Biochemical and Biophysical Analyses of Recombinant Forms of Human Topoisomerase I*

Amino acid sequence comparisons of human topoi- somerase I (Topo I) with seven other cellular Topo I enzymes reveal that the enzyme can be divided into four major domains: the unconserved NH 2 -terminal domain (24 kDa), the conserved core domain (54 kDa), a poorly conserved linker region (5 kDa), and the highly conserved COOH-terminal domain (8 kDa), which contains the active site tyrosine. To investigate this predicted domain organization, recombinant baculoviruses were engineered to express the 91-kDa full-length enzyme, a 70-kDa NH 2 -terminally truncated enzyme that is miss- ing the first 174 residues, and a 58-kDa NH 2 - and COOH-terminally truncated core fragment encompassing resi- dues 175–659. The specific activity of the full-length and Topo70 enzymes are indistinguishable from the native human Topo I purified from HeLa cells. Each protein is inhibited by camptothecin, topotecan, and 9-amino- camptothecin, but not by ATP. Activity is stimulated by Mg 2 (cid:49) , Ba 2 (cid:49) , Ca 2 (cid:49) , Mn 2 (cid:49) , spermine, and spermidine. The magnitude of the stimulatory effect of Mg 2 (cid:49) is inversely proportional to the salt concentration. Furthermore, at KCl concentrations of 300 m M or greater, the addition of Mg 2 (cid:49) is inhibitory. The effects of Mg 2 (cid:49) and the polycations spermine and spermidine are partially additive, an indication that the stimulatory mechanisms of the two substances are different. Activity was strongly inhibited or abolished by Ni 2 (cid:49) , Zn 2 (cid:49) , Cu 2 (cid:49) , Cd 2 (cid:49) , and Co 2 (cid:49) . An examination of the hydrodynamic properties of full- length Topo I, Topo70, and Topo58 demonstrates that the core, linker, and COOH-terminal domains fold into a globular structure, while the NH 2 -terminal domain is highly extended. A comparison of the circular dichroism spectra of full-length Topo I and Topo70 demonstrates that residues 1–174 ( (cid:59) 21 kDa) of Topo I are largely if not completely unfolded. This observation is consistent with the fact that the NH 2 -terminal domain is dispensable for activity. Recombinant baculoviruses were generated by co-transfecting Spodoptera frugiperda (Sf9) cells with linearized wild type Auto-graphica californica multiple nucleocapsid nuclear polyhedrosis virus DNA (Invitrogen) together with transfer vector DNAs, and plaque purified according to standard procedures provided by Invitrogen. To confirm that the recombinant viruses were expressing the appropriate protein, infected cells were analyzed by SDS-polyacrylamide gel elec- trophoresis (SDS-PAGE) and immunoblotting. Sf9 cells were maintained in Inc.) supplemented with 10% fetal yeastolate g/liter), lactalbumin hydrolysate g/liter), units/ml of penicillin, g/ml of streptomycin, of nystatin. Cells were cultured in and and at a in an O The 1-liter flasks microcarrier (Bellco) that were to the air-liquid

Eukaryotic topoisomerase I (Topo I) 1 is capable of relaxing both negatively and positively supercoiled DNA. The enzyme catalyzes changes in the superhelical state of duplex DNA by transiently breaking a single strand, allowing for unwinding of positively supercoiled DNA or rewinding of negatively supercoiled DNA (reviewed in Ref. 1). No metal cation or energy cofactor is required for Topo I activity, although Mg 2ϩ and Ca 2ϩ , as well as the polycation spermidine, have been shown to stimulate activity (2)(3)(4)(5). Phosphodiester bond energy is preserved during the nicking-closing cycle by the formation of a phosphotyrosine bond between the active-site tyrosine and the 3Ј-end of the broken strand (6 -8). This covalent intermediate can be trapped by denaturing the enzyme during catalysis with either SDS or alkali (9 -11). Sequence analyses of a large number of SDS-induced breakage sites indicated that the cellular Topo I enzymes will cleave at specific sequences (9,12,13), but there is only limited sequence similarity between such sites (9,(12)(13)(14)(15)(16)(17). The SDS-induced cleavage at many breakage sites is enhanced by camptothecin, a plant alkaloid that inhibits the cellular enzymes by reversibly binding to the covalent Topo I-DNA intermediate in a manner that slows the the religation step of catalysis (17)(18)(19)(20)(21)(22).
The human Topo I is composed of 765 residues with a predicted molecular mass of 91 kDa. Sequence comparisons of cellular eukaryotic Topo I proteins demonstrate that the human Topo I can be divided into four domains (Fig. 1). 2 Residues Met 1 -Lys 197 (24 kDa) comprise the unconserved NH 2 -terminal domain, which is highly charged (Asp ϩ Glu ϭ 27%; His ϩ Lys ϩ Arg ϭ 68%) and contains four putative nuclear localization signals (24). Residues Glu 198 -Ile 651 (54 kDa) form the conserved core domain, which is followed by a short positively charged linker domain of unconserved residues Asp 652 -Glu 696 (5 kDa). Finally, residues Gln 697 -Phe 765 (8 kDa) make up the highly conserved COOH-terminal domain, which contains the active site tyrosine at position 723 (25,26). 2 Previous reports have demonstrated that the NH 2 -terminal domain is sensitive to proteolyis (3) and that residues 1-230 can be removed with little if any consequence for Topo I activity (25,27). In contrast, a 5-amino acid deletion from the COOH terminus abolishes activity. 3 We have used the baculovirus-infected insect cell system to overproduce full-length, as well as NH 2 -and COOH-terminal deletions of human topoisomerase I. The purified recombinant Topo I is by all biochemical criteria tested identical to the native enzyme purified from human cells. Furthermore, we find the activities of the full-length protein and an aminoterminally deleted enzyme (Topo70, missing residues  are identical in every respect. The two proteins are inhibited by camptothecin and its derivatives, are stimulated by Mg 2ϩ with a magnitude that is inversely proportional to the salt concentration (up to 250 mM KCl), and are not affected by ATP. By comparing the circular dichroism spectra and hydrodynamic properties of Topo70 and the full-length protein, we demonstrate that the amino-terminal domain is largely if not completely unfolded, while the remainder of the molecule is relatively globular. In the accompanying paper (40), we use limited proteolysis to further examine the domain structure of human Topo I.

