Different structural requirements at specific proline residue positions in the conserved proline-rich region of cytochrome P450 2C2.

Cytochrome P450 is anchored to the endoplasmic reticulum membrane by an N-terminal transmembrane sequence with the catalytic domain facing the cytoplasmic side. Within the peptide sequence linking these two domains is a highly conserved proline-rich region. In cytochrome P450 2C2, this region has the sequence 30PPGPTPFP37. To examine the structural requirements at these proline residues, each proline was replaced with alanine, glycine, valine, or an acidic amino acid, and the activities of the mutated proteins were determined in transfected COS-1 cells. Lauric acid 1ω-hydroxylase activities of Pro30 and Pro33 mutants were less than 10% of wild type for each substitution except for alanine, which was 25-30%. In striking contrast, substitutions at Pro31, including an acidic residue, did not substantially alter activity. At positions 35 and 37, acidic amino acid substitutions reduced activity to less than 10% of wild type while substitution of the other three amino acids had little effect. The tolerance of substitutions of charged residues at Pro31 suggests that the side chain at this position is exposed to a polar environment; conversely, the reduced activity with charged substitutions, but not with uncharged substitutions at positions 35 and 37, suggests that these residues are exposed to a hydrophobic environment, presumably within the folded protein. The loss of activity with substitutions at Pro30 and Pro33 implies that the motif PXXP is important for the formation of a functional cytochrome P450 and that this sequence might have a helical structure with a repeat of three, as in the left-handed poly-L-proline II helix. Insertion of alanine between positions 29 and 30 did not substantially affect activity, but insertions between either 33 and 34 or 37 and 38 resulted in activity less than 25% of wild type. These data indicate that the position of PXXP, relative to the sequence flanking it on the C-terminal side, may be important for its function.

Cytochrome P450 is anchored to the endoplasmic reticulum membrane by an N-terminal transmembrane sequence with the catalytic domain facing the cytoplasmic side. Within the peptide sequence linking these two domains is a highly conserved proline-rich region. In cytochrome P450 2C2, this region has the sequence 30 PPG-PTPFP 37 . To examine the structural requirements at these proline residues, each proline was replaced with alanine, glycine, valine, or an acidic amino acid, and the activities of the mutated proteins were determined in transfected COS-1 cells. Lauric acid 1-hydroxylase activities of Pro 30 and Pro 33 mutants were less than 10% of wild type for each substitution except for alanine, which was 25-30%. In striking contrast, substitutions at Pro 31 , including an acidic residue, did not substantially alter activity. At positions 35 and 37, acidic amino acid substitutions reduced activity to less than 10% of wild type while substitution of the other three amino acids had little effect. The tolerance of substitutions of charged residues at Pro 31 suggests that the side chain at this position is exposed to a polar environment; conversely, the reduced activity with charged substitutions, but not with uncharged substitutions at positions 35 and 37, suggests that these residues are exposed to a hydrophobic environment, presumably within the folded protein.
The loss of activity with substitutions at Pro 30 and Pro 33 implies that the motif PXXP is important for the formation of a functional cytochrome P450 and that this sequence might have a helical structure with a repeat of three, as in the left-handed poly-L-proline II helix. Insertion of alanine between positions 29 and 30 did not substantially affect activity, but insertions between either 33 and 34 or 37 and 38 resulted in activity less than 25% of wild type. These data indicate that the position of PXXP, relative to the sequence flanking it on the Cterminal side, may be important for its function.
Cytochrome P450 (P450) 1 refers to a superfamily of proteins that metabolize a wide variety of endogenous and xenobiotic compounds and are located in the mitochondria or microsomes of mammalian cells (1,2). The microsomal forms are integral membrane proteins that are targeted to the ER membrane by a hydrophobic N-terminal sequence (3)(4)(5)(6). This sequence also acts as a stop-transfer signal and anchors the protein to the membrane so that the catalytic domain of the protein is on the cytoplasmic side of the membrane. The sequence immediately following the N-terminal domain, which links it to the cytoplasmic domain, may play an important role in the generation of a functional P450 by initiating the folding of the protein during synthesis at the interface of the membrane and cytoplasm. In addition, P450 must interact with its redox partner, P450 reductase, which is also an integral membrane protein, and with hydrophobic substrates possibly entering the protein from the membrane (7). A linker region may be required to allow the catalytic domain to achieve its proper orientation in the membrane for these interactions. While x-ray structures of four bacterial P450s have been useful in guiding and interpreting mutagenic studies probing the structural elements important for substrate recognition and catalytic activity (8), these structures are not as useful for studying structural aspects of the P450s important for membrane interactions since the bacterial P450s are soluble proteins.
