Yeast Tdp1 regulates the fidelity of nonhomologous end joining

Abstract

Tyrosyl-DNA-phosphodiesterase 1 (Tdp1) can disjoin peptides covalently bound to DNA. We assessed the role of Tdp1 in nonhomologous end joining (NHEJ) and found that linear DNA molecules with 5′ extensions showed a high frequency of misrepair in Δtdp1 cells. The joining errors in Δtdp1 cells were predominantly 2-4 nucleotide insertions. Ends with 3′ extensions or blunt ends did not show enhanced frequencies of errors, although Δtdp1 cells repaired blunt DNA ends with greater efficiency than WT cells. We found that insertions required Ku80 and DNA ligase IV, as well as polymerase IV. Our results show that yeast Tdp1 is a component of the NHEJ pathway. We suggest that Tdp1p 3′ nucleosidase activity regulates the processing of DNA ends by generating a 3′ phosphate, thereby restricting the ability of polymerases and other enzymes from acting at DNA ends. In support of this model, we found that overexpression of Tpp1, a yeast DNA 3′ phosphatase, also leads to a higher frequency of insertions, suggesting that the generation of a 3′ phosphate is a key step in Tdp1-mediated error prevention during NHEJ.

Nonhomologous end joining (NHEJ) is a critical pathway for repairing DNA double-strand breaks. In higher eukaryotes, it functions as a primary repair pathway for repairing double-strand breaks from exogenous DNA damage and is also required for gene rearrangements in the immune system (1). This pathway is conserved in lower eukaryotes and bacteria where it functions mainly as a secondary repair pathway for the repair of double-strand breaks (2, 3). Because NHEJ does not rely on DNA homology for carrying out repair, it is an intrinsically error-prone pathway. The detailed steps of how accuracy is maintained during NHEJ processes remain poorly understood.

NHEJ is carried out by a set of well-conserved proteins including the Ku70/Ku80 proteins that bind to DNA ends, serving as a scaffold for subsequent repair reactions, and DNA ligase IV (4). In mammalian cells, a DNA-dependent protein kinase DNA-PKcs also plays critical roles (5). In addition to the core factors that are absolutely required for the NHEJ pathway, other factors such as nucleases and DNA polymerases participate in NHEJ (6, 7).

Tdp1p was identified on the basis of its ability to remove peptides covalently bound to DNA (8–10). Yeast Tdp1 can remove phosphotyrosyl-linked peptides from both the 3′ and 5′ ends of DNA (8, 11). Tdp1 also has the ability to remove damaged and undamaged nucleosides from the 3′ end of DNA through an activity that is mechanistically identical to the esterase activity that removes peptides (12). Because human Tdp1 is mutated in a DNA repair disorder [spinocerebellar ataxia with axonal neuropathy (13)], we assessed additional roles of this enzyme by using yeast repair mutants. In this paper, we demonstrate that yeast Tdp1 plays an important and unique role in NHEJ and affects the accuracy of the formation of repair junctions.

Results

Yeast Tdp1 is Required for Accurate Joining of some NHEJ Substrates.

We measured NHEJ by transfecting a linearized yeast plasmid into cells and determining transformation efficiency. Repair of transfected plasmid DNA requires NHEJ functions when the break site lacks homology to sequences found in the cell (2, 14). No significant differences in repair efficiency between WT and isogenic Δtdp1 cells were observed when the transfected plasmid was linearized with HindIII, PstI, or EcoRI (Fig. 1A). In WT cells, repair of linearized DNA with cohesive ends is accurate (15, 16). However, a large difference in repair accuracy was detected between WT and Δtdp1 strains when the transfected DNA was linearized with HindIII or EcoRI (Fig. 1B). In WT cells, the error frequency was 4/100 transformants analyzed, and the four colonies showing misrepair carried 1-2 nucleotide deletions at the junction. By contrast, in Δtdp1 cells transfected with HindIII-linearized DNA, 23/100 colonies showed misrepair (p < 0.0001 by using Fisher’s exact test). Of the misrepaired DNAs, 18 of the 23 contained nucleotide insertions, with 9/23 carrying four nucleotide insertions. All the insertion events were consistent with filling in part of the four base 5′ extension generated by HindIII. Similar results were seen when Δtdp1 cells were transfected with EcoRI-linearized DNA, with 1/100 misrepair events in WT cells versus 26/100 misrepair events in Δtdp1 cells. A different pattern was seen with DNA linearized with PstI, an enzyme that generates a 3′ extension. The error frequencies were similar in WT and Δtdp1 cells, with no insertions observed. Similar results were obtained with KpnI, another enzyme that generates 3′ extensions (Fig. S1a).

