A Method for the Purification of DNA/Protein Complexes Applied to DNA Topoisomerase II Cleavage Sites R. J. Anderson, Molecular


C. Delgado,

Cell Pathology



D. Fisher,

J. M. Cunningham,

Royal Free Hospital


United Kingdom

20, 1990

The object of this study was to devise a purification method for DNAltopoisomerase II complexes, with which to examine the enzyme’s cleavage site specificity in cellular differentiation. Retinoic acid-induced differentiation involves topoisomerase II-mediated transient changes in DNA supercoiling, but it is not known whether this occurs at specific sites in the genome. Topoisomerase II forms a covalent DNA enzyme complex as it acts, which can be recovered by the sodium dodecyl sulfate (SDS)/KCl precipitation method, but this method fails to recover significantly more DNA from cells induced to differentiate. This may in part reflect the low numbers of retinoic acid-induced proteinlinked breaks in DNA and also the method’s relative inefficiency for DNA with few attached topoisomerase molecules. This suggested that an additional purification method would be required to enrich sufficiently for cleavage site DNA to address the issue of site specificity. The principle of our method is to couple poly(ethylene glycol) (PEG) to topoisomerase while it is covalently attached to DNA and then to use phase partitioning in an aqueous two-phase system of PEG and phosphate to separate free DNA from DNA bound to PEG-modified topoisomerases (which have high affinities for the phosphate-rich and PEG-rich phases, respectively). The method can be used in conjunction with DNase protection and, unlike the SDS/KC1 method, can fractionate short fragments of DNA to which single protein molecules are attached. Using the SDS/KC1 precipitation and new method in series, we have recovered protein-linked DNA from HL60 cells induced to differentiate to the granulocyte lineage (by retinoic acid) or to the monocyte/macrophage lineage (by phorbol myristate acetate) and have demonstrated that specific sequences become protein linked, probably to topoisomerase II, during induced differentiation. Q 1991 Academic Press,

and G. E. Francis

School of Medicine,


Two distinct types of change in chromatin structure precede overt hemopoietic cell differentiation. Studies 0003.2697/91 $3.00 Copyright Q 1991 by Academic Press, All rights of reproduction in any form

using nucleoid sedimentation (1,2), alkaline filter elution (2), and the fluorescent alkaline DNA unwinding (FADU)’ technique (3) demonstrate that chromatin structural changes which are lineage specific (i.e., changes dependent on the cell type being induced) occur less than 1 h after exposure to both physiological and pharmacological differentiation inducers. These include both increased and transiently decreased DNA supercoiling (l), ligation of preexisting DNA breaks, and the opening and closure of new DNA breaks (l-3). DNA topoisomerase II has been implicated in some of these changes (2) and ADP-ribosyltransferase in others (1,3), and both enzymes have themselves been implicated in the differentiation process (2,4). DNA topoisomerase II mediates the transient relaxation of DNA supercoiling and the opening and closure of new DNA breaks occurring during induction of granulocytic and monocytic differentiation (2). Since we have shown that inhibitors of topoisomerase II antagonize the differentiation-inducing action of the retinoids, it appears that this change is necessary for granulocytic differentiation (2). Similar inhibitor studies indicate a possible involvement in mono&c differentiation (5). Chromatin structural changes could have a variety of permissive/regulatory roles in differentiation and in order to identify what these might be, we need to determine their distribution in the genome. Since estimates of the number of topoisomerase-mediated DNA breaks occurring in differentiation are at the borderline of detection by nucleoid sedimentation, alkaline filter elution, and the FADU technique (1-3) (the former detects circa 5 rad equivalents in the presence of DNA repair inhibitors), these chromatin structural changes may

’ Abbreviations used: FADU, fluorescent alkaline DNA unwinding SDS, sodium dodecyl sulfate; PEG, poly(ethylene glycol); MPEG, monomethoxy-PEG; TMPEG, tresylated MPEG; BSA, bovine serum albumin; PMSF, phenylmethylsulfonyl fluoride; PMA, phorbol 12myristate 13-acetate; EGTA, ethylene glycol bis(@-aminoethyl ether) N,N,N’,N’-tetraacetic acid; DTT, dithiothreitol; DMSO, dimethyl sulfoxide; RA, retinoic acid. 101

Inc. reserved.



well be occurring at limited, specific sites in the genome. Topoisomerases, by binding covalently to DNA as they act (6,7), potentially provide an opportunity to examine site specificity. The DNA topoisomerase strand cleavage reaction can be arrested by SDS leaving the enzyme coupled to DNA. There are, however, limitations to the most widely used method for the purification of DNA/ topoisomerase II complexes. This method, the SDS/ KC1 precipitation method of Trask and co-workers (8), recovers DNA bound covalently (or in a detergent-resistant fashion) to protein, in practice largely topoisomerase-bound DNA. The precipitation of DNA attached to noncovalently bound proteins (e.g., histones or DNA polymerases) is minimal (8). Several observations suggest that this method is et%cient only where multiple protein molecules are attached. Experiments with purified topoisomerase I (8) imply that lo-20 molecules are required per DNA fragment for efficient precipitation to occur. This is probably the basis of the observed enrichment in SDS/KC1 precipitates of replication fork DNA, on which topoisomerase II is highly active (possibly solving the topological problem posed by the nascent daughter strands (9)). The exact nature of the latter complexes is somewhat puzzling since arrest of the topoisomerase II reaction by SDS has been assumed to cleave both strands, leaving one topoisomerase molecule attached to each 5’ end (7). However, examination of the DNA fragment size of SDS/KC1 precipitates shows almost exclusively large fragments (>20 kb), suggesting that recovered replication forks are not highly fragmented. Thus, since the influence on precipitation efficiency seems unlikely to be dramatically different for topoisomerases I and II, either arrest of the enzyme is at the single strand break stage (allowing several molecules to coexist on the same DNA strand) or cleavage is at the double strand stage giving topoisomerase II monomers at each end, complemented by internally placed topoisomerase I molecules (which only produces single strand nicks). When topoisomerase II cleavage has been examined in mammalian cells in vitro and in whole cells, using known genes as models, the sites are remarkably specific but are mainly solitary and widely spaced (usually in flanking sequences), or in very closely spaced clusters of three to five cleavage sites (10,ll). In terms of the recovery strategy, these clusters are unlikely to represent sites where multiple topoisomerase II molecules bind simultaneously since only a few base pairs intervene between each site of the cluster. Most of these analyses have been performed using either epipodophyllotoxinor amsidine-induced stabilization of topoisomerase II DNA complexes or have examined the cleavage reaction of purified enzyme. Since drugs are known to influence site specificity of cleavage (12) and purified enzyme is acting in a highly artificial milieu, it is not clear whether these sites represent the physiologi-

