A Single Amino Acid Substitution Confers Progesterone 6 @-Hydroxylase Activity to Rabbit Cytochrome P 450 2 C 3 *

A cDNA encoding a naturally occurring variant of cytochrome P450 (P450) 2C3 that catalyzes the 66and l6a-hydroxylation of progesterone exhibits six differences of nucleotide sequence leading to five amino acid substitutions from that encoding 2C3, a progesterone lea-hydroxylase that does not catalyze GB-hydroxylation. Analysis of chimeric and mutant enzymes indicates that a Ser/Thr difference at position 364 underlies the difference between the two enzymes in GB-hydroxylase activity as well as ensitivity to the inhibitor, 16a-methylprogesterone. In addition, an Ile/ Met difference at position 178 influences the apparent K,,, for progesterone. The two mutations, S364T and I178M, together convert 2C3 to a form that exhibits kinetic properties which are similar to the 2C3v enzyme, and the reciprocal mutations in 2C3v convert it to an enzyme that resembles 2C3. Interestingly, position 364 of 2 C 3 maps to a substrate-contacting domain suggested by models for mammalian P450 enzymes based on the structure of P450cam. Ile”* is highly conserved among mammalian microsomal P450s with the exception of CYP4A and CYP19 enzymes which exhibit a Met at this alignment position.

residues that affect changes in the catalytic properties of P450s align at a position which maps to a substrate-contacting surface loop in the bacterial enzyme P450cam (4). The region containing these key amino acid residues is highly variable among closely related P450 enzymes which have distinct catalytic functions. We have proposed a framework model for P450 enzymes in which substrate-contacting surface loops readily accommodate genetic changes that lead to changes in substrate specificity without altering the basic topological organization of the enzymes (3)(4)(5). Other regions are more highly conserved and may be part of basic topological features, such as the heme binding site that is required for the reduction of oxygen. Based on this model, we have demonstrated that we could transfer a hypervariable domain from P450 2C5 to P450 2C1, two enzymes that share less than 75% amino acid identity, and confer a new catalytic activity, progesterone 21 hydroxylation, to 2C1 (3).
The characterization of naturally occurring variants of P450 has been especially useful for identifying critical amino acids for substrate specificity and enzymatic activity. We have characterized the biochemical properties of closely related rabbit liver P450 enzymes that catalyze 60-and 16a-hydroxylation of progesterone (6). Only one of these forms, designated as SP', catalyzes the 60-hydroxylation of progesterone, whereas the other, designated as 6p-, does not (6). Both forms catalyze 16a-hydroxylation, but the 60' form exhibits a higher catalytic efficiency for this activity. In addition, 16a-methylprogesterone selectively inhibits both the 6p-and 16a-hydroxylase activities catalyzed by the 60+ form, whereas it slightly stimulates the 16a-hydroxylase activity of the 6pform. This activation of the 60-form is more apparent for 5/3-pregnane-3@,20a-diol, a naturally occurring catabolic product of progesterone (7). Preparations of 2C3 isolated from outbred New Zealand White or inbred III/J rabbit liver appear to be a mixture of both the 6p-and 6p' forms (6,8). In contrast, preparations of 2C3 isolated from inbred IIIVO/J or B/J rabbit liver do not contain the 6p+ form (6,8).
The complete nucleotide sequence of the cDNA encoding 2C3 has been derived from a partial cDNA and gene sequences for 2C3 (9,10). Characterization of the cognate protein expressed in COS-1 cells or in Escherichia coli indicates that it encodes the 60-form (11). Based on the high degree of structural similarity expected for the two forms of 2C3, we reasoned that a PCR based approach could be utilized to isolate and identify a cDNA encoding the 6p' form of 2C3 (2C3v) by using primers corresponding to the reported sequence of the 2C3 gene (9) and first strand cDNAs prepared from New Zealand White rabbit liver RNA. In this report, we describe the successful isolation of a cDNA encoding 2C3v that catalyzes progesterone GB-hydroxylation and identify five differences in its predicted amino acid sequence relative to 6939 6940 Characterization of a Variant Form of P450 2C3 2C3. Expression of chimeric constructs from 2C3 and 2C3v in COS-1 cells and in E. coli revealed the amino acid difference at residue 364 is the determinant of the 6P-hydroxylation of progesterone. Sequence alignments (12, 13) predict that this amino acid position corresponds to a region that forms a portion of the substrate binding site within the polypeptide chain of P450cam.

