Reprinted from Proc. Nat. Acad. Sei. USA Vol. 69, No. 2, pp. 435-439, February 1972 RNA in Human Leukemic Cells Related to the RNA of a Mouse Leukemia Virus (leukocytes/RNA-DNA hybridization/Rauscher virus/polysomal RNA) R. HEHLMANN, D. KUFE, AND S. SPIEGELMAN Institute of Cancer Research, College of Physicians and Surgeons, Columbia University, New York City, N.Y. 10032 Contributed by S. Spiegelman, December 7, 1971 ABSTRACT Molecular hybridization with radioactively labeled DNA complementary to the RNA of the Rauscher leukemia virus was used to probe for homologous RNA in the polysome fraction of human leukemic cells. The leu- kocytes of 24 out of 27 patients examined contained RNA possessing homology to that of the mouse leukemia agent, but not to that of the unrelated viruses causing mammary tumors in mice or myeloblastosis in chickens. Further, no control human leukocytes or other adult and fetal tissues showed significant levels of the leukemia-specific RNA. It would appear that human leukemic cells contain RNA sequences homologous to those found in a viral agent known to cause leukemia in an experimental animal. The fact that human sarcomas have also been shown to con- tain this type of RNA points to a remarkable parallelism in the leukemias and sarcomas of mice and men. The search for evidence implicating oncogenic RNA viruses in human neoplasia constitutes a major endeavor in present- day cancer research. One approach recently explored stemmed from an attempt to determine via RNA-DNA hybridiza- tion (1) whether human breast carcinomas contain RNA mole- cules homologous to an analogous virus of proven oncogenic potential. Human breast cancer was chosen for the initial effort because of the recent observations (2) in human milk of particles similar to the mouse mammary tumor virus and the demonstration (ref. 3 and manuscript in preparation) that the human particles also contain the enzyme and high molecular weight (70s) RNA characteristic of the animal RNA tumor viruses. As a necessary prelude to attempts with human material, we examined the technical feasibility of using the molecular hybridization technique with mouse mammary tumors and the corresponding viral agent as a suitable experimental model. Hybridizations with radioactively labeled DNA comple- mentary to the RNA of the mouse mammary tumor virus revealed (4) that it was in fact possible to identify in tumor extracts viral-specific RNA in the fraction of ribosomes actively engaged in protein synthesis. The experiments with the mouse model produced the de- tailed technology required to handle human tissue specimens. The mouse mammary tumor virus was chosen to produce the radioactively labeled DNA needed to search for homologous RNA in human tumors. The results obtained (5) were satis- fyingly clear-cut. Of 29 malignant breast tumors examined, 67% showed unmistakable evidence of RNA that can hybridize with mouse mammary tumor virus DNA. None of the 30 preparations from normal or nonmalignant breast tis- sues contained detectable quantities of this type of RNA. Abbreviation: RLV, Rauscher leukemia virus Further, the malignant RNA samples that hybridized with mouse mammary tumor virus DNA showed no ability to com- plex with DNA complementary to the RNA of two leukemo- genic agents, the avian myeloblastosis virus and the murine Rauscher leukemia virus. These last observations in particular gave added credence to the belief that the positive responses observed between RNA from malignant human breast tumors and the mouse-viral DNA were specific and meaningful. In any event, as soon as it became apparent that the experiments with breast tumor material were yielding interesting results, the present investi- gation into human leukemia, and a parallel one in human sarcomas (manuscript in preparation), were instituted. A particle of established leukemogenic potential, the murine Rauscher leukemia virus (RLV) , was chosen to pro- duce the radioactively labeled DNA required. We report here that 89% of leukemic patients in the active phase of the disease contained, in their leukocytes, RNA possessing homology