome P-450: Structure, Mechanism and Biochemistry, Plenum, New York. 2 McMurry, T.J. and Groves, J.T. (1986) Cytochrome P-450; Structure, Mechanism and Biochemistry, Ch. 1, Plenum, New York. 3 Poulos, T.L., Finzel, B.C., Gunsalus, I.C., Wagner, G.C. and Kraut, J. (1985) J. Biol. Chem. 260, 16122-16310. 4 Imai, Y., Komori, M. and Sato, R. (1988) Biochemistry 27, 80-88. 5 Goto, O. and Fujii-Kutiyama, Y. (1988) in Frontiers in Biotransformation (Ruckpaul, K., ed.), Akademie Verlag, Berlin. 6 Imai, Y. and Nakamura, M. (1988) FEBS Lett. 234, 313-315. 7 Imal, Y. and Nakamura, M. (1989) Biochem. Biophys. Res. Commun. 158, 717-722. 8 Spiro, T.G. (ed.) (1978) Biological Application of Raman Spectroscopy, Vol. 3, Wiley & Sons, New York. 9 Kitagawa, T. (1986) in Advances in Spectroscopy (Clark, R.J.H. and Hester, R.E., eds.), Vol. 13, pp. 443-481, Wiley, Chichester. 10 Imai, Y. (1988) J. Biochem. 103, 143-148. 11 Uno, T. and Imai, Y. (1989) J. Biochem. 106, 569-574. 12 Imai, Y. (1987) J. Biochem. 101, 1129-1139. 13 Abe, M., Kitagawa, T. and Kyogoku, Y. (1978) J. Chem. Phys. 69, 4526-4534. 14 Kitagawa, T. (1987) Pure Appl. Chem. 59, 1285-1294. 15 Spiro, T.G. and Strekas, T.C. (1974) J. Am. Chem. SOc. 96, 338-345. 16 Kitagawa, T., Kyogoku, Y., Iizuka, T. and Ikeda-Saito, M. (1976) J. Am. Chem. SOc. 98, 5169-5173. 17 Shimizu, T., Kitagawa, T., Mitani, F., Iizuka, T. and Ishimura, Y. (1981) Biochim. Biophys. Acta 670, 236-242. 18 Champion, P.M., Gunsalus, I.C. and Wagner, G.C. (1978) J. Am. Chem. SOc. 100, 3743-3751. 19 Ozaki, Y., Kitagawa, T., Kyogoku, Y., Imai, Y., HashimotoYutsudo, C. and Sato, R. (1978) Biochemistry 17, 5826-5831. 20 Anzenbacher, P., Stepanek, J., Baumruk, V., Janig, G.-R. and Ruckpaul, K. (1987) Stud. Biophys. 118, 183-188. 21 Ozaki, Y., Kitagawa, T., Kyogoku, Y., Shimada, H., lizuka, T. and Ishimura, Y. (1976) J. Biochem. 80, 1447-1451. 22 Tsubaki, M. and Ichikawa, Y. (1985) Biochim. Biophys. Acta 827, 268-274.
216 23 Tsubaki, M., Srivastava, R.B. and Yu, N.T. (1982) Biochemistry 21, 1132-1140. 24 Li, X.-Y., Spiro, T.G. (1988) J. Am. Chem. Soc. 110, 6024-6033. 25 Uno, T., Nishimura, Y., Makino, R., Iizuka, T., Ishimura, Y. and Tsuboi, M. (1985) J. Biol. Chem. 260, 2023-2026. 26 Tsubaki, M., Hiwatashi, A. and Ichikawa, Y. (1987) Biochemistry 26, 4535-4540.
27 Yu, N.T., Kerr, E.A., Ward, B. and Chang, C.K. (1983) Biochemistry 22, 4534-4540. 28 Kerr, E.A., Mackin, H.C. and Yu, N.T. (1983) Biochemistry 22, 4373-4379. 29 Nagai, K., Luisi, B., Shih, D., Miyazaki, G., Imai, K., Poyart, C., DeYoung, C., Kwiatkowsky, L., Noble, R.W., Lin, S-H. and Yu, N.T. (1987) Nature 329, 858-860.