MATERIALS AND METHODS
Production of P450 2C3 cDNAs by PCR-The synthesis of cDNA from total RNA prepared from a New Zealand White rabbit liver (14) and its subsequent amplification by PCR using the GeneAmp PCR Kit followed the method described by Perkin-Elmer Cetus. Briefly, first strand cDNAs were prepared with MuLV reverse transcriptase (GIBCO/BRL) using either random 9-mers (Stratagene) or oligo(dT) (Boehringer Mannheim) as primers. Double-stranded cDNAs were then synthesized by PCR using Taq DNA polymerase and two oligonucleotide primers which correspond to the 5' and 3' ends of the coding region of 2C3 based on the published sequence (9,10). The upstream primer, 5'-GAAGATCTGCCATGGATCTCCTCATTA-TCTTG-3', corresponds t o t h e n d of the coding strand of 2C3 (nucleotides -11 to +21) and contains an additional BglII site at its 5' end (underlined). The downstream primer, 5"GCTCTAGATCA-GACTGGAACAAAACACAGCTCA-3', corresponds t o e n d of the complementary strand (nucleotides 1478-1510) and contains an additional XbaI site which is underlined. The PCR reaction utilized an Ericomp thermal cycler and 35 repetitions of the following cycle: 94 "C 1 min (denature), 55 "C 1 min (anneal), 72 "C 2 min (extend) followed by a single incubation for 7 min at 72 "C. The approximately 1.5-kilobase pair PCR products were isolated from 1% agarose gels (SeaKem GTG grade, FMC BioProducts) and purified using the Geneclean Kit (BiolOl). Purified PCR products were digested with BglII and XbaI and ligated into the expression vector, pCMV (15), using T4 DNA ligase (Bethesda Research Laboratories). The ligated DNAs were used to transform competent E. coli DH5a (Bethesda Research Laboratories). Initial analysis utilized DNA purified by a method employing hexadecyl trimethyl ammonium bromide (16).
Cloning of 2C3 cDNAs from a Rabbit Liver L i b r a r y " probe, CAGTGTTTGACAGAGTCACC, corresponding to nucleotides 302-321 of the 2C3 cDNA, was synthesized and used to screen a rabbit liver cDNA library provided by Dr. D. Russell of the University of Texas Southwestern Medical School (17). The oligonucleotide was end-labeled with [ Y -~* P ]~A T P (>3000 Ci/mmol) using T4 polynucleotide kinase (Stratagene). Four hybridization positive clones were checked for the presence or absence of a specific SpeI restriction site. The nucleotide sequences of the longest representative SpeI+ and SpeI-cDNAs were determined.