ET AL. cal sites of topoisomerase action. There are, however, recurring themes in the location of such cleavage sites, suggesting they may have a regulatory role, in that they are often in DNase-hypersensitive regions (ll), near enhancers in some genes (cf. immunoglobulin genes (13)) and precisely located around the binding footprints of known transcriptional modulators (cf. topoisomerase II sites in c-fos (10,14)). We therefore suspected that the topoisomerase II cleavage reactions observed in, and apparently necessary for, neutrophil granulocyte differentiation (2) might similarly be isolated or in small clusters. A method capable of fractionating DNA attached to a single topoisomerase molecule is thus desirable. In practice, as demonstrated below, the Trask technique does not recover significantly more DNA from cells induced to differentiate than from undifferentiated controls and is unlikely therefore, on its own, to provide sufficient enrichment of DNA-protein complexes from differentiation-associated breakage reunion reactions to address the issue of site specificity. Given the estimated low number of topoisomerase II mediated breakage/ reunion reactions operative in differentiation, it also seemed unlikely that a single purification step would suffice. The objective of this study was to devise a new method for the fractionation of topoisomerase-linked DNA, specifically one capable of separating DNA to which a single topoisomerase II molecule is attached. Immunoaffinity methods are not readily applicable because of the restricted availability of anti-topoisomerase II antibody. We recently demonstrated (15,16) that poly(ethylene glycol) (PEG) may be covalently linked to proteins, under mild conditions, using tresyl monomethoxy-PEG (TMPEG). Such attachment allows affinity partitioning of modified proteins (or even whole cells) in PEG containing aqueous biphasic polymer systems. Here we show that TMPEG may be used to couple MPEG to topoisomerases covalently attached to DNA and that a phase-partitioning system can be devised where only DNA attached to protein is recovered in the PEG phase. The method can successfully fractionate DNA fragments to which single protein molecules are attached, allowing DNase protection to be employed to restrict recovered DNA to that at the protein binding sites. With the new method, after preliminary purification by SDS/KC1 precipitation, we were able to detect increased recovery of protein-linked DNA from cells induced to differentiate by retinoic acid. The rationale for the use of the SDS/KC1 method as a preliminary step, despite reservations about its efficiency for the recovery of DNA to which few molecules of topoisomerase are attached, is as follows. Free DNA remains in the supernatant whereas protein-attached DNA precipitates with varying degree of efficiency depending on the num-




ber of attached protein molecules. Thus even if a protein-attached fragment precipitates very inefficiently (because only few or single protein molecules are attached), its relative concentration in the precipitate with respect to free DNA must be higher than that in the start preparation, because free DNA remains in the supernatant. In addition, since replication forks are randomly distributed, solitary sites will be coincident with multiple sites in a proportion of fragments (unless attachment never occurs in S-phase). Thus the SDS/ KC1 procedure, applied as a first step, should significantly reduce the content of non-protein-bound DNA. With the two-step procedure, the recovered DNA from differentiating cells was sufficiently enriched in specific sequences to show significant differences in hybridization between topoisomerase-associated and topoisomerase-free DNA from differentiating and undifferentiated cells. Such a hybridization pattern is consistent with cleavage sites being at specific sites in the genome. MATERIALS



Chemicals and reagents. All chemicals were of analytical grade and were purchased from Sigma Chemicals (UK) and all tissue culture reagents were purchased from GIBCO Ltd. (UK), except where stated. Monomethoxy-poly(ethylene glycol) was purchased from Union Carbide Chemicals (U.S.A.), and poly(ethylene glycol) 6000 and molecular sieve A3 were obtained from BDH Chemicals Ltd. (UK). Nylon filters were obtained from DuPont Research Products (UK), and the filter paper used was from Whatman Ltd. (UK). VP16-213 (vepesid) was purchased from Bristol Meyers Ltd. (UK), and 0.22- and 0.45~pm filters were obtained from Gelman Science (U.S.A.). The plasmid pBR322 and EcoRI were obtained from Boehringer-Mannheim. Selecting the phase system. Initially separation was attempted by using a PEG/dextran system (17). However, on addition of phenol/chloroform to the aqueous phases (to extract DNA after proteinase K digestion) dextran precipitates. This can be overcome (18) by transferring the DNA from the dextran-rich phase to a fresh PEG-rich phase (which has been optimized for DNA extraction by the addition of salts) and then removing PEG by extraction with chloroform (18). However, PEG phases are not exclusively PEG, thus some dextran still contaminates the DNA. An alternative is to use ammonium sulfate extraction (19). The requirement for a multiple extraction procedure led us to investigate a less commonly exploited system using PEG and phosphate to form the immiscible phases. We constructed a PEG: phosphate system which separates proteins and DNA (20). To reduce the affinity of the PEG phase for free DNA, we selected a comparatively high polymer molecular weight (PEG 6000). The final phase



composition selected and used throughout this paper was 10% (w/w) PEG 6000 and 14% (w/w) inorganic phosphate. To increase the specificity of the system, we elected to include an affinity ligand by coupling MPEG to proteins and hence dramatically increase their affinity for the PEG phase. Coupling PEG to topoisomerase. Coupling of PEG to proteins is usually achieved by appropriate derivatization of the hydroxyl groups of PEG with a suitable reagent that can be substituted by nucleophilic groups in the protein (the NH, groups of lysine side chains are particularly accessible sites and either few or many sites can be modified) (21). In order to devise a PEG coupling method suitable for delicate proteins like cytokines, we recently developed a new coupling method using tresyl chloride (2,2,2-trifluoroethanesulfonyl chloride) to activate PEG, based on Nillson and Mosbach’s method for coupling enzymes and affinity ligands to solid phases bearing hydroxyl groups (22). By activating monomethoxy-PEG (MPEG) at its single free derivatizable hydroxyl group, the resultant MPEG tresylate gives a reactive species capable of linking to only one protein molecule. We have demonstrated that with tresylated MPEG the coupling of MPEG to both antibodies (16) and albumin (15) can occur under mild conditions (pH 7.5 phosphate buffer, at room temperature). Activation of MPEG with tresyl chloride. This was performed as described elsewhere (15) using dried MPEG (1M, 5000; 18 g, 3.5 mmol) dissolved in dry dichloromethane (45 ml) to which 1.125 ml (14 mmol) of pyridine and 1 ml (9 mmol) of ice-cold tresyl chloride were added dropwise. The reaction was continued at room temperature with constant stirring for 1.5 h before the dichloromethane was removed by evaporation under reduced pressure, leaving the tresylated MPEG. Pyridine was removed as previously described (15) and the TMPEG was stored desiccated at 4°C prior to use (within 3-4 months). PEG modification of DNA - protein complexes. TMPEG, 400 mg/ml in 0.05 M Na phosphate, 0.125 M NaCl buffer, pH 7.5, was mixed with cell lysate or DNAprotein complexes at a volume ratio of 1:l on a rotating mixer for 2 h at room temperature. Since we do not know the number of lysine molecules due to topoisomerase II and the other proteins present, we chose to add TMPEG in excess (i.e., sufficient to modify all protein present). Experiments with albumin indicate maximum partitioning to the PEG phase with a TMPEG:lysine molar ratio of 8:l and significantly increased partitioning is achieved at much lower values. The ratios of DNA-protein complexes to TMPEG used in these experiments therefore represent a gross excess of TMPEG (being 0.8 mg of TMPEG/pg of protein), but this may have additional advantageous benefits because of the effects of PEG concentration on DNA conforma-



