Yeast Gene for a Tyr-DNA Phosphodiesterase that Repairs Topoisomerase I Complexes

See allHide authors and affiliations

Science  15 Oct 1999:
Vol. 286, Issue 5439, pp. 552-555
DOI: 10.1126/science.286.5439.552


Covalent intermediates between topoisomerase I and DNA can become dead-end complexes that lead to cell death. Here, the isolation of the gene for an enzyme that can hydrolyze the bond between this protein and DNA is described. Enzyme-defective mutants of yeast are hypersensitive to treatments that increase the amount of covalent complexes, indicative of enzyme involvement in repair. The gene is conserved in eukaryotes and identifies a family of enzymes that has not been previously recognized. The presence of this gene in humans may have implications for the effectiveness of topoisomerase I poisons, such as the camptothecins, in chemotherapy.

Topoisomerases are cellular enzymes that are crucial for replication and readout of the genome; they work by breaking the DNA backbone, allowing or encouraging topological change, and resealing the break (1). The enzymes are efficient because DNA breakage is accompanied by covalent union between protein and DNA to create an intermediate that is resolved during the resealing step. This mechanism, although elegant, also makes topoisomerases potentially dangerous. If the resealing step fails, a normally transient break in DNA becomes a long-lived disruption, one with a topoisomerase protein covalently joined to it. Unless a way is found to restore the continuity of DNA, the cell will die.

In virtually all topoisomerases, the heart of the covalent complex is a phosphodiester between a specific tyrosine residue of the enzyme and one end of the break (the 3′ end for eukaryotic topoisomerase I and the 5′ end for topoisomerases II and III). The high-energy nature of this bond normally ensures the resealing step. But failure of resealing is markedly increased by several drugs, such as camptothecin (CPT), a promising anticancer agent that specifically targets eukaryotic topoisomerase I (2). Protein-linked breaks also accumulate when topoisomerases act on DNA containing structural lesions such as thymine dimers, abasic sites, and mismatched base pairs (3). To the extent that such lesions arise during the normal life of a cell, topoisomerase-associated damage may be unavoidable.

Repair of topoisomerase covalent complexes is of obvious value to the cell, but the subject remains largely unexplored. A plausible pathway invokes hydrolysis of the bond joining the topoisomerase to DNA; release of the topoisomerase would then permit the cleaved DNA to undergo conventional modes of break repair (4). Although no such hydrolysis has been reported for covalent complexes of topoisomerases II or III, we described (5) an activity that specifically hydrolyzes the type of bond found in complexes between DNA and topoisomerase I (Fig. 1A). The specificity of this activity and its conservation from yeast to man suggested that it might be part of a repair pathway. But without specific inhibitors or mutants, no assessment of its function could be made. We now report the identification of the gene encoding this enzyme and the demonstration of its importance for topoisomerase metabolism.

Figure 1

Molecular genetics of tyrosyl-DNA phosphodiesterase (TDP) activity. (A) Enzymatic transformations. The jagged line represents the single-strand 18-mer oligonucleotide of oHN279Y. TDP activity removes the tyrosine from this chemically synthesized substrate (5) and leaves a 3′-terminal phosphate. In crude extracts, subsequent action by unidentified phosphatases can produce a 3′-terminal hydroxyl. (B) Denaturing gel analysis of TDP activity in yeast strains. Incubations with 5′-radiolabeled oHN279Y were for 12 min as described (5) with buffer (lane 1) or extract (150 μg/ml) from the following strains: HNY102 and KYY337 (lanes 2 and 3); E17 and E6, two haploid segregants derived from KYY337 after four rounds of back crossing (lanes 4 and 5); HNY243 and HNY244,rad9::hisG derivatives of HNY102 and E6 (lanes 6 and 7); HNY244 containing plasmid pL10-13 (lane 8); and HNY383, a derivative of HNY243 with a disruption of the gene for ORF YBR223c (lane 9). The positions of the labeled substrate (Y) and oligonucleotides terminated by phosphate (P) and hydroxyl (O) residues are marked. Total TDP activity is best judged as the ratio P + O/Y + P + O. (C) TDP activity in E. coli. Radiolabeled oHN279Y was incubated as above with buffer (lane 1) or sonic extracts (10 ng/ml) of strain BL21(DE3) transformed either with plasmid vector (lane 2) or vector plus the coding region of YBR223c (lane 3). wt, wild type.

