In vitro translation of RNA molecules that are partially de- graded, rendering .... HiNDI II. RIb +. + DCIP + DrrP. IS-NUCLEASE. IS1-NUCLEASE. LIGASE. 4DD;j1.
Proc. Nat Acad. Sci. USA Vol. 79, pp. 800-804, February 1982
Biochemistry
Expression of transforming region of Moloney murine sarcoma virus in Escherichia coli as a fusion protein with small tumor antigen of polyoma virus (recombinant DNA/transforming retrovirus/nucleotide sequence/lac operon/mos gene)
DANIEL J. DONOGHUE AND TONY HUNTER Tumor Virology Laboratory, The Salk Institute, P.O. Box 85800, San Diego, California 92138
Communicated by Dan L. Lindsley, October 28, 1981
ABSTRACT Bacterial expression of the transforming region of Moloney murine sarcoma virus, designated mos, was obtained as a fusion protein with a portion of the small tumor antigen of polyoma virus. This was accomplished by fusing the entire mos open reading frame, encoding a 41,000-dalton protein, with a plasmid that expresses a (3-galactosidase-polyoma fusion. protein under lac operon control. The resulting plasmid directed synthesis of the predicted polyoma antigen-sarcoma virus fusion protein of 59;000 daltons. This protein was immuneprecipitated by an antipolyoma tumor antigen antiserum that recognized polyoma determinants at the NH2 terminus of the hybrid protein. This protein was also immunoprecipitated by an antiserum directed against a synthetic peptide containing the 12 COOH-terminal amino acids encoded by the mom open reading frame. This work confirms the existence of a long open reading frame in the mos gene and resolves a discrepancy between different nucleotide sequences for its COOH-terminal coding region. Moloney murine sarcoma virus (Mo-MuSV) is a replication-defective retrovirus that transforms fibroblasts in vitro and induces fibrosarcomas in vivo with a short latent period. MoMuSV apparently arose by recombination between the nondefective Moloney murine leukemia virus (Mo-MuLV) and normal mouse cellular information (1-3). The "acquired cellular sequence of Mo-MuSV consists of an uninterrupted sequence ofabout 1200 nucleotides near the 3' terminus ofthe Mo-MuSV genome which has been substituted for the env gene of MoMuLV (4, 5) (see Fig. 1). Previously referred to as src, the "acquired cellular sequence" of Mo-MuSV is currently termed
MuLV LTR
pol
gag
LTR
env
N\
N% N
N
mos ==
A=
=I
1 24-MSV Sal I Bgl 11 Hind Ill
clone partial reverse transcript in pBR322
Sal 1 _
*s _ Hind IlI
Bgl 11
Xba I
FIG. -1. Relationship of124-MuSV to Mo-MuLV. The transforming gene mos is associated with the cellular sequence that replaces env in the single large substitution. The plasmid pDDO contains the Sal I-HindEI region of 124-MuSV inserted into pBR322. The transforming activity of the mos gene present in pDDO has been demonstrated by ligation to a long terminal repeat sequence and transfection into NIH/3T3 cells (7).
mos.
Transfection with subgenomic DNA fragments of Mo-MuSV has provided some evidence as to the physical location of the transforming function. By using focus formation on NIH/3T3 cells as an assay, Andersson et al. (6) located the transforming function between the unique Sal I and HindIII sites (see map, Fig. 1). In another study, Canaani et al. (8) placed the transforming function to the right of the Bgl II site. Together, these studies place the transforming function between the Bgl II and HindIII sites on the map of 124-MuSV, which is the specific clonal isolate of Mo-MuSV produced by G8-124 cells (9). This is consistent with DNA transfection studies performed with two other clonal isolates of Mo-MuSV, HT1-MuSV and ml-MuSV, although.the delimiting restriction sites vary somewhat (10, 11). In vitro translation of RNA molecules that are partially degraded, rendering internal methionine codons available for initiation of translation, was used by Papkoff et aL (12) to examine the coding potential of 124-MuSV virion RNA. They observed proteins that had Mrs of approximately 37,000, 33,000, 24,000,
.and 18,000, shared overlapping amino acid sequences, were unrelated to the gag, pol, or env gene products of Mo-MuLV, and appeared to be encoded by mos. More recently, Cremer et aL (13) described a family of proteins with Mrs of 43,000, 40,000, 31,000, and 24,000 generated by in vitro translation of the mos gene of 124-MuSV. By partial cyanogen bromide cleavage, they also showed that these proteins are related in their COOH-terminal regions. It seems clear that the 43,000-dalton family described by Cremer et aL (13) is identical to the 37,000dalton family described by Papkoff et aL (12). The nucleotide sequence of. mos has been determined by three independent groups using different cloned DNAs. The
The publication costs ofthis article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. ยง1734 solely to indicate this fact.
