ABSTRACT
The pattern of sites for cleavage mediated by topoisomerase II was determined in 830 kb of cloned DNA from the Drosophila X chromosome, with the objectives of comparing it with mapped structural and functional landmarks and examining if the correlations with such landmarks reported in individual loci can be generalized to a region ~100 times longer. The relative frequencies of topoisomerase II cleavage sites in 247 restriction fragments from 67 clones were quantified by hybridization with probes prepared from DNA fragments which abutted all cleavage sites in each clone, selected through the covalently bound topoisomerase II subunit; the specificity and quantitative nature of this method were demonstrated using a plasmid DNA model. The 12 restriction fragments with strong nuclear scaffold attachment (SAR) activity, of which seven possess autonomous replication (ARS) activity, show statistically strong coincidence or contiguity (P <= 0.11) with regions of high topoisomerase II cleavage site frequency. These regions show no correlation with repetitive sequence or A/T or C/G content and some extend over >10 kb; their sensitivity is therefore unlikely to be due to alternating purine-pyrimidine repeats or regions of Z conformation, which are preferred motifs. The hypothesis that they possess intrinsic curvature is consistent with the similarity of their length and spacing to regions of predicted curvature in the 315 kb DNA of Saccharomyces cerevisiae chromosome III and with the reported strong binding preference of topoisomerase II for curved DNA. The topoisomerase II cleavage pattern in this DNA further shows that its relationships to functional properties seen in individual loci, especially to MAR/SAR and ARS activity and to the restricted accessibility of DNA to topoisomerase II in vivo, can be generalized to much longer regions of the genome.
The selection of sites for binding and reaction of topoisomerase II on DNA is determined by sequence and secondary structure motifs (1 -11 ). Sequences of alternating purine-pyrimidine repeats provide preferred targets (1 -4 ) and the enzyme binds strongly to regions with intrinsic curvature (5 -7 ); its affinity for curved and Z conformations is two orders of magnitude greater than that for the B conformation (5 -7 ). Topoisomerase II also binds to points of crossover of two duplex DNAs (8 ,9 ) and to regions in hairpin (10 ) and tetraplex (11 ) conformations. Binding sites can be revealed by inducing DNA cleavage by denaturation of the reaction intermediate, in which both strands are transiently interrupted and bound covalently to the enzyme (12 ). In genomic DNA, the cleavage sites which have been mapped in a number of loci show certain correlations with functional properties (13 -24 ); for example in the Drosophila hsp70 (14 ,19 ) and histone (4 ,21 ) genes they flank the boundaries of coding regions, in the human and mouse c-fos gene they are clustered in promoters and enhancers (20 ) and they are highly reiterated in regions of DNA which bind to the nuclear matrix (MARs) and nuclear scaffold (SARs) (4 ,16 ,22 ,23 ,25 ).
This evidence that topoisomerase II recognizes functional and conformational motifs led us to examine its cleavage pattern in a region of genomic DNA two orders of magnitude longer than the individual loci previously studied, a cloned 830 kb continuum from region 14B-15B of the Drosophila X chromosome (26 ) in which a number of structural and functional landmarks have been mapped. These include 12 periodically spaced restriction fragments with strong SAR activity (27 ), 11 of which are attached to the nuclear matrix in vivo (28 ), 27 with ARS activity (29 ) and >80 transcription units (26 ), as well as structural features including repetitive sequences (30 ) and regions enriched in A/T or C/G (31 ; Jullien et al., submitted for publication). Correlations between certain of these properties have been shown by statistical analysis; for example SARs generally frame transcription units and strong SAR and ARS activities frequently coincide (27 ,29 ). Here we have mapped the relative frequency of topoisomerase II cleavage sites in restriction fragments covering this region using a new approach and examined its relationship to their structural and functional properties.