P450 2C2 is present in the liver microsomes and catalyzes the hydroxylation of fatty acids (9). The sequence of P450 2C2 following the N-terminal hydrophobic anchor contains a Glyrich region that is not conserved among P450s and a Pro-rich region that is conserved. The sequence PPGP is highly conserved with the identical sequence present in family 2 P450s. The function of these Pro residues is not known, but they could be important for a turn out of the membrane, which determines the orientation of the protein relative to the membrane, for the folding or incorporation of heme in newly synthesized P450, or for the stability or catalytic activity of the mature protein.
Mutations at the equivalent of the first Pro in this sequence reduces the activity of P450 21A2 resulting in nonclassical adrenal hyperplasia, and analogous mutations reduced the activity of human P450 2D6 (10,11). In P450 2C2, deletion of the PPGP sequence did not alter the location of the protein in the ER membrane but eliminated enzymatic activity in transfected COS-1 cells (12). Substitutions of Ala for the first Pro in the PPGP sequence of P450 2C11, or for combinations of two or three Pro residues including the first or last Pro in each case, resulted in mutant proteins that were expressed in yeast but that did not exhibit the characteristic difference spectrum at 450 nm (13). These data indicated that the mutant proteins did not fold to form a functional P450 and supported the conclusion that this region was important for the folding of the protein or the incorporation of the heme moiety.
To examine the role of the Pro-rich region further, each of the Pro residues in this region was replaced with either Ala, Gly, Val, or an acidic residue, and the activities of the mutants were determined. The results demonstrate different structural requirements at specific Pro residue positions and suggest that a PXXP motif may be important for the formation of a functional P450 2C2.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Construction of plasmid pc2A containing P450 2C2 cDNA in pTZ18R has been described (4). Single strand DNA mutagenesis was performed as described (14). The following oligonucleotides were used as primers, and uracil-containing single-stranded DNA from plasmid pc2A grown in Escherichia coli strain CJ236 was used as template: Pro 30  In order to construct Pro mutants in the pCMV5 mammalian expression vector, each pc2A-pTZ mutant was digested with KpnI and ApaI, and the isolated fragment containing a mutation in the Pro-rich region was ligated to the pc2A-pCMV vector cut with KpnI and ApaI. The mutations were confirmed by sequencing the KpnI-ApaI segment.
Cell Culture and Assay of Laurate Hydroxylase Activity-COS-1 cells were cultured and transfected with plasmid DNA as described (15). Laurate hydroxylase activity was assayed in whole cell lysates of transfected COS-1 cells, and lauric acid metabolites were separated by high performance liquid chromatography as described (15) except that the reaction was incubated for 30 min.
Immunoprecipitation of Expressed Proteins-Forty-eight h after transfection, cells were incubated for 4 h with 50 Ci/ml Tran 35 S-label in Met-and Cys-free minimal essential medium. Cells were lysed, and radioactive proteins were immunoprecipitated from cell lysates and analyzed by SDS-PAGE as described previously (15).

Effect of Mutations on the Level of Expression in Transfected
Cells-The Pro-rich region of P450 2C2 contains five Pro residues in the sequence from amino acids 30 -37 ( Fig. 1). In order to examine the functional importance of the Pro residues in this region, each Pro was replaced with Ala, Gly, Val, and either Asp or Glu depending on the third base of the codon in the cDNA sequence. The four amino acid residues were chosen for the following reasons. Ala has a small side chain and is uncharged, properties that make it a "neutral" substitution in mutagenesis; Gly, like Pro, is often present in turns rather than ␣-helical structures but results in less stable structures; Val is a small hydrophobic amino acid with a tertiary ␤-carbon, which decreases the flexibility of the peptide backbone; and Asp and Glu are charged amino acids that dramatically change the hydrophobicity of the side chain. Each of the mutants was transfected into COS-1 cells, and lauric acid hydroxylase activ-ity in cell lysates was determined.
