Assignment of Functional Amino Acids around the Active Site of Human DNA Topoisomerase IIα*

An expression library for active site mutants of human topoisomerase IIα (TOP2α) was constructed by replacing the sequence encoding residues 793–808 with a randomized oligonucleotide cassette. This plasmid library was transformed into a temperature-sensitive yeast strain (top2-1), and viable transformants were selected at the restrictive temperature. Among the active TOP2α mutants, no substitution was allowed at Tyr805, the 5′ anchor of the cleaved DNA, and only conservative substitutions were allowed at Leu794, Asp797, Ala801, and Arg804. Thus, these 5 residues are critical for human TOP2α activity, and the remaining mutagenized residues are less critical for function. Using the x-ray crystal structure of yeast TOP2 as a structural model, it can be deduced that these 5 functionally important residues lie in a plane. One of the possible functions of this plane may be that it interacts with the DNA substrate upon catalysis. The side chains of Ser803 and Lys798, which confer drug resistance, lie adjacent to this plane.

Type II DNA topoisomerases are essential cellular enzymes that are required for cell proliferation. They catalyze topological change of DNA molecules through the transient breakage and rejoining of double-stranded DNA; mechanistically, the enzyme makes a gate in one double-strand DNA segment and allows another DNA segment to pass through the gate. These reactions relax supercoiled DNA and catenate or decatenate covalently closed circular DNA molecules (1). In addition, type II DNA topoisomerases are clinical targets for antibiotics and anticancer drugs.
Prokaryotic DNA TOP2, also known as DNA gyrase, is composed of two subunits, A protein (GyrA) and B protein (GyrB), and the active form is an A 2 B 2 heterodimer. Eukaryotic TOP2 1 is a homodimer composed of a monomer with two domains, BЈ and AЈ, corresponding to DNA gyrase B and A subunits, respectively (2). All TOP2 enzymes share sequence similarity (3); in addition, similarity has been observed in the x-ray crystal structures of TOP2 proteins (4,5). Because TOP2 is conserved through evolution, limited genetic interchangeability for TOP2 from different species is observed (6 -8).
The enzymatic mechanism of TOP2 has been studied exten-sively, and it is thought to involve multistep conformation changes of the enzyme (9). Because only a small number of crystal structures of TOP2 have been published, it is difficult to understand the enzymatic mechanism fully from this structural information alone. However, biochemical experiments have been carried out to provide information on the interactions between the enzyme and its DNA substrate. For example, Worland and Wang (10) showed that the active site Tyr 783 attaches covalently to the 5Ј end of the cleaved DNA chain. Recent work on DNA gyrase suggested specific interactions between the enzyme and its DNA substrate at distinct steps of the catalytic cycle by probing the topology of the enzyme-bound DNA segment (11).
To understand enzymes and their biological function, it is important to characterize both their catalytic and their structural properties. Several experimental approaches are commonly used for this purpose. One approach is to determine the primary structures (i.e. protein sequences) of related proteins from different species and align them with one another. Such a sequence alignment is useful to identify the evolutionarily conserved amino acids and motifs that may play functional roles in a group of related proteins. X-ray crystallography and nuclear magnetic resonance are used to study the three-dimensional structure of a protein at the atomic level. These structural methods provide information on the structure of the entire protein and its subdomains and subunits. Such information is necessary to understand how enzymes interact with substrates, other molecules, inhibitors, and activators, etc., and to evaluate the biological roles of the proteins in vivo. Once a structural model is available, site-directed mutagenesis of specific residues is often undertaken to determine the structure/ function relationships of the protein in detail.
Ala scanning is a commonly used mutagenesis approach, in which protein residues of interest (i.e. in the active site, ligand binding site, or protein interface) are systematically changed one by one from their identity in the wild type protein to Ala in a corresponding mutant. This is useful to create mutant proteins, but the effects of specific amino acid substitutions are not generally predictable. Another approach is to randomly mutagenize a codon or group of codons by random substitution of nucleotides; the group of mutants can then be subjected to genetic selection (12,13). If a particular property of a side chain (length, hydrophobicity, etc.) is important at a given position, only side chains with conservative amino acid substitutions will be allowed in functional mutants (i.e. those that survive the selection process). This method reinforces the interpretation of x-ray crystal structures and helps to define the functional roles of each amino acid residue in vivo.
