[CANCER RESEARCH 41, 2483-2490, June 1981] 0008-5472/81 /0041-0000$02.00 Evidence for Double Replication of Chromosomal DMA Segments as a General Consequence of DMA Replication Inhibition David M. Woodcock1 and Ian A. Cooper Haemato/ogy Research Unit, Cancer Institute, 481 Little Lonsdale Street, Melbourne, ABSTRACT We have previously presented evidence that a transient inhibition of DNA synthesis by a pulse of 1-/J-D-arabinofuranosylcytosine (ara-C) results in a disruption of the pattern of replication of the chromosomal DNA of cultured human cells, resulting in some DNA segments being replicated more than once in a single S phase. Further evidence is presented in this paper that this effect is not a specific property of the ara-C molecule in that a similar effect is produced in cells by a pulse of 9-yS-D-arabinofuranosyladenine (ara-A) and also by a pulse of cycloheximide. The activated form of ara-A and ara-C (the triphosphates) both inhibit DNA synthesis at the level of the polymerase. Double replication following an ara-A pulse dem onstrates that double replication after an ara-C pulse is not caused by some specific property of the ara-C molecule which might be unrelated to any effect on DNA synthesis. However, cycloheximide is an inhibitor of mammalian protein synthesis and inhibits DNA synthesis only indirectly, probably through a consequent deficiency of DNA-packaging proteins. Hence, the occurrence of double replication of chromosomal DNA seg ments following a pulse of cycloheximide is consistent with this phenomenon being a general and nonspecific consequence of the freezing of DNA replication forks. INTRODUCTION A transient inhibition of DNA synthesis in mammalian cells results in chromosome aberrations (10, 12), cell death (4), and even oncogenic transformation (1). A compound which is a potent inhibitor of DNA synthesis but which has minimal effect on RNA synthesis is ara-C2 (3, 4). ara-C exhibits S-phasespecific cytotoxicity of mammalian icity is likely to be the consequence induced by the ara-C treatment induced chromosome aberrations vious explanation since, following cells (5), and this cytotoxof chromosome aberrations (10, 11 ). However, ara-Cand cell death have no ob a pulse of ara-C, cells re cover their ability to synthesize new DNA chains and ligate them to high molecular weight (10, 27), and also cells treated with ara-C do not show degradation of their preformed DNA (19, 28). Hence, more subtle events inside the cell must be causing the chromosome damage and cell killing resulting from a pulse of ara-C. We have presented evidence that the mechanism controlling the pattern of replication of the chromosomal DNA of cultured 'Recipient of a grant from the National Health and Medical Research Council of Australia. To requests for reprints should be addressed. 2 The abbreviations used are: ara-C, 1-/î-D-arabinofuranosylcytosine; CHO, Chinese hamster ovary; ara-A, 9-/3-p-arabinofuranosyladenine; BrdUrd, 5-bromodeoxyuridine; FdUrd, 5-fluorodeoxyuridine; dCyd, deoxycytidine; dGuo, deoxyguanosine; dThd, thymidine. Received March 3, 1980; accepted February 13, 1981. JUNE 1981 Victoria 3000. Australia cells of human origin is disrupted following a pulse of ara-C, resulting in some DNA segments being replicated more than once in a single S phase (26, 28). These data indicated that, following a pulse of ara-C, DNA strands were synthesized off DNA template strands which had themselves been synthesized only a few hr prior to the ara-C pulse. This can be visualized as a partial endoreduplication of the chromosomal DNA occurring within a single S phase. We have presented evidence against a number of alternative explanations of this effect including DNA repair synthesis, cells entering a second S phase, multiple rounds of mitochondrial or Mycoplasma DNA synthesis, and DNA recombinational and exchange events (26, 28). This effect has been observed in 4 different mammalian cells lines. As well as the 2 cell lines for which data were presented in previous publications (26, 28), this effect has also been observed with the K562 human leukemic cell line and with CHO cells.3 In this paper, we demonstrate that a similar effect on DNA replication is also produced by a pulse of ara-A which, like araC, is a direct inhibitor of DNA polymerase (6) and also by a pulse of cycloheximide, an inhibitor of protein synthesis (15, 17) which inhibits DNA synthesis by an indirect mechanism probably related to its causing a deficiency in proteins neces sary to package newly replicated DNA (20). Hence, we con clude that disruption of the pattern of replication of the chro mosomal DNA resulting in double replication of some chromo somal DNA segments is a general and nonspecific conse quence of a temporary interruption to DNA replication. MATERIALS AND METHODS Cells and Culture Conditions. Experiments used Crow cells, a human cell line derived from a retroorbital hemangioma in the laboratory of Dr. T. R. Bradley (Cancer Institute, Melbourne, Australia). Cells were grown in suspension culture in a-medium supplemented with 10% fetal calf serum (Flow Laboratories, Stanmore, New South Wales). Cells grew with a doubling time of approximately 24 hr. Cell cultures were subdivided twice a week. Chemicals. Radiochemicals were purchased from the Radiochemical Centre, Amersham, United Kingdom. Nucleosides, nucleoside analogs, cycloheximide, and Colcemid were pur chased from the Sigma Chemical Co., St. Louis, Mo. All chem icals were analytical reagent grade. Isolation of Nuclei. Nuclei were isolated from whole cells by the method of Lerner ef ai. (14). Detection of DNA Repair Synthesis. The method used to detect DNA repair synthesis was similar to that used previously (28). BrdUrd (10 JIM), FdUrd (1 JIM), and dCyd (10 fiM) were added to a culture of logarithmically growing Crow cells which were allowed to continue DNA synthesis for 60 min to allow 3 D. M. Woodcock and I. A. Cooper, unpublished observations. 