Any of the mutations could alter activity of the P450 expressed in COS-1 cells by affecting the rate of synthesis or degradation of the protein. To examine the levels of the proteins expressed in the COS-1 cells, transfected cells were labeled for 4 h with Tran 35 S-label, and P450 2C2 and the mutant proteins were immunoprecipitated and analyzed by SDS-PAGE. In mock transfected cells, only a weak band comigrating with P450 was detected, and a second protein migrating more rapidly was present, which served as a useful internal control for labeling and immunoprecipitation (Fig. 2). Since the halflife of the P450 expressed in COS-1 cells is less than 1 h (15), labeling the cells for 4 h provides a reasonable measure of the steady-state levels of the protein. Differences of only 2-fold or less in the amount of radioactive protein immunoprecipitated for any of the mutants and wild-type P450 2C2 were detected in 2 or 3 independent experiments. Therefore, the changes in activity of the mutant P450s cannot be explained by different levels of protein expressed in COS-1 cells. While these experiments detect P450 apoprotein, it is possible that the folding, heme incorporation, or stability of functional forms of the protein might be affected by the mutations. Attempts to express the P30A mutant, which has decreased activity in COS-1 cells, in E. coli in order to distinguish among these possibilities have been unsuccessful using conditions that allow expression of wild-type P450 2C2 and other P450 2C2 mutants that appear to have reduced stability. 2 Laurate 1-Hydroxylase Activity of Single Pro Mutations in the Pro-rich Region-Within the highly conserved PPGP sequence, the effects of the mutations at Pro 30 and Pro 33 on lauric acid hydroxylase activity were similar and were remarkably different from effects of mutations at Pro 31 (Fig. 3). At Pro 30 , no activity was detected with Gly, Val, or Asp substitutions, and less than 10% activity compared with wild type was detected with these substitutions at Pro 33 . Ala substitutions at these two positions were tolerated slightly better with 25-30% activity of wild type. These results indicate that the structural requirements for Pro at these positions are very stringent since the Gly substitution, which should be tolerated if only a turn structure is required, reduced activity. In striking contrast, all of the substitutions at Pro 31 were well tolerated including the acidic residue, Asp, with a relatively large charged side chain, suggesting that the side chains at this position are on the surface and pointing toward the aqueous environment. In contrast, the side chains of positions 30 and 33 may be involved in packing of the PPGP peptide against another peptide structure.
Substitutions at Pro 35 and Pro 37 exhibited a third pattern, with Ala, Gly, or Val mutations being well tolerated and Asp or Glu mutations reducing activity to 15% or undetectable for 2 B. Doray, C. Chen, and B. Kemper, unpublished data. Pro 35 and Pro 37 , respectively. The reduced activity with charged substitutions, but not with uncharged substitutions, suggests that these Pro residues are in a relatively hydrophobic environment, very likely buried inside the molecule. Alternatively, the size of Glu may sterically alter the structure, but this would also suggest that the side chain is packed against other parts of the protein structure.
Laurate 1-Hydroxylase Activity of a Secondary Mutant Restoring a PXXP Motif-The effects of substitutions at Pro 30 , Pro 31 , and Pro 33 suggest that a PXXP motif is important in the structure-function relationship of P450 2C2. If the PXXP motif itself is important for activity, then re-creating a PXXP motif in the proteins with mutations at Pro 30 or Pro 33 might restore activity. To accomplish this, a double mutant, P30A/T34P was constructed in which Pro at positions 31 and 34 provide a PXXP motif (Fig. 1). The activity of this double mutant, however, was similar to the P30A mutation alone so that activity was not restored by the second mutation (Fig. 4). This result might be explained either by the fact that the position of the PXXP was moved one residue relative to its flanking regions or by some restrictions on the amino acids between or immediately flanking the two Pro residues.
Laurate 1-Hydroxylase Activity of Ala Insertion Mutants Flanking the PXXP Motif-In order to determine whether the position of the PXXP motif was critical, Ala was inserted on either the N-or C-terminal sides of the PXXP motif (Fig. 1). Insertion of an Ala between residues 29 and 30 on the Nterminal side of the PXXP only slightly reduced activity to about 75% of wild type (Fig. 4). On the other hand, insertions of Ala between residues 33 and 34 or 37 and 38 reduced activity to less than 25% of the activity of wild type, similar to the Ala substitutions at Pro 30 and Pro 33 . These results suggest that the position of the PXXP motif relative to flanking sequence on the C-terminal side is important for the formation of functional P450 2C2, while the position relative to N-terminal sequences is not.

