- Email: [email protected]
NMR studies of putidaredoxin: associations of putidaredoxin with NADH-putidaredoxin reductase and cytochrome P450cam
Biochimica et Biophysica Acta 1386 (1998) 168^178
NMR studies of putidaredoxin: associations of putidaredoxin with NADH-putidaredoxin reductase and cytochrome P450cam Masaaki Aoki, Koichiro Ishimori, Isao Morishima * Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, Kyoto 606-8501, Japan Received 20 January 1998; accepted 16 March 1998
Abstract To characterize the electron-transfer reaction in the P450cam monooxygenation system, the binding regions of putidaredoxin (Pdx) to NADH-putidaredoxin reductase (PdR) and P450cam were investigated using isotope-filtered NMR experiments in which uniformly 15 N-labeled Pdx ([U-15 N]Pdx) is mixed with unlabeled PdR and P450cam. By addition of PdR to Pdx, site specific signal broadening was observed for the N-H correlation peaks from Val-28, Glu-72, Ile-88, and Gln105. Although previous studies have suggested the contribution from acidic amino acid residues on the G-helix of Pdx to the binding with PdR, no site specific broadening was observed for the resonances from these residues except for Glu-72. The lesser contribution of electrostatic interactions to the Pdx/PdR complex formation was also suggested by our previous study (M. Aoki, K. Ishimori, H. Fukada, K. Takahashi, I. Morishima, Biochim. Biophys. Acta 1384 (1998) 180^188), which is in sharp contrast to the complex formation between adrenodoxin and adrenodoxin reductase. Upon the complex formation between Pdx and P450cam, the site specific NMR line broadening was observed for several amino acid residues distributed near the iron-sulfur cluster, corresponding to the large binding site in the complex formation with P450cam. Since some of the amino acid residues included in the binding site are not conserved for the electron-transfer iron-sulfur proteins such as ferredoxin and adrenodoxin, the interactions formed by these amino acid residues would be highly specific to the binding with P450cam, consistent with very low cross-reactivity to other iron-sulfur proteins in the P450cam monooxygenation system. ß 1998 Elsevier Science B.V. All rights reserved. Keywords: Putidaredoxin; Cytochrome P450cam; Nuclear magnetic resonance; Isotope labeling; Electron-transfer complex; Proteinprotein interaction
1. Introduction Putidaredoxin (Pdx) is a globular protein consisting of a single 106 residue polypeptide (molecular Abbreviations: AdR, adrenodoxin reductase; Adx, adrenodoxin; GARP, globally optimized alternating rectangular pulse; HSQC, heteronuclear single quantum correlation; PdR, NADHputidaredoxin reductase; Pdx, putidaredoxin * Corresponding author. Fax: +81 (75) 751-7611; E-mail: [email protected]
weight 11 500) and a [2Fe-2S] prosthetic group, and serves as a one-electron shuttle between NADH-putidaredoxin reductase (PdR; molecular weight 43 500) and cytochrome P450cam (molecular weight 45 000) [1,2]. Oxidized Pdx associates with reduced PdR and accepts one electron from the reductase as shown in Fig. 1. After the dissociation from oxidized PdR, reduced Pdx binds with P450cam and donates an electron to P450cam which catalyzes the oxygenation of d-camphor to 5-exo-hydroxycamphor. The electron-transfer process from Pdx to
0167-4838 / 98 / $19.00 ß 1998 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 9 8 ) 0 0 0 9 1 - 0
BBAPRO 35672 16-7-98 ^ Pagina 168 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
P450cam is crucial for the P450cam oxygenation cycle, in that the electron transfer is rate limiting for the catalytic cycle [2]. It should also be noted here that Pdx is the only iron-sulfur protein which can reduce the ferrous oxygenated adduct for the P450cam reaction, while a number of reducing agents are competent to reduce ferric P450cam [3]. It was, therefore, concluded that Pdx acts not only as a redox shuttle, but also as a speci¢c e¡ector in the camphor hydroxylation [3]. Although the high speci¢c requirements of Pdx to facilitate the catalytic cycle are of great importance in the P450cam system, the mechanism for the binding of Pdx with the redox partners in the electron-transfer reaction has not yet been clear. Many investigators have tried to reveal the details of the speci¢c electron-transfer complexes, the Pdx/ PdR and Pdx/P450cam complexes, in the P450cam oxygenation system formed by Pdx [4^11]. Using a molecular dynamics-derived model for the Pdx/ P450cam complex, Pochapsky et al. [12] proposed that speci¢c salt bridges between negatively charged amino acid residues on Pdx (aspartic acids 34 and 38, and C-terminal carboxylate) and positively charged groups on P450cam (Arg-79, Arg-109 and Arg-112) promote the protein association. The site-directed mutagenesis studies of Pdx have also revealed that the loop structure constructing the active site (including Asp-34 and Asp-38) and C-terminal region of Pdx is a potential binding site for P450cam [5,9]. The C-terminal residue, Trp-106, on Pdx was also considered to be important for the complex formation between Pdx and PdR, since the substitution of the Trp residue highly inhibited the electron-transfer reaction [9]. In addition to the region near the iron-sulfur cluster, the `negatively charged patch' formed by acidic amino acid residues on the G-helix would be an alternative binding site for PdR or P450cam [7,13]. The chemical modi¢cation of Pdx has shown that neutralization of the negative charges of Asp-58, Glu-65, Glu-67, Glu-72, and Glu-77 severely discouraged the electron transfer from PdR to Pdx [7]. In the vicinity of the acidic region, Cys-73 of Pdx was also identi¢ed as one of the crucial residues for the interaction of Pdx with PdR [9]. However, our thermodynamic study on the associ-
169
ation of Pdx with PdR and P450cam has suggested that associations of Pdx with the redox partners are not strongly dominated by electrostatic interactions [11]. The replacements of the positively charged amino acid residues on P450cam with the corresponding neutral residues did not seriously a¡ect the electrontransfer reaction in the P450cam monooxygenase system, supporting the lesser contribution of the electrostatic interaction to the electron-transfer reaction [4]. Instead of the electrostatic interactions, we have pointed out that the hydrophobic interaction and/or van der Waals interaction play key roles in the formation of the electron-transfer complex in the P450cam monooxygenation system [10,11]. Thus, the interactions and binding sites responsible for the complex formation in the electron-transfer reaction are still controversial. In this study, to identify the amino acid residues on Pdx constructing the binding site with PdR or P450cam, we applied a heteronuclear multi-dimensional NMR technique to the PdR/Pdx and Pdx/ P450cam complexes. The large molecular weight of the complexes between Pdx and the redox partners would prevent us from analyzing the proton signals in one- and two-dimensional 1 H-NMR spectra. However, recent progress in protein expression and puri¢cation has enabled us to prepare enough amount of the isotope-labeled protein for the multinuclear multi-dimensional NMR spectroscopy, which is one of the potent spectroscopies for structural analysis of large molecular weight proteins and protein complexes. If one protein in a complex is labeled with a stable isotope having nuclear spins and the other unlabeled, we can preferentially observe the NMR signals from the isotope in one of the proteins and discuss the structural changes accompanied with the complex formation. For Pdx, Pochapsky and co-workers [12] have already synthesized 15 N-labeled Pdx and measured its two-dimensional 1 H-15 N-NMR spectra. They have assigned the 1 H-15 N correlation peaks for the 15 Nlabeled Pdx [14]. Although they have done some preliminary NMR measurements for the complex of Pdx with P450cam, and indicated that many residues of Pdx might be involved in the association with P450cam, the association site on Pdx for P450cam has not yet been clearly proposed and the complex
BBAPRO 35672 16-7-98 ^ Pagina 169 Cyaan Magenta
Geel
Zwart
170
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