211
Elsevier BBAPRO 33717
Resonance Raman study on mutant cytochrome P-450 obtained by site-directed mutagenesis T. Egawa 1, y . Imai
2, T. Ogura
1 and T. Kitagawa 1
1 Institute for Molecular Science, Okazaki'National Research Institutes, Myodaiji, Okazaki and 2 Department of Veterinary Science, Unioersity of Osaka Prefecture, Mozuumemachi, Sakai (Japan)
(Received 28 December1989)
Key words: ResonanceRaman; CytochromeP-450; Laurate(to- 1)-hydroxylase;Carbonmonoxideadduct
Resonance Raman spectra were observed for the threonine-301 to serine or valine mutant as well as the wild type of rabbit liver microsomal cytochrome/'-450 [laurate(0:- 1)-hydroxylase] (P-450(o:- 1)), which were prepared through site-directed mutagenesis. The high-spin marker resonance Raman (RR) bands became similarly stronger for all the P-450s examined in the oxidized form upon addition of laurate, and the RR spectra in the higher frequency region of the oxidized, reduced and CO-adduct forms did not distinctly differ among the P-450s examined. Nevertheless, the Fe-CO stretching mode (Pw-co) of the CO adduct exhibited an upshift for the valine mutant, suggesting positional proximity of Thr-301 to bound CO like Thr-252 of P-450~m, in agreement with the expectation from the sequence analysis. The 1,w.co band was shifted to higher frequency upon binding of normal alkyl fatty acids with Ct0 or longer alkyl chain but little affected by binding of shorter fatty acids.
Introduction Cytochrome P-450 (P-450) is a family of heme proteins which have a monoxygenase activity and give the Soret band around 450 nm in their reduced CO adduct [1]. All P-450s are believed to have a cysteinate as the fifth ligand of the heine iron, but other structural characteristics of P-450s are not known. Since the synthetic ferryloxo iron-porphyrins exhibit the monoxygenase activity like P-450 [2], despite the fact that the ferryloxo heme is found for intermediates of peroxidase catalysis, we are curious to know what kinds of residue in the heine proximity play an essential role in the P-450 catalysis. X-ray crystallographic analysis was completed only for P-450c~ isolated from P s e u d o m o n a s p u t i d a [3]. The catalytic site is surrounded by hydrophobic residues. The X-ray study located Thr-252 at the distal heme
Abbreviations: P-450, cytoehrome P-450; P-450(to-1), P-450 [laurate(to-1)-hydroxylase]; P-450~, P-450 from Pseudomonas putida; P-450scc, adrenal cortex mitochondrial P-450 which catalyzes side-chain cleavage of cholesterol; P-4481, a high spin type /-450 purified from PB- or MC-induced rabbit liver mierosomes; PB, phenobarbital; MC, 3-methylcholantlirene; RR, resonance Raman; Hb, hemoglobin. Correspondence: T. Kitagawa, Institute for Molecular Science, Okazaki National Research Institutes, Myodaiji, Okazaki, 444 Japan.
surface, which might serve as a key residue in the P-450 catalysis. The corresponding threonine is highly conserved in various P-450s [4,5]. The sequence alignment of rabbit liver microsomal P-450 [laurate(to- 1)hydroxylase] [P-450(to - 1)] suggested that Thr-301 corresponds to the distal threonine. In fact, replacement of Thr-301 to an other residue affected the catalytic activity of this enzyme [6,7]. Since P - 4 5 0 ( t o - 1) has not been crystallized, it is interesting to examine spectroscopically whether Thr-301 of P - 4 5 0 ( t o - 1) is really located in the vicinity of the sixth ligand of the heme iron. Resonance Raman (RR) spectroscopy has provided detailed structural information about heme vicinity of various heme proteins and their reaction intermediates (see Refs. 8, 9 for a review). Accordingly, in the present study, we investigated R R spectra for a few Thr-301 mutants of P-450(to - 1) obtained by the sitedirected mutagenesis and found that the F e - C O stretching mode (vFe.CO) is affected by replacement of Thr-301 with Val but not by the replacement with Ser. We also found that binding of normal alkyl fatty acids with longer carbon chains than C10 causes an upshift of the ~'Fe-co frequency.
