Signals for the incorporation and orientation of cytochrome P450 in the endoplasmic reticulum membrane

Cytochrome P450b is an integral membrane protein of the rat hepatocyte endoplasmic reticulum (ER) which is cotranslationally inserted into the membrane but remains largely exposed on its cytoplasmic surface. The extreme hydrophobicity of the amino-terminal portion of P450b suggests that it not only serves to initiate the cotranslational insertion of the nascent polypeptide but that it also halts translocation of downstream portions into the lumen of the ER and anchors the mature protein in the membrane. In an in vitro system, we studied the cotranslational insertion into ER membranes of the normal P450b polypeptide and of various deletion variants and chimeric proteins that contain portion of P450b linked to segments of pregrowth hormone or bovine opsin. The results directly established that the amino-terminal 20 residues of P450b function as a combined insertion-halt-transfer signal. Evidence was also obtained that suggests that during the early stages of insertion, this signal enters the membrane in a loop configuration since, when the amino-terminal hydrophobic segment was placed immediately before a signal peptide cleavage site, cleavage by the luminally located signal peptidase took place. After entering the membrane, the P450b signal, however, appeared to be capable of reorienting within the membrane since a bovine opsin peptide segment linked to the amino terminus of the signal became translocated into the microsomal lumen. It was also found that, in addition to the amino-terminal combined insertion-halt-transfer signal, only one other segment within the P450b polypeptide, located between residues 167 and 185, could serve as a halt-transfer signal and membrane-anchoring domain. This segment was shown to prevent translocation of downstream sequences when the amino-terminal combined signal was replaced by the conventional cleavable insertion signal of a secretory protein.     

Insertion of a multispanning membrane protein occurs sequentially and requires only one signal sequence

To study the insertion of multispanning membrane proteins into the endoplasmic reticulum, we constructed novel proteins on the cDNA level by repeating, up to four times, the internal signal-anchor domain of the asialoglycoprotein receptor H1. Upon in vitro translation in the presence of microsomes, these polypeptides are indeed inserted as polytopic membrane proteins. The first hydrophobic domain functions as a signal and the second as a stop-transfer sequence, while the third initiates a second translocation process, halted again by the fourth. We were able to demonstrate that insertion occurs sequentially, starting with the first apolar segment from the amino terminus. By replacing the original signal-anchor domains by a mutant sequence not recognized by signal recognition particle (SRP), it was shown that only the first hydrophobic domain needs to be a signal sequence and that the second translocation event does not require SRP.     

The amino-terminal structures that determine topological orientation of cytochrome P-450 in microsomal membrane.

We previously showed that the amino-terminal region of P-450 is responsible not only for targeting to endoplasmic reticulum (ER) membrane but also for stable anchoring to the membrane. In the present study, we introduced several mutations or deletions into the signal-anchor region of the chimeric proteins in which the amino-terminal regions of two forms of cytochrome P-450 were fused to the mature portion of interleukin 2. The amino-terminal acidic amino acid residues were replaced with basic amino acid residues or the hydrophobic core sequences were partially deleted, and these mutant proteins were assayed in vitro for their capacity to be inserted into or translocated across the ER membrane. The proteins that received the former manipulations were processed and the IL-2 portion was translocated across the membrane. In one case, the processing did not occur, thereby enabling the chimeric protein to anchor on the luminal side of the ER. Those that received the latter manipulation were also processed and the IL-2 portion translocated across the ER. These results strongly suggest that the signal-anchor function is determined both by the amino-terminal charged amino acid residues and by the length of the hydrophobic stretch.     

Cloning and expression of three rabbit kidney cDNAs encoding lauric acid omega-hydroxylases

cDNAs encoding three cytochrome P-450 enzymes were cloned from a rabbit kidney cDNA library. These three cDNAs exhibit greater than 90% nucleotide sequence identity across the coding region. This degree of sequence identity is also seen with P450IVA4, an enzyme that catalyzes the omega-hydroxylation of prostaglandins and that is elevated during pregnancy and induced by progesterone in rabbit lung. The 3' untranslated regions of the three cDNAs display very little sequence identity, suggesting that they are the products of distinct genes. The predicted amino acid sequences derived from each cDNA and for P450IVA4 exhibit about 85% identity. Each cDNA was inserted into an expression vector for transient transfection of COS-1 cells. The transfected cells each expressed a protein recognized by antibodies to P450IVA4. Microsomes isolated from the cells transfected with each cDNA efficiently catalyzed the omega-hydroxylation of lauric acid with rates that greatly exceed that catalyzed by microsomes isolated from the host cell line. One of the cDNAs encodes an enzyme that omega-hydroxylates prostaglandin A1; however, the specific activity was 2 orders of magnitude lower than that for lauric acid. Our results indicate that the substrate selectivity of the kidney P-450s encoded by these cDNAs is distinct from that of the lung P450IVA4 and that multiple enzymes comprise P-450 class IVA in the rabbit.     