to that of the Rauscher leukemia virus but not to the RNAs of the unrelated mouse mammary tumor virus or the avian myeloblastosis virus. Further, no control leukocytes or other tissues examined showed significant levels of this specific RNA. METHODS Leukocytes were obtained by fractionation of citrated blood samples in inverted syringes. Leukemic cells prepared in other laboratories were sent to us frozen in dry ice. The cells were collected by centrifugation, washed in saline, and then sus- pended in 0.01 M Tris.HC1 (pH 7.4)-0.15 M NaC1-2 mM MgClz containing 5y0 sucrose. They were ruptured with 5-12 strokes of a Dounce homogenizer. Cell membranes and nuclei were removed and the cytoplasmic pellets were collected and used for the isolation of polysomal RNA (4). All annealings were at 37°C in 50% formamide, to minimize fragmentation of the RNA during the hybridization. Hybrids were detected by isopycnic separation in CszS04 gradients (6). RESULTS For comparison, and to see what might be optimally expected with human material, it was of interest to examine the out- comes of analogous hybridizations in instances of two animal leukemias known to be associated with a viral agent. Fig. IA shows the result of hybridization of RLV-[3H]DNA to poly- soma1 plus monosomal RNA, hereafter designated polysomal RNA, prepared from the spleen of an RLV-infected mouse. A peak of [3H]DNA, corresponding to about 3% of the input DNA, is found in the RNA region of the density gradient. The density position of the hybrid indicates that the RNA 435 436 Microbiology: Hehlmann el al. Proc. Nut. Acad. Sci. USA 69 (1978) DNA 7, DNA 17 16 13 10 15 20 25 Fraction Number 5 10 15 20 25 Fraction Number FIG. 1. (A) CszSO4 density profile of Rauscher leukemia virus (RLV) [3H]I>NA hybridized to polysomal RNA obtained from the spleen of an RLV-infected mouse. RLV-13H]DNA was syn- thesized as follows: A 1-ml reaction mixture, containing 100 pg of viral protein purified (6) from plasma, 50 pmol of Tris. IICl (pH 8.3)40 pmol KC1-6 pmol MgC12-2.5 pmol dithio- threitol-0.00125yo NP-40 detergent (Shell Chemical Co.)- 100 pmol (each) of dGTP, dATP, and dCTP-5 X 104 pmol of [WITTP (8000 cpm/pmol), was incubated at 37°C for 180 min. After addition of 0.5% sodium dodecyl sulfate and extraction with an equal volume of phenol-cresol, the 13H]DNA product was purified by Sephadex G-50 chromatography and treated with 0.5 M NaOH at 43OC for 24 hr to hydrolyze any viral RNA pres- ent. In studies such as these, it must be shown before use that the radioactive DNA product bands solely in the DNA density region of a CszSo4 gradient and that it hybridizes with homol- ogous RNA and not to normal cellular RNA. The RNA (designated a5 pltNA) used in the hybridizations is derived from a cytoplasmic pellet consisting of a mixture of monosomes and polysomes (4). Suitable care must be exercised to insure the complete removal of protein contaminants. For the preparation of polysomal RNA, the spleen was disrupted by a Silverson homogenizer at 4°C in two volumes of 5y0 sucrose in TNM buffer [0.01 M Tris.HC1 (pH 7.4)-0.15 M NaCl-2 mM MgCIz]. The suspension was centrifuged at 20,000 g for 15 min at 0°C. The supernatant fluid was then layered on 20 ml of 25% sucrose in TNM buffer containing 200 pg of polyvinyl sulfate per ml and spun for 180 min at 180,000 X g in a 60 Ti rotor (Spinco). The pellet was resuspended in TNM buffer with 0.5yo sodium dodecyl sulfate and the RNA was extracted twice with an equal volume of cresol-phenol (pH 8.0). Nucleic acid was precipitated from the aqueous phase by addition of two volumes of ethanol and 0.1 volume of 4 M NaCl. The polysomal RNA was redissolved in .50%) for1namide-.507~ 5 niM EDTA. The [3H]DNA was denatured by incubation at 80°C for 15 min in 7oy0 form- amide and subsequent quick chilling. 480 pg of polysomal RNA were hybridized to 2000 cpm of RLV-[3H]DNA in 60 pl of 0.4 M NaCL containing 50% formamide. The reaction was incubated for 18 hr at 37"C, mixed with 10 ml of 50%-saturated CszSO4 (starting density 1.52) and centrifuged at 44,000 cpm for 60 hr at 20°C in a ,SO Ti rotor (Spinco). 