Characterization of then-Alkane and Fatty Acid Hydroxylating Cytochrome P450 Forms 52A3 and 52A4

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS

Vol. 328, No. 2, April 15, pp. 245–254, 1996 Article No. 0170

Characterization of the n-Alkane and Fatty Acid Hydroxylating Cytochrome P450 Forms 52A3 and 52A41 Ulrich Scheller,2 Thomas Zimmer, Eva Ka¨rgel, and Wolf-Hagen Schunck Max-Delbru¨ck-Center for Molecular Medicine, Robert-Ro¨ssle-Straße. 10, D-13122 Berlin-Buch, Federal Republic of Germany

Received October 19, 1995; and in revised form January 29, 1996

cies and preferences for either n-alkanes or fatty acids. Two enzymes, P450 52A3 (P450Cm1) and 52A4 (P450Cm2), the genes of which belong to the CYP52 multigene family occuring in the alkane-assimilating yeast Candida maltosa, have been characterized biochemically and compared in terms of their substrate specificities. For this purpose, both the P450 proteins and the corresponding C. maltosa NADPH-cytochrome P450 reductase were separately produced by expressing their cDNAs in Saccharomyces cerevisiae, purified, and reconstituted to active monooxygenase systems. Starting from microsomal fractions with a specific content of 0.75 nmol P450Cm1, 0.34 nmol P450Cm2, and 10.5 units reductase per milligram of protein, respectively, each individual recombinant protein was purified to homogeneity. P450 substrate difference spectra indicated strong type I spectral changes and high-affinity binding of n-hexadecane (Ks Å 26 mM) and n-octadecane (Ks Å 27 mM) to P450Cm1, whereas preferential binding to P450Cm2 was observed using lauric acid (Ks Å 127 mM) and myristic acid (Ks Å 134 mM) as substrates. These substrate selectivities were further substantiated by kinetic parameters, determined for n-alkane and fatty acid hydroxylation in a reconstituted system, which was composed of the purified components and phospholipid, as well as in microsomes obtained after coexpressing each of the P450 proteins with the reductase. The highest catalytic activities within the reconstituted system were measured for n-hexadecane hydroxylation to 1-hexadecanol by P450Cm1 (Vmax Å 27 mM 1 min01, Km Å 54 mM) and oxidation of lauric acid to 16-hydroxylauric acid by P450Cm2 (Vmax Å 30 mM 1 min01, Km Å 61 mM). We conclude that P450Cm1 and P450Cm2 exhibit overlapping but distinct substrate specificities due to different chain-length dependen-

1 This work was supported by Grant 0310257A from the Bundesministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie. 2 To whom correspondence should be addressed. Fax: /49-30-94063760.

q 1996 Academic Press, Inc.

Key Words: P450 52A3; P450 52A4; NADPH-cytochrome P450 reductase; yeast; high-level expression; purification; substrate binding; reconstitution; catalytic activity; kinetics; alkane and fatty acid hydroxylation.

Cytochromes P450 constitute a superfamily of ubiquitous heme-thiolate proteins that are involved in the oxidative metabolism of a wide variety of endogenous and xenobiotic compounds (1). According to a recent update of P450 gene nomenclature, nearly 500 genes of the P450 superfamily have been identified and classified into 65 gene families (2). Among the P450 families that exist in microorganisms, the CYP52 family is unique considering its remarkable extent, approaching that of some mammalian P450 families involved in drug metabolism. Members of this family have been detected in Candida tropicalis (3–5), Candida maltosa (6–11), and more recently also in Candida apicola (12). The presently available 21 sequences constitute five subfamilies (CYP52A-E) and a given species, like C. maltosa, may contain at least eight different P450 forms, not counting the allelic variants present (11). Interestingly, some of the P450 genes were found to be tandemly arranged, suggesting that the CYP52 family may have evolved from an ancestral gene by gene duplications and divergent evolution (10). Most of the CYP52 genes were shown to be inducible by long-chain aliphatic hydrocarbons such as n-alkanes, alkenes, fatty alcohols, and fatty acids. The degree of induction was found to vary from gene to gene and to depend on the chemical structure of the inducer (5, 9, 11). So far, two different enzymatic functions can be attributed to the alkane-inducible P450 systems: the terminal hydroxylation of n-alkanes, which represents 245

0003-9861/96 $18.00 Copyright q 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

/ 6b15$$9382

03-12-96 18:50:45

arca

AP: Archives

246

SCHELLER ET AL.

the first and rate-limiting step of the n-alkane degradation pathway, and the v-hydroxylation of fatty acids (13–15). However, information is lacking regarding the actual enzymatic properties of the individual P450 forms, which is needed to understand structure–function relationships within the CYP52 family. All these P450 forms are probably integral membrane proteins of the endoplasmic reticulum (16) and require electron transfer from NADPH supported by an FAD and FMN containing NADPH-cytochrome P450 reductase to catalyze monooxygenase reactions (17). As shown for some of the P450 genes from C. tropicalis (5) and C. maltosa (8), heterologous expression in Saccharomyces cerevisiae allows individual P450 forms to be produced in a functionally active state and thus provides a suitable basis for their characterization. The aim of the present study was to obtain two of the C. maltosa P450 proteins, P450Cm1 (product of a gene classified to CYP52 A3) and P450Cm2 (CYP52 A4), showing an amino acid sequence identity of 57.5%, in a highly purified state in order to compare in detail their spectral properties and substrate specificities. P450Cm1 represents the major P450 form during growth of C. maltosa on long-chain n-alkanes (18). Its primary structure was found to be nearly identical to that of P450Alk1A, the major alkane-inducible P450 of another C. maltosa strain (6). Disruption of the P450Alk1A gene did not abolish, however, the ability of alkane utilization, demonstrating that it can be functionally replaced by other P450 forms (9). P450Cm2 may be one of the candidates. This protein became available only recently after heterologous expression in S. cerevisiae and was suggested to prefer fatty acid substrates. This assumption was based on the relatively high activity of the microsomal protein toward lauric acid (8, 19). We describe here the purification of P450Cm1, P450Cm2, and the corresponding reductase after highlevel heterologous expression in S. cerevisiae and the efficient reconstitution of active monooxygenase systems based on these components. The results show that the two P450 forms differ significantly in their abilities to bind and hydroxylate n-alkanes and fatty acids. Moreover, the P450 forms were found to have distinct selectivities regarding the chain-length of these compounds leading to partially overlapping substrate specificities. MATERIALS AND METHODS Chemicals. Unlabeled n-alkanes and fatty acids having a purity of ú98% in gas chromatography were obtained from Fluka (NeuUlm, Germany). 1-14C-labeled n-alkanes (dodecane, hexadecane, and octadecane with specific activities of 3.7, 57, and 54 mCi/mmol, respectively) and fatty acids (lauric acid, palmitic acid, and stearic acid with specific radioactivities of 57, 54.0, and 51 mCi/mmol, respec-