Plasmids
All plasmids were constructed by common subcloning techniques and propagated in either the Sure TM (Stratagene) or TOP10FЈ (Invitrogen) strains of Escherichia coli. Nucleotides of the cDNA clone encoding human topoisomerase I are numbered according to D'Arpa et al. (25).
pAc-Topo I and pAc-Topo I(Y/F)-The wild type and active site mutant Y/F human Topo I cDNA sequences from the plasmids pKM16 and pKM18, respectively (26), were inserted into the blunt-ended NheI cloning site of pBlueBac (Invitrogen) to generate pAc-Topo I and pAc-Topo I(Y/F). Restriction analyses were carried out to confirm proper orientation of the Topo I reading frames with respect to the polyhedrin gene promoter.
pTopo70-start-The polymerase chain reaction was used to amplify a segment of the human Topo I cDNA from position 731 to 1112. The 5Ј end of the amplification product contained an XbaI restriction site followed by an initiating methionine codon immediately adjacent to the codon for residue Lys 175 . The polymerase chain reaction product was digested with XbaI (which cuts 18 base pairs upstream from the initiating ATG) and NdeI (which cuts at position 901 in the Topo I cDNA) to generate a 194-base pair fragment that was subcloned into the plasmid pET11a (Novagen) to generate pTopo70-start. The subcloned sequences that had been subjected to polymerase chain reaction were sequenced to confirm proper construction of the initiation signal and to ensure that no other mutations were present.
pET11a-Topo70 -The plasmid pTopo70-start was linearized with HindIII, which cuts the pET11a vector backbone at a site just downstream of the Topo I sequence. The HindIII 5Ј overhangs were filled in with T4 DNA polymerase and all four dNTPs. This DNA was digested with NdeI, which cuts at position 901 in the Topo I coding sequences. The resulting fragment was ligated with a 2.7-kilobase pair NdeI 901 -NotI (blunt-ended) fragment, which contains the NdeI 901 -EcoRI 3641 Topo I sequences followed by multiple cloning site sequences from EcoRI to NotI of the plasmid pSKϩ (Stratagene). The resulting plasmid, called pET11a-Topo70, carries the coding sequence for a ϳ70-kDa NH 2terminally truncated form of Topo I called Topo70 that starts at a methionine immediately 5Ј of Lys 175 and ends at the natural Topo I COOH terminus.
pAc-Topo70 -The baculovirus transfer vector pBlueBac (Invitrogen) has a unique NheI cloning site, located immediately downstream of the polyhedrin gene promoter, into which we inserted an XbaI-XbaI fragment from pET11a-Topo70, which carries the entire Topo70 coding sequence. Diagnostic restriction digests were performed to confirm proper orientation of the Topo70 reading frame with respect to the polyhedrin gene promoter. The resulting plasmid is named pAc-Topo70.
pAc-Topo70(Y/F)-The NdeI 901 -AvrII 2386 sequences from pKM18 (26) were exchanged for the NdeI 901 -AvrII 2386 sequences of pAc-Topo70, to generate pAc-Topo70(Y/F). This plasmid is identical to pAc-Topo70 except that it carries a mutation that converts the active site tyrosine 723 codon to a phenylalanine codon. pAc-Topo58 -A self-complementary NheI-stop oligonucleotide (5Ј-CTAGCCTAGGCCTAGG-3Ј) was inserted at the unique NheI 2183 restriction site of the plasmid pAc-Topo70 to generate pAc-Topo58. The NheI-stop oligonucleotide introduces a diagnostic AvrII restriction site in place of the NheI site and positions a stop codon immediately 3Ј of the codon for residue Ala 659 of the Topo I cDNA.

Plasmid Relaxation Assays
Unless stated otherwise, stocks of protein were serially diluted 2-fold in standard buffer (150 mM KCl, 10 mM Tris-hydrochloride, pH 7.5, 1 mM DTT, 1 mM EDTA, 0.1 mg/ml BSA), and reactions were initiated by the addition of 5 l of the diluted enzyme to 15 l of the appropriate buffer containing 0.5 g of supercoiled pKSIIϩ plasmid substrate (Stratagene). The final reaction conditions are indicated in the table and figure legends. The reactions were incubated at 37°C for 10 min and terminated with 5 l of a stop mix containing 2.5% SDS, 15% Ficoll, 0.03% bromphenol blue, 0.03% xylene cyanol, and 25 mM EDTA. The products were fractionated by 0.8% agarose gel electrophoresis and visualzed by ethidium bromide staining. The inhibitors camptothecin (Sigma), topotecan (NCI, National Institutes of Health), and 9-aminocamptothecin (NCI) were dissolved in Me 2 SO and stored at Ϫ20°C.