with PdR has not been examined [12]. With an aim to the elucidation of the molecular mechanism in the electron-transfer reaction for the P450cam monooxygenation system, we have also prepared uniformly 15 N-labeled Pdx ([U-15 N]Pdx) and measured the 1 H15 N correlation spectra of [U-15 N]Pdx in the presence or absence of one of the redox partners, PdR or P450cam, to determine the binding regions and amino acid residues on Pdx crucial for the binding with PdR and P450cam. 2. Materials and methods 2.1. Expression of Pdx in
15
N-enriched media
Escherichia coli TG1 cell was transformed with a plasmid which contains a clone of the gene encoding Pdx under transcriptional control of a tac promoter [10,11], plated on 2UTY agar, and grown overnight at 28³C. Colonies were picked up and grown in 10 ml of LB media at 28³C. This culture was centrifuged, and the pellet resuspended in 250 Wl of deionized water. This suspension was added to one liter of M63 minimal medium containing (15 NH4 )2 SO4 as sole nitrogen source [15]. The culture was grown for approx. 30 h at 37³C, then harvested by centrifugation. The cell yield was approx. 2 g/l medium. The cell pellet was frozen at 380³C prior to puri¢cation. 2.2. Puri¢cation of [U-15 N]Pdx The cell pellet was resuspended to 50 mM potassium phosphate (pH 7.4) and lysed with lysozyme in the presence of sodium deoxycholate followed by addition of streptomycin sulfate (approx. 0.01 mg/ ml). Cell debris was removed by centrifugation (9.5 krpm for 20 min at 4³C), and the supernatant loaded onto a DE-52 anion exchange column (P 25 mmU6 cm) equilibrated with 50 mM potassium phosphate (pH 7.4) containing 10 mM 2-mercaptoethanol. After loading, the column was washed with the same bu¡er. Pdx was eluted by 50 mM potassium phosphate (pH 7.4) containing 200 mM KCl and 10 mM 2-mercaptoethanol. The Pdx-containing brown elution was concentrated using an Amicon
concentrator equipped with a YM3 membrane. The concentrated Pdx was loaded onto a Superdex 75 (P 10 mmU30 cm) mounted on a Pharmacia FPLC system equilibrated with 50 mM potassium phosphate (pH 7.4) containing 100 mM KCl and 10 mM 2-mercaptoethanol. Brown fractions were assayed for purity by spectrophotometry. The homogeneous protein has a 325^280 nm absorbance ratio of 0.68 [1]. Fractions for which this ratio is greater than 0.6 were combined and concentrated. Concentration of Pdx was calculated using extinction coe¤cients of 10.4 mM31 cm31 at 455 nm for its oxidized form [1]. Protein yields were approx. 800 Wl of 1 mM Pdx from 6 l of culture, i.e., approx. 2 mg/l medium. 2.3. Preparations of PdR and P450cam PdR and P450cam were also expressed in E. coli and puri¢ed by the procedures previously described [1,10,11]. A puri¢ed sample of PdR with a 280 nm/ 455 nm absorbance ratio of less than 9.5 was employed in this study, which corresponds to approx. 70% purity with respect to total protein [1]. The concentration of PdR was calculated using extinction coe¤cients of 8.5 mM31 cm31 at 480 nm [1]. The purity of the P450cam enzyme was estimated from its RZ (A391 /A280 ) value. Puri¢ed crystalline preparations exhibit an RZ value of 1.63 [1], and all of the P450cam samples used here had RZ values greater than 1.4, which correspond to its purity of approx. 90%. We used an extinction coe¤cient of 102.0 mM31 cm31 at 391 nm to estimate the concentration of the ferric camphor-bound cytochrome P450cam [1]. 2.4. NMR measurements All measurements were recorded on a Bruker AVANCE DRX500 spectrometer equipped with a Silicon Graphics Indy workstation. The spectrometer operates at 500.13 and 50.68 MHz for 1 H and 15 N, respectively. Bu¡ers used for NMR samples of oxidized Pdx were 50 mM potassium phosphate (pH 7.4) containing 10 mM 2-mercaptoethanol as protecting agent in H2 O/D2 O 90/10. The sample volume was approx. 350 Wl. In the Pdx/PdR and Pdx/ P450cam mixtures, the thiol agent was removed to
BBAPRO 35672 16-7-98 ^ Pagina 170 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
171
Fig. 1. Reaction cycle of cytochrome P450cam and the associated electron cascade from NADH via PdR and Pdx, where R-H is a substrate, d-camphor, R-OH is the product, 5-exo-hydroxycamphor, Pdxox and Pdxred are oxidized and reduced putidaredoxin, respectively, and PdRox and PdRred are oxidized and reduced NADH-putidaredoxin reductase, respectively.
avoid possible interference with the redox partners.1 One mM d-camphor was contained in the mixture of Pdx with P450cam. The [U-15 N]Pdx concentration was 0.76 mM for the mixtures with PdR. The concentrations of PdR for the Pdx:PdR = 3:1 and 1:1 mixtures were 0.27 and 0.76 mM, respectively. The [U-15 N]Pdx:P450cam = 3:1 mixture contained 0.50 mM Pdx and 0.17 mM P450cam, respectively. The 1:1 mixture of [U-15 N]Pdx with P450cam contained 0.50 mM of each protein. The sample temperature in the probe was adjusted to 17.0 þ 0.1³C and controlled by the temperature control unit of the spectrometer. The 1 H-15 N HSQC experiments were performed by the published sequence [16] with 15 N decoupling using GARP [17] during the 1 H acquisition period. A total recycling time of approx. 3 s was used in order to suppress a heat-up by decoupling. 1 H-15 N coupling evolution and refocusing decays were set using a value of 2.25 ms (1/4 JHN ). A 1 H spectral width of 8090.6 Hz was used, and the spectral width for 15 N dimension was 8109 Hz. The 1 H dimension was referred to a water signal at 4.7 ppm at 17³C, and the 15 N carrier frequency was placed at 120 ppm. Quadrature detection in the 15 N dimension was obtained using time-proportional phase incrementation 1 The thiol agent, 2-mercaptoethanol, used as a preservative might attach covalently to a solvent exposed cysteine, Cys-73, of Pdx [13]. Although structural changes of Pdx caused by the removal of the thiol agent cannot be excluded, the resonance from Cys-73 covalently bound to the thiol was not drastically a¡ected by the removal of the thiol agent, implying that the e¡ect of the removal on the resonances of [U-15 N]Pdx is not so large.
(TPPI). Free induction decays were acquired with 2048 complex data points, and 512 increments in the 15 N dimension. The 1 H dimension was multiplied by a sine-bell prior to Fourier transformation. A squared sine-bell was applied to interferograms in the 15 N dimension, and data sets were linear-predicted twice prior to transformation in order to improve resolution and remove truncation artifacts [18]. The NMR spectra were analyzed using FELIX version 2.3 (Biosym Technologies, San Diego, CA) on a Silicon Graphics Indigo workstation. The NH correlation cross-peaks were optimized by a Gaussian line shape function in order to obtain the line widths for the 1 H and 15 N dimensions [18]. 3. Results and discussion 3.1. Residues observed in 1 H-15 N HSQC spectra of [U-15 N]Pdx The cross-peaks colored with black in Fig. 2 represent the 1 H-15 N HSQC spectra for 0.50 mM oxidized [U-15 N]Pdx in 50 mM potassium phosphate (pH 7.4) containing 2-mercaptoethanol at 17³C. Assignment of the residues of the 15 N-labeled Pdx was achieved on the basis of the published chemical shifts reported by Lyons et al. [14]. As shown in Fig. 2, 78 connectivities between 15 N and the amide proton of the backbone appeared in the 1 H-15 N HSQC spectra of [U-15 N]Pdx as previously reported [14]. In addition to the resonances of the main chain, the NH correlations of the side chains have also been detected [14]; they include the amino groups of four
BBAPRO 35672 16-7-98 ^ Pagina 171 Cyaan Magenta
Geel
Zwart
172
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
Fig. 2. (A) Overlay of 15 N-1 H HSQC spectra of 0.50 mM [U-15 N]Pdx in the absence (black) and presence of PdR (red) in the mole ratio of 3:1. The sample of [U-15 N]Pdx/PdR mixture contained 0.76 mM [U-15 N]Pdx and 0.27 mM PdR. Spectra were recorded with 15 N decoupling during acquisition in 50 mM potassium phosphate (pH 7.4) at 17³C. The [U-15 N]Pdx alone sample contained 2-mercaptoethanol as protecting agent. Cross-peaks that are not observed at all by the addition of the redox partner are labeled with amino acid sequence number adjacent to the peak. (B) Overlay of 15 N-1 H HSQC spectra of 0.50 mM [U-15 N]Pdx in the absence (black) and presence of P450cam (red) in the mole ratio of 3:1. The mixture of [U-15 N]Pdx with P450cam contained 0.50 mM [U-15 N]Pdx and 0.17 mM P450cam. One mM d-camphor was contained in the mixture. Other conditions weres the same as in A.
