Nucleic Acids Research, Vol. 19, No. 6 1235

Eukaryotic topoisomerase I - DNA interaction is stabilized by helix curvature Susanne Krogh, Uffe H.Mortensen, Ole Westergaard and Bjarne J.Bonven* Department of Molecular Biology and Plant Physiology, University of Aarhus, C.F.M0llers Alle 130, 8000 Aarhus C, Denmark Received December 28, 1990; Revised and Accepted February 27, 1991

ABSTRACT The influence of DNA structure on topoisomerase 1DNA interaction has been investigated using a high affinity binding site and mutant derivatives thereof. Parallel determinations of complex formation and helix structure in the absence of superhelical stress suggest that the interaction is intensified by stable helix curvature. Previous work showed that a topoisomerase I binding site consists of two functionally distinct subdomains. A region located 5' to the topoisomerase I cleavage site is essential for binding. The region 3' to the cleavage site is covered by the enzyme, but not essential. We report here that the helix conformation of the latter region is an important modulator of complex formation. Thus, complex formation is markedly stimulated, when an intrinsically bent DNA segment is installed in this region. A unique pattern of phosphate ethylation interferences in the 3'-part of the binding site indicates that sensing of curvature involves backbone contacts. Since dynamic curvature in supercoiled DNA may substitute for stable curvature, our findings suggest that topoisomerase I is able to probe DNA topology by assessment of writhe, rather than twist. INTRODUCTION Readout of the sequence information in DNA by proteins can occur directly or indirectly. Direct readout, termed 'digital recognition' (1), is based on formation of specific contacts, typically hydrogen bonding, between amino acids and bases. This mode is characteristic of highly sequence specific proteins, such as the C l repressor of bacteriophage lambda (2). Indirect readout termed 'analogue recognition' (1) relates to the fact that DNA helix structure is sequence dependent. Proteins can read sequence information translated into structural microheterogeneity by contacting unspecific components of the helix, such as backbone phosphates. This usually results in relatively promiscuous interaction. The two modes of recognition are not mutually exclusive and are used in combination by many DNA binding proteins. *

To whom correspondence should be addressed

Analogue recognition may be the major mechanism in topoisomerase I (topo I)-DNA interaction. Thus, the enzyme interacts with DNA in a non-specific, but sequence dependent manner (3,4,5). The minimal binding site requirements defined by the 4-5 bp degenerate consensus sequence give no indication of a specific sequence code for recognition (6,7). Functional analyses have shown that the consensus region corresponds to the essential part of the binding site. The essential region is operationally defined as the part of the binding site, where abasic sites or partial single-strandedness inhibit enzyme action (8,9). Topo I binding selectively stabilizes the DNA duplex in this region, indicating that the enzyme forms tight contacts to this part of the binding site (9). The recent observation (10,11) that a conserved structural motif is associated with the degenerate consensus, suggests that local aspects of helix geometry set the minimal requirements for recognition. A topo I binding site encompasses 20-25 bp (8,12), in which the topo I cleavage site is centrally located. Yet the enzyme contacts the binding site in a highly asymmetrical manner, as the essential region lies almost entirely 5' to the cleavage site. No function has previously been ascribed to the 3' part of the binding site (8,9), which has therefore been designated the 'non-essential region'. Here, we examine the influence of the non-essential region and find that topo I-DNA interaction is modulated by the helix conformation in this binding site subdomain. Sequence-directed conformational variations in DNA have come into focus in recent years (13), and the stable curvature associated with oligo (dA) . (dT)-tracts (14,15,16) has drawn particular attention. This structural trait can affect several processes, such as transcription regulation (17,18,19) and sitespecific recombination (20,21). Curvature dependent DNA binding has been reported for a diverse group of proteins, including SV40 large T antigen (22), restriction endonuclease Eco RI (23), the Crithidia nicking enzyme (24), and the [Su(Hw)] zink-finger protein from Drosophila (25). Two lines of evidence suggest that topo I may belong to this group: (i) High affinity binding sites have been found to contain oligo (dA) (dT)-tracts (26,27), and (ii) topological domains harboring an oligo (dA) (dT)-rich, intrinsically curved segment of Crithidia kinetoplast DNA are preferentially topoisomerized by topo I in vitro (28). It remains to be rigorously established, however, -

-

1236 Nucleic Acids Research, Vol. 19, No. 6 whether helix curvature per se or some other structural property of oligo (dA) * (dT)-tracts are causal in determining the efficacy of topo I-DNA interaction. We have therefore analyzed two relevant parameters, intrinsic curvature (14,15,16) and conformational rigidity (29), in relation to topo I-DNA interaction. The results suggest that helix curvature in the nonessential part of the topo I binding site stabilizes this interaction.