tion (23,24). The latter is likely to facilitate dissociation of noncovalently bound proteins, although many will have been dissociated by prior exposure to 1% SDS in the cell lysate preparation or SDS/KC1 procedure (8). TMPEG-treated material may contain unreacted TMPEG capable of coupling to enzymes used in subsequent processing of the DNA. Lysine or albumin can be used to prevent further undesired reaction of free TMPEG after the coupling step (15,16). In practice we found that proteinase K is robust and, provided overnight incubations at 37°C are used, is little affected by the addition of TMPEG-treated material. However, where DNase was to be used after TMPEG exposure, 1 M free base lysine dissolved in coupling buffer (1 vol) was added to the reaction mixture (6 vol) and the mixture incubated at room temperature for a further hour. Phase partitioning. The phase system constructed using 10% (w/w) poly(ethylene glycol) 6000, 14% phosphate (ratio of 16.86 g KH,PO, and 40.20 g K,HPO, * 3H,O), and 76% distilled deionized sterile water. It was mixed and allowed to settle at 25°C then the two phases were separated, and the phosphate-rich lower phase was filtered through a 0.22-pm filter and the PEG-rich upper phase through a 0.45-pm filter. Both were then aliquoted and stored at -20°C. This was done because it is difficult to sample aliquots of the mixed-phase system without taking varied proportions of the two phases. Although these are strictly phosphate-rich and PEG-rich phases (since they each contain a minor proportion of the other phase constituent), they will, for brevity, be referred to as the PEG and phosphate phases hereafter. The phase systems were reconstructed using various volume ratios of PEG:phosphate phases (indicated with each experiment) with no more than 15% the total volume of the phase system comprising the TMPEGtreated DNA/protein extract. For experiments fractionating small DNA fragments the ratio of PEG:phosphate phases were decreased to 250:750 ~1 and multiple rounds of phosphate extraction were used. The phases were mixed by briefly vortexing and allowed to settle at 25°C for 10 min. For multiple phosphate extractions, the PEG-phase was then transferred to a fresh phosphate phase and the procedure repeated (the number of rounds of extraction is indicated with each experiment). Similarly, where DNA depleted of topoisomerase II attachment sites was required, phosphate phases were repeatedly extracted with fresh PEG phases. Recovery of DNA from the phases and quantitation. We evaluated several means of estimating the amount of DNA in the phases. Many commonly used methods are not applicable to the DNA while in the phases themselves. Chemiluminescence proved problematic for the use of [3H]DNA and scintillation count-

ET AL. ing. The influence of PEG on fluorescence prevented the use of Hoechst 33258 to estimate DNA fluorometritally. We therefore recovered DNA prior to estimation using overnight proteinase K digestion, phenol/chloroform extraction. This procedure effectively removes PEG but not phosphate and the latter was removed using Sephadex G-50 columns (110 ~1 of phase extract per 1.2-ml column). DNA was recovered from the eluate by precipitation in 2 vol of absolute alcohol at -70°C for 1 h followed by centrifugation at ll,OOOg for 10 min. This DNA was either estimated by agarose gel electrophoresis and scanning densitometry (where we wished to examine the size distribution of the loaded and recovered DNA) or, for rapid quantitation, samples were spotted onto an agarose slab, photographed on a transilluminator, and compared to a set of DNA standards. Sonication of DNA and DNA-protein complexes. Aliquots (500 ~1) of TMPEG-treated HL60 cell DNAprotein precipitates, whole cell extracts, or salmon sperm DNA (10 pg/ml in 10 mM Tris . HCl, 1 mM EDTA, pH 7.4) were sonicated for 5-60 s at 12 pm peak to peak, 20 kHz, 150 W in an MSE bench sonicator, on ice, taking care to avoid frothing. DNase treatment. For the preparation of short fragment DNA-protein complexes from which probes for hybridization were prepared (see below), protein-associated DNA prepared by the method of Trask et al., equivalent to 5 pg DNA, was exposed to 2.5 pg bovine pancreatic DNase I, for a period of 5 min in 250 ~1 reaction mixture (50 mM Tris. HCl, pH 8.0, lo%, v/v, glycerol, 0.1% bovine serum albumin (BSA), 25 mM MgCl,, 1 mM CaCl,, and 20 mM KCl). For the experiment of Fig. 4b (main panel), DNAprotein complexes equivalent to 25 Kg DNA (prepared by the Trask method) were exposed to 9 pg DNase I (bovine pancreatic DNase) in 500 ~1 of reaction mixture (defined above). After a IO-min incubation (or 1, 5, 10 min; Fig. 4b, inset), the reaction was stopped using a final concentration of 1% SDS and 15 mM EDTA. Preparation of HL60 cell lysates andprotein-associated DNA. HL60 cells in log phase growth were harvested by centrifugation at 400g for 6 min and resuspended in serum-free medium. For the preparation of whole cell lysates cells were resuspended at 5 X lo6 cells/ml and were lysed in the presence of protease inhibitors by the addition of aqueous SDS to a final concentration of 1% with 1% (v/v) Triton X-100, 15 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride (PMSF). The cell lysate was vortexed, aliquoted, and stored at -20°C. For SDS/ KC1 precipitation of DNA/protein complexes, circa 1 x 10’ log phase cells were used. The method was performed essentially as described by Trask (8) except that calf thymus DNA was omitted from the SDS buffer and vigorous vortexing was not used (so that long DNA fragments were not sheared).






