Reprinted from Proc. Nat. Acad. Sei. USA Vol. 69, No. 12, pp. 3737-3741, December 1972 Nuclear DNA Sequences Present in Human Leukemic Cells and Absent in Normal Leukocytes (70s RNA/hybridization/RNA tumor viruses/oncogene/protovirus) W. G. BAXT AND S. SPIEGELMAN Institute of Cancer Research and Department of Human Genetics and Development, College of Physicians and Surgeons, Columbia University, New York, N.Y. 10032 Contributed by S. Spiegelman, October 27, 1972 ABSTRACT The central purpose of the present study was to test the proposition that the nuclear DNA of every human cell contains whatever information is necessary and sufficient for transformation to malignancy. The experiments were made possible by our earlier identifica- tion in human leukemic cells of particulate elements en- capsulating 70s RNA and RNA-directed DNA polymerase. The [3H]DNA synthesized by these particles was used as a probe, through molecular hybridization, to normal and leukemic DNA. The results obtained establish that leu- kemic nuclear DNA contains particle-related sequences that cannot be detected in normal leukocytes. This out- come does not support the virogene-oncogene theory, which postulates the inclusion of at least one complete copy of oncogenic information in the genome of every normal cell. The data suggest that we may not be forced to cope with an omnipresent DNA segment coding for malignancy. Under the circumstances, we can perhaps entertain more hopeful pathways leading to the control and cure of cancer. We have over the past few years focused our efforts on an attempt to determine whether the knowledge gained from animal viral oncology is relevant to the etiology of human cancer. We first searched in human tumors for RNA molecules related in sequence to those found in the RNA tumor viruses known to cause corresponding cancers in experimental ani- mals. Radioactive DNAs synthesized on the appropriate viral RNA templates with RNA-directed DNA polymerase were used as probes. The data obtained revealed a pattern of specificities that agreed remarkably with what was known from the animal experimental systems. Human breast car- cinomas were found (1) to contain RNA homologous to that of the mouse mammary tumor virus. As expected, the human mammary tumor RNA showed no homology to that of the unrelated Rauscher leukemia virus. This observation gained in interest and significance when we turned our attention to human leukemias (2), sarcomas (3), and lymphomas (4). Here we found RNA homologous to that of the Rauscher leu- kemia virus, an agent related to a group that causes similar neoplasias in mice. Finding the pertinent RNA molecules related to oncogenic viruses in the different human tumors did not, of course, establish a viral etiology. We next had to design and perform experiments that could answer the following questions with respect to the RNL4 molecules we had identified: (a) Were they as large as the 70s-RNA molecules characteristic of the RNA tumor viruses?, (b) Were they physically associated in a particle containing RNA-directed DNA polymerase, also characteristic of RNA tumor viruses?, and (c) Did they possess the density characteristic of tumor viruses? 1 I 3737 Answering these questions became feasible with the de- Jelopment of an assay for the simultaneous detection of a 70s-RNA template associated with an RNA-directed DNA iolymerase (5). hlouse mammary tumor was used to resolve ;he technical problems encountered in applying this tech- iique to malignant tissue (6). The methodology thus developed 3ermitted us to establish that the viral-related RNA en- :ountered in human mammary carcinomas (7) and leukemic :ells (8) contained molecules 70s in size, and physically en- :apsulated with a RNA-directed DNA polymerase in a particle possessing a density between 1.16 and 1.19 g/ml (Hehlmann, R. and Spiegelman, S., in preparation). Further- more, the DNA synthesized endogenously by these complexes hybridized uniquely to the RNAs of the viral agents that caused the corresponding