BBAPRO 35672 16-7-98 ^ Pagina 172 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
173
Fig. 2. (continued)
asparagines (Asn-30, Asn-53, Asn-64, and Asn-81), NO H of arginines (Arg-12, Arg-13, and Arg-83), NN1 H of histidines (His-8 and His-49), and NO1 H of Trp-106. However, the correlations for Arg-13 and Leu-71, and NR H1 and NR H2 of Arg-83 could not be observed in the HSQC spectra at 500.13 MHz for 1 H
in this study (Fig. 2), which have been assigned using the three-dimensional NMR techniques [14]. Although the residues around the redox active center of Pdx were assumed to be essential for the binding with recognition for PdR and P450cam [9,12], the resonances of these residues, which are colored with
BBAPRO 35672 16-7-98 ^ Pagina 173 Cyaan Magenta
Geel
Zwart
174
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
yellow in Fig. 4, were missing due to paramagnetic broadening by the iron-sulfur cluster [13,14,19,20]. 3.2. Residues of Pdx a¡ected by the presence of PdR In order to estimate the binding sites on Pdx with PdR, 1 H-15 N HSQC spectra of [U-15 N]Pdx in the presence and absence of PdR were compared as illustrated in Fig. 2A. In Fig. 2A, the 1 H-15 N HSQC spectrum of 0.76 mM [U-15 N]Pdx in the presence of 0.27 mM PdR in a molar ratio of 3:1 excess of Pdx is presented with coloring in red. Due to the large molecular mass for the complexes (about 55.5 kDa), the rotational correlation time would increase, which induces line broadening for the NMR resonances. In addition to the line broadening by the increase of the molecular weight, some cross-peaks completely disappeared by addition of PdR. The line broadening of individual amino acid residues in the presence of 1:3 molar amount of PdR is summarized in Fig. 3A. In the presence of PdR, while most of the resonances result in a moderated change in line width of less than 5 Hz (Fig. 3A), the resonances from Val-28, Glu-72, Ile-88, and Gln-105, which are represented by a black bar in Fig. 3A, are missing in the spectrum, suggesting that these amino acid residues are involved in the binding site with PdR. It is of interest that the resonances of the indole ring and backbone NH of the C-terminal tryptophan residue were not seriously a¡ected by the addition of PdR (Fig. 2A and Fig. 3A), since the C-terminal region on Pdx has been proposed as one of the binding sites for PdR [9]. The contribution of the C-terminal Trp residue to the association of Pdx with PdR, therefore, would be moderate, although the adjacent amino acid residue, Glu-105, participates in interactions for the binding with PdR. As clearly indicated by the experiment of the chemical modi¢cation of Pdx, the acidic region nearby the active site, which includes Asp-58, Glu-65, Glu-67, Glu-72, and Glu-77, also plays an important role in the electron transfer from PdR to Pdx [7]. Our recent mutagenetic studies, however, revealed that most of the removals of the single negative charge listed above induced only slight inhibition to the electron-transfer reaction from PdR to Pdx [10,26] and the contribution of the acidic amino acid residues to the binding with PdR is still controversial. In the 1 H-
Fig. 3. The e¡ects of line widths of amide proton (vvXNH : upper) and nitrogen (vvXN : lower) of [U-15 N]Pdx in the presence of 1:3 molar amounts of the redox partners. (A) E¡ects in the presence of PdR. (B) E¡ects in the presence of P450cam. Black bars indicate that the corresponding NH cross-peak disappeared after the addition of the partner as shown in Fig. 2. The N-terminal residue, Ser-1, and proline residues are located at positions 61, 80, 92, and 102, which are not observed and are represented as open circles. The resonances which are not observed due to paramagnetic broadening, M24, Q25, from V36 to H49, and from L84 to Q87, are represented as closed circles.
BBAPRO 35672 16-7-98 ^ Pagina 174 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
175
Fig. 4. (A) Residues of Pdx a¡ected by the addition of PdR. (B) Residues of Pdx a¡ected by the addition of P450cam. The redox active center, 2Fe-2S cluster, is represented as a space ¢lling model. Those residues whose amide 1 H/15 N resonances are a¡ected by additions of the 1:3 and 1 molar equivalents of the redox partners are colored red and blue, respectively. The residues colored yellow are four proline and the N-terminal residues and the residues which are not observed due to paramagnetic broadening. 15
N HSQC spectra of [U-15 N]Pdx, these acidic residues on the negatively charged cluster, Asp-58, Glu65, Glu-67, and Glu-77, were not dramatically affected by the addition of PdR as shown in Fig. 2A and Fig. 3A. Glu-72 is the only residue whose signal width was speci¢cally increased and whose resonance was diminished. The contribution of the negatively charged patch on the G-helix to the association with PdR would be less important than we expected [7,9]. As we proposed in our previous papers [10,26], the severe inhibition of the electron transfer from PdR to Pdx by the chemical modi¢cation of the negatively charged amino acid residues would be attributable to the introduction of the bulky side chain, glycine ethyl ester, into the acidic cluster on the Pdx surface by that modi¢cation [7]. It would be the steric hindrance rather than the loss of the negative charges on the Ghelix that discourages the electron-transfer reaction from PdR to Pdx. It might be rather surprising that the association of Pdx with PdR is not dominated by electrostatic interactions, since most of the electron-transfer complex formation mediated by iron-sulfur proteins has been considered to be driven by electrostatic interactions [21]. Our thermodynamic analysis, however, in-
dicates that the hydrophobic interactions are crucial for the formation of the complex between PdR and Pdx rather than the electrostatic interactions [11]. The present NMR results also show that the two hydrophobic residues at positions 28 and 88 were a¡ected by the addition of PdR, supporting the contribution of the hydrophobic interaction to complex formation. The contribution of the hydrophobic interactions instead of the electrostatic interactions would be re£ected in the low cross-reactivity to PdR by other iron-sulfur proteins. The reduction of adorenodoxin (Adx), one of the [2Fe-2S]-type iron-sulfur proteins having a tertiary structure similar to that of Pdx [12], by PdR is very slow: the reduction rate of Pdx by PdR is 4.5 WM/min, whereas that of Adx by PdR is only 0.2 WM/min [7]. Since previous studies [22] have shown that the interactions responsible for the complex formation between Adx and its reductase (adorenodoxin reductase: AdR) are dominated by electrostatic interactions, not by hydrophobic interactions, such di¡erent kinds of interaction would be responsible for a di¡erence in the electron-transfer properties between the two iron-sulfur proteins, resulting in low cross-reactivity.
BBAPRO 35672 16-7-98 ^ Pagina 175 Cyaan Magenta
Geel
Zwart
176
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
Fig. 5. The amino acid sequences of Pdx and various ferredoxins. The sequence number is based on that of Pdx. The four cysteine residues for ligation of the 2Fe-2S cluster are underlined. In the sequence of Pdx, shaded residues denote the residues which disappeared by the addition of 1:3 molar of PdR, and boxed ones are for the residues which disappeared by the addition of 1:3 molar of P450cam. Shaded and/or boxed amino acid residues for adrenodoxins and ferredoxins represent the corresponding amino acid residues in the sequences of adrenodoxins and ferredoxins.