In all hybridization experiments aliquots of cells were pretreated for 70 min with either 10e6 M all-trans-retinoic acid or 2 X lo-* M phorbol12-myristate 13-acetate (PMA). This timing was selected because previous studies indicate that protein-associated DNA breaks induced by retinoic acid during differentiation are present after this time interval (2). Controls were sham treated with the relevant diluent for the same times. To increase DNA-topoisomerase cross-linking, HL60 cells were exposed to VP16-213 lo-’ to lop5 M for 15 min and to reduce topoisomerase II activity, HL60 cells were preincubated with lop3 M novobiocin for 60 min at 37°C under standard culture conditions (viability was >95% as assessed by nigrosin dye exclusion).

pBR322 as substrate. It is characteristic of topoisomerase II that its catenation activity is dependent on ATP and spermidine. Controls were therefore performed with the latter omitted for the reaction mixture. A 5-/J sample of nuclear extract was incubated at 33°C for 1 h with 5 ~1 of reaction buffer containing 0.1 pg of the supercoiled DNA plasmid pBR 322, 10 mM Tris * HCl, pH 8.1, 10 mM MgCl,, 20 mM KCl, 10 mM ATP, 10 mM spermidine, 1 mM DTT, 0.5 mM EDTA, 15% glycerol, and 30 pg/ml bovine serum albumin. The reaction was stopped by the addition of 2.5 ~1 of a 1% SDS/I5 mM EDTA solution. The reaction products were electrophoresed through 1% agarose gels containing ethidium bromide to visualize the DNA.

Preparation and phase partitioning of topoisomerase II plasmid complexes. To prepare the topoisomerase II containing nuclear lysate, log phase HL60 ceils were harvested and lysed by the addition of a solution containing 10 mM Tris-PO, (pH 6.75), 0.1 mM disodium EDTA, 0.2 mM ethylene glycol bis(P-aminoethyl ether) N,N,N’,N’-tetraacetic acid (EGTA), 1 mM 2-mercaptoethanol, 10% (v/v) glycerol, 0.5% (v/v) Triton X-100, 1 mM PMSF, 1 mM dithiothreitol (DTT), 10 mM caminocaproic acid, and 0.5% (v/v) Nonidet-P40. This mixture was allowed to stand on ice for 5 min before nuclei were extracted by centrifugation at 1OOOgfor 8 min. The supernatant was discarded, the pellet was washed in nuclear wash buffer containing 10 mM Tris. HCl (pH 7.4), 10 mM NaCl, 1.5 mM MgCl,, 1 mM PMSF, and 1 mM DTT and recentrifuged at 1OOOgfor 8 min, and the supernatant was again discarded. The pellet was then dissolved in 200 ~1 of nuclear lysis solution (10 mM Tris. HCl (pH 7.4), 500 mM NaCl, 1.5 mM MgCl,, 1 mM PMSF, and 1 mM DTT) and then left to stand for 45 min before being further centrifuged at 9OOOgfor 15 min. The supernatant was reserved on ice until use as the nuclear extract. EcoRI-digested pBR322, which had been labeled with [a-32P]dCTP by random priming (25), was incubated with an aliquot of the lysate for 30 min in the presence of 1 mM ATP and in the presence of either 100 pM VP16-213 or a DMSO diluent control. The reaction was stopped by the sequential addition of prewarmed SDS and EDTA to 1% (v/v) and 25 mM, respectively. Aliquots (400 ~1) were then incubated overnight at 37°C with proteinase K (10 pg) or the relevant diluent, before being allowed to react with 400 mg/ml TMPEG, as described above. Excess TMPEG was allowed to react with 1 M free base lysine for a further hour before half of the conditions were exposed to bovine pancreatic DNase 1, as above, for 2 min. The DNA was recovered as above and run on a 1% agarose gel, dried down, and autoradiographed on X-ray film. To confirm the topoisomerase activity of the nuclear extract, catenation activity was measured using

Differential hybridization. Probes containing the putative specific sequences were prepared from differentiation induced cells using the two-step procedure with 60-s sonication and DNase digestion to shorten attached DNA between the SDS/KC1 and phase-partitioning steps. Maximum shortening is important to reduce contamination by adjacent DNA (which may contain repetitive sequences), because being heterogeneous these probes will be prone to a low signal-to-noise ratio. Ret+ and PMA+ probes refer to those prepared from retinoic acid- and phorbol myristate acetate-induced cells, respectively. To examine differential hybridization with the above probes, four target DNA samples were prepared per experiment: repeatedly extracted PEG-phase recovered DNA from differentiating and undifferentiated cells and their complementary phosphate phases (also multiply extracted to reduce cross-contamination). These filter samples were sonicated for 20 s after the Trask precipitation step to produce longer fragments (this allows regions adjacent to the breaks sites to be examined with known gene probes and does not restrict samples to the immediate vicinity of the binding site). The filters, either nitrocellulose or nylon GeneScreen Plus, were prepared by the dot-blot method of Kafatos et al. (26). Four serial dilutions (usually 50, 25, 12.5, and 6.25 ng DNA) were loaded onto the nylon (or nitrocellulose) with a dot-blot device and allowed to stand at room temperature for at least 30 min prior to being aspirated by vacuum. Filters were then dried between Whatman 3MM paper and then baked for 2 h at 80°C under vacuum (nitrocellulose filters). The filters were then prehybridized, for 3.5 h with agitation, using 6X SSC (1X is 0.15 M NaCl, 0.15 M sodium citrate, pH 7.0), 0.5% SDS, 5X Denhart’s solution (1X is 0.1 g Ficoll, 0.1 g polyvinylpyrrolidone, and 0.1 g BSA in 500 ml) and 100 pglml of highly sonicated salmon sperm DNA, allowing 1 ml/cm2 of filter. This and all subsequent hybridization/washing steps were carried out at 68°C with agitation. Hybridization fluid (6X SSC, 0.5% SDS, 5~ Denhart’s solution,



2 B 5 :: x s

800 700 600 500








100 0

FIG. 1. Exposure of whole cells to VP-16 for 15 min was used to increase the level of topoisomerase II complexed to DNA. The results are means + SE of PEG phase yields for eight independent experiments, expressed with respect to untreated controls.

