DNAVolume 6, Number 6, 1987Mary Ann Liebert, Inc., PublishersPp. 515-517

MEETING

Cytochrome P450

OLIVER HANKINSON*

The genes for cytochrome P450s constitute a super-family encoding heme-thiolate proteins involved in the

metabolism of a vast range of foreign chemicals (activatingsome to carcinogenic and toxic derivatives, and inacti-vating others), and endogenous compounds, such as ste-roid hormones. The transcriptional rate of many P450genes has been shown to be increased by their substrates,hormones, or other inducers. The superfamily is, there-fore, of interest from the viewpoints of molecular evolu-tion, toxicology, carcinogenesis, endocrinology, and generegulation. The following are some of the highlights of theSecond International Workshop on P450 Gene Regulationthat was held in Airlie, Virginia, April 5-8, 1987, under theauspices of the National Institute of Child Health and Hu-man Development.The nomenclature for the P450s has hitherto been very

confusing. The sequences of many P450 proteins, genes,and cDNAs have now been determined, and this encour-

aged several of the prominent workers in the field to agreeon a systematic nomenclature based on sequence homol-ogy and presumed evolutionary relationships. The pro-posed system assigns the approximately 60 known P450s to10 families. The system will undoubtedly have to be ad-justed as additional P450s are discovered and sequenced.Nevertheless the basic structure of the nomenclature is inplace (Nebert et al., DNA 6, 1-11, 1987).

P4501A1 (previously known as P,450, P450c, or form 6)is directly involved in the activation of polycyclic aromatichydrocarbons to carcinogenic intermediates. This P450 ishighly inducible by polycyclic hydrocarbons and by2,3,7,8-tetrachlorodibenzo-p-dioxin (dioxin). Considerableprogress in defining regulatory sequences for the P4501A1gene was reported. There are at least two dioxin-responsive

"Laboratory of Biomédical and Environmental Sciences, Uni\i Department of Physiology and Biophysics, University of Uli:

REPORTGene Regulationand BYRON KEMPERt

domains upstream from the mouse P450IA1 cap site, andalso there is evidence for the existence of a repressor-re-sponsive domain located between the dioxin domains andthe cap site. There is no sequence homology between thetwo mouse dioxin domains (J.P. Whitlock, Stanford;D.W. Nebert, Bethesda). Fujii-Kuriyama's group (Tokyo,Japan) provided evidence for three domains responsiblefor dioxin induction in the rat gene. A common short se-

quence (xenobiotic responsive element, or XRE) wasfound associated with the dioxin domains. Interestingly,the XRE sequences bear homology to the glucocorticoidresponsive enhancer consensus sequence. Characterizationof the repressor-responsive site upstream from the ratP450IA1 gene, if indeed such a site exists, has not yet beencarried out.Clones carrying mutations in the P450IA1 gene have

been isolated from cultured cells. The nucleotide altera-tions in two of these mutants were reported by D.W. Ne-bert and O. Hankinson (Los Angeles). The analysis ofadditional mutants derived in vivo should complement theresults obtained from mutants of the P450IA1 gene gener-ated in vitro in providing insight into the mode of action ofthe P450IA1 protein.The aromatic hydrocarbon (Ah) receptor mediates in-

duction of the P450IA1 gene by polycyclic hydrocarbonsand dioxin. It was reported that mutations in yet a thirdgene can abolish function of the Ah receptor in culturedcells (Hankinson). In contrast, only one gene has beenidentified that affects functioning of the apparently verysimilar glucocorticoid receptor. A fragment of one of theabove three genes has been cloned.The barbiturates were some of the earliest-recognized in-

ducers of P450. Induction of drug metabolism by these

iity of California at Los Angeles, Los Angeles, CA 90024.is, Urbana, IL 61801.

515

516 HANKINSON AND KEMPKR

agents can be the basis for human drug interactions as ex-

emplified by the interaction between barbiturates and oralanticoagulants. Compared to the polycyclic hydrocarbonsystem, progress in understanding the mechanism of thebarbiturate induction has been slower, in part because a

continuous cell line that retains expression and inducibilityof the phenobarbital-responsive genes is not available. Incontrast to the polycyclic hydrocarbon-inducible family,which contains only two known inducible genes in mostspecies, the phenobarbital-inducible genes are members ofa large family (P450II) that includes five subfamilies. Mostof the phenobarbital-inducible genes are present in sub-families P450IIB and P450IIC. The P450IIC subfamilycontains at least five to seven members in rodents and rab-bits, but the subfamily appears to be much less complex inhuman beings (C.R. Wolf, Edinburgh) and chickens (B.May, Adelaide), as determined by Southern blotting. It isnot yet clear whether comparable differences in complexityare present at the functional P450 protein level or whethersome of the genes may be pseudogenes. In addition to