Such mutagenesis studies have exclusively used Escherichia coli as the host for expression and selection of mutant proteins.
Large numbers of mutant proteins are easily analyzed in E. coli; however, high molecular weight mammalian enzymes that are not able to be expressed in E. coli cannot be analyzed by this approach in this host. In the present study, random mutagenesis was carried out on human topoisomerase 2␣ (TOP2␣). To characterize mutants of TOP2␣, they were expressed in a yeast strain carrying a temperature-sensitive endogenous yeast TOP2, which can be genetically complemented by the gene encoding wild type human TOP2␣. This mutagenesis study specifically targets amino acid residues 793-808 around the active site of human TOP2␣. This highly conserved region is of particular interest because it is involved in both catalytic activity and drug sensitivity. The results presented here provide useful information on the roles of these active site residues. Based on the amino acid substitutions allowed in active mutants and their drug sensitivity profiles, possible functions of the targeted amino acid residues are proposed and discussed.
Plasmid Constructions-The short XbaI-HindIII fragment of pYES2 (Invitrogen, Groningen, Netherlands) was replaced with a synthetic oligomer 5Ј-CTA GGC TCG AGA TGC TAC GTA AGT CAG CCC GGG CTG CGG CCG CT, which is annealed with 5Ј-AGC TAG CGG CCG CAG CCC GGG CTG ACT TAC GTA GCA TCT CGA GC (LLpYES-PGAL1). LLpYES-PGAL1 contains a new restriction site for NotI and lacks XbaI and KpnI sites. The full length of human TOP2␣ cDNA was transferred into the large NotI-XhoI fragment of LLpYES-PGAL1 (LLpYES hTOP2-PGAL1).
Nonfunctional Dummy TOP2␣ was constructed by insertion of a synthetic oligomer 5Ј-GTT ACC ATC GCA TGC AAG CTT GCT CAG C (DUM-U) annealed with 5Ј-GCT GAG CAA GCT TGC ATG CGA TGG TAC C (DUM-D) in the large HpaI-KpnI fragment of PSL1190hTOP2. The NotI-XhoI fragment was further replaced with the corresponding site of LLpYES hTOP2-PGAL1 (LLpYES hTOP2DUM-PGAL1 or Dummy Vector).
Random Oligonucleotides-An equimolar mixture of random oligonucleotides was made from 16 synthetic oligomer pools with substitutions at codons 788 -813 of human TOP2␣, 5Ј-GGT CAG TTT GGT ACC (AGG CTA CAT GGT GGC AAG GAT TCT GCT AGT CCA CGA TAC ATC TTT ACA) ATG CTC AGC TCT TTG. In each primer, either one of the target codons, between Arg 793 and Thr 808 , shown in parentheses in the above sequence, was randomized by including all four nucleotides in equimolar amounts at the appropriate synthetic cycle. We also introduced a silent mutation at a codon downstream of each target for 15 out of 16 random oligomers (Arg 793 -Phe 807 ). Two silent mutations in Arg 793 and His 795 were placed in the random oligomer that targeted residue Thr 808 . For example, the Arg 793 random oligomer had the sequence 5Ј-GGT CAG TTT GGT ACC NNN CTg CAT GGT GGC AAG GAT TCT GCT AGT CCA CGA TAC ATC TTT ACA ATG CTC AGC TCT TTG, where the N represents an equimolar mixture of the G, C, A, and T and the lowercase letter g denotes a silent mutation. The silent mutations were used to identify the origins of the oligomer when the wild type sequence was restored during synthesis. As a result, the random oligonucleotides theoretically contain equimolar amount of the 1024 possible mutations as single amino acid substitutions.
Library Construction-To construct a human TOP2␣ mutant plasmid library, the mixture of random oligonucleotides was used as a forward PCR primer. A reverse primer was designed complementary to the XbaI site of hTOP2cDNA sequence (5Ј-CAT GGG TTC TAG AAC TTG TTC). Using PSL1190hTOP2, as a template, 25 cycles of PCR (94°C for 30 s, 45°C for 30 s, and 72°C for 30 s) were carried out using the proofreading-competent Pyrobest DNA polymerase (TaKaRa, Kyoto, Japan) with appropriate buffers. The 380-base pair PCR product was isolated by electrophoresis in a 0.8% agarose gel and purified by QIAEXII gel extraction kit (Qiagen, Valencia, CA). This fragment was treated with restriction enzymes KpnI and XbaI, and purified again by the same procedure (Randomized Cassette). In some experiments, one of 16 random oligomers was used as a primer for library construction (see below).