Isolation of Recombinant Baculoviruses and Large Scale Sf9
Infection-Recombinant baculoviruses were generated by co-transfecting Spodoptera frugiperda (Sf9) cells with linearized wild type Autographica californica multiple nucleocapsid nuclear polyhedrosis virus DNA (Invitrogen) together with transfer vector DNAs, and plaque purified according to standard procedures provided by Invitrogen. To confirm that the recombinant viruses were expressing the appropriate protein, infected cells were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting. Sf9 cells were maintained in TC100 medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum, yeastolate (3.3 g/liter), lactalbumin hydrolysate (3.3 g/liter), 100 units/ml of penicillin, 100 g/ml of streptomycin, and 100 units/ml of nystatin. Cells were cultured in 100 ml and 1-liter spinner flasks (Bellco) and stirred at a rate of 60 rpm in an atmosphere of 50% O 2 , 50% air at 27°C. The 1-liter flasks were assembled with microcarrier impellers (Bellco) that were adjusted to break the airliquid interface. 4 The maximum volume of medium used in the spinner flasks was 80 and 500 ml for the 100-ml and 1-liter flasks, respectively. Typically, cells were seeded at 0.5-0.8 ϫ 10 6 cells/ml and cultured until the density reached 3-3.5 ϫ 10 6 cells per ml (ϳ3 days). To infect the cells, they were first pelleted by centifugation at 600 ϫ g at room temperature for 5 min and then resuspended in fresh medium at a density of 1 ϫ 10 7 cells/ml. The appropriate volume of virus stock was added to ensure a multiplicity of infection of approximately 10 plaqueforming units/cell. After stirring for 1 h at room temperature, fresh medium was added to the cells such that the final density was 3 ϫ 10 6 cells/ml.

Protein Purification
Purification of Topo I from Baculovirus-infected Insect Cells-All purification steps except those involving room temperature high pressure liquid chromatography (Mono Q, Mono S, and POROS columns) were carried out at 4°C. At 48 h postinfection, approximately 3 ϫ 10 9 Sf9 cells were harvested by centrifugation for 5 min at 400 ϫ g. The cells were resuspended in 1 liter of ice-cold phosphate-buffered saline and centrifuged for 5 min at 400 ϫ g. This wash procedure was repeated twice with 250 ml of phosphate-buffered saline. The washed cells were resuspended by vigorous shaking in 40 ml of lysis buffer (50 mM KCl, 10 mM Tris-hydrochloride, pH 7.5, 2 mM MgCl 2 , 1% Triton X-100, 15 mM DTT, 0.15 mg/ml phenylmethylsulfonyl fluoride, 0.05 mg/ml aprotinin). 4 S. G. Graber, personal communication. The nuclei were washed twice in 80 ml of lysis buffer minus Triton X-100 and resuspended in 40 ml of resuspension buffer (50 mM KCl, 10 mM Tris-hydrochloride, pH 7.5, 2 mM MgCl 2 , 25 mM DTT, 0.4 mg/ml phenylmethylsulfonyl fluoride, 0.12 mg/ml aprotinin). The nuclei were adjusted to 10 mM EDTA and then lysed by the addition of 50 ml of 2 ϫ nuclear extraction buffer (2 M NaCl, 80 mM Tris-hydrochloride, pH 7.5, 20% glycerol, 2 mM EDTA). The nuclear extract was stirred for 5 min at ϳ200 rpm. With continued stirring, 50 ml of polyethylene glycol (PEG) buffer (18% PEG 8000, 1 M NaCl, 10% glycerol) was added dropwise in order to precipitate the DNA (30). After stirring for 40 min, the PEG precipitate was pelleted by centrifugation at 10,000 ϫ g for 10 min. The resulting PEG supernatant was dialyzed against 4 liters of potassium phosphate buffer (PPB) (250 mM KPO 4 , pH 7.4, 1 mM DTT, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride). The dialyzed PEG supernatant was clarified by centrifugation at 10,000 ϫ g for 10 min and passed over a 10-ml bed volume of phosphocellulose (Whatman P11) equilibrated with PPB. The P11 column was washed with 50 ml of PPB and then step-eluted with 30 ml of PPB containing 700 mM KPO 4 , pH 7.4. The P11 eluate was dialyzed overnight against 2 liters of PPB and passed over a 6-ml bed volume of phenyl-Sepharose (PS) (Pharmacia) that was equilibrated with PPB. The Topo I flowed through the PS column and was collected together with a 10-ml wash. An equal volume of water was then added to the PS flow-through, which was passed over a Mono Q (5H/R; Pharmacia Biotech Inc.) column that had been equilibrated with K100 buffer (100 mM KPO 4 , pH 7.4, 1 mM DTT, and 1 mM EDTA). The Topo I flowed through the Mono Q column and was loaded onto a Mono S column (5H/R) that was equilibrated with K100. The Mono S column was eluted with a 25-ml 50 -200 mM KPO 4 , pH 7.4, gradient. Topo I eluted as a single major peak at ϳ150 mM KPO 4 . The peak Mono S fractions were pooled, concentrated with an Amicon ultrafiltration cell, dialyzed into storage buffer (50% glycerol, 10 mM Tris-hydrochloride, pH 7.5, 1 mM DTT, 1 mM EDTA), and stored at Ϫ20°C. The high level expression achieved in the baculovirus-infected insect cell system yielded ϳ10 mg of protein from 3 ϫ 10 9 cells.
Purification of Recombinant Topo70 and Topo58 -The purification of recombinant Topo70 was as described above for the full-length Topo I. The initial steps in the purification of recombinant Topo58, up to and including elution of the P11 column, were as described for the fulllength enzyme. The P11 eluate was then dialyzed against K100, filtered through a 0.45-m syringe filter, and loaded onto a POROS SP20 (4.6/100) column (PerSeptive Biosystems) that was equilibrated with K100. The POROS SP20 column was eluted with a linear 100 -500 mM KPO 4 , pH 7.4, gradient, and Topo58 was found to elute at 300 mM KPO 4 . The peak fractions were pooled, diluted with an equal volume of water, and loaded onto a Mono S column that was eluted with a 100 -300 mM KPO 4 , pH 7.4, gradient. Approximately 70% of the Topo58 eluted from the Mono S at 200 mM KPO 4 , while the remainder eluted at 250 mM KPO 4 . The 200 mM peak fractions were pooled, diluted with an equal volume of water, and passed over a Mono Q column that was equilibrated with K100. The flow-through fractions were concentrated with an Amicon ultrafiltration cell, dialyzed into storage buffer and stored at Ϫ20°C.