Modification interference DNA modification interference experiments followed the procedure for the cleaved strand of topo I-DNA complexes outlined in (8). Chemical modification with dimethyl sulfate or ethyl nitrosourea and subsequent cleavage chemistry were according to (33).

EXPERIMENTAL PROCEDURES

Electrophoresis and densitometry Analytic and preparative DNA sequencing gel electrophoresis were in 6% or 12% polyacrylamide/8M urea DNA sequencing gels. Densitometric scanning of autoradiograms were as previously described (30).

Plasmids Oligonucleotides were synthesized using a DNA synthesizer model 381A from Applied Biosystems. Plasmids were constructed by inserting synthetic oligonucleotides between the Sph I- and Hin CII-sites of pUC19 or between the Eco RV- and Sph I-sites of pBR322. The pUC19 constructs were named pUCNC(number of insert), and the pBR322 clones were named pNC(number of insert). Gel purified restriction fragments of these plasmids, labeled at recessed 3'-ends using [ct-32p] dATP and E. coli DNA polymerase I, Klenow fragment, were used as substrates in all experiments.

Topo I cleavage reactions Topo I was purified from exponentially grown cultures of Tetrahymena thermophila, strain B1868-VII, according to (30). Single-end labeled restriction fragments (3-5 fmol) were incubated with 0.2-2.4 ng topo I in 30 /1 10 mM Tris-HCl pH 8.0, 3 mM CaCl2, 1 mM MgCl2, 0.2 M sucrose, 1 mM dithiothreitol, 100 lAg/ml bovine serum albumin for 15 min at 30°C. Reactions were terminated by addition of SDS and EDTA to final concentrations of 1% and 10 mM. Samples were heated briefly at 45°C, supplemented with NaCl to 0.8 M, and ethanol precipitated. Pellets were dissolved in 4 1l 10 mM Tris-HCl pH 8.0, 1 mM EDTA and digested with proteinase K (300 jig/ml) for 60 min at 45°C. One volume of deionized formamide, 0.05% bromophenol blue, 0.03% xylene xyanol, 5 mM EDTA, pH 8.5 was added, the samples were heated briefly at 95°C and loaded onto DNA sequencing gels.

Methidiumpropyl-EDTA*Fe(ll) cleavage MPE Fe(II) cleavage of DNA was according to (31). The MPE Fe(HI) complex solution was made immediately before use by adding 5 ILI 1 mM MPE and 5 IA 1 mM Fe(NH4)2(SO4)2 to 40 ,d 15 mM Tris-HCl pH 7.4 and 1 yd 0.5 M dithiothreitol. 5 ll of this mixture was added to 40 fmol end-labeled restriction fragment in 80 yd 10 mM Tris-HCl pH 8.0, 50 mM NaCl, 3 mM CaCl2, 1 mM MgCl2 containing 1 mM H202. After 2 min of incubation at 30°C the reaction was stopped by addition of 8.5 !d bathophenantroline. After repeated ethanol precipitation the samples were processed for sequencing gel electrophoresis.

EDTA*Fe(II) cleavage Generation of hydroxyl radical by reacting EDTA Fe(II) with H202 followed the principles in (32). A freshly prepared mixture containing 8 l1 0.3 mM Fe(NH4)2.(SO4)2, 8 Al 0.6 mM EDTA pH 7.4, 6 Al 0.3% H202 was added to 40 pmol endlabeled fragment in 70 ,ul Tris-HCl pH 8.0, 50 mM NaCl, 3 mM CaCl2, 1 mM MgCl2. After 2 min at 30°C the formation of hydroxyl radical was quenched by sequential addition of 10 Id A1 0.3 M thiourea, 32 0.2 M EDTA, 200 Aul 0.3 M NaCl, and 750 1l ethanol. The DNA was reprecipitated twice and processed for sequencing gel electrophoresis.