Isolation and sequence analysis of three cloned cDNAs for rabbit liver proteins that are related to rabbit cytochrome P-450 (form 2), the major phenobarbital-inducible form

We have isolated from rabbit liver three cDNA clones of 1400-1800 base pairs that hybridize selectively to RNA from animals treated with phenobarbital. The nucleotide sequences of the cDNAs have been determined. In the protein coding region the nucleotide sequences of two of the cDNAs are 88% homologous, and the third cDNA is about 72-74% homologous to the other two. All three are 55-60% homologous to rat liver cytochrome P-450b cDNA. The amino acid sequences derived from the cDNA sequences are about 50% homologous to those of rat liver cytochrome P-450b and rabbit liver cytochrome P-450 (form 2). The degree of homology differs substantially in different regions of the protein. The hydrophobicity profiles of these five mammalian cytochromes P-450 are very similar and contain up to eight regions of hydrophobicity that are long enough to span a membrane. These results indicate that these three cDNAs code for rabbit liver cytochromes P-450 which are different from any rabbit liver cytochrome P-450 for which amino acid sequence information is published. These cDNAs are part of a family of genes that are related to rabbit liver cytochrome P-450 (form 2) and rat liver cytochrome P-450b which are the major phenobarbital-inducible forms. The divergence of amino acid sequence between the rat and rabbit forms and the divergence of nucleotide sequences of silent sites in the two most closely related rabbit forms suggest that cytochromes P-450 have a relatively high rate of amino acid divergence compared to many other vertebrate proteins.     

Comparison of primary structures deduced from cDNA nucleotide sequences for various forms of liver microsomal cytochrome P-450 from phenobarbital-treated rabbits

cDNA clones, termed pHP2, b32-3, b43, and b43-1, encoding cytochromes P-450 that are expressed in the liver of phenobarbital- (PB-) treated rabbits were isolated, and their nucleotide sequences were determined. pHP2 cDNA contains an open reading frame for a 490-residue protein and is a full-length counterpart of pP-450PBc2 [Leighton, J. K., Debrunner-Vossbrinck, B. A., & Kemper, B. (1984) Biochemistry 23, 204-210]. The b32-3 insert has a sequence for a protein whose primary structure is 91% similar to that of progesterone 21-hydroxylase P-450 1, though this cDNA lacks the sequence encoding the amino-terminal 110 residues. The overlapping clones b43 and b43-1 together encode an ethanol-inducible form of cytochrome P-450, though the amino-terminal five or more residues are missing in the composite b43/b43-1 sequence. Northern blot analysis showed that the b43/b43-1 protein is more strongly inducible by polycyclic aromatic hydrocarbons and isosafrole than by PB, in contrast to the case of the HP2 and b32-3 proteins. A comparison of the primary structures of eight forms of cytochrome P-450, including the HP2, b32-3, and b43/b43-1 proteins, that are expressed in the liver of PB-treated rabbits showed that 149 out of 487-492 amino acid residues are conserved in these cytochromes P-450. The eight forms can be assigned to three rabbit cytochrome P-450 gene subfamilies, P450IIB, P450IIC, and P450IIE. It was also shown that the members of the rabbit P450IIC subfamily can be further classified into three subgroups on the basis of their sequence similarity.     