2483 Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. D. M. Woodcock and I. A. Cooper replicons synthesizing DNA at the beginning of the experiment to have their replication completed in the presence of unlabeled BrdUrd. At this point, inhibitor (either ara-A or cycloheximide) was added to one-half of the culture (test cells), and 1 hr later, these cells were resuspended in fresh medium containing 10 ¡UM BrdUrd, 1 ¡UM FdUrd, and [3H]dGuo plus [3H]dCyd, both at 0.5 ¡uCi/ml.Control cells were transferred in this medium after the 1 hr in the presence of unlabeled BrdUrd. Cells were collected 4 hr after being placed in the radioactive medium. All procedures were performed using minimal lighting, especially of fluorescents. DNA extraction was performed as described previously (26). The extracted DNA was dissolved in 0.15 M NaCI:0.015 M trisodium citrate and further purified by treatment with 50 /ig RNase-A per ml (Worthington Biochemical Corp., Freehold, N. J.) at 37°for 45 min, followed by 100 /ig pronase per ml (Calbiochem, Carlingford, New South Wales; nucleasefree grade; self-digested) for a further 15 min. The aqueous phase was then extracted with chloroform:octanol (24:1, v/v) for 10 min at room temperature. After centrifugation, DNA was precipitated from the resultant aqueous phase with cold ethanol. CsCI isopyknic gradient fractionation of DNA was performed as described previously (26). Before CsCI gradient fractionation, DNA was sheared by 6 passages through a 26gauge needle which produced DNA fragments of about 6000 base pairs (26). The light-light DNA fraction from each DNA sample was purified through 2 cycles of neutral CsCI gradients, with 0.1 -ml aliquote being taken from each of the fractions from the first round of neutral CsCI gradients (diluted previously with 0.5 ml of buffer for A26o readings) for the determination of 3H content as described previously (28). The purified light-light DNA was analyzed on alkaline CsCI gradients as described previously (26, 28). Detection of Double Replication. In this study, the method used to detect aberrant double replication of DNA segments was the second of the 2 methods described previously (26, 28). Each experiment used a logarithmically growing culture of Crow cells which were labeled with [3H]dCyd at 0.5 ¡uCi/ml(22 odology to demonstrate double replication is used in the ex periments reported in this paper. This method is based on the basic nature of DNA semiconservative synthesis whereby, in a given cell cycle, each of the strands of the parental DNA duplex acts as a template for the synthesis of a daughter strand, with the resultant replicated duplex containing one parental and one daughter DNA strand. After a DNA segment is replicated in a particular S phase, each resultant DNA duplex must contain one parental and one newly synthesized strand. If a duplex is present with a daughter strand opposite a daughter strand, it could not be the product of normal semiconservative synthesis in a single S phase. However, daughter-daughter duplexes could be formed by: (a) cells entering a second S phase; (to) DNA repair synthesis; (c) DNA recombinational and exchange events; (cOmultiple rounds of mitochondrial DNA synthesis; (e) multiple rounds of Mycoplasma DNA synthesis; or ( f) abnormal reinitiation of DNA synthesis in DNA segments already repli cated earlier in that S phase, resulting in double replication of those DNA segments. To demonstrate unequivocally that the production of daughter-daughter duplexes was due to double replication (Mechanism f), all other possible mechanisms must be eliminated. To demonstrate the presence of daughterdaughter duplexes following a pulse of ara-C, cells were given a pulse of [3H]dCyd followed by a 1.5-hr chase with a 2000fold excess of unlabeled dCyd. After the 1.5-hr chase period, one-half of the cells was made 100 ¡UM in ara-C and 1 hr later transferred to fresh medium containing BrdUrd which resulted in any DNA synthesized after the ara-C pulse being density labeled. Control cells were placed directly into this medium after the 1.5-hr chase. The light-heavy DNA fraction from control and test cells was purified using 2 cycles of neutral CsCI gradients. The purified light-heavy DNA which was the DNA synthesized after the time of the ara-C pulse was disso ciated, and the strands were analyzed on alkaline CsCI gra dients. With the DNA from control cells, effectively all of the 3H label was present at the density of the heavy strands, the daughter strands which were synthesized after the time of the addition of BrdUrd (26, 28). That any 3H label was present in Ci/mmol) for 1 hr. The cells were transferred to fresh medium containing 50 ¡UM unlabeled dCyd for 1.5 hr to chase the [3H]- the heavy strand of this light-heavy DNA fraction of control dCyd. After the 1.5-hr chase period, inhibitor was added to cells might be due to (a) the chase procedure not being 100% one-half of the culture (test cells). One hr later, test cells were effective and (to) interspersion of DNA segments replicated at transferred to fresh medium containing 10 /¿MBrdUrd, 1 JUM different times during S phase (9). Note that neither of these FdUrd, and 10 ¿IM dCyd. Control cells were placed directly into mechanisms would result in nucleosides being incorporated into a DNA template strand. However, when the light-heavy the BrdUrd-containing medium after the 1.5-hr chase. If ColDNA from ara-C-pulsed cells was banded in alkaline CsCI cemid was included in the protocol, it was added at 1 ¡ug/mlto all media from the end of the 3H-labeling period onward. Cells gradients, a high proportion (one-third to one-half) of the 3H were collected from the BrdUrd-labeling medium at the times label was present at the light-strand density, the density of the template strand from which the BrdUrd-containing daughter indicated in the captions of the appropriate charts. DNA was extracted as described above, and the light-heavy DNA frac strand had just been synthesized (26, 28). In the case of araC, evidence has been presented that these 3H-labeled template tions was purified through 2 cycles of neutral CsCI gradients. The strands of the purified light-heavy DNA were dissociated strands were not due to some cells entering a second S phase since a similar distribution of 3H label was obtained whether or and analyzed on alkaline CsCI gradients. Procedures were as not Colcemid had been added to the medium to prevent mitosis described previously (26). Conceptual Basis of Method to Demonstrate Aberrant (26). Nor was it due to mitochondrial DNA synthesis since Double Replication. In previous publications (26, 28), we have similar results were obtained with DNA from whole cells or presented evidence that, following a temporary inhibition of isolated nuclei (26). Cells were routinely tested (21 ) and found DNA synthesis by a pulse of ara-C, some DNA synthesis is to be free of Mycoplasma contamination. No evidence could be found with the cell line used (GK cells) for DNA repair reinitiated in DNA segments replicated earlier in that same S synthesis following a pulse of ara-C under experimental con phase, resulting in double replication of some segments of the chromosomal DNA. Two types of experiments were presented ditions where this aberrant form of DNA synthesis could be in support of this hypothesis. The second more definitive meth readily demonstrated (28). Also direct experimental evidence 2484 CANCER