RESULTS Topo I cleavage on substrate variants Despite the loose sequence-specificity of topo I, the enzyme acts on some sites with exceptional efficiency. Such sites were originally found in Tetrahymena and Physarum rDNA spacers (26,27) and later in non-rDNA spacer sequences (34). These sequences are variations of a common motif, and their consensus corresponds to the sequence shown in Fig. IA. The salient feature of the motif is that two oligo (dA) * (dT)-tracts flank a core region obeying the degenerate minimal consensus. Fig. IA summarizes previous data on the location of the cleavage site and the essential region within the binding site. Note that both (dA) * (dT)-tracts are part of the binding site, and the tract located 5' to the cleavage site is partially included in the essential region. To elucidate the role of the (dA) (dT)-tracts, we cloned a series of synthetic mutant derivatives with substitutions in these regions and the core part maintained constant (see Fig. IB). Binding of topo I to this family of sequences was quantitated by cleavage titration as previously described (30). Topo I catalysis involves transient

A. 5'- CAT GAAAi

kGIAA AA AT TT TT AAA -3'

3'- G TA C T T V T

CTTTTTAAAAATTT -5'

B. 5'-AAAAAAGACTTAGTTT TTTTTTAAAAA-3' 5'-TATATAGACTTAGATATATATATATAG -3' 5'-CA CA CAGACTT A GACACACACACA CAG -3' 5'-AAAAAAGACTTAGAAAAATTTTTAAAG-3' 5'-TA TATAGAC T TA GAAAAAT TT TT AAAG-3' 5-AAAAAAGACT TAGATATATATAT A TAG-3' 5'-TATATAGACTTAGAGCAAAAATTT TTAAAG-3' 5'-TATA TAGAC T TA GAGCATGCATCGAAAAA T T TAAAG-3' 5'-AAAAAA GACTTA GAGCAAAAATT T TTAAAG-3' 5'-AAAAAAGAC T TAT CGATGATAAGC TGT CAA-3' 5'-AAAAAAGACTTAGTTTTTAAAAAG-3'

5'-AGGGTGACGGTGCCGAGGATGACGATG-3' 5'-AGGGT GACGGT GCAAAAATTT TT AAAG-3'

NC 3 NC12 NC 13

NC 14 NC 15 NC16 NC18 NC20 NC21 NC22 NC25 NC27 NC28

Fig. 1. The topo I binding site. (A) The extent of the binding site, as defined by footprinting (8) is indicated by wavy lines. The essential region delimited by modification interference analysis (8) is shown in the shaded box. The stippled line shows the extent of the minimal duplex region required for topo I cleavage (9). The arrow denotes the topo I cleavage site (26). (B) Mutant binding sequences. The underlining indicates deviations from the wild type (NC14).

Nucleic Acids Research, Vol. 19, No. 6 1237 formation of a single-strand cleaved intermediate. Cleaved intermediates can be trapped by addition of SDS (sodium dodecyl sulfate), and the amount of cleavage at a given site is proportional to the fractional binding at the site. Fig. 2 shows the results of a series of titration experiments, where appropriate restriction fragments, each containing one of the sequences NC3, 12,13,14 15,16, or 25, were single-end labeled in the scissile strand and incubated with varying amounts of topo I. The reactions were subsequently terminated with SDS and the cleavage products separated by denaturing gel electrophoresis. Following autoradiography, the percentage of substrate molecules cleaved at the inserted sequences was calculated from densitometer scans and plotted against the amount of topo I added. The wild type recognition sequence, NC 14 (crosses), is cleaved efficiently by topo I. Approximately 35% of the substrate molecules suffer cleavage in the recognition sequence at saturating levels of enzyme. From the half-saturation value we estimate that KD = 10-10 M, consistent with previous determinations (8,30). In NC 15, NC 16, and NC 12, respectively, the upstream, downstream, or both (dA) (dT)-tracts were replaced by runs of alternating AT. In the upstream tract this change causes a twofold reduction of the maximal cleavage level (NC 15, filled triangles). The overall shape of the titration curve is similar to that of the wild type, but half-saturation occurs with approximately two fold less enzyme, and the maximal cleavage level is reduced to abut 15%. Replacement in the downstream tract leads to a markedly reduced cleavage (NC 16, filled boxes). Replacement in both tracts (NC 12, open circles) further reduces the cleavage intensity. The melting temperature of alternating AT is lower than that of homopolymeric AT (35). The reduced cleavage noted above could therefore be due to reduced stability in the regions flanking the essential part of the binding site. To address this, GC-pairs were introduced in the (dA) (dT)-tract

regions (NC 13, closed circles). This did not improve cleavage relative to NC 12, and hence, stability alone cannot confer efficient cleavage. To further elucidate the effect of changes in the non-essential region, the orientation of the (dA) (dT)-tract

-

30-

x CD

20-

Ne

10~~~~~

05

10

15

20

25

ng Topo I

Fig. 2. Titration curves. Hin dllI-Pvu II fragments of the plasmids pNC3-pNC 16 (see 'Materials and Methods') were 32p labelled at the recessed Hin dIll-end. Each of the fragments were incubated with varying amounts of topo I for 15 min at 30°C. Reactions were terminated by addition of SDS to 1%, final concentration, and the samples were processed for sequencing gel electrophoresis (30). Gels were autoradiographed, and the percentage of cleavage was calculated from densitometer scans as the fraction of substrate molecules cleaved sequence specifically in the recognition sequence. NC3, open boxes; NC 12, open circles; NC 13, filled circles; NC 14, crosses; NC 15, filled triangles; NC 16, filled boxes; NC25, open triangles.