/ 6b15$$9382

03-12-96 18:50:45

arca

tively) were purchased either from Sigma (Deisenhofen, Germany) or Amersham-Buchler (Braunschweig, Germany). v-Amino-n-octylSepharose 4B was prepared by the method of Cuatrecasas (20). Fastflow hydroxyapatite was purchased from Calbiochem (Bad Soden, Germany), DEAE-Toyopearl 650M from Tosoh (Tokyo, Japan), 2*,5*ADP-Sepharose from Pharmacia (Uppsala, Sweden). Glass backed silica gel plates (20 1 20 cm, 0.25 mm thick) for TLC3 were obtained from Merck (Darmstadt, Germany). NADPH, glucose-6-phosphate, glucose-6-phosphate dehydrogenase, n-dodecylmaltoside, and CHAPS were purchased from Boehringer Mannheim (Germany) and Emulgen 911 from Kao-Atlas (Tokyo, Japan). The nonionic detergent Pra¨wozell WON-100 used for purification of the reductase and measurement of the alkane hydroxylase activities is equivalent to Emulgen 913, Kao-Atlas (Tokyo, Japan). All other chemicals were obtained locally and were of analytical grade or better. Strains, cell cultivation, and heterologous protein expression. The yeast strain Saccharomyces cerevisiae GRF18 (MATa, his 3-11, his 3-15, leu 2-3, leu 2-112, canr) and the expression vector YEp51 were kindly provided by D. Sanglard (3) and J. R. Broach (21), respectively. The S. cerevisiae tranformants used in this work are listed in Table I. To express the authentic P450Cm2 protein in S. cerevisiae, both CTG triplets of its cDNA corresponding to amino acid positions 99 and 389 were exchanged by the serine triplet TCT using recombinant PCR (22). Yeast transformants GRF18/YEp51Cm1, GRF18/YEp51Cm2, and GRF18/YEp51-R were cultivated in a 20-liter bioreactor and expression of the recombinant proteins was induced by addition of galactose as described previously (23, 24). S. cerevisiae transformants GRF18/ YEp51Cm1-R and GRF18/YEp51Cm2-R for coexpression of P450 and NADPH-cytochrome P450 reductase were cultivated under semianaerobic conditions at 287C in 100 ml yeast minimal medium using 500-ml shaking flasks (19). The yeast strain C. maltosa EH15 (Institute of Technical Chemistry, Leipzig, Germany) was grown on a mixture of C11 –C19 n-alkanes to produce the wild-type P450Cm1 as described previously (18). Microsome preparation. The harvested cells (about 300 g wet weight) were washed twice with ice-cold 50 mM Tris/HCl buffer, pH 7.4, and suspended in the same buffer containing 40% glycerol, 5 mM DTT, and 10 mM EDTA (for preparation of microsomes containing the reductase additionally containing 0.25 mM PMSF and 0.5 mM FAD/ FMN) to give a final volume of 400 ml. Cells were then disrupted mechanically in a cooled Dyno-mill (W.A. Bachofen Maschinenfabrik, Basel, Switzerland) with glass beads (0.5–0.8 mm diameter). The disrupted cells were diluted with 10 vol of 50 mM Tris/HCl buffer, pH 7.4, containing 0.4 M sorbitol and centrifuged at 5,000g for 5 min. The supernatant was then centrifuged for 15 min at 10,000g and the microsomal fractions containing P450 and/or reductase were pelleted from the supernatant after addition of a 2 M solution of CaCl2 to a final concentration of 20 mM and subsequent centrifugation for 20 min at 20,000g. The microsomes were then resuspended in the Tris buffer containing 0.4 M sorbitol to give final P450 and reductase concentrations of about 15 nmol and 300 units per milliliter, respectively. Purification of P450Cm1. P450Cm1 was purified either from C. maltosa EH15 cells (18) or after heterologous expression in S. cerevisiae GRF18 (23) using described methods. Purification of P450Cm2. S. cerevisiae microsomes containing the heterologously expressed P450Cm2 were diluted to a protein concentration of 10 mg/ml in 250 mM potassium phosphate buffer, pH 7.3, containing 20% glycerol, 0.5 mM DTT, and 1 mM EDTA (buffer A). Solubilization was initiated by addition of a 10% (w/v) sodium cholate

3 Abbreviations used: TLC, thin-layer chromatography; CHAPS, 3-[(3-cholamido-propyl)dimethylammonio]-1-propanesulfonic acid; DTT, dithiothreitol; PMSF, phenylmethanesulfonyl fluoride; recombinant PCR, recombinant polymerase chain reaction.