RESEARCH Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. VOL. 41 Aberrant DNA Replication after DNA Synthesis Inhibition was presented against DMA recombinational and exchange events being the explanation of these results (26). Hence, it was concluded that the most feasible explanation for these results was that, following the temporary interruption to DNA synthesis due to the pulse of ara-C, the mechanism which controls the sequence of replication of the chromosomal DNA was disturbed, with some DNA synthesis being abnormally reinitiated in DNA segments which had already been replicated earlier in that S phase, resulting in aberrant double replication of some DNA segments. S 0.6 18 0.4 1.7 I •-* 1.6 0.15 1.1 6.000 0.10 1.7 ¡..000 0.05 1.6 2.000 0 0 1.9 IS 1.8 10 1.7 RESULTS To test whether this effect was a peculiarity of the ara-C molecule itself or was perhaps a more general phenomenon, similar experiments were performed with ara-A. ara-A is a purine nucleoside analog whereas ara-C is a pyrimidine analog. The activated form of ara-A, like that of ara-C, inhibits DNA replication directly at the level of the polymerase (6). Using the experimental methodology as described above, logarithmically growing Crow cells were labeled with a 1-hr pulse of [3H]dCyd, and the label was chased for 1.5 hr with excess unlabeled dCyd. After the chase period, ara-A at 0.5 mw was added to one-half of the cells. One hr later, these cells were transferred to fresh BrdUrd-containing medium. Crow cells after 1 hr in the presence of 0.5 mM ara-A had [3H]dThd incorporation reduced by more than 90% (not illustrated). The light-heavy DNA frac tion from control and ara-A-pulsed cells were isolated and purified through 2 cycles of neutral CsCI gradients, and the strands of the purified light-heavy DNA fractions were sepa rated and banded in alkaline CsCI gradients (Chart 1). As was the case with ara-C (26, 28), a pulse of ara-A resulted in a high proportion (approximately one-third) of the 3H label in the isolated light-heavy DNA from the cells treated with inhibitor banding as a peak at light-strand density (Chart 1F), the density of the template strand from which the BrdUrd-containing strand had been synthesized. This peak was absent in control DNA (Chart 1E). In the experiment illustrated, Colcemid jug/ml (to inhibit mitosis) had been included in all media the end of the 3H-labeling period onward. Similar results cell at 1 from have also been obtained without the addition of Colcemid to the experimental protocol (not illustrated). Any 3H label in the daughter strands in the alkaline CsCI gradients of Chart 1, £and F, could be present as noted above because of the chase procedure not being totally effective as well as through interspersion of DNA segments replicated at different times during the S phase (9). The amount of this 3H <o N -1 I x ca TS 10 20 Fraction Number Chart 1. Testing for aberrant reinitiation and double replication following a pulse of ara-A. Logarithmically growing Crow cells were labeled with a 1-hr pulse of [3H]dCyd, and the label was chased for 1.5 hr with a 2000-fold excess of unlabeled dCyd. After the chase period. ara-A at 0.5 mw was added to one half of the cells (test cells). One hr later, the test cells were transferred to fresh medium containing 10 fiM BrdUrd. 1 fiM FdUrd. and 10 JIM dCyd. The other half of the cells (control cells) was transferred directly to this medium after the 1.5-hr chase. Control cells were incubated for 3 hr in this density-labeling medium and test cells for 4 hr to allow a comparable proportion of DNA to be replicated in the presence of BrdUrd in control and test cells. Colcemid (1 fig/ml) was present in all media from the end of the 3H-labeling period onward. DNA was extracted and purified as described in "Materials and Methods." The light-heavy DNA fraction was purified through 2 cycles of neutral CsCI gradients, and the strands of the purified light-heavy duplex were dissociated and separated in alkaline CsCI gradients. A. and 6. initial neutral CsCI gradients; C and D. second cycle of neutral CsCI gradients; E and F, alkaline CsCI gradients. A, C, and £.control cell DNA; B, D. and F. test cell DNA. The fractions pooled from the neutral CsCI gradients are indicated by the oars. The 10 fractions on the light-density side of fractions pooled from the light-heavy peak from the second neutral CsCI gradients were assayed for 3H content to check for contamination of the light-heavy fraction with light-light DNA. Total cpm recovered from the alkaline CsCI gradients were 3896 in E and 2582 in F. label is only a very small proportion of the total label incorpo rated (see Chart 2 of Ref. 28). In order to characterize this 3H longed low level of incorporation of [3H]nucleoside. To deter mine the amount of unincorporated [3H]dThd remaining in the label in the daughter strands and to see whether variations in chase conditions and in the type of 3H-labeled nucleoside could reduce the amount of 3H present in this fraction, the use of [3H]dThd instead of [3H]dCyd was investigated. Cells were labeled for 1 hr with 0.5 /iCi of [3H]dThd per ml and then cells at increasing times after removal from the label, BrdUrd (10 /UM)plus FdUrd (1 /IM) were added to portions of the culture after 0.5, 1, and 1.5 hr in fresh medium. The cells were allowed to synthesize DNA for 4 hr in the presence of BrdUrd before being collected. As has been demonstrated to occur with other mammalian cells (22), Crow cells show a preferential utilization of dThd over BrdUrd since the light-heavy hybrid DNA pro duced by these cells in the presence of FdUrd and BrdUrd plus low levels of dThd was found to band in neutral CsCI gradients at a lower density than would be expected from the molar ratio of BrdUrd and dThd in the medium (not illustrated). Hence, 4 hr of DNA synthesis in the presence of BrdUrd and FdUrd transferred to fresh medium without unlabeled nucleosides. The use of excess unlabeled nucleosides as a chase procedure was not used because of the inhibitory effect of excess dThd on cell cycle progression (29) and because of the objection that high levels of exogenous nucleosides might cause a large expansion in the intracellular pool of dThd, diluting unincor porated intracellular [3H]dThd and hence resulting in a pro- JUNE 1981 Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. 