Fig. 3. Structure analysis. The Eco RI-Pvu II fragments of pUCNC3-pUCNC 16

were 32P-end labeled at the recessed Eco RI-end and cleaved in the presence of either EDTA * Fe(ll) (OH*) or MPE * Fe(II) (MPE). The cleavage products were separated on 12% polyacrylamide/8 M urea DNA sequencing gels. Densitometer scans of autoradiograms are shown. The sequences are indicated in lettering above the scans in each case.

1238 Nucleic Acids Research, Vol. 19, No. 6 nearest to the cleavage site was inverted (NC3, open boxes; NC25, open triangles). Again, cleavage was clearly reduced relative to the wild type. In conclusion, these results show that the regions flanking the essential part of the binding site are important determinants of cleavage efficiency.

Structure analysis To investigate the possible relation between helical structure and cleavage efficiency, we analyzed two structural parameters on the collection of sequences used in Figure 2. Stable curvature was probed by assessment of the reactivity along the helix towards hydroxyl radical generated with EDTA Fe(g) and hydrogen peroxide (32). The conformational rigidity of poly (dA) * (dT) is manifested by its relative resistance to intercalation (36,37,38) and propagated deformations induced by intercalation (39). To monitor the degree of intercalation, DNA was reacted with MPE* Fe(ll) (31). In principle, this compound is a hydroxyl radical generator linked to an intercalator targeting the reactive species to DNA. The reactivity pattern will therefore reflect the variation in intercalation along the DNA. Note that the two methods employ the same reactive species, the hydroxyl radical. However, differential targeting to DNA allows detection of distinct structural features. In Fig. 3, the sequences in the NC3-NC16 series were subjected to cleavage with either agent and electrophoresed through a DNA sequencing gel. Densitometer scans of the resulting autoradiograms are shown. The cleavage pattern obtained with the wild type sequence (NC14) in the presence of EDTA * Fe(II) shows that the reactivity in the core region is similar to that of mixed sequence vector DNA. This indicates that the core is maintained in a standard B conformation. Both (dA) (dT)-tracts show a characteristic pattern of smoothly decreasing reactivity in the 5' to 3' direction. This anomaly is unique to (dA) (dT)-tracts and diagnostic of stable curvature (31). Thus, a core region with normal conformation seems to be placed between two bends in the high affinity binding

site. The MPEFe(ll) cleavage pattern on NC14 shows reduced reactivity in both (dA) (dT)-tracts and normal reactivity in the core. This indicates that the rigidity characteristic of the (dA) (dT)-tracts is not propagated into the core. Substitution of alternating AT for the homopolymeric tracts normalizes the reactivity in these regions (NC12, NC 15, and NC 16), indicating that both curvature and rigidity have been eliminated. This is also true for NC 13, where the (dA) * (dT)-tracts are replaced by alternating AC. The pattern obtained with NC3 in the presence of EDTA * Fe(II) exhibits the anticipated bend-specific reactivity in the upstream (dA) (dT)-tract, while the substituted downstream tract shows little if any sign of curvature (compare NC3 to NC14 and NC 15). This region apparently behave similarly to the 5'-TnAn-3' family of sequences with respect to curvature (40,41). Nevertheless, the pattern generated with MPE Fe(II) shows that the rigid structure is maintained in the downstream tract of NC3. 'This implies that rigidity and curvature can be uncoupled. Taken together with the results presented in Fig. 2, these data show that (i) the 5'-AGACTTAGA-3' core region alone is insufficient to achieve efficient cleavage, since both NC12 and NC13 are poor substrates, (ii) there is no requirement for (dA) (dT)-tract specific structure in the upstream region, as NC15 is an efficient substrate, (iii) the structure of the downstream tract is a crucial determinant of cleavage efficiency (compare NC 14 and NC 16), and (iv) stable curvature, not rigidity, is the important structural property, because NC3, where the downstream tract is stiff and almost straight, is a poor substrate. Consistently, the sequence NC25, containing the non-curved 5'-T5A5-3'-tract (40), is an inefficient substrate.