The KpnI-XbaI large fragment of the LLpYES hTOP2DUM-PGAL1 and the Randomized Cassette were ligated and used for transformation of DH5␣. An aliquot of the transformed E. coli was plated on 2ϫ YT (16 g/liter tryptone, 10 g/liter yeast extract, 5 g/liter NaCl, pH 7.3) containing carbenicillin (100 g/ml) to determine the total number of transformants, and the remainder was inoculated into 500 ml of 2ϫ YT and cultured at 37°C overnight. Plasmids were recovered and used as a mutant plasmid library. The random library consists of approximately 6000 independent clones.
Complementation Assay-The yeast strain SD1-4 was used for transformation with the human TOP2␣ mutant plasmid library by Frozen-EZ Yeast Transformation II (Zymo Research, Orange, CA). The transformants were cultured on solid medium, SD-URA (6.7 g/liter yeast nitrogen base without amino acids, 5 g/liter casamino acids, 20 g/liter glucose, 20 mg/liter adenine sulfate, 20 mg/liter tryptophan, 15 g/liter Bacto agar) at 25°C for 48 h, replicated on a plate containing SDGAL-URA by using Replica Plater (TaKaRa, Kyoto, Japan), in which glucose of the SD-URA medium was replaced by 20 g/liter galactose. The replicated colonies were cultured at 37°C for 7 days. Colonies were picked and cultured in tubes with 2 ml of SD-URA at 25°C overnight. The yeast cells were collected, resuspended in 200 l of lysis buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris-HCl, and 1 mM Na 2 EDTA), and lysed by vigorous mixing for 10 min with 300 mg of acid-washed glass beads (300 microns; Sigma) and 200 l of Phenol CIA (phenol:chloroform:isoamyl alcohol ϭ 25:24:1, mixed pH 6.7; Nacalai Tesque, Kyoto, Japan). After centrifugation and ethanol precipitation, the purified DNA was used for transformation of DH5␣ by electroporation according to Bio-Rad. The transformed cells were incubated on 2ϫ YT plate containing carbenicillin (100 g/ml) at 37°C overnight. Colonies were picked up and inoculated into 2 ml of 2ϫ YT solution containing carbenicillin (100 g/ml) at 37°C overnight. Plasmids were extracted and purified by Wizard Plus Mini Preps DNA purification system (Promega, Madison, WI) and used for a second transformation of the SD1-4 strain to confirm colony formation.
DNA Sequencing-The target regions of active mutants were sequenced using the Taq Dye Terminator Cycle Sequence Kit, with a Perkin-Elmer-Applied Biosystems 373A DNA Sequencer (Foster City, CA). An oligomer, 5Ј-GACAGTGGTGAAATGT, was used for the sequencing primer.
Complementation Assay Using the Independent Libraries-Construction of independent libraries and complementation assays were carried out using the same procedures as described above, except that each single random oligomer was used for the library construction. Each library contains 64 codon variations at the targeted position.
Site-directed Mutagenesis-CelII restriction sites outside the TOP2␣ gene of the Dummy Vector were knocked out by a CelII partial restriction and treatments with Klenow fragment and T4 ligase. Oligomers between the KpnI and CelII sites of human TOP2␣ with desired mutations were chemically synthesized and used for replacement with the corresponding fragment of the Dummy Vector.
Etoposide Resistance Assay-To investigate etoposide resistance, the permeable yeast strain JN394top2-4 was transformed by the functional mutants of human TOP2␣ by means of electroporation (Gene Pulser II, Richmond, CA). After 7 days of culture on SD-URA plates at 25°C, each transformant was picked and streaked on SDGAL-URA plate with or without etoposide (100 g/ml; Sigma). Etoposide resistance was determined after culture at 35°C for 7 days.