The [lambda] clones of Drosophila DNA covering this 830 kb region have been described (26 ). Plasmid pRYG, a derivative of pUC19 containing a 54 bp insert with strong topoisomerase II cleavage sites (2 ), was obtained from TopoGen (Columbus, OH). DNA was cleaved by topoisomerase II (bovine [alpha]-isoform; TopoGen) by incubation in 20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 5 mM MgCl2, 1 mM DTT, 5% glycerol, 30 [mu]g/ml BSA, 250 [mu]g/ml ATP. The ratio of enzyme to DNA (by mass) was 1:10 and the topoisomerase II religation inhibitor VM-26 was added to 10 [mu]g/ml, except where indicated in Figure 1 C. After 15 min at 37oC the reaction was terminated by adding SDS to 0.5% and EDTA to 10 mM; proteolytic enzymes were not used in order to conserve the topoisomerase II subunit covalently bound to the end of the DNA fragments which abutted cleavage sites (31 ). These fragments were recovered specifically by binding to a glass fibre filter (29 ); the sample was diluted in 10 ml 10 mM Tris-HCl, pH 8.0, 0.3 M NaCl, 0.4 M guanidine-HCl, 10 mM EDTA, 0.01% sarkosyl, and passed under gravity through a moistened Whatman GF/C filter. After washing with the same buffer the bound DNA fragments were deproteinized and eluted in 1 ml 10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.5% SDS, 500 [mu]g/ml proteinase K for 16 h at 37oC and purified by extraction with phenol/chlorofom.
Two different probes were hybridized successively to restriction fragments from cloned Drosophila DNA inserts, separated by electrophoresis in individual lanes of a 0.8% agarose gel and transferred to a Hybond N+ membrane (Amersham). The first (total DNA) probe served to quantify the relative amount of DNA in each fragment and was prepared from DNA of the same clones restricted with EcoRI, HindIII and BamHI. The second (topo II site) probe served to identify fragments which contained topoisomerase II cleavage sites and to quantify these sites and was prepared by further topoisomerase II cleavage and selection of DNA fragments which abutted cleavage sites, as described above. Probes were labelled by random priming with [32P]dCTP using the Multiprime system (Amersham). Hybridization was at high stringency conditions (30 ) in 4* SET (20* SET is 0.6 M Tris-HCl, pH 8.0, 3 M NaCl, 40 mM EDTA), 2* Denhardt's solution, 0.2% SDS, 100 [mu]g/ml sheared denatured herring sperm DNA at 65oC for 18 h. Membranes were washed at room temperature with 1* SSC, 0.1% SDS, for 10 min, twice at 65oC with 3* SET, 0.2% SDS for 30 min and once at 65oC with 1* SET, 0.2% SDS for 30 min. The autoradiographic signals from the total DNA probe were quantified by scanning with a Vernon PH15 densitometer under conditions of linear response and the probe was removed by incubating the membrane in 0.5 M NaOH, 1.5 M NaCl for 30 min at 45oC and then in 1.5 M Tris-HCl, pH 8.0, 1.5 M NaCl for 30 min at 45oC; the efficiency of this stripping procedure is demonstrated in Figures 1 and 2 . The membrane was rehybridized with the topo II site probe under the same high stringency conditions and this autoradiograph was scanned after adjusting the signal from a vector DNA fragment (Fig. 2 ) to the value given by the first probe. The cleavage site frequency was expressed as the ratio of the integrated signals (topo II site probe/total DNA probe) for each restriction fragment.
The use and interpretation of statistical tests to examine correlations between different properties of restriction fragments from this region of DNA have been described previously (26 -31 ). Initial comparisons of the topoisomerase II cleavage site frequency in each fragment with a second property were made using the MacMul+GraphMu factorial corrrespondence analysis program (34 ) and correlations indicated by this test were analysed quantitatively by the standard homogeneity and Wilcoxon tests. The properties compared (26 -31 ) were the repetitive sequence content determined by cross-hybridization (30 ), the number of sites for restriction enzymes which cleave in A/T-rich (31 ) or C/G-rich sequences (Jullien et al., submitted for publication), the level of transcription in 0-18 h old Drosophila embryos determined by hybridization with reverse transcripts of total mRNA (26 ), strong SAR activity determined by a standard procedure (27 ) and ARS activity determined by transformation of Saccharomyces cerevisiae (28 ). For statistical analyses the fragments were ranked in classes according to their cleavage site frequency, a histogram was constructed to show the number of fragments in each class which possessed or lacked the second property and the probability that this distribution was random was calculated by the homogeneity and Wilcoxon tests and expressed as p1 and p2 respectively. Relationships of proximity between fragments with different cleavage site frequencies and these properties were examined by the same method after attributing to each fragment (n) the cleavage site frequency of the neighbouring fragment, (n + 1), (n + 2) or (n + 3), while keeping the other parameter unchanged.