Materials and Methods Wild-type and Thr-301 mutated P - 4 5 0 ( t o - 1) were prepared from the transformed yeast [7,10] as described
0167-4838/90/$03.50 © 1990 ElsevierSciencePublishers B.V. (BiomedicalDivision)
212 previously [11] with a small modification for large scale preparation. The yeast precultivated in the synthetic medium (2% glucose and 0.67% yeast nitrogen bases without amino acids supplemented with amino acids and bases [12]) was inoculated to YEPD medium (1% glucose, 2% Bactopeptone and 2% yeast extract). Cultivation was carried out aerobically at 30 ° C for 17 h and the cell was harvested. The P-450s were purified from the microsomal fraction with successive chromatography on ~-aminooctyl Sepharose, hydroxyapatite and CM-Sephadex C-50 columns. P-450 was reduced by a small amount of sodium dithionite and its R R spectrum was measured at around 0 ° C under N2 atmosphere. Since the reduced form appeared to be very unstable, its R R spectrum was measured before completion of reduction with a minimum laser power (5-7 mW at the sample point) and a spinning cell (1800 rpm) within a few minutes (46-115 s). Accordingly, the sample of the reduced form was actually a mixture of unreduced species, reduced P-450 and reduced P-420. After the Raman measurements, CO gas was introduced into the cell and its absorption spectrum was measured to quantify an amount of P-420. Under the present experimental conditions, 70% of the enzyme remained as the active form after the Raman measurements. The RR spectra of the CO adduct of the substrate bound form were measured with the spinning cell after adding CO gas and dithionite to the same samples as used for the measurements of visible absorption spectra for the oxidized form. P-420 was obtained by leaving the dithionite containing enzyme in the N 2filled Raman cell at 4 ° C overnight and it was confirmed with absorption spectrum for the CO adduct. Raman scattering was excited by an H e / C d laser (Kinmon Electrics CDR80SG) or a Kr + ion laser (Spectra Physics, Model 2026) and recorded on a JEOL 400D Raman spectrometer with a cooled photomultiplier (RCA 31034a) for stable samples and a Spex 1404 double monochromator with a diode array detector (PAR 1420) operated by an OMA-2 system (PAR 1215) for unstable samples. The frequencies of Raman bands were calibrated with indene for each measurement. Results and Discussion Fig. 1 shows the R R spectra of oxidized P-450(oa - 1) excited at 406.7 nm. The strongest band at 1372 cm -1 is assigned to the v4 mode [13], which reflects electron delocalization from the iron ion to the porphyrin ~r *(es) orbitals and therefore is sensitive to an axial ligand [14]. There is no shift of this frequency between the wild type and mutants. Some other R R bands can be used as diagnostic markers for the coordination number and spin states of the heme iron [15,16]; the /'Pl0 and v3 bands generally appear around 1640-1636 cm -1 and 1503-1499 cm -1
1625
t6~. Wild
1565 15~
,t4Tl\ A
1"-83
1372
v.9~,
Raman s h i t t / c m -1 Fig. 1. Resonance Raman spectra of oxidized P-450(o - 1)s. (A) wild type, (B) T301V, (C) T301S. S denotes a Raman band of glycerol.The solid and broken vertical lines indicate the high-spin and low-spin marker bands, respectively. Laser, 406.7 nm, 5 mW at sample point; instruments, OMA systems in which accumulation time 660 s.
for the hexa-coordinate ferric low-spin hemes, respectively, around 1631-1627 cm -1 and 1500-1495 cm -1 for the penta-coordinate ferric high-spin hemes, and 1620-1608 and 1485-1480 cm -1 for hexa-coordinate ferric high-spin hemes, although their frequencies are occasionally found outside of the typical regions depending on the axial ligands. R R spectra of various P-450s were previously classified on the basis of such an empirical rule [17]. The R R spectra shown in Fig. 1 gave the low-spin marker bands at 1636 (vl0) and 1499 cm -1 (/"3) and the high-spin ones at 1625 (Vl0) and 1483 cm -1 (v3). The vl0 and u3 frequencies of the high spin species are close to those of P-450sc c (1617 and 1485 cm - t ) [17] and P-450¢~ (1623 and 1488 cm -a) [18], which are considered to adopt the six-coordinate structure in the solution [17], but are distinct from the Vl0 frequencies of P-4481(PB ) (1629 cm -1) [19] and P-4481(MC ) (1630 c m - 1 ) [19] which are very close to those of typical five coordinate high-spin hemes (v 3 is not reported). Therefore, the high-spin species of P-450(oa- 1) is categorized into the six-coordinate heme from R R spectroscopy. The coexistence of the low-spin and high-spin hemes remain unaltered after mutation of Thr-301, although the relative population of the low-spin heme seems to be slightly decreased in mutants. When a saturating amount of laurate was added to the oxidized P-450 solutions, relative intensities of the marker bands were altered. Fig. 2 shows the difference spectra of the substrate-bound form minus its free form. Upon addition of laurate the strongest v4 band exhibits appreciable narrowing and the high-spin marker bands are markedly intensified at the expense of the low-spin bands. Note that the frequencies of the high-spin bands became more reliable and are unaltered from those in Fig. 1. This means that the substrate binding simply shifted the spin equilibrium toward the high-spin form without changing the geometry of the high-spin heme,
213 1625 Wild
~,
P-450/P-420 and
1565
lse~= !
