Proc. Nadl. Acad. Sci. USAVol. 83, pp. 2340-2344, April 1986Biochemistry

Harvey ras genes transform without mutant codons, apparentlyactivated by truncation of a 5' exon (exon -1)

(in vitro recombination/DNA transfection/nudeotide sequence analysis/ras protooncogene activation)

KLAUS CICHUTEK AND PETER H. DUESBERGDepartment of Molecular Biology and Virus Laboratory, University of California, Berkeley, CA 94720

Communicated by Harry Rubin, December 11, 1985

ABSTRACT The hypothesis is tested that the ras gene ofHarvey sarcoma virus (Ha-SV) and the proto-ras DNAs fromcertain tumor cells derive transforming function from specificcodons in which they differ from normal proto-ras genes.Molecularly cloned Harvey proviral vectors carrying viral ras,normal rat proto-ras, and recombinant ras genes in which thevirus-specific ras codons 12 and 59 were replaced by proto-rasequivalents each transformed aneuploid mouse 3T3 cells afterlatent periods that ranged from 4 to 10 days. Viruses with orwithout virus-specific ras codons all transformed diploid ratcells in 3-5 days equally well. However, in the absence of virusreplication, mutant codons were beneficial for transformingfunction. Deletion of non-ras regions of Ha-SV did not affecttransforming function. We conclude that specific ras codonsare not necessary for transforming function. Comparisons ofthe ras sequences of Ha-SV, BALB SV, and Rasheed SV withsequences of proto-ras genes from rat and man revealed anupstream proto-ras exon, termed exon -1. The 3' end of thisexon is present in all three viruses and in a ras pseudogene ofthe rat. Since ras genes transform without mutation and sinceexon -1 is truncated in viral ras genes and all transformingproto-ras DNAs of the Harvey and the Kirsten ras family, wepropose that ras genes are activated by truncation of exon -1either via viral transduction or artificially via cloning andtransfection. The proposal implies that untruncated proto-rasgenes with point mutations may not be cellular cancer genes.

Harvey sarcoma virus (Ha-SV) causes sarcomas anderythroleukemias in mice and rats and, in culture, transformsdiploid cells from many vertebrate species, including man(1-3). The viral ras gene is an autonomous and dominantoncogene that initiates and maintains transformation in onestep (2-4). It encodes a protein of 189 amino acids, termedp21, that is colinear with a p21 coding sequence of cellularproto (p)-ras genes (5, 6). Therefore, viral ras and cellular rasgenes are assumed to be isogenic (5-14).The apparent paradox that the viral protein has obligatory

transforming function, whereas the cellular protein does not,was initially explained in terms of a quantitative hypothesis(5, 6). This hypothesis suggests that the increased level ofp21in virus-infected cells causes malignant transformation. Theobservation that synthetic, virus-type p-ras genes, construct-ed from Ha-SV promoters and p21 coding regions of molec-ularly cloned normal rat and human p-ras, can transformaneuploid mouse NIH 3T3 cells (6, 7) is in accord with thishypothesis. A competing hypothesis that is currently favoredsuggests that viral and cellular ras genes owe transformingfunction to specific ras codons that differ from normal p-rascodons. The basis for this hypothesis is that p21-encodingp-ras DNA species or clones from certain rodent or humantumors have dominant transforming function, when trans-fected into cells, due to one of at least 50 different mutations

in at least five different ras codons, notably codon 12,whereas equivalent normal DNAs do not (8-18, 42). Thecoincidence that the 12th ras codons of Ha-SV (19), Rasheed(Ra) SV (20) and BALB murine SV (21) each differ from the12th codon ofnormal human (8-10) and rat (22) p-ras and thatthe 12th ras codon of the related Kirsten (Ki) SV also differsfrom Ki-p-ras (23-25) further supported the mutational ras-activation hypothesis. However, the paradox that a normalcellular gene would be activated to a dominant cancer gene byso many different point mutations although mutations typicallyare silent or inactivating (26) remained unresolved. It is alsounclear why normal p-ras would transform under some (6, 7)but not other conditions (8-18).

RESULTSTransforming Function of ras Genes Without Mutant

Codons. A comparison of the nucleotide sequence of Ha-SV(19) with the sequence ofnormal rat p-ras-1 (ref. 22; M. Ruta,personal communication) shows that ras codons 12 (Arg vs.Gly in p-ras), 59 (Thr vs. Ala), and 122 (Cys vs. Ala) areHa-SV-specific; the comparison also defines the 5' and 3'borders of the ras sequence of Ha-SV at positions 917 and1832, for a total of915 nucleotides (numbered as in ref. 19 andFig. 1). The ras sequence is embedded in sequences derivedfrom a 30S defective rat retrovirus (30S DRV) (27, 28), whichin turn is embedded in terminal elements from Moloneyleukemia virus (5, 19) (Fig. 1).