Integration of membrane proteins into the endoplasmic reticulum requires GTP

We have examined the requirement for ribonucleotides and ribonucleotide triphosphate hydrolysis during early events in the membrane integration of two membrane proteins: the G protein of vesicular stomatitis virus and the hemagglutinin-neuraminidase (HN) glycoprotein of Newcastle disease virus. Both proteins contain a single transmembrane-spanning segment but are integrated in the membrane with opposite orientations. The G protein has an amino-terminal signal sequence and a stop-transfer sequence located near the carboxy terminus. The HN glycoprotein has a single sequence near the amino terminus that functions as both a signal-sequence and a transmembrane-spanning segment. Membrane insertion was explored using a cell-free system directed by transcribed mRNAs encoding amino-terminal segments of the two proteins. Ribosome-bound nascent polypeptides were assembled, ribonucleotides were removed by gel filtration chromatography, and the ribosomes were incubated with microsomal membranes under conditions of defined ribonucleotide content. Nascent chain insertion into the membrane required the presence of both the signal recognition particle and a functional signal recognition particle receptor. In the absence of ribonucleotides, insertion of nascent membrane proteins was not detected. GTP or nonhydrolyzable GTP analogues promoted efficient insertion, while ATP was comparatively ineffective. Surprisingly, the majority of the HN nascent chain remained ribosome associated after puromycin treatment. Ribosome-associated HN nascent chains remained competent for membrane insertion, while free HN chains were not competent. We conclude that a GTP binding protein performs an essential function during ribosome-dependent insertion of membrane proteins into the endoplasmic reticulum that is unrelated to protein synthesis.     

Both amino- and carboxy-terminal portions are required for insertion of yeast porin into the outer mitochondrial membrane

Yeast porin, the major outer mitochondrial membrane protein, is synthesized without a cleavable extension peptide and post-translationally inserted into the membrane. When inserted into the membrane, it acquires resistance to externally added trypsin. To locate the sequences responsible for membrane insertion and topogenesis in the primary structure of yeast porin, we constructed several deletion and chimeric mutants of the porin cDNA. These cDNAs were expressed in vitro and the products were assayed for capacity to be correctly inserted into isolated mitochondria. It was thus found that deletion of the segment spanning residues 37-98 did not appreciably impair the insertion competence and the inserted protein became resistant to trypsin. On the other hand, the porin mutant lacking the segment consisting of residues 17-98 did not acquire the trypsin resistance, though it could bind to mitochondria specifically. Deletion of the carboxy-terminal 62 amino acid residues also abolished the capacity to be correctly inserted into mitochondria. We conclude that information required for membrane insertion and intramembranous topogenesis of the porin molecule is stored not only in the amino-terminal region but also in the carboxy-terminal portion.     

Deletion analysis of the internal signal-anchor domain of the human asialoglycoprotein receptor H1.

The human asialoglycoprotein receptor H1 is a single-spanning membrane protein with the amino terminus facing the cytoplasm and the carboxy terminus exposed on the exoplasmic side of the plasma membrane. It has been shown earlier that the transmembrane segment, residues 38-65, functions as an internal signal directing protein synthesis to the endoplasmic reticulum and initiating membrane insertion. This process is co-translational and mediated by signal recognition particle (SRP). To identify subsegments within this region containing the signal information, we prepared deletion mutants at the level of the cDNA and analysed them in a wheat germ in vitro translation system with microsomes as the target membrane. Insertion and membrane anchoring were judged by the glycosylation of the protein, its resistance to exogenous protease and the extent to which it can be extracted from the microsomes by alkaline treatment. It was found that very small deletions already reduce the stability of membrane anchoring. However, nearly half of the transmembrane domain can be deleted, both from the amino-terminal and from the carboxy-terminal side, without completely abolishing membrane insertion. Several mutants, although not inserted, still interact with SRP. The results support the notion that the main feature of a signal sequence is a hydrophobic stretch of sufficient length (10-12 residues in our sequence), and indicate that recognition by SRP is not sufficient for membrane insertion.     

Synthesis and insertion, both in vivo and in vitro, of rat-liver cytochrome P-450 and epoxide hydratase into Xenopus laevis membranes

We described whole cell and cell-free systems capable of inserting into membranes cytochrome P-450 and epoxide hydratase made under the direction of rat liver RNA. The systems have been used to study the pathways followed by newly made secretory and integral membrane proteins. The cell-free system contains Xenopus laevis embryo membranes, and demonstrates competition for a common receptor between cytochrome P-450 and epoxide hydratase, and normal secretory proteins: evidence is provided for differential membrane receptor affinity. Thus, synthesis of secretory and membrane proteins appears to involve a common initial pathway. Microinjection of rat liver RNA into whole oocytes suggests that membrane insertion is neither cell type nor species specific, because functional rat liver enzymes are found inserted in the endoplasmic reticulum of the frog cell. Nonetheless, insertion is highly selective since albumin and several other proteins made under the direction of the injected liver RNA are sequestered within membrane vesicles and are then secreted by the oocyte, whilst epoxide hydratase and cytochrome P-450 are inserted into membranes but are not secreted.     