neoplastic diseases in mice. These experiments provided evidence for the existence in human neoplasias of particulate elements possessing four features diagnostic of the homologous oncogenic murine vi- ruses. Though compelling, the data did not establish that the particles thus identified are causative, or even contributory. Final proof requires the demonstration that the human par- ticles are in fact infectious and transforming agents. Never- theless, the results were sufficiently impressive to warrant the initiation of the next stage of our investigation into the ap- plicability of animal viral oncology to human cancer. It is true that virologists can satisfy Koch's postulates with purified animal tumor viruses. The question, however, is being raised with increasing frequency of whether this fact is relevant to the natural history of cancer. Indeed, the "viro- gene-oncogene" hypothesis (9, 10) implicitly uses this argu- ment, since it suggests that normally the disease is caused not by an external virus, but by the activation of an oncogene DNA segment already present, and one that is vertically transmitted (inherited) in the germ lines of all organisms prone to cancer. An alternative is the provirus theory (11, 12), which postulates the exogenous introduction of new DNA se- quences through viral infection with an RNA tumor virus. The insertion of this viral-related information into the genome is presumed to involve the synthesis of the requisite DNA by a DN,4 polymerase directed by the viral RNA. It is obvious that an experimental decision between the provirus arid virogerie hypotheses is a necessity for the de- velopment of a rationale for coritrolliiig the disease. The two hypotheses can be distinguished by their logical consequences. The virogeiie hypothesis predicts that normal DNA includes at least one complete copy of the information required for transformation and virus production. The provirus hy- pothesis denies this, and stipulates that if some of this infor- 3738 Medical Sciences: Baxt and Spiegelman Proc. Nat. Acad. Sci. USA 69 (1972) mation previously exists in normal cells, it is incomplete, and that infection with a virus is a necessary requisite to provide the information needed for tumorigenesis. The issue can be settled by experiments designed to answer the following ques- tion: Does the DNA of transformed cells contain viral-related sequences that are not found in the DNA of normal cells? We explored the technical feasibility of performing the necessary experiments with an animal system by use of DNA. DNA hybridization because of its superior sensitivity. The data obtained showed (Goodman, N., Ruprecht, R. M., Sweet, R., Deinhardt, F., Massey, R., and Spiegelman, S., in preparation) that it is readily possible to demonstrate with norlindigenous RNA oncogenic viruses that the UNA of transformed cells contains sequences that are not present be- fore transformation. These results encouraged us to apply this technology to human leukemia, and the purpose of this paper is to describe the outcome. We have shown (8) that human leukemic cells contain viral-related 70s RNA encapsulated with an RNA- directed DNA polymerase. These particulate elements can be used to generate radioactive DNA probes, which in turn can be used to detect viral-specific iiiforniation in human leuko- cyte genomes. The experiments to he de leukemic nuclear DNA contains sequences that cannot be de- tected in the DNA of normal leukocytes. METHODS Preparation of [ 3H]DNA. Leukemic leukocytes were gently opened with a Dourice homogenizer, and the extract was centrifuged at 16,000 X g for 15 min at 2" to remove nuclei aiid mitochondria. The resultant supernatant was layered on 25% sucrose-TNE buffer [0.01 Ai Tris.HC1 (pT-1 8.3)-0.15 AI NaC1-0.01 Ai EDTA] arid centrifuged in a Ueckmari SW-41 rotor at 40,000 rpm for 1 hr at 2". The resulting pellet was resuspended in 1 In1 of 0.01 Ai Tris.HCI (pH 8.3) aiid used in a typical eiidogerious RNA-directed DNA polymerase reaction to generate [3H]I)NX (8). The product was brought to 0.4 M NaOH and incubated at 37" for 2 lir to destroy RNA. To provide acceptable backgrounds for such probes, it is imperative that all self-complementary material be removed before use. To this end, the alkali- treated material is neutralized with 0.4 M IlCl and 1 Ai Tris. HC1 (pH 7.4) and then self-annealed at 60' for 48 hr, cor- responding to a Cot value of 0.01-0.25, as defined by Britten and Kohrie (18). The annealed material is then passed over a 2-nil bed volume of hydroxyapatite (Uiorad) at 60". The column was washed with 100 ml of 0.01 kI NaHP04 (pH 6.8), and the single-stranded [3H]DNX was eluted with 4 ml of 0.15 AI NaHP04 (pII 6.8), a technique that recovers 90% of the input [3H]I)NA. The DNA was then layered upon a Sephades G-50 column (coarse) of 100-ml bed volume. The DNA region of the column was pooled, and the [3H]DNA was precipitated with 2 pg/ml of carrier yeast RNA and two volumes of !%a/, ethanol. The pellet was dis- solved in 3 mM EDTA and stored at 2". The [3H]DNA banded in Cs2S04 gradients at densities between 1.42 and 1.45. Nuclear DNA was prepared as detailed elsewhere (Good- man, N., Sweet, R., Ruprecht, R. M., and Spiegelman, s., manuscript in preparation). The ratios of A260 to A280 of the prelnrations were all close to 2; on denaturation, these prepa- rations reannealed to better than 80%. Annealing Conditions. Annealing reaction mixtures con- tained 6 units of cellular DNA, 0.2-1.0 pmol of [3H]DNA, and 15 pmol NaHP04 (pH 7.2), in a final volume of 0.1 ml. The reaction was brought to 98" for 60 see and 0.04 mmol of NaCl was added. The reaction mixture was then incubated at 60", and aliquots were removed at the indicated intervals. The reaction was stopped by the addition of 2 ml of 0.01 M NaHi'04 (pH 6.8). The sample was then passed over a column of hydroxyapatite of 3-ml bed volume (15, 16) at 60". The column \vas washed with 20 ml of 0.15 hi NaHP04 (pH 6.8) at 60", 80", 88", and 95". 4-ml Fractions were collected, the A260 of each fraction was read, and the DNA was pre- cipitated with 2 pg/ml of carrier yeast RNA arid 10% tri- chloroacetic acid. The precipitate was collected on Millipore filters, which mere dried and counted (8). The method identifies unpaired strands that elute at 60" and poorly paired duplexes that disassociate at 80". Only the duplexes dissociating and eluting at 88"-95" are counted here as properly hybridized. Recycling of Probe on Normal DNA to Separate Leukemic from Normal Sequences. The reactioiis were performed with normal DNA\ as above and the eluates were collected in 1-ml fractions at 60", 80", 88", and 95". The peak fractions were combined and the resulting pools were passed over a Sephades G-50 column of IO-ml bed volume. The DNA4 regions were collected, and the DNA was precipitated with 2 pg/ml of carrier yeast RNA and 2 volumes of 95% ethanol. The ma- terial eluted at 88"-95" was pooled and used for the deter- mination of sequences common to normal and leukemic DNA. The material eluted at 60" was used as the recycled product unable to hybridize to normal DNA4. K E SULT S The strategy of the present investigation may be outlined in the following steps: (a) Isolate from leukemic cells the frac- tion containing the particles eiicapsulating the 70s RNA and RNA-direc ted DNA polymerase (8) ; (b) Use this fraction to generate [3H]DNA endogenously synthesized in the presence of high concentrations of actinomycin D to inhibit host and viral DNA-directed DNA synthesis (9, 15, 17); (c) Purify the [ 3H]DNA by hydroxyapatite and Sephades chromatography, with care being exercised to remove all self-complementary material; (d) Use the [3H]DNA as a probe to detect comple- mentary sequences in normal- and leukemic-leukocyte DNA; (e) If sequences are detected in both types of leukocyte, re- move those found in normal DNA by exhaustive hybridiza- tion; and (f) Test the residue for specific hybridizability to leukemic DNA. Fig. 1 shows typical Cot reannealing curves (18) determined by hydroxyapatite