Crude extracts of the yeast Saccharomyces cerevisiae contain readily detectable amounts of tyrosyl-DNA phosphodiesterase (TDP) activity (5). We disrupted (6) four yeast genes—RAD9, RAD17, RAD52, andTOP1—that we suspected might encode or control the activity, but none of the disruptions affected activity in extracts (Fig. 1B) (7). To search for previously unknown genes, we assayed extracts from colonies of chemically mutagenized yeast (8); this screen yielded a single strain, KYY337, with very low TDP activity (Fig. 1B). In back crosses to the parental line, the enzyme defect appeared to reflect a single mutation (denoted here as enz). That is, when a diploid between the parental line and a defective line was sporulated and haploid colonies were assayed at random (8), about equal numbers were found with normal and with low enzyme activity. The activity of representative colonies after four rounds of back crossing is shown in Fig. 1B.

To assess the role of TDP activity in repair of topoisomerase damage, we compared strains for sensitivity to killing by CPT (9). Despite the marked difference in TDP activity, the parental line and the back-crossed enz mutant were both insensitive to CPT (Fig. 2A, bars 1 and 2). We reasoned that, as for other kinds of damage (4), repair of topoisomerase lesions might take place by multiple pathways. If so, a genetic background in which some of these pathways were disabled might reveal a role for TDP activity. Indeed, when combined with a disruption of the RAD9 gene, the CPT sensitivity of the low activity mutant (strain HNY244), was increased by a factor of 12 relative to the rad9 derivative of the parental strain, HNY243 (Fig. 2A, bars 3 and 4); the same difference was seen after the mutant had undergone two additional rounds of back crossing (7).

Figure 2

Influence of TDP activity on cell survival after drug treatment. The indicated yeast strains were exposed to drug for 24 hours, diluted, and plated (9). Killing by the drug is calculated from the relative change in colony-forming units (CFUs), the number of colonies obtained from a portion of the culture after drug treatment divided by the number in a portion of the starting culture. (A) CPT was added at 100 μg/ml to strains HNY102, E6, HNY243, HNY244, and HNY383 (bars 1 to 5). (B) MMS was added at 0.01% to strains HNY243 and HNY244 (bars 1 and 2). (C) CPT was added at 100 μg/ml to strain HNY244 containing either a control plasmid, pX1, or plasmid pL10-13 (bars 1 and 2).

The RAD9 gene is needed both for the operation of DNA damage checkpoints and for expression of a set of DNA damage–inducible genes (10). The loss of these functions in a rad9mutant not only increases the sensitivity of the cell to killing by CPT, it apparently leaves TDP activity as a principal remaining source of repair of CPT-induced damage. Under these circumstances, killing by CPT still reflects topoisomerase trapping; when the TOP1gene of HNY244 was disrupted, survival increased nearly 1000-fold (7). The mutant line was not sensitized to all sources of DNA damage; killing by methyl methane sulfonate (MMS), an alkylating agent, was indistinguishable in HNY243 and HNY244 (Fig. 2B).

Mutations in yeast topoisomerase I have been isolated that depress rejoining and thereby lead to accumulation of covalent complexes (11). We used these mutants for an independent test, one without recourse to drugs and the attendant questions concerning uptake, of the importance of TDP activity for in vivo repair of topoisomerase-DNA adducts. Indeed, overexpression of a mutant (but not the wild-type) TOP1 gene was more toxic to the strain with low TDP activity than to its control (Fig. 3). A second mutant, top1R517G, with a similar defect (11), was similarly hypertoxic in the strain with low TDP activity (7).

Figure 3

Cell growth with a toxic topoisomerase. Strains HNY243 top1Δ and HNY244 top1Δ were transformed with derivatives of plasmid YCpGAL1 bearing either a wild-type TOP1 gene or the Thr722 → Ala (T722A) mutant (11). These strains were serially diluted and spotted on uracil-deficient minimal plates containing either 2% glucose (Glu) or galactose (Gal) to repress or induce the plasmid-borne gene.