Rather than detecting topoisomerase II cleavage sites in each of the 247 fragments resulting from restriction of this region of DNA with EcoRI, HindIII and BamHI (26 ) by direct or indirect end-labelling (4 ,13 -20 ), a more rapid procedure was devised to compare the relative frequency of cleavage sites in restriction fragments from each clone by hybridization with a probe which contained sequences abutting all topoisomerase II cleavage sites in the clone. This probe, termed a topo II site probe, was prepared from DNA fragments which bore a terminal covalently bound enzyme subunit, selected specifically by a filter binding procedure (33 ) after cleavage of restricted DNA from the clone by topoisomerase II; it therefore contained sequences complementary to those at all cleavage sites at a relative abundance which reflected the frequency of cleavage, so that the hybridization signal on each target restriction fragment was proportional to the number of cleavage sites which the fragment contained and to the frequency with which they were cleaved. This strategy greatly reduced the number of experiments required and is analogous to using a probe containing a mixture of DNA sequences of different abundance, reverse transcribed from a heterogeneous population of mRNAs, to compare the relative frequency of transcribed sequences in different restriction fragments (see for example 26 ).
The specificity of this approach has been demonstrated in studies of the chicken [alpha]-globin locus, where a single restriction fragment known to contain cleavage sites was identified among six contiguous fragments (35 ). Its quantitative nature was assessed using a plasmid containing a 54 bp insert with strong topoisomerase II cleavage sites, pRYG (2 ), which was restricted to produce three fragments of which one contained the insert (Fig. 1 A). After further cleavage by topoisomerase II the fragments abutting the cleavage site were selected to prepare a topo II site probe; this probe recognized only the restriction fragment which contained topoisomerase II cleavage sites (Fig. 1 A). To simulate differences in the frequency of cleavage sites in a particular fragment, varying amounts of pRYG were mixed with pUC19, which lacks the insert but is otherwise identical (2 ), restricted and subjected to electrophoresis; the hybridization signal from the topo II site probe relative to that from the total DNA probe was proportional to the amount of the pRYG fragment containing topoisomerase II cleavage sites (Fig. 1 B). This procedure thus specifically detects restriction fragments which contain topoisomerase II cleavage sites and allows the relative frequency of these sites to be quantified. To obtain topo II site probes for all 67 clones of Drosophila DNA employed, the amount of topoisomerase II required was prohibitively high and we therefore examined whether the religation inhibitor VM-26 (32 ) could be used to increase the yield of cleaved DNA to prepare these probes. Probes prepared using VM-26 produced higher hybridization signals than those prepared without VM-26 but using a 10-fold greater quantity of enzyme (Fig. 1 C). The signals on 21 restriction fragments from five representative clones were higher by a factor of 4.0; the small standard deviation of +-0.9 shows that the relative cleavage site frequencies determined by this low resolution approach using fragments of average length 3.4 kb were not significantly distorted by the use of probes prepared using VM-26 (see Discussion).
Only strong SAR activity and ARS activity showed a correlation with cleavage site frequency by factorial correspondence analysis (34 ) and none was detected with repetitive sequence content, A/T or C/G richness, weak SAR activity or transcriptional activity (not shown). These correlations were examined quantitatively by the homogeneity and Wilcoxon tests, in which the number of fragments in each cleavage frequency class which possessed or lacked a second property and the probability (P) that these numbers could arise from a random distribution were calculated.
The topoisomerase II cleavage site frequency was not correlated with the known structural properties of repetitive sequence content (Fig. 4 , upper row) or A/T or C/G richness (not shown) nor with the functional properties of weak SAR or transcriptional activity (not shown). Restriction fragments with strong SAR activity were clearly over-represented in classes with higher cleavage site frequencies (Fig. 4 , second row, left panel). The correlation between these properties was statistically highly significant, as shown by the low p1 and p2 values (P <= 0.11); this correlation was still seen after a relative displacement by one or two fragment lengths (one average fragment length is 3.4 kb). Fragments with ARS activity were also over-represented in the classes with a high cleavage site frequency and the correlation between these properties was statistically highly significant (P <= 0.06; Fig. 4 , bottom row).