Construction of Chimeras and Site-directed Mutagenesis-Chimeras were constructed from the 2C3 and 2C3v cDNAs in pCMV by exchanging restriction fragments, generated using the restriction enzymes: BsaBI, MscI, and PpuMI. The resulting constructs were verified by complete sequence analysis. The S364T mutation in 2C3 (2C3:S364T) was generated using a two-step PCR procedure for sitedirected mutagenesis developed by Landt et al. (18). The first PCR reaction utilized as primers the aforementioned upstream primer CATGGGGCAAAGTAGTGGGG, which is complementary to nucle-(nucleotides -11 to +21) and a specific mutagenic oligonucleotide, otides 1083-1102 2 the 2C3 cDNA. The underlined nucleotide indicates the mutation introduced into 2C3. The first PCR reaction consisted of 30 cycles of 1 min at 94 "C, 1 min at 45 "C and 2 min at 72 "C, followed by a 7-min incubation at 72 "C. The gel-purified PCR product and the downstream primer described above (nucleotides 1478-1510) were used as the primers for a second PCR reaction. The second PCR reaction consisted of 35 cycles as described above for the amplification of the 2C3 cDNAs. Both PCR reactions utilized the 2C3 cDNA as template and Vent DNA polymerase (New England Biolabs). The final PCR product was isolated from a 1% agarose gel, purified with the Geneclean Kit (BiolOl), digested with BglII and XbnI, and ligated into the plasmid pCMV5 for expression in COS-1 cells. The insert of the resulting construct was sequenced in its entirety, and the loss of the SpeI site was also verified by restriction mapping.
Expression in COS-1 Cells-The expression vector, pCMV5, obtained from Dr. D. Russell of the University of Texas Southwestern Medical School and used with the permission of Dr. M. Stinski, University of Iowa, was employed for the expression of 2C3-related proteins from their respective cDNAs in COS-1 cells (ATCC). DNA for transfection was prepared by the alkaline lysis method followed by a CsCl centrifugation (14). COS-1 cells were cultured and transiently transfected as described previously (3). The 6p-and 1601hydroxylation of progesterone was determined at 72 h after transfection by supplementation of the culture medium for 2 h with 10 PM ["C]progesterone (57.2 or 60 Ci/mol, Du Pont-New England Nuclear), followed by extraction and analysis by thin layer chromatography (TLC) on silica gel (5). Metabolites and substrate were separated by TLC as described (19), with one modification to improve resolution. Extracted substrate and products were first separated using benzene:ethyl acetate (3:l) prior to the use of two solvent systems employed in earlier work (6). Metabolites were quantified by liquid scintillation counting.
Sequence Analysis-Nucleotide sequencing utilized the dideoxynucleotide chain termination method (20) with [ Y -~~S I~A T P~S (>lo00 Ci/mmol, Amersham Corp.) and T7 DNA polymerase based sequencing Kits (Pharmacia LKB Biotechnology Inc. and United States Biochemical Corp.). Fourteen oligonucleotide primers corresponding to the 2C3 cDNA were utilized to verify the complete sequence on both strands of all constructs. Sequencing gels consisted of 6% acrylamide and 7.8 M urea. The electrophoresis buffer was 1 X TBE (14). In some cases, the gel was run for 1 h with 0.5 X TBE in the upper buffer chamber and 1 X TBE in the lower buffer chamber and then the buffer in the lower chamber was altered by addition of 0.5 volume of 3 M sodium acetate, pH 5.0, to increase the number of readable bases per reaction (21).
Heterologous Expression of 2C3 Enzymes in E. coli-The pCW expression vector was obtained from Dr. R. Dahlquist (Institute of Molecular Biology, University of Oregon, Eugene, OR) and used for the expression of 2C3 proteins in E. coli. Details for the construction of the 2C3 cDNA in pCW has been described elsewhere (11). An internal EcoRI site (nucleotides 431-436) and the 3'-flanking XbaI site were utilized for constructing the recombinant plasmids in pCW. The EcoRI and XbaI digested and purified fragments from the various chimeric constructs in pCMV were transferred into the pCW 2C3 ( l l ) , previously digested with the same restriction enzymes. All constructs were verified by restriction mapping and sequence analysis. These plasmids were used to transform the E. coli strain XL-1 Blue (Stratagene) which served as the expression host.