-

30-

x

Analysis of binding site subdomain interrelations The effect of small insertions between the curve and the essential region was examined in cleavage titration experiments (Fig. 4). A three base pair increment of the distance virtually eliminates cleavage, irrespective of whether the upstream region is homopolymeric (dA)* (dT) (NC21, filled circles) or alternating AT (NC 18, open boxes) (mutants are listed in Fig. 1). A similar result is obtained with an eleven base pair distance increment (NC20, open circles) or complete removal of the curved segment by substitution with vector DNA (NC22, filled boxes). These findings emphasize the importance of the non-essential region as a determinant of cleavage efficiency and may suggest that

w20-

10

os

1.0 1.5 ng Topo I

20

25

Fig. 4. Insertions in the non-essential region. Topo I cleavage titration experiments, conducted as in Fig. 2, with the single-end labeled Hin dIII-Pvu II fragments of pNC14 and pNC18-pNC22. NC14, crosses; NC18, open boxes; NC20, open circles; NC21, filled circles; NC22, filled boxes.

10

20

30

ng Topo I Fig. 5. Curvature stimulate cleavage at a foreign site. Cleavage titration experiments with the single-end labeled hin dIII-Pvu II fragments of pNC27 (filled circles) and pNC28 (open circles). The vertical scale is in arbitrary units.

Nucleic Acids Research, Vol. 19, No. 6 1239 optimal activity requires the curve to be in correct distance and rotational orientation relative to the essential region. The notion that curvature in the non-essential region acts to stimulate cleavage holds the prediction that an unrelated site should become activated when juxtaposed to an A-tract. To test this, we chose the sequence NC27 (Fig. 1). The consensus region of this sequence reads 5 '-CGGT^G-3' (A denotes the position of

3 4

1 2

_

*_ 9 B

:

A

1* .*

_* T

A

Z A

?

A

2

G

AG u.

A

t

B

C

5 -CpTpTp

cleavage) and thus clearly differs from the 5'-ACTT^A-3' motif present in the other substrates studied here. NC27 is a very poor substrate (30-50 fold less efficient than NC 14, our unpublished observation), and therefore a high enzyme to DNA ratio is required to obtain detectable levels of cleavage (Fig. 5, closed symbols). Insertion of a curved segment two base pairs downstream from the cleavage site (NC28) leads to a 5 fold enhancement of cleavage (Fig. 5, open symbols). This result clearly indicates that local curvature can stimulate cleavage at a foreign essential region. However, NC28 is still a much less efficient substrate than NC14. Therefore, the curvature-dependent activation is a relative phenomenon, while the basal cleavage level seems to depend on properties of the essential region. In this context it is noteworthy that the consensus region of NC27/28 conforms less well to the structural consensus (10,11) than that of NC14 (our unpublished result).

OH-ApGpApApApApApTpTpTpTpTp-33

Fig. 6. Methylation and ethylation interference. The Eco RI-Pvu fragment of pUCNCl4, 32P-labeled at the recessed Eco RI-end, was premodified with dimethyl sulfate or ethyl nitrosourea according to (Siebenlist & Gilbert, 1980) and cleaved with topo I. The labeled subfragment generated by topo I cleavage as well as the intact scissile strand from substrate surviving cleavage were isolated by denaturing gel electrophoresis (8). Modifications were converted to strand breaks according to (33). The resulting mixture of fragments was analyzed on a 12% polyacrylamide/8 M urea DNA sequencing gel (A), methylated DNA; (B), ethylated DNA. Lanes 1 and 3, intact DNA control; Lanes 2 and 4, DNA cleaved by topo I. (C) Schematic representation of interference results. The scissile strand is shown in its cleaved state. The shaded ball symbolizes topo I covalently attached to T(-1) via a phosphodiester bridge. Nucleotide coordinates refer to the cleavage site. Filled triangles, pointing up: positive methylation interference. Open triangles, pointing up: positive ethylation interference. Open triangles, pointing down: negative ethylation interference.