Purification of HeLa Topo I-The starting material for purification of native human Topo I was 3 ϫ 10 9 HeLa S3 cells that were doubling every 20 -24 h in S-modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum, 100 units/ml of penicillin, 100 g/ml of streptomycin, and 50 units/ml of nystatin. The initial steps in purification of HeLa Topo I, up to the point of isolating the clarified dialyzed PEG supernatant, were identical to that described above for the recombinant enzyme. The dialyzed PEG supernatant was diluted with an equal volume of water, filtered through a 0.45-m filter, and loaded onto a POROS SP20 (4.6/100) column that was equilibrated with cation exchange buffer (7 mM MES, 7 mM HEPES, 7 mM sodium acetate, pH 7.5) plus 100 mM NaCl. The SP20 column was eluted with a 30-ml linear salt gradient from 100 to 800 mM NaCl. Plasmid relaxation assays were performed to identify the peak Topo I fractions, which were found to elute at ϳ700 mM NaCl. The Topo I fractions were pooled, dialyzed against 2 liters of PPB, and passed over a 2-ml bed volume of PS that was equilibrated with PPB. The PS flow-through was diluted with an equal volume of water and loaded onto a Mono S column (5H/R) that was equilibrated with K100. The Mono S column was eluted as described for the recombinant enzyme, and the peak fractions were dialyzed into storage buffer and stored at Ϫ20°C.
Purification of f-Topo70 and f-Topo75 Fragments of Recombinant Topo I-During a 3-week room temperature incubation of the clarified PEG supernatant from a preparation of baculovirus-expressed Topo I, all of the full-length enzyme was proteolytically cleaved into a roughly equal mixture of 70-and 75-kDa forms of Topo I (designated f-Topo70 and f-Topo75, respectively). At this time, the PEG supernatant was clarified by centrifugation at 10,000 ϫ g for 10 min and dialyzed against 4 liters of PPB. The dialyzed PEG supernatant was clarified by centrifugation at 10,000 ϫ g for 10 min and loaded onto a POROS SP20 (4.6/100) column that was equilibrated with cation exchange buffer plus 400 mM NaCl. The proteins were eluted with a linear 50-ml NaCl gradient from 400 mM to 1 M. The f-Topo70 and f-Topo75 co-eluted at 500 mM NaCl. The peak SP20 fractions were pooled, dialyzed against 2 liters of K100, and loaded onto a Mono S column (5H/R) that was equilibrated with K100. The column was eluted with a 20-ml 100 -200 mM KPO 4 , pH7.4, gradient. The f-Topo75 eluted at 140 mM KPO 4 , while the f-Topo70 eluted at 180 mM KPO 4 . The peak fractions of each protein were pooled, dialyzed against Mono Q Buffer (100 mM NaCl, 1 mM DTT, 1 mM EDTA, 20 mM Tris-hydrochloride, pH 7.5), and passed through a Mono Q column that was equilibrated with Mono Q buffer. Finally, the purified f-Topo70 and f-Topo75 fractions were dialyzed into storage buffer and maintained at Ϫ20°C.

Amino-terminal Sequencing
The f-Topo75 and f-Topo70 proteins were fractionated by SDS-PAGE and then transferred to Immobilon-P (Millipore) membranes in 10 mM CAPS, pH 10. The proteins were visualized by staining the membranes with Coomassie Blue. The appropriate bands were excised and sent to DNA Express (Fort Collins, CO) for amino-terminal sequencing.

Gel Filtration and Glycerol Gradient Sedimentation
Fast protein liquid chromatography (Pharmacia) gel filtration analyses were performed at 25°C with a flow rate of 0.75 ml/min using a Superose 12 (Pharmacia) column that was equilibrated with gel filtration buffer (200 mM KPO 4 , pH 7.4, 1 mM DTT, 1 mM EDTA). Purified protein samples of 10 -100 g were diluted into gel filtration buffer and injected in a total volume of 200 l. Elution profiles of both Topo I constructs and molecular weight markers were monitored by UV absorbance at 280 nm and analyzed by SDS-PAGE. Glycerol gradient sedimentation analyses were performed by layering 250-l samples of protein in 200 mM KPO 4 , pH 7.4, onto a 3.8-ml linear 10 -30% glycerol gradient containing gel filtration buffer. The gradients were centrifuged at 50,000 rpm in an SW60 rotor (Beckman) for 16 h at 25°C. Fractions of 300 l were collected from the bottom of the gradient tube through a small puncture and analyzed by SDS-PAGE and silver staining. The experimental gradients each included carbonic anhydrase as an internal standard. Parallel gradients were used to determine the sedimentation profiles of the marker proteins.

Circular Dichroism
Proteins were extensively dialyzed into 10 mM KPO 4 , pH 7.4. Exact molar concentrations of each protein were calculated from the A 280 measurements of fully denatured protein in 6 M guanidine hydrochloride using molar extinction coefficients that were predicted by the Genetics Computing Group (GCG) software. CD spectra were obtained at room temperature, using a Jasco 3000 spectrapolarimeter with a 0.1-cm path length cell. The molar ellipticity spectrum for each sample was taken as the average of 8 -12 individual scans. To eliminate the dichroism that is contributed by the sample buffer, the molar ellipticity spectrum for the dialysis buffer was subtracted from the molar ellipticity spectrum for each sample. The molar ellipticity values were normalized for the concentration of amide bonds in each sample and then converted to ␦⑀ values (31).