~,~l, s !
~3~ , i
,
I i
Raman
i
i i
,
shift/cm
-1
Fig. 2. Resonance Raman spectra of laurate-bound forms of oxidized P-450(~0 - 1) which were obtained as difference spectra. (A) wild type, (B) T301V, (C) T301S. Meanings of S, and solid and broken vertical lines and experimental conditions are the same as those in the caption of Fig. 1.
in agreement with the observation in the absorption spectra [10]. This trend remained practically unchanged in the mutated P-450(to - 1), although the T301S species seems to resist slightly against the spin conversion. Reduced P-450(to - 1)s were not so stable that their RR spectra were observed with a low laser power (5-7 mW at the sample point) within a few minutes (46-115 s) after addition of dithionite. Nevertheless, P-420 were contaminated in addition to the unreduced P-450. Since high-quality RR spectra could be obtained for each of reduced P-420 and oxidized P-450, the R R spectra of P-450 were determined as the difference between the observed spectra and those of reduced P-420 and oxidized P-450. Fig. 3 shows an example of the reduced
oxidized P-450 mixture (B), pure reduced P-420 (C), oxidized P-450 (D) and their difference (A) of the T301S mutant. After the measurement of spectrum (B) the absorption spectrum of its CO adduct was measured. This indicated that 70% of the enzyme remained as the active P-450. The RR spectra of the reduced forms of the wild type and the T301V mutant were similar to Fig. 3A, which is close to those of other reduced P-450s [18-22]; the v4 band appears at an unusually low frequency (1342-1340 cm-1), and when it is converted to P-420, the v4 band returns to the normal frequency of reduced hemeproteins. These characteristics of P-450 are not altered by replacement of Thr-301 with Val or Ser. When CO gas was introduced into the cell containing the reduced P-450, all of the mutant P-450s similarly gave the Soret band at 450 nm and their v4 frequencies were coincident within ___1 cm -1 at 1365 cm -1. The RR spectra in the low frequency region are shown in Fig. 4, where the 441.6 nm excited RR spectra in the 600-300 cm -1 region of the wild type (a), T301V (b) and T301S (c) species are displayed. The band at 473-476 cm -1 exhibited an isotopic frequency shift by - 8 cm-1 with ~2ClsO and by - 3 . 5 cm -~ with 13C160. Therefore, the band was assigned to the F e - C O stretching vibration (VF¢_CO). This frequency is distinctly lower tfian the corresponding modes of other heme proteins, which appear generally around 490-520 c m - ] [23,24], but are close to those of other P-450s [25,26].
1628 158/,
-~)
Rarnan shift/cm-1
Fig. 3. Resonance Raman spectra of the T301S species. (A) Spectrum of P-450 (Ire 2+ ) obtained as difference [ = ( B ) - ( C ) - (D)]. (B) Observed spectrum for the mixture of oxidized P-450, reduced P-450 and reduced P-420. (C) Spectrum of reduced P-420. (D) Spectrum of oxidized P-450. Laser, 406.7 nm, 7 mW at sample point; instrument, OMA system in which accumulation time is 66 s.
Raman shift/c rn"t Fig. 4. Resonance Raman spectra of the CO adduct of reduced P-450(00-1)s. (A) Difference spectrum [ = spectrum ( c ) - spectrum (b)]; (B) difference spectrum [ = spectrum ( a ) - spectrum (c)]; (a) wild type, (b) T301V and (c) T301S. Spectra (a), (b) and (c) were observed with a scanning spectrometer and an excitation line at 441.6 nm (12 roW). Scanning speed, 10 cm-1/min.