In preliminary experiments, the role of 30S DRV se-quences and of noncoding ras regions on transformingfunction ofHa-SV was tested with deletion mutants preparedfrom infectious proviral DNAs, cloned in the plasmidpBR322. One ofthese, pA, is colinear with the viralRNA (27)and is flanked by two LTRs (ref. 29; P. Tambourin and D.Lowy, personal communication). The 7.8-kb viral insertbegins at a HindIII site at position 3495 of the viral genome,spans the complete provirus, and terminates at a BamHI siteat position 378 (Fig. 1). The other, pH-1, is permuted withregard to an EcoRI site at position 2540 (19, 29). One suchdeletion mutant, pAASc, lacks a Sac I-resistant region fromposition 814 to 1044 of Ha-SV that includes most rassequences 5' of the p21 coding region (construction in legendto Fig. 1). Two others lack 1.7 kb of noncoding ras and 30SDRV sequences 3' of the p21 region, between two Pst I sitesat positions 1759 and 3436, or 2.5 kb of DRV sequence 3' ofras, between the two HincII sites at positions 2168 and 4795(Fig. 1). They were constructed by digesting the pH-1 clonewith Pst I (29) or HincIl. The digested DNAs were directlytransfected into 3T3 cells. Each deletion mutant generatedfoci of transformed 3T3 cells in 4 to 6 days (see below).Indeed, foci generated by the Sac I deletion mutant werebetter delineated than wild-type-transformed controls.Virus recovered from transfected cells transformed rat pri-mary cells. Thus, the viral ras gene is not dependent on

Abbreviations: p-ras, proto-ras; SV, sarcoma virus; Ha, Harvey; Ki,Kirsten; Ra, Rasheed; DRV, defective rat retrovirus; LTR, longterminal repeat; kb, kilobase(s); bp, base pair(s).

2340

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natd. Acad. Sci. USA 83 (1986) 2341

cowxo~~~w_c1_In_____ec en

N

0o C

(8amis=3CLA is 8 .5Am PICHA 'xO&X-zmxd I? a,

I

pA 12-1/-2

pA 1259-1 ___

~'. cwr o-cy

= -= z9

3' rat p-ras 1

pA 1259-2ASc

IK ~_pA pras 1-X

900 . . . . 950 . (ace)hp-ras X-1 Barn... ( 205 codons) ... Coo gag tot cog cog co0 gtg co0 tgc a T cgp-ras-2 ogoo aa age co cooo cococooooccooo g c c coog ag cg oag c oa g g SaCII a cg c c ag tcg cgHaSV HaSV tgt get tat goo ogo age cog age ogo ace ogo ogo gga egg age cca tgc gog gac cca gtc ggc

RasVBaBy aco t gog

hp-ras X-1 a a 0

hp-ras XIp-ras-2 gtc ao goap-raa-1

HaSY

1000RaSY....... oto ogg agg t

t g g c

(goo~otg)go g t

a

t tg a gtgagtg... SDSA...gocotcottggeag Xl g

1050

ao

aa

(gga 0cc ogg goo)ta ggaT g c tga ga

S...attotcattggoag Xl SacI a

ace cgt co0 ego ace ogo cot goo cog goo cog goo cog ggg goa gto gog coa goa age ggt ggg gea aga get act ggt ttg goa gee cot gta gaa gog

FIG. 1. (Upper) Restriction enzyme sites (referred to in the text) of the 5579-nucleotide sequence of Ha-SV (19) and of rat p-ras-1 (5-7). Themap of the Ha-SV sequence in pA shows Moloney leukemia virus-derived (stippled), 30S DRV-derived (hatched), and p-ras-derived coding(elevated) and noncoding sequences (both white). The map ofp-ras-1 (black) shows the coding region (elevated) ofthe four known exons. pA12-1and -2 were constructed by replacing a viral ras region between the HindIHl site at position 1088 and the Pvu II site at 1139 with an equivalentfrom rat p-ras-1 or -2. For this purpose, a viral region from the BamHI site S' of ras at position 409 to an artificial Sal I site within ras of a mutantprovirus (29) at position 1435 was subcloned into a pMLd plasmid that lacks a HindIII and a Pvu II site (30). After replacing the virus-specificHindIII-Pvu II region by the p-ras-1 or -2 equivalent, a 0.9-kilobase (kb) mutant HaSV ras region between BamHI at 409 and Fsp I at 1290 wasexcised from this clone. This piece was then ligated with a 6.1-kb Sal I-BamHI pA fragment that includes the 5' long terminal repeat (LTR)and a 4.9-kb Sal I-Fsp I pA fiagment that includes the 3' LTR, to generate pA12-1 or -2. pA1259-1 was constructed by cloning a 2.2-kbviral-ras-containing region, extending from BamHI at 409 to EcoRI at 2540, into the pMLd vector. A 0.8-kb viral ras region of this clone, from SacII at position 940 to Nhe I at 1695, was exchanged for the p-ras-2 equivalent. The 0.9-kb BamHI-Fsp I ras region of pA was then replaced by therecombinant equivalent from this clone to construct pA1259-1 as described above. To construct pA1259-2 (not shown), the 2.2-kbBamfl-EcoRI viralras region from pA12-2 was subcloned in pMLd. A 0.6-kb Pvu ll(1139)-Nhe I (1695)-resistant viral ras fragment was then exchanged for a p-ras-2