Fig. 1.

NHEJ efficiency and repair fidelity in WT and tdp1 deleted strains. (A) Plasmid YCplac111 linearized with HindIII, EcoR1, or Pst1 were transfected into BY4741 (WT) or a Δtdp1 derivative by using electroporation. Repair frequencies for each genotype are expressed as the ratio of colonies obtained with linear DNA divided by colonies obtained with uncut DNA × 100. The results shown are the mean of at least three independent transfections performed with the same reagents; error bars indicate SEM. (B) Accuracy of repair of linear DNA with 5′ extensions (HindIII and EcoRI) or 3′ extensions (PstI) was determined in WT and Δtdp1 strains. After transfection with YCplac111-linearized DNA, plasmids were isolated from colonies as described in Methods. Joints were amplified by PCR and redigested by the same restriction enzyme used to linearize the DNA. PCR samples that redigested were deemed accurately repaired. Samples that failed to digest were analyzed by DNA sequencing. The experiments analyzed 100 colonies per condition by design. Accurately repaired junctions are indicated in the line marked with a star, with spaces to allow the alignment of insertions. Junction sequences in red indicate insertions based on the left side filling in, and sequences in blue are insertions based on right-side fill-in reactions, although some assignments are arbitrary. A single nontemplated insertion of a cytosine was recovered in the Δtdp1 strain when transfected with HindIII-linearized DNA (indicated by the underlined C). The number of identical sequences recovered is indicated for each sequence.

We tested a variety of conditions to verify that the effect seen was only because of the deletion of TDP1. The experiments shown in Fig. 1 were carried out by using strains derived from BY4741. The same effect was seen by using other isogenic pairs of strains unrelated to BY4741 (Fig. S1b and c). Because the restriction sites in YCplac111 were part of a polylinker, we also assessed the repair of a unique HindIII site in YCp50. Repair was highly accurate in WT cells, whereas colonies from Δtdp1 transformants exhibited insertions like those seen with linearized YCplac111 (Fig. S1d).

Yeast Tdp1 Acts Through a Ku70/Ku80, DNA Ligase IV, and DNA Polymerase IV Dependent Pathway.

One hypothesis to explain our results is that Tdp1p is an inhibitor of an alternative end-joining pathway and that elimination of TDP1 allows this alternate pathway to function. We compared the repair efficiency and accuracy for HindIII-linearized YCplac111 in strains with deleted TDP1 that also carry deletions of genes encoding critical NHEJ components. Deletion of YKU80 or DNL4 causes a substantial decrease of efficiency of transformation with linearized plasmids (Fig. 2A), similar to results reported in refs. 16 and 17. Strains carrying Δtdp1 with either Δyku80 or Δdnl4 have the same reduced transformation efficiency as the NHEJ single mutants (Fig. 2A and Fig. S2a). Furthermore, we observed that Δyku80 or Δdnl4 single mutants have accurate repair of HindIII-linearized DNA. However, inaccurate repair in NHEJ mutants such as Δyku80 or Δdnl4 has been reported by others, suggesting that strain background may contribute to accuracy in some contexts (reviewed in ref. 18). Importantly, in our experiments, Δtdp1 Δyku80 or Δtdp1 Δdnl4 strains show the same accuracy of repair as the NHEJ single mutants, with no detected insertions (Fig. 2B). Similar results were also seen in other strains carrying deletions of TDP1 along with deletions of genes encoding other NHEJ components (Fig. S2b and c). We conclude that TDP1 prevents insertions that utilize canonical NHEJ components.

Fig. 2.