Abbreviations: Mo-MuSV, Moloney murine sarcoma virus; Mo-MuLV, Moloney murine leukemia virus; t antigen, small tumor antigen of polyoma. 800
Biochemistry: Donoghue and Hunter mos sequence of the biologically active clone pDDO (7) agrees with that of Van Beveren et al. (14) for the clone pMSV-1; these sequences predict a single large open reading frame that could encode a protein of Mr 41,000. However, these two sequences differ from the mos sequence of Reddy et al. (15) as to the COOH terminus of the predicted polypeptide. In order to resolve this discrepancy, we wished to express the mos open reading frame as a bacterial protein in Escherichia coli. To this end, we exploited the plasmid pgt-7 in which the small tumor antigen (t antigen) of polyoma virus is expressed under lac operon control (16). In the hybrid plasmid pDD21, the entire mos open reading frame was fused to the coding region of the polyoma antigen. Characterization of the predicted t antigen-MuSV fusion protein, encoded by pDD21, serves to resolve the discrepancy concerning the sequence of the mos open reading frame.
MATERIALS AND METHODS Construction of pDD21. The plasmid pDDO containing the Sal I-HindIII Mo-MuSV fragment was constructed by Stephen Goffand Mitchell Goldfarb (at Massachusetts Institute of Technology). Plasmid pgt7-wt6f, abbreviated pgt-7 (16), and the bacterial strain HB101 (17) were provided by Art Horwich and Walter Eckhart. The desired Xba I-HindII1 fragment of pDDO was purified by agarose gel electrophoresis, and DNA was recovered from the gel slice by the glass powder technique (18). The Xba I end was partially filled in by incubation for 20 min at 17'C in 50 mM Tris HCl, pH 7.6/10 mM MgCl2/L mM dithiothreitol/20 A.tM dCTP/5 ,uM dTTP with 3 units of E. coli DNA polymerase I. After ethanol precipitation, the DNA was resuspended in 30 mM NaOAc, pH 4.6/0.16 M NaCl/0.5 mM ZnCl2 and was digested with S1 nuclease. After incubation at 370C for 30 min, EDTA was added to a final concentration of 2 mM. The sample was extracted once with phenol and three times with ether and was precipitated with ethanol prior to ligation. The DNA of pgt7 was cleaved with Sst I, after which gel purification, S1 nuclease treatment, and inactivation of the S1 nuclease were as for the pDDO DNA above. The resuspended Xba I-HindIII fragment of pDDO was ligated (19) to the resuspended Sst I fragment of pgt-7. After transfection, ampicillin-resistant colonies were screened (20) for Mo-MuSV-specific sequences by using a probe prepared by nick-translation (21) of an aliquot of the gel-purified Xba I-HindIII fragment. Several clones were obtained with the desired insert in either orientation with respect to the polyoma sequence; pDD21 bears the insert in the correct orientation for bacterial expression. The mos insert in pDD21 was characterized by digestion with Ava I, Bgl II, Kpn I, Pst I, Hinfl, and Hha I and appeared to be identical to the mos insert in the parental plasmid pDDO. Immunoprecipitations from Bacterial Lysates. In general, 1 ml of exponentially growing bacteria at 2 x 108 cells per ml were labeled with [3S]methionine (100 ,uCi/ml; 1 Ci = 3.7
10'0 becquerels) for 45 min in the presence of 1 mM isopropyl thiogalactoside (16). Immunoprecipitates of lysates, prepared x
described (16), were analyzed by electrophoresis in NaDodSOJpolyacrylamide gels containing 15% acrylamide and 0.09% bisacrylamide. 'S-Labeled proteins were detected by fluorography.