Natural Variants of the PXXP motif in Microsomal P450 -
The sequence PPGP is present in P450s of widely divergent organisms, ranging from plants to mammals as shown in the alignment of selected sequences in Table I. PPGP is completely conserved in family 2 P450s, and PPGP, or a variant PXXP, is present in most other mammalian P450s (Table II). Another variant, I/LPGP, is also commonly present, particularly in P450s 1A2, 3A, and 19. To examine whether Ile or Leu can substitute for the first Pro in the PPGP motif, mutants P30I and P30L were constructed. Replacing Pro 30 with either Ile or Leu resulted in less than 10% activity of wild type (Fig. 4). Similarly an Leu mutation at this position in human P450 21A2 resulted in activity about 30% of normal for progesterone hydroxylation and represents a potential nonclassical steroid 21-hydroxylase deficiency allele (10). These data suggest that the first Pro in the PXXP motif of P450s cannot be replaced by amino acids present at this position in other P450s without substantial loss of activity and provide further evidence that Pro 30 is important for formation of a fully functional P450 2C2. If the PXXP motif packs against other peptide segments of the P450, then the requirement for Pro at position 30 may be altered by different sequences in the interacting regions of those P450s with Ile or Leu at the first Pro position. DISCUSSION Mutations of the Pro residues in the Pro-rich region of P450 2C2 indicate that these residues can be divided into two structural regions, the highly conserved PPGP sequence and the more C-terminal Pro 35 and Pro 37 . The latter Pro tolerated uncharged amino acid substitutions, but not charged substitutions, which is consistent with their location in a hydrophobic environment within the folded protein. In contrast, the PPGP sequence appears to be exposed on the surface of the protein or as a linker between the membrane anchor and catalytic domain since the second Pro could be replaced by both hydrophobic and charged amino acids without loss of activity. The presence of Pro and Gly residues in this sequence suggests a loop or turn structure for this region, which might be required for the proper orientation of the catalytic cytoplasmic domain relative to the membrane anchor. However, while PG is a common sequence at positions 2 and 3 of a type II ␤-turn, Pro at position 4, as in PPGP, is very rare (16). Furthermore, the inactivity of mutants with Gly substitutions at positions 30 and 33 indicates that the function of the Pro is not simply to allow a bend and that either increased stability produced by a Pro or a structure different from a turn is required.
The decreased activity of all the mutants with substititions at the Pro 30 and Pro 33 positions suggests that a PXXP motif is critical for formation of a functional P450 2C2. The presence of PPGP in P450s of widely divergent organisms and the conservation of a PXXP motif in many others (Tables I and II) provide further evidence that this motif plays an important role in many P450s. A common functional role associated with a PXXP motif has been protein-protein interactions. A PX 2 PXGX 3 PP motif was found in the N-terminal domain of the vesicle-associated membrane protein/synaptobrevin, which is involved in vesicle docking and/or membrane fusion processes through protein-protein interaction (17). A novel protein binding module, the WW domain, which has a role in mediating protein-protein interactions via Pro-rich sequences containing PXXP motifs, has been characterized (18). Interactions of Src homology 3 (SH3) domains with Pro-rich sequences play important roles in many intracellular signaling pathways (19,20). A Pro-rich sequence was found in the prolactin receptor that mediated association of this protein to Jak2 kinase (21). Mutagenesis of a PPXP sequence in an interleukin-5 receptor showed that signal transduction mediated by this receptor required the first and last Pro, but not the second (22), exactly the requirements of the P450 2C2 PPGP sequence for activity. Interactions of a synthetic Pro-rich polypeptide, containing a PPXP sequence, with SH3 ligands were critically dependent on the two end position Pro. In addition, flanking sequences outside of the Pro-rich core increased binding affinity and specificity (23). The distance of the PXXP motif from these flanking amino acids was fixed so that PXXP motifs incorrectly spaced did not mediate high affinity binding. Similarly, in the present study, insertion of Ala to the C-terminal side of the PPGP sequence resulted in reduced activity suggesting that the position of the PXXP motif relative to a flanking sequence on the C-terminal side was important for its function.