Effects of alterations in DNA repair pathways on inaccurate end joining when TDP1 is deleted. (A) Derivatives of BY4741 carrying deletions in NHEJ functions and/or TDP1 were constructed, and the efficiency of repair of HindIII-linearized YCplac111 was assessed. As in Fig. 1A, repair frequencies for each genotype are expressed as the ratio of colonies obtained with linear DNA divided by colonies obtained with uncut DNA × 100. Note that the transformation efficiency is shown on a logarithmic scale. (B) Accuracy of the repair of HindIII-linearized YCplac111 DNA was assessed in Δyku80 and Δyku80 Δtdp1 strains. As indicated, no misrepaired plasmids were recovered. Similar results were obtained with Δdnl4 and Δdnl4 Δtdp1 strains (Fig. S2). Accurately repaired junctions are indicated in the line marked with a star. (C) Accuracy of the repair of HindIII-linearized YCplac111 DNA was assessed in Δpol4 and Δpol4 Δtdp1 strains. (D) Accuracy of the repair of HindIII-linearized YCplac111 DNA was assessed in Δrad52 and Δrad52 Δtdp1 strains. (E) A derivative of JN362a was constructed carrying either a URA3 disruption of Δtdp1 or a replacement of the WT TDP1 gene with an allele encoding a His182Ala missense mutation. A more complete description of the junction sequences obtained is presented in Fig. S4. (F) TPP1 overexpression leads to misrepair of linearized plasmids in WT cells. By4741 cells carrying either pTW375 (WT yeast TPP1 under the control of the yeast Adh1 promoter) (24) or pTW375^D35A (as pTW375, but with a Tpp1 mutation changing Asp 35 to Ala) (24) were transfected with YCplac111 that had been linearized with HindIII. Accuracy of repair of the linearized DNA was determined as described in the legend for Fig. 1. The overall repair efficiency was similar to results presented in Fig. 1 for WT cells.

The insertions observed in the Δtdp1 mutants are consistent with the action of a DNA polymerase. Yeast DNA polymerase IV (POL4) participates in some NHEJ events (19, 20). Therefore, we tested the efficiency and accuracy of end joining in strains carrying a deletion of the POL4 alone or in combination with a deletion of TDP1. As observed previously (19, 21), we observed no change in the repair efficiency using linearized plasmids in Δpol4 strains compared to WT strains (Fig. 2A). In Δpol4 strains, repair of HindIII-linearized DNA was accurate, with 4/100 misrepair events (Fig. 2C). Δpol4 Δtdp1 strains had the same frequency of misrepair events (4/100) and the same frequency of insertions as the Δpol4 single mutant. These results show that the additions that Tdp1 prevents in the plasmid repair assay do not occur in the absence of Pol4p.

Although the events we examined depend primarily on NHEJ, we also examined whether homologous recombination functions are important for the misrepair of transfected linear DNA. Strains lacking RAD52, encoding a protein essential for homologous recombination in yeast, showed efficient and accurate repair of HindIII-linearized DNA (1/100 misrepaired, Fig. 2D). Strains lacking TDP1 and RAD52 had 18/100 misrepair events (p < 0.0001 compared to rad52 single mutant) with 10 insertions (p = 0.0015); therefore, the insertions do not require a functional RAD52 pathway.

To demonstrate that Tdp1p plays an enzymatic role in NHEJ rather than simply a structural role, we took advantage of the known residues that participate in Tdp1 catalysis. The enzyme active site includes two histidines, with His182 absolutely required for enzyme activity (10, 11, 22, 23). We replaced the chromosomal TDP1⁺ gene with a mutant version that changes His182 to Ala and assessed the accuracy of NHEJ in cells expressing the mutant gene. Previous work showed that the His182Ala was stably expressed in yeast (23). The His182Ala mutant showed a high error frequency when transfected with YCplac111 that had been linearized with HindIII. Cells carrying Tdp1(His182Ala) showed a frequency of insertions similar to the Δtdp1 strain (13/100 for His182Ala versus 15/100 for Δtdp1). Cells carrying Tdp1(His182Ala) also exhibited a higher frequency of deletions at the repair junction compared to cells carrying a complete deletion of TDP1 (19/100 compared to 3/100 for the isogenic Δtdp1 strain, p = 0.002, Fig. 2E and Figs. S3 and S4). This result suggests that the inactive Tdp1 protein may interfere with NHEJ more than the absence of the protein but, importantly, that the inactive mutant leads to insertions as does the complete absence of the protein.

Overexpression of the 3′ DNA Phosphatase Tpp1 Results in Inaccurate End Joining.