Modification interference analysis The curvature dependence in topo I-DNA interaction implies that the enzyme somehow is able to contact the non-essential region. Previous studies provided no evidence for major groove contacts (8). The possibility that topo I could form contacts to the minor groove was therefore tested using methylation interference analysis (33). A sporadically methylated fragment containing NC14 was single-end labeled in the scissile strand and cleaved with topo I. The products were denatured and separated electrophoretically (8). The subfragment downstream to the cleavage was isolated and analyzed on a DNA sequencing gel after cleavage at methylated positions in NaOH (Fig. 6A, lane 2). The intensities of the bands corresponding to nucleotide coordinates 1, 2, and 4 (as defined in Fig. 6C) are increased relative to the control DNA not cleaved by topo I (lane 1; interferences are marked by asterisks). Since the substrate was saturated with topo I in the cleavage reaction, this cannot be explained by a methylation induced increase in affinity. In a topo I-DNA equilibrium system, modifications inhibitory to forward processes, i.e. binding and cleavage, will reduce band intensities. Modifications inhibitory to reverse processes, such as religation, will lead to accumulation of cleaved intermediates and thereby increase band intensities in the interference assay. We therefore favour the explanation that methylations of adenine N3 in the minor groove immediately downstream to the cleavage site are inhibitory to religation. Methylations further downstream have no discernible effect on binding and cleavage. In the absence of base contacts, we sought for contacts to the exterior of the helix. Phosphate contacts were defined by ethylation interference analysis (33). The experimental procedure was similar to that in panel A, except that DNA was ethylated. Fig. 6B, lane 4 displays the enzyme cleaved DNA. The bands corresponding to positions 1, 2, 3, and 4 appear enhanced relative to the control in lane 3, indicating that ethylation at these positions impedes religation. Reduced band intensities occur at positions 6, 8, 9, and 10, suggesting that masking of phosphates at these positions is partially inhibitory to binding and/or cleavage. The interference results are summarized in panel C.

DISCUSSION In a continuing effort to understand the DNA binding properties of eukaryotic topo I, we have investigated the role of DNA helix structure in topo I-DNA interaction. Previous results (10,11) showed that the enzyme requires a distinct conformation within

1240 Nucleic Acids Research, Vol. 19, No. 6 the region defined as 'consensus' by homology studies (6,7,42) or 'essential' by functional analysis (8,9). This region is characterized by a unique constellation of twist and roll angles between adjacent base pairs. In particular, a large positive roll (opening the minor groove) is found at the scissile bond (10,11). Mutational elimination of this structural feature abolishes topo I cleavage and catalysis (8,30,43). Accordingly, we find in Fig. 5 of this paper that an essential region fitting the structural consensus poorly can only support low levels of cleavage. Thus, the basal level of cleavage appears to be determined by properties of the essential region. However, our observation that mutations outside the consensus region profoundly influence cleavage shows that the structural consensus model (10,1 1) does not provide an adequate description of topo I-DNA interaction. We find that the helix conformation in the adjacent non-essential moiety of the topo I binding site is an important modulator of cleavage efficiency. Cleavage is significantly potentiated, when the nonessential region is stably curved. The stimulatory effect is observed with different essential regions and may depend on correct positioning of the bend relative to the essential region. Our results do not exclude a hypothetical role of curvature in the upstream part of the binding site. According to published phasing rules (16), the two (dA)-(dT)-tracts in the sequence NC14 do not bend in the same direction. It would therefore be of interest to see, whether further stimulation can be achieved by curving both binding site subdomains in the same direction. The phosphate ethylation interference results suggest that topo I senses the path of the helix by formation of a distinct set of contacts to phosphates in the non-essential region. These contacts are located 6-10 base pairs downstream of the cleavage site. We have recently used a high-resolution footprinting technique to show that topo I protects the essential region and the region, where the phosphate contacts are clustered. In contrast, the DNA segment connecting the two is exposed to nucleolytic attack (J. Q. Svejstrup, unpublished results). The role of curvature may therefore be to provide a steric arrangement that optimizes interactions in both of the contacted regions. In general, topo I prefers supercoiled over relaxed DNA (44,45,46,47). Furthermore, the enzyme exhibits a topologyindependent preference for stably curved DNA (28). That is, topo I acts preferentially on curved DNA, be it stably or dynamically bent. This implies that eukaryotic topo I senses writhe rather than twist. Since writhe is relatively independent of the sign of supercoiling (48), this interpretation is consistent with the fact that eukaryotic topo I relaxes both positively and negatively supercoiled DNA (3,4,5). In contrast, prokaryotic topo I senses twist due to a preference for single stranded DNA. As a consequence, this enzyme relaxes only negative supercoils (49). The effect of topology on eukaryotic topo I action may in part be exerted via the essential binding site subdomain, since the local structure in this region will be affected by superhelical stress. The results reported here suggest that local sensing of writhe may in addition involve the non-essential region. The notion that topo I-DNA interaction is controlled by writhe may be of relevance to the intranuclear localization of this enzyme. The religation inhibitor campthothecin has been used to trap enzymatically active topo I at the sites of action in vivo (27,50,51,52). The results showed that topo I activity coincides with active transcription. Furthermore, photo crosslinking (53) and immunofluorescence studies (54) have shown that topo I is physically present only at transcriptionally active loci. It therefore appears that the intranuclear topo I activity is controlled at the level enzyme distribution, and not by adjustment of the activity