214 We emphasize that the gee-co frequency of the T301V species is higher by 3 cm-1 than those of the wild type and T301S. Since the band is broad, one may doubt the observation of the frequency shift. In order to demonstrate it more explicitly, difference spectra were calculated. As shown by the upper traces in Fig. 4, spectrum (c) minus spectrum (b) yields a derivative-like curve around 473 cm-1 as shown by trace (A) while spectrum (a) minus spectrum (c) gives a flat curve in that region as shown by trace (B). Although the derivative-like curve is obscured by the inclined base line, the presence of the band shift is evident from trace (A). The frequency shift was observed for all isotopes of ~3C160 and 12C180 and repeatedly confirmed with different spectrometers. Note that the difference between spectra (b) and (a) gave a similar differential type curve around 473 cmbut with a less clear shape than that of (c)-(b), suggesting that the wild type contains a small amount of the higher frequency component besides the main band of the lower-frequency counterpart. Since the RR spectra of the oxidized and reduced forms scarcely depended on the mutation of Thr-301, the structure of a whole protein would be little different between T301V and others. Accordingly, the frequency shift of VFo_COis probably due to some local effect near bound CO. One of the plausible interpretations is to assume that the frequency shift occurs owing to direct interaction between bound CO and the distal residue. Yu and co-workers [27,28] systematically studied the correlation between the VFe.CO frequencies and steric hindrance to bound CO and concluded that the increase of the steric hindrance raises the VFe.CO frequency; the vFe_COband is observed at 509 cm -1 for the 15-strapped heroes with less steric hindrance and at 514 cm -1 for 13-strapped heroes with larger steric hindrance. For P-450sc c, which gives the VF,.CO band at 477 cm -1 in the substrate-free state, the vFe.co frequency was up-shifted to 487 and 483 cm -1 upon binding of 22(S)-hydroxycholesterol and 25-hydroxycholesterol, respectively, but not shifted upon binding of 20(S)-hydroxycholesterol and 22-ketocholesterol [26], and the up-shift of VFe_COwas ascribed to steric hindrance. For P-450cam, the presence of camphor raised the VF~_COfrequency from 464 to 481 cm -1 [25]. On the basis of these observations, it is inferred that the T301V protein imposes appreciable steric interaction to the sixth coordination site and accordingly bound CO is more tilted in T301V than in T301S and the wild type. This postulates direct interaction of the methyl side chain of Val-301 with CO. An alternative interpretation assumes that the /)Fe,-CO frequency depends upon the hydrophobicity around CO, This idea arised from the observation of the vF~.co bands for mutant COHbs [29]; Val-Ellfl of Hb occupies a part of the CO binding site of the fl-chain of normal Hb, and when Val-Ellfl was replaced by a larger or smaller nonpolar residue, its band position
remained unshifted but when it was replaced by a polar group, it exhibited a frequency shift. Based on this idea, removal of the side-chain OH group from the distal residue of P-450 is considered to cause a change of hydrophobicity at the CO binding site and thus the frequency shift. However, it is puT~,ling that deletion of the OH group caused an upshift for P-450(0~ - 1) contrary to the fact that mutation of distal histidine (HisE7fl) of Hb to a nonpolar Gly or Val lowered the vFe.CO frequency [29]. In the Thr to Val mutation of P-450(to- 1), the increase of steric hindrance by the c n 3 group should raise the VFe_COfrequency but the hydrophilic to hydrophobic change of environments should lower its frequency. It is most probable that both effects are competing and the magnitude of the former is slightly larger in this case. Then it is understandable that the Thr to Ser mutation does not