100 pg/ml denatured salmon sperm DNA, and 0.1 M EDTA) was added allowing 0.5 ml/cm2 filter with a probe labeled with [cu-32P]dCTP by the random primer method (26,25). Hybridization was for 15-20 h. Filters were then washed twice for 45 min and then twice for 30 min, in 1 ml/cm2 1X SSC, 1X Denhart’s solution, and 0.1% SDS (prewarmed to 68°C). Finally, filters were rinsed twice for 20 min, in 1 ml/cm2 of prewarmed 0.1X SSC, were then blotted with Whatman 3MM paper and were exposed to Fuji RX 100 X-ray film with two intensifying screens at -70°C.




To demonstrate that the method is applicable to topoisomerase II, the amount of DNA complexed to the enzyme was varied by exposing HL60 cells to increasing concentrations of VP16-213. This drug stabilizes the enzyme at the stage where it is covalently bound to DNA (27) and prevents completion of the reaction (the ligation and dissociation steps). Figure 1 shows a dose-related increase in DNA recovery. Phase partitioning was performed using equal volumes of PEG-rich and phosphate-rich phases (multiple phosphate extractions were not used in these experiments). Results are means + SE of eight independent experiments (using either unsonicated or briefly sonicated samples) and are expressed as a percentage of the amount of DNA recovered without exposure of cells to VP16-213. Although the extent of the increment in PEG-phase DNA recovery varied considerably from experiment to experiment a trend of increasing recovery is evident. An increment at lo-’ M VP16-213 was detected in two of two briefly sonicated (2-5 s) samples and three of six unsonicated samples, but otherwise responses were similar.



Before attempting to demonstrate the extent to which PEG-phase recovery is due to the coupling of PEG to the protein attached to the DNA, it is important to appreciate that PEG-phase recoveries of free DNA are highly influenced by DNA fragment length (18). This therefore needs to be consideredwhen makingcomparisons to demonstrate affinity partitioning and also when selecting partitioning conditions (ratio of PEG:phosphate phase volumes and required rounds of phosphate extraction) to render the system more specific for PEG-protein-DNA complexes. Figure 2 shows the influence of sonication on PEGphase yield of protein free DNA (salmon sperm DNA). As the molecular weight of DNA falls with progressive sonication the yield in the PEG phase increases. These results are in agreement with those expected on theoretical grounds since with small molecules the partition coefficients approach unity (i.e., equal distribution between the two phases) (18) and the specificity of the partitioning is lost. Given the known partitioning coefficients of PEG-proteins (28) which can approach infinity (i.e., 100% in the PEG phase), we anticipated that the desired affinity-partitioning could still be achieved, even if DNA was highly fragmented (and this is often desirable on theoretical grounds; see below), merely by repeatedly extracting the PEG phase with additional rounds of phosphate phase (Fig. 2, squares). To limit the number of rounds required, phase systems can also be constructed with larger phosphate than PEG phases. We next examined, over a range of partitioning conditions, the extent to which additional rounds of phosphate extraction remove small fragments of protein free DNA by measuring the protein-associated DNA in the PEG phase. Whole genomic or DNA-protein complex enriched DNA was partitioned with additional rounds of phosphate extraction and varying phase volume ra-














FIG. 2. The influence of sonication on PEG phase yield of sonicated salmon sperm DNA using 1:l PEG:phosphate phase volume ratio and a single round of partitioning (circles) or seven rounds with fresh phosphate phases (squares). Modal fragment sizes were >25 kbp with no sonication and circa 0.7 kbp after a 10-s sonication.




tios (Table 1 and Fig. 3). Figure 3 shows progressive extraction of DNA with repeated exposure to new phosphate phases (mean results f SE of four experiments). In this case, using DNA-protein complexes as the start material, a 10-s sonication, and a 500:500 ~1 phase volume ratio, after three rounds of phosphate extractions, very little further DNA (0.2% of total) was recovered in the phosphate phase. Table 1 shows results for similar experiments with different start samples and conditions. Proteinase K-treated and -untreated aliquots were compared to determine what proportion of the recovered PEG-phase DNA was protein linked. In order to establish the extent to which PEG-phase yield is dependent on the PEG linked to the protein or the protein itself, PEG-modified and sham-treated DNA-protein complexes were compared, as well as proteinase K-treated samples and protein-free DNA derived from aliquots of the same complexes. ProteinDNA covalent complexes (largely topoisomerase-bound DNA, prepared by SDS/KC1 precipitation (8)) were either coupled to MPEG using TMPEG or sham treated with coupling buffer and untresylated MPEG (which cannot link to the protein). This sham treatment resulted in a much reduced recovery in the PEG phase (96.3 ? 4.3% reduction of yield, mean + SE of three independent experiments, ranging from 88.8% in which



Dependence of DNA PEG Phase Yield on Protein Attachment DNA Yield (% loaded) [pg recovered] Start material

Sonication (s)



















0 (+DNasel










0.46 [1.47] 0.70 [0.47] 1.21 [0.87] 0.56

0.00 [O.OO] 0.00 [O.OO] 0.00 [O.OO] 0.00



0.36 [O.ZO] 7.89 [x34] 11.42

0.01 [0.005] 0.00 [O.OO] 0.00







Note. A, whole genomic DNA from HL60 cells induced by retinoic acid; B, DNA-protein complexes prepared by SDS/KC1 precipitation from HL60 cell induced by retinoic acid; C, equivalent complexes from uninduced cells. PK+ and PKindicate the presence and the absence, respectively, of proteinase K treatment prior to partitioning; N.A., not assessed.










FIG. 3. The effect of additional rounds of extraction with phosphate phase on the recovery of DNA in the PEG and phosphate phases. Results are means _+ SE of four independent experiments.

one round of phosphate extraction was used to 100% with five rounds of extraction). Where aliquots of the same PEG-modified DNA-protein complexes were subjected to proteinase K treatment prior to partitioning PEG-phase yields were reduced by 99.7 + 0.4% and where proteinase K treatment was followed by phenol/ chloroform extraction to remove all protein, no DNA was detected in the PEG phase in all three experiments. Recoveries of DNA from the two phases varies and is circa 38.0 -t 8.1% of load for the PEG phase and 49.7 + 8.4% for the phosphate phase using the standard procedure outlined under Materials and Methods (using desalting columns and ethanol precipitation). Results are means f SE for six independent experiments and the difference is statistically significant (P < 0.01; paired t test). This will be important in some applications where the amount of recovered DNA is to be expressed in terms of loaded DNA and not with respect to a similarly handled control (i.e., DNA dissolved in and then extracted from the relevant phase). We next compared the SDS/KC1 precipitation method and the new technique in parallel. We evaluated their performance on differentiating and undifferentiated cells (Table 2). Since yield with the new procedure is highly dependent on the average length of the DNA fragments, no definitive yield for the method can be calculated. With 5-s sonication the new procedure produces a higher yield than for the Trask procedure. That this occurred despite a smaller fragment size in the 5-s-sonicated samples indicated that the new method recovers significantly more DNA fragments (and since each fragment represents one or more DNA/protein attachment sites the method is thus apparently more efficient with respect to those sites). This size differential is difficult to estimate precisely: the Trask material’s DNA molecular weight is typically over 20 kb (cf. Fig. 4b), with progressive sonication size is reduced. This