0.4-ml @-drop) Fractions were collected through an 18-gauge needle from the bottom of the tube and assayed for Cl&COOH-precipitable radioactivity. 1.7 1.6 8 q.5 1.4 1.3 E 0 1.3 Fraction Number Fraction Number DYA DNA 61 0 I 1.4 1.3 1 5 10 15 20 25 5 10 15 20 25 Froction Number Froction Number FIG. 2. (A-D) CszSOI density profiles of RLV-[3H]DNA hybridized to polysomal RNAs of four leukemic samples. Poly- somal RNA was isolated from buffy coats of patients showing clinical manifestation of acute lymphocytic leukemia (A, C, and D) and of acute myelogenous leukemia (B). The cells were dis- rupted with a Dounce homogenizer, and cytoplasmic pellets were prepared as described under Fig. 1B. 300 pg of polysomal RNA were hybridized to RLV-[3H]DNA in 60-pl volumes, and the reactions were analyzed by Cs2S04 density centrifugation. molecule in the complex is much larger than the 4 S-6 S aver- age size of the product DNA. Fig. 1B shows the similar out- come with a rat ascitic monocytic leukemia, originally in- duced (7) with 2O-methy1cholanthreneJ that produces low levels of a C-type particle (8) of unknown specificity. Again, about 3-4y0 of the input t3H]DNA is found in the RNA region at the concentration of polysomal RNA used in the annealing reaction. Similar results have been obtained in hybridizations of RLV-t3H]DNA with polysomal RNA from cat lymphocytes infected with feline leukemia virus and hamster cells (HT2) transformed with the Moloney sarcoma virus. Fig. 2A-D shows representative profiles in Cs2SO4 gradients of annealing reactions between RLV-[3H jDNA and poly- somal RNA preparations derived from human-leukemic leukocytes (buffy coats). As may be seen, the results are com- parable to those obtained with the animal systems (Fig. 1A and B). In Fig. 2A, 15y0 of the input RLV-DNA is hybridized; the other three samples gave 3-5% formation of the hybrid complex. Again, as in the hybridizations with polysomal RNA from the animal leukemic material (Fig. 1A and B) , the loca- tions of the complexes in the gradients of Fig. 2 indicate that the RNA is much larger than the DNA, and determines the density of the DNA-RNA hybrid structure. The positive responses observed (Fig. 2A-D) with the leukemic polysomal RNAs are in contrast to the negative Data are expressed in this and subsequent figures as cplOm, counts per 10 min. (B) . CsnSOa equilibrium gradient centrifugation of RLV- [3H]DNA annealed to 600 pg of polysornal RNA obtained from ascites cells of a rat leukemia. The cells were collected from the rat ascitic fluid, washed in phosphate-buffered saline (pH 7.4) by low-speed centrifugation (2500 rpm, Sorvall), and resuspended in two volumes of TNM buffer containing 5% sucrose. The cells were disrupted by 5-12 strokes of a Dounce homogenizer, and treated as in part A. Proc. Nut. Acad. Sei. USA 69 (1972) Viral RNA in Human Leukemia Cells 437 reactions with polysomal RNAs from normal leukocytes (Fig. 3A), from phytohemagglutinin-stimulated lymphocytes (Fig. 3B), from leukocytes of a leukemic patient in clinical remission (Fig. 3C), and from human fetal lung (Fig. 30). In none of these samples does one find significant amounts of taH]DNA in the RNA region of the density gradient. An informative way to compare positive and negative re- actions is via a concentration curve, as shown in Fig. 4. Here, the same amount of RLV-taH]DNA is annealed to increasing concentrations of polysomal RNA and the percent of [aH]- DNA complexed to the RNA is determined by isopycnic sep- aration in C&04 gradients. It is clear that no detectable re- sponse with polysomal RNA from normal leukocytes is ob- served, even at 14 mg of RNA per ml. On the other hand, the percent of RLV-DNA complexed to polysomal RNA from the buffy coat of a patient with acute myelogenous leukemia shows no evidence of saturation within the concentration range tested. Several leukemic buffy coats were obtained in amounts sufficient to permit the determination of such concentration curves, and comparable results were obtained. It should be further noted that similar concentration curves were also run with polysomal RNA