The strategy used here to examine topoisomerase II-mediated cleavage in a region of DNA much longer than those studied previously is more rapid than direct or indirect end-labelling methods (4 ,13 -24 ,36 -41 ), which, further, are not suited to quantitative measurement or comparison of cleavage site frequencies because rapid cleavage at sites close to the labelled extremity or to the probe can mask cleavage occuring at more distant sites. This approach shows the relative frequency of cleavage sites in different fragments, but does not distinguish those with a few strong cleavage sites from those with more frequent but weaker sites. The use of probes prepared using the religation inhibitor VM-26 had little influence on the relative frequency of cleavage sites at this relatively low resolution using fragments 3.4 kb in average length (Fig. 1 C), when the differential stimulation seen at nucleotide resolution (35 -41 ) is likely to be masked, consistent with reports that cleavage patterns mapped by indirect end-labelling do not show significant differences in the presence of VM-26 (17 ,19 ,20 ). The strong preference of topoisomerase II for binding and cleaving in alternating purine-pyrimidine sequences is not influenced by VM-26 (42 ,43 ). Topoisomerase II of bovine origin was employed to cleave Drosophila DNA because of its ready availability, but the sequence preference of the enzyme does not show significant species differences (2 ,42 ) and a heterologous enzyme is therefore commonly used in binding and cleavage studies (2 ,16 ,17 ,20 ,21 ,37 ,43 ).
Of the functional properties mapped previously in this 830 kb region of DNA, only strong SAR and ARS activities show a statistically strong correlation (P <= 0.11; Fig. 4 ) with a higher frequency of topoisomerase II cleavage sites. A proximity relationship was seen by statistical analysis in cases where a SAR fragment did not itself show frequent cleavage sites, with one exception (Figs 3 and 4 ), consistent with the contiguity in the chicken lysozyme locus (44 ) of a MAR and a strongly curved region which would provide a preferred cleavage target (5 -7 ). The present results thus extend the evidence that strong cleavage sites occur in a number of individual MARs and SARs (4 ,16 ,17 ,22 ,23 ,25 ) to the 12 strong SARs in this 830 kb region, considerably strengthening the hypothesis that this is a general characteristic of MARs/SARs. The inverse relationship was not seen, however; a high frequency of cleavage sites is not predictive of SAR activity (Fig. 3 ). ARS activity also showed a strong correlation with higher cleavage site frequency (P <= 0.1, Fig. 4 ), possibly reflecting the implication in ARS function of cruciform conformations (45 ,46 ) which offer cleavage targets (8 ).
Topoisomerase II associated with the nuclear matrix can cleave this DNA at 11 positions in the region between 300 and 800 kb, which have been interpreted as positions where DNA is anchored to the nuclear matrix (28 ). The frequency of cleavage sites in vitro in some of the corresponding restriction fragmentsis only average (Fig. 3 ), extending to this much longer region of DNA the evidence from a number of individual loci (20 ,25 ,28 ) that the accessibility of cleavage sites in vivo is determined predominantly by chromatin structure and by the manner in which DNA is organized within the nucleus.
Most regions with a frequency of topoisomerase II cleavage higher than the mean extend over several contiguous restriction fragments and are >10 kb in length (Fig. 3 ), suggesting that their sensitivity is unlikely to be caused by sequence motifs alone. Their length and absence of correlation with repetitive sequence content (P <= 0.59; Fig. 4 ) or with A/T or C/G content (not shown) suggest that these regions are unlikely to be enriched in sequences of alternating purine-pyrimidine repeats or in the Z conformation which provide preferred cleavage targets. In view of the marked preference of topoisomerase II for binding to curved DNA (5 -7 ) their distribution could, however, be explained by the hypothesis that their frequent cleavage is a consequence of curvature; their length and spacing of ~13 and ~28 kb (Fig. 3 ) are comparable with the values of ~10 and ~18 kb for the interspersed strongly curved regions seen in the computed spatial conformation of the 315 kb DNA of chromosome III of S.cerevisiae (47 ) and they are coincident or contiguous with strong SARs, which show intrinsic curvature (49 ), and with ARSs, which lie in curved regions of S.cerevisiae chromosome III (46 ) (Fig. 4 ). As well as raising this hypothesis, which remains to be explored, this study of the topoisomerase II cleavage pattern in 830 kb of Drosophila DNA shows that its relationships to functional properties reported in individual loci, especially to MAR/SAR and ARS activity, and to the restricted accessibility of genomic DNA to topoisomerase II in vivo, can be generalized to much longer regions of the genome.
We thank Bristol-Myers Squibb for VM-26. This work was supported by the MRC of Canada (grant MT-1172), the CNRS, France and by Visiting Scientist Fellowships from the Coopération France-Québec to R.M. and the Fonds de la Recherche en Santé du Québec to S.V.R.
REFERENCES