chromatography (13, 14). The purpose here is to compare annealing of [3H]DNA probe to normal (Fig. 1A) arid leukemic DNA (Fig. 1B). Since each reaction mixture contains large amounts of nuclear DNA, the be- havior of this component can be readily followed by ab- sorbance readings at 260 nm, and can be used as an internal control for the adequacy of the annealing reaction. As seen from the A260 readiiigs, the rates and extents of annealing are the same for both normal and leukemic nuclear DNA. How- ever, comparison of Fig. 1A and B shows that annealing of the [3H]D?jA probe to leukemic DNA appears to be faster and more extensive than to normal DNA. The observation that the [3H]I)NX synthesized by the Proc. Nut. Acad. Sci. USA 69 (1972) 1100- 1000 900 800 700 600 500 & 400- 0 I 300- 200 100- Nuclear DNA Sequences 3739 - - - - - - - 0- leukemic particles hybridized to nuclear DNA of normal leu- kocytes was not surprising in view of previous esperience with murine and avian systems (19-26). Whenever indigenous RNA tumor viruses were used, DNA from normal cells were found to contain viral-related sequences. We have already noted that such findings cannot decide between the provirus and virogene mechanisms. What is noteworthy here is the quantitative difference revealed in Fig. 1 between the re- sponses of normal and leukemic DNh. The greater estent and rate of aiinealing of the leukemic [3H]l)NB with leukemic iiuclear DNA suggests either one, or a combination, of the following two possibilities; (a) Leukemic DNA contains more of tlie same sequences detected by the [3H]IlNA in normal DNA, or (b) Leukemic DNA contains additional unique se- quences not found in normal DN.1 aiid detectable by the [3H]DNA. A decision between the two possibilities is readily made by the last two steps of the experimental plan outlined above. Sequences in [3H]I)Nil common to both normal and leukemic nuclear 1)NX are first removed by esliaiistive annealing with normal Dn'h in vast excess. The residue of unreactive [ 313]DN.1 can then be isolated and purified by hydrosyapatite chromatography, as described in Jlethods. These recovered strands should no longer contain sequences complementary to those found in normal DNA. Exclusive hybridizability of such recycled [3H]DNA to leukemic DNA would establish that the genome of leukemic cells contains specific sequences not detectable in normal DNA. In such experiments, strands complementary to normal- leukocyte DNA are removed by aiinealing to a Cot of greater than 10,000. Fig. 2 shows a typical outcome of hybridizing such recycled [3H]DNA to normal and leukemic DNA. It is evident that no complexes stable at temperatures above 88" are formed with normal DNA. On the other hand, 57'30 of the recycled [3H]DNA added forms well-paired duplexes with the leukemic DNA. Table 1 summarizes the data obtained in a series of esperi- ments with [3H]DNA and nuclear DNA obtained from eight untreated patients with either acute myelogenous leukemia or chronic myelogenous leukemia. In all instances the leu- 73 73 60- + v) al 6 2 40- a-" 73 20 loo[ 80 A - /// /' d' ,/ 2 3 4 log cot B I I 3 4 log cot FIG. 1. Annealing curves of [3H] DiVA synthesized by leukemic particles to leukocyte nuclear DNA from normal pooled blood (A) and from a leukemic patient (B). The eluates from the hydroxy- apatite columiis at 88" and 95" were slimmed and counted as hy- brids. The abscissas represent log Cot, where Co is mol/liter of nucleotides and t is time in seconds (18). 0- - -0, A260; .