Comparison SDS/KC1

Yield, Cell

pretreatment (HL60 cells)




of the New Precipitation


Method with Method % loaded


6.0 k 3.0 (0.16 + 0.46)

[31 None

6.4 + 1.6 (1.00 f 0.33)




o.ig DNA) Trask

[n] method

2.61 + 0.52 (70.7 k 29.7) ]251 2.37 21 0.48 (61.6 +- 20.7) ]251



rounds of phosphate extraction used here, but the ratio of volumes of the two phases was increased to 250:750 ~1 PEG:phosphate. Topoisomerase II protects DNA fragments of circa 140 base pairs in DNase protection assays, using DNA gyrase (29). Thus protein-associated fragments of this size are unlikely to have more than a single molecule of enzyme attached to the DNA. DNase was also used to digest the DNA-topoisomerase complexes prepared by the Trask method. The protein (which may include a proportion of noncovalently

a u Using 5-s sonication and four rounds of phosphate extraction (to compensate for the smaller DNA fragment size induced by sonication; see Materials and Methods).

material is somewhat more robust and less reproducible in its response than the salmon sperm DNA in Fig. 2 (data not shown), but a circa 20-fold average size differential seemslikely, indicating a circa 40-fold differential in the number of fragments recovered. However, although there is independent experimental evidence (2) for DNA/topoisomerase II associations during induction of differentiation, neither technique, used singly, showed significant differences in the amount of DNA recovered from induced and uninduced cells (Table 2). On theoretical grounds shortening DNA should maximize the partitioning of topoisomerase II-DNA complexes by reducing the affinity of the attached DNA for the phosphate phase. In addition, as discussed above, a method capable of recovering DNA to which only a single protein molecule is attached would have advantages over the Trask method for the recovery of isolated topoisomerase II cleavage sites. We evaluated both sonication and DNase digestion as a means of shortening DNA. The former has the disadvantage that vigorous sonication might detach protein molecules and the latter that it is difficult to control. We established that significant protein-linked DNA remains after a 60-s sonication in Table 1 (on the basis of the difference between recoveries in proteinase K-treated and -untreated samples). Figure 4a shows an example of affinity partitioning of small PEG-protein-DNA complexes. DNA-protein complexes were sonicated for 60 s to fragment the DNA. Small fragments (mainly between molecular weight markers of 564 and 125 bp) were recovered from the PEG phase after partitioning. That these DNA fragments were recovered on the basis of their attached protein (not merely because of the tendency of small DNA molecules to escape to the PEG phase) is demonstrated by a control sample which was treated with proteinase K before the coupling step linking PEG to the proteinDNA complexes. Because of the anticipated very small DNA fragment sizes, not only were an additional three

I\ i-pa ,983,831100UBLETI

125 /











" 212


Ii 201

II I 1581 056 1-9 133 h B4A MOLECVLAR




FIG. 4. (a) Scanning densitometer traces of agarose gel electrophoresis of molecular weight markers (top trace), DNA from partitioned, highly sonicated DNA-protein complexes recovered from the PEG phase, and DNA from a similar sample but with pretreatment prior to partitioning with proteinase K. (b) Inset: DNase digestion of DNAprotein complexes precipitated by the Trask procedure where the protein “protects” a low-molecular-weight band of DNA from digestion. Lane (left to right) 1, HindIWEcoRI X DNA, lanes 2-4, l-, 5-, and lo-min DNase digestions of protein-DNA complexes (additional lanes, not shown, had no detectable remaining DNA after proteinase K digestion prior to DNase digestion for 1, 5, and 10 min). Main panel: DNase-digested DNA-protein complexes prepared by the Trask procedure, were fractionated by the new procedure. The figure shows densitometer traces of agarose gel electrophoresis of the DNA recovered from the PEG phase: (1) undigested protein-associated DNA; (2) DNase-treated protein-associated DNA; (3) as (2) but with pretreatment prior to partitioning with proteinase K (lower, flat trace).





FIG. 5. (a) Partial DNase I digestion of topoisomerase II cleaved [axP]dCTP-labeled pBR322. The plasmid was exposed to TMPEG to couple PEG to protein and then DNase to reveal protected fragments of DNA (protected by the attached PEG-protein conjugate) and the DNA phase partitioned by the new method. The recovered DNA revealed a single low-molecular-weight band, seen only in the presence of VP16, which enhances formation of DNA-topoisomerase II complexes (lane 2). In the absence of VP16 (lane 1) this band is undetectable and only a small amount of whole plasmid (P) was recovered from the PEG phase. Lane 3 shows molecular weight markers (EcoRIIHindIII digest of labeled X DNA). In (b) The ATP- and spermidine-dependent catenation activities of a comparable extract exposed to supercoiled pBR322. The supercoiled plasmid (S) is converted to catenanes (C) and relaxed (R) forms. Lane 1, plasmid; lanes 2 and 3, plasmid plus extract with spermidine and ATP, lanes 4 and 5, as for lanes 2 and 3 without spermidine; lanes 6 and 7, as for lanes 2 and 3 without ATP.