P450 expression was verified using the carbon monoxide difference spectrum of whole cell suspensions with the addition of 16.7 PM methyl viologen to accelerate the reduction of P450 (22). The cultures were grown in Terrific broth (14) at 30 "C and harvested after 48 h. The P450 proteins were purified as described (11) with the following modifications. The P450 proteins were eluted from the first column of HA-Ultrogel (IBF Biotechnics, Inc., Columbia, MD) using 0.15 M KPEG buffer (potassium phosphate buffer, pH. 7.4, containing 0.1 mM EDTA and 20% glycerol), with 0.3% Nonidet P-40, and dialyzed against 10 mM KPEG overnight. The dialyzed P450 solutions were concentrated, and the detergent was removed using calcium phosphate gel, as described earlier for preparations of 2C3 from rabbit liver (23). The proteins were eluted from calcium phosphate gel with 0.5 M KPEG and dialyzed against 10 mM KPEG overnight. Protein concentrations were determined using the Pierce BCA protein assay kit (Pierce Chemical Co.) employing bovine serum albumin as the standard. Preparations used for kinetic characterization had specific contents of P450 exceeding 11 nmol/mg. P450 (10 pmol), dilauroyl-L-a-lecithin (30 gg), and 0.3 unit of purified rabbit liver P450 reductase were reconstituted and assayed for progesterone metabolism in the presence or absence of 10 PM 5P-pregnane-3P,ZOa-diol (diol) or 5 PM 16-01-methylprogesterone as described (6) with modifications to the TLC procedure as described above.

RESULTS
Isolation of cDNA Encoding the 6@* Form of 2C3"In order to isolate and characterize the 6@' variant of 2C3, PCR primers were designed, based on the sequence of the gene (9) encoding the 6/3-form of 2C3, which would yield a complete coding region and contained restriction sites for unidirectional insertion into the pCMV5 vector for the subsequent transfection of COS-1 cells. First strand cDNAs were generated from total liver RNA obtained from an outbred rabbit and served as templates for the polymerase chain reaction. The products were expected to include both the 6p+ and 6p-forms of 2C3.

Characterization of a Variant Form
The PCR products were isolated, ligated into pCMV, and 22 transformants of E. coli were chosen for preliminary characterization by limited restriction mapping using BamHI, BspHI, EcoRI, NheI, PpuMI, SmaI, and SpeI. Interestingly, a SpeI site was not present in four of the 22 clones, suggesting that they might encode a variant of 2C3. When the four SpeIclones were expressed in COS-1 cells, progesterone was hydroxylated a t both the 6p-and l6a-positions (Fig. l ) , indicating that these cDNAs encoded the 6/3+ form of 2C3. In contrast, cells transfected with the SpeI+ clones exhibited only 16a-hydroxylase activity as expected for 2C3 (11). If cells transfected with the latter clones exhibited GP-/lGa-hydroxylase activity ratios similar to that of the SpeI-clones, 6Phydroxylase activity would have been readily detected in these experiments based on the levels of 16a-hydroxylation detected for cultures transfected with the SpeI+ clones. Thus, the SpeI' and SpeI-clones reflect variants of 2C3 which differ in their capacity to catalyze the GB-hydroxylation of progesterone.
Complete sequence analysis of three SpeI+ and three SpeI-PCR derived cDNAs revealed six nucleotide differences between 2C3 (SpeI+) and 2C3v (SpeI-) which resulted in five amino acid differences and one silent mutation ( Table I) Forty eight hours after transfection, the medium was removed and replaced by medium containing 10 PM ['4-C]progesterone. After a 2h incubation at 37 "C, the medium was removed and analyzed. An autoradiogram of the thin layer chromatogram for one example of each is shown. The mobilities of unmetabolized progesterone and of the metabolites, 60-hydroxyprogesterone (GB-OH-P) and 16a-hydroxyprogesterone (16a-OH-P), are indicated at the right. nucleotide difference that is responsible for the loss of the SpeI site in 2C3v. All six nucleotide differences were observed in three completely sequenced SpeI-cDNAs obtained from three independent PCR reactions, whereas the SpeI+ cDNAs corresponded to 2C3. We sought to confirm the natural occurrence of all six of these nucleotide differences in partial cDNAs obtained from an independent rabbit liver cDNA library which had been generated without PCR amplification. SpeI digestion was used to distinguish the partial cDNAs corresponding to either 2C3v or 2C3. Of the four clones examined, one was found to be SpeI-. A representative clone for each cDNA was completely sequenced. All six of the nucleotide differences observed between the PCR-derived 2C3 and 2C3v cDNAs were also found to exist in the corresponding cDNAs obtained from the rabbit liver cDNA library. No additional differences between the SpeI-and SpeI+ cDNAs were noted in the 3"untranslated regions.