When an equimolar amount of PdR was added to [U-15 N]Pdx, additional line broadening was observed. These resonances arise from the backbone of Val-4, Val-17, Ala-18, Gly-20, Ser-22, Leu-23, Ala-27, Ser-29, Gly-31, Ile-35, Val-50, Tyr-51, Ile-68, Cys-73, Thr-75, Lys-79, Val-101, and Trp-106, and the side chain of Asn-30. These are located at both sides of the redox active site, the D- and G-helices, in the Pdx structure as mapped with blue in Fig. 4A. Although we cannot exclude the possibility that these additional amino acid residues participate in the binding site for the PdR/Pdx complex, the signal broadening widely distributed on the protein surface would not correspond to the speci¢c interactions for the formation of the electrontransfer complex of Pdx with PdR. 3.3. Residues of Pdx a¡ected by the presence of P450cam We also measured the 1 H-15 N HSQC spectra of the complex of [U-15 N]Pdx with P450cam with a 3:1 (0.50 mM Pdx:0.17 mM P450cam) excess of
Pdx and compared them with those of Pdx alone (Fig. 2B). The line widths of most resonances increased by the addition of P450cam as shown in Fig. 3B, due to the increase in molecular weight by complex formation. Several 1 H-15 N correlation resonances disappeared by the addition of P450cam (Figs. 2B and 3B) as found for the Pdx/PdR complex. The residues whose resonances disappeared are Leu-23, Val-28, Ser-29, Asn-30, Ile-32, Asp-34, Val50, Glu-67, Gln-105, and the indole NO1 H of Trp106. Many of these residues are located on and nearby the D-helix and the C-terminal region as depicted in the right and back sides, respectively, in Fig. 4B. The involvement of the amino acid residues on and nearby the D-helix and C-terminal region in the binding with P450cam was also suggested by the molecular dynamics simulation of the complex in which the residues at positions 25, 28, 29, 33^42, 44^47, 66, 70, 75, and 104^106 construct the binding interface of the Pdx/P450cam model complex [12]. It is should be noted here that some of the amino acid residues mentioned above are not highly conserved in the iron-sulfur protein family mediating
BBAPRO 35672 16-7-98 ^ Pagina 176 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
electron transfer and speci¢c for Pdx as listed in Fig. 5. Ser-29 is substituted for Gln or Glu in adrenodoxins and ferredoxins and Ile-32 is deleted for other iron-sulfur proteins. One of the hydrophilic amino acid residues, Gln-105, located at the penultimate position, is replaced by a hydrophobic alanine in porcine Adx and human ferredoxin. It is plausible that such low homology of these amino acid residues between Pdx and other iron-sulfur proteins reduces the cross-reactivity to P450cam by other iron-sulfur proteins [23]. By the replacement of Pdx with Adx in the P450cam monooxygenation cycle, the catalytic cycle is completely blocked [23], although the redox potential for Adx is almost similar to that for Pdx [24,25]. The highly speci¢c interactions between Pdx and P450cam, therefore, would depend on the amino acid residues located on and nearby the D-helix we listed here. Another characteristic feature for the complex formation between Pdx and P450cam is the spread binding site on the surface of Pdx. Comparison with the binding site for the Pdx/PdR complex indicates that the binding area for the Pdx/P450cam complex would be much larger, since the amino acid residues whose NMR resonances are a¡ected by the formation of the complex with P450cam are delocalized on the protein surface near the iron-sulfur cluster. The interactions participated by many amino acid residues also correspond to the high speci¢city for the complex formation between Pdx and P450cam. In the presence of an equimolar amount of P450cam, most of the NMR signals in the 15 N-1 H HSQC spectrum were broadened and disappeared. The residues a¡ected by the addition of equimolar amounts of P450cam are also represented with blue in Fig. 4B. The residues from which the resonances did not vanish in the spectra of Pdx:P450cam = 1:1 mixture are His-8, Asp-19, Asn-53, Asp-58, Val-60, Ala-62, Asn-64, Cys-73, Glu-77, Val-99, and Asp103. Some residues from which the resonances disappeared are found at the sites further away from the D-helix and the C-terminal region in the vicinity of the redox active center and distributed over the entire Pdx, indicating that Pdx has multiple non-speci¢c interaction sites for P450cam on the NMR time scale. These multiple and non-selective associations of Pdx with P450cam in the presence of an equimolar
177
amount of P450cam would imply the conformational gating of electron transfer in the Pdx/P450cam complex, in which the electron-transfer reaction is initiated by the formation of the complex having the speci¢c con¢guration for electron transfer after random collision of the two proteins. 4. Conclusions Using 15 N-labeled Pdx, we have proposed the amino acid residues on Pdx involved in the binding with the redox partner, Pdx or P450cam. In the binding with PdR, four amino acid residues were identi¢ed as the interaction sites. These amino acid residues include only one negatively charged acidic amino acid residue and two of them are hydrophobic residues, although previous results have suggested the key role of electrostatic interaction in the complex formation of Pdx and PdR. The minor contribution of electrostatic interaction would be re£ected in the low cross-reactivity with the Adx-AdR system, in which electrostatic interaction is dominant for complex formation. On the other hand, the binding site on Pdx for P450cam was constructed by several amino acid residues near the iron-sulfur cluster. The amino acid residues we identi¢ed as interaction sites for P450cam are not well conserved in the iron-sulfur electron-transfer protein family, which allow us to speculate that such low homology in the binding site leads to the speci¢c binding between Pdx and P450cam and low cross-reactivity with other ironsulfur proteins. Acknowledgements We are grateful to Professor Tadao Horiuchi and Dr Takanori Yasukochi (Kyushu University) for the gift of the plasmids encoding wild type Pdx and PdR and to Professor Yuzuru Ishimura (Keio University) for supplying the P450cam gene.
References [1] I.C. Gunsalus, G.C. Wagner, Methods Enzymol. 52 (1978) 166^188.
BBAPRO 35672 16-7-98 ^ Pagina 177 Cyaan Magenta
Geel
Zwart
178
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
[2] M.J. Hintz, D.M. Mock, L.L. Peterson, K.M. Tuttle, J.A. Peterson, J. Biol. Chem. 257 (1982) 14324^14332. [3] J.D. Lipscomb, S.G. Sligar, M.J. Namtvedt, I.C. Gunsalus, J. Biol. Chem. 251 (1976) 1116^1124. [4] P.S. Stayton, S.G. Sligar, Biochemistry 29 (1990) 7381^7386. [5] M.D. Davies, S.G. Sligar, Biochemistry 31 (1992) 11383^ 11389. [6] P.W. Roome Jr., J.C. Philley, J.A. Peterson, J. Biol. Chem. 258 (1983) 2593^2598. [7] L. Geren, J. Tuls, P. O'Brien, F. Millett, J.A. Peterson, J. Biol. Chem. 261 (1986) 15491^15495. [8] M. Unno, H. Shimada, Y. Toba, R. Makino, Y. Ishimura, J. Biol. Chem. 271 (1996) 17869^17874. [9] M. Holden, M. Mayhew, D. Bunk, A. Roitberg, V. Vilker, J. Biol. Chem. 272 (1997) 21720^21725. [10] M. Aoki, K. Ishimori, I. Morishima, Y. Wada, Inorg. Chim. Acta 272 (1998) 80^88. [11] M. Aoki, K. Ishimori, H. Fukada, K. Takahashi, I. Morishima, Biochim. Biophys. Acta 1384 (1998) 180^188. [12] T.C. Pochapsky, T.A. Lyons, S. Kazanis, T. Arakaki, G. Ratnaswamy, Biochimie 78 (1996) 723^733. [13] T.C. Pochapsky, X.M. Ye, G. Ratnaswamy, T.A. Lyons, Biochemistry 33 (1994) 6424^6432. [14] T.A. Lyons, G. Ratnaswamy, T.C. Pochapsky, Protein Sci. 5 (1996) 627^639.
[15] F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, K. Struhl, Short Protocols in Molecular Biology, 3rd edn., John Wiley, New York, 1995. [16] T.J. Norwood, J. Boyd, J.E. Heritage, N. So¡e, I.D. Campbell, J. Magn. Reson. 87 (1990) 488^501. [17] A.J. Shaka, P.B. Barker, R. Freeman, J. Magn. Reson. 64 (1985) 547^552. [18] J.C. Hoch, A.S. Stern, NMR Data Processing, John Wiley, New York, 1996. [19] X.M. Ye, T.C. Pochapsky, S.S. Pochapsky, Biochemistry 31 (1992) 1961^1968. [20] T.C. Pochapsky, G. Ratnaswamy, A. Petera, Biochemistry 33 (1994) 6433^6441. [21] A.R. De Pascalis, I. Jelesarov, F. Ackermann, W.H. Koppenol, M. Hiroasawa, D.B. Kna¡, H.R. Bosshard, Protein Sci. 2 (1993) 1126^1135. [22] L.M. Geren, P. O'Brien, J. Stonehuerner, F. Millett, J. Biol. Chem. 259 (1984) 2155^2160. [23] C.A. Tyson, J.D. Lipscomb, I.C. Gunsalus, J. Biol. Chem. 247 (1972) 5777^5784. [24] J.J. Huang, T. Kimura, Biochemistry 12 (1973) 406^409. [25] S.G. Sligar, I.C. Gunsalus, Proc. Natl. Acad. Sci. USA 73 (1976) 1078^1082. [26] M. Aoki, K. Ishimori, I. Morishima, Biochim. Biophys. Acta 1386 (1998) 157^167.