Proc. NatL Acad. Sci. USA 79 (1982)
801
Bgl II and HindIII sites (having the same sense as the viral RNA genome). A single large open reading frame is revealed in frame 1, containing five methionine residues and encoding a protein of Mr 41,000. Fig. 2B schematically aligns the predicted in vitro translation products of frame 1 with the sequence. Five in vitro translation products are predicted, differing at their NH2 termini but sharing a common COOH terminus, with calculated Mrs of 41,000, 38,000, 32,000, 22,000, and 8000. The sizes of these predicted proteins are in good agreement with those observed by in vitro translation of 124-MuSV virion RNA (12, 13). However, the schema presented in Fig. 2B disagrees with the model presented by Cremer et al. (13); they propose that the 41,000-dalton family of proteins has a different COOH terminus. This is based on the mos sequence of Reddy et al. (15), which lacks two base pairs present in the sequences of pDDO and pMSV-1; these are located 1226 and 1232 base pairs from the Bgl II site. As a result, their proposed reading frame would consist of the NH2-terminal region of reading frame 1 joined to the COOH-terminal region of reading frame 3, which would then terminate downstream from the HindIII site. This would render each of the predicted proteins slightly larger than those shown in Fig. 2B because they would be longer (and differ in amino acid composition) at their COOH termini. Construction of the Polyoma-MuSV Hybrid Plasmid. Given
these different nucleotide sequences of mos and the potential biological significance of the mos reading frame, we wished to confirm this reading frame by some technique other than DNA sequence analysis. Our approach involved fusing the entire mos open reading frame to the coding region of some protein that A 1 11 I
L
.
.
kMI
II
loll I
.
2011 3 o ,
.
Il-II.1k II III1
I
so
Distance, bases
.
BGLII
Il
I
I 1000
.
XBA I
1
HI NDI II
B 11II
'
'
ATG ATG
ATG
.
I
ATG
ATG Met
Met Met
lt-t 32
Met
Met
DERIVED FROM
_
TGA 8-
-8-
22 _
_
_
_
_
_
_
38
41 DERIVED FROM C-MOS
Mo-MULV
as
RESULTS MuSV Reading Frames. Fig. 2A presents the methionine codons and terminator codons for the three reading frames of the + strand of the mos sequence of pDDO between the unique
FIG. 2. (A) Position of initiator and terminator codons. The three + strand reading frames refer to the sequence read from the Bgl II site
to the HindII site-i.e., the same
sense as the Mo-MuSV viral RNA (B) Predicted in vitro translation products. Each ATG in the mos open reading frame potentially could be used by the in vitro reticulocyte lysate translation system to yield the proteins indicated (sizes shown in kilodaltons). The first ATG is actually donated by the env gene of Mo-MuLV-i.e., the recombination point between MoMuLV and c-mos occurs 15 base pairs downstream from the start of the mos open reading frame. *, ATG; I, TAG, TAA, or TGA.
genome.