The simplest hypothesis that explains the role of Tdp1p in NHEJ is that the nucleosidase activity of the protein removes nucleosides during the NHEJ process. A unique aspect of the nucleolytic activity of Tdp1p is that the removal of a nucleoside leaves a 3′ PO₄, a unique DNA end that requires further processing before subsequent repair reactions. We hypothesized that the important role of Tdp1 in NHEJ is the generation of ends with a 3′ PO₄. If the 3′ PO₄ is relevant to the accuracy of end joining, we predicted that accelerating the removal of the 3′ PO₄ would have the same phenotypic effect as the absence of Tdp1p. Wilson and colleagues showed that Tpp1, a yeast phosphatase homolog of polynucleotide kinase, is involved in removal of the 3′ PO₄ following the repair of Top1-mediated DNA damage. Tpp1 has 3′ DNA phosphatase activity and lacks 5′ DNA phosphatase, 5′ polynucleotide kinase, and 3′ exonuclease activities; i.e., its only known activity is 3′ DNA phosphatase (24, 25). We overexpressed either WT Tpp1 or a mutant Tpp1 with a change at a key catalytic residue (Asp35 mutated to Ala). In cells overexpressing WT Tpp1, repair of a plasmid linearized with HindIII was inaccurate (Fig. 2F), with 10/100 samples showing filling-in reactions similar to those seen with Δtdp1 strains. The Tpp1 overexpressing cells also showed larger deletions than those seen in either WT or Δtdp1 strains (6/100 carried deletions of ≥19 nucleotides versus no deletions of that class in WT cells). By contrast, overexpressing the Tpp1 (D35A) mutant showed a significantly lower frequency of misrepair with 3/100 insertions (p = 0.031 comparing the overall frequency of misrepair in cells overexpressing WT versus mutant Tpp1). This result indicates that overexpressed enzymatically active Tpp1 reduces the accuracy of end joining and leads to insertions in the same way as deletion of TDP1. Overexpression of WT Tpp1 did not change the frequency of errors in a Δtdp1 strain, suggesting that the effect of Tpp1 overexpression is epistatic to loss of function of TDP1 (Fig. S5). We conclude that the preservation of accuracy in NHEJ is assisted by the transient generation of DNA ends with a 3′ PO₄ and that reducing the lifetime of the 3′ PO₄ largely eliminates the enhanced accuracy.

Effect of Tdp1 Deletion on Substrates that Require Processing Prior to Ligation.

In the substrates examined above, accurate repair can be accomplished by direct ligation of the DNA substrate. We also examined repair of a plasmid substrate that absolutely requires processing prior to ligation. We digested YCplac111 DNA with both HindIII and SalI and transfected the linear DNA into WT and Δtdp1 strains. Processing will necessarily result in sequence alterations at the repair junctions, so we directly carried out sequence analysis of 50 independent clones from WT and Δtdp1 strains. The majority of events in Δtdp1 strains were filling-in events, with 33/50 being a complete filling in of both the HindIII and SalI 5′ extensions (Fig. 3A and Fig. S6a). The remaining 17/50 isolates carried plasmids with partial fill-in reactions. There were no deletions seen in any of the plasmids isolated from the Δtdp1 strain. By contrast, filling in also occurred in WT cells, although less efficiently (only 1/50 showed complete filling in of both the HindIII and SalI 5′ extensions). In addition to the partial fill-in reactions (34/50), we also observed a substantial portion of large deletions (15/50). Therefore, in DNA with 5′ extensions that are not directly ligatable, there is a major difference in the repair between WT and Δtdp1 strains. Loss of TDP1 function eliminated the deletions seen in WT cells. Loss of Tdp1 activity also suppressed deletions occurring because of the repair of a plasmid linearized with HindIII and PstI (one end with a 5′ extension and one end with a 3′ extension, Fig. 3B and Fig. S6b.)

Fig. 3.