induce a frequency shift, since the hydrophobicity remains unchanged. The presence of the higher frequency component in the wild type may suggest the presence of two rotamers for the CH3-CH(OH )- side-chain of Thr; the CH 3 group does not impose the steric hindrance to CO in one rotamer, but it does in the other and the latter rotamer is a minor component. The minor component of the rotamer ceases to exist upon removal of the CH 3 group, in agreement with the observed symmetric band shape of the Ser mutant. The VFe.CO frequency of P - 4 5 0 ( ~ - 1) (473 cm -1) is higher than those of P-450cam (464 cm -1, [25]) and rabbit liver microsomal P-450a (467 cm-1, Egawa et al., unpublished data) but slightly lower than that of bovine mitochondrial P-450sc c (477 cm-1, [22]). It was noted previously for P-450scc that when the vvo_co band appears at a higher frequency as in the substrate bound forms, the bound CO is more photodissociable [26]. The CO adduct of P-450(t~- 1) was considerably photostable and may correspond to the least photodissociable adduct of P-450sc c. This may suggest that the relatively low frequency of vF~_CO implies weaker sterical hindrance in addition to a more hydrophobic environment. The VF~.CO band exhibited a frequency shift upon binding of normal alkyl fatty acids. The normal fatty acids with the carbon number of C9 to Ca3 serve as a substrate for this P-450(t0- 1) (Imai, unpublished resuits). It is known that binding of a substrate causes a shift of Sorer band from 417 to 393 nm for the oxidized P-450(o~- 1) [10] and accordingly the relative amount of the substrate-bound P-450(~0 - 1) was monitored by the ratio of absorbances at 393 and 417 nm. The lower part (B) of Fig. 5 plots of A393//A417against the carbon number of the fatty acids observed when an excess amount of them were added to the oxidized P-450(to 1) solution. The VF¢.COfrequencies which were observed after CO gas and dithionite were incorporated into the
215 ,80
5
\
[ . . . . . . -]- --- ~
subs.free
t ,0tI
"o
0.5t . . . . . . . . . . . . T--tl
130N subs. freei
i
8
9
2
not exactly parallel with the sterical interaction of CO with the substrate. In conclusion, it is deduced that the up-shift of the VFe.CO frequency upon binding of fatty acids longer than Ct0 and C12 arises from the steric hindrance by those substrates. Similarly, the small up-shift of the 1,Fe.CO frequency of the T301V mutant would also arise from the steric hindrance by the methyl group of Val-301 to bound CO. It became certain that Thr-301 of P4 5 0 ( t o - 1 ) is located in the vicinity of the oxygen binding site similar to Thr-252 of P-450cam. Acknowledgments
t
10 1'1 12
Carbon number Fig. 5. The Fe-CO stretching frequencies and the absorbance ratios at 393 and 417 run for the CO adduets of two kinds of P-450(t0 -1)s in the presence of the saturated amount of normal alkyl fatty acids. (A) The Fe-CO stretching frequencies vs. the carbon number of fatty acids. Error bars denote general reproducibility of band positions in this measurement. (B) The ratio of absorbances at 393 and 417 nm vs. the carbon number of fatty acids. Circles and triangles denote the data for wild type and the T301V species, respectively. Errors in absorbance are as small as 0.01 and therefore error bars are smaller than individual markers. Solid lines and broken lines indicate the values for the substrate free form of each P-450.