preparations from phytohemagglutinin- stimulated lymphocytes with results not significantly different from those seen with normal leukocytes (Fig. 4). Since it is impractical to present the Cs2S04 gradient profile of every sample examined, a more convenient recording of our findings was used. After correction for background counts, the sum of the tritium counts in the RNA density region (be- tween densities 1.62 and 1.68 g/ml) was used to measure the amount of DNA complexed to RNA. To achieve the accuracy desired, 10-min counts were taken on each sample. An opera- tional mean background and its standard deviation (S) were empirically determined for individual machines by the total counts/lO min of three tubes in the negative regions (e.g., tubes 2, 3, and 4) of each of 60 gradients. The convention was adopted that all specimens with less than three standard deviations of the background mean counts/lO min would be considered negative. w. 0". Du 6L c Froctian Number Fmclion Number FIG. 3. (A-D) CszSC.4 density centrifugation of RLV-L3H]- DNA after annealing to polysomal RNA isolated from (A) normal white buffy coat, (B) phytohemagglutinin-stimulated lympho- cytes (C), buffy coat of a leukemic patient in clinical remission, and (D) fetal lung. The polysomal RNA input was 300 pg, ex- cept in (C) where 1000 pg were used per 60-p1 hybridization re- action. Subsequent analysis on CszSOa was described in Fig. 1. Table 1 lists the leukemic buffy coats tested for RNA com- plementary to RLV-DNA, with results recorded as the total counts/lO rnin (corrected for background) in the RNA densit,y region, and as multiples of the mean background standard deviation. The samples came from patients in the active phase of the disease, as determined by differential counts of their peripheral leukocytes, and include acute myelogenous leukemia, acute lymphocytic leukemia, chronic lymphocytic TABLE 1. Test for viral-specific RNA in human leukemic cells RNA-region Counts per Tissue Counts/lO min 10 min/S Reaction 402 515 387 233 1413 308 832 393 1271 478 328 631 813 564 912 118 682 450 684 361 322 665 220 1315 528 590 550 4.7 + 6.0 + 4.5 + 2.7 - 16.4 + 3.6 + 9.7 + 4.6 + 14.8 + 5.6 + 3.8 + 7.3 + 9.5 + 6.6 + 10.6 + 1.4 7.9 + 5.2 + 8.0 + 4.2 + 3.7 + 7.7 + 2.6 - 15.3 + - 6.1 + 6.9 + 6.4 + Results of hybridization reactions between RLV-[3H] DNA and polysomal RNA isolated from human leukemic cells. In two cases (If and LS) cells were taken in 1- and 3-year intervals. 200- 1000 pg of polysomal RNA of each sample were hybridized to 2000 cpm of RLV-[SH]DNA and the reactions were analyzed by CszSO4 equilibrium centrifugation. The amount of DNA banding in the RNA region of the gradient (between densities 1.62 and 1.68) was then determined. The results are expressed as counts/ 10 rnin (corrected for background) banding in the RNA region for each RNA sample tested, and as multiples of S, the operational standard deviation (see text and legend to Fig. 5). The annealing reaction is considered positive only if the counts/l0 rnin per RNA region is greater than 3S, thus providing 99.9% confidence sta- tistically. Of the 27 samples tested, 89% were positive. 438 Microbiology: Hehlmann et al. Proc. Nut. Acad. Sei. USA 69 (1972) /. ,/ a 3, Normal Leukocytes I60 260 360 460 560 660 760 860 960 RNA (pg/6Opl) FIG. 4. Comparison of hybridization reactions between RLV-[3H]DNA, leukemic polysomal RNA, and normal buffy- coat polysomal RNA at various RNA concentrations. The in- dividual annealing reactions were analyzed by CszS04 density centrifugation, and the percent DNA hybridized was determined by the counts/lO min of [3H]DNA (corrected for background) banding in the RNA region (between densities 1.62 and 1.68) of the gradients. leukemia, and monocytic leukemia. Fig. 5 presents a conve- nient pictorial summary of the reactions observed with leuke- mic (Table 1) and various control tissues. Of the 27 leukemic samples examined, only three fell in the negative range. The remaining 24 showed positive reactions, with some exceeding those observed with the animal leukemic material (Fig. IA and B). In contrast, none of the