attached proteins) protects a low-molecular-weight band of DNA from digestion (Fig. 4b). Figure 4b, inset, shows a protected band appearing with increasing DNase treatment (l-10 min): proteinase K treatment before DNase, to remove attached protein, allows almost complete digestion of the DNA (lanes not shown). Using DNase digestion to reduce DNA size prior to partitioning (Fig. 4b, main panel), affinity partitioning of the recovered DNA was confirmed by a proteinase K-digested control: no DNA was recovered if the PEG-protein was removed from the DNA by proteinase K digestion before partitioning (note that in the experiment of Fig. 4b, inset, proteinase K was applied before DNase treatment to confirm that the band was due to protein protection; here it was applied after DNase). It must be emphasized that the new method is not specific for topoisomerase II (any covalently attached protein will be recovered). However, since we were interested in the performance of the method with this enzyme, we exposed a nuclear extract with topoisomerase II activity to a [32P]dCTP-labeled and then partially digested with DNase and finally the products were partitioned (Fig. 5a). Only in the presence of VP16 (which traps the enzyme at the stage where it is covalently complexed to protein) was a “protected” DNA band recovered in the PEG phase. In the absence of VP16 this band is not evident and less plasmid was recovered. Both the relative specificity of VP16 and the ATP and spermidine dependence of the catenation activity of an identical nuclear extract suggest that it is topoisomerase II complexed to DNA that is responsible for the single DNA band seen. The size of the protected fragment



is somewhat larger than that seen in Fig. 4b, but this is expected since PEG was attached to the protein prior to DNase digestion in this experiment and after digestion in the former experiment (hence protecting additional DNA). The presence of a single, low-molecular-weight, DNA band in Fig. 5 and of protein-specific affinity partitioning of fragments in the size range 564-125 bp in Fig. 4b (which includes the known protection size for a single topoisomerase II molecule), both indicate that the method is effective where only a single protein or topoisomerase II molecule is attached to the DNA. The reservation over the relative efficiencies of the Trask procedure for replication fork and single topoisomerase cleavage sites, does not of course mean that it cannot be used in series with the new technique to produce a two-step fractionation procedure for differentiation-associated topoisomerase cleavage sites. Even if the efficiency of SDS/KC1 precipitation is low, differentiation sites with one or few topoisomerase molecules are not likely to be less represented in the SDS/KC1 precipitate than in the supernatant. We therefore examined the two methods in series. Using the Trask then the new method, providing the DNA was short enough in the partitioning procedure (achieved using a 60-s sonication or sonication with additional DNase treatment) (Table 3), there was a significant increase in the proportion of PEG-phase DNA recovered from retinoic acidtreated cells when the start material was SDS/KC1 precipitated. This enhanced recovery of DNA from differentiating cells was dependent on the fragment size selected for the phase partitioning, since it was not detectable when using only 20-s sonication. This is probably a function of the DNA freed by sonication, since with no sonication, virtually all Trask material (which when prepared avoiding shearing consists mainly of fragments over 20 kb) carries to the PEG phase (data not shown) thus precluding the possibility of differential recovery. Finally we investigated whether the method had achieved its objective, i.e., whether it sufficiently enriched protein-linked DNA to allow us to address the issue of site specificity of the differentiation-induced attachment sites of topoisomerase II (and possibly other covalently attached proteins). If the topoisomerase II cleavage sites induced during differentiation are at specific sites in the genome, then, using a DNA preparation containing these putative specific sequences as a probe, the following “differential hybridization” patterns would be expected: (i) topoisomerase-associated DNA from differentiating cells would show a more intense hybridization than that from undifferentiated cells; (ii) DNA depleted of topoisomerase cleavage sites from differentiating cells would show a less intense hybridization than a similar preparation from undifferentiated cells. To construct the DNA preparation with putative specific sequences to be used as hybridization probes,






Yield, % whole genomic (pg DNA recovered) Cell pretreatment (HL60 cells) Retinoic




0.027 (0.047

P (paired

0.020 (0.018

+ 0.015 f 0.026)





0.40 + 0.14 (8.86 * 1.77)


20-s sonication

1.46 f 0.33 (14.96 + 0.33)


+ 0.008 k 0.009)

11.79 f 2.34 (42.42 f 11.50)


0.57 IL 0.28 (8.21 f 1.85)


Note. N.S., not significant,

Yield, % Trask DNA (pg DNA recovered) [n]

20-s sonication

181 t test)


DNA [n]

181 None






0.52 + 0.17 (21.25 f 6.86)

16.09 + 6.20 (47.00 f 12.75)







P > 0.05. digestion

in three


we produced two-step purified protein-associated DNA from retinoic acid (RA)- or PMA-induced differentiating cells and labeled and hybridized this DNA to DNA enriched for, or depleted of, protein-DNA from both undifferentiated HL60 cells and cells induced to differentiate by RA or PMA. Figure 6 shows representative hybridization experiments. There were significant and reciprocal differences in hybridization signals between the PEG- and phosphate-phase DNA prepared from induced and uninduced cells. To exclude any chance influences on differential hybridization due to fragment size and filter loading, differential hybridization was checked using independently prepared samples. Using three different probe samples from retinoic acid-induced cells and six independently prepared filter DNA sample pairs, from induced and uninduced cells, in 57 hybridization signal comparisons there was significantly higher hybridization where the filter DNA was prepared from retinoic acid-induced cells than from uninduced cells (P = 0.015, paired t test; 112.3% versus 87.7% of the average signal, respectively). Similarly using one probe from PMA-induced cells with three independently prepared filter DNA pairs and 12 hybridization signal comparisons there was a significantly higher hybridization signal PEG phase

where the filter DNA was prepared for PMA-induced rather than uninduced cells (P = 0.00005, paired t test; 130.6 versus 69.4% of the average signal, respectively). This differential hybridization is consistent with translocation to the PEG phase of selected DNA sequences and indicates that the method produces substantial enrichment of differentiation-associated protein-linked DNA which is nonrandomly distributed in the genome. Differential hybridization was lost when cells were preincubated with novobiocin. This is an inhibitor of DNA topoisomerase II that acts in a different manner to the epipodophyllotoxins and intercalators in that, unlike the latter, it does not stabilize cleavable complexes of enzyme and DNA, but tends to inhibit binding of the enzyme to DNA. Filter DNA from retinoic acid-induced cells preexposed to 1 X 10e3 M novobiocin for 60 min gave a hybridization signal not significantly different from that for undifferentiated cells (0.27 + Q.04 versus 0.28 + 0.05 absorbance/mm2/ng DNA, respectively; P = 0.74, paired t test for 6 comparisons). Similarly, filter DNA from PMA-induced cells preexposed to novobiotin gave a hybridization signal not significantly different from that for undifferentiated cells (0.27 f 0.01 versus 0.28 + 0.02 absorbance/mm2/ng DNA, respectively; P = 0.48 for 12 comparisons). Thus although neither the

Phosphate phase

PEG phase

Phosphate phase

50 DNA cont. (w.

25 1 12.5 6.25

Ret+ FIG. 6. Dot-blot hybridization of protein-associated DNAs from induced and uninduced cells.