of enzyme bound constitutively throughout the genome. The activities of topo I and RNA polymerase II are quantitatively and temporarily linked during an induced transcriptional wave on the human c-fos gene (55). The two enzymes do not appear to be physically associated (56). Instead, supercoiling generated by advancing polymerases may act to enhance topo I-DNA interaction. According to the twin-loop model for generation of superhelical stress during transcription (57,58,59), DNA is overwound in front of a travelling polymerase, and underwound behind it. If twist and writhe are free to equilibrate in vivo, high values of writhe will accompany transcription. The preference for curved DNA may therefore contribute in targeting the enzyme to sites, where its activity is required.

ACKNOWLEDGEMENTS This work was supported by contract BI-6-0171-DK with EURATOM (CEC, Brussels), The Danish Cancer Society (88-060), The Danish National Agency of Technology (1985-133/001-85.521), and the Aarhus University Bioregulation Center.

REFERENCES 1. 2. 3. 4. 5. 6.

7.

8. 9.

10.

11. 12. 13. 14. 15.

16. 17.

18. 19.

20. 21. 22. 23. 24.

25. 26. 27. 28. 29. 30.

Drew, H. R. and Travers, A. A. (1985) Nucleic Acids Res. 13, 44455-4467. Jordan, S. R. & Pabo, C. 0. (1988) Science 242, 893-899. Vosberg, H.-P. (1985) Curr. Topics Microbiol. Immunol. 114, 19-102. Wang, J. C. (1985) Annu. Rev. Biochem. 54, 665-697. Maxwell, A. and Gellert, M. (1986) Adv. Prot. Chem. 38, 69-107. Edwards, K. A., Halligan, B. D., Davis, J. L., Nivera, N. L., and Liu, L. F. (1982) Nucleic Acids Res. 10, 2565-2576. Been, M. D., Burgess, R. R., and Champoux, J. J. (1984) Nucleic Acids Res. 12, 3097-3114. Stevnsner, T., Mortensen, U. H., Westergaard, O., and Bonven, B. J. (1989) J. Biol. Chem. 264, 10110-10113. Svejstrup, J. Q., Christiansen, K., Andersen, A.H., Lund, R. and Westergaard, 0. (1990). J. Biol. Chem. 265, 12529-12535. Kjeldsen, E., Bendixen, C., Thomsen, B., Christiansen, K., Bonven, B. J., Nielsen, 0. F., and Westergaard, 0. (1991) In Ross W. and Potmesil, M. (eds), DNA Topoisomerases and Cancer. Oxford University Press. In press. Shen, C. C. and Shen, C.-K. J. (1990) J. Mol. Biol. 212, 67-78. Trask, D. K. and Muller, M. T. (1983) Nucleic Acids Res. 11, 2779-2800. Wells, R. D. (1988) J. Biol. Chem. 263, 1095-1098. Trifonov, E. N. (1986) CRC Crit. Rev. Biochem. 19, 89-106. Diekmann, S. (1987) In Eckstein, F. & Lilley, D. (eds), Nucleic Acids and Molecular Biology. Springer, Berlin, pp. 138-156. Crothers, D. M., Haran, T. E., and Nadeau, J. G. (1990). J. Biol. Chem. 265, 7093-7096. Achtberger, C. F. and McAllister, E. C. (1989) J. Biol. Chem. 264, 10451 -10456. Bracco, L., Kotlarz, D., Kolb, A., Diekmann, S., and Buc, H. (1989) EMBO J. 8, 4289-4296. Collis, C. M., Molloy, P. L., Both, G. W., and Drew, H. P. (1989) Nucleic Acids Res. 17, 9447-9468. Goodman, S. D. and Nash, H. A. (1989) Nature 341, 251-254. Snyder, U. K., Thompson, J. F. and Landy, A. (1989) Nature 341, 255-257. Ryder, K., Silver, S., DeLucia, A. L., Fanning, E., and Tegtmeyer, P. (1986) CeUl 44, 719-725. Diekmann, S. and McLaughlin, L. W. (1988) J. Mol. Biol. 202, 823-834. Linial, M. and Schlomai, J. (1987) Proc. Natl. Acad. Sci. USA 84, 8205-8209. Spana, C. and Corces, Y. G. (1990). Genes & Develop. 4, 1505-1515. Bonven, B. J., Gocke, E., and Westergaard, 0. (1985) Cell 41, 541 -551. Ness, P. J., Koller, T., and Thoma, F. (1988) J. Mol. Biol. 200, 127-139. Caserta, M., Amadei, A., DiMauro, E., and Camilloni, G. (1989) Nucleic Acids Res. 17, 8463-8474. Travers, A. A. (1989) Annu. Rev. Biochem. 58, 427-452. Thomsen, B., Mollerup, S., Bonven, B. J., Frank, R., Blocker, H., Nielsen, 0. F., and Westergaard, 0. (1987) EMBO J. 6, 1817- 1823.