equivalent, and the 0.9-kb BamHI-Fsp I ras fragment of this clone was introduced into pA as described above. pA1259-2ASc was constructed as

described above except that a BamHI-Fsp I pA1259 ras fiagment was used from which a region between two Sac I sites at positions 814 and 1044was deleted. The Sac I region was deleted from a viral BamHI-EoRI region cloned in pMLd. pAprasl-X was constructed by replacing a 0.9-kbHindIU(1088)-Xba I(2023)-flanked viral region from within the aboveBamHI-EcoRI subclone with a 2-kbHindIl-Xba Ifagment ofp-ras-1. A 2.8-kbBamHI-Xba I fiagment from this construct was then ligated with a 5.7-kb Spe I-HindIH fragment of pA containing the 3' LTR and a 2.3-kbHindlI-BamHI fragment ofpA containing the 5' LTR, to generate pAprasl-X. Preparation ofDNA fiagments, subsequent purification by Elutip(Schleicher & Schuell) chromatography, and ligation followed published procedures (31). (Lower) The nucleotide sequence 5' of the p21 codingsequence ofHa-SV, from position 900 to 1075 (numbered as in ref. 19) is compared to ras sequence equivalents ofRa-SV (20), BALB SV (21), human(h) p-ras (15, 32), rat p-ras-1, and rat pseudogene p-ras-2 (refs. 5-7, 22; M. Ruta, personal communication; P. Seeburg, K.C. and P.H.D., unpublisheddata). X-1 and X1 are exons -1 and 1; SA and SD are splice acceptor and donor sites.

noncoding ras and 30S DRV sequences for transformingfunction.To test the role of Ha-SV-specific ras codons on trans-

forming function, we have constructed recombinant virusesfrom pA, in which Ha-SV-specific ras codons were replacedby equivalents from normal rat p-ras DNA. In pA12-1 and -2the 12th ras codon was replaced by the equivalent of normalrat p-ras-1 (Fig. 1) or the rat pseudogene p-ras-2, both clonedin pBR322 (6, 7). In two others, pA1259-1 and -2, the 12th and59th ras codons were replaced by p-ras-2 equivalents (con-struction in legend to Fig. 1). Both ofthese recombinants alsocontained the p-ras-2-specific codons 34 (Ser) and 53 (Met),instead ofthe viral and p-ras-1 equivalents which are Pro andLeu (ref. 19; M. Ruta, personal communication). pA1259-1differs from pA1259-2 in that it contains a 5' noncoding rasregion from p-ras-2 (Fig. 1). To verify exchange of ras codon12 in these recombinants, we tested for a Mae I-sensitiveCTAG sequence at position 1107 that is characteristic ofwild-type Ha-SV (19). Viruses with the p-ras-specific codon12 (Gly) contain the Mae I-resistant CTGG sequence (22).

Finally, a recombinant was constructed, pAprasl-X, inwhich the viral ras gene was entirely replaced by the ratp-ras-1 equivalent. For this purpose, proviral DNA from a

HindIII site at the fifth ras codon to a Spe I site near the 3'LTR, at positions 1088 and 4732, was replaced by a 2-kb

HindIII-Xba I region of rat p-ras-1 that includes the probablepolyadenylylation signal of rat p-ras-1 (Figs. 1 and 3; M.

Ruta, personal communication). The viral non-ras regionlacking in this recombinant was shown above to be dispens-able for transforming function.Recombinant proviral clones were transfected into mouse

NIH 3T3 cells together with cloned proviral DNA of helperMoloney leukemia virus (33) by the calcium phosphateprecipitation method (34, 35). Four to six days aftertransfection with clones pA12-1 or -2 or wild-type pA, >1000foci of transformed cells per jug of DNA were observed(Table 1). In the absence of helper virus, focus formation wasdown by a factor of 100 for pA and by a factor of 1000 forpA12, and the latent period was 2-3 weeks (Table 1). Thetransforming efficiency ofrecombinants with p-ras codons 12and 59 (pA1259-1 and -2) was lower by a factor of 3-10, andthe latent period was slightly longer than those ofpA or pA12(Table 1). However, the pA1259-2ASc deletion mutant,which lacks the same noncoding ras region 5' of the p21coding region as the above-described pAASc, transformed3T3 cells essentially as well as did pA or pA12 (Table 1). ThepAprasl-X recombinant generated about 10% as many foci asdid the wild-type virus, after latent periods of 10-14 days.Mouse 3T3 and primary Fischer rat embryo cells were

transformed within 3-5 days by virus present in undiluted

-EZZ0 ..