Construction of Yeast Expression Library for Mutants of Human TOP2␣-
The primary and secondary structures of the active site region of TOP2 are conserved among species from bacteria to human ( Fig. 1 and Refs. [3][4][5] and are essential for catalysis (1). The active site includes a critical tyrosine residue that becomes covalently linked to DNA during catalysis (Tyr 805 in human TOP2␣). In Saccharomyces cerevisiae TOP2, this active site Tyr residue is located at the C-terminal edge of a loop structure on the surface of the AЈ domain (4). To study the functions of individual residues in this region of human TOP2␣, random substitutions were made for amino acids 793-808, and mutants with enzymatic activity were selected by genetic complementation (Fig. 2).
Sixteen oligonucleotide pools were synthesized that are complementary to the 78 bases around the TOP2 active site. Each oligonucleotide pool contains randomized nucleotides at one of the 16 codons in amino acids 793-808. A 380-base pair fragment of the TOP2 gene was amplified using a mixture of these 16 "mutant" oligonucleotide pools as one PCR primer and a wild type oligonucleotide as the other PCR primer. The PCR product was purified, digested by KpnI and XbaI, and used to replace the same restriction fragment of the Dummy Vector (see "Experimental Procedures"). In theory, the library of plasmids recovered from ligating the PCR product with the Dummy Vector direct synthesis of wild type human TOP2 and 19 single amino acid substitution mutants at each residue in the targeted region in human TOP2␣. However, the representation of different mutants in the library depends on the efficiency of synthesis of each mutant oligonucleotide and the efficiency with which it is used in the PCR reaction. To test the actual composition of the library, 49 clones were arbitrarily selected, and their DNA sequences were determined (Fig. 3). All theoretically possible mutations were found within the randomized region, and all isolates had single amino acid substitutions at 13 positions of 16. Although in five cases the same substitutions were recovered twice, each contained different nucleotide sequences with a few exceptions (i.e. two Leu substitutions at Asp 799 and two Gly substitutions at Ser 802 ).
Genetic Selection of Functional Human TOP2␣ Mutants-The temperature-sensitive S. cerevisiae strain SD1-4 (top2-1) was used to select active variants of TOP2 from the library of mutants by in vivo complementation assay. Previous studies have demonstrated that the temperature-sensitive phenotype in yeast can be complemented by transformation with a plasmid that expresses either mouse (6) or human TOP2 (14). A plasmid that expresses wild type human TOP2␣ also suppresses the temperature-sensitive phenotype of SD1-4 at 37°C, although a plasmid expressing an inactive human TOP2␣ (i.e. Y805L) or the vector alone, does not complement the host strain (Fig. 4).
In vivo selection was carried out to identify functional human TOP2␣ mutants in the plasmid library; however, accurate discrimination between functional and nonfunctional mutants requires a sensitive and strict complementation specificity. To test this system using SD1-4, cells were screened at 37°C after being transformed with plasmid DNA mixtures containing wild type human TOP2-expressing plasmid and the Dummy Vector at the following ratios: 100:0, 50:50, 25:75, or 0:100. Colonies were seen on plates only when the mixture included the wild type vector. Plasmid DNA was isolated from 20 individual clones from each transformation, and their sequences were determined. All plasmids from the 100:0 mixture and 19 of 20 from the 50:50 and 25:75 mixtures were wild type plasmids. Thus, the system provides a way to discriminate the nonfunctional clone when a homogeneous plasmid was used for transformation, but the rate of false positives is 5% when a mixed population of input DNA is screened. To completely exclude nonfunctional mutants in the subsequent experiments, each plasmid recovered from the first round of selection was subjected to a second round of selection in the same strain. After the second round of selection, the false positive clones were effectively removed.
Active Mutants of Human TOP2␣-This random library was  transformed into yeast, and 200 independent transformants were grown at 30°C (permissive temperature). Colonies were transferred to fresh plates by replica plating and cultured for a week at 37°C (restrictive temperature). Approximately half of the colonies that grew at 37°C also grew at 30°C. A total of 108 plasmids were isolated from the temperature-resistant colonies, and the DNA sequence of the mutated region was determined. Six of the 108 plasmids unexpectedly contained double mutations, and these were not studied further. Four mutants did not form colonies when the plasmid was retransformed and were regarded as false positives. As a result, 98 clones encoding active mutants of human TOP2␣ were obtained for further study.