Analysis of plasmids obtained following transformation with DNA with incompatible ends. (A) Plasmid yCPlac11 was cut with HindIII and Sal1 and introduced into WT and Δtdp1 strains, which were analyzed by DNA sequencing for the nature of the repaired junctions. The events were rationalized by assuming fill-in reactions of the HindIII end or the SalI end of the molecule. Filling in of the HindIII end is shown as red nucleotides, and filling in of the SalI end is shown with blue nucleotides. For WT cells, 27/50 showed additions consistent with partial filling in of one of the two sides and 1/50 showed a complete filling in of the extensions By contrast, 33/50 of the plasmids from tdp1 showed complete filling in, and 11 more showed almost complete filling in. No deletions were recovered from tdp1 cells. A complete sequence analysis is presented in Fig. S6a. (B) Colonies obtained from transformation with yCPlac11 digested with HindIII and Pst1 were analyzed for the nature of the repaired junctions. Events on the HindIII cut end were rationalized as before by assuming filling in of the 5′ extension (shown as red nucleotides), whereas events at the Pst1 side were rationalized on the basis of preserving the 3′ extension (shown with blue nucleotides). Seven large deletions (> 10 nucleotides) were recovered in WT cells compared to none in tdp1 cells. A complete sequence analysis is presented in Fig. S6b.

We also tested whether Δtdp1 strains have an altered efficiency of repairing DNA double-strand breaks with blunt ends. As previously observed (26), linear plasmids with blunt ends transform yeast cells less efficiently than linear DNA with cohesive ends (Fig. 4A). However, we found a 5- to 6-fold increase in transformation efficiency of YCplac111 cut with Sma1 in a Δtdp1 strain compared to an isogenic WT strain. The enhanced efficiency was also observed in strains carrying the His182Ala allele of TDP1 (Fig. S4). Because we observed an increase in repair efficiency, we examined the possibility that the events in Δtdp1 strains occurred by a pathway that was independent of NHEJ. Combining mutations in NHEJ functions along with Δtdp1 reduced the transformation efficiency to the level seen in NHEJ single mutants. Therefore, the enhanced efficiency of repairing blunt ends occurs through the NHEJ pathway.

Fig. 4.

NHEJ efficiency and repair fidelity in WT and tdp1-deleted strains transfected with DNA with blunt ends. (A) YCplac111 cut with Sma1 was transfected into yeast by using the same methods as in Figs. 1 and 2. (B) Strains transfected with DNA with blunt ends show considerably less accurate repair than DNA with cohesive ends. Accurately repaired junctions are indicated in the line marked with a star. For WT cells 61/100 were repaired accurately, whereas 67/100 were repaired accurately in Δtdp1 strains, an insignificant difference (p = 0.462 by using Fisher’s exact test). There was no obvious preference for the size of the deletions in WT versus Δtdp1 strains. (C) YCplac111 was cut with both HindIII and SalI. All samples were analyzed by DNA sequencing, with 50 independent plasmids each from WT and Δtdp1 strains analyzed. The figure shows a summary of the insertion and deletion events; a more complete listing of the recovered events is shown in Fig. S3. The overall frequency of repaired plasmids with insertions was statistically different between WT and Δtdp1 strains (all Δtdp1 strains carried insertions, p < 0.0001) as was the frequency of deletions in WT cells.

A surprising result was seen when we tested transformation of a Δtdp1 Δpol4 strain with DNA with blunt ends. Δpol4 single mutants exhibited the same efficiency of repairing blunt ends as WT cells. Although there is no requirement for POL4 in repairing blunt ends, the Δtdp1 Δpol4 strain did not show an enhancement of joining blunt ends (Fig. 4A). Therefore, POL4 is required for TDP1-mediated stimulation of blunt end repair. We also examined the repair fidelity of DNA digested with SmaI in WT and Δtdp1 strains. Both strains showed approximately 60%–70% accurate repair, with no significant difference between the two strains (Fig. 4B). However, the low frequency of accurate repair with blunt ends suggests that substantial processing frequently occurs prior to religation. Our results suggest that Pol4 plays an important role in effective processing blunt ends in NHEJ in yeast and that WT TDP1 may antagonize the ability of Pol4 to effectively function with blunt ends. The results taken together indicate that Tdp1 acts on a wide variety of DNA substrates during NHEJ, and the absence of Tdp1 leads to major changes in the products of NHEJ reactions.

Discussion

NHEJ mediates rejoining of DNA breaks with a minimal dependence on DNA homology, resulting in repair that is intrinsically error-prone. Whereas this repair process is likely to make errors, both in choice of partners for rejoining and in preservation of the DNA sequence at the break junction, the accessory proteins that function in this pathway influence overall error rates and patterns. We showed that yeast Tdp1 is an accessory protein for NHEJ and that loss of Tdp1 activity mainly affects end joining by changing the accuracy at repair junctions. All misrepair events that are prevented by Tdp1p activity require Ku70/Ku80 and DNA ligase IV indicating that Tdp1 is acting as part of the NHEJ pathway, rather than through an alternate end-joining pathway.