AP: Archives

CHARACTERIZATION OF P450 52A3 AND 52A4 solution to give a final concentration of 0.8% w/v and was carried out for 30 min at 47C. After ultracentrifugation (100,000g, 60 min, 47C), the supernatant was loaded onto a v-amino-n-octyl-Sepharose 4B column, (P450:gel ratio of 3 nmol/1 ml packed gel) which had been equilibrated with buffer A containing 0.8% (w/v) sodium cholate. After binding, the column was washed twice with 3 vol of buffer A, containing 0.1 and 0.5% sodium cholate, respectively. P450Cm2 was then eluted with 0.5% (v/v) Tween 20 in buffer A containing 0.5% sodium cholate. The pooled P450 fractions were dialyzed overnight against 10 mM potassium phosphate buffer, pH 7.3, containing 0.3% sodium cholate, 20% glycerol, 0.5 mM DTT, and 1 mM EDTA (buffer B) and then loaded onto a hydroxyapatite column, previously equilibrated with buffer B. The column was washed using a continuous gradient of 10–700 mM potassium phosphate in buffer B. Finally, P450 was eluted with 0.3% (v/v) Emulgen 911 in 700 mM buffer B. Residual emulgen was removed on hydroxyapatite as described above, except that P450 was eluted in one step with 800 mM potassium phosphate in buffer B after extensively washing the column with buffer B. Purification of reductase. Microsomes containing the NADPH-cytochrome P450 reductase from C. maltosa after heterologous expression in S. cerevisiae were diluted to a final protein concentration of 10 mg/ml in 100 mM Tris/HCl buffer, pH 7.7, containing 30% glycerol, 1 mM EDTA, 1 mM DTT, 0.5 mM PMSF, and 5 mM FAD/FMN. A solution of 10% (w/v) CHAPS was added to give a final concentration of 1.5% (w/v). After stirring for 30 min at 47C, the solubilized mixture was centrifuged at 100,000g for 60 min. The supernatant was diluted to a protein concentration of 3 mg/ml with 25 mM potassium phosphate buffer, pH 7.7, containing 0.3% Pra¨wozell WON-100, 0.1% sodium cholate, 1 mM EDTA, 0.1 mM DTT, 0.25 mM PMSF, 1 mM FAD/ FMN (buffer C), and loaded onto a DEAE-Toyopearl 650M column, previously equilibrated with buffer C. After washing with 10 vol of buffer C containing 0.1 M KCl, the reductase was eluted with a linear gradient of 0.1–0.3 M KCl in this buffer. The pooled fractions were diluted 1:10 in buffer C, not containing Pra¨wozell WON-100, and loaded onto a hydroxyapatite column, which was equilibrated with this buffer. After washing, the reductase was eluted with a linear gradient of 25–300 mM potassium phosphate in this buffer. Finally, reductase was purified by affinity chromatography on 2*,5*-ADPSepharose. Conditions used for loading, washing, and elution of the affinity resin were identical to those previously described (25). Substrate binding studies. Spectrophotometric titration of P450Cm1 and P450Cm2 with n-alkanes and fatty acids of different chain lengths (C10 –C18) was performed in an optimized buffer system containing 1 M sodium citrate buffer, pH 7.4, 30% glycerol, and 0.1% ndodecylmaltoside (buffer D) in order to improve the solubility of these sparingly water-soluble substrates. For this purpose, each substrate was first added as a 100 mM ethanolic solution to buffer D at 377C to give a final concentration of 2 mM, sonicated for 1 min, and then diluted with buffer D to prepare stock solutions of each of the substrates with different concentrations between 7.8 mM and 1 mM. Titration was performed at 307C. After registration of the baselines (substrate solution in buffer D against buffer D), 5 ml of P450 were added to each cuvette (final concentration 1 nmol/ml). The difference spectra were recorded between 350 and 500 nm. These conditions ensured that the incubation mixtures remained optically clear in the presence of all substrates tested even at concentrations up to 1 mM, suggesting a complete substrate dissolution. Both the components and the ionic strength of buffer D itself had no effect on the P450 spin equilibrium in the absence of substrates. In contrast, the phosphate or Tris buffer systems initially selected for the substrate titration studies (10 or 200 mM potassium phosphate, pH 7.4, or 50 mM Tris/buffer pH 7.4, both containing 20% glycerol and 0.3% sodium cholate) were not suited due to turbidity problems occuring at high concentrations of the lipophilic substrates. Activity measurements. The hydroxylation activities of P450Cm1 and P450Cm2 were measured both with intact microsomes con-

/ 6b15$$9382

03-12-96 18:50:45

arca

247

taining the individual P450 form and the reductase after coexpression and in a reconstituted system. Stock solutions of n-alkanes and fatty acids of different chain lenghts (C12 , C16 , and C18) were prepared in ethanol to give a final concentration of 100 mM and specific radioactivities of 1–2 1 103 dpm/nmol of substrate. The reconstituted system contained 0.2 nmol purified P450, 0.8 nmol purified reductase, and 0.35 mg of total microsomal phospholipid extracted from S. cerevisiae GRF18/YEp51Cm1 microsomes by the method of Bligh and Dyer (26). A cholate dialysis method was used to incorporate the enzymes into phospholipid vesicles as described previously (19). The reaction mixture (0.9 ml final volume) contained 200 mM potassium phosphate, pH 7.4, 0.5 mM DTT, and the constituents of an NADPH-regenerating system (15). Ten microliters of 14C-substrate (stock solution or further dilutions with ethanol) was added to give final concentrations of each 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, and 1 mM, respectively. To improve the solubility of the long-chain n-alkanes, 0.02% Pra¨wozell WON-100 was included in the corresponding reaction mixtures. After preincubation at 307C for 10 min (n-alkanes as substrate) or 2 min (fatty acids as substrate), reactions were initiated by addition of 100 ml NADPH (0.5 mM final concentration). After incubation at 307C for 5 min, the reactions were stopped with 0.5 ml 0.8% (v/v) H2SO4 . In control experiments either NADPH or P450 and/or reductase were omitted. Analysis of the n-alkane and fatty acid metabolites by thin-layer chromatography. The stopped reaction mixtures were vigorously mixed with 5 vol of chloroform/methanol (1:2) and extracted with 4 vol chloroform/0.1 M KCl (1:1) and then twice with 5 vol chloroform. The pooled extracts were evaporated to dryness and redissolved in 100 ml chloroform. Product separation was performed by TLC using silica gel plates and a solvent system containing n-hexane/diethylether/acetic acid (40:60:1). TLC plates were counted for radioactivity on a Berthold TLC Linear Analyzer LB 285, equipped with a CHROMA 2D computer program for quantitative analysis of the radiographic density plots. Enzyme activities were determined for each substrate by calculating the sum of the radioactivity of the product peaks as percentage of the total radioactivity, multiplied by the number of nanomoles of substrate originally added to the assay. For product identification, the Rf-values of the radiolabeled n-alkane and fatty acid metabolites were compared with those of authentic standards visualized with 2*,7*-dichlorofluorescein under UV light. Analytical methods. Established methods were used for the determination of P450 (27) and protein concentrations (28) and for SDS–polyacrylamide gel electrophoresis (29). Reductase activity was determined in 50 mM Tris/HCl buffer, pH 7.7, 0.05 mM cytochrome c, 0.1 mM NADPH, 3.3 mM KCN at 227C, based on an extinction coefficient of 21 mM01 1 cm01 at 550 nm (17). One unit of reductase is defined as the activity that reduces 1 mM cytochrome c per min under the above conditions. Phospholipid content of the extracted microsomal phospholipids was estimated by the method of Broekhuyse (30). Spectroscopic studies were carried out on an Uvicon 941plus double beam spectrophometer (Kontron Instruments). The NH2-terminal amino acid sequences of P450Cm1 and P450Cm2 were determined by automated Edman degradation, using a Model 477A/120A Gas-Liquid Phase Protein sequencer from Applied Biosystems.