Among the 98 active mutants of TOP2␣ selected by this screen, there were 83 mis-sense and 15 silent mutations. Some of the mis-sense substitutions were nonredundant at the nucleotide or amino acid level, so that the pool of active TOP2 mutants represented 68 different codon substitutions and 50 different amino acid substitutions in the targeted active site region (Fig. 5).
The amino acid substitutions were not random. Leu 798 , Ser 802 , and Pro 803 were among the residues that were most tolerant of variation; active mutants were selected in which these residues were replaced by amino acids with variable charge and side chain lengths. In contrast, Leu 794 , Gly 797 , Asp 799 , Ala 801 , Arg 804 , Tyr 805 , Ile 806 , Phe 807 , and Thr 808 tolerated only subtle changes (Fig. 5B) and tended to be unchanged or were replaced only by very similar amino acids (conservative changes); however, silent mutations were found in the codons for Arg 804 and Phe 807 (Fig. 5A).
Active Mutants in Independent Libraries-Amino acids that allow only limited substitution without resulting in loss of enzyme activity may play a strictly defined role in the function of that protein (12,13,15). Although the initial screen described above identified 50 allowable amino acid substitutions in active variants of human TOP2␣, it was not a comprehensive screen of such mutants. Using a mixture of the 16 random cassettes as a library source, thousands of positive clones would have to be screened, isolated, and sequenced to make statistically valid conclusions about all possible active variants in the library. Therefore, nine additional independent libraries were prepared, each of which included random substitution at the codon for one of the following amino acids: Leu 794 , Gly 797 , Asp 799 , Ala 801 , Arg 804 , Tyr 805 , Ile 806 , Phe 807 , and Thr 808 . Complementation experiments were carried out until at least 200 transformants were screened from each library; this num-ber is more than three times higher than the number required to screen one copy of each possible substitution at a target codon (Fig. 6).
All active clones isolated from the library of mutants at Tyr 805 library had a Tyr residue at the targeted position (Fig.  6B). Because a Dummy Vector was used to construct these libraries, wild type clones cannot be contaminated by incomplete restriction digestion. Therefore, the active variants with a Tyr codon at 805 were generated from the randomly substituted oligonucleotide pool. In the Tyr 805 random mutant library, it is predicted that 6.3 Tyr codons would be found among the 200 colonies screened (assuming the library contains all codons in equimolar proportions). Consistent with this prediction, six wild type clones were identified by screening the Tyr 805 mutant library (Fig. 6A).
Among the other amino acids studied, four positions allowed only conservative substitutions to conserve enzyme activity (16). Leu 794 could be substituted with Ile or Val, the other two aliphatic side chains. Asp 799 could be substituted with the acidic side chain of Glu, Ala 801 could be substituted with the small amino acids Gly and Ser, and Arg 804 could be substituted with the basic amino acid Lys (Fig. 6B). These data show that Tyr 805 is immutable and that Leu 794 , Asp 799 , Ala 801 , and Arg 804 are nearly immutable residues of human TOP2␣, as detected by this genetic complementation assay.
Mutants That Are Resistant against Etoposide-By screening these nine libraries of mutants at individual codons in the active site of human TOP2␣, 66 functional mutants of human TOP2␣ were identified in total. Because these substitutions were introduced around the active site, these mutants are likely to include enzymes with altered catalytic specificities and properties, such as altered drug sensitivity. This possibility was tested by transforming the mutant plasmids into the JN394t2-4 strain and plating the transformants on agar plates containing 100 g/ml etoposide. Three mutants were selected as etoposide-resistant clones. One mutant, P803S, showed a weak resistance to etoposide, as reported previously (7,17). The other mutants, K798L and K798P, also formed etoposideresistant colonies. The relative resistance of these mutants were in the order K798P Ͼ K798L Ͼ P803S (Fig. 7).