Biochemistry: Cichutek and Duesberg

2342 Biochemistry: Cichutek and Duesberg

media from 3T3 cells transformed by wild-type or recombi-nant proviral DNA. Optimally transforming virus stockswere obtained by biological cloning, from cultures derivedfrom foci of virus-transformed 3T3 cells. 3T3 cells trans-formed by virus from pA1259-2 and rat cells transformed byvirus from pA1259-2ASc and pAprasl-X are shown in Fig. 2.Foci of pAprasl-X-transformed rat cells had either round orfusiform morphology (Fig. 2 C and D). 3T3 cells transformedby any, and rat primary cells transformed by most, recom-binant Ha-SVs were morphologically indistinguishable fromcells infected by the wild type. In some experiments, rat cellstransformed by A1259-1 and -2 virus were individually morespindle-shaped and, as a group, more parallel to each otherthan cells transformed by wild-type virus (data not shown).Comparison of p-ras and Viral ras Sequences Reveals an

Upstream p-ras Exon. In comparing various ras sequences,we detected a common 5' border ofhomology of both rat (22)and human (9, 15, 32) p-ras with Ha-SV (19), BALB SV (21),Ra-SV (20), and the rat pseudogene p-ras-2, at position 1035(Figs. 1 and 3). For this matching, an allowance offour humanp-ras-specific triplets was made at position 1056. Upstream ofposition 1035, the three viruses and the pseudogene sharevarious amounts of additional ras sequences with each other(Figs. 1 and 3). Moreover, positions 1035 of human and ratp-ras-1 are preceded by exactly the same canonical spliceacceptors (Fig. 1). Thus, this comparison suggests position1035 as the 5' border of human and rat p-ras exon 1 andsignals the existence of an upstream p-ras exon from whichthe common viral and rat pseudogene sequences 5' ofposition 1035 must have been derived by the same splicingevents. The postulated upstream exon must be conserved inrats and mice, since Ha-SV (5) and Ra-SV (20) originatedfrom rats, and BALB SV (4) from a mouse. Indeed, theavailable sequence data on human p-ras show a sequence thatfits the description of an upstream exon 1040 bp 5' of humanp-ras exon 1 (15, 32) (Fig. 3). The 3'-terminal nucleotides ofthis sequence (upstream of position 1034 in Fig. 1) overlapwith the 5'-terminal 144 nucleotides of the known sequenceof rat p-ras-2 and the 5' 117, 54, and 24 ras nucleotides ofHa-SV, BALB SV, and Ra-SV, respectively (Figs. 1 and 3).The human homolog differs from rat p-ras-2 and the viralequivalents only in scattered point mutations, in two specifictriplets at position 1006 and in another at position 967 (Fig. 1).Upstream of position 890 (Fig. 1) there are only a few

Table 1. Transformation of mouse NIH 3T3 cells by molecularlycloned ras genes, including Harvey viral ras, normal rat p-ras,and recombinants in which virus-specific (v) ras codons arereplaced by p-ras (p) equivalents.

LatentFoci, no period,

Clone 12 59 122 Mo-MuLV per Ig dayspA v v v + 1000-2000 4

v v v - 25/40 14/21pA12-1 p v v + 1000-2000 4

p v v - 2/4 14/21pA1259-1 p p v + 100-200 6-8pA1259-2 p p v + 300-400 5-6pA1259-2ASc p p v + 1000-1500 4pAprasl-X p p p + 140-160 10-14About 2 x 106 NIH 3T3 cells in a 10-cm tissue culture dish were

transfected with 1-2 Ag of ras genes cloned in the Harvey proviralvector, with or without 0.5 Aug of Moloney murine leukemia provirus(Mo-MuLV) DNA clone ZAP (33), in the presence of 10,ug of salmonsperm DNA (34, 35). Two days later the cells were transferred 1:4.Medium with 10% fetal bovine serum was initially changed daily andat 1- or 2-day intervals after 5 days posttransfection. Foci thatappeared after latent periods of 4-21 days are recorded; numbersreflect the ranges of two or more experiments.

FIG. 2. (A) Mouse NIH 3T3 cells (from S. Aaronson, NationalCancer Institute) transformed by Ha-SV with p-ras specific codons12 and 59 from clone pA1259-2. (B-D) Primary cultures of Fischer ratembryo cells (Cell Culture Facility, University of California, SanFrancisco) transformed by virus from clone pA1259-2ASc (B) or byvirus with the complete p21 coding unit of normal rat p-ras-1 fromclone pAprasl-X [C (round) andD (fusiform)]. Sparsely seeded cells(10) in a 5-cm tissue culture dish were infected by incubation for 24hr, in the presence of 1 ug of Polybrene per ml of medium, with 1-2ml of filtered (0.45 ,um) medium from 3T3 cells that had beentransformed by the respective proviral DNA (Table 1). Cells werepropagated in Dulbecco's modified Eagle's medium supplementedwith 10%6 heat-inactivated (30 min, 56°C) fetal bovine serum. Photoswere taken 7 days after infection. (x 10.)