RESULTS AND DISCUSSION

High-level expression of the recombinant proteins. P450Cm1, P450Cm2, and NADPH-cytochrome P450 reductase were separately produced by expressing their respective cDNAs under control of the galactoseinducible GAL10 promoter in S. cerevisiae (compare Table I). A main prerequisite for the production of the authentic P450Cm2 protein was the exchange of both CTG triplets in the CYP52A4 cDNA by TCT, encoding

AP: Archives

248

SCHELLER ET AL. TABLE I

Strains and Plasmids Used for the Heterologous Expression of P450Cm1, P450Cm2, and NADPHCytochrome P450 Reductase from Candida maltosa S. cerevisiae strain/plasmid

Expressed proteins

References

GRF18/YEp51Cm1 GRF18/YEp5184Vb GRF18/YEp51Cm2 GRF18/YEp51R GRF18/YEp51Cm1-R GRF18/YEp51Cm2-R GRF18/YEp51

P450Cm1 P450Cm2 [Ser(99, 389)Leu] P450Cm2 Reductase P450Cm1 and reductase P450Cm2 and reductase No (control strain)

(8)a (8)a (22)a (24)a (19) (19) (8)

a The cDNA sequences encoding P450Cm1 (CYP52A3), P450Cm2 (CYP52A4), and NADPH-cytochrome P450 reductase from C. maltosa are available from the EMBL nucleotide sequence data base under the Accession Nos. X51931, X51932, and X76226, respectively. b The plasmid contains the original P450Cm2 cDNA which encodes in the host S. cerevisiae GRF18 a mutant protein due to a deviation from the universal genetic code in C. maltosa (22).

serine, since the C. maltosa donor strain exhibits a deviation from the universal genetic code by translating CUG as serine instead of leucine (22, 31). Maximal expression levels were obtained by performing a two-step cultivation procedure that included (i) production of most of the biomass during growth on glucose and (ii) production of the recombinant proteins after induction by galactose. Cultivations in a 20-liter bioreactor resulted in the following expression levels (calculated per 108 cells): 0.18 nmol P450Cm1, 0.10 nmol P450Cm2, and 2.5 units reductase. Cultures were harvested at a cell density of 3 1 108/ml after induction times of 22 h (P450) or 18 h (reductase). In each case, a biomass of nearly 300 g yeast wet weight was obtained containing about 8000 nmol P450Cm1, 4500 nmol P450Cm2, and 86000 units reductase, respectively. Microsomes prepared from the different yeast strains contained per milligram of protein 0.75 nmol P450Cm1, 0.34 nmol P450Cm2, and 10.5 units reductase. For comparison, expression in S. cerevisiae of the original C. maltosa P450Cm2 cDNA containing the unmodified CTG triplets resulted in a significantly lower microsomal P450 content (0.25 nmol/mg). This lower expression level in S. cerevisiae was probably caused by a more rapid degradation of the Ser(99,389)Leu mutant of P450Cm2 compared to the authentic P450Cm2 protein produced after the described codon exchange (22). Analyzing microsomes from the control strain GRF18/YEp51, the amount of host-own P450 (£20 pmol/mg) and NADPH-cytochrome P450 reductase (0.1 units/mg) were almost negligible. Taken together, these results show that the selected host/vector system is well suited for the production of large amounts of individual components of the C. maltosa monooxygenase system as required for their fur-

/ 6b15$$9382

03-12-96 18:50:45

arca

ther purification and detailed investigations on structure–function relationships. In contrast, microsomes prepared from alkane-grown C. maltosa cells contained a number of closely related P450 isoforms, and except P450Cm1, it has been difficult to isolate them separately and to distinguish their enzymatic function solely on the basis of biochemical analysis. The total specific contents of P450 and reductase in these microsomes were estimated to be 0.27 nmol and 0.78 units per mg of microsomal protein, respectively (18). Thus, the heterologous expression system allowed not only a definite separation of different P450 proteins, but yielded in addition even higher P450 and reductase expression levels than the original donor strain. The expression levels reached in this study belong to the highest reported for P450 expression in yeast (5, 32) and productivities (nmol P450 per liter cell culture) are in the same range as achieved recently for the expression of membrane-bound P450 forms in Escherichia coli (33–36). To establish highly active microsomal monooxygenase systems, coexpression of reductase with each of the individual P450 forms was performed using a modified YEp51 plasmid carrying two independent expression units (19). Microsomes prepared after P450Cm1/reductase coexpression contained per milligram of protein 0.26 nmol P450 and 3.5 units reductase, whereas the respective values for P450Cm2/reductase were 0.27 nmol and 3.3 units. Thus, P450/reductase molar ratios of about 1:3 were reached providing a suitable basis for a direct comparison of enzymatic activities of the microsomal and purified reconstituted P450 systems. Purification of the recombinant proteins. P450Cm1 and P450Cm2 were purified as summarized in Table II. Initial attempts to solubilize microsomal P450Cm2 under conditions identical to P450Cm1 with 0.8% sodium cholate in a 10 mM potassium phosphate buffer failed due to low efficiencies of solubilization (13% compared to 89% for P450Cm1). By increasing the potasTABLE II

Purification of Microsomal Cytochromes P450Cm1 and P450Cm2 after Heterologous Expression in S. cerevisiae in a 20-liter Bioreactor

Total P450 content (nmol)

Specific P450 content (nmol per mg of protein)

Yield (%)

Preparation

Cm1

Cm2

Cm1

Cm2

Cm1

Cm2

Microsomes AO-Sepharosea Hydroxyapatite

3699 1687 911

1839 883 262

0.75 7.20 16.1

0.34 6.30 15.8

100 46 25

100 48 14

a

AO-Sepharose, v-amino-octyl Sepharose 4B.