Tolerance to Amino Acid Substitutions in the Genetic
Complementation-We have introduced single amino acid substitutions into the 16 amino acids around the active site of the enzyme, and these mutant enzymes were used to complement FIG. 4. Functional complementation of a TOP2 ts yeast strain SD1-4 by human TOP2␣. S. cerevisiae SD1-4 (top2-1) was transformed with a vector pYES, LLpYEShTOP2DUM-PGAL1, which carries inactive TOP2␣; LLpYESh-TOP2-PGAL1, which carries wild type human TOP2␣; or LLpYES-PGAL1-Y805L, which carries a mutant TOP2␣ at the active center. Transformed cells were streaked on an SD-URA plate and incubated at 25°C (A) and on SDGAL-URA plate at 37°C (B). In between the panels, names of the clones are indicated. LpYes, pYES2-PGAL1; Dummy, LLpYES hTOP2DUM-PGAL1; WT, LLpYESh-TOP2-PGAL1; Y805L, Leu substitution at Tyr 805 . Open arrowhead indicates the wild type culture. a TOP2 temperature-sensitive mutation in S. cerevisiae. Tolerance to amino acid substitutions at each residue was determined by using libraries including either a mixture of mutants at the 16 targeted amino acids or including mutants at only one of the 16 targeted amino acid residues (Figs. 5 and 6).
The results of the complementation assay suggest that the targeted amino acids can be categorized into four groups. The first group includes only one member, the active site Tyr 805 . This Tyr residue is involved in transesterification to the DNA phosphodiester bond; no active mutant could be isolated with any substitution in this position in experiments with the mixed codon and the single codon targeted mutant libraries. As shown in Fig. 3, the plasmid libraries include variants that synthesize proteins with amino acid substitutions at these positions; however, these nonfunctional substitutions do not complement the temperature-sensitive yeast host strain. Such mutants were also reported previously and were created by site-directed mutagenesis at this site. For example, a DNA topoisomerase IV mutant that has a conservative substitution of His for Tyr 805 carries out nicking of single-stranded DNA but does not carry out double-stranded DNA cleavage (18). That study agrees with the results presented here, in that it also demonstrates a strict requirement for a Tyr residue at position 805 to retain TOP2 activity.
The second group of residues identified in this mutant screen includes Leu 794 , Asp 799 , Ala 801 , and Arg 804 ; these four residues were susceptible to conservative amino acid substitutions. Leu 794 can be substituted with other aliphatic amino acids, Asp 799 can be substituted with another acidic side chain Glu, Ala 801 can be substituted with amino acids Gly and Ala that are small, and Arg 804 can be substituted with another basic side chain Lys (Fig. 6B). These amino acids play such important roles in enzyme function that only extremely limited substitutions can occur without loss of catalytic efficiency. Arg 804 is immediately adjacent to the active site Tyr 805 . In DNA gyrase, a similar set of amino acids has been proposed to form the active site of the breakage-reunion reaction (5). In addition, Arg 781 from S. cerevisiae TOP2, which corresponds to Arg 804 in the human enzyme, has been subjected to site-directed mutagenesis (19). The results of that study suggest that Arg 781 might be involved in anchoring the 5Ј side of the broken DNA (19). Our results are consistent with those results and also  Fig. 3. Solid triangles indicate the amino acid residues where either no substitution or only conservative amino acid substitutions were found. All substitutions were single amino acid replacements. In A, the number of substitution at each amino acid is listed under the wild type sequence. Codons and the corresponding amino acids are at the left. Silent substitutions are shown in italics. In B, substituted amino acid residues at each site are listed. Silent substitutions are not listed. When the same amino acid substitution was isolated more than once, the number of isolates is also indicated. demonstrate that another basic residue, Lys, could substitute for Arg at this position, although Arg is strictly conserved in naturally occurring proteins from various species (Fig. 1). Interestingly, in this group of four residues that allow conservative substitutions, Ala 801 and Arg 804 are highly conserved in all species including prokaryotes, whereas Leu 794 and Asp 799 , which are located farther from the active site, are conserved only in higher eukaryotes in the primary structure.