polysomal RNA preparations from 33 normal tissues gave a reaction that could be unambiguously designated as positive. These prep- arations include RNAs from normal leukocytes, phytohemag- glutinin-stimulated lymphocytes, and various normal human adult and fetal tissues. The fact that 89% of the leukemic polysomal RNAs yielded unmistakable positive hybridizations with RLV-DNA , whereas none of the 33 control tissues exhibited this type of response, argues for the specificity of the annealing reactions. Further support for this conclusion can be obtained by the use of a [3H]DNA complementary to the RNA of the mouse mammary tumor virus or to that of the avian myeloblastosis virus. We have shown (manuscript in preparation) that RLV-DNA and the homologous viral RNA do not crosshy- bridize significantly with the corresponding nucleic acids of either the mammary tumor virus or the avian agent. If the annealing reaction is specific, one would not expect a leukemic polysomal RNA, positive for a reaction with RLV-DNA to show the ability to hybridize either with mouse mammary tumor viral-DNA or avian myeloblastosis viral-DNA. Fig. 6 shows that these expectations are realized. None of three other leukemic polysomal RNAs giving a positive (550-910 counts/lO min above background) reaction with RLV-DNA responded to either of the other two viral DNAs. Stimulated by the known viral etiology of animal leukemias (9, 10) , others have sought for similar evidence in the human disease. Particles resembling the C-type viruses of the avian and murine leukemias have been occasionally detected by electron microscopy (11). Of perhaps more immediate rele- vance is the report (12) of components in human leukemic cells and sera that can interact with fluorescent-labeled antibodies against the Rauscher leukemia virus. The present investigation is not directly aimed at resolving the issue of a viral etiology of human cancer. It is concerned rather with determining whether viral-related information, in terms of sequence similarities, can be detected in human neoplastic cells. It is worth explicitly noting the questions that remain un- answered by the data described here. They provide no measure of the degree of homology between the RLV-DNA and the RNA detected in human leukemic cells. To settle this issue will require the synthesis of RLV-DNA in amounts adequate to permit comparative saturation curves of labeled RLV-RNA and the relevent RNA strands from human leukemic cells. The fact that hybrids are found in the RNA region of the density gradient implies that the RNA is much larger than the DNA product. However, experiments must now be designed and performed that will determine how large this RNA is, how much viral sequence it contains, and whether the viral information is covalently linked to normal cellular message. The latter point is particularly relevant to whether the viral information is integrated into the genome of the cancer cell. Finally, more highly purified mRNA fractions will be re- quired to make the quantitative aspects more certain. In par- ticular, it is apparent that a negative outcome of a hybridiza- Human Leukemias Normal Human Tissues FIG. 5. Results of hybridization reactions with RLV-[3H]- DNA and polysomal RNA from leukemic and normal human cells. The RNA was derived from leukemic and normal leuko- cytes, phytohemagglutinin-stimulated lymphocytes (A), a human lymphocyte cell line, NC37 (V), lymph nodes, other adult tissues: liver (A), spleen (X), intestine (0), striated muscle (0), and fetal tissues: liver (A), lung (V), limbs (0), placenta (0). The reactions were then subjected to CszS04 equilibrium density centrifugation as described under Fig. 1. The amount of [3H] DNA, expressed as cplOm corrected for background banding in the density region of RNA (between densities 1.62 and 1.68), was determined for each reaction. An operational mean and standard deviation (8) were determined for each counter by the total cplOm of three tubes [e.g., 2, 3, 41 of each of 60 gradients. The cpl0m [3H]DNA corrected for background banding in the RNA region of the gradient was then divided