P-450(to-1) (wild-type and T301V) solutions, are plotted in the upper part (A) of Fig. 5. It is apparent that the C8 acid (caprylic acid) binds little to P-450(to - 1) and its ~'F,-co frequency is almost unshifted. The increase of ,/1393/,4417 implies that an appreciable amount of C 9 acid (pelargonic acid) binds to the enzyme, but its ~'F~-COfrequency is practically the same as that of the Ca acid. The C10 (capric acid) and C12 acids (lauric acid) definitely bind to the enzyme (approx. 60% of the enzyme are in the bound form from the Aa93/A417 value). We note that the ~'F~-COfrequencies for the C~o and C12 acid-bound enzymes are alike and distinctly higher than those of the shorter acidbound ones. In the case of the C12 acid, only the to - 1 position was hydroxylated but for the Clo acid both to and t o - 1 positions were hydroxylated, although the latter has higher activity [10]. These results suggest that the normal fatty acids are held by the enzyme at the COOH terminus in the geometry of directing its toterminus toward proximity of the heine iron and the to-terminus of the fatty acids shorter than C 9 does not reach the proximity sufficient for the bound CO to interact with it. This is compatible with the fact that the C 9 to C13 fatty acids can he hydroxylated by this P-450(to - 1). Since the binding geometry of 02 is fairly different from that of CO, it is not unreasonable that the oxygenation activity for a variety of fatty acids does
This study was supported by a Grant-in-Aid for Scientific Research on Priority Areas (63635005) from the Ministry of Education, Science and Culture. T.E. is indebted to Professor T. Masuda of Tokyo Metropolitan University for encouragement during the course of this study. References 1 0 r t i z de Montellano, P.R. (ed.) (1986) Cytochrome P-450: Structure, Mechanism and Biochemistry, Plenum, New York. 2 McMurry, T.J. and Groves, J.T. (1986) Cytochrome P-450; Structure, Mechanism and Biochemistry, Ch. 1, Plenum, New York. 3 Poulos, T.L., Finzel, B.C., Gunsalus, I.C., Wagner, G.C. and Kraut, J. (1985) J. Biol. Chem. 260, 16122-16310. 4 Imai, Y., Komori, M. and Sato, R. (1988) Biochemistry 27, 80-88. 5 Goto, O. and Fujii-Kutiyama, Y. (1988) in Frontiers in Biotransformation (Ruckpaul, K., ed.), Akademie Verlag, Berlin. 6 Imai, Y. and Nakamura, M. (1988) FEBS Lett. 234, 313-315. 7 Imal, Y. and Nakamura, M. (1989) Biochem. Biophys. Res. Commun. 158, 717-722. 8 Spiro, T.G. (ed.) (1978) Biological Application of Raman Spectroscopy, Vol. 3, Wiley & Sons, New York. 9 Kitagawa, T. (1986) in Advances in Spectroscopy (Clark, R.J.H. and Hester, R.E., eds.), Vol. 13, pp. 443-481, Wiley, Chichester. 10 Imai, Y. (1988) J. Biochem. 103, 143-148. 11 Uno, T. and Imai, Y. (1989) J. Biochem. 106, 569-574. 12 Imai, Y. (1987) J. Biochem. 101, 1129-1139. 13 Abe, M., Kitagawa, T. and Kyogoku, Y. (1978) J. Chem. Phys. 69, 4526-4534. 14 Kitagawa, T. (1987) Pure Appl. Chem. 59, 1285-1294. 15 Spiro, T.G. and Strekas, T.C. (1974) J. Am. Chem. SOc. 96, 338-345. 16 Kitagawa, T., Kyogoku, Y., Iizuka, T. and Ikeda-Saito, M. (1976) J. Am. Chem. SOc. 98, 5169-5173. 17 Shimizu, T., Kitagawa, T., Mitani, F., Iizuka, T. and Ishimura, Y. (1981) Biochim. Biophys. Acta 670, 236-242. 18 Champion, P.M., Gunsalus, I.C. and Wagner, G.C. (1978) J. Am. Chem. SOc. 100, 3743-3751. 19 Ozaki, Y., Kitagawa, T., Kyogoku, Y., Imai, Y., HashimotoYutsudo, C. and Sato, R. (1978) Biochemistry 17, 5826-5831. 20 Anzenbacher, P., Stepanek, J., Baumruk, V., Janig, G.-R. and Ruckpaul, K. (1987) Stud. Biophys. 118, 183-188. 21 Ozaki, Y., Kitagawa, T., Kyogoku, Y., Shimada, H., lizuka, T. and Ishimura, Y. (1976) J. Biochem. 80, 1447-1451. 22 Tsubaki, M. and Ichikawa, Y. (1985) Biochim. Biophys. Acta 827, 268-274.
216 23 Tsubaki, M., Srivastava, R.B. and Yu, N.T. (1982) Biochemistry 21, 1132-1140. 24 Li, X.-Y., Spiro, T.G. (1988) J. Am. Chem. Soc. 110, 6024-6033. 25 Uno, T., Nishimura, Y., Makino, R., Iizuka, T., Ishimura, Y. and Tsuboi, M. (1985) J. Biol. Chem. 260, 2023-2026. 26 Tsubaki, M., Hiwatashi, A. and Ichikawa, Y. (1987) Biochemistry 26, 4535-4540.
27 Yu, N.T., Kerr, E.A., Ward, B. and Chang, C.K. (1983) Biochemistry 22, 4534-4540. 28 Kerr, E.A., Mackin, H.C. and Yu, N.T. (1983) Biochemistry 22, 4373-4379. 29 Nagai, K., Luisi, B., Shih, D., Miyazaki, G., Imai, K., Poyart, C., DeYoung, C., Kwiatkowsky, L., Noble, R.W., Lin, S-H. and Yu, N.T. (1987) Nature 329, 858-860.