AP: Archives

CHARACTERIZATION OF P450 52A3 AND 52A4

sium phosphate concentration to 250 mM, a solubilization rate greater than 80% was achieved. We also found that P450Cm2 bound only poorly to v-amino-octyl Sepharose using the P450Cm1 purification protocol. Again, an increase of the buffer concentration to 250 mM potassium phosphate significantly improved P450 binding and recovery. The purified P450Cm1 and P450Cm2 proteins were homogeneous in SDS–polyacrylamide gel electrophoresis and migrated with an apparent Mr of 55,000 and 56,500, respectively (Fig. 1). For comparison, the molecular weights calculated from the deduced amino acid sequences were 59,800 for P450Cm1 and 61,800 for P450Cm2. The high specific contents of the final preparations (Table II) indicate that both P450 forms were obtained almost completely in the native, heme-containing state. Partial amino acid sequencing revealed identical NH2-terminal sequences of the P450Cm1 proteins purified from C. maltosa and S. cerevisiae. The determined and cDNA-deduced NH2-terminal amino acid sequences of P450Cm1 as well as of P450Cm2 were identical except for the starting methionine, demonstrating that a further processing occurred neither in the donor nor in the host organism (Fig. 2). The reductase from C. maltosa was purified starting from microsomal fractions after heterologous expression in S. cerevisiae. After purification, the preparations were free of other constituents of the microsomal electron transfer system, e.g., P450, cytochrome b5 , and NADH-cytochrome b5 reductase. In SDS–polyacrylamide gel electrophoresis (see Fig. 1), the reductase migrated as a homogeneous protein with an apparent Mr of 79,000 (compared to 77,100, calculated from the

FIG. 1. SDS–polyacrylamide gel electrophoresis of purified P450Cm1, P450Cm2, and NADPH-cytochrome P450 reductase. The purified proteins were analyzed on a 10% SDS–polyacrylamide gel and visualized with Coomassie blue. Lane 1, molecular weight standards (92,500, 68,000, 45,000, and 29,000); lane 2, P450Cm1 (2 mg); lane 3, P450Cm2 (1 mg), lane 4, NADPH-cytochrome P450 reductase (1 mg).

/ 6b15$$9382

03-12-96 18:50:45

arca

249

FIG. 2. NH2-terminal amino acid sequences of P450Cm1 and P450Cm2. (A) cDNA deduced sequences; (B) sequences determined for the purified proteins after heterologous expression; (C) sequence determined for the P450Cm1 protein obtained after purification from alkane-grown C. maltosa cells. The number in parentheses corresponds to the total number of amino acids deduced from the corresponding cDNA sequence.

deduced amino acid sequence). The highly purified protein showed a specific activity of 63 units/mg of protein which is very close to the respective data determined for NADPH-cytochrome P450 reductases from rat liver microsomes (37–39). Based on this value, one nmol of reductase was assumed to correspond to 4.6 units. Spectral properties and substrate binding studies. Purified P450Cm1 and P450Cm2 exhibited nearly identical spectral characteristics with Soret peaks of the oxidized and dithionite-reduced forms at 418 and 415 nm, respectively (Figs. 3A and 3B), and an absorbance maximum of the reduced CO-complex at 448 nm (Fig. 3C). Furthermore, both P450 forms had well resolved a- and b-bands at 568 and 539 nm, respectively. A marked change of the absorbtion spectra was observed upon addition of n-octadecane to P450Cm1 and lauric acid to P450Cm2 (Soret peak at 392 nm, compare Fig. 3). Following the interpretations of other authors, this spectral shift is typical for a low-spin to high-spin transition of the heme iron (40). As known in detail from crystallographic studies with P450cam , the low-spin state of the heme iron is characterized by the presence of a cluster of six water molecules in the active site, one of which occupies the sixth iron coordination site (41, 42). The binding of substrates is generally assumed to lead to water exclusion resulting in the five-coordinated high-spin heme complex (43). The extent of low-spin to high-spin transition seems to depend on the steric structure of the substrate and its influence on the accessibility of the heme pocket for water molecules. Thus, it may be used as a measure to evaluate the capacity of the substrate to fit into the recognition pocket and to convert the enzyme into its active conformation (44). The high-spin content of substrate-bound P450Cm1 and P450Cm2 was calculated by decomposing the absorption spectra into the overlapping bands of the pure low-spin state and the pure high-spin state using the curve fit procedure described by Jung et al. (44). The high-spin content in the absence of substrate was approximately 16 and 11% for P450Cm1 and P450Cm2,

AP: Archives

250

SCHELLER ET AL.

FIG. 3. Comparison of spectral properties of purified P450Cm1 and P450Cm2. Absolute spectra of the oxidized low-spin and reduced forms of P450Cm1 (A) and P450Cm2 (B) were recorded with 5 nmol P450 in 200 mM potassium phosphate buffer, pH 7.3, containing 30% glycerol, 1 mM EDTA, 0.5 mM DTT, and 0.1% Tween-20. Spectra of the P450 high-spin states were measured in 1 M sodium citrate buffer, pH 7.3, containing 30% glycerol, 0.1% n-dodecylmaltoside, and 1 mM n-octadecane (C18) or lauric acid (LA) as substrates for P450Cm1 and ) and P450Cm2 (rrrrrrr). P450Cm2, respectively. C shows difference spectra of the reduced carbon monoxide complexes of P450Cm1 (

FIG. 4. Difference spectra of P450Cm1 (A) and P450Cm2 (B) obtained upon incubation with n-octadecane and lauric acid, respectively. Titration was performed in 1 M sodium citrate buffer, pH 7.3, containing 30% glycerol, 0.1% n-dodecylmaltoside. Difference spectra were recorded after incubation of 1 nmol P450 with different concentrations of n-octadecane (for P450Cm1) or lauric acid (for P450Cm2). Spectra shown in A were measured at the following n-octadecane (C18) concentrations: 7.8, 15.6, 62.5, 250, and 1000 mM. Spectra shown in B were recorded after titration of P450Cm2 with 15.6, 31.2, 62.5, 125, 250, 500, and 1000 mM lauric acid (LA), respectively. The inserts represent the reciprocal plots of DA389 – 421 versus the corresponding substrate concentration (Lineweaver–Burk plot).