BBAPRO 35672 16-7-98 ^ Pagina 178 Cyaan Magenta
Geel
Zwart
NMR studies of putidaredoxin: associations of putidaredoxin with NADH-putidaredoxin reductase and cytochrome P450cam Masaaki Aoki, Koichiro Ishimori, Isao Morishima * Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, Kyoto 606-8501, Japan Received 20 January 1998; accepted 16 March 1998
Abstract To characterize the electron-transfer reaction in the P450cam monooxygenation system, the binding regions of putidaredoxin (Pdx) to NADH-putidaredoxin reductase (PdR) and P450cam were investigated using isotope-filtered NMR experiments in which uniformly 15 N-labeled Pdx ([U-15 N]Pdx) is mixed with unlabeled PdR and P450cam. By addition of PdR to Pdx, site specific signal broadening was observed for the N-H correlation peaks from Val-28, Glu-72, Ile-88, and Gln105. Although previous studies have suggested the contribution from acidic amino acid residues on the G-helix of Pdx to the binding with PdR, no site specific broadening was observed for the resonances from these residues except for Glu-72. The lesser contribution of electrostatic interactions to the Pdx/PdR complex formation was also suggested by our previous study (M. Aoki, K. Ishimori, H. Fukada, K. Takahashi, I. Morishima, Biochim. Biophys. Acta 1384 (1998) 180^188), which is in sharp contrast to the complex formation between adrenodoxin and adrenodoxin reductase. Upon the complex formation between Pdx and P450cam, the site specific NMR line broadening was observed for several amino acid residues distributed near the iron-sulfur cluster, corresponding to the large binding site in the complex formation with P450cam. Since some of the amino acid residues included in the binding site are not conserved for the electron-transfer iron-sulfur proteins such as ferredoxin and adrenodoxin, the interactions formed by these amino acid residues would be highly specific to the binding with P450cam, consistent with very low cross-reactivity to other iron-sulfur proteins in the P450cam monooxygenation system. ß 1998 Elsevier Science B.V. All rights reserved. Keywords: Putidaredoxin; Cytochrome P450cam; Nuclear magnetic resonance; Isotope labeling; Electron-transfer complex; Proteinprotein interaction
1. Introduction Putidaredoxin (Pdx) is a globular protein consisting of a single 106 residue polypeptide (molecular Abbreviations: AdR, adrenodoxin reductase; Adx, adrenodoxin; GARP, globally optimized alternating rectangular pulse; HSQC, heteronuclear single quantum correlation; PdR, NADHputidaredoxin reductase; Pdx, putidaredoxin * Corresponding author. Fax: +81 (75) 751-7611; E-mail: [email protected]
weight 11 500) and a [2Fe-2S] prosthetic group, and serves as a one-electron shuttle between NADH-putidaredoxin reductase (PdR; molecular weight 43 500) and cytochrome P450cam (molecular weight 45 000) [1,2]. Oxidized Pdx associates with reduced PdR and accepts one electron from the reductase as shown in Fig. 1. After the dissociation from oxidized PdR, reduced Pdx binds with P450cam and donates an electron to P450cam which catalyzes the oxygenation of d-camphor to 5-exo-hydroxycamphor. The electron-transfer process from Pdx to
0167-4838 / 98 / $19.00 ß 1998 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 4 8 3 8 ( 9 8 ) 0 0 0 9 1 - 0
BBAPRO 35672 16-7-98 ^ Pagina 168 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
P450cam is crucial for the P450cam oxygenation cycle, in that the electron transfer is rate limiting for the catalytic cycle [2]. It should also be noted here that Pdx is the only iron-sulfur protein which can reduce the ferrous oxygenated adduct for the P450cam reaction, while a number of reducing agents are competent to reduce ferric P450cam [3]. It was, therefore, concluded that Pdx acts not only as a redox shuttle, but also as a speci¢c e¡ector in the camphor hydroxylation [3]. Although the high speci¢c requirements of Pdx to facilitate the catalytic cycle are of great importance in the P450cam system, the mechanism for the binding of Pdx with the redox partners in the electron-transfer reaction has not yet been clear. Many investigators have tried to reveal the details of the speci¢c electron-transfer complexes, the Pdx/ PdR and Pdx/P450cam complexes, in the P450cam oxygenation system formed by Pdx [4^11]. Using a molecular dynamics-derived model for the Pdx/ P450cam complex, Pochapsky et al. [12] proposed that speci¢c salt bridges between negatively charged amino acid residues on Pdx (aspartic acids 34 and 38, and C-terminal carboxylate) and positively charged groups on P450cam (Arg-79, Arg-109 and Arg-112) promote the protein association. The site-directed mutagenesis studies of Pdx have also revealed that the loop structure constructing the active site (including Asp-34 and Asp-38) and C-terminal region of Pdx is a potential binding site for P450cam [5,9]. The C-terminal residue, Trp-106, on Pdx was also considered to be important for the complex formation between Pdx and PdR, since the substitution of the Trp residue highly inhibited the electron-transfer reaction [9]. In addition to the region near the iron-sulfur cluster, the `negatively charged patch' formed by acidic amino acid residues on the G-helix would be an alternative binding site for PdR or P450cam [7,13]. The chemical modi¢cation of Pdx has shown that neutralization of the negative charges of Asp-58, Glu-65, Glu-67, Glu-72, and Glu-77 severely discouraged the electron transfer from PdR to Pdx [7]. In the vicinity of the acidic region, Cys-73 of Pdx was also identi¢ed as one of the crucial residues for the interaction of Pdx with PdR [9]. However, our thermodynamic study on the associ-
169
ation of Pdx with PdR and P450cam has suggested that associations of Pdx with the redox partners are not strongly dominated by electrostatic interactions [11]. The replacements of the positively charged amino acid residues on P450cam with the corresponding neutral residues did not seriously a¡ect the electrontransfer reaction in the P450cam monooxygenase system, supporting the lesser contribution of the electrostatic interaction to the electron-transfer reaction [4]. Instead of the electrostatic interactions, we have pointed out that the hydrophobic interaction and/or van der Waals interaction play key roles in the formation of the electron-transfer complex in the P450cam monooxygenation system [10,11]. Thus, the interactions and binding sites responsible for the complex formation in the electron-transfer reaction are still controversial. In this study, to identify the amino acid residues on Pdx constructing the binding site with PdR or P450cam, we applied a heteronuclear multi-dimensional NMR technique to the PdR/Pdx and Pdx/ P450cam complexes. The large molecular weight of the complexes between Pdx and the redox partners would prevent us from analyzing the proton signals in one- and two-dimensional 1 H-NMR spectra. However, recent progress in protein expression and puri¢cation has enabled us to prepare enough amount of the isotope-labeled protein for the multinuclear multi-dimensional NMR spectroscopy, which is one of the potent spectroscopies for structural analysis of large molecular weight proteins and protein complexes. If one protein in a complex is labeled with a stable isotope having nuclear spins and the other unlabeled, we can preferentially observe the NMR signals from the isotope in one of the proteins and discuss the structural changes accompanied with the complex formation. For Pdx, Pochapsky and co-workers [12] have already synthesized 15 N-labeled Pdx and measured its two-dimensional 1 H-15 N-NMR spectra. They have assigned the 1 H-15 N correlation peaks for the 15 Nlabeled Pdx [14]. Although they have done some preliminary NMR measurements for the complex of Pdx with P450cam, and indicated that many residues of Pdx might be involved in the association with P450cam, the association site on Pdx for P450cam has not yet been clearly proposed and the complex
BBAPRO 35672 16-7-98 ^ Pagina 169 Cyaan Magenta
Geel
Zwart
170
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