scattered regions of homology between human p-ras and ratp-ras-2 (the p-ras-2 sequence upstream of the Sac I site atposition 1050 is unpublished data from P. Seeburg, K.C. andP.H.D.; downstream from the Sac I site, it is from M. Ruta). As

Human proto-rasorf

4,Bam HiK SD

StopjLAUG (Start)1040 bp SA -2T-3

I Stop

X4

poly (A)

4,AUG (Start ?) IStop

Rat proto-ras SA

Start JAUG I Stop

Rasheed SV \\a-

N\AUG (Start) lStopBalb SV

\:AUG (Start) tStop

Harvey SV 0

JAUG (Start 7)Rat pseudo-ras

FIG. 3. Relationship between the human and rat p-ras genes andthe ras sequences of Ha-SV, Ra-SV, and BALB SV and the ratpseudogene p-ras-2. X- 1 to X4 are p-ras exons. Agag is a deletionmutant ofthe gene encoding the viral group-specific antigen. SA andSDare splice acceptorand donor sites. The p21 protein is encoded by p-ras,Ha-SV and BALB SV; orf is an open reading frame of 205 codons.

a I-..

Proc. Nad. Acad Sci. USA 83 (1986)

p21

Proc. Natl. Acad. Sci. USA 83 (1986) 2343

expected for an exon, the human sequence is followed by acanonical splice donor, shown at position 1035 in Fig. 1.We deduce that rat, mouse, and human p-ras share, in

addition to the four exons defined previously, at least oneconserved exon, which we refer to as p-ras exon -1 (Figs. 1and 3). The existence of such an exon in human p-ras hasbeen considered previously (15). Since all three viral rasgenes contain terminal elements of this exon, they must havebeen generated by recombination between this exon and therespective progenitor viruses.

DISCUSSIONIs Transforming Activity of ras Genes due to Truncation of

p-ras Exon -1? Our data show that specific mutant codonsare not essential for transforming function of p21-encodingHa-ras genes. Further, we demonstrate that p-ras exon -1 istruncated in all ras-containing SVs, as well as in character-ized p-ras DNAs with transforming function (see below).Therefore, we propose that ras genes are activated bytruncation of p-ras exon -1 or possibly other upstreamexons. In the case of the viral ras genes, this truncation wasaccomplished by selective transduction, and in the case ofp-ras DNAs with transforming function, by fragmentation viatransfection or molecular cloning.Here we itemize examples of how the proposal fits our

results and those of others. (i) There is an apparent discrep-ancy between the results that unmutated p21 has transform-ing function, as reported here for rat and previously for ratand human p-ras (refs. 6 and 7; P. Tambourin and D. Lowy,personal communication), and the results that it does not(8-17). We suggest that the negative results reflect therelative inefficiencies of the assay systems used. The repli-cating retroviral vector system used by us not only allowsoptimal recovery of transfected DNA but also provides foroptimal expression and transmission by virus spread. Forinstance, omission of the helper-virus DNA upontransfection reduced focus-forming activity of wild-typeHa-SV by a factor of 100 and transformation by Ha-SVwithout virus-specific codon 12 by a factor of about 1000(Table 1). Thus, <1% of cells transfected become stablytransformed in the absence of virus replication. Moreover, avisible focus appears 4-6 days after transfection, due to virusspread, compared to 2-3 weeks after transfection withouthelper virus. It follows that transforming activity of ras innonreplicating vectors (single provirus per cell) is at least 10times more dependent on point mutation than in replicatingvectors (multiple proviruses per cell). Thus, the more inef-ficiently ras is expressed, the more it benefits from mutationfor transforming activity. The work ofothers is in accord withthis conclusion. For example, when linked to a nonretroviralpromoter or to a retroviral promoter (LTR), the 6.6-kbBamHI-resistant human p-ras DNA with a mutant codon 12and its unmutated equivalent each generate colonies of "im-mortalized" rat primary cells with equal efficiencies. Thesecolonies were selected by drug-resistance genes cotransfectedwith ras. However, only the colonies generated by the mutatedras linked to the LTR are reported to be morphologicallytransformed (16). We propose that all other ras-immortalizedcells are transformants with a fusiform morphology like thoseshown in Fig. 2D. The negative results reported for other clonesofnormal human p-rasDNA could then be negative because noreplicating vectors or enhancers were used and because therewas no selection for ras-transfected cells by drug resistance(8-11, 13). It is unclear, however, why one other study failed todetect transforming function with unmutated but truncatedp-ras in a retroviral vector similar to ours (12).