Biochemistry: Donoghue and Hunter
802
Proc. Nad Acad. Sci. USA 79 (1982)
could be expressed in E. coli and that could be immunoprecipitated with available antisera. Horwich et al. (16) described a plasmid, pgt-7, in which the polyoma t antigen coding region is fused to the NH2 terminus of f-galactosidase. When the lac operon is induced, this plasmid directs synthesis of a 26,000dalton protein consisting of a small NH2-terminal portion of /3galactosidase (eight amino acids) fused to the t antigen. This plasmid offered a convenient point of departure for bacterial expression of the mos open reading frame. The protocol used to construct the required hybrid is diagrammed generally in Fig. 3 and in detail in Fig. 4. The plasmid pgt-7 was opened with Sst I in the t antigen encoding region and then treated with SI nuclease; then it was ligated to the Xba 1-HindIII fragment of Mo-MuSV that had been appropriately modified at its termini. The closest restriction site upstream from the start of the mos open reading frame is the Xba I site TCTAGA Unfortunately, this site contains an amber codon (TAG) which is in frame with the initiator ATG ofthe mos open reading frame. This problem was overcome by partially filling in the Xba I end with E. coli DNA polymerase I in the presence of only dCTP and dTTP (see Fig. 4). Treatment with Si nuclease then removed the 5'-terminal C and T of this site, including the T of the amber codon. This also left the correct number of bases so that the mos open reading frame would be in frame with the t antigen reading frame after ligation. The ligated mixture was transfected into competent E. coli and one clone that contained the desired Mo-MuSV insert in the correct orientation, pDD21, was chosen for further study. The nucleotide sequence at the polyoma-MSV junction was determined on the - strand-i.e., the complement of the viral RNA genome-at the position shown in Fig. 4. The sequencing
t antigen sequence
mos sequence
HIS AG GLIU LEU LYS
-CAT AGA GAG CrC MA-GTA TCT CrC GG mrrSSTI
-CAT A GA CT --GTA TCT C
SI-MMEASE -CAT AGA G
-GTA TCT C
MET ALA HIS
-CTC TAG AC1 GC ATG GCG CAT--C ATC TGA CTG TAC CGC GAXBAI
C TAG ALT GC ATG GCG CATTGA CrG TC OGC GTAPOL I M + DTil C TAG ACT[AC ATG GCG CAT-TC TGA CTG TAC CGC GTA-
AGLT AC ATG GCG CATTC TGA CTG TAC OC GTALIGASE HIS AR GLIJ TM ASP MET ALA HIS
GM ACT GAC ATG POLY--XT AA CTC IA TCT TGA CMG TAC CGTA
GCGM4
FIG. 4. Joining of the polyoma (left column, sequence at Sst I site in t antigen-encoding region) and Mo-MuSV (right column, sequence at Xba I site at start of 41,000-dalton open reading frame) termini during construction of pDD21. Note the presence of a TAG terminator (***) in frame with the ATG that initiates the mos open reading frame. The two bases filled in by E. coli DNA polymerase I at the Xba I end are indicated by the small arrows. The polyoma coordinates of Friedmann et al. (22) were used. Broken line at bottom, nucleotide sequence at junction for which the sequencing gel is shown in Fig. 5.
pDDO
t
I
SSTI
HiNDI II
XBAI
SSTI
RIb
+
+ DCIP + DrrP
IS-NUCLEASE
IS1-NUCLEASE LIGASE
4DD;j1 59,000-dalton fusion protein FIG. 3. Construction of the polyoma-MuSV hybrid. The details of joining the polyoma and Mo-MuSV termini, during which the correct reading frame had to be preserved, are presented in Fig. 4. The derivation of the 528 total amino acids in the 59,000-dalton fusion protein is as follows: 8 from the NH2 terminus of f3galactosidase, 3 from a linker, 8 of polyoma preceding the t antigen coding region, 133 from the NH2 terminus of polyoma t antigen, 2 from the polyoma-MuSV joint, and 374 representing the entire mos open reading frame.