phenobarbital-inducible forms, the P450IIC subfamily in-cludes constitutive enzymes (E. Johnson, La Jolla; B. Kem-per, Urbana) and sex-dependent enzymes that are regulatedby growth hormone (J.-A. Gustafsson, Huddinge). F. Gon-zalez (Bethesda) reported the cloning of rat cDNAs or genesfor members of the P450IIA subfamily (testosterone 7a-hy-droxylase), and the P450IIE subfamily (alcohol inducible).Induction of gene transcription is the principal mecha-

nism of barbiturate induction in both the P450IIB (Atchin-son and Adesnik, J. Biol. Chem. 258, 11280-11284, 1983;Hardwick et al., J. Biol. Chem. 258, 8081-8085, 1983;Pike et al., Biochem. Pharmacol. 34, 2489-2494, 1985) andP450IIC (Kemper) subfamilies. In chickens two P450IICcDNAs have been cloned and the mRNAs are inducedabout 70-fold by 2-allyl-2-isopropylacetamide (a barbitu-rate-like inducer), providing an excellent system to studythe induction process. Using oligonucleotide probes to dis-tinguish between rat P450IIB1 and P450IIB2 (98% homol-ogous), C. Omiecinski (Seattle) showed that developmentalregulation of these isozymes was qualitatively similar, butsome quantitative differences were noted. This approachmay be useful for sorting out the regulation of the otherclosely related members of the P450II family. Taking a ge-netic approach, M. Negishi (Research Triangle Park) re-

ported that a P450 (detected with a rat P450IIB1 cDNAprobe) catalyzing the 16a hydroxylation of testosteronewas present in high levels and was not inducible by phéno-barbital in female BALB/cJ mice, but was present in lowlevels in female 129/J mice and was inducible to levelscomparable to those of the BALB/cJ mice. Since in un-

treated animals this isozyme is repressed in female 129/Jmice but is expressed in female BALB/cJ mice, Negishisuggested that phénobarbital acts by a derepression mecha-nism. A single locus (Rip) on chromosome 7 regulates therepression in 129/J mice and an association of the locuswith the derepression was demonstrated.M. Waterman (Dallas) discussed the regulation of ste-

roidogenic forms of P450 and presented evidence for threemajor types of regulation of these P450s. The mRNAs offour P450s involved in the synthesis of glucocorticoids in

the adrenal are all increased by ACTH through a cAMP-dependent mechanism which probably involves a labile ste-roid hormone-inducing protein (SHIP), since cyclohexi-mide blocks the effect. Similar putative labile factors were

reported for the induction of P450XIXAl(aromatase)(E.R. Simpson, Dallas) and the glucocorticoid-inducibleP450s (Guzelian, Richmond), but W.L. Miller (San Fran-cisco) reported that cycloheximide did not inhibit cAMPstimulation of P450XXIIA1 (P450scc) mRNA in humangranulosa cells. In human and sheep fetuses, high levels ofsteroid hydroxylase mRNA and protein are found at a timewhen ACTH is undetectable, indicating a cAMP-indepen-dent mechanism (Waterman). Tissue-specific regulationwas also observed. The regulation of these steroidogenichormones is clearly complex and multifactorial.Because of their activities toward a broad range of hy-

drophobic compounds, the P450s could potentially be usedfor eliminating environmental pollutants and for drug syn-thesis. In an attempt to obtain elevated P450 activity inyeast, T. Sakaki and co-workers (Hyogo, Japan) fused thegenes coding for P450 reducíase and rat P450IA1 (the for-mer is essential for activity, as it transfers electrons to thelatter). Yeast cells transformed with the fused gene exhib-ited about a fourfold greater activity than those trans-formed with the rat P450IA1 gene alone (DNA 6, 189-196,1987). A. Fulco (Los Angeles) described a P450 of Bacillusmegaterium in which both activities naturally reside on thesame molecule. This fused enzyme system has a catalyticcenter activity of 5000 min"1,, which is several orders ofmagnitude greater than that for known mammalian P450s.A second interesting property of this P450 is that it is in-ducible by barbiturates. Since the mechanism involved inthe induction of the mammalian P450s remains obscure,further studies on the B. megaterium enzyme, which hasnow been cloned and expressed in Escherichia coli, mayprovide considerable insight into this problem. Althoughexpression of mammalian P450s in bacterial cells has notyet been reported, S. Sligar (Urbana) reported the expres-sion in E. coli of large amounts of the hemoproteins, cyto-chrome ¿>s, and myoglobin. Total genes for these proteinswere synthesized, and the selection of high-frequency bac-terial codons was important for efficient expression.