2485 D. M. Woodcock and I. A. Cooper would have resulted in any remaining [3H]dThd being incorpo method used, described in detail in "Materials rated into the BrdUrd-containing is similar to, but more sensitive than, the method used previ ously to test for repair-type synthesis following a pulse of araC (28). The rationale for the method to detect DNA repair was to test for the incorporation after treatment with a potentially DNA-damaging agent of short patches of nucleosides into regions of the DNA not involved in semiconservative replication during the time of the experiments. These short patches of nucleosides incorporated independently of replication were taken as regions of presumptive DNA repair. Cells were incu bated in medium containing unlabeled BrdUrd for 1 hr before addition of 0.5 mw ara-A. After 1 hr in the presence of the ara- daughter strands. DNA from these cells was extracted and banded in alkaline CsCI gra dients. The proportion of total [3H]dThd present in the heavy strands progressively declined with time of chase, with 1.5 hr of chase resulting in 0.58% of total cpm present in this region of the gradient (Table 1). In attempts to reduce this apparent residual incorporation of [3H]dThd, it was not possible to in crease the rate of utilization of unincorporated intracellular [3H]dThd by chasing in medium containing dialyzed fetal calf serum since Crow cell viability after alyzed serum. After [3H]dCyd dCyd, the amount cells showed growth retardation and loss of even short periods in media containing di labeling and chase with excess unlabeled of [3H]dCyd present in heavy strands after and Methods," A, cells were suspended in fresh medium containing BrdUrd and [3H]dCyd plus [3H]dGuo. The control cells were transferred chase procedure, this level was obtained after only 1 hr of chase. Hence, we conclude that labeling with [3H]dCyd fol directly into this medium following the 1 hr in the presence of unlabeled BrdUrd. Cells were incubated initially in nonradioactive medium containing BrdUrd so that, after DNA extraction and shearing of the DNA for banding in CsCI gradients, no DNA fragments would be present which contained an interface be tween a 3H-labeled BrdUrd-substituted region and an unlabeled lowed by chase with excess unlabeled dCyd was superior to that using [3H]dThd followed by a chase without unlabeled nucleosides. The similar level of 3H label in daughter strands unsubstituted region. The prior addition of unlabeled BrdUrd ensured that, in effectively all fragments where there was an interface between a 3H-labeled and unlabeled region in a DNA after 1.5 hr of chase with both procedures would be consistent with interspersion being the primary explanation of this appar ent residual incorporation. However, about 0.5% of total label incorporated may simply represent a practical lower limit of efficiency of any chase procedure which will always be limited by factors such as reincorporation of labeled nucleosides from cells damaged during experimental manipulation. However, neither interspersion or continued incorporation could intro duce 3H label into template strands. To ensure that the peak of 3H-labeled light strands from ara- strand, both regions were BrdUrd substituted. [Note: DNA was sheared to fragments of approximately 6000 base pairs (26).] Hence, all [3H]nucleosides incorporated by normal semiconservative DNA replication would be present in the light-heavy 0.5, 1, or 1.5 hr of chase were similar to those obtained with [3H]dThd (Table 1) with a level of 0.59% at 1.5 hr compared to 0.58% with [3H]dThd. However, with [3H]dCyd labeling and A-pulsed cells was not present due to contamination of the light-heavy DNA fraction with light-light DNA, the 10 fractions on the light-density side of the fractions pooled from the lightheavy peak from the second cycle of neutral cell gradients were assayed for 3H content (Chart 1C, control cell DNA; Chart 1D, ara-A-pulsed cell DNA). In both control and test cell the light-heavy DNA was well separated from light-light and the possible 3H label cross-contamination between DNA species would be too low to account for the amount DNA, DNA, the 2 of 3H label in template strands in the light-heavy DNA of ara-A-pulsed cells. Since DNA repair synthesis is the most feasible alternative explanation of these results, these cells were tested for the induction by a pulse of ara-A of DNA repair-type synthesis. The Table 1 Comparison of 3H-labeling procedures Cells were incubated with 0.5 nC\ of [3H]dThd or [3H]dCyd per ml for 1 hr. Cells were then transferred to fresh medium. For the [3H]dCyd-labeled cells, this medium contained 50 /IM unlabeled dCyd. At 0.5-hr intervals, BrdUrd (10 /IM) and FdUrd (1 /IM) were added to portions of each culture. After 4 hr in the presence of BrdUrd, cells were collected, and the DNA was extracted and banded in alkaline CsCI gradients. In all gradients, total label recovered exceeded 10°cpm. % of total 3H cpm in heavy strands [3H)dCyd labeling with Time of chase (hr) [3H]dThd labeling excess unlabeled dCyd chase 0.5 1 1.5 1.13 0.72 0.58 1.02 0.57 0.59 2486 DNA fraction of the CsCI gradients, with none in DNA which banded at the density of light-light However, if DNA repair synthesis was occurring DNA during the time the [3H]deoxynucleosides these 3H-labeled nucleosides would be inserted the bulk of the duplex DNA. in the cellular were present, into segments of the chromosomal DNA not undergoing semiconservative DNA replication at that time. Since the size of DNA repair patches would be small compared to the sheared DNA fragment size (8), the DNA duplexes containing regions of DNA repair synthesis would still band at light-light duplex DNA density in neutral CsCI gradients. The light-light duplex DNA from control and ara-A-pulsed cells from an experiment as described above was purified through 2 cycles of neutral CsCI gradients from control and ara-A-pulsed cells, respectively (Chart 2). In the initial neutral CsCI gradients (Chart 2, A and ß),the 3H cpm were of 0.1 -ml samples taken from each fraction. (Each fraction had been diluted previously with 0.5 ml of buffer for determi nation of A26o's). Total cpm in each fraction would have been approximately 7.5 times the values shown in Chart 2, A and B. The purified light-light duplex DNA was dissociated, and the separated strands were banded in alkaline CsCI gradients (Chart 2, £and F; control and test, respectively) in order to determine whether any 3H label had been incorporated into the nonreplicating light-density strands due to short patches of DNA repair synthesis. In the DNA of control cells (Chart 2£), some presumptive DNA repair synthesis was present (3H label at light-strand density). This presumptive DNA repair synthesis was also present in DNA from ara-A-pulsed cells but to a lesser extent (Chart 2F). The presence of repair-type synthesis in control Crow cells is at variance with the lack of detectable repair-type synthesis in untreated GK cells, another cell line of human origin (28). It is also in disagreement with Gautschi ef al. (8) who found no evidence for DNA repair synthesis in untreated mammalian cells using a very similar methodology. CANCER RESEARCH Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. VOL. 41 Aberrant DNA Replication after DNA Synthesis Inhibition 200.000- 0.0 1.0 0.4 tOOMO 17 0.2 0 sis. Hence, the result of experiments to demonstrate double replication following a pulse of ara-A such as is presented in O.I «D eg parental DNA (7, 13). These patches of presumptive DNA repair synthesis might thus be small regions of semiconservative replication involved in this delayed ligation. Moreover, there is less of this presumptive repair synthesis in ara-Apulsed cells (Chart 2F) than control cells (Chart 2£).This would be consistent with it being part of the process of semiconservative replication. However, whatever the true nature of this apparent repair synthesis, that less is present after a pulse of ara-A indicates that ara-A does not induce DNA repair synthe 0.4 1.7 0.2 •1.0 0 O.S too 1.0 1.8 I). 