Nucleic Acids Research, Vol. 19, No. 6 1241 31. Van Dyke, M., W., Hertzberg, R. P., and Dervan, P. B. (1982) Proc. Natl. Acad. Sci. USA 79, 5470-5474. 32. Burkhoff, A. M. and Tullius, T. D. (1987) Cell 48, 935-943. 33. Siebenlist, U. and Gilbert, W. (1980) Proc. Natl. Acad. Sci. USA 77, 122-126. 34. Orii, H., Tanaka, Y., and Yanagisawa, K. (1989) Nucleic Acids Res. 17, 1395-1408. 35. Arnott, S. and Selsing, E. (1974). J. Mol. Biol. 88, 1376-1380. 36. Bresloff, J. L. and Crothers, D. M. (1981) Biochemistry 20, 3547-3553. 37. Wilson, D. W., Wang, Y.-H., Krishnamoorthy, C. R., and Smith, J. C. (1985) Biochemistry 24, 3991-3999. 38. Fox, K. R. & Waring, M. J. (1987) Nucleic Acids Res. 15, 491-507. 39. Huang, Y., Rehfuss, R. P., La Plante, S. R., Boudreau, E., Borer, P. N., and Lane, M. J. (1988) Nucleic Acids Res. 16, 11125-11139. 40. Hagerman, P. J. (1986) Nature 321, 449-450. 41. Burkhoff, A. M. and Tullius, T. D. (1988) Nature 331, 455-457. 42. Perez-Stable, C., Shen, C. C., and Shen, C.-K. J. (1988) Nucleic Acids Res. 16, 7975-7993. 43. Busk, H., Thomsen, B., Bonven, B. J., Kjeldsen, E., Nielsen, 0. F., and Westergaard, 0. (1987) Nature 327, 638-640. 44. Muller, M. T. (1985) Biochim. Biophys. Acta 824, 263 -267. 45. Camilloni, G., DiMartino, E., Caserta, E., and DiMauro, E. (1988) Nucleic Acids Res. 16, 7071-7085. 46. Camilloni, G., DiMartino, E., DiMauro, E., and Caserta, M. (1989) Proc. Natl. Acad. Sci. USA 86, 3080-3084. 47. Caserta, M., Amadei, A., Camilloni, G., and DiMauro, E. (1990). Biochemistry 29, 8152-8157. 48. Travers, A. A. (1990). Cell 60, 177-180. 49. Kirkegaard, K. and Wang, J. C. (1985) J. Mol. Biol. 260, 625-637. 50. Gilmour, D. S. and Elgin, S. C. R. (1987) Mol. Cell. Biol. 7, 141-148. 51. Stewart, A. F. and Schutz, G. (1987) Cell 50, 1109-1117. 52. Zhang, H., Wang, J. C., and Liu, L. F. (1988) Proc. Natl. Acad. Sci. USA 85, 1060-1064. 53. Gilmour, D. S., Pflugfelder, G., Wang, J. C., and Lis, J. T. (1986) Cell 44, 401-407. 54. Fleischmann, G., Pflugfelder, G., Steiner, E. K., Javaherian, K., Howard, G. C., Wang, J. C., and Elgin, S. C. R. (1984) Proc. Natl. Acad. Sci. USA 81, 6958-6962. 55. Stewart, A. F., Herrera, E. H., and Nordheim, A. (1990) Cell 60, 141 - 149. 56. Rose, K. M., Szopa, J., Han, F.-S., Cheng, Y. C., Richter, A., and Scheer, U. (1988) Chromosoma %, 411-416. 57. Liu, L. F. and Wang, J. C. (1987) Proc. Natl. Acad. Sci. USA 84, 7024-7027. 58. Wu, J.-Y., Shyy, S., Wang, J. C., and Liu, L. F. (1988) Cell 53,433-440. 59. Tsao, Y.-P., Wu, H.-Y., and Liu, L. F. (1989) Cell 56, 111-118.