(ii) It is improbable that reduced transformation efficiencyof some mutant viruses reflects back-mutations, becausesome mutants retain genetically stable markers that affect thephenotype and the kinetics of transformation and because the

frequency of mutation in retroviruses [10-_-10-4 per nucle-otide per replicative cycle (36)] is much too low to account forthe 30%- to 90%-reduced transforming efficiencies of somemutant viruses (Table 1). Instead, less efficient transforma-tion may reflect less efficiently transforming p21 proteins.Further, the later after transfection a focus forms, the moreits formation is delayed by interference with helper virus. Athigh multiplicity of infection with biologically cloned virus,this effect was not apparent, probably because it was com-pensated for by high p21 dosage. The 10-day latency ofpAprasl-X containing normal p-ras-1 could reflect a neces-sity for a second round of virus replication, because RNAtranscribed from unintegrated p-ras DNA with introns maynot be spliced (Table 1).

(iii) Truncation of p-ras exon -1 in a human 6.6-kbBamHI-resistant clone (Fig. 3; refs. 7-10) or rat p-ras plasmidclones (6, 7, 22) with transforming function is in accord withour proposal. It is uncertain whether additional DNA carriedby some X phages with transforming function belongs to thep-ras gene, since these phages were derived from 3T3 cellstransformed by human (37-39) or rat (22) DNA.

(iv) Deletion ofa 0.8-kb Sac I-resistant region that extendsfrom within exon -1 to intron -1 (Fig. 3) unexpectedlyinactivates transforming function of the 6.6-kb human p-rasclone (17). We suggest that an as yet undefined sequence ofexon -1 functions as promoter or that intron -1 includessequences that inhibit translation ofp21 mRNA. The deletionwould either eliminate the promoter or preserve the transla-tion-inhibitory sequence by preventing splicing.

(v) Comparison of the sequences of Ki-SV (23) and p-Ki-ras DNA (24) reveals that Ki-ras may also be activated bytruncation, because Ki-SV also contains the 3' end of anupstream exon (24). Moreover, both genomic and cDNAclones ofp-Ki-ras DNA with transforming function lack all ormost of this upstream exon (25). In contrast, the upstreamKi-ras exon is part of normal p-Ki-ras mRNA and also of ap-Ki-ras pseudogene (24, 25).

(vi) There is only indirect evidence that p-Ha-ras exon -1is transcribed, such as the existence of the rat p-ras pseu-dogene, the size of the mRNAs, or analogy with p-Ki-rasmRNA. p-Ha-ras mRNA species of 1.4-2 kb and 5 kb thathave been described in human (39, 40) and rodent cells (41),and polyadenylylation signals have been located about 400nucleotides downstream of the p21 stop codon in rat (M.Ruta, personal communication) and in human (15, 32) p-ras.Thus each of these mRNAs contains more than the known0.6-kb coding and 0.4-kb noncoding ras sequences. Theabsence of a suitable promoter in the known human (15, 32)or rat p-ras sequences upstream ofthe p21 sequence [only 162nucleotides are known (ref. 6; Fig. 3) also suggests that p-rasmRNA originates from an upstream location.

Deletion of the viral ras region that is 5' of the p21 codingsequence and that is derived from exon -1 enhanced trans-forming function of Ha-SV in 3T3 cells. This result isconsistent with a noncoding function for this region. How-ever, the following are consistent with a coding function. (a)p-ras sequences upstream of the p21 sequence are coding inRa-SV (Figs. 1 and 3). (b) Human p-ras exons -1 and 1 forma large open reading frame of at least 205 codons that extendsfrom the BamHI site at the beginning of the known humanp-ras sequence to a stop codon that maps three codonsupstream of the human p21 sequence (Figs. 1 and 3). (c)Specific nucleotides of human p-ras 5' of p21 all occur intriplets when compared to viral ras or rat p-ras (Fig. 1). Thus,human p-ras may be a gene with two cistrons.

Conclusions. (i) All viral ras genes lack a complete p-rasexon -1. Thus viral ras genes and intact p-Ha- and p-Ki-rasgenes are not isogenic. It is unclear whether p-ras exon -1 is acoding element or a noncoding translational control element.

Biochemistry: Cichutek and Duesberg

2344 Biochemistry: Cichutek and Duesberg

(ii) The hypothesis that ras genes are activated by trun-cation is based on a complete structural correlation with allknown viral ras genes and p-ras DNAs with transformingfunction. It preempts and reconciles the hypotheses that rasgenes are activated by point mutation or elevated expression.These would each describe relative activations of ras genesthat owe absolute transforming activity to truncation. Ac-cording to this view, virus-specific ras codons are theproducts of selection for optimal transforming function.These codons were either selected from mutant, perhapstumor, cells or, more likely9 were generated after transduc-tion, when they were subject to the high mutation frequencyof retroviruses (36). Point mutations in protooncogenes mayalso condition transforming function of other viral onco-genes. For instance, different viral myb (44), fps (45), andmyc (46) genes each contain different strain-specific, but nocommon virus-specific, mutations compared to the corre-sponding protooncogenes. Since it is unlikely that nonspe-cific mutations convert protooncogenes to autonomoustransforming genes, they are not considered essential for theoncogenic properties of these viruses (4, 44-46).