Heterologous Expression of 2C3u"Although the expression of the 2C3v in COS cells established that the enzyme catalyzes the 6P-hydroxylation of progesterone, it is difficult to determine the concentrations of P450 enzymes expressed in COS-1 cells and, thus, to precisely define the turnover number of the enzyme. Expression of these enzymes in E. coli, and their subsequent isolation and characterization provides a means for a more complete determination of their enzymic properties for comparison to preparations of P450 3b from rabbit liver microsomes. In addition, larger amounts of enzyme can be obtained more economically to facilitate this characterization. For this purpose, the N-terminal coding sequence of the 2C3v cDNA was modified to facilitate heterologous expression in E. coli as described earlier for 2C3 (11). The 2C3v-encoded protein expressed in E. coli was purified and reconstituted with reductase. The enzyme exhibits an apparent K,,, of 1.2 p~ and a VmaX of 5.6 pM/min/pM P450 for the 6P-hydroxylation of progesterone and a K,,, of 1.4 p M and a V,,, of 2.0 pM/min/pM P450 for the 16a-hydroxylation of progesterone. Preparations of 2C3 from rabbits that contain both the 6/3+ and 6P-forms of 2C3 exhibit a K,,, estimated to be less than 1 pM, with a Vmax in the range of 1-3 pM/min/pM P450 for 6P-hydroxylation of progesterone (6,8). Kinetic parameters for the high-efficiency 16a-hydroxylase activity catalyzed by 6P+ preparations of P450 3b purified from rabbit liver had been estimated by subtraction of the component arising from the low efficiency 6P-enzyme to yield values for K,,, of 0.3 p M and for Vmax of 0.5 pM/min/pM P450.
The inhibitor, 16a-methylprogesterone, was found to inhibit the 6p-and 16a-hydroxylation catalyzed by reconstituted 2C3v (Fig. 2). These results mirror those found for preparations of P450 2C3 from rabbit liver that contain the 6p' form and support the conclusion that 2C3v encodes the form of 2C3 catalyzing progesterone 60-hydroxylation.
When 10 p~ 5P-pregnane-3@,20a-diol was included in the reconstitution assay, the efficiency of the 16a-hydroxylation catalyzed by 2C3v was increased reflecting a lower K, for progesterone and a higher V,,,,,. A similar effect of 5P-pregnane-3@,20a-diol on the 6P-hydroxylase activity of 2C3v was also observed (Fig. 2), although this effect was relatively small and had not been reported previously for preparations of P450 2C3 from rabbit liver that contain the 6/3+ form (8).
T364S Is Necessary for the 66-Hydroxylation of Progesterone by 2C3u"Chimeric cDNAs were constructed by exchanging restriction fragments between 2C3 and 2C3v in the mammalian expression vector pCMV5 in order to identify which of the amino acid differences between 2C3 and 2C3v determine the ability of 2C3v to catalyze 6p-hydroxylation. Expression in transfected COS-1 cells combined with an in vivo progesterone assay was used to qualitatively assess the ability of each chimeric protein to catalyze progesterone 6P-hydroxylation. These constructs and their capacity to catalyze the 6P-hydroxylation of progesterone are summarized in Fig. 3. Hybrids in which Nor C-terminal regions of 2C3 (containing changes a t positions 178 and 256 or 472 and 476, respectively) were replaced with those of 2C3v did not confer 6P-hydroxylase activity. In contrast, substituting a larger C-terminal fragment of 2C3v for that of 2C3, including the S364T change, did confer the 6P-hydroxylase activity. These results indicate that the capacity to catalyze 66-hydroxylation is dependent on the presence of a threonine residue at position 364 rather than a serine. To confirm this, the reciprocal point mutations were made in 2C3 and 2C3v. The single mutation, S364T, in 2C3 confers 6P-hydroxylase activity to 2C3, whereas the reciprocal mutation in 2C3v deletes this activity.