Several human genetic polymorphisms have been de-scribed. If the polymorphic P450 is involved in biosynthe-sis of endogenous compounds such as hormones or in me-

tabolism of toxic chemicals, dramatic clinical conse-

quences can result. The best understood of these is a defi-ciency in the functional P450XXIA2 gene which catalyzesthe C21 hydroxylation of 17-0//-progesterone in the Corti-sol biosynthetic pathway. About 25% of these deficiencieshave been thought to be associated with the deletion of theP450 gene, but W.L. Miller made the controversial sugges-tion that these apparent deletions were due to gene conver-

sions or polymorphism and that loss of DNA was not ap-parent if analyses with multiple restriction enzymes were

performed. P. White (Cornell) reported on two patientswith point mutations resulting in a substitution at isoleu-cine-172 (which is strongly conserved in other P450s) inone case and a termination codon at 318 in the other. In-terestingly, these mutated nucleotides are normally present

CYTOCHROME P450 MEETING 517

in P450XXIA1, a closely linked pseudogene ofP450XXIA2, so that the mutations in the functional genesmay have been introduced by a gene conversion event.Debrisoquine 4-hydroxylase activity is polymorphic in

rats and humans. In humans the metabolism of multipledrugs is affected by the polmorphism and high activity hasbeen associated with enhanced susceptibility to broncho-genie carcinoma, as well as cancer of the liver and gastro-intestinal tract. F. Gonzalez has cloned cDNAs for twogenes, termed P450IID1 and P450IID2, of the probablefour active genes in the rat and has cloned the orthologoushuman P450IID1 gene. In the rat, the polymorphism is notassociated with changes in mRNA levels and, in strainswith low activity, is probably due to a structural differencein the protein. The human P450IID1 protein is 73% simi-lar in amino acid sequence to the rat P450IID1 protein. Inthree individuals with low debrisoquine 4-hydroxylase ac-

tivity, characterization of the human cDNA indicated thatseveral defective P450 mRNAs occur as the result of aber-rant splicing of the pre-mRNA.The complexities of analyzing multiple closely related

P450 genes was demonstrated by E. Johnson's (La Jolla)studies of polymorphism in rabbit P450I1C5, a liver C21hydroxylase different from the adrenal C21 hydroxylasediscussed above. A restriction fragment length polymor-phism (RFLP) associated with the changes in activity wasdetected with cDNA probes of P450I1C5. However, offour closely related genomic clones, the only cloned genewith available sequence in the exon 1 region identical toP450I1C5 did not correspond to the genomic DNA exhibit-ing the RFLP.The process by which P450s are inserted into the mem-

brane is known to be dependent on the signal recognitionparticle (SRP). Furthermore, the amino termini of P450sresemble hydrophobic signal sequences of secreted pro-teins, but also resemble membrane anchor domains, sincethis region is more hydrophobic and longer than the typi-cal signal sequence. M. Adesnik (New York) and B. Kem-per (Urbana) showed that the amino termini of particularP450s, when substituted for the signal peptide of normallysecreted proteins, did not support translocation of the hy-brid protein across the microsomal membranes, but actedas SRP-dependent combination insertion-halt transfer sig-

nals. R. Sato (Osaka) noted that his laboratory also hadevidence supporting this conclusion. Replacing the singlenegative charge at the amino terminus with two positivecharges apparently selectively eliminated the halt transfersignal since the mutant hybrid protein was translocatedacross the membrane (Kemper). In reciprocal experimentsin which the P450 amino terminus was replaced by thegrowth hormone signal, translocation of the hybrid pro-tein was stopped only by one area of the P450 molecule(amino acids 165-185) (Adesnik). This constitutes a poten-tial second anchor segment, although it is not knownwhether it functions as one in vivo. This work and the re-cent demonstration that antigenic determinants other thanthose at the amino terminus of P450 are on the cytoplas-mic side of the endoplasmic reticulum (Delemos-Chiaran-dini el al., J. Cell Biol. 104, 209-219, 1987) indicate thatonly the amino terminus may penetrate through the mem-brane and that the P450 molecule does not loop throughthe membrane several times as predicted from the hydrop-athy profiles.The characterization of multiple new P450s reported at

this meeting extended the major theme of the First Interna-tional Workshop. It is clear that the study of this complexfamily of enzymes may provide insight into a variety offields. Understanding of the regulation of these genes is atan early stage, and it is likely that c/s-acting and trans-act-ing regulatory elements will be defined in the next fewyears. Expression of the mammalian proteins in microbialorganisms will likely lead to new understanding of the en-

zymatic mechanisms of these enzymes, both through phys-icochemical analysis and in vitro mutagenic studies. Howthis complex group of enzymes with multiple substratespecificities and many different physiological regulatorscould have evolved from a presumed single gene remains a

fascinating question.

Address reprint requests to:Byron Kemper

524 Burrill HallUniversity of Illnois

407 S. Goodwin A ve.

Urbana, IL 61801