(iii) The hypothesis that p-ras genes of certain tumors aredominant cancer genes due to at least 50 different pointmutations in 5 different codons (8-18, 42) needs to bereevaluated because mutations typically inactivate genes orare silent (26) and because the ras mutations are not neces-sary for transforming activity. Since in tumors p-ras geneswith mutant codons are neither overexpressed (8) nor knownto be truncated, there is no experimental evidence that theyare cancer genes. Conceivably, these mutations do not"activate" the as yet unknown function of intact p-ras genes.The observations that some ras mutations originated subse-quent to carcinogenesis (47-49) argue that they are notnecessary for transformation, and those that benign skin papil-lomas with mutated p-ras genes spontaneously revert to normal(50) argue that they are not sufficient. As p-ras mutations areonly very rarely associated with spontaneous tumors (4), theycould be one of many other permissible mutations resultingfrom genetic instability of tumor cells (51).Complete definition of p-ras genes is necessary to test the

truncation hypothesis.

Note Added in Proof: A p-ras cDNA that includes part of exon -1 asdefined here has been described and four initiation sites for rastranscription between positions 1000 and 1020 (Fig. 1) have beendeduced (43). We question these initiation sites because sequences5' of position 1000 are in the rat pseudogene (Figs. 1 and 3) andbecause the 80 p-ras-derived nucleotides of HaSV upstream of theseinitiation sites do not function as promoters.

We thank Monika Lusky for critical discussions and generous helpand Lorrine Chao for excellent assistance. We are grateful to D.DeFeo-Jones and E. Scolnick for providing clones and generousadvice, to D. Lowy and P. Tambourin for the pA clone and forcommunicating results similar to ours, to M. Ruta for providingunpublished rat p-ras sequences, to C. Romerdahl for an initial efforton this project, to S. Aaronson and M. Barbacid for advice, to M.Kriegler for pMLV, and to P. Seeburg for help and advice. This workwas supported by National Cancer Institute Grant CA11426 and byCouncil for Tobacco Research Grant 1547. K.C. is supported by theDeutsche Forschungsgemeinschaft.

1. Harvey, J. J. & East, J. (1971) Int. Rev. Exp. Pathol. 10, 265-360.2. Aaronson, S. A. & Todaro, G. J. (1970) Nature (London) 225, 458-459.3. Levy, J. A. (1975) Nature (London) 253, 140-142.4. Duesberg, P. H. (1985) Science 228, 669-677.5. Ellis, R. W., Lowy, D. k. & Scolnick, E. M. (1982)Adv. Viral Oncol. 1,

107-126.6. DeFeo, D., Gonda, M. A., Young, H. A., Chang, E. H., Lowy, D. R.,

Scolnick, E. M. & Ellis, R. W. (1981) Proc. Natl. Acad. Sci. USA 78,3328-3332.

7. Chang, E. M., Furth, M. E., Scolnick, E. M. & Lowy, D. R. (1982)Nature (London) 297, 479-483.

8. Tabin, C. J., Bradley, S. M., Bargmann, C. I., Weinberg, R. A.,Papageorge, A. G., Scolnick, E. M., Dhar, R. D., Lowy, D. R. &Chang, E. M. (1982) Nature (London) 300, 143-149.

9. Reddy, E. P., Reynolds, R. K., Santos, E., Barbacid, M. (1982) Nature(London) 300, 149-152.

10. Taparowsky, E., Suard, Y., Fasano, O., Shimizu, K., Goldfarb, M. &Wigler, M. (1982) Nature (London) 300, 762-765.

11. Seeburg, P. H., Colby, W. W., Capon, P. J., Goeddel, D. V. &Levinson, A. D. (1984) Nature (London) 312, 71-75.

12. Tabin, C. J. & Weinberg, R. A. (1985) J. Virol. 53, 260-265.13. Fasano, O., Aldrich, F., Tamanoi, F., Taparowsky, E., Furth, M. &

Wigler, M. (1984) Proc. Natl. Acad. Sci. USA 81, 4008-4012.14. Varmus, H. E. (1984) Annu. Rev. Genet. 18, 553-612.15. Capon, D. J., Chen, E. Y., Levinson, A. D., Seeburg, P. H. & Goeddel,

D. V. (1983) Nature (London) 302, 33-37.16. Spandidos, D. A. & Wilkie, N. M. (1984) Nature (London) 310,

469-475.17. Puga, A., Gomez-Marquez, J., Brayton, P. R., Cantin, E. M., Long,

L. K., Barbacid, M. & Notkins, A. L. (1985) J. Virol. 54, 879-881.18. Bos, J. L., Toksoz, D., Marshall, C. J., Verlaan-de Vries, M., Veene-

man, G. H., van der Eb, A., van Boom, J. H., Janssen, J. W. G. &Steenvoorden, A. C. M. (1985) Nature (London) 315, 726-730.