The two single-mutant proteins, 2C3:S364T and 2C3v:T364S, were expressed in E. coli with a modified Nterminal membrane anchor domain (11) and purified for detailed kinetic characterization following reconstitution with P450 reductase. The Ser/Thr difference at position 364 not only determines the enzyme's capacity for 6P-OH, but it is also a determinant for selective inhibition by 16a-methylprogesterone. For the single mutants, this inhibitor was found to inhibit progesterone SP-and 16a-hydroxylations catalyzed by 2C3:S364T as is seen for 2C3v, but under the same conditions, it did not inhibit the 16a-hydroxylation catalyzed by the 2C3v:T364S as is observed for 2C3 (not shown). 2C3:S364T was also seen to exhibit similar values of Vmax for the 6p-and 16a-hydroxylase activities as 2C3v. However, two results indicate the involvement of other amino acids in the kinetic differences observed between 2C3 and 2C3v. The K,,, values for progesterone observed for 2C3:S364T were about %fold higher for 16a-hydroxylation and 2-fold higher for 6P-hydroxylation than the respective K,,, values obtained from 2C3v (Table 11). Moreover, the 2C3v:T364S mutant exhibited a higher catalytic activity than 2C3 over the range of substrate concentrations examined (Fig. 4). These results suggest that one or more of the four remaining amino acid differences influence the relative values of apparent K,,, for progesterone exhibited by the two enzymes.
Additional Amino Acid Differences Contribute to the Distinct Enzymic Properties of 2C3 and 2C3u"In order to identify which of the four remaining amino acid differences contribute to differences between the single mutants and the parental enzymes in their apparent K,,, for progesterone, additional chimeras were constructed and heterologously expressed in E. coli. Characterization of these chimeras indicated that the I178M difference contributes to differences in the apparent Km for progesterone between 2C3, 2C3v, and the single mutants in which residue 364 is exchanged. As shown in Fig. 4, the mutation I178M increases the catalytic efficiency of 2C3, and this single mutant (2C33178M) exhibits kinetic properties similar to that of the 2C3v:T364S mutant. Introduction of the M178I mutation into 2C3v:T364S converts the latter into an enzyme that is similar to 2C3 (Fig. 4). The reciprocal mutations in 2C3 (S364T and I178M) convert 2C3 into an enzyme which exhibits V,,, and K,,, values for both the 16aand 6P-hydroxylations of progesterone that are very similar to those exhibited by 2C3v (Table 11).

DISCUSSION
The results reported herein demonstrate that 2C3 and 2C3v are structurally highly related P450s that differ in their capacity to catalyze the 6P-hydroxylation of progesterone. The cDNA corresponding to the 6p' form (2C3v) differs a t only six nucleotide positions from the reported sequence of 2C3 (9), and these nucleotide changes result in 5 amino acid differences between 2C3v and 2C3. Analysis of hybrid enzymes expressed from chimeric constructs between the 2C3 and 2C3v cDNA in COS-1 cells indicates that the S364T difference is responsible for the phenotypic difference in the expression of progesterone 6P-hydroxylase activity between 2C3 and 2C3v. However, when the kinetic properties of the 2C3v and 2C3:S364T proteins purified from E. coli were compared, the single S364T change introduced into 2C3 did not fully mimic the enzymic characteristics of 2C3v. Although 2C3v and 2C3:S364T each had a similar Vmax for the 6P-and 16a-hydroxylase activities, the 2C3:S364T mutant displayed a higher apparent K,,, for both progesterone hydroxylase activities than 2C3v (Table 11). This ObSeNatiOn suggested that some or all of the other 4 amino acid differences between 2C3 and 2C3v contribute to these differences. This situation is also evident when comparing the 16a-hydroxylase activity of 2C3v:T364S with that of 2C3. 2C3v:T364S displays a higher catalytic efficiency than 2C3 but is not as efficient as 2C3v. Characterization of additional chimeric constructs indicates that the I178M mutation combined with the S364T mutation   M178I,T364S (V), and 2C3:1178 M (0). As observed for preparations from B/J and IIIVO/J rabbits (6)(7)(8), 2C3 cannot be readily characterized by V,, and K,,, parameters in the absence of positive effectors (11). The lines shown for 2C3 and 2C3v:M178I,T364S are linear regression lines.