The last two categories of residues in the active site region of human TOP2␣ are much more tolerant of amino acid substitution. Ile 806 and Phe 807 are tolerant to all substitutions except those with charged side chains, and the last group, His 795 , Lys 798 , Ser 800 , Ser 802 , Pro 803 , and Thr 808 , are tolerant to a variety of amino acid substitutions including alterations in side chain length, hydrophobicity and charge. Gly 796 is tolerant of various substitutions and belongs either in the third or fourth groups. Residues in the third or fourth group are not essential for TOP2 function. Nevertheless, it is surprising that these residues have not randomly drifted in the course of evolution ( Fig. 1) but instead are represented by a rather narrow ranges of amino acids in naturally occurring TOP2 sequences. One explanation for this is that some substitutions at these positions might have little or no direct effect on enzyme activity but may alter other characteristics of the enzyme, such as its sensitivity to environmental stress, that are not evident in the screen carried out here.
Etoposide-resistant Mutants-It was somewhat surprising to find that Lys 798 was highly mutable by this screening procedure. It has been proposed that the corresponding residue in S. cerevisiae TOP2 interacts with the single-stranded region of the substrate DNA (19,20). If this interaction were to be essential, it would be predicted that Lys 798 would be relatively immutable. One possible explanation is that Lys 798 is involved in an interaction with DNA but that this interaction is dispensable for efficiency of the catalytic functions. Such interactions have been described in Taq DNA polymerase I. For example, Thr 664 in this polymerase interacts with the template DNA in the closed ternary complex (21) and was highly mutable in a genetic complementation assay (15). However, the substitution was not totally without effect on the enzyme, because it decreased the fidelity in DNA replication (22). If a comparable situation applies to Lys 798 of TOP2, then some mutants of Wild type (WT) and mutants in human TOP2␣ (K798P, K798L, K798C, and P803S) were used to transform JN394t2-4 (ISE2, top2-4). Each clone that carries either wild type or mutant TOP2␣ was cultured and streaked on an SDGAL-URA plate with (A) or without (B) 100 g/ml etoposide (VP16). Drug sensitivity was determined after incubation at 35°C for 1 week. In between the panels, names of the clones are indicated. P803S is indicated by an open arrowhead, and K798P and K798L are indicated by filled arrowheads .   FIG. 8. Structural model of human TOP2␣. A, the winged motif of the AЈ subdomain in the S. cerevisiae crystal structure (AA702-AA792) is shown with the side chains that belong to group1 (red) and the group 2 (purple). Two amino acid residues that are involved in etoposide sensitivity are also shown (black). The polypeptide backbone between Arg 793 and Thr 808 is highlighted in gray, and the remaining region is in green. The names of ␣5-helix ␤3-sheet, N-terminal site (N), and some of amino acids are indicated in the figure. B, the target loop is shown in a stereo view with all side chains. In both panels, side chains are replaced with the corresponding human amino acid residues, and referred to as the human positions (Arg 793 -Thr 798 ). The coordinate sets were obtained from the Protein Data Bank (38). The drawing was made by using the program package InsightII (Molecular Simulations Inc, San Diego, CA). The polypeptide backbone between Arg 793 and Thr 808 is shown as a gray ribbon. Red, group 1 (Tyr 805 ); purple, group 2 (Leu 794 , Asp 797 , Ala 801 , and Arg 804 ); yellow, group 3 (Ile 806 and Phe 807 ); light blue and black, group 4 (Arg 793 , Lys 798 , Ser 800 , Ser 802 , and Thr 808 ). Among the group 4 side chains, two that were involved in etoposide sensitivity are shown in black (Lys 798 and Pro 803 ).
Lys 798 may have altered biochemical properties that have not been identified in this study. Support for the above hypothesis was found by testing the active variants of TOP2␣ identified in this study for drug resistance. Among the three etoposide-resistant mutants that were identified, two mutants had amino acid substitutions of Leu or Pro at Lys 798 (K798L and K798P; Fig. 7). These mutants also exhibited the resistance in an in vitro experiment using overexpressed and purified proteins. 2 Another mutant at this position, K798N, has been described previously; thismutation occurred during selection of a cloned human leukemic cell line (23). However, the relationship between the substitution and drug resistance has not been proven for K798N.