We hypothesize that Tdp1 acts on DNA ends, either directly counteracting polymerase additions by immediate removal of the added nucleoside or by removing a nucleoside prior to any polymerization and leaving a 3′ PO₄, which cannot be extended by a polymerase (Fig. 5). We propose that the unique quality of Tdp1 that makes this enzyme suitable for regulating DNA metabolic events is that it is a 3′ nucleosidase (shown for the human enzyme in ref. 12 and for the yeast enzyme in Fig. S7). The enzyme cannot act as a processive nuclease without an additional activity such as the phosphatase Tpp1 or in mammalian cell polynucleotide kinase (12). The reaction of Tdp1 temporarily restricts other enzymes that act at DNA ends including DNA polymerases and DNA ligase. Once Tdp1 removes a nucleoside, the DNA end cannot be used by ligase until the 3′ PO₄ is removed. Therefore, reactivation of 3′ ends is a critical step in all pathways that involve Tdp1.

Fig. 5.

A model for suppression of additions by Tdp1 nucleosidase activity. We suggest that Tdp1 acts directly on 3′ DNA ends during NHEJ. The nucleosidase activity of Tdp1 removes a single nucleoside, leaving a terminal 3′ phosphate. This blocks polymerization and ligation until the phosphate is removed. Complete repair of the double-strand break will require removal of the phosphate by Tpp1 or removal of the nucleotide phosphate by a nuclease activity such as Apn1. Polymerization is required to repair the gap generated by Tdp1 nucleosidase activity prior to ligation. Recent results suggest that polymerase activity is required for efficient sealing of cohesive ends with 5′ extensions (33), supporting the hypothesis that some nucleotides are removed during the repair of DNA with cohesive ends. When Tdp1 activity is absent, DNA polymerases can extend a primed template, leading to nucleotide insertions. The simplest hypothesis based on our results is that the insertions result from additions by Pol4p.

Wilson and colleagues assessed the path of 3′ end activation following Tdp1 reaction in the context of camptothecin sensitivity (25, 27). In addition to Tpp1, yeast 3′ end processing can be carried out by the apurinic endonucleases Apn1 and Apn2 (24, 25). Mutations in TPP1, APN1, or APN2 confer enhanced camptothecin sensitivity, but only in cells with a WT TDP1 gene. Therefore, these three enzymes are capable of processing 3′ PO₄ ends that arise from removal of Top1 covalently bound to DNA by Tdp1p. Whereas many 3′ to 5′ exonucleases are unable to act on DNA ends terminated by a 3′ PO₄, endonucleases may also be able to reactivate DNA ends acted upon by Tdp1 (28).

Because Tpp1 overexpression in WT cells leads to insertions, we think that the most important aspect of Tdp1 is the generation of 3′ PO₄ end rather than removal of nucleotides at DNA ends. We hypothesize that Tpp1 overexpression reduces the amount of time that ends are inactive because of the presence of a 3′ PO₄ end. There is no other known mechanism in yeast for generating a 3′ PO₄ at a DNA end besides Tdp1 action, and Tpp1 has no other known activity besides a 3′ phosphatase. Because we observe that WT cells that overpress Tpp1 also have a higher level of deletions than cells without Tpp1 overexpression, we think that deletions in this context may arise from the concerted action of Tdp1 and Tpp1 to processively remove nucleotides from DNA ends. This may also be a mechanism for Tdp1-dependent deletions that arise when DNA ends cannot be directly ligated (as seen in the experiments in Fig. 3).

Because the concerted action of Tdp1 and Tpp1 results in a futile cycle (albeit with the removal of one nucleotide), how might Tdp1 action contribute to the overall accuracy of end joining? A speculative possibility is that Tdp1 acts on ends prior to synapsis mediated by the end-joining machinery. In this view, Tpp1 or Apn1 reactivates the 3′ ends only after synapsis is established. This hypothesis suggests that proteins capable of reactivating DNA ends might be regulated by Ku70/Ku80 or other proteins that are involved in establishing synapsis.