with PdR has not been examined [12]. With an aim to the elucidation of the molecular mechanism in the electron-transfer reaction for the P450cam monooxygenation system, we have also prepared uniformly 15 N-labeled Pdx ([U-15 N]Pdx) and measured the 1 H15 N correlation spectra of [U-15 N]Pdx in the presence or absence of one of the redox partners, PdR or P450cam, to determine the binding regions and amino acid residues on Pdx crucial for the binding with PdR and P450cam. 2. Materials and methods 2.1. Expression of Pdx in
15
N-enriched media
Escherichia coli TG1 cell was transformed with a plasmid which contains a clone of the gene encoding Pdx under transcriptional control of a tac promoter [10,11], plated on 2UTY agar, and grown overnight at 28³C. Colonies were picked up and grown in 10 ml of LB media at 28³C. This culture was centrifuged, and the pellet resuspended in 250 Wl of deionized water. This suspension was added to one liter of M63 minimal medium containing (15 NH4 )2 SO4 as sole nitrogen source [15]. The culture was grown for approx. 30 h at 37³C, then harvested by centrifugation. The cell yield was approx. 2 g/l medium. The cell pellet was frozen at 380³C prior to puri¢cation. 2.2. Puri¢cation of [U-15 N]Pdx The cell pellet was resuspended to 50 mM potassium phosphate (pH 7.4) and lysed with lysozyme in the presence of sodium deoxycholate followed by addition of streptomycin sulfate (approx. 0.01 mg/ ml). Cell debris was removed by centrifugation (9.5 krpm for 20 min at 4³C), and the supernatant loaded onto a DE-52 anion exchange column (P 25 mmU6 cm) equilibrated with 50 mM potassium phosphate (pH 7.4) containing 10 mM 2-mercaptoethanol. After loading, the column was washed with the same bu¡er. Pdx was eluted by 50 mM potassium phosphate (pH 7.4) containing 200 mM KCl and 10 mM 2-mercaptoethanol. The Pdx-containing brown elution was concentrated using an Amicon
concentrator equipped with a YM3 membrane. The concentrated Pdx was loaded onto a Superdex 75 (P 10 mmU30 cm) mounted on a Pharmacia FPLC system equilibrated with 50 mM potassium phosphate (pH 7.4) containing 100 mM KCl and 10 mM 2-mercaptoethanol. Brown fractions were assayed for purity by spectrophotometry. The homogeneous protein has a 325^280 nm absorbance ratio of 0.68 [1]. Fractions for which this ratio is greater than 0.6 were combined and concentrated. Concentration of Pdx was calculated using extinction coe¤cients of 10.4 mM31 cm31 at 455 nm for its oxidized form [1]. Protein yields were approx. 800 Wl of 1 mM Pdx from 6 l of culture, i.e., approx. 2 mg/l medium. 2.3. Preparations of PdR and P450cam PdR and P450cam were also expressed in E. coli and puri¢ed by the procedures previously described [1,10,11]. A puri¢ed sample of PdR with a 280 nm/ 455 nm absorbance ratio of less than 9.5 was employed in this study, which corresponds to approx. 70% purity with respect to total protein [1]. The concentration of PdR was calculated using extinction coe¤cients of 8.5 mM31 cm31 at 480 nm [1]. The purity of the P450cam enzyme was estimated from its RZ (A391 /A280 ) value. Puri¢ed crystalline preparations exhibit an RZ value of 1.63 [1], and all of the P450cam samples used here had RZ values greater than 1.4, which correspond to its purity of approx. 90%. We used an extinction coe¤cient of 102.0 mM31 cm31 at 391 nm to estimate the concentration of the ferric camphor-bound cytochrome P450cam [1]. 2.4. NMR measurements All measurements were recorded on a Bruker AVANCE DRX500 spectrometer equipped with a Silicon Graphics Indy workstation. The spectrometer operates at 500.13 and 50.68 MHz for 1 H and 15 N, respectively. Bu¡ers used for NMR samples of oxidized Pdx were 50 mM potassium phosphate (pH 7.4) containing 10 mM 2-mercaptoethanol as protecting agent in H2 O/D2 O 90/10. The sample volume was approx. 350 Wl. In the Pdx/PdR and Pdx/ P450cam mixtures, the thiol agent was removed to
BBAPRO 35672 16-7-98 ^ Pagina 170 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
171
Fig. 1. Reaction cycle of cytochrome P450cam and the associated electron cascade from NADH via PdR and Pdx, where R-H is a substrate, d-camphor, R-OH is the product, 5-exo-hydroxycamphor, Pdxox and Pdxred are oxidized and reduced putidaredoxin, respectively, and PdRox and PdRred are oxidized and reduced NADH-putidaredoxin reductase, respectively.
avoid possible interference with the redox partners.1 One mM d-camphor was contained in the mixture of Pdx with P450cam. The [U-15 N]Pdx concentration was 0.76 mM for the mixtures with PdR. The concentrations of PdR for the Pdx:PdR = 3:1 and 1:1 mixtures were 0.27 and 0.76 mM, respectively. The [U-15 N]Pdx:P450cam = 3:1 mixture contained 0.50 mM Pdx and 0.17 mM P450cam, respectively. The 1:1 mixture of [U-15 N]Pdx with P450cam contained 0.50 mM of each protein. The sample temperature in the probe was adjusted to 17.0 þ 0.1³C and controlled by the temperature control unit of the spectrometer. The 1 H-15 N HSQC experiments were performed by the published sequence [16] with 15 N decoupling using GARP [17] during the 1 H acquisition period. A total recycling time of approx. 3 s was used in order to suppress a heat-up by decoupling. 1 H-15 N coupling evolution and refocusing decays were set using a value of 2.25 ms (1/4 JHN ). A 1 H spectral width of 8090.6 Hz was used, and the spectral width for 15 N dimension was 8109 Hz. The 1 H dimension was referred to a water signal at 4.7 ppm at 17³C, and the 15 N carrier frequency was placed at 120 ppm. Quadrature detection in the 15 N dimension was obtained using time-proportional phase incrementation 1 The thiol agent, 2-mercaptoethanol, used as a preservative might attach covalently to a solvent exposed cysteine, Cys-73, of Pdx [13]. Although structural changes of Pdx caused by the removal of the thiol agent cannot be excluded, the resonance from Cys-73 covalently bound to the thiol was not drastically a¡ected by the removal of the thiol agent, implying that the e¡ect of the removal on the resonances of [U-15 N]Pdx is not so large.
(TPPI). Free induction decays were acquired with 2048 complex data points, and 512 increments in the 15 N dimension. The 1 H dimension was multiplied by a sine-bell prior to Fourier transformation. A squared sine-bell was applied to interferograms in the 15 N dimension, and data sets were linear-predicted twice prior to transformation in order to improve resolution and remove truncation artifacts [18]. The NMR spectra were analyzed using FELIX version 2.3 (Biosym Technologies, San Diego, CA) on a Silicon Graphics Indigo workstation. The NH correlation cross-peaks were optimized by a Gaussian line shape function in order to obtain the line widths for the 1 H and 15 N dimensions [18]. 3. Results and discussion 3.1. Residues observed in 1 H-15 N HSQC spectra of [U-15 N]Pdx The cross-peaks colored with black in Fig. 2 represent the 1 H-15 N HSQC spectra for 0.50 mM oxidized [U-15 N]Pdx in 50 mM potassium phosphate (pH 7.4) containing 2-mercaptoethanol at 17³C. Assignment of the residues of the 15 N-labeled Pdx was achieved on the basis of the published chemical shifts reported by Lyons et al. [14]. As shown in Fig. 2, 78 connectivities between 15 N and the amide proton of the backbone appeared in the 1 H-15 N HSQC spectra of [U-15 N]Pdx as previously reported [14]. In addition to the resonances of the main chain, the NH correlations of the side chains have also been detected [14]; they include the amino groups of four
BBAPRO 35672 16-7-98 ^ Pagina 171 Cyaan Magenta
Geel
Zwart
172
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
Fig. 2. (A) Overlay of 15 N-1 H HSQC spectra of 0.50 mM [U-15 N]Pdx in the absence (black) and presence of PdR (red) in the mole ratio of 3:1. The sample of [U-15 N]Pdx/PdR mixture contained 0.76 mM [U-15 N]Pdx and 0.27 mM PdR. Spectra were recorded with 15 N decoupling during acquisition in 50 mM potassium phosphate (pH 7.4) at 17³C. The [U-15 N]Pdx alone sample contained 2-mercaptoethanol as protecting agent. Cross-peaks that are not observed at all by the addition of the redox partner are labeled with amino acid sequence number adjacent to the peak. (B) Overlay of 15 N-1 H HSQC spectra of 0.50 mM [U-15 N]Pdx in the absence (black) and presence of P450cam (red) in the mole ratio of 3:1. The mixture of [U-15 N]Pdx with P450cam contained 0.50 mM [U-15 N]Pdx and 0.17 mM P450cam. One mM d-camphor was contained in the mixture. Other conditions weres the same as in A.