4 50 0.3 1.7 0.2 0.1 10 20 20 Fraction Number Chart 2. Testing for DNA repair synthesis after a pulse of ara-A. BrdUrd (10 /IM), FdUrd (1 /IM), and dCyd (10 JIM) were added to cultures of logarithmically growing Crow cells. One hr later, ara-A at 0.5 rriM was added to one-half of the cells (test cells). One hr later, these cells were transferred to fresh medium containing BrdUrd, FdUrd, [3H]dCyd, and [3H]dGuo. The other half of the cells (control cells) was placed directly into this medium after the 1 hr in the presence of BrdUrd. Both control and test cultures were incubated for 4 hr in the Hlabeling medium before the cells were collected. DNA was extracted and purified through 2 cycles of neutral CsCI gradients, and the strands of purified light-light duplex DNA were dissociated and separated in alkaline CsCI gradients. A and B. initial CsCI gradients; C and D, second cycle of CsCI gradients; £and F. alkaline CsCI gradients. A, C, and E, control cell DNA; B, D, and F, test cell DNA. Fractions pooled from the neutral CsCI gradients are indicated by the bars. Aliquots of 0.1 ml from fractions of the initial neutral CsCI gradients were assayed for 3H content (fractions diluted previously with 0.5 ml of buffer for A2K>deter minations). The total cpm in each fraction of the initial neutral CsCI gradients would have been approximately 7.5 times the cpm shown in A and B. However, the methodology used in this study should be more sensitive than that used in the 2 studies quoted above, since in this case [3H]dCyd plus [3H]dGuo was used to label DNA repair patches, neither of which nucleoside had to compete with unlabeled BrdUrd for sites of incorporation as would be the case with the [3H]BrdUrd or [3H]dThd used in the previous studies. The peak of 3H label which is present in the light strands of untreated Crow cells (Chart 2E) coincides with the absorbance peak and shows no significant displacement to a higher density. Since the DNA was sheared to approximately 6000 base pair fragments before any CsCI gradient fractionation, the patches of presumptive DNA repair synthesis could have been at most a few hundred nucleotides long and prob ably much shorter than that. An alternative explanation to its being DNA repair synthesis is that it might be regions of delayed semiconservative synthesis. There is evidence in eukaryotic cells that there is a delay of some hours in the ligation of semiconservatively replicated DNA to the molecular weight of Chart 1 cannot be explained on the basis of it being an artifact due to DNA repair. Whether ara-C or ara-A is used in these experiments, DNA replication is being inhibited directly at the level of DNA polymerase (6). To test whether aberrant reinitiation and double replication is simply a general consequence of DNA replication being temporarily halted, DNA replication was inhibited by an indirect method. Cycloheximide is a specific inhibitor of protein synthesis acting at the level of the ribosome (17), inhibiting initiation and translocation of the ribosomes along the mRNA (15). Addition of the cycloheximide to mammalian cells results in a rapid decrease in the rate of protein synthesis with a parallel and quantitatively similar decrease in the rate of DNA synthesis (20). Although the precise mechanism has not been definitely demonstrated, it is likely that the parallel decrease in DNA synthesis is due to lack of packaging proteins (both histones and nonhistone chromosomal proteins) for the newly replicated DNA (20). Hence, cycloheximide was considered to be a good example of a compound producing an inhibition of DNA syn thesis by an indirect mechanism. In an experiment similar to that described above to demon strate reinitiation and double replication after a pulse of ara-A, Crow cells were labeled for 1 hr with [3H]dCyd, and the label was chased for 1.5 hr with unlabeled dCyd. After the chase period, cycloheximide at 100 /xg/ml was added to one-half of the cells. One hr later, these cells were transferred to fresh BrdUrd-containing medium. Control cells were transferred di rectly to the BrdUrd-containing medium after the 1.5-hr chase. A 1-hr treatment with cycloheximide at 100 jug/ml reduced the incorporation by Crow cells of [3H]dThd into cold acid-insoluble material by greater than 90% (not illustrated). The light-heavy DNA fraction from control and cycloheximide-pulsed cells was purified through 2 cycles of neutral CsCI gradients, and the strands of the isolated fractions of light-heavy duplex DNA were dissociated and analyzed on alkaline CsCI gradients (Chart 3). As was the case with ara-C and ara-A, a high proportion (approximately one-half) of 3H label from the lightheavy DNA fraction banded as a peak at the density of the template strands (Chart 3F), with this peak being absent from the DNA of control cells (Chart 3£). Hence, this aberrant reinitiation of DNA replication also occurs following inhibition of synthesis by an indirect mechanism, namely, inhibition of protein synthesis. Again, to ensure that 3H label at template strand density was not present due to contaminating light-light DNA in the lightheavy fraction, the 10 fractions from the second neutral CsCI gradients (Chart 3, C and O) on the light-density side of the samples pooled from the light-heavy peak were assayed for 3H content. Again, the light-heavy fractions were well separated JUNE 1981 Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. 