From a library of yeast genomic fragments screened (12) for the ability to improve the CPT resistance of HNY244 and restore its TDP activity, we obtained plasmid pL10-13 (Figs. 1B and2C). Several subclones of the ∼8-kb insert in this plasmid retained full activity (7). The smallest of these subclones contains a single open reading frame (ORF), YBR223c, a protein of 544 amino acids and relative molecular mass ∼ 62,000. Into strain HNY243, we generated by polymerase chain reaction (PCR) (13) a disruption that removed all but the first 32 amino acids of the ORF. The resulting strain had an enzymatic defect and CPT sensitivity very similar to that of HNY244 (Figs. 1B and 2A), indicating that YBR223c is involved in TDP activity. To distinguish whether YBR223c encodes or controls TDP activity, we introduced a histidine-tagged version (14) into Escherichia coli, which by itself has no detectable TDP activity. Induction of bacteria bearing this construct (but not a control construct) apparently resulted in massive overproduction of TDP because crude extracts of such cells had a specific activity >10,000 times as high as that of extracts from a standard yeast strain (Fig. 1, B and C). Moreover, most of the induced activity could be bound to a tag-specific column; specific elution released >75% of the bound activity, resulting in a fraction with a single Coomassie-stainable band of the expected molecular size (7). We conclude that YBR223c encodes the enzyme and have accordingly renamed its gene TDP1.

Database searches failed to reveal homology betweenTDP1 and any genes of known function. Even individualized comparisons to motifs identified in various phosphodiesterases and phosphatases were, at best, marginal. On this basis, we conclude thatTDP1 encodes a previously unknown enzyme. However, eukaryotic (but not prokaryotic) databases contain several unannotated sequences that match TDP1, a finding consistent with the distribution of enzyme activity (5). The complete genome sequence of the nematode Caenorhabditis elegans contains a single ORF with substantial similarity to TDP1. Probing expressed sequence tag (EST) databases with the yeast and nematode proteins revealed many unambiguous matches (Fig. 4). In mouse and man, there are several EST entries that can be aligned to make up a single ORF with substantial similarity to the carboxyl-terminal half of TDP1. To see if the homology extends further, we carried out a PCR on a collection of human cDNAs (Marathon-Ready; Clontech Laboratories, Palo Alto, CA) with a primer complementary to the human EST sequence and a primer complementary to the tag affixed to the 5′ end of the cDNAs. We cloned the resulting 5′-RACE (rapid amplification of cDNA ends) products; the sequence of one of the longest clones (Fig. 4) aligns well to most of the 5′ half of the yeast and nemotode ORFs. We conclude that theTDP1 gene is highly conserved in eukaryotes.

Figure 4

Alignment of TDP homologs from various organisms. sc, S. cerevisiae gene YBR223c (GenBankZ36092.1); ce, C. elegans gene F52C12.1 (GenBankAF100657.2); hs5′, 5′-RACE of human cDNA (22); hs3′, assembly of human ESTs (GenBank AA477148, AA489121, and AI480141); mm, assembly of mouse ESTs (GenBank AA940134, W89267 and W13117); dm,Drosophila melanogaster EST (GenBank AI517253). Black boxes, identities; shaded boxes, similarities. A region of uncertain sequence in GenBank entry AA489121 is marked by x residues. Together with the 5′-RACE, the product of a 3′-RACE (22) confirms the sequence of the human ESTs and shows that the sequence in the region of ambiguity is identical to that shown for the mouse protein. The product of the 5′-RACE extends for 79 amino acids upstream of the sequence shown but still may not include the start of the full-length human protein. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Isolation of the TDP1 gene will allow studies of the enzymology and cell biology of a kind of DNA repair that has previously been hard to analyze. The gene also provides a potential tool to improve chemotherapy with camptothecins and other topoisomerase I poisons. Although these are promising anticancer drugs, their value is often limited by resistance of tumor cells or sensitivity of nontumor cells (or both). Repair of the topoisomerase lesion is likely to be one of the factors that determine the level of cellular sensitivity to topoisomerase poisons (15). With the TDP1 gene in hand, one can readily assess the expression of this enzyme in individual patients and possibly predict the likelihood of therapeutic success. Moreover, if genetic or biochemical techniques can be used to alter enzyme activity, the efficacy and safety of the drugs may be improved.

  • * To whom correspondence should be addressed. E-mail: nash{at}


Stay Connected to Science

Navigate This Article