Cloning and sequence analysis of a rat liver cDNA coding for a phenobarbital-inducible microheterogenous cytochrome P-450 variant: regulation of its messenger level by xenobiotics

A rat liver cDNA library was prepared from total polyribosomal poly(A)+ RNA extracted from phenobarbital-treated animals. A cDNA clone coding for a phenobarbital-inducible cytochrome P-450 (PB P-450) was identified by differential colony hybridization to cDNAs synthesized from liver poly(A)+RNAs isolated from phenobarbital-treated rats for positive selection and cDNAs from either untreated rats or beta-naphthoflavone-treated rats as negative controls, followed by hybrid-selected translation and analysis of the translation products by immunoprecipitation. As the cloning and screening strategies involve no prior enrichment for specific mRNAs, they also permit the identification of sequences coding for phenobarbital-induced proteins other than cytochromes P-450. This relatively straightforward approach is generally applicable to the molecular cloning of sequences coding for other inducible cytochromes P-450. Nucleic acid sequencing data indicated that the cloned PB P-450 cDNA codes for a cytochrome P-450 variant [designated P-450e(U.C.)] that is very similar, but not identical, to P-450e. Sequence analysis of the section of cDNA specifying the 3'-non-coding region of the mRNA revealed that it lacked the usual poly(A) addition site signal sequence but contained three inverted repeat structures. Solution hybridization analysis demonstrated that PB P-450 mRNA is increased 20-fold by phenobarbital treatment and decreased 3-fold by beta-naphthoflavone treatment.     

Segmental homologies in the coding and 3' non-coding sequences of rat liver cytochrome P-450e and P-450b cDNAs and cytochrome P-450e-like genes

The nucleotide sequence of a cloned cDNA insert carried by pHDQ14 was determined and found to code for the 107 C-terminal amino acids of rat liver cytochrome P-450e. Comparison of the pHQ14 cDNA sequence with those of cloned cDNAs for cytochrome P-450b and of 2 P-450e-like genes revealed segmental homologies that may have resulted from gene conversion. These results suggest that gene conversion may generate sequence variants of genes for rat liver cytochrome P-450s.     

The influenza hemagglutinin insertion signal is not cleaved and does not halt translocation when presented to the endoplasmic reticulum membrane as part of a translocating polypeptide

The co-translational insertion of polypeptides into endoplasmic reticulum membranes may be initiated by cleavable amino-terminal insertion signals, as well as by permanent insertion signals located at the amino-terminus or in the interior of a polypeptide. To determine whether the location of an insertion signal within a polypeptide affects its function, possibly by affecting its capacity to achieve a loop disposition during its insertion into the membrane, we have investigated the functional properties of relocated insertion signals within chimeric polypeptides. An artificial gene encoding a polypeptide (THA-HA), consisting of the luminal domain of the influenza hemagglutinin preceded by its amino-terminal signal sequence and linked at its carboxy-terminus to an intact prehemagglutinin polypeptide, was constructed and expressed in in vitro translation systems containing microsomal membranes. As expected, the amino-terminal signal initiated co-translational insertion of the hybrid polypeptide into the membranes. The second, identical, interiorized signal, however, was not recognized by the signal peptidase and was translocated across the membrane. The failure of the interiorized signal to be cleaved may be attributed to the fact that it enters the membrane as part of a translocating polypeptide and therefore cannot achieve the loop configuration that is thought to be adopted by signals that initiate insertion. The finding that the interiorized signal did not halt translocation of downstream sequences, even though it contains a hydrophobic region and must enter the membrane in the same configuration as natural stop-transfer signals, indicates that the HA insertion signal lacks essential elements of halt transfer signals that makes the latter effective membrane-anchoring domains. When the amino-terminal insertion signal of the THA-HA chimera was deleted, the interior signal was incapable of mediating insertion, probably because of steric hindrance by the folded preceding portions of the chimera. Several chimeras were constructed in which the interiorized signal was preceded by polypeptide segments of various lengths. A signal preceded by a segment of 111 amino acids was also incapable of initiating insertion, but insertion took place normally when the segment preceding the signal was only 11-amino acids long.(ABSTRACT TRUNCATED AT 400 WORDS)     

Critical region in the extension peptide for the import of cytochrome P-450(SCC) precursor into mitochondria