SDS-Polyacrylamide Gel Electrophoresis and Autoradiography
SDS-PAGE was performed according to Laemmli (32). Proteins were visualized by Coomassie Blue or silver staining (33). Autoradiography was performed by exposing dried gels to Kodak XAR film.

Expression and Purification of Recombinant Forms of Hu-
man Topo I-With the long term goal of investigating the domain structure of human Topo I, we developed procedures to obtain large quantities of purified enzyme. This was achieved by the generation of recombinant baculoviruses that express wild type and active-site mutant (Y723F) versions of the fulllength human Topo I and a 70-kDa protein (Topo70) which is missing the first 174 NH 2 -terminal amino acids (Fig. 2A). The recombinant Topo70 retains at least one (residues 192-198) of the four potential nuclear localization signals that reside in the NH 2 -terminal domain (24). We also generated a recombinant baculovirus that expresses a NH 2 -and COOH-terminally truncated 58-kDa protein (Topo58), which encompasses the conserved core domain (Fig. 2A). The recombinant proteins have been purified to apparent homogeneity (Fig. 2B). The wild type full-length Topo I and Topo70 proteins have the same specific activity as the native HeLa enzyme (Table I, line 1). The active site Y723F mutant protein and Topo58 are at least 5000-fold less active than the wild type enzyme. The very low level of activity present in the mutant preparations is due to trace amounts of contaminating insect cell Topo I.
The recombinant full-length Topo I and HeLa cell Topo I co-migrate in SDS-PAGE analyses (Fig. 2B, compare lanes 2  and 3). Although the full-length proteins are predicted to be 91-kDa (25), they migrate anomalously in SDS gels with an apparent molecular mass of ϳ100 kDa. In contrast, the recombinant Topo70 migrates appropriately with an apparent molecular mass of ϳ70 kDa (Fig. 2B, lane 6). Since it has previously been observed that negatively charged residues retard the migration of proteins in SDS gels (34), it is likely that the 48 negatively charged residues in the first 174 NH 2 -terminal amino acids are responsible for the slow migration of the fulllength protein.
While designing the purification scheme, we noticed that long term storage of nuclear extracts leads to proteolysis of the full-length enzyme into a fragments of 75 and 70 kDa, designated f-Topo75 and f-Topo70, respectively (Fig. 2B, lanes 4 and  5). Both of these fragments were purified and found to display activity equal to that of the full-length protein (Table I, line 1), indicating that they retain the active site tyrosine, which is very close to the COOH terminus. This observation, together with the fact that amino acids downstream of the active site are essential for activity (27), 3 indicated that f-Topo75 and f-Topo70 represent NH 2 -terminally deleted forms of Topo I. Amino terminal sequencing confirmed that f-Topo75 was missing the first 137 residues while f-Topo70 was missing the first 174 residues.
The Effect of Divalent Metal Cations and Polycations on Topo I Activity-Although not required for Topo I activity, divalent metal cations are known to stimulate the activity as much as 25-fold (2,3). Therefore to further characterize the recombinant full-length Topo I and Topo70, we examined the effect of a variety of divalent cations on the activity of the two proteins. In the standard relaxation assay buffer (which contains 150 mM KCl), Mg 2ϩ , Mn 2ϩ , Ba 2ϩ , and Ca 2ϩ were all found to stimulate Topo I activity approximately 16-fold (Table I, lines 2, 10, 19, 20, and 21). In contrast, Cd 2ϩ , Zn 2ϩ , Co 2ϩ , and Cu 2ϩ completely inhibit Topo I (Table I, lines [22][23][24][25], while Ni 2ϩ was found to inhibit activity 16-fold at 5 mM (Table I, line 26).
To further investigate the stimulatory effect of Mg 2ϩ we thought it would be informative to determine the level of Mg 2ϩ stimulation obtained over a range of salt concentrations (Fig.  3). Initially, we varied the salt concentration in the absence of Mg 2ϩ and found that the optimal KCl concentration was 200 -250 mM, with very little activity detected at KCl concentrations less than 10 mM or greater than 400 mM (Fig. 3A, and data not shown). We then assayed the effect of 10 mM Mg 2ϩ over a range of KCl concentrations from 25 to 350 mM. This revealed that Mg 2ϩ had its largest stimulatory effect (50-fold) at low salt concentrations (25 mM KCl). As the KCl concentration was increased, the stimulatory effect of Mg 2ϩ steadily dropped, and at ϳ250 mM Mg 2ϩ was slightly inhibitory. With higher salt concentrations, 300 and 350 mM, the addition of Mg 2ϩ was 3and 7-fold inhibitory, respectively. The full-length and NH 2terminally truncated Topo70 enzymes behaved identically in their responsiveness to salt and Mg 2ϩ (data not shown). For both forms of Topo I, there was an inverse relationship between the fold-effect of Mg 2ϩ on activity and the KCl concentration. This is best depicted graphically as a logarithmic plot of the ratio of enzyme activity with and without Mg 2ϩ versus the KCl concentration (Fig. 3B).
Like divalent metal cations, polycations such as spermine and spermidine have also been shown to stimulate Topo I activity (5,35). However, previous reports did not describe the effect of a combination of Mg 2ϩ and a polycation, which might shed light on the stimulatory mechanism of each alone. For example, if the Mg 2ϩ and spermidine effects were additive, then this might suggest separate mechanisms for activation. Thus, we examined the ability of spermidine and spermine to stimulate Topo I activity in the presence and absence of 10 mM MgCl 2 (Table I). In the absence of Mg 2ϩ , spermine (1 mM) and spermidine (5 mM) were found to stimulate Topo I activity 8-fold at 150 mM KCl. However, the combination of 10 mM Mg 2ϩ and 1 mM spermidine stimulated the activity 64-fold. This suggests that the effects of polycations and Mg 2ϩ are at least partially additive and that the two substances may stimulate Topo I activity by different mechanisms.