gel shown in Fig. 5 confirmed that the E. coli DNA polymerase I and S1 nuclease reactions of Figs. 3 and 4 worked as expected. Thus, the plasmid pDD21 contained the mos open reading frame fused, in frame, downstream from the antigen encoding region. Synthesis of the Predicted Fusion Protein in E. coli. E. coli carrying the plasmid pDD21, or pgt-7 as a control, were examined for the expression of proteins related to polyoma t antigen. A protein of Mr 26,000 was specifically immunoprecipitated from E. coli strain HB101 carrying the parental plasmid pgt-7 (Fig. 6A). In contrast, a protein of Mr 54,000 was specifically immunoprecipitated from bacteria carrying the hybrid plasmid HB101[pDD21]. This is in reasonable agreement with the calculated Mr of 59,000 for the polyoma-MSV hybrid protein. The hybrid protein represented approximately 0.05% of the total soluble protein in HB101[pDD21] under the experimental conditions used (data not shown). This level of expression is comparable to that observed for expression of the polyoma t antigen in HB101[pgt-7] (16). Also shown in Fig. 6A are the immunoprecipitated products of HB101[pDD14]. This plasmid contained only a portion of the Mo-MuSV XbaI-HindIII fragment inserted at the polyoma Sst I site, as shown by restriction endonuclease mapping (data not shown). Plasmid pDD14 directed synthesis of a truncated form of polyoma t antigen, Mr 17,000, arising by termination of translation upon entering the Mo-MuSV sequence in an incorrect reading frame. Recently, another antiserum became available that is directed against a synthetic peptide containing the COOH-terminal 12 amino acids of the mos reading frame (24). This antiserum was used to immunoprecipitate [ S]methionine-labeled
Biochemistry: Donoghue and Hunter
Proc. Natl. Acad. Sci. USA 79 (1982)
A
B
1 2 3
TG
4-+54
-
1 2 3
M^a._
_--4-54 q*e 5e
No
-AT
4N;m-
803
E
-T R
-_
d.
*26 ..-. J.....
--
-G -A -
G
A
C
C
-
%*
+17
C
C
T
FIG. 5. Nucleotide sequence across the polyoma-MuSV junction. DNA ofplasmid pDDO was 32P-labeled at the 5' termini of the HhaI site GcG shown in Fig. 4, and the sequence was determined upstream into the t antigen encoding region. The products of Maxam-Gilbert (23) reactions were analyzed by electrophoresis in a 24% polyacrylamide sequencing gel. The location of this sequence is indicated by the dashed arrow at the bottom of Fig. 4. The two bases at the Xba I site that were filled in by E. coli DNA polymerase I are indicated by the small arrows. To isolate a DNA fragment uniquely labeled at the Hha I site, an Alu I site at polyoma position 539 (22) was utilized for secondary digestion.
lysates of HB101 harboring either pDD21 or pgt-7 as a negative control. This antiserum precipitated a 54,000-dalton protein from lysates of HB101[pDD21] (Fig. 6B), which comigrated with the protein specifically immunoprecipitated from the same lysate by the polyoma antitumor serum. As expected, the COOH-terminal antiserum fitiled to immunoprecipitate the 26,000-dalton t antigen produced in HB101[pgt-7]. DISCUSSION
There exists the possibility of a frameshift error in any nucleotide sequence of significant length, particularly when a sequence is determined with no a priori knowledge of the gene product. In the case of Mo-MuSV, we have used bacterial expression of the mos gene product as a means of confirming the open reading frame predicted by the nucleotide sequence. This entailed construction of a hybrid between the coding re-
FIG. 6. Immunoprecipitations of [35S]methionine-labeled bacterial lysates. (A) Polyoma antitumor serum was used to immunoprecipitate lysates of HB101[pgt-7] (lane 1), HB101[pDD21I (lane 2), and HB101[pDD14I (lane 3). (B) An antiserum against a synthetic peptide containingthe COOH-terminal 12 amino acids of the mos open reading frame was used to immunoprecipitate lysates of HB101[pgt-71 (lane 1) and HB101[pDD21I (lane 2). For comparison, lane 3 shows HB101[pDD21I immunoprecipitated with polyoma antitumor antiserum.