by the appropriate oper- ational standard deviation. Any reaction with less than 3s cpl0m in the RNA region is considered negative, thus providing 99.9% confidence that those reactions retained as positive (greater than 38) are statistically significant. ALL, acute lym- phatic leukemia; CLL, chronic lymphatic leukemia; AML, acute myelogenous leukemia; ML, monocytic leukemia. Proc. Nat. Acad. Sei. USA 69 (1972) Viral RNA in Human Leukemia Cells 439 ?w::I 5 '* - a :2 14 I 13 ;2b I 5 IO 15 20 25 17 E I6 g 0 15 2. Bl l4F; 13 I 5 IO 15 20 25 Fraction Number I FIG. 6. CszSOa equilibrium gradient centrifugation of (A) mouse mammary tumor virus [3H]DNA and (B) avian myelo- blastosis virus [3H]DNA hydridized (each) to 400 pg of human leukemic polysomal RNA. Hybridization conditions and CspSOa gradient analytjis as in Fig. 1. tion test cannot be accepted as evidence for absence of the relevant RNA. From the data presented, one can conclude at best that the probability of finding RNA homologous to Rauscher leukemia viral RNA is much greater in human leukemic cells than in normal tissues. Indeed, if the provirus (13) or oncogene hypotheses (14) are valid, some part of this information might be found with more sensitive tests in presumably normal cells derived from fetal tissues or phytohemagglutinin-stimulated lymphocytes, both of which have been reported (14, 15) to possess the group-specific antigen of the mammalian leukenio- genic viruses. While the experiments described here do not constitute definitive proof of a viral etiology of human leukemia, they provide rather compelling evidence for the presence in human leukemic cells of RNA posessing sequences homologous to those found in a viral agent known to cause leukemia in an experimental animal. From what we know of animal systems, one would predict that a similar situation would be found in human sarcomas, a prediction that has been confirmed and will be detailed in another publication. We thank the following for supplying cell material: Dr. M. Stoker (Imperial Cancer Research Fund, London), Dr. R. Powles (Royal Marsden Hospital, London), Dr. R. Gallo (National In- stitutes of Health, Bethesda, Md.), Dr. J. F. Holland (Roswell Park Memorial Institute, Buffalo, N.Y.), Drs. J. H. Burchenal and H. F. Oettgen (Sloan-Kettering Institute, New York), and Drs. E. F. Osserman and M. Farhangi (InstlitUte of Cancer Re- search, Columbia University, New York). We greatly appreciate the excellent technical assistance of Jeanne W. Meyers, Lee E. Hindin, and Mildred Rothenberg. This research was supported by the National Institutes of Health, National Cancer Institute, Special Virus Cancer Program Contract 70-2049 and Research Grant CA-02332. Hall, B. D. & Spiegelman, S. (1961) Proc. Nut. Acad. Sei. Schlom, J., Spiegelman, S. & Moore, D. H. (1971) Nature Schlom, 3. & Spiegelman, S. (1971) Science 174, 840-843. Axel, R., Schlom, J. & Spiegelman, S. (1972) Proc. Nut. Acad. Sci. USA 69, in press. Axel, R., Schlom, J. & Spiegelman, S., Nature, in press. Spiegelman, S., Burny, A., Das, M. It., Keydar, J., Schlom, J., Travnicek, M. & Watson, K. (1970) Nature 227, 563- 567. 1. 2. 3. 4. 5. 6. USA 47,137-146. 231,97-100. 7. 8. 9. 10. 11. 12. 13. 14. 15. Shay, H., Gruenstein, M., Marx, H. E. & Glazer, L. (1951) Cancer Res. 11,29-34. Weinstein, R. S. & Moloney, W. C. (1965) Proc. SOC. Exp. Biol. Med. 118, 4.59-461. Eilermann, V. & Bang, 0. (1908) Centralb. f. Bakt., Abty. I Gross, L. (1970) in Oncogenic Viruses (Pergamon Press, Oxford, England). Dmochowski, L., Yumoto, T., Grey, C. E., Hales, R. L., Langford, P. L., Taylor, H. G., Freireich, E. J., Schullen- berger, C. C., Shively, J. A. & Howe, C. D. (1967) Cancer Fink, M. A,, Karon, M., Rauscher, F. J., Malmgren, R. A. & Orr, H. C. (1965) Cancer, 18, 1317-1321. Temin, H. M. (1971) J. Xat. Cancer Inst. 46,111-VI1. Huebner, R. J., Kelloff, G. J., Sarma, P. S., Lane, W. T., Turner, H. C., Gilden, It. V., Oroszlan, S., Meier, H., Myers, D. D. & Peters, R. L. (1970) Proc. Nut. ilcad. Sci. USA 67, 366376. Hellman, A. & Fowler, A. K. (1971) Nature New Biol. 46, 595-610. 20, 760-777. 233, 142-144.