Effect of ATP on Topo I Activity-Topo I does not require ATP or any other energy source for activity. However, it has been reported that physiological concentrations of ATP (ϳ2 mM) can inhibit human Topo I (36). In another report, ATP was shown to inhibit human Topo I only in the presence of 1 mM KPO 4 (37). These conflicting reports prompted us to examine the effect of ATP and KPO 4 on the activity of HeLa Topo I in the presence or absence of 10 mM Mg 2ϩ (Table I). In the absence of Mg 2ϩ , 4 mM ATP had no detectable effect on Topo I activity whether 3 mM KPO 4 was included or not (Table I, lines 6 and 7). In the presence of 10 mM Mg 2ϩ , 4 mM ATP was found to inhibit the activity 2-fold regardless of the presence or absence of 3 mM KPO 4 (Table I, lines 8 and 9). Since ATP was only inhibitory in the presence of Mg 2ϩ , it seemed possible that the inhibitory effect of ATP could be the consequence of its ability to bind Mg 2ϩ , which would effectively lower the concentration of free Mg 2ϩ . To test this possibility, we assayed the activity in the presence of 6 mM Mg 2ϩ , which is the expected concentration of free Mg 2ϩ in a mixture of 10 mM MgCl and 4 mM ATP. Activity in the presence of 6 mM Mg 2ϩ was reduced 2-fold relative to activity in the presence of 10 mM Mg 2ϩ (Table I, compare lines  2 and 10). Similar results were obtained for both recombinant full-length Topo I and Topo70. Thus, we conclude that the inhibitory effect of ATP is not the result of binding to the enzyme but rather is the consequence of the ability of ATP to reduce the free Mg 2ϩ concentration.
The Effect of Camptothecin and Its Derivatives on Topo I Activity-To further characterize the activity of the recombinant full-length Topo I and Topo70, we examined the inhibitory effects of camptothecin and its derivatives, topotecan and 9amino-camptothecin. When camptothecin was included in the reactions at 50 M, activity was inhibited 8 -16-fold in the presence or absence of 10 mM Mg 2ϩ , 1 mM spermine, or 5 mM spermidine (Table I, lines [11][12][13][14]. Topotecan and 9-aminocamptothecin were also found to inhibit relaxation 8 -16-fold in the presence or absence of Mg 2ϩ (Table I, lines [15][16][17][18]. Hydrodynamic Properties of Full-length Topo I and Topo70 -The sensitivity of the NH 2 -terminal one-fourth of Topo I to proteolysis suggested that it might be in an extended conformation, while the remaining three-quarters of the protein, which is more resistant to proteolysis, might be more globular. To test this notion, we subjected full-length Topo I, Topo70, and Topo58 to gel filtration and glycerol gradient sedimentation analyses (Fig. 4). As previously reported (38), human Topo I chromatographed through gel filtration with an apparent molecular mass of ϳ300 kDa (Fig. 4A). In contrast, the same protein sedimented in a glycerol gradient with an apparent molecular mass of ϳ66 kDa (Fig. 4B). The discrepancy in the molecular mass estimates by the two methods suggests that Topo I has an extended shape and thus a larger frictional coefficient than would be expected for a globular protein of 91 kDa. In contrast, the NH 2 -terminally truncated Topo70 was found to have an apparent molecular mass of ϳ66 kDa by sedimentation and ϳ96 kDa by gel filtration, indicating that the NH 2 -terminal region is largely responsible for the asymmetric shape of the protein.
Since full-length Topo I and Topo70 both co-sediment with BSA in a glycerol gradient, we can assume that the sedimentation coefficients for the two proteins are approximately equal to that of BSA (s 20,w 0 ϭ 4.55) (39). Thus, the frictional coefficient (f) for each enzyme can be estimated from the formula f ϭm(1 Under standard conditions (150 mM KCl, 10 mM Tris-hydrochloride, pH 7.5, 1 mM DTT, 1 mM EDTA, 0.1 mg/ml BSA), 1.0 ng of the full-length Topo I is sufficient to fully relax 0.5 g of supercoiled plasmid DNA in 10 min at 37°C in a 20-l reaction volume. This level of activity was assigned the value of unity. All of the other activities are given as the dilution factor required to achieve the same level of activity as one unit of full-length Topo I under standard conditions. The enzymes that were assayed are listed above each column. Substances that were added to standard buffer conditions are listed in the first column under the heading "Additives." The Ͻ0.005 value indicates that no enzyme activity was detected. Blank areas indicate that the assays were not done. The reactions that contained camptothecin, topotecan, or 9-amino-camptothecin also contained 0.5% Me 2 SO in the final assay buffer. This quantity of Me 2 SO (0.5%) was found to have no effect on the activity of full-length Topo I or Topo70 under otherwise standard conditions (data not shown). Ϫ v)/s, (where m is molecular mass, v is the partial specific volume, is the density of the solvent, and s is the sedimentation coefficient (39). The partial specific volumes for full-length Topo I and Topo70 are both ϳ0.74, as predicted from the amino acid content of each protein (39). Thus, taking as equal to 1.0 g/cm 3 (the density of water at 20°C), the frictional coefficients for the full-length protein and Topo70 are calculated to be 8.7 ϫ 10 Ϫ8 and 6.6 ϫ 10 Ϫ8 (g ϫ s Ϫ1 ), respectively. Given these frictional coefficients, the axial ratio (39) of the full-length protein is approximately 10, as compared with a value of 5 for Topo70 (BSA is 6), confirming the elongated shape for full-length Topo I.