--e, 3H. kemic [3H]DNA probe hybridized to normal nuclear DNA as in Fig. 1A. However, in every case, after it was recycled by exhaustive annealing to normal DNA, the residual [3H]DNA formed stable duplexes only with leukemic DNA, in agree- ment with the experiment of Fig. 1B. To provide further information on the quantitative aspects of tlie annealing reactions, we performed esperiments to see whether the hybridization with leukemic nuclear DNA re- moved tlie leukemia-specific strands from the [3H]IlNh probe. To accomplish this, initial hybridizations were performed be- tween the [3H]DNh probes and their homologous nuclear leukemic DNA. The residual strands were then isolated and rechallenged agaiiist, tlie same leukemic DNAs. It is clear from the results summarized in Table 2 that no further hybridiza- tions occur, a result that implies that the first annealing to leukemic DNA had indeed removed virtually all of the [3H]DNA strands complementary to leukemic DNA. The data of Table 2, combined with the information on the same five patients in Table 1, permit another quantitative comparison of independently performed hybridizations on the same set of materials. The percent hybridization to leukemic 1)N.k (column 2 of Table 2) should approsimate the sums estimated from the serial hybridizations to normal arid leu- kemic DNA (columns 2 and 4 of Table 1). The resulting sums are given in column 5 of Table 2; comparisons with column 2 of the same table indicate reasonable agreement. Normal '2001 Leukemic I-- '60°'88' 88 95" ' 68' 88 88"' 95' E I uti on Temperature FIG. 2. ITydroxyapatite elution profile of a hybridization reac- tion of the recycled leukemic [3H]I>NA probe to nuclear DNA from normal leukocytes and leitkocytes from the same leukemic patient. The backgroiind was 30 cpm. 3740 Medical Sciences : Baxt and Spiegelman Proc. Nat. Acad. Sci. USA 69 (19722) TABLE 1. Exhaustive hybridization of i3H]DN.4 probe synthesized by leukemic particles with normal-leukocyte nuclear DNA, followed by hybridization of the wnhybridizing recycled leukemic [3H]DNA probe to normal DNA and to leukocyte nuclear DAVA from the same leukemic patient Recycled leukemic [ 3H] DNA hybridized to leukocyte DNA Leukemic [3H] DNA hybridized to normal-leukocyte DNA Leukemic Normal cPm yo Hybridization CPm Yo Hybridization cPm yu Hybridization Ne (AML) Ha (CML) Co (CML) Mi (CML) Go (AML) Si (AML) El (AML) Mac (AML) 3020 1350 2580 510 1100 4520 390 1450 61 40 51 45 43 49 42 49 523 1020 43 1 101 303 1130 45 510 56 51 35 36 48 46 69 52 Background was 30 cpm and all counts recorded represent cpm above background. CML = Chronic myelogenous leukemia. A MI, = Acute myelogenous leukemia. First column indicates initials of patients. The concordance we have just noted includes the sequences in the [3H]DNA probes that are common to both normal and leukemic DNA, and implies that tlie proportion of these shared sequences in the two types of DNA will not be very different. This conclusion can be checked more directly by the technique of Gelb, Kohne, arid Martin (27). Exhaustive hy- bridizatioii of [3H ]DNA to normal nuclear DNA was used to isolate the duplexes stable above 80". This material was then annealed in the presence of either normal or leukemic nuclear TINA. If either iiuclear DNA contained a significantly higher content of these shared sequences, its addition would result in a significantly greater acceleratioii of the annealing reac- tion. As can be seen from the two experiments described ill Fig. 3, the addition of normal and leukemic DNA4 made no obvious difference on the rate of reassociation of the [3H]I>NA probe containing sequences common to both. This result sug- gests that tlie [3H]DNA sequences shared by normal and leu- kemic DNA are present in equal numbers in the genomes of leukemic and normal cells. DISCU SSlON It is informative to consider the sensitivity of the hybridiza- tion experiments summarized ill Tables 1 arid 2. The specific activity (50.1 Ci/mmol) of the [3H]TTl' used, when cor- rected for counting efficiency and composition of the product, results in a specific activity of 3 x 107 cpm/pg of [3H]DNA synthesized. If we assume that only 10% of the 70s RNA has been effectively transcribed, the ratio of the molecular weight of tlie [311]DNA to that of a liumaii genome is 5 X lop7. If there were one probe equivalent per genome, the 250 pg of nuclear D?iA used in tlie hybridizations would coiitaiii 120 pg of probe sequeiices, arid woiild correspond to about 3.7 X lo3 cpm, if all were complexed