BBAPRO 35672 16-7-98 ^ Pagina 172 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
173
Fig. 2. (continued)
asparagines (Asn-30, Asn-53, Asn-64, and Asn-81), NO H of arginines (Arg-12, Arg-13, and Arg-83), NN1 H of histidines (His-8 and His-49), and NO1 H of Trp-106. However, the correlations for Arg-13 and Leu-71, and NR H1 and NR H2 of Arg-83 could not be observed in the HSQC spectra at 500.13 MHz for 1 H
in this study (Fig. 2), which have been assigned using the three-dimensional NMR techniques [14]. Although the residues around the redox active center of Pdx were assumed to be essential for the binding with recognition for PdR and P450cam [9,12], the resonances of these residues, which are colored with
BBAPRO 35672 16-7-98 ^ Pagina 173 Cyaan Magenta
Geel
Zwart
174
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
yellow in Fig. 4, were missing due to paramagnetic broadening by the iron-sulfur cluster [13,14,19,20]. 3.2. Residues of Pdx a¡ected by the presence of PdR In order to estimate the binding sites on Pdx with PdR, 1 H-15 N HSQC spectra of [U-15 N]Pdx in the presence and absence of PdR were compared as illustrated in Fig. 2A. In Fig. 2A, the 1 H-15 N HSQC spectrum of 0.76 mM [U-15 N]Pdx in the presence of 0.27 mM PdR in a molar ratio of 3:1 excess of Pdx is presented with coloring in red. Due to the large molecular mass for the complexes (about 55.5 kDa), the rotational correlation time would increase, which induces line broadening for the NMR resonances. In addition to the line broadening by the increase of the molecular weight, some cross-peaks completely disappeared by addition of PdR. The line broadening of individual amino acid residues in the presence of 1:3 molar amount of PdR is summarized in Fig. 3A. In the presence of PdR, while most of the resonances result in a moderated change in line width of less than 5 Hz (Fig. 3A), the resonances from Val-28, Glu-72, Ile-88, and Gln-105, which are represented by a black bar in Fig. 3A, are missing in the spectrum, suggesting that these amino acid residues are involved in the binding site with PdR. It is of interest that the resonances of the indole ring and backbone NH of the C-terminal tryptophan residue were not seriously a¡ected by the addition of PdR (Fig. 2A and Fig. 3A), since the C-terminal region on Pdx has been proposed as one of the binding sites for PdR [9]. The contribution of the C-terminal Trp residue to the association of Pdx with PdR, therefore, would be moderate, although the adjacent amino acid residue, Glu-105, participates in interactions for the binding with PdR. As clearly indicated by the experiment of the chemical modi¢cation of Pdx, the acidic region nearby the active site, which includes Asp-58, Glu-65, Glu-67, Glu-72, and Glu-77, also plays an important role in the electron transfer from PdR to Pdx [7]. Our recent mutagenetic studies, however, revealed that most of the removals of the single negative charge listed above induced only slight inhibition to the electron-transfer reaction from PdR to Pdx [10,26] and the contribution of the acidic amino acid residues to the binding with PdR is still controversial. In the 1 H-
Fig. 3. The e¡ects of line widths of amide proton (vvXNH : upper) and nitrogen (vvXN : lower) of [U-15 N]Pdx in the presence of 1:3 molar amounts of the redox partners. (A) E¡ects in the presence of PdR. (B) E¡ects in the presence of P450cam. Black bars indicate that the corresponding NH cross-peak disappeared after the addition of the partner as shown in Fig. 2. The N-terminal residue, Ser-1, and proline residues are located at positions 61, 80, 92, and 102, which are not observed and are represented as open circles. The resonances which are not observed due to paramagnetic broadening, M24, Q25, from V36 to H49, and from L84 to Q87, are represented as closed circles.
BBAPRO 35672 16-7-98 ^ Pagina 174 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
175
Fig. 4. (A) Residues of Pdx a¡ected by the addition of PdR. (B) Residues of Pdx a¡ected by the addition of P450cam. The redox active center, 2Fe-2S cluster, is represented as a space ¢lling model. Those residues whose amide 1 H/15 N resonances are a¡ected by additions of the 1:3 and 1 molar equivalents of the redox partners are colored red and blue, respectively. The residues colored yellow are four proline and the N-terminal residues and the residues which are not observed due to paramagnetic broadening. 15
N HSQC spectra of [U-15 N]Pdx, these acidic residues on the negatively charged cluster, Asp-58, Glu65, Glu-67, and Glu-77, were not dramatically affected by the addition of PdR as shown in Fig. 2A and Fig. 3A. Glu-72 is the only residue whose signal width was speci¢cally increased and whose resonance was diminished. The contribution of the negatively charged patch on the G-helix to the association with PdR would be less important than we expected [7,9]. As we proposed in our previous papers [10,26], the severe inhibition of the electron transfer from PdR to Pdx by the chemical modi¢cation of the negatively charged amino acid residues would be attributable to the introduction of the bulky side chain, glycine ethyl ester, into the acidic cluster on the Pdx surface by that modi¢cation [7]. It would be the steric hindrance rather than the loss of the negative charges on the Ghelix that discourages the electron-transfer reaction from PdR to Pdx. It might be rather surprising that the association of Pdx with PdR is not dominated by electrostatic interactions, since most of the electron-transfer complex formation mediated by iron-sulfur proteins has been considered to be driven by electrostatic interactions [21]. Our thermodynamic analysis, however, in-
dicates that the hydrophobic interactions are crucial for the formation of the complex between PdR and Pdx rather than the electrostatic interactions [11]. The present NMR results also show that the two hydrophobic residues at positions 28 and 88 were a¡ected by the addition of PdR, supporting the contribution of the hydrophobic interaction to complex formation. The contribution of the hydrophobic interactions instead of the electrostatic interactions would be re£ected in the low cross-reactivity to PdR by other iron-sulfur proteins. The reduction of adorenodoxin (Adx), one of the [2Fe-2S]-type iron-sulfur proteins having a tertiary structure similar to that of Pdx [12], by PdR is very slow: the reduction rate of Pdx by PdR is 4.5 WM/min, whereas that of Adx by PdR is only 0.2 WM/min [7]. Since previous studies [22] have shown that the interactions responsible for the complex formation between Adx and its reductase (adorenodoxin reductase: AdR) are dominated by electrostatic interactions, not by hydrophobic interactions, such di¡erent kinds of interaction would be responsible for a di¡erence in the electron-transfer properties between the two iron-sulfur proteins, resulting in low cross-reactivity.
BBAPRO 35672 16-7-98 ^ Pagina 175 Cyaan Magenta
Geel
Zwart
176
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
Fig. 5. The amino acid sequences of Pdx and various ferredoxins. The sequence number is based on that of Pdx. The four cysteine residues for ligation of the 2Fe-2S cluster are underlined. In the sequence of Pdx, shaded residues denote the residues which disappeared by the addition of 1:3 molar of PdR, and boxed ones are for the residues which disappeared by the addition of 1:3 molar of P450cam. Shaded and/or boxed amino acid residues for adrenodoxins and ferredoxins represent the corresponding amino acid residues in the sequences of adrenodoxins and ferredoxins.