gion of polyoma t antigen and the mos open reading frame. E. coli harboring this plasmid directed synthesis of a protein of the expected size. This protein was specifically immunoprecipitated by polyoma antitumor antiserum through recognition of polyoma determinants at its NH2 terminus. The hybrid protein was also immunoprecipitated by an antiserum directed against the predicted COOH terminus of the mos open reading frame. These immunoprecipitation studies demonstrate that the termini of the mos open reading frame are as predicted by the nucleotide sequences of pDDO and of pMSV-1 (14). The alternative COOH terminus predicted by Reddy et al. (15) is incompatible with these results. The results reported here are in agreement with recent results of Papkoff et at (24). Using the aforementioned COOHterminal mos antiserum directed against a synthetic peptide predicted by the sequences of pDDO and pMSV-1, they were able to immunoprecipitate in vitro translation products of 124MuSV virion RNA. What is the biological significance ofthis open reading frame in mos, with regard to transformation? For the following reasons we suggest that the 41,000-dalton reading frame is used in vivo to encode a transforming gene product. First, all retroviral transforming genes examined thus far are devoid of intervening sequences in their coding regions, although intervening sequences are often found in the cellular homologues of these
804
Biochemistry: Donoghue and Hunter
genes-e.g., Rous sarcoma virus (25), Abelson murine leukemia virus (26), and Harvey murine sarcoma virus (27). By analogy, one expects the coding region for the Mo-MuSV transforming function to be uninterrupted by intervening sequences. As shown in Fig. 2A, reading frame 1 is the sole reading frame that contains a large uninterrupted coding region. Second, the NH2-terminal region of the mos open reading frame is derived from the env gene of Mo-MuLV (7, 14, 15)-i.e., the recombinational event between Mo-MuLV and the cellular homologue c-mos occurred 15 base pairs downstream from the env gene NH2 terminus. Recently, we examined the nucleotide sequence of other isolates of murine sarcoma virus, including Gazdar murine sarcoma virus (28) and mlMuSV; in these cases, the NH2 terminus of the Mo-MuLV env gene initiates the mos open reading frame (unpublished data). Thus, we infer that proper expression of the viral mos gene is contingent upon its fusion to the NH2 terminus ofthe env gene, analogous to many other transforming retroviruses in which cellular information has been fused to the retroviral gag gene -.g., Abelson murine leukemia virus (29), feline sarcoma
virus (30), or Y73 virus (31). At the present time, no conclusive data exist concerning a mos-specific mRNA in transformed cells. Recently, however, evidence has been obtained that the 41,000-dalton product is expressed in Mo-MuSV-transformed cells (J. Papkoff, personal communication). Ultimately, characterization of mutants that interrupt the mos open reading frame will be required to resolve its significance. Note Added in Proof. Recently, Reddy et aL (32) revised their earlier sequence of mos (15). In their latest sequence, three additional base pairs are present. Two ofthese base pairs, at 1226 and 1232 nucleotides from the Bgl II site, are now in agreement with the sequences of pDDO (7) and pMSV-1 (14). However, the third additional base pair in their revised sequence (32) was not observed in these other sequencs of mos. Thus, both sequences of Reddy et aL (15, 32) are in disagreement with the results presented here as to the correct reading frame in the COOHterminal region of mos. We thank Mary Anne Hutchinson and Walter Eckhart for providing the polyoma antitumor serum, Jackie Papkoff and Inder Verma for providing the COOH-terminal Mo-MuSV antiserum, Art Horwich for providing pgt-7 DNA, Steve Goff for providing the Mo-MuSV-specific plasmid, Peter Geiduschek for valuable criticism of the manuscript, and Arnie Berk, Jon Cooper, Art Horwich, Bart Sefton, and Dennis Templeton for stimulating conversations. D.J.D. gratefully acknowledges a Helen Hay Whitney postdoctoral fellowship. This work was supported by U.S. Public Health Service Grants CA-14195, CA-17096, and CA28458. 1. Moloney, J. B. (1966) NatL Cancer Inst. Monogr. 22, 139-142. 2. Scolnick, E. J., Howk, R. A., Anisowicz, A., Peebles, P. T., Scher, C. D. & Parks, W. P. (1975) Proc. NatL Acad. Sci. USA 72, 4650-4654.