/ 6b15$$9382

03-12-96 18:50:45

arca

AP: Archives

CHARACTERIZATION OF P450 52A3 AND 52A4

respectively, demonstrating that both P450 forms were isolated predominantly in the low-spin state. Maximal detectable low-spin to high-spin shifts were measured upon addition of n-octadecane to P450Cm1 (81% highspin content) and lauric acid to P450Cm2 (85% highspin content). Substrate binding was further studied by means of substrate difference spectroscopy (40). Examples for the type I spectral changes observed are shown in Fig. 4. The spectra indicate a low-spin to high-spin shift of P450 upon substrate binding and are characterized by an absorption peak centered at 389 nm and a trough at 421 nm. Titration of purified P450Cm1 and P450Cm2 with C10 –C18 n-alkane and fatty acid substrates allowed the calculation of the apparent spectral dissociation constants (Ks) in a Lineweaver–Burk plot (Fig. 4). Moreover, the maximal high-spin contents reached under substrate saturation conditions were estimated as a further parameter to characterize P450– substrate interactions. In general, substrates having the highest affinities (lowest apparent Ks-values) were also those that induced the strongest type I spectral changes. Considering these two parameters, P450Cm1 and P450Cm2 clearly differed in their substrate preferences and chain-length dependencies (Fig. 5). P450Cm1 showed highest affinities and spectral effects with n-octadecane and n-hexadecane. Among the fatty acids tested, only stearic acid caused a weak type I spectral change, whereas reverse type I spectra were observed with fatty acids of shorter chain lengths. Quite different results were obtained with P450Cm2. Clearly preferred substrates of this P450 form were lauric acid and myristic acid (Fig. 5). Interestingly, similar chain-length optima around C12 –C14 were found for the binding of fatty acids and n-alkanes. Compared to P450Cm1, interactions of P450Cm2 with n-alkanes were characterized, however, by relatively low affinities and weak type I spectral changes. Hydroxylase activities toward n-alkanes and fatty acids. Active monooxygenase systems were obtained in two different ways. First, the highly purified P450 and reductase components were combined in the presence of phospholipids by means of a cholate-dialysis method (reconstituted system). As previously shown for P450Cm1, maximal activities were reached at a three- to fourfold molar excess of reductase (23). And second, the monooxygenase systems were directly established in the endoplasmic reticulum of S. cerevisiae by coexpression of the required components. The derived membrane fractions contained P450 and reductase in molar ratios of about 1:3 (microsomal system). It should be noted that replacement of the total microsomal phospholipid fraction by 1,2-dilauroylglyceryl-3-phosphocholine resulted in turnover numbers

/ 6b15$$9382

03-12-96 18:50:45

arca

251

FIG. 5. Comparison of the apparent spectral dissociation constants (Ks) and the substrate-dependent maximal high-spin shift of purified P450Cm1 and P450Cm2 during interaction with various n-alkanes and fatty acids. Spectral titration of P450Cm1 (l) and P450Cm2 (s) with C10 –C18 alkane and fatty acid substrates were performed as described in ‘‘Materials and Methods.’’ From the reciprocal plot of the absorbance change DA389 – 421 in the substrate difference spectra versus the corresponding substrate concentration (Lineweaver– Burk plot), the intercept on the x-axis was taken as the negative inverse of Ks . The high-spin content of the purified substrate-free proteins is marked by arrows. The maximal high-spin content of P450Cm1 and P450Cm2 after incubation with each substrate was calculated by comparison of the signal response DA389 – 421 with the maximal absorbance difference for P450Cm1/octadecane and P450Cm2/lauric acid, which correspond to a high-spin content of 81 and 85%, respectively (see text). The apparent Ks-values were estimated from three independent titration experiments and represent the mean { the SD of the dissociation constant. The determination of the apparent Ks-values for fatty acid binding of P450Cm1 was hindered by the appearance of reverse type I spectral effects upon addition of C10 –C14 substrates.

about fivefold lower than those obtained for n-hexadecane hydroxylation by P450Cm1 (23). Short-term reactions (reaction times not longer than 5 min) revealed only one major product of n-alkane (the respective n-alkyl-1-alcohol) and fatty acid (the v-hydroxylated fatty acid) hydroxylation catalyzed by either the reconstituted or the microsomal system (data

AP: Archives

252

SCHELLER ET AL. TABLE III

Comparison of Kinetic Parameters for Hydroxylation of n-Alkanes and Fatty Acids with Different Chains Lengths in a Reconstituted System. Vmax (mM 1 min01) P450

C12

Km (mM)

C16

C18

C12

C16

C18

Alkane hydroxylation P450Cm1 P450Cm2

11 (14 11 (13

{ { { {

0.7 0.9) 0.5 0.7)

27 (40 10 (12

{ { { {

1.5 1.2) 0.6 0.5)

13 (16 6.1 (7.9

{ { { {

0.4 1.9) 0.9 1.2)

65 (57 92 (87

{ { { {

18 2) 18 14)

54 (48 114 (108

{ { { {

17 7) 25 11)

51 (39 235 (195

{ { { {

13 11) 97 95)

93 (70 104 (99

{ { { {

8 2) 9 28)

211 (163 273 (234

{ { { {

23 11) 17 15)

Fatty acid hydroxylation P450Cm1 P450Cm2

5.4 { 0.6 (7.4 { 0.8) 30 { 1.8 (47 { 7.3)

9.4 (14 5.5 (5.8

{ { { {

0.3 0.1) 0.1 0.1)

8.1 (12 0.3 (0.5

{ { { {

1.9 2.4) 0.1 0.2)

194 (182 61 (71

{ { { {

27 36) 13 5)

Note. Purified P450Cm1 and P450Cm2 were reconstituted with NADPH-cytochrome P450 reductase in the presence of microsomal phospholipid. Activity measurement was performed as described under Materials and Methods. Data obtained from duplicate substrate titration experiments were fitted against the Michaelis–Menten equation using the ENZFIT computer software. The values in brackets represent Vmax and apparent Km obtained with the respective microsomes prepared after P450/reductase coexpression.

not shown). These products were identical to those earlier detected upon hydroxylation of n-hexadecane and lauric acid with C. maltosa microsomes (15). When reaction times were prolonged (ú20 min), up to three additional products were detected in the case of n-alkane hydroxylation by P450Cm1 as well as by P450Cm2. They were identified in TLC by comigration with nonlabeled standards as palmitic acid, 1,16-hexadecanediol and 1,16-hexadecanedioic acid. v-Hydroxy fatty acids were also found to be further oxidized resulting in the formation of dicarboxylic acids. A characterization of these secondary products by gas-chromatography/mass spectrometry and the mechanism of their formation will be published elsewhere. The reconstituted as well as the microsomal systems were used to compare the substrate specificities of P450Cm1 and P450Cm2 in terms of the kinetic parameters for the hydroxylation of several n-alkanes and fatty acids (Table III; Fig. 6). Both systems revealed statistically insignificant differences in the apparent Km-values whereas the Vmax-values obtained for the microsomal systems were higher by 20–30% on average. It should be noted that all enzymatic reactions were NADPH-dependent and not supported by NADH. If one of the constituents of the reconstituted system was omitted, no activity was detectable (data not shown). In the case of P450Cm1, n-alkanes were clearly the preferred substrates. For each of the chain-lengths tested, the calculated Vmax/Km ratios (Fig. 6) of n-alkane hydroxylation were at least five times higher than those with fatty acids. Compounds with a chain-length