In order to establish the role of the extension peptide of the precursor of P-450(SCC), a mitochondrial inner membrane protein, in the import into the organella, three deletion mutants of the precursor, in which the deletions were in the mature portion, were constructed. These mutant precursors were imported into mitochondria in vitro as efficiently as the original precursor, indicating that the extension peptide contains sufficient information for the import of the precursor into mitochondria. To investigate which portion of the extension peptide contains the mitochondrial targeting signal, various lengths of the amino-terminal portion of the extension peptide of P-450(SCC) precursor were fused to the mature portion of adrenodoxin. The fusion proteins consisting of 44 and 19 amino-terminal amino acids and mature adrenodoxin were imported into mitochondria, whereas those containing 14, 7, and 2 amino-terminal amino acid residues were not. The importance of the amino-terminal portion of the extension peptide was confirmed by the deletion from the amino-terminal end of a fusion protein consisting of the amino-terminal 44 amino acid residues of P-450(SCC) precursor and mature adrenodoxin, SCC44RAd. The amino-terminal deletions abolished the import of the fusion proteins into mitochondria. Substitution of all of the three basic amino acids, Arg(4), Arg(9), and Lys(14) in the extension peptide of SCC44RAd to Ser or Thr inhibited the binding of the fusion protein to mitochondria as well as its import.(ABSTRACT TRUNCATED AT 250 WORDS)     

Aberrant membrane insertion of a cytoplasmic tail deletion mutant of the hemagglutinin-neuraminidase glycoprotein of Newcastle disease virus.

The hemagglutinin-neuraminidase (HN) protein of Newcastle disease virus (NDV) is a type II glycoprotein oriented in the plasma membrane with its amino terminus in the cytoplasm and its carboxy terminus external to the cell. We have previously shown that the membrane insertion of HN protein requires signal recognition particle SRP, occurs cotranslationally, and utilizes the same GTP-dependent step that has been described for secretory proteins, type I proteins, and multispanning proteins (C. Wilson, R. Gilmore, and T. Morrison, Mol. Cell. Biol. 7:1386-1392, 1987; C. Wilson, T. Connolly, T. Morrison, and R. Gilmore, J. Cell Biol. 107:69-77, 1988). The role of the amino-terminal cytoplasmic domain in the faithful membrane insertion of this type II protein was explored by characterizing the membrane integration of a mutant lacking 23 of the 26 amino acids of the cytoplasmic domain. The mutant protein was able to interact with SRP, resulting in translation inhibition, membrane targeting, and membrane translocation, but the efficiency of translocation was considerably lower than for the wild-type HN protein. In addition, a significant proportion of the mutant protein synthesized in the presence of SRP and microsomal membranes was associated with the membrane in an EDTA- and alkali-insensitive manner yet integrated into membranes with its carboxy-terminal domain on the cytoplasmic side of membrane vesicles. Membrane-integrated molecules with this reverse orientation were not detected when the mutant protein was synthesized in the absence of SRP or a functional SRP receptor. Truncated mRNAs encoding amino-terminal segments of the wild-type and mutant proteins were translated to prepare ribosomes bearing arrested nascent chains. The arrested mutant nascent chain, in contrast to the wild-type nascent chain, was also able to insert into membranes in a GTP- and SRP-independent manner. Results suggest that the cytoplasmic domain plays a role in the proper membrane insertion of this type II glycoprotein.     

cDNA cloning and functional expression in yeast Saccharomyces cerevisiae of beta-naphthoflavone-induced rabbit liver P-450 LM4 and LM6

A cDNA library was constructed from liver mRNA of a beta-naphthoflavone-induced rabbit. Two clones pLM4-1 and pLM6-1 containing 2.2-kbp inserts that hybridized at low stringincy with a mouse P1 P-450 probe were selected. The clone pLM4-1 was fully sequenced and found to contain a full-length cDNA coding for cytochrome P-450 LM4. Partial sequence and restriction mapping made it possible to identify pLM6-1 as coding for the major part of cytochrome P-450 LM6. Cloned LM4-1 cDNA was reformed by deletion of the 5' and 3' non-coding regions before insertion into yeast expression vectors PYe DP1/10. A similar operation was performed on pLM6-1 cDNA after replacement of the missing N-terminus-coding sequences by homologous sequences form the pLM4-1 clone resulting in a chimeric cytochrome P-450 coding sequence. Expression of cloned rabbit cytochrome P-450 into transformed yeast was optimized by studying the effect of the nature of the DNA sequence just preceding the initiation codon on the level of cytochrome P-450 production. Yeast synthesized cytochromes P-450 were characterized by immunoblotting, spectra and catalytic activity determinations. Cloned cytochrome P-450 LM4 was found by all criteria to be identical to the authentic rabbit one. The chimeric cytochrome P-450 that contains the 143 N-terminal amino acids of cytochrome P-450 LM4 and the remaining 375 amino acids of cytochrome P-450 LM6 was found to exhibit most of the authentic cytochrome P-450 LM6 catalytic properties. Enzymatic and evolutionary implications of these results are discussed.     