Our results also provide an explanation for a surprising aspect of NHEJ in yeast. Whereas yeast efficiently repairs cohesive ends in a Ku70/Ku80/DNA ligase IV dependent pathway, repair of blunt ends is inefficient and largely NHEJ-independent. Results presented in Fig. 4A support this view. However, in the absence of TDP1, yeast cells can use blunt ends for NHEJ as efficiently as staggered ends. This enhanced repair efficiency required NHEJ functions along with POL4. We suggest that joining blunt ends requires POL4 following an initial processing step. In WT cells, Tdp1 action blocks POL4 and prevents processing that leads to effective end joining. When Tdp1 is absent, Pol4 is able to promote efficient joining of blunt ends.

Whereas we propose regulation of Pol4 (or other polymerases) by Tdp1 action, the regulation of events at DNA ends may include negative regulation of DNA ligase IV, because the 3′ PO₄ would also prevent ligation. The regulation of events at DNA ends may also include accessibility to exonucleases. Several 3′ to 5′ exonucleases from mammalian cells Trex1 and Wrn are inactive on ends with a 3′ PO₄ group (29, 30). Therefore, many of the enzymes acting on DNA will be blocked following Tdp1 action. Whereas our experiments have tested the role of Tdp1 in NHEJ, we suggest that the mechanisms proposed here might also be relevant for other DNA metabolic processes including homologous recombination and utilization of error-prone DNA polymerases.

Materials and Methods

Yeast Strains.

All yeast strains were constructed by using either BY4741 or JN362a. Strains were constructed by using standard yeast genetic approaches. Details of the genotypes of the strains used and construction of strains are presented in SI Text.

Plasmid and Yeast Transformation Procedures.

The yeast centromeric plasmids YCplac111 or YCplac33 were digested with indicated restriction enzymes. Linearized DNA was purified by electrophoresis using a 0.8% agarose gel, followed by extraction using QIAquick Gel Extraction kit (Qiagen). Transformation with an undigested plasmid was performed in parallel to determine transformation efficiency. For yeast transformation assays, cells were grown overnight in yeast extract, peptone, dextrose, adenine medium (YPDA), diluted the following morning, and grown until the suspension reached an OD₆₀₀ of approximately 1. Typically, 100 ng of linearized or supercoiled DNA were transformed into yeast by using electroporation (31). We chose to use a carrier-free transformation procedure to eliminate possible complications from plasmid recombination reactions with carrier DNA (19, 32). Electroporation conditions were 0.75 V, 25 μF, and 200 Ω by using cuvettes with a 0.1-cm gap width (Bio-Rad). After electroporation, cells were plated onto appropriate selective media. The plates were incubated at 30 °C for 3–4 days, and colonies were counted. Relative repair efficiency was expressed as the ratio of transformants obtained by using linearized plasmids divided by the number of colonies obtained with circular plasmids × 100.

Breakpoint Junction Analysis.

Plates containing less than 100 colonies were chosen, and all the colonies on the plate were used for DNA isolation. Plasmid DNA was extracted from the colonies after transformation using the Zymolase method. Briefly, 1 mm³ of cells from a single colony was resuspended in 50 μL of a solution containing 1.2 M sorbitol, 0.1 M Na PO₄, pH 7.4, and 60 U/mL of Zymolase 100 T in a 1.7-mL polypropylene tube and incubated at 37° for 30 min. The extracts were then incubated in a boiling water bath for 10 min and then used directly (typically 2.5 μL) for PCRs. The region containing the restriction enzyme site used to linearize the plasmid was amplified. The specific PCR primers used for yCPLac111 were (5′-TAGCCGTAGTTAGGCCACCAC-3′) and (5′-ACCGCACAGATGCGTAAGGAG-3′). PCR products were then digested with the same restriction enzyme initially used to linearize the plasmid. Digested products were scored as accurately repaired, whereas PCR products that were resistant to digestion were sequenced by using the primer (5′-CCAATACGCAAACCGCCTCTCC-3′). For analyses where accurate repair led to regeneration of a restriction site, 100 individual colonies were analyzed for the accuracy of plasmid repair from each yeast transformation. For assays where repair would not be expected to regenerate a restriction site, 50 colonies were analyzed by DNA sequencing. The significance of differences in the accuracy of repair (or the frequency of particular classes of events) was assessed by using Fisher’s exact test.