/ 6b15$$9382

03-12-96 18:50:45

arca

of C16 gave the highest values independent of the substrate class. In contrast, the substrate specificity of P450Cm2 was characterized by (i) a clear chain length optimum both for alkanes and fatty acids with 12 carbon atoms (ii) a strong preference for the lauric acid substrate and (iii) a lower chain-length selectivity within the n-alkane substrates compared to the fatty acids. Except the data obtained for octadecane and stearic acid hydroxylation by P450Cm1, there was a good correlation between the degree of the low-spin to highspin shift upon substrate binding and the corresponding Vmax-values in the enzymatic assay. As studied in detail with other P450 forms (45), this is probably caused by a change of the redox potential of the P450 heme iron upon substrate binding leading to increased rates of electron transfer from NADPH-cytochrome P450 reductase (46). We assume, that the estimated binding and catalytic constants reflect significant differences in the substrate specificity of P450Cm1 and P450Cm2 since they were determined under the same definite conditions, i.e., identical concentrations of P450, phospholipid, substrate, and buffer composition (pH, ionic strength, detergent concentration). However, it should be emphasized in this context that all the kinetic parameters determined have to be considered as apparent values since the actual substrate concentrations available to the membrane-bound P450 enzymes are unknown. Thus the C10 –C18 alkane and fatty acid substrates tested can be expected to partition with different coef-

AP: Archives

CHARACTERIZATION OF P450 52A3 AND 52A4

253

substrate specificity of P450Cm1 and P450Cm2. Therefore, it can be concluded that coexpression of an individual C. maltosa P450 form and the reductase is in general a well-suited alternative strategy to rapidly elucidate the functional role of the other CYP52 gene products. The functional characterization of P450Cm1 and P450Cm2 may provide a suitable basis for further investigations on the structural basis of their substrate specificities. Using for example the three-dimensional crystal structure of P450BM3 as an active site model for a fatty acid hydroxylating P450 (49), we speculate that P450Cm2, but not P450Cm1, may contain an arginine residue in its substrate binding region that ‘‘fixes’’ the carboxyl group of the fatty acid substrate similarly as shown for other fatty acid binding proteins (50, 51). According to a recent alignment of the primary structures, Arg104 in P450Cm2 corresponds to Arg47 in P450BM3 and may serve as such a functionally important amino acid residue. This would give a reasonable explanation for the strong selectivity in binding and hydroxylation of C12 –C14 fatty acid substrates by P450Cm2 compared to less pronounced chain-length dependencies and rather moderate activities toward the completely hydrophobic n-alkanes. In contrast, the substrate binding site of P450Cm1 may represent a large hydrophobic pocket, the length of which is indicated by the findings, that only fatty acids having at least 18 carbon atoms are able to induce type I spectral changes. This hypothesis can be tested in further studies by site-directed mutagenesis. FIG. 6. Substrate selectivity of P450Cm1 and P450Cm2 by the hydroxylation of n-alkanes and fatty acids with different chain lengths. The bars represent the Vmax/Km values estimated from kinetic studies in a reconstituted system as shown in Table III.

ficients at least between the following phases of the reaction mixture: detergent micelles, aqueous phase, and lipid membrane. Further complicating the interpretation of the data, it is not clear whether the substrates enter the P450 active site from the membrane or aqueous phase. These problems generally apply to the binding and conversion of lipophilic substrates by microsomal P450 enzymes and have already been extensively discussed by other authors (47, 48). Taken together, the comparison of P450Cm1 and P450Cm2 gives a detailed insight into the development of different enzymatic functions among the members of the CYP52 family. The results show that P450Cm1 and P450Cm2 exhibit distinct though partially overlapping substrate specificities due to the different chainlength dependencies and preferences for either n-alkanes or fatty acids. It was further demonstrated that both the reconstituted as well as the microsomal system containing the heterologously expressed redox components yielded identical results concerning the

/ 6b15$$9382

03-12-96 18:50:45

arca

ACKNOWLEDGMENTS The authors are indebted to Drs. K.-L. Schro¨der and R. Kraft for support in large-scale cell cultivation and protein sequencing, respectively, and E. Honeck, H. Schoth, and P. Neubert for their excellent technical assistance. They are also grateful to Dr. C. Jung for support in calculation of the P450 spin states and Drs. H. Honeck, S. Mauersberger, and C. Jung for helpful discussions and critical reading of the manuscript.

REFERENCES 1. Porter, T. D., and Coon, M. J. (1991) J. Biol. Chem. 266, 13469– 13472. 2. Nelson, D. R., Kamataki, T., Waxman, D. J., Guengerich, F. J., Estabrook, R. W., Feyereisen, R., Gonzalez, F. J., Coon, M. J., Gunsalus, I. C., Gotoh, O., Okuda, K., and Nebert, D. W. (1995) P450 World Wide Web site on D. R. Nelson’s PC (Macintosh) at the University of Tennessee (Memphis), available under http:// drnelson.utmem.edu/homepage.html. 3. Sanglard, D., and Loper, J. C. (1989) Gene 76, 121–136. 4. Seghezzi, W., Sanglard, D., and Fiechter, A. (1991) Gene 106, 51–60. 5. Seghezzi, W., Meili, C., Ruffiner, R., Ku¨enzi, R., Sanglard, D., and Fiechter, A. (1992) DNA Cell Biol. 11, 767–780. 6. Takagi, M., Ohkuma, M., Kobayashi, N., Watanabe, M., and Yano, K. (1989) Agric. Biol. Chem. 53, 2217–2226. 7. Schunck, W.-H., Ka¨rgel, E., Gross, B., Wiedmann, B., Mauersberger, S., Ko¨pke, K., Kießling, K., Strauss, M., Gaestel, M., and

AP: Archives

254

8.

9. 10. 11. 12.

13. 14. 15.

16. 17. 18. 19. 20. 21.

22. 23. 24. 25. 26. 27. 28.

SCHELLER ET AL.