The requirements for a PXXP motif are consistent with a helical structure for this sequence with three amino acids/ helical turn as is present in right-handed 3 10 or left-handed PPII helices (24). The 3 10 helix is stabilized by hydrogen bonding between residues 1 and 4, which is prevented if these two residues are Pro as in PPGP (25). In contrast, the structures of Pro-rich polypetides, with PPXP sequences, complexed with SH3 domains have been shown to be PPII helices (26), and a PPII helical structure is present in the HIV viral protein Nef before interaction with a SH3 domain (27). Thus, if the PXXP requirement reflects a helical structure for PPGP in P450 2C2, it is likely a left-handed PPII helix as illustrated in Fig. 5. Most of the PPII helices are a maximum of four residues, and this structure is compatible with immediate transitions to ␣-helical or ␤-sheet structures at either end (24). If PPGP assumes this structure, the side chains of Pro 30 and Pro 33 would point in the same direction and that of Pro 31 would point 120°in the opposite direction (Fig. 5). The effects of the mutations indicate that the Pro 31 helix face would be exposed to the aqueous environment while the Pro 30 and Pro 33 face would interact with or pack against other peptide structures (Fig. 5).
The functional significance of a PXXP requirement is not clear but could involve intra or interprotein interactions that are important for folding of newly synthesized P450, for heme incorporation, or for stability or catalytic activity of the folded protein. Mutations at the first or last Pro positions in the PPGP motif of P450 2C11 either alone or in combination with mutations at other Pro positions in the Pro-rich region resulted in mutant proteins that were expressed in yeast but were inactive (13). None of the proteins exhibited the characteristic reduced CO difference spectrum at 420 or 450 nm of the P450 hemoprotein, indicating that heme was not incorporated into the protein, and it was concluded that these Pro residues were  important in folding of the molecule. Further support for malfolding was the observation that the inactive proteins segregated to a separate ER compartment compared to wild-type P450 when expressed in COS-7 cells (28). Our inability to express the P30A mutation in E. coli is also consistent with a defect in folding of the mutant protein. Since PXXP motifs are known to participate in high affinity protein-protein interactions, the PXXP motif may form a stable interaction with another segment in the nascent P450 polypeptide chain, or another protein, which is required for the folding of the rest of the protein as it is synthesized. In P450s without the PXXP motif, secondary mutations in the target sequence may result in a stable interaction, or other stabilizing interactions may substitute for the PXXP interaction. Although the evidence favors a role for the PPGP region in folding or assembly of P450, the critical evidence that the mutated proteins are not expressed in a hemoprotein form in yeast is not conclusive since folding and assembly in the envi-ronment of the yeast cell may be different from that in the mammalian cell. Alternatively, the PXXP requirement could reflect the necessity of this sequence to pack against the P450 molecule, and mutations might affect the stability or local conformation of the folded protein. For example, according to the alignment by Hasemann et al. (8), the PXXP motif is located adjacent to the AЈ ␣-helix that contains amino acids that are part of a postulated site for the initial interaction of the protein with its substrate (29 -31). Alteration of the PXXP motif could indirectly alter the position of the AЈ ␣-helix and affect substrate interaction. Additional experiments will be needed to distinguish definitively among the possible effects of these mutations. FIG. 5. The PPGP motif modeled as a left-handed PPII helix. A, structure with the axis of the helix parallel to the paper; B, structure with the helical axis nearly perpendicular to the paper. The gray and black shading are to aid in visual interpretation only. C, schematic representation of the PPGP PPII helix. A PPII helical structure is consistent with the requirement for Pro at 30 and 33 and the tolerance for substitutions of amino acids, including acidic residues, at 31. The PPII helix is a left-handed helix with a repeat of three amino acids per turn and, thus, forms a triangular structure with side chains pointing in three directions as indicated by the arrows in panel B. The critical Pro 30 and Pro 33 are on one side of the helix and form a hydrophobic protrusion that might provide specificity or steric surfaces important for formation of functional P450. Alternatively, packing along the surface formed by Gly 32 , Pro 30 , and Pro 33 , as illustrated in panel C, might occur.