When an equimolar amount of PdR was added to [U-15 N]Pdx, additional line broadening was observed. These resonances arise from the backbone of Val-4, Val-17, Ala-18, Gly-20, Ser-22, Leu-23, Ala-27, Ser-29, Gly-31, Ile-35, Val-50, Tyr-51, Ile-68, Cys-73, Thr-75, Lys-79, Val-101, and Trp-106, and the side chain of Asn-30. These are located at both sides of the redox active site, the D- and G-helices, in the Pdx structure as mapped with blue in Fig. 4A. Although we cannot exclude the possibility that these additional amino acid residues participate in the binding site for the PdR/Pdx complex, the signal broadening widely distributed on the protein surface would not correspond to the speci¢c interactions for the formation of the electrontransfer complex of Pdx with PdR. 3.3. Residues of Pdx a¡ected by the presence of P450cam We also measured the 1 H-15 N HSQC spectra of the complex of [U-15 N]Pdx with P450cam with a 3:1 (0.50 mM Pdx:0.17 mM P450cam) excess of
Pdx and compared them with those of Pdx alone (Fig. 2B). The line widths of most resonances increased by the addition of P450cam as shown in Fig. 3B, due to the increase in molecular weight by complex formation. Several 1 H-15 N correlation resonances disappeared by the addition of P450cam (Figs. 2B and 3B) as found for the Pdx/PdR complex. The residues whose resonances disappeared are Leu-23, Val-28, Ser-29, Asn-30, Ile-32, Asp-34, Val50, Glu-67, Gln-105, and the indole NO1 H of Trp106. Many of these residues are located on and nearby the D-helix and the C-terminal region as depicted in the right and back sides, respectively, in Fig. 4B. The involvement of the amino acid residues on and nearby the D-helix and C-terminal region in the binding with P450cam was also suggested by the molecular dynamics simulation of the complex in which the residues at positions 25, 28, 29, 33^42, 44^47, 66, 70, 75, and 104^106 construct the binding interface of the Pdx/P450cam model complex [12]. It is should be noted here that some of the amino acid residues mentioned above are not highly conserved in the iron-sulfur protein family mediating
BBAPRO 35672 16-7-98 ^ Pagina 176 Cyaan Magenta
Geel
Zwart
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
electron transfer and speci¢c for Pdx as listed in Fig. 5. Ser-29 is substituted for Gln or Glu in adrenodoxins and ferredoxins and Ile-32 is deleted for other iron-sulfur proteins. One of the hydrophilic amino acid residues, Gln-105, located at the penultimate position, is replaced by a hydrophobic alanine in porcine Adx and human ferredoxin. It is plausible that such low homology of these amino acid residues between Pdx and other iron-sulfur proteins reduces the cross-reactivity to P450cam by other iron-sulfur proteins [23]. By the replacement of Pdx with Adx in the P450cam monooxygenation cycle, the catalytic cycle is completely blocked [23], although the redox potential for Adx is almost similar to that for Pdx [24,25]. The highly speci¢c interactions between Pdx and P450cam, therefore, would depend on the amino acid residues located on and nearby the D-helix we listed here. Another characteristic feature for the complex formation between Pdx and P450cam is the spread binding site on the surface of Pdx. Comparison with the binding site for the Pdx/PdR complex indicates that the binding area for the Pdx/P450cam complex would be much larger, since the amino acid residues whose NMR resonances are a¡ected by the formation of the complex with P450cam are delocalized on the protein surface near the iron-sulfur cluster. The interactions participated by many amino acid residues also correspond to the high speci¢city for the complex formation between Pdx and P450cam. In the presence of an equimolar amount of P450cam, most of the NMR signals in the 15 N-1 H HSQC spectrum were broadened and disappeared. The residues a¡ected by the addition of equimolar amounts of P450cam are also represented with blue in Fig. 4B. The residues from which the resonances did not vanish in the spectra of Pdx:P450cam = 1:1 mixture are His-8, Asp-19, Asn-53, Asp-58, Val-60, Ala-62, Asn-64, Cys-73, Glu-77, Val-99, and Asp103. Some residues from which the resonances disappeared are found at the sites further away from the D-helix and the C-terminal region in the vicinity of the redox active center and distributed over the entire Pdx, indicating that Pdx has multiple non-speci¢c interaction sites for P450cam on the NMR time scale. These multiple and non-selective associations of Pdx with P450cam in the presence of an equimolar
177
amount of P450cam would imply the conformational gating of electron transfer in the Pdx/P450cam complex, in which the electron-transfer reaction is initiated by the formation of the complex having the speci¢c con¢guration for electron transfer after random collision of the two proteins. 4. Conclusions Using 15 N-labeled Pdx, we have proposed the amino acid residues on Pdx involved in the binding with the redox partner, Pdx or P450cam. In the binding with PdR, four amino acid residues were identi¢ed as the interaction sites. These amino acid residues include only one negatively charged acidic amino acid residue and two of them are hydrophobic residues, although previous results have suggested the key role of electrostatic interaction in the complex formation of Pdx and PdR. The minor contribution of electrostatic interaction would be re£ected in the low cross-reactivity with the Adx-AdR system, in which electrostatic interaction is dominant for complex formation. On the other hand, the binding site on Pdx for P450cam was constructed by several amino acid residues near the iron-sulfur cluster. The amino acid residues we identi¢ed as interaction sites for P450cam are not well conserved in the iron-sulfur electron-transfer protein family, which allow us to speculate that such low homology in the binding site leads to the speci¢c binding between Pdx and P450cam and low cross-reactivity with other ironsulfur proteins. Acknowledgements We are grateful to Professor Tadao Horiuchi and Dr Takanori Yasukochi (Kyushu University) for the gift of the plasmids encoding wild type Pdx and PdR and to Professor Yuzuru Ishimura (Keio University) for supplying the P450cam gene.
References [1] I.C. Gunsalus, G.C. Wagner, Methods Enzymol. 52 (1978) 166^188.
BBAPRO 35672 16-7-98 ^ Pagina 177 Cyaan Magenta
Geel
Zwart
178
M. Aoki et al. / Biochimica et Biophysica Acta 1386 (1998) 168^178
[2] M.J. Hintz, D.M. Mock, L.L. Peterson, K.M. Tuttle, J.A. Peterson, J. Biol. Chem. 257 (1982) 14324^14332. [3] J.D. Lipscomb, S.G. Sligar, M.J. Namtvedt, I.C. Gunsalus, J. Biol. Chem. 251 (1976) 1116^1124. [4] P.S. Stayton, S.G. Sligar, Biochemistry 29 (1990) 7381^7386. [5] M.D. Davies, S.G. Sligar, Biochemistry 31 (1992) 11383^ 11389. [6] P.W. Roome Jr., J.C. Philley, J.A. Peterson, J. Biol. Chem. 258 (1983) 2593^2598. [7] L. Geren, J. Tuls, P. O'Brien, F. Millett, J.A. Peterson, J. Biol. Chem. 261 (1986) 15491^15495. [8] M. Unno, H. Shimada, Y. Toba, R. Makino, Y. Ishimura, J. Biol. Chem. 271 (1996) 17869^17874. [9] M. Holden, M. Mayhew, D. Bunk, A. Roitberg, V. Vilker, J. Biol. Chem. 272 (1997) 21720^21725. [10] M. Aoki, K. Ishimori, I. Morishima, Y. Wada, Inorg. Chim. Acta 272 (1998) 80^88. [11] M. Aoki, K. Ishimori, H. Fukada, K. Takahashi, I. Morishima, Biochim. Biophys. Acta 1384 (1998) 180^188. [12] T.C. Pochapsky, T.A. Lyons, S. Kazanis, T. Arakaki, G. Ratnaswamy, Biochimie 78 (1996) 723^733. [13] T.C. Pochapsky, X.M. Ye, G. Ratnaswamy, T.A. Lyons, Biochemistry 33 (1994) 6424^6432. [14] T.A. Lyons, G. Ratnaswamy, T.C. Pochapsky, Protein Sci. 5 (1996) 627^639.
[15] F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith, K. Struhl, Short Protocols in Molecular Biology, 3rd edn., John Wiley, New York, 1995. [16] T.J. Norwood, J. Boyd, J.E. Heritage, N. So¡e, I.D. Campbell, J. Magn. Reson. 87 (1990) 488^501. [17] A.J. Shaka, P.B. Barker, R. Freeman, J. Magn. Reson. 64 (1985) 547^552. [18] J.C. Hoch, A.S. Stern, NMR Data Processing, John Wiley, New York, 1996. [19] X.M. Ye, T.C. Pochapsky, S.S. Pochapsky, Biochemistry 31 (1992) 1961^1968. [20] T.C. Pochapsky, G. Ratnaswamy, A. Petera, Biochemistry 33 (1994) 6433^6441. [21] A.R. De Pascalis, I. Jelesarov, F. Ackermann, W.H. Koppenol, M. Hiroasawa, D.B. Kna¡, H.R. Bosshard, Protein Sci. 2 (1993) 1126^1135. [22] L.M. Geren, P. O'Brien, J. Stonehuerner, F. Millett, J. Biol. Chem. 259 (1984) 2155^2160. [23] C.A. Tyson, J.D. Lipscomb, I.C. Gunsalus, J. Biol. Chem. 247 (1972) 5777^5784. [24] J.J. Huang, T. Kimura, Biochemistry 12 (1973) 406^409. [25] S.G. Sligar, I.C. Gunsalus, Proc. Natl. Acad. Sci. USA 73 (1976) 1078^1082. [26] M. Aoki, K. Ishimori, I. Morishima, Biochim. Biophys. Acta 1386 (1998) 157^167.
BBAPRO 35672 16-7-98 ^ Pagina 178 Cyaan Magenta
Geel
Zwart