2487 D. M. Woodcock and I. A. Cooper alkaline CsCI gradients (Chart 4£,light-light DNA from control cells; Chart 4P, light-light DNA from cycloheximide-pulsed cells). As was the case with the experiment illustrated in Chart 2, 3H label was present in the light-density strands from the light-light DNA of control cells, indicating insertion of short regions of 3H-labeled nucleotides into DNA segments not in 10.000 5.000 1 10 20 Fraction 1 10 Number Chart 3. Testing for reinitiation and double replication following a pulse of cycloheximide. Logarithmically growing Crow cells were labeled with a 1-hr pulse of [3H]dCyd followed by a 1.5-hr chase with a 2000-fold excess of unlabeled dCyd. After the chase period, cycloheximide (100 fig/ml) was added to one-half of the cells (test cells). One hr later, the test cells were transferred to fresh medium containing 10 JUMBrdUrd. 1 IÕMFdUrd, and 10 UM dCyd. The other half of the cells (control cells) was transferred directly into this medium after the 1.5hr chase. Control cells were incubated for 3 hr in the density-labeling medium and test cells for 4 hr to allow approximately equal proportions of DMA to be replicated in the presence of BrdUrd in control and test cultures. Colcemid (1 /jg/ml) was present in all media from the time of the 3H-labeling period onward. DNA was extracted and purified as described in "Materials and Methods." The light-heavy DNA fraction was purified through 2 cycles of neutral CsCI gradients, and the strands of the purified light-heavy duplex DNA were dissociated and separated in alkaline CsCI gradient. A and B, initial neutral CsCI gradients; C and D, second cycle of neutral CsCI gradients; E and F, alkaline CsCI gradients. A, C, and £,control cell DNA; B, D. and F, test cell DNA. Fractions pooled from the neutral CsCI gradients are indicated by the oars The 10 fractions on the lightdensity side of the fractions pooled from the light-heavy peak from the second neutral CsCI gradients were assayed for 3H content to check for contamination of the light-heavy fraction with light-light DNA. Total cpm recovered from the alkaline CsCI gradients were 862 in £and 512 in F. from the remaining light-light material (Chart 3, C and D). As with ara-A, we tested whether a pulse of cycloheximide induced DNA repair-type synthesis in Crow cells which might have accounted for results such as those in Chart 3. Using the same experimental method used with ara-A (Chart 2), Crow cells were given a 1-hr pulse with 100 /¿gcycloheximide per ml, and the light-light DNA fraction was purified through 2 cycles of neutral CsCI gradients. Aliquots of 0.1 ml were taken from the first neutral CsCI gradients and assayed for 3H cpm. The first neutral CsCI gradients are shown in Chart 4 (A, control cell DNA; 8, cycloheximide-pulsed cell DNA). The total 3H cpm in each of the fractions would have been approximately 7.5 times the cpm shown on Chart 4, A and ß.The strands of the purified light-light DNA fraction were separated and banded in 2488 volved in semiconservative synthesis during the time of the experiment. As discussed above, this might be true DNA repair synthesis, or it might be short regions of delayed ligation. However, in cells given a pulse of cycloheximide, there is much less of this presumptive DNA repair synthesis (Chart 4P) than in the control cell DNA (Chart 4E). Again, the lower amount of this presumptive repair synthesis following a period of reduced DNA semiconservative synthesis would be consistent with its being related in some way to semiconservative synthesis (i.e.. <o M 0.50 0.25 10 20 10 20 Fraction Number Chart 4. Testing for DNA repair synthesis after a pulse of cycloheximide. BrdUrd (10 JIM), FdUrd (1 UM), and dCyd (10 ;uM) were added to cultures of logarithmically growing Crow cells. One hr later, cycloheximide (100 fig/ml) was added to one-half of the cells (test cells). One hr later, the test cells were transferred to fresh medium containing BrdUrd, FdUrd. [3H]dCyd, and [3H]dGuo. The other half of the cells (control cells) was placed directly into this medium after the 1 hr in the presence of BrdUrd. Both control and test cultures were incubated for 4 hr in the 3H-labeling medium before the cells were collected. DNA was extracted as described in "Materials and Methods." The light-light DNA fraction was purified through 2 cycles of neutral CsCI gradients, and the strands of the purified light-light duplex DNA were dissociated and separated in alkaline CsCI gradients. A and B, initial CsCI gradients; C and D, second cycle of CsCI gradients; £and F, alkaline CsCI gradients. A, C, and £.control cell DNA; B, D, and F, test cell DNA. Fractions pooled from the neutral CsCI gradients are indicated by the bars. Aliquots of 0.1 ml from the fractions of the initial neutral CsCI gradients fractions were assayed for 3H content (fractions diluted previously with 0.5 ml of buffer for A26odeterminations). The total cpm in each fraction of the initial neutral CsCI gradients would have been approximately 7.5 times the cpm shown in A and B. CANCER RESEARCH Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. VOL. 41 Aberrant DNA Replication after DNA Synthesis Inhibition delayed ligation). In any case, the reduced extent of this presumptive repair synthesis following a pulse of cycloheximide argues against repair synthesis being the cause of the in creased amount of 3H label present in template strands of DMA duplexes replicated subsequent to a pulse of cycloheximide in experiments such as those illustrated in Chart 3. As was done previously with ara-C-pulsed cells (26), we have determined the amount of double replication in DNA extracted from whole cells and isolated nuclei following a pulse of cycloheximide. There was quantitatively the same distribu tion of 3H label between heavy and light strands of light-heavy DNA extracted from whole cells or isolated nuclei from cycloheximide-pulsed cells (not illustrated). Hence, multiple rounds of mitochondrial DNA synthesis cannot be a trivial explanation of these results. Also cells were routinely monitored for Mycoplasma (21), obviating Mycoplasma tion. replication as an explana DISCUSSION We have previously presented data which indicates that, following a pulse of ara-C, the mechanism controlling the pattern of replication of the chromosomal DNA is disrupted, resulting in some DNA segments being replicated more than once in a single S phase (26, 28). This can be visualized as some localized regions of the chromosomes undergoing endoreduplication in a single S phase. This paper presents evi dence that the induction of double replication is not just a peculiarity of the ara-C molecule itself but can also be induced by transiently inhibiting DNA synthesis with ara-A and also with cycloheximide. In the case of ara-C, we have specifically ob tained evidence that this effect is not due to DNA recombinational and exchange events or to multiple rounds of DNA replication in mitochondria or Mycoplasma (26, 