The lines shown for 2C3v:T364S and 2C3:1178M are predicted from a nonlinear least squares fit of the data to the Michaelis-Menton equation. Estimated values for V,, are 31.8 and 19.8 min" and for K, are 62.8 and 45.6 PM, respectively.
can confer a K, to 2C3 that is similar to that of 2C3v. These results suggest that methionine 178 improves the enzymesubstrate interaction, whereas 6P-hydroxylase capacity is conferred by threonine 364. Amino Acid When the sequence of 2C3v is compared with that of P450cam for which a three-dimensional structure is available (24), using the sequence alignments of either Laughton et al. (13) or Nelson and Strobe1 (12), Thr3'j4 corresponds to one of the proposed substrate contact residues in P450cam, Valzg5. Atkins and Sligar (25) found that alterations of Valzg5 in P450cam to an Ile or Ala residue decreased the stoichiometry between product formation and oxygen consumption and altered the regiospecificity of product formation from camphor, 1-methylnorcamphor, and norcamphor (25). In addition, others have reported single amino acid substitutions in mammalian P450s in this region that alter catalytic activity. A L365M substitution which aligns in close proximity to T364 of 2C3v confers the coumarin hydroxylase activity of P450 2A5 to P450 2A4 (1). In addition, an I380F substitution restores the selective loss of bufuralol metabolism to a mutant form, 2Dlv, of the rat debrisoquine hydroxylase, 2D1 (2). These differences lie within SRS-5, one of six substrate recognition sites proposed by Gotoh (26) for mammalian P450s based on sequence alignments with P450cam generated from comparisons of amino acid similarity, hydrophobicity, and predicted secondary structure. Thus, our observation that an amino acid within SRS-5 underlies the difference in 60hydroxylase activity exhibited by variant forms of 2C3 and that this position occurs in close proximity to other key amino acid differences determined experimentally supports the assignment of Gotoh (26).
In contrast, the I178M difference between 2C3 and 2C3v that underlies differences between the two variants in their apparent K,,, for progesterone falls outside of SRS boundaries proposed by Gotoh (26). Ile is highly conserved at this alignment position among all members of P450 families 1, 2, 3, 17, and 21 (12). A single mutation (1172N) at this alignment position in human CYP2lA results in the loss of the steroid 21-hydroxylase activity that underlies some forms of adrenal hyperplasia (27,28). It is interesting to note that a Met is found at this position in the aromatase enzyme and among family 4A P450s (12). Based on alignments with P450cam (12, 13, 26), residue 178 falls in helix E which is far from the heme-binding region and proposed substrate pocket (12,25). We speculate that a mutation in helix E could alter its contact with helix I, resulting in a shift in the positions of the helix F-helix G loop which includes SRS-2 and SRS-3. Such a shift might contribute to the observed difference in K,.
Gotoh (26) has noted that the SRS regions of family 2 enzymes often exhibit higher frequencies of nonsynonymous substitutions when compared with other regions. In this regard, it is worth noting that five of the six nucleotide substi-