Topo58 and Topo70 eluted from a gel filtration column with similar apparent molecular masses of ϳ83 and ϳ96 kDa, respectively. However they sedimented with apparent molecular masses of ϳ59 and ϳ66 kDa. The ϳ30-kDa disparity between molecular mass estimates by the two methods indicates that the shared 58 kDa domain of the two proteins either contains a small highly extended region or is by itself somewhat elongated. Evidence from limited proteolysis studies (see accompanying paper (40)) indicates that the former possibility is more The results are presented graphically as log (molecular mass) versus the ratio of observed elution volume (Ve) to excluded volume (Vo). The calculated apparent molecular masses for full-length Topo I, Topo70, and Topo58 are ϳ300, ϳ96, and ϳ83 kDa, respectively. Panel B, proteins were fractionated by sedimentation through 10 -30% glycerol gradients. Fractions were collected and analyzed by SDS-PAGE and silver staining. Molecular mass standards (Sigma) were ␤-amylase (200 kDa), alcohol dehydrogenase (150 kDa), bovine serum albumin (66 kDa), ovalbumin (45 kDa), and carbonic anyhdrase (29 kDa). The results are presented graphically as molecular mass versus fraction number. The calculated apparent molecular masses for full-length Topo I, Topo70, and Topo58 are ϳ66, ϳ66, and ϳ60 kDa, respectively. likely to be correct. For example limited trypsin digestion of full-length Topo I generates a proteolytically resistant 55-kDa fragment that starts at residue Lys 197 and ends somewhere very close to residue Lys 654 (Fig. 1A in Ref. 40). The sensitivity to proteolysis strongly suggests that the NH 2 -terminal segment of the Topo58 protein is likely extended (residues Lys 174 -Lys 197 ), providing the hydrodynamic feature that leads to the ϳ30-kDa discrepancy.
CD studies of Full-length Topo I and Topo70 -The gel filtration and sedimentation analyses indicated that human Topo I has a highly extended NH 2 terminus, which minimally includes the first 174 residues but probably extends up to residue 197. To further investigate this structural feature, we obtained the far UV CD spectra for both the wild type and active-site mutant (Y723F) versions of full-length Topo I and Topo70 (Fig.  5). Within experimental error, the CD spectra for the wild type and mutant versions of the two proteins are identical (Fig. 5, A  and B). Hence, as expected the active-site mutation Y723F does not appreciably alter the structure of the enzyme as measured in this way. In contrast, the averaged CD spectra of Topo70 and full-length Topo I are different (Fig. 5C). To minimize experimental error in this comparison, we averaged the mutant and wild type spectra obtained for two different preparations of each protein. Topo70 has a significantly stronger polarization in the 190-nm range than the full-length protein. Since polarization at 190 nm is a very reliable predictor of ␣-helical content in proteins (31,41), it is immediately apparent that the Topo70 protein has a greater percentage of ␣-helical content than the full-length protein. To confirm this notion, we compared the percentage secondary structures of the two proteins as predicted by the program VARSLC1 (41). The predicted molecular mass of ␣-helix (ϳ28 kDa) and parallel ␤-sheet (ϳ3 kDa) are very similar for the two proteins. However, the fulllength protein is predicted to contain an extra 10 kDa of unfolded regions, 8 kDa of turns, and 4 kDa of antiparallel ␤-sheet. Hence, while the NH 2 terminus appears to be largely unfolded, it also appears to have a substantial quantity of turns and antiparallel ␤-sheet. The turns and antiparallel ␤-sheet could either be folded into a stable contiguous domain or instead could be a reflection of transiently formed secondary structure in an otherwise random coil (41). To assess which of the explanations is more likely, we analyzed the first 174 residues Topo I with the PEPTIDESTRUCTURE function of the GCG software package (Fig. 5D). This program predicts only very short regions of secondary structure for the NH 2terminal domain. Taking all of the information into account, gel filtration, sedimentation, and CD analyses, we conclude that the amino-terminal 174 resides of Topo I are largely unfolded and are comprised of very little if any extended secondary structure.

DISCUSSION
The Effects of Metal Cations and Polycations-The recombinant human Topo I produced in the baculovirus-infected insect cell system displays the same apparent molecular weight and specific activity as the native enzyme purified from HeLa cells. Consistent with earlier findings (3,25,27), the full-length Topo I and the NH 2 -terminally truncated Topo70 display identical activities. Both enzymes are inhibited by camptothecin, topotecan, and 9-amino-camptothecin but not by ATP. The activities of both are stimulated by Mg 2ϩ , Ba 2ϩ , Ca 2ϩ , and Mn 2ϩ , but are strongly inhibited by Ni 2ϩ , Zn 2ϩ , Cu 2ϩ , Cd 2ϩ , and Co 2ϩ . The stimulatory effect of Mg 2ϩ was found to increase with decreasing salt concentration. Under low salt conditions of (25 mM), the 50-fold stimulatory effect of Mg 2ϩ resulted in a final level of activity that was still 10-fold less than that achieved with the most favorable condition, a combination of 10 mM Mg 2ϩ and 200 mM KCl. This suggests that Mg 2ϩ may stimulate the activity by two mechanisms, one similar to that achieved by monovalent cations alone and another that further stimulates activity in the presence of monovalent cations up to 200 mM KCl. At higher KCl concentrations (Ն 250 mM) Mg 2ϩ was found to be slightly inhibitory, as has been observed for the rat liver and vaccinia Topo I enzymes (4,35).