Proc. Nad Acad. Sci. USA 79 (1982) 3. Frankel, A. E. & Fischinger, P. J. (1976) Proc. NatL Acad. Sci. USA 73, 3705-3709. 4. Hu, S., Davidson, N. & Verma, I. M. (1977) Cell 10, 469-477. 5. Donoghue, D. J., Sharp, P. A. & Weinberg, R. A. (1979) J. Virol 32, 1015-1027. 6. Andersson, P., Goldfarb, M. P. & Weinberg, R. A. (1979) Cell 16, 63-75. 7. Donoghue, D. J. (1982)J. ViroL, in press. 8. Canaani, E., Robbins, K. C. & Aaronson, S. A. (1979) Nature (London) 282, 378-383. 9. Ball, J. K., McCirter, J. A. & Sunderland, S. M. (1973) Virology 56, 268-284. 10. Blair, D. G., McClements, W. L., Oskarsson, M. K., Fischinger, P. J. & Van de Woude, G. F. (1980) Proc. NatL Acad. Sci. USA 77, 3504-3508. 11. Oskarsson, M., McClements, W. L., Blair, D. G., Maizel, J. V. & Van de Woude, G. F. (1980) Science 207, 1222-1224. 12. Papkoff, J., Hunter, T. & Beemon, K. (1980) Virology 101, 91-103. 13. Cremer, K., Reddy, E. P. & Aaronson, S. A. (1981)J. Virol 38, 704-711. 14. Van Beveren, C., Galleshaw, J. A., Jonas, V., Berns, A. J. M., Doolittle, R. F., Donoghue, D. J. & Verma, I. M. (1981) Nature (London) 289, 258-262. 15. Reddy, E. P., Smith, M. J., Canaani, E., Robbins, K. C., Tronick, S. R., Zain, S. & Aaronson, S. A. (1980) Proc. Natl Acad. Sci. USA 77, 5234-5238. 16. Horwich, A., Koop, A. H. & Eckhart, W. (1980) J. Virol 36, 125-132. 17. Boyer, H. W. & Roulland-Dussoix, D. (1969) J. Mol Biol 41, 459-472. 18. Vogelstein, B. & Gillespie, D. (1979) Proc. Nati Acad. Sci. USA 76, 615-619. 19. Ferretti, L. & Sgaramella, V. (1981) Nucleic Acids Res. 9, 85-93. 20. Grunstein, M. & Hogness, D. S. (1975) Proc. Natl Acad. Sci. USA 72, 3461-3465. 21. Rigby, P. W. J., Dieckmann, M., Rhodes, C. & Berg, P. (1977) J. Mol Biol 113, 237-247. 22. Friedmann, R., Esty, A., LaPorte, P. & Deininger, P. (1979) Cell 17, 715-724. 23. Maxam, A. & Gilbert, W. (1980) Methods Enzymol 65, 499-560. 24. Papkoff, J., Lai, M. H.-T., Hunter, T. & Verma, I. M. (1981) Cell 27, 109-119. 25. Shalloway, D., Zelenetz, A. D. & Cooper, G. M. (1981) Cell 24, 531-541. 26. Goff, S. P., Gilboa, E., Witte, 0. N. & Baltimore, D. (1980) Cell 22, 777-785. 27. 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. 28. Gazdar, A. F., Chopra, H. C. & Sarma, P. S. (1972) Int. J. Cancer 9, 219-233. 29. Witte, 0. N., Rosenberg, N., Paskind, M., Shields, A. & Baltimore, D. (1978) Proc. Natl Acad. Sci. USA 75, 2488-2492. 30. Barbacid, M., Lauver, A. V. & Devare, S. G. (1980)J. Virol 33, 196-207. 31. Kawai, S., Yoshida, M., Segawa, K., Sugiyama, H., Ishizaki, R. & Toyoshima, K. (1980) Proc. NatL Acad. Sci. USA 77, 6199-6203. 32. Reddy, E. P., Smith, M. J. & Aaronson, S. A. (1981) Science 214, 445-450.