Ret+ DNA





cells to PEG

pMA+ phase


and phosphate






SDS/KC1 nor the new method is specific for topoisomerase II, the transfer of DNA sequences to the PEG phase in differentiating cell preparations is likely to be due to an association with that enzyme rather than other proteins that covalently attach to DNA. Although at this concentration of novobiocin direct effects on topoisomerase I are unlikely, we cannot exclude indirect effects via other inhibitory actions of novobiocin. Whether the cross-hybridizing DNA (DNA common to both samples) is due to topoisomerase II (e.g., at replication forks which are randomly distributed), topoisomerase I, other covalently attached proteins, or noncovalently attached proteins is obscure. However, given that the bulk of DNA is associated with noncovalently attached proteins, the very low yield of recovered DNA suggests that their contribution is negligible. Although the differential hybridization between the differentiating and undifferentiated preparations is modest, it is reproducible and the tell-tale differential hybridization pattern should allow us to identify, in clones from DNA libraries, the DNA sequences that become topoisomerase bound as differentiation is induced. Given the proposed roles of modulation of DNA supercoiling in gene transcription (modulation of the binding of transcriptional modulatory via supercoiling status; relief of torsional stress in the transcriptional process) these DNA sequences may lie in or around differentiation genes. There are not many simple strategies for the detection of differentiation genes and few have yet been identified. The problem lies in the fact the these genes essentially code for cell death and cannot therefore be recovered by the transfection approach used so successfully for oncogenes. Insertional mutagenesis (analyzing the site of insertion of introduced DNA in cell clones where differentiation has been lost) is effective but time consuming. The construction of subtraction libraries of cDNA (from differentiating and undifferentiated cells) favors more abundant mRNAs and hence the products of, rather than the mediators of, the gene expression changes in differentiation programs. If regulatory genes exist in differentiation they seem unlikely to produce large quantities of mRNA. Chromatin structural changes have not to our knowledge yet been exploited to identify unknown genes, but given their observed lineage specificity and restricted nature in differentiation (l-3) they may provide a viable new approach. ACKNOWLEDGMENT This


was supported

by the Wellcome

3. Francis, G. E., Ho, A. D., Gray, D. A., Berney, J. J., Wing, M. A., Yaxley, J. C., Ma, D. D., and Hoff&and, A. V. (1984) Leukaemia Res. 8,407-415. 4. Francis, G. E., Gray, D.A., Berney, J. J., Wing, M. A., Guimaraes, J. E. T., and Hoffbrand, A. V. (1983) Blood 62, 1055-1062. 5. Sayhoun, N., Wolf, M., Besterman, J., Heish, T.-S., Sander, M., Levine, H., Chang, K.-J., and Cuatrecasas, P. (1986) Proc. N&l. Acad. Sci. USA 83,1603-1607. 6. Tse, Y. C., Kirkegaad, 255,5560-5565. 7. Osheroff,

N. (1989)

K., and Wang, Pharmacol.

8. Trask, D. K., DiDonato, 3,671-676. 9. Nelson, W. G., Liu, don) 322, 187-189. 10. Darby, (1986)

J. C. (1980)

L. F., and Coffey,

Vilarem, M.-J., Larsen, NCI Monogr. 4, 41-47.

M. T. (1984)

D. S. (1986)

M. K., Herrera, R. E., Vosberg, EMBO J. 9, 2257-2265.

11. Riou, J.-F., G. F. (1987)


R. (1986)

15. Delgado, Biotechnol.



M. T. (1989)

E. L.,


Cell 44,



567-574. D. (1990)

G. E., and Fisher, D. (1989) in Separations Systems. Applications in Cell Biology and D., and Sutherland, I. A., Eds.), pp. 211-

J., and Pettijohn,

D. E. (1967)


J. Biochem.

3, 33-41.

18. Miiller, M. (1985) in Partitioning in Aqueous Two Phase Systems. Theory, Methods, Uses and Applications in Biotechnology (Walter, H., Brookes, D. E., and Fisher, D., Eds.), pp. 227-266, Academic Press, New York. 19. Albertsson, molecules,

P.-A. Wiley

(1986) Partitioning Interscience, New

of Cell Particles York.

and Macro-

20. Husted, H., Kroner, K. H., and Kula, M.-R. (1985) in Partitioning in Aqueous Two Phase Systems. Theory, Methods, Uses and Applications in Biotechnology (Walter, H., Brookes, D. E., and Fisher, D., Eds.), pp. 529-587, Academic Press, New York. 21. Harris, J. M. (1985) Reu. Mucromol. Chem. Phys. C25, 325-373. 22. Nilsson, K., and Mosbach, mun. 102,449-457. 23. Maniatis, Biol. 84, 24. Jordan, (London),

T., Venable, 37-64.

K. (1981)


J. H., and Lerman,

C. F., Lerman, L. S., and Venable, New Biol. 236, 67-70.


Res. Com-

L. S. (1974)

J. Mol.

J. H. (1972)


25. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd ed., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. A. (1979)

27. Chen, G. L., Yang, L., Rowe, T. C., Halligan, B. D., Tewey, and Liu, L. F. (1984) J. Biol. Chem. 259, 13,560-13,566.



E., and Riou,

C., Patel, J. N., Francis, G. E., and Fisher, Appl. Biochem. 12, 119-128.

16. Delgardo, C., Francis, Using Aqueous Phase Biotechnology (Fisher, 213, Plenum, London. 17. Favre,

Cell 46,


and Nordheim,

C. J., Multon,

12. Spitzner, J. R., Chung, I. K., and Miiller, Acids Res. 18, l-11. 13. Cockerill, P. N., and Garrad, W. T. (1984) 14. Treisman,

J. Biol.



J. A., and Muller,

26. Kafatos, F. C., Jones, C. W., and Efstratiadis, Acids Res. 7, 1541-1552.


1. Khan, Z., and Francis, G. E. (1987) Blood 69,1114-1119. 2. Francis, G. E., Berney, J. J., North, P. S., Khan, Z., Wilson, Jacobs, P., and Ali, M. (1987) Leukaemia 1,653-659.



28. Kerr, L. J., Shafer, S. G., Harris, Snyder, R. S. (1986) J. Chromatogr. 29. Liu, L. F., and Wang, J. C. (1978)

J. M., Van Alstine, 354, 269-282. Cell 15,979-984.

Nucleic K. M.,

J. M.,


protein complexes applied to DNA topoisomerase II cleavage sites.

The object of this study was to devise a purification method for DNA/topoisomerase II complexes, with which to examine the enzyme's cleavage site spec...
2MB Sizes 0 Downloads 0 Views