There are several potential mechanisms whereby a divalent cation such as Mg 2ϩ could effect a large stimulation of Topo I activity. For example Mg 2ϩ is known to effectively shield the negative charge of the phosphate backbone of duplex DNA, which in addition to allowing the two strands to wind tighter (42) also reduces the effective diameter of the double helix (43,44), making it more favorable for two duplexes to lie on top of each other to form a node (45). Since it has been shown that Topo I has a preference for binding to nodes (46,47), it could be envisioned that the presence of Mg 2ϩ -facilitated nodes recruits Topo I to supercoiled DNA, thereby effectively increasing activity. Alternatively, Topo I may simply prefer to relax DNA with a Mg 2ϩ -shielded phosphate backbone. Another possibility is that Mg 2ϩ binds to the enzyme effecting some allosteric activation. However, the fact that Mg 2ϩ does not influence the patterns of limited proteolysis of either the free or DNA-bound protein (see accompanying paper (40)), suggests that Mg 2ϩ does not have any major effect on enzyme structure. For the vaccinia Topo I, Mg 2ϩ has been shown to stimulate activity by accelerating the release of DNA substrate following topoisomerization, which is the rate-limiting step in steady state catalysis under low salt conditions (48). Since the viral enzyme has no detectable divalent metal cation binding site (48), this effect is presumably mediated by metal cation binding to DNA. Given the known effects of monovalent cations on processivity (4) and the similarity between the stimulatory effects of monovalent cations and Mg 2ϩ , it seems likely that the cellular enzymes are stimulated by Mg 2ϩ in a manner similar to the viral enzyme. The possibility of direct participation of a divalent metal cation in phosphodiester bond cleavage would seem to be excluded by the fact that metal cations are not required for Topo I activity.
The stimulatory effects of Mg 2ϩ and the polycations (spermine and spermidine) were found to be at least partially additive, suggesting that the two substances may influence Topo I activity by separate mechanisms. Spermine and spermidine can effectively shield the phosphate backbone of duplex DNA by binding in the minor groove (49). This information and the fact that certain minor goove binding compounds are known to be inhibitors of Topo I (50) can be taken to suggest that the configuration of the minor groove can influence Topo I activity. This hypothesis is further supported by the observed weak consensus sequence for Topo I cleavage (5Ј-(A/T)(G/C)(A/T)T-3Ј), which is suggestive of protein contacts within the minor groove (9,16). Alternatively, spermine and/or spermidine may have an allosteric effect on Topo I. Further experiments are needed to better define the effects of both divalent and polyvalent cations on Topo I activity.
The NH 2 -terminal Domain-The role of the unconserved NH 2 -terminal domain remains elusive. All cellular eukaryotic topoisomerase I enzymes have this highly charged domain, which in every case examined is dispensable for in vitro activity (3,24,27,30,51). However, since this domain contains nuclear localization signals, it is nevertheless required for the in vivo function of Topo I (24). Of the four putative nuclear localization signals residing in NH 2 -terminal domain of the human en-zyme, residues Lys 150 -Asp 156 have been suggested to be the most important for nuclear localization based on sequence comparisons with other cellular eukaryotic Topo I enzymes (24). However, we find that Topo70, which starts at a methionine immediately 5Ј to residue Lys 175 , is transported to the nucleus of insect cells. Hence, the single remaining intact nuclear localization signal (Lys 192 -Glu 198 ) must be sufficient for nuclear transport, at least in insect cells.
Aside from its role in nuclear localization, what is the function of the NH 2 terminus? In the human enzyme, we find that residues 1-174 of the NH 2 -terminal domain are largely if not completely unfolded and that the activity of the Topo70 enzyme (which is missing these residues) is indistinguishable from that of the full-length enzyme by every criterion tested. The highly extended nature of the NH 2 -terminal domain was observed using analytical techniques that involved conditions of both low salt (circular dichroism, 10 mM) and high salt (sedimentation and gel filtration, 200 mM). Hence, the extended nature of the NH 2 terminus persists under variable salt concentrations and is not an artifact of any single analytical technique. With its high density of both negatively and positively charged residues (67% charged residues), the NH 2 terminus is almost zwitterionic in nature. Accordingly, we have observed that the fulllength enzyme can be concentrated to Ͼ50 mg/ml with no signs of precipitation. In contrast, Topo70 can only be concentrated to ϳ5 mg/ml before it begins to precipitate (data not shown). Hence, the NH 2 -terminal domain acts as a solubilizing element in vitro. A conservative estimate for the quantity of Topo I in HeLa cells can be made based on our recovery of 250 g of purified Topo I from 10 9 HeLa cells. Since greater than 95% of the cell protein is associated with the nuclear compartment (data not shown), which has an average free water volume of approximately 3000 M 3 for HeLa S3 cells (52), we estimate that the minimal nuclear concentration of Topo I is ϳ100 g/ml. This concentration is 50-fold below the solubility limit of Topo70. However, Topo I is known to be highly concentrated in regions of chromatin that are undergoing high levels of transcription such as activated heat-shock loci and the nucleolus (28,29,53,54). Furthermore, UV cross-linking and SDS-induced trapping of Topo I-DNA complexes has revealed that Topo I is enriched at least 20-fold on rDNA relative to total DNA (29) and is enriched 20-fold at induced heat shock loci relative to the uninduced loci (23). Thus, the possibility exists that local concentrations of Topo I could approach 5 mg/ml, which might necessitate a solubility factor such as the zwitterionic NH 2 terminus.
The Shape of Topo I-In addition to revealing the extended nature of the NH 2 -terminal domain, the comparison of the hydrodynamic properties of full-length Topo I to those of Topo70 and Topo58 revealed that the core, linker, and COOHterminal domains fold into a globular structure. Hence, the catalytically active domain of Topo I is largely globular, while the NH 2 -terminal domain is largely unfolded, consistent with it being dispensable for activity. Since the Topo58 core domain was capable of folding into a well behaved globular molecule, it probably represents a subdomain of Topo I that is capable of folding independent of the amino-terminal, linker, or core domains. Finally, the availability of milligram quantities of purified full-length Topo I, Topo70, and Topo58 greatly facilitated further studies of the structure-function relationships of Topo I. In the accompanying paper, we describe the use of limited proteolysis to further characterize the domain structure of the protein (40). POROS purifications were performed using the SPRINT high pressure liquid chromatography system from Perseptive Biosystems, kindly provided by Ken Walsh.