A tripartite structure of the signals that determine protein insertion into the endoplasmic reticulum membrane

Multilineage colony stimulating factor is a secretory protein with a cleavable signal sequence that is unusually long and hydrophobic. Using molecular cloning techniques we exchanged sequences NH2- or COOH-terminally flanking the hydrophobic signal sequence. Such modified fusion proteins still inserted into the membrane but their signal sequence was not cleaved. Instead the proteins were now anchored in the membrane by the formerly cleaved signal sequence (signal-anchor sequence). They exposed the NH2 terminus on the exoplasmic and the COOH terminus on the cytoplasmic side of the membrane. We conclude from our results that hydrophilic sequences flanking the hydrophobic core of a signal sequence can determine cleavage by signal peptidase and insertion into the membrane. It appears that negatively charged amino acid residues close to the NH2 terminal side of the hydrophobic segment are compatible with translocation of this segment across the membrane. A tripartite structure is proposed for signal-anchor sequences: a hydrophobic core region that mediates targeting to and insertion into the ER membrane and flanking hydrophilic segments that determine the orientation of the protein in the membrane.     

Alcohol-inducible cytochrome P-450IIE1 lacking the hydrophobic NH2-terminal segment retains catalytic activity and is membrane-bound when expressed in Escherichia coli

We have expressed in Escherichia coli a cDNA encoding rabbit liver cytochrome P-450IIE1, the ethanol-inducible P-450. The expressed P-450 is located primarily in the bacterial inner cell membrane and comprises 3% of the E. coli total membrane protein. The partially purified cytochrome exhibits a reduced CO difference spectrum with a maximum at 452 nm, characteristic of P-450IIE1, and solubilized membranes or partially purified P-450 preparations reconstituted with NADPH-cytochrome P-450 reductase and phosphatidylcholine catalyze the deethylation of N-nitrosodiethylamine with a turnover number equal to that of purified liver P-450IIE1 (approximately 4.5 nmol/min/nmol of P-450). A modified IIE1 cDNA that encodes a protein lacking amino acids 3-29, a proposed membrane anchor for cytochrome P-450, was also expressed in E. coli and, unexpectedly, the shortened protein was also found to be predominantly located in the bacterial inner membrane rather than the cytosol. Like the full-length protein, this truncated cytochrome has a reduced CO difference spectrum characteristic of P-450IIE1 and is fully active in the deethylation of N-nitrosodiethylamine. These results demonstrate that the NH2-terminal hydrophobic segment is not solely responsible for attachment to the membrane and evidently is not required for proper protein folding or catalytic activity.     

Complementary DNA and amino acid sequence of rat liver microsomal, xenobiotic epoxide hydrolase

The coding nucleotide sequence for rat liver microsomal, xenobiotic epoxide hydrolase was determined from two overlapping cDNA clones, which together contain 1750 nucleotides complementary to epoxide hydrolase mRNA. The single open reading frame of 1365 nucleotides codes for a 455 amino acid polypeptide with a molecular weight of 52,581. The deduced amino acid composition agrees well with those determined by direct amino acid analysis of the rat protein, and the amino acid sequence is 81% identical to that of rabbit epoxide hydrolase. Analysis of codon usage for epoxide hydrolase, and that of rabbit epoxide hydrolase. Analysis of codon usage for epoxide hydrolase, and comparison to codon usage for NADPH-cytochrome P-450 oxidoreductase and cytochromes P-450b, P-450d, and P-450PCN, suggest that epoxide hydrolase is more conserved than cytochromes P-450b and P-450PCN; comparison of the extent of sequence conservation for 12 homologous proteins between the rat and rabbit, including cytochrome P-450b, supports this hypothesis, and indicates that much of epoxide hydrolase is constrained to maintain its hydrophobic character, consistent with its intramembranous location. The predicted membrane topology of epoxide hydrolase delineates 6 membrane-spanning segments, less than the 8 or 10 predicted for two cytochrome P-450 isozymes; the lower number of membrane-spanning segments predicted for epoxide hydrolase correlates with its lesser dependence on the membrane for maintenance of its tertiary structure and catalytic activity.