28). In the cases of ara-A and cycloheximide, we have tested whether this double replication phenomenon might simply be an artifact due to DNA repair synthesis. In both cases, we have found no evidence for any stimulation of DNA repair-type synthesis fol lowing a pulse of either inhibitor. The situation in relation to repair synthesis is complicated by the increased sensitivity of the method used in the present study to detect DNA repairtype synthesis. This has allowed the detection of what appears to be DNA repair synthesis in previously unmanipulated control cells (Charts 2 and 4). It is not unreasonable that there should be a constant background of DNA repair in normal untreated cells from spontaneous events at 37°.However, this presump tive repair synthesis is reduced after DNA synthesis has been transiently blocked by ara-A (Chart 2) or cycloheximide (Chart 4). This apparent repair synthesis is also reduced after a pulse of ara-C. This effect has also been observed in 2 other cell lines (K562 and CHO cells) as well as Crow cells.3 Since the enzymes involved in DNA repair synthesis are in general less sensitive to inhibitors than are the enzymes involved in semiconservative DNA replication (6), the observed reduction in the amount of apparent repair synthesis after DNA replication inhibition is more consistent with its being related to semiconservative replication than to repair. The explanation for this apparent repair synthesis that we favor is that it is short regions of normal semiconservative synthesis which is involved in de layed ligation of replicated DNA to very high molecular weight (7, 13). However, whatever the true explanation of this phe nomenon, this apparent repair synthesis decreases after DNA JUNE synthesis inhibition. Hence, it cannot account for the increase in 3H label from purified light-heavy DNA present at template strand density in alkaline CsCI gradients in experiments testing for double replication of chromosomal DNA segments (Charts 1 and 3). As to other alternative explanations of the results shown in Charts 1 and 3, cells were routinely screened for Mycoplasma and were always found to be negative. As we have shown for ara-C, the double replication after a pulse of cycloheximide cannot be due to multiple rounds of mitochondrial DNA synthe sis since similar results were obtained with DNA from whole cells and from isolated nuclei. With the Crow cells used in these experiments having a doubling time in logarithmic growth of approximately 24 hr, the time of the experiment was too short for cells in one S phase at the beginning of the experiment (during the [3H]dCyd pulse) passing through G2, M, and Gt during the time of the experiment. As an added precaution against this possibility, Colcemid at 1 jug/ml was included in all media from the time of the end of the [3H]dCyd pulse onwards. It was not specifically tested experimentally as was done with ara-C inhibition whether double replication after a pulse of araA or cycloheximide could have been due to DNA recombinational and exchange events. As has been fully argued previ ously (26), the possibility that recombination or exchange could explain results such as those in Charts 1 and 3 is extremely remote. Hence, we conclude that the most feasible explanation for the results presented in this paper is that double replication of some chromosomal DNA segments is not just the result of transiently inhibiting DNA synthesis by ara-C but also results after similar levels of inhibition by ara-A and also by cyclohex imide. That this phenomenon occurs after a pulse of ara-A as well as ara-C is not surprising since the activated forms of both compounds (the triphosphates) act to inhibit DNA synthesis at the level of the polymerase (6). However, what the result with ara-A does show is that aberrant double replication after an ara-C pulse is most likely mediated via the inhibition of DNA synthesis and that the double replication is not the result of some specific property of the ara-C molecule which might not even have been related to any effect of 1-/S-D-arabinofuranosylcytosine triphosphate on DNA polymerase a (23). What is more interesting is that double replication also occurs after DNA synthesis is inhibited indirectly with cycloheximide due to a consequent deficiency in the proteins necessary for pack aging newly replicated DNA duplexes (20). We have suggested that the double replication of some chromosomal DNA seg ments following transient DNA synthesis inhibition by ara-C may be the cause of the chromosome aberrations (28) and hence of the cell death resulting from the pulse of ara-C (10, 11). Hence, an important test of this hypothesis is whether other compounds which induce this double replication phenom enon also induce chromosome aberrations and cell death, araA is a cytotoxic compound which has been shown to induce chromosome aberrations in S-phase cells (16). However, araA is such a similar compound to ara-C as not to constitute a particular stringent test of this hypothesis. However, the protein synthesis inhibitor, cycloheximide, which affects DNA synthesis only indirectly, constitutes a more appropriate test of this model. If double replication following transient DNA synthesis inhibition by ara-C is a major primary cause of chromosome aberrations and subsequent S-phase-specific cytotoxicity, then 1981 Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. 2489 D. M. Woodcock and I. A. Cooper cycloheximide should also cause chromosome aberrations and exhibit S-phase-specific cytotoxicity. It has already been shown that protein synthesis inhibitors are cytotoxic to cells and that this cytotoxicity is S-phase specific (5). Indeed, Bhuyan and Frazer (3) have shown that streptovitacin A (acetoxycycloheximide) is not only S-phase specific but will protect cells in other phases of the cell cycle from the toxicity of other S-phasespecific cytotoxics. S-phase cells were shown to be killed by either the protein synthesis inhibitor or an S-phase-specific agent such as ara-C while the protein synthesis inhibitor stopped other cells from entering S-phase (3, 5). Hence, one of our predictions of the consequences of double replication following a cycloheximide pulse is already well documented. However, for the second prediction, namely that this cycloheximide-induced double replication will cause chromosome aberrations, we were unable to find any report of cycloheximide inducing chromosome aberrations. On the contrary, there are reports that cycloheximide is not clastogenic (2, 24). However, in subsequent experiments with Crow cells, we have shown that a 1-hr pulse of cycloheximide at 100 /ig/ml (the conditions shown above to induce double