Implication for a Structural Model of TOP2␣-The amino acid sequence of human TOP2␣ is nearly 50% homologous to the sequence of S. cerevisiae TOP2 (3). Furthermore, human TOP2␣ is functionally interchangeable with S. cerevisiae TOP2 in the yeast cell (this study and Ref. 14). These results suggest that these two proteins share a common conformation at the level of tertiary structure, especially around the active site (i.e. at the CAP-like folds; see Refs. 4,24,and 25). Based on this assumption, the functional data presented here were superimposed on structural information for S. cerevisiae TOP2 (4). Fig. 8 shows the locations of human amino acid residues Arg 793 -Thr 808 on a schematic diagram of the S. cerevisiae TOP2 polypeptide backbone. Side chains are shown in four colors as categorized above: red, purple, yellow, and light blue for groups 1 (immutable), 2 (conservative substitutions only), 3 (highly mutable, but not to charged side chains), and 4 (highly mutable), respectively. In S. cerevisiae TOP2, the first three targeted amino acids are in the ␣5-helix (Fig. 8A). This structure may also exist in human TOP2␣, because the corresponding region (Arg 793 -His 795 ) is located between the three conserved glycines at 791, 796, and 797 in both human and S. cerevisiae enzymes. Ile 806 and Phe 807 could not be substituted with any hydrophilic side chains, which may indicate that they are buried toward the hydrophobic interior of the molecule (26,27). In agreement with this observation, in S. cerevisiae TOP2, the corresponding amino acids extend toward the inside of the protein (␤3 sheet; Fig. 8A). Thus the first three and last three residues of the target loop ( Fig. 1) of human TOP2␣ superimpose well on the yeast TOP2 structure (Fig. 8).
In wild type human TOP2␣, two Gly residues and a Pro residue are found between Gly 796 and Pro 803 . These two amino acids are not favored to form either ␣-helix (28 -30) or ␤-sheet (31,32). The number of Gly and Pro residues can be further increased by substitutions without loss of activity (K798G, K798P, and S802G). Therefore, this region may form a random coil structure in human TOP2␣, just as seen in S. cerevisiae TOP2. Charged residues Lys 798 , Asp 799 , and Ser 802 , which were substituted by Arg (Fig. 5B), must be exposed on the surface of the protein.
The four amino acids (Leu 794 , Asp 799 , Ala 801 , and Arg 804 ) and the active site Tyr 805 must be functionally important because they are well conserved or unchangeable and because they form a plane on the polypeptide backbone (Fig. 8B). In the crystal structure of DNA gyrase A, Morias Cabral et al. (5) modeled a double-stranded DNA substrate on the surface of the protein and found that the active Tyr residues of each GyrA subunit, located approximately 27 Å apart, must move to achieve appropriately staggered positions for cleavage of the 5Ј ends of the duplex. In DNA gyrase A, the side chain positions of Ala 118 , Arg 121 , and Tyr 122 can be superimposed on the corresponding positions of Ala 801 , Arg 804 , and Tyr 805 (data not shown). Interestingly, Leu 794 and Asp 799 of human TOP2␣ also come very close to Ile 112 and Asp 115 of DNA gyrase A, respectively, in the same structure (data not shown). Therefore, it would be an attractive picture that this plane, including the active Tyr, performs catalysis through the conformational change in TOP2␣. According to this hypothetical model, the DNA molecule is held, cut, and rejoined on this planar surface.
The etoposide-resistant clones identified in this study may support this model. In bacterial DNA gyrases that are sensitive to synthetic quinolone compounds (33,34), resistant mutants have been mapped near the putative interaction site with DNA (5). Human TOP2␣ is inhibited by etoposide through formation of the cleavable complex (35), like bacterial DNA gyrases by quinolone compounds. Therefore, amino acids that are associated with etoposide sensitivity might come close to the DNA substrate in this region.
Our model is also consistent with the co-crystal structures of proteins that share the winged-helix motif with TOP2 (5,24). In the HNF-3/fork head DNA recognition motif (36), a part of the wing (W1) of HNF-3, which may be structurally homologous to the target loop structure in human TOP2␣, is considered to interact with DNA. The loop stem is stabilized by ␤-sheet hydrogen bonding and side chain-backbone interactions.
In the absence of any binary structures of human TOP2␣ and the substrate DNA, however, we do not rule out the other possibilities. For example, an alternative explanation is that the plane is involved in the interaction with the BЈ domain, although this is less likely because such amino acids may tolerate nonconservative substitutions (37). More plausible story is that some of the conservative amino acids are required for the efficient conformational changes during the catalysis or critical in keeping random coil structure of the loop region.