with [3H]DNA. Since hybridiza- tions were carried to a (:ot of 10,000, well beyond the point where the nucleitr DNA has reamiealed, one woiild espect to see at least 50% of the reaction coml)leted, correspoiiding to 1.8 X lo3 cpm. It should, however, be noted that in many in- stances the amount of [3H ]I)NA I)robe available was riot suf- ficient to saturate cellular IINX contaiiiiiig one equivalent per genome. Comparison of the cpm observed with recycled [3H 1- DNA on normal and leukemic nuclear DNA (Table l), and assuming we c:tii easily detect 30 cpm over R hackgrourid of the same value, one can conclude from the two instalices where over 1000 cpm were complexed to leukemic DNA that normal DNA contains coiisiderably less than 2.5% of a probe eyuiva- lent per genome. We would like to explicitly emphasize tlie biologiml dif- ference between our earlier studies and the present one. We previously focused (1-8) attention on the RNA content of tumors and, therefore, oiily on active gene transcripts or 011 autonomous extrachromosomal RNA entities. The experi- ments reported here center on the nuclear DNA and, as such, include all genes-active or silent. It is evident from the re- sults described that at least a portion of the viral-related se- TABLE 2. Exhaustive hybridization Gf leukemic [3H]DN,1 probe to leukocyte nuclear DNA, from the same patient, folloued by hybridization of the residue to nuclear DNA Estimated total % hybridization from serial Leukemic [3H] DNA hybridized to Recycled leukemic [ 3H] DNA leukemic-leukocyte DNA hybridized to leukemic-leukocyte DNA hybridizations* __ yo Hybridization yu Hybridization CPm yo Hybridization CPm Ne (AML) 1310 Ha (CML) 4450 Co (CML) 4040 Mi (CML) 372 Go (AML) 1250 79 63 63 5 9 75 82 70 68 65 70 Background was 30 cpm and all counts recorded represent cpm above background. * See columris 2 and 4 of Table 1. Proc. Nat. Acad. Sci. USA 69 (1972) Nuclear DNA Sequences 3741 2'ool 80 A 2 3 42 3 4 - log cot log cot FIG. 3. Cot curves of [3H]DNA probe recovered from normal DNA at 88" and 93" and hybridized to normal DNA and the same leukemic patient's DNA. A and I? represent two different patients. 0- - -0, acute myelogenous leukemia DNA; 0-0, Normal leukocyte DNA. quences we detected in the RNA are also to be found in the nuclear DNA of the tumor cells. This finding establishes a generic relation between RNA and DNA viral-related se- quences, and makes it difficult to argue that the viral-specific RNA we detected (1-8) is associated with an irrelevant pas- senger in the cancer cell. The fact that we cannot detect these sequences in the DNA of normal leukocytes argues against the virogene-oncogene theory as an explanation of human leukemia. Our data are more consistent with either the provirus hypothesis, or the more recent protovirus concept (28), and-at the very least- imply a requirement to add new genetic information for con- version of a normal cell to a cancer cell, in agreement with our studies of simian cells transformed by nonindigenous tumor viruses (Goodman, N., Sweet, R., Ruprecht, R. M., and Spiegelman, S., manuscript in preparation). One striking feature exhibited by the experiments described is the necessity of first removing shared sequences before at- tempting to use such probes to search for sequences unique to a neoplastic cell. Many of the previous attempts (19-26) to de- tect significant differences between normal and transformed cells probably failed because they did not include such a re- cycling step. The experiments reported here generate the possibility for various potentially illuminating investigations. Thus, one would like to know whether [TIIDNA probes made with one type of leukemia can be used to detect sequences in the DNA of other leukemias, or related neoplastic disorders. In addition, it is of key importance to determine whether the DNA in the peripheral cells of a leukemic patient in remission still con- tain the leukemia-specific sequences. Further, one would like to understand the significance of