Mu¨ller, H.-G. (1989) Biochem. Biophys. Res. Commun. 161, 843– 850. Schunck, W.-H., Vogel, F., Gross, B., Ka¨rgel, E., Mauersberger, S., Ko¨pke, K., Gengnagel, C., and Mu¨ller, H.-G. (1991) Eur. J. Cell Biol. 55, 336–345. Ohkuma, M., Hikiji, T., Tanimoto, T., Schunck, W.-H., Mu¨ller, H.-G., and Takagi, M. (1991) Agric. Biol. Chem. 55, 1757–1764. Ohkuma, M., Tanimoto, T., Yano, K., and Takaki, M. (1991) DNA Cell Biol. 10, 271–282. Ohkuma, M., Muraoka, S.-I., Tanimoto, T., Fujii, M., Ohta, A., and Takagi, M. (1995) DNA Cell Biol. 14, 163–173. Lottermoser, K., Asperger, O., and Schunck, W.-H. (1994) in Proceedings of the 8th International Conference on Cytochrome P450 (Lechner, M. C., Ed.), pp. 643–646, John Libbey Eurotext, Paris. Ka¨ppeli, O. (1986) Microbiol. Rev. 50, 244–258. Schunck, W.-H., Mauersberger, S., Huth, J., Riege, P., and Mu¨ller, H.-G. (1987) Arch. Microbiol. 147, 240–244. Blasig, R., Mauersberger, S., Riege, P., Schunck, W.-H., Jockisch, W., Franke, P., and Mu¨ller, H.-G. (1988) Appl. Microbiol. Biotechnol. 28, 589–597. Vogel, F., Gengnagel, C., Ka¨rgel, E., Mu¨ller, H.-G., and Schunck, W.-H. (1992) Eur. J. Cell Biol. 57, 285–291. Honeck, H., Schunck, W.-H., Riege, P., and Mu¨ller, H.-G. (1982) Biochem. Biophys. Res. Commun. 106, 1318–1324. Riege, P., Schunck, W.-H., Honeck, H., and Mu¨ller, H.-G. (1981) Biochem. Biophys. Res. Commun. 98, 527–534. Zimmer, T., Kaminski, K., Scheller, U., Vogel, F., and Schunck, W.-H. (1995) DNA Cell Biol. 14, 619–628. Cuatrecasas, P. (1970) J. Biol. Chem. 245, 3059–3065. Broach, J. R., Li, Y.-Y., Wu, L.-C. C., and Jayaram, M. (1983) in Experimental Manipulation of Gene Expression. (Inoye, M., Ed.), pp. 83–117. Academic Press, New York. Zimmer, T., and Schunck, W.-H. (1995) Yeast 11, 33–41. Scheller, U., Kraft, R., Schro¨der, K.-L., Schunck, W.-H. (1994) J. Biol. Chem. 269, 12779–12783. Ka¨rgel, E., Menzel, M., Honeck, H., Vogel, F., Bo¨hmer, A., and Schunck, W.-H. (1996) Yeast, in press. Ardies, C. M., Lasker, J. M., Bloswick, B. P., and Lieber, C. S. (1987) Anal. Biochem. 162, 39–46. Bligh, E. G., and Dyer, W. J. (1959) Can. J. Biochem. Physiol. 37, 911–917. Omura, T., and Sato, R. (1964) J. Biol. Chem. 239, 2370–2378. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randal, R. J. (1951) J. Biol. Chem. 193, 265–275.

/ 6b15$$9382

03-12-96 18:50:45

arca

29. Laemmli, U. K. (1970) Nature 227, 680–685. 30. Broekhuyse, R. M. (1968) Biochim. Biophys. Acta 152, 307–315. 31. Sugiyama, H., Ohkuma, M., Masuda, Y., Park, S.-M., Ohta, A., and Takagi, M. (1995) Yeast 11, 43–52. 32. Gonzalez, F. J., and Korzekwa, K. R. (1995) Annu. Rev. Pharm. Toxicol. 35, 369–390. 33. Barnes, H. J., Arlotto, M. P., and Waterman, M. R. (1991) Proc. Natl. Acad. Sci. USA 88, 5597–5601. 34. Fisher, C. W., Caudle, D. L., Martin-Wixtrom, C., Quattrochi, L. C., Tukey, R. H., Waterman, M. R., and Estabrook, R. W. (1992) FASEB J. 6, 759–764. 35. Sagara, Y., Barnes, H. J., and Waterman, M. R. (1993) Arch. Biochem. Biophys. 304, 272–278. 36. Richardson, T. H., Jung, F., Griffin, K. J., Wester, M., Raucy, J. L., Kemper, B., Bornheim, L. M., Hassett, C., Omiecinski, C. J., and Johnson, E. F. (1995) Arch. Biochem. Biophys. 323, 87–96. 37. Yasukochi, Y., and Masters, B. S. S. (1976) J. Biol. Chem. 251, 5337–5344. 38. Dignam, J. D., and Strobel, W. (1977) Biochemistry 16, 1116– 1123. 39. Shephard, E. A., Pike, S. F., Rabin, B. R., and Phillips, I. A. (1983) Anal. Biochem. 129, 430–433. 40. Jefcoate, C. R. (1978) Methods Enzymol. 52, 258–279. 41. Poulos, T. L., Finzel, B. C., and Howard, A. J. (1986) Biochemistry 25, 5314–5322. 42. Thomann, H., Bernardo, M., Goldfarb, D., Kroneck, P. M. H., and Ullrich, V. (1995) J. Am. Chem. Soc. 117, 8243–8251. 43. Raag, R., and Poulos, T. L. (1989) Biochemistry 28, 914–922. 44. Jung, C., Ristau, O., and Rein, H. (1991) Biochim. Biophys. Acta 1076, 130–136. 45. Fisher, M. T., and Sligar, S. G. (1983) J. Am. Chem. Soc. 107, 5018–5019. 46. Poulos, T. L., and Raag, R. (1992) FASEB J. 6, 674–679. 47. Parry, G., Palmer, D. N., and Williams, D. J. (1976) FEBS Lett. 67, 123–129. 48. Vermeir, M., Boens, N., and Heirwegh, K. P. M. (1992) Biochim. Biophys. Acta 1107, 93–104. 49. Ravichandran, K. G., Boddupalli, S. S., Hasemann, C. A., Peterson, J. A., and Deisenhofer, J. (1993) Science 261, 731–736. 50. Cheng, L., Qian, S. J., Rothschild, C., d’Avignon, A., Lefkowith, J. B., Gordon, J. I., and Li, E. (1991) J. Biol. Chem. 266, 24404– 24412. 51. Borchers, T., and Spencer, F. (1993) Mol. Cell. Biochem. 123, 23–27.

AP: Archives