19. Soeda, E. & Yasuda, S. (1985) in RNA Tumor Viruses, eds. Weiss, R.,Teich, N., Varmus, H. & Coffin, J. (Cold Spring Harbor Laboratory,Cold Spring Harbor, NY), pp. 928-939.

20. Rasheed, S., Norman, S. L. & Heidecker, G. (1983) Science 221,155-159.

21. Reddy, E. P., Lipman, D., Andersen, P. R., Tronick, S. R. &Aaronson, S. A. (1985) J. Virol. 53, 984-987.

22. Sukumar, S., Notario, V., Martin-Zanca, D. & Barbacid, M. (1983)Nature (London) 306, 658-661.

23. Tsuchida, N., Ryder, T. & Ohtsubo, E. (1982) Science 217, 937-939.24. McGrath, J. P., Capon, D. J., Smith, D. H., Chen, E. Y., Seeburg,

P. H., Goeddel, D. V. & Levinson, A. D. (1983) Nature (London) 304,501-506.

25. Capon, D. J., Seeburg, P. H., McGrath, J. P., Hayflick, J. S., Edman,U., Levinson, A. D. & Goeddel, D. V. (1983) Nature (London) 304,507-513.

26. Watson, J. D. (1970) Molecular Biology of the Gene (Benjamin, NewYork).

27. Maisel, J., Klement, V., Lai, M. M.-C., Ostertag, W. & Duesberg, P.(1973) Proc. Natl. Acad. Sci. USA 70, 3536-3540.

28. Scolnick, E. M., Vass, W. C., Howk, R. S. & Duesberg, P. H. (1979) J.Virol. 29, 964-972.

29. Ellis, R. W., DeFeo, D., Maryak, J. M., Young, H. A., Shih, T. Y.,Chang, E. H., Lowy, D. R. & Scolnick, E. M. (1980) J. Virol. 36,408-420.

30. Lusky, M. & Botchan, M. (1981) Nature (London) 293, 79-89.31. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning:A

Laboratory Manual (Cold Spring Harbor Laboratory, Cold SpringHarbor, NY).

32. Reddy, E. P. (1983) Science 220, 1061-1063.33. Shoemaker, C., Hoffmann, J., Goff, S. P. & Baltimore, D. (1981) J.

Virol. 40, 164-172.34. Graham, F. L. & Van der Eb, A. J. (1973) Virology 54, 536-539.35. Zhou, R.-P., Kan, N., Papas, T. & Duesberg, P. H. (1985) Proc. Natl.

Acad. Sci. USA 82, 6389-6393.36. Coffin, J. M., Tsichlis, P. N., Barker, C. S., Voynow, S. & Robinson,

H. L. (1980) Ann. N. Y. Acad. Sci. 354, 410-425.37. Pulciani, S., Santos, E., Lauver, A.-V., Long, L. K., Robbins, K. &

Barbacid, M. (1982) Proc. Nat!. Acad. Sci. USA 79, 2845-2849.38. Shih, C. & Weinberg, R. A. (1982) Cell 29, 161-169.39. Goldfarb, M., Shimizu, K., Perucho, M. & Wigler, M. (1982) Nature

(London) 296, 404-409.40. Parada, L. F., Tabin, C. J., Shih, C. & Weinberg, R. A. (1982) Nature

(London) 297, 474-478.41. Ellis, R. W., DeFeo, D., Furth, M. E. & Scolnick, E. M. (1982) Mol.

Cell. Biol. 2, 1339-1345.42. Der, C. J., Finkel, T. & Cooper, G. M. (1986) Cell 44, 167-176.43. Ishi, S., Merlino, G. T. & Pastan, I. (1985) Science 230, 1378-1380.44. Nunn, M., Weiher, H., Bullock, P. & Duesberg, P. (1984) Virology 139,

330-339.45. Pfaff, S. L., Zhou, R.-P., Young, J. C., Hayflick, J. & Duesberg, P.

(1985) Virology 146, 307-314.46. Hayflick, J., Seeburg, P. H., OhIsson, R., Pfeifer-Ohlsson, S., Watson,

D., Papas, T. & Duesberg, P. (1985) Proc. Natl. Acad. Sci. USA 82,2718-2722.

47. Albino, A. P., Le Strange, R., Oliff, A. I., Furth, M. E. & Old, L. J.(1984) Nature (London) 308, 69-72.

48. Vousden, K. M. & Marshall, C. J. (1984) EMBO J. 3, 913-917.49. Tainsky, M. A., Cooper, C. S., Giovanella, B. C. & Vande Woude,

G. F. (1984) Science 225, 643-645.50. Balmain, A., Ramsden, M., Bowden, G. T. & Smith, J. (1984) Nature

(London) 307, 658-660.51. Dracopoli, N. C., Houghton, A. N. & Old, L. J. (1985) Proc. Natl.

Acad. Sci. USA 82, 1470-1474.

Proc. Natl. Acad Sci. USA 83 (1986)