the indigenous viral-related sequences shared between normal and cancer cells. The fact that they do not appear to be amplified (Fig. 3) in the leu- kemic cells and that are not necessary for transportation (Goodman, N., Sweet, R., Ruprecht, R. &I., and Spiegelman, S., manuscript in preparation) suggests that these sequences may ultimately tell us more about the origin and evolutionary history of oncogenic viral elements than about the mechanism of their pathogenesis. We thank Drs. David E. Kohne and John J. Holland for helpful discussions during the preparation of this manuscript and Dr. James F. Holland, Roswell Park Memorial Institute, Buff alo, N.Y., for supplying the cell material. This research was sup- ported by the National Institutes of Health, National Cancer In- stitute, Special Virus Cancer Program Contract 70-2049, and by Research Grant CA-02332. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. Axel, R., Schlom, J. & Spiegelman, S. (1972) Nature 235, Hehlmann, R., Kufe, D. & Spiegelman, S. (1972) Proc. Nut. Kufe, D., Hehlmann, R. & Spiegelman, S. (1972) Science Hehlmann, R., Kufe, D. & Spiegelman, S. (1972) Proc. Nut. Schlom, J. & Spiegelman, S. (1971) Science 174,840-843. Gulati, S. C., Axel, R. & Spiegelman, S. (1972) Proc. Nut. Axel, R., Gulati, S. C. & Spiegelman, S. (1972) Proc. Nat. Acad. Sei. USA 69, in press. Baxt, W., Hehlmann, R. & Spiegelman, S. (1972) Nature, in press. Huebner, R. & Todaro, G. (1969) Proc. Nut. Acad. Sei. USA 64,1087-1094. Todaro, G. J. & Huebner, R. J. (1972) Proc. Nut. Acad. Sei. Temin, H. M. (1964) Nat. CancerInst. Monogr. 17,557-570. Temin, H. M. (1964) Proc. Nat. Acad. Sei. USA 52, 323-329. Bernardi, G. (1965) Nature 206,779-783. Miyazawa, Y. & Thomas, C. A,, Jr. (1965) J. Mol. Hid. 11, Manly, K. F., Smoler, D. F., Bromfield, E. & Baltimore, D. (1971) J. Virol. 7, 106-111. Varmus. H. E.. Levinson. W. E. & BishoD. J. M. (1971) Na- 32-36. Acad. Sei. USA 69,435-439. 175, 182-185. Acad. Sei. USA 69, 1727-1731. Acad. Sei. USA 69, 2020-2024. USA 69, 1009-1015. 223-237. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. .I ture Neul Biol. 223, 19-21.' Ruprecht, R. M., Goodman, N. C. & Spiegelman, S. (1972) Biochirn. Riophys. Acta, in press. Britten, R. J. & Kohne, D. E. (1968) Science 161,529-540. Baluda, M. A. & Nayak, D. P. (1970) Proc. Nut. Acad. Sei. Rosenthal, P. N., Robinson, H. L., Robinson, W. S., Hanaf- usa, T. & Hanafusa, H. (1971) Proc. Nat. Acad. Sci. USA Baluda, M. A. (1972) Proc. Nut. Acad. Sei. USA 69,576-580. Wilson, D. E. & Bauer, H. (1967) Virology 33,754-757. Yoshikawa-Fukada, M. & Ebert, J. P. (1969) Proc. Nut. Acad. Sei. USA 64,870-877. Gelb, L. D., Aaronson, S. A. & Martin, M. (1971) Science Varmus, H. E., Weiss, R. A,, Friis, R. R., Levinson, W. & Bishop, J. M. (1972) Proc. Nut. Acad. Sei. USA 69,20-24. Varmus, H. E. & Bishop, J. M. (1972) Nature New Biol. Gelb, L. D., Kohne, D. E. & Martin, M. A. (1971) J. Mol. Temin, H. (1971) J. Nut. Cancer Inst. 46,IIILVII. USA 66,329-336. 68,2336-2340. 172, 1353-1355. 238,189-191. Bioi. 87, 129-145. Reprinted from the Proceedings of The National Academy of Sciences Vol. 70, No. 2, p. 627, February, 1973. Correction. In the article "Nuclear DNA Sequences Present in Human Leukemic Cells and Absent in Normal Leukocytes," by Uaxt, W. G. & Spiegelman, S., which appeared in the December 1972issne of the Proc. Nut. Acud. Sci. USA 69, 3737-3741, the second line, under Annealing Conditions, top right-hand column, p. 3738, should read : "tained 60 A2M) units of cellular DNA,. . . ," and the last sentence of the text, p. 3741, bottom, left-hand column, should read: "The fact that they (shared sequences) do not appear to be amplified (Fig. 3) in the leukemic cells and that they are not necessary for transformation (Good- man, N., Sweet, R., Ruprecht, R. M. & Spiegelman, s., manuscript in preparation) suggests that these sequences may ultimately tell us more about the origin and evolu- tionary history of oncogenic viral elements than about the mechanism of their pathogenesis."