replication) does in fact induce chromosome damage (25). That the clastogenicity of cyclohex imide was not detected previously was probably due to the prolonged delay of cells with aberrations in reaching mitosis (25). Hence, both of the predictions of our hypothesis as to the consequences of double replication induced by a pulse of cycloheximide have now been documented. Therefore, from present evidence, we suggest that a tran sient block to DNA replication, irrespective of the agent causing this effect, disrupts the mechanism controlling the pattern of replication of the chromosomal DNA, resulting in double repli cation of some chromosomal DNA segments. Further, we sug gest that this double replication induced in cells in S phase results in chromosome aberrations and subsequent loss in cell viability. We note that this hypothesis is entirely consistent with the data of Ramseier ef a/. (18) who have shown a similar decrease in cell viability of synchronized S-phase CHO cells when DNA synthesis was inhibited to similar extents by hydroxyurea, FdUrd, excess dThd, or cycloheximide. In all cases, almost total inhibition of DNA synthesis was necessary for large reductions in cell viability with these authors suggesting that the killing of S-phase cells was related to complete blocking of DNA replication forks. ACKNOWLEDGMENTS We would like to thank Jill Adams for able technical assistance and Barbara Caldecoat for typing the manuscript. REFERENCES 1. Benedict, W. F., and Jones, P. A. Mutagenic, clastogenic and oncogenic effects of 1-/S-o-arabinofuranosylcytosine. Mutât.Res., 65. 1-20. 1979. 2. Benedict, W. F., Rucker, N., and Karon, M. 1-/î-o-Arabinofuranosylcytosineinduced chromatid breakage: effect of inhibition of DNA synthesis. J. Nati. 2490 Cancer Inst., 54: 431 -433. 1975. 3. Bhuyan, B. K . and Frazer, T. J. Antagonism between DNA synthesis inhibitors and protein synthesis inhibitors in mammalian cell cultures. Cancer Res., 34. 778-782, 1974. 4. Bhuyan, B. K., Frazer, T. J., Grey, L. G., Kuentzel, S. L., and Neil, G. L. Cell kill kinetics of several S-phase-specific drugs. Cancer Res.. 33.888-894, 1973. 5. Bhuyan, B. K., Scheldt, L. G., and Frazer, T. J. Cell cycle specificity of antitumor agents. Cancer Res., 32. 398-407, 1972. 6. Cozzarelli, N. R. The mechanism of action of inhibitors of DNA synthesis. Annu. Rev. Biochem., 46.641-668, 1977. 7. Funderud, A., Andreassen, R., and Haugli, F. Size distribution and matura tion of newly replicated DNA through S and G2 phases of Physarum polycephalum. Cell, 15: 1519-1526, 1978. 8. Gautschi, J. R., Young, B. R., and Painter, R. B. Evidence for DNA repair replication in unirradiated mammalian cells—is it an artifact? Biochim. Biophys. Acta, 287. 324-328, 1972. 9. Hori, T., and Lark. K. G. Effect of puromycin on DNA replication in Chinese hamster cells. J. Mol. Biol., 77. 391-404, 1973. 10. Jones, P. A., Baker, M. S., and Benedict, W. F. The effect of 1-/3-Darabinofuranosylcytosine on cell viability, DNA synthesis, and chromatid breakage in synchronized hamster fibrosarcoma cells. Cancer Res., 36. 3789-3797. 1976. 11. Karon, M., Benedict, W. F.. and Rucker, N. Mechanism of 1-/3-o-arabinofuranosylcytosine-induced lethality. Cancer Res., 32. 2612-2615, 1972. 12. Kihlman, B. A. Deoxyribonucleotide synthesis and chromosome breakage. Chromosomes Today, 1: 108-117. 1964. 13. Kowalski, J., and Cheevers, W. P. Synthesis of high molecular weight DNA strands during S-phase. J. Mol. Biol., 104: 603-615, 1976. 14. Lerner, R. A., Meinke, W., and Goldstein, D. A. Membrane-associated DNA in the cytoplasm of diploid human lymphocytes. Proc. Nati. Acad. Sei. U. S. A., 68. 1212-1216, 1968. 15. Lin, S. Y., Mosteller, R. D., and Hardesty. B. The mechanism of sodium fluoride and cycloheximide inhibition of haemoglobin biosynthesis in cell free reticulocyte system. J. Mol. Biol., 21: 51-69. 1966. 16. Nichols, W. W. in vitro chromosome breakage induced by arabinosyladenine in human leukocytes. Cancer Res., 24. 1502-1505, 1964. 17. Poche, H., Zakrzewski, S., and Nierhaus, K. H. Resistance against cyclo heximide in cell lines from Chinese hamster and human cells is conferred by the large subunit of cytoplasmic ribosomes. Mol. Gen. Genet., 775. 181185, 1979. 18. Ramseier. H. P., Burkhalter, M.. and Gautschi, J. R. Survival of CHO cells that replicated DNA in the presence of hydroxyurea. Exp. Cell Res.. 705. 445-453, 1977. 19. Ross, W. E.. Glaubiger, D., and Kohn. K. W. Qualitative and quantitative aspects of intercalator-induced DNA strand breaks. Biochim. Biophys. Acta, 562. 41-50, 1979. 20. Stimac, E., Housman, D., and Huberman, J. A. Effects of inhibition of protein synthesis in DNA replication in cultured mammalian cells. J. Mol. Biol., 115: 485-511, 1977. 21. Tomkins, G. A.. McGregor, A., Pye, D., and Atkinson, M. Rapid detection and isolation of mycoplasmas from cell cultures. Aust. J. Exp. Biol. Med. Sci.. 53. 257-263, 1975. 22. Weintraub, H , Campbell, G. L.. and Holtzer. H. Identification of a develop mental program using bromodeoxyuridine. J. Mol. Biol., 70. 337-350,1972. 23. Wist, E. The role of DNA polymerases n, ßand y in nuclear DNA synthesis. Biochim. Biophys. Acta, 562: 62-69, 1979. 24. Wolff. S. The splitting of human chromosomes into chromatids in the absence of either DNA or RNA synthesis. Mutât.Res., 8: 207-214, 1969. 25. Woodcock, D. M., Adams, J. K., and Cooper, I. A. An explanation for the Sphase specificity of the cytotoxicity of protein synthesis inhibitors. Eur. J. Cancer, 17: 173-177, 1981. 26. Woodcock, D.M., and Cooper. I. A. Aberrant double replication of segments of chromosomal DNA following DNA synthesis inhibition by cytosine arabinoside. Exp. Cell Res., Õ23: 157-166, 1979. 27. Woodcock, D. M.. and Fox, R. F. The molecular mechanism of cell killing by DNA synthesis inhibition. Proc. Aust. Biochem. Soc., 10: 68, 1977. 28. Woodcock, D. M., Fox, R. D., and Cooper, I. A. Evidence for a new mechanism of cytotoxicity of 1-/?-o-arabinofuranosylcytosine. Cancer Res., 39. 1416-1424, 1979. 29. Zeros, N. Deoxyriboside control and synchronization of mitosis. Nature (Lond.), Õ94. 682-683, 1962. CANCER RESEARCH Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research. VOL. 41 Evidence for Double Replication of Chromosomal DNA Segments as a General Consequence of DNA Replication Inhibition David M. Woodcock and Ian A. Cooper Cancer Res 1981;41:2483-2490. Updated version E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/41/6/2483 Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected]. To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]. Downloaded from cancerres.aacrjournals.org on June 16, 2017